mariadb/mysys/my_bitmap.c

709 lines
18 KiB
C
Raw Normal View History

2011-07-04 01:25:49 +02:00
/*
2011-11-03 19:17:05 +01:00
Copyright (c) 2001, 2011, Oracle and/or its affiliates.
Copyright (C) 2009- 2011 Monty Program Ab
Update copyright Fixed memory leak on shutdown (Affects the embedded version & MyODBC) client/client_priv.h: Update copyright client/completion_hash.cc: Update copyright client/completion_hash.h: Update copyright client/connect_test.c: Update copyright client/errmsg.c: Update copyright client/get_password.c: Update copyright client/insert_test.c: Update copyright client/list_test.c: Update copyright client/my_readline.h: Update copyright client/mysql.cc: Update copyright client/mysqladmin.c: Update copyright client/mysqlbinlog.cc: Update copyright client/mysqlcheck.c: Update copyright client/mysqldump.c: Update copyright client/mysqlimport.c: Update copyright client/mysqlmanager-pwgen.c: Update copyright client/mysqlmanagerc.c: Update copyright client/mysqlshow.c: Update copyright client/mysqltest.c: Update copyright client/password.c: Update copyright client/readline.cc: Update copyright client/select_test.c: Update copyright client/showdb_test.c: Update copyright client/sql_string.cc: Update copyright client/sql_string.h: Update copyright client/ssl_test.c: Update copyright client/thimble.cc: Update copyright client/thread_test.c: Update copyright div/deadlock_test.c: Update copyright extra/comp_err.c: Update copyright extra/my_print_defaults.c: Update copyright extra/perror.c: Update copyright extra/replace.c: Update copyright extra/resolve_stack_dump.c: Update copyright extra/resolveip.c: Update copyright fs/database.c: Update copyright fs/libmysqlfs.c: Update copyright fs/mysqlcorbafs.c: Update copyright fs/mysqlcorbafs.h: Update copyright fs/mysqlcorbafs_test.c: Update copyright heap/_check.c: Update copyright heap/_rectest.c: Update copyright heap/heapdef.h: Update copyright heap/hp_block.c: Update copyright heap/hp_clear.c: Update copyright heap/hp_close.c: Update copyright heap/hp_create.c: Update copyright heap/hp_delete.c: Update copyright heap/hp_extra.c: Update copyright heap/hp_hash.c: Update copyright heap/hp_info.c: Update copyright heap/hp_open.c: Update copyright heap/hp_panic.c: Update copyright heap/hp_rename.c: Update copyright heap/hp_rfirst.c: Update copyright heap/hp_rkey.c: Update copyright heap/hp_rlast.c: Update copyright heap/hp_rnext.c: Update copyright heap/hp_rprev.c: Update copyright heap/hp_rrnd.c: Update copyright heap/hp_rsame.c: Update copyright heap/hp_scan.c: Update copyright heap/hp_static.c: Update copyright heap/hp_test1.c: Update copyright heap/hp_test2.c: Update copyright heap/hp_update.c: Update copyright heap/hp_write.c: Update copyright include/config-win.h: Update copyright include/dbug.h: Update copyright include/errmsg.h: Update copyright include/ft_global.h: Update copyright include/getopt.h: Update copyright include/hash.h: Update copyright include/heap.h: Update copyright include/m_ctype.h: Update copyright include/m_string.h: Update copyright include/md5.h: Update copyright include/merge.h: Update copyright include/my_alarm.h: Update copyright include/my_base.h: Update copyright include/my_bitmap.h: Update copyright include/my_dir.h: Update copyright include/my_global.h: Update copyright include/my_list.h: Update copyright include/my_net.h: Update copyright include/my_no_pthread.h: Update copyright include/my_nosys.h: Update copyright include/my_pthread.h: Update copyright include/my_sys.h: Update copyright include/my_tree.h: Update copyright include/myisam.h: Update copyright include/myisammrg.h: Update copyright include/myisampack.h: Update copyright include/mysql.h: Update copyright include/mysql_com.h: Update copyright include/mysql_embed.h: Update copyright include/mysqld_error.h: Update copyright include/mysys_err.h: Update copyright include/nisam.h: Update copyright include/queues.h: Update copyright include/raid.h: Update copyright include/sslopt-case.h: Update copyright include/sslopt-longopts.h: Update copyright include/sslopt-usage.h: Update copyright include/sslopt-vars.h: Update copyright include/t_ctype.h: Update copyright include/thr_alarm.h: Update copyright include/thr_lock.h: Update copyright include/violite.h: Update copyright isam/_cache.c: Update copyright isam/_dbug.c: Update copyright isam/_key.c: Update copyright isam/_locking.c: Update copyright isam/_packrec.c: Update copyright isam/_page.c: Update copyright isam/_search.c: Update copyright isam/_statrec.c: Update copyright isam/changed.c: Update copyright isam/close.c: Update copyright isam/create.c: Update copyright isam/delete.c: Update copyright isam/extra.c: Update copyright isam/info.c: Update copyright isam/isamchk.c: Update copyright isam/isamdef.h: Update copyright isam/log.c: Update copyright isam/open.c: Update copyright isam/panic.c: Update copyright isam/range.c: Update copyright isam/rfirst.c: Update copyright isam/rkey.c: Update copyright isam/rlast.c: Update copyright isam/rnext.c: Update copyright isam/rprev.c: Update copyright isam/rrnd.c: Update copyright isam/rsame.c: Update copyright isam/rsamepos.c: Update copyright isam/sort.c: Update copyright isam/static.c: Update copyright isam/test1.c: Update copyright isam/test2.c: Update copyright isam/test3.c: Update copyright isam/update.c: Update copyright isam/write.c: Update copyright libmysql/conf_to_src.c: Update copyright libmysql/dll.c: Update copyright libmysql/errmsg.c: Update copyright libmysql/get_password.c: Update copyright libmysql/libmysql.c: Update copyright libmysql/manager.c: Update copyright libmysql/net.c: Update copyright libmysql/password.c: Update copyright libmysqld/lib_sql.cc: Update copyright libmysqld/lib_vio.c: Update copyright libmysqld/libmysqld.c: Update copyright merge/mrg_close.c: Update copyright merge/mrg_create.c: Update copyright merge/mrg_def.h: Update copyright merge/mrg_delete.c: Update copyright merge/mrg_extra.c: Update copyright merge/mrg_info.c: Update copyright merge/mrg_locking.c: Update copyright merge/mrg_open.c: Update copyright merge/mrg_panic.c: Update copyright merge/mrg_rrnd.c: Update copyright merge/mrg_rsame.c: Update copyright merge/mrg_static.c: Update copyright merge/mrg_update.c: Update copyright myisam/ft_boolean_search.c: Update copyright myisam/ft_dump.c: Update copyright myisam/ft_eval.h: Update copyright myisam/ft_static.c: Update copyright myisam/ft_stem.c: Update copyright myisam/ft_stopwords.c: Update copyright myisam/ft_test1.h: Update copyright myisam/mi_cache.c: Update copyright myisam/mi_changed.c: Update copyright myisam/mi_check.c: Update copyright myisam/mi_checksum.c: Update copyright myisam/mi_close.c: Update copyright myisam/mi_create.c: Update copyright myisam/mi_dbug.c: Update copyright myisam/mi_delete.c: Update copyright myisam/mi_delete_all.c: Update copyright myisam/mi_delete_table.c: Update copyright myisam/mi_dynrec.c: Update copyright myisam/mi_extra.c: Update copyright myisam/mi_info.c: Update copyright myisam/mi_key.c: Update copyright myisam/mi_locking.c: Update copyright myisam/mi_log.c: Update copyright myisam/mi_open.c: Update copyright myisam/mi_packrec.c: Update copyright myisam/mi_page.c: Update copyright myisam/mi_panic.c: Update copyright myisam/mi_range.c: Update copyright myisam/mi_rename.c: Update copyright myisam/mi_rfirst.c: Update copyright myisam/mi_rlast.c: Update copyright myisam/mi_rnext_same.c: Update copyright myisam/mi_rrnd.c: Update copyright myisam/mi_rsame.c: Update copyright myisam/mi_rsamepos.c: Update copyright myisam/mi_scan.c: Update copyright myisam/mi_search.c: Update copyright myisam/mi_static.c: Update copyright myisam/mi_statrec.c: Update copyright myisam/mi_test1.c: Update copyright myisam/mi_test2.c: Update copyright myisam/mi_test3.c: Update copyright myisam/mi_unique.c: Update copyright myisam/mi_update.c: Update copyright myisam/mi_write.c: Update copyright myisam/myisamchk.c: Update copyright myisam/myisampack.c: Update copyright myisammrg/myrg_close.c: Update copyright myisammrg/myrg_create.c: Update copyright myisammrg/myrg_def.h: Update copyright myisammrg/myrg_delete.c: Update copyright myisammrg/myrg_locking.c: Update copyright myisammrg/myrg_open.c: Update copyright myisammrg/myrg_panic.c: Update copyright myisammrg/myrg_rsame.c: Update copyright myisammrg/myrg_static.c: Update copyright myisammrg/myrg_update.c: Update copyright myisammrg/myrg_write.c: Update copyright mysql-test/r/gcc296.result: Update of benchmark results mysql-test/r/innodb.result: Update of benchmark results mysql-test/r/join_outer.result: Update of benchmark results mysql-test/r/myisam.result: Update of benchmark results mysys/array.c: Update copyright mysys/charset.c: Fix for restart of character sets mysys/checksum.c: Update copyright mysys/default.c: Update copyright mysys/errors.c: Update copyright mysys/getopt.c: Cleanup mysys/getvar.c: Update copyright mysys/hash.c: Update copyright mysys/list.c: Update copyright mysys/make-conf.c: Update copyright mysys/md5.c: Update copyright mysys/mf_brkhant.c: Update copyright mysys/mf_cache.c: Update copyright mysys/mf_casecnv.c: Update copyright mysys/mf_dirname.c: Update copyright mysys/mf_fn_ext.c: Update copyright mysys/mf_format.c: Update copyright mysys/mf_getdate.c: Update copyright mysys/mf_iocache.c: Update copyright mysys/mf_iocache2.c: Update copyright mysys/mf_keycache.c: Update copyright mysys/mf_loadpath.c: Update copyright mysys/mf_pack.c: Update copyright mysys/mf_path.c: Update copyright mysys/mf_qsort.c: Update copyright mysys/mf_qsort2.c: Update copyright mysys/mf_radix.c: Update copyright mysys/mf_same.c: Update copyright mysys/mf_sleep.c: Update copyright mysys/mf_sort.c: Update copyright mysys/mf_soundex.c: Update copyright mysys/mf_stripp.c: Update copyright mysys/mf_tempfile.c: Update copyright mysys/mf_unixpath.c: Update copyright mysys/mf_util.c: Update copyright mysys/mf_wcomp.c: Update copyright mysys/mf_wfile.c: Update copyright mysys/mulalloc.c: Update copyright mysys/my_alarm.c: Update copyright mysys/my_alloc.c: Update copyright mysys/my_append.c: Update copyright mysys/my_bit.c: Update copyright mysys/my_bitmap.c: Update copyright mysys/my_chsize.c: Update copyright mysys/my_clock.c: Update copyright mysys/my_compress.c: Update copyright mysys/my_copy.c: Update copyright mysys/my_create.c: Update copyright mysys/my_delete.c: Update copyright mysys/my_div.c: Update copyright mysys/my_dup.c: Update copyright mysys/my_error.c: Update copyright mysys/my_fopen.c: Update copyright mysys/my_fstream.c: Update copyright mysys/my_getwd.c: Update copyright mysys/my_init.c: Free 'once_alloc' memory at shutdown. mysys/my_lib.c: Update copyright mysys/my_lock.c: Update copyright mysys/my_lockmem.c: Update copyright mysys/my_lread.c: Update copyright mysys/my_lwrite.c: Update copyright mysys/my_malloc.c: Update copyright mysys/my_messnc.c: Update copyright mysys/my_mkdir.c: Update copyright mysys/my_net.c: Update copyright mysys/my_once.c: Update copyright mysys/my_open.c: Update copyright mysys/my_pread.c: Update copyright mysys/my_pthread.c: Update copyright mysys/my_quick.c: Update copyright mysys/my_read.c: Update copyright mysys/my_realloc.c: Update copyright mysys/my_redel.c: Update copyright mysys/my_rename.c: Update copyright mysys/my_seek.c: Update copyright mysys/my_static.c: Update copyright mysys/my_static.h: Update copyright mysys/my_symlink.c: Update copyright mysys/my_symlink2.c: Update copyright mysys/my_tempnam.c: Update copyright mysys/my_thr_init.c: Update copyright mysys/my_vsnprintf.c: Update copyright mysys/my_wincond.c: Update copyright mysys/my_winthread.c: Update copyright mysys/my_write.c: Update copyright mysys/mysys_priv.h: Update copyright mysys/ptr_cmp.c: Update copyright mysys/queues.c: Update copyright mysys/raid.cc: Update copyright mysys/safemalloc.c: Update copyright mysys/string.c: Update copyright mysys/test_charset.c: Update copyright mysys/test_dir.c: Update copyright mysys/test_fn.c: Update copyright mysys/testhash.c: Update copyright mysys/thr_alarm.c: Update copyright mysys/thr_lock.c: Update copyright mysys/thr_mutex.c: Update copyright mysys/thr_rwlock.c: Update copyright mysys/tree.c: Update copyright mysys/typelib.c: Update copyright pstack/debug.c: Update copyright pstack/debug.h: Update copyright pstack/demangle.h: Update copyright pstack/ieee.c: Update copyright pstack/ieee.h: Update copyright pstack/pstack.c: Update copyright readline/bind.c: Cleanup empty lines readline/complete.c: Cleanup empty lines readline/display.c: Cleanup empty lines readline/funmap.c: Cleanup empty lines readline/histexpand.c: Cleanup empty lines readline/histfile.c: Cleanup empty lines readline/history.c: Cleanup empty lines readline/history.h: Cleanup empty lines readline/input.c: Cleanup empty lines readline/kill.c: Cleanup empty lines readline/readline.c: Cleanup empty lines readline/readline.h: Cleanup empty lines readline/vi_mode.c: Cleanup empty lines sql/cache_manager.cc: Update copyright sql/cache_manager.h: Update copyright sql/convert.cc: Update copyright sql/custom_conf.h: Update copyright sql/derror.cc: Update copyright sql/field.cc: Update copyright sql/field.h: Update copyright sql/field_conv.cc: Update copyright sql/filesort.cc: Update copyright sql/frm_crypt.cc: Update copyright sql/ha_berkeley.cc: Update copyright sql/ha_heap.cc: Update copyright sql/ha_heap.h: Update copyright sql/ha_innobase.cc: Update copyright sql/ha_isam.cc: Update copyright sql/ha_isam.h: Update copyright sql/ha_isammrg.cc: Update copyright sql/ha_isammrg.h: Update copyright sql/ha_myisam.cc: Update copyright sql/handler.cc: Update copyright sql/hash_filo.cc: Update copyright sql/hash_filo.h: Update copyright sql/hostname.cc: Update copyright sql/init.cc: Update copyright sql/item.cc: Update copyright sql/item.h: Update copyright sql/item_buff.cc: Update copyright sql/item_cmpfunc.cc: Update copyright sql/item_cmpfunc.h: Update copyright sql/item_create.cc: Update copyright sql/item_create.h: Update copyright sql/item_func.cc: Update copyright sql/item_strfunc.cc: Update copyright sql/item_sum.cc: Update copyright sql/item_sum.h: Update copyright sql/item_timefunc.cc: Update copyright sql/item_timefunc.h: Update copyright sql/item_uniq.cc: Update copyright sql/item_uniq.h: Update copyright sql/key.cc: Update copyright sql/lex_symbol.h: Update copyright sql/lock.cc: Update copyright sql/log.cc: Update copyright sql/log_event.cc: Update copyright sql/log_event.h: Update copyright sql/matherr.c: Update copyright sql/mf_iocache.cc: Update copyright sql/mini_client.cc: Update copyright sql/mini_client.h: Update copyright sql/my_lock.c: Update copyright sql/mysqld.cc: Update copyright sql/net_pkg.cc: Update copyright sql/net_serv.cc: Update copyright sql/opt_sum.cc: Update copyright sql/password.c: Update copyright sql/procedure.cc: Update copyright sql/procedure.h: Update copyright sql/records.cc: Update copyright sql/repl_failsafe.cc: Update copyright sql/slave.cc: Update copyright sql/slave.h: Update copyright sql/sql_acl.cc: Update copyright sql/sql_acl.h: Update copyright sql/sql_analyse.cc: Update copyright sql/sql_analyse.h: Update copyright sql/sql_base.cc: Update copyright sql/sql_cache.cc: Update copyright sql/sql_class.cc: Update copyright sql/sql_class.h: Update copyright sql/sql_crypt.cc: Update copyright sql/sql_crypt.h: Update copyright sql/sql_db.cc: Update copyright sql/sql_delete.cc: Update copyright sql/sql_handler.cc: Update copyright sql/sql_insert.cc: Update copyright sql/sql_lex.cc: Update copyright sql/sql_lex.h: Update copyright sql/sql_list.cc: Update copyright sql/sql_list.h: Update copyright sql/sql_load.cc: Update copyright sql/sql_map.cc: Update copyright sql/sql_map.h: Update copyright sql/sql_parse.cc: Update copyright sql/sql_rename.cc: Update copyright sql/sql_repl.cc: Update copyright sql/sql_select.h: Update copyright sql/sql_string.cc: Update copyright sql/sql_string.h: Update copyright sql/sql_table.cc: Update copyright sql/sql_test.cc: Update copyright sql/sql_udf.cc: Update copyright sql/sql_udf.h: Update copyright sql/stacktrace.c: Update copyright sql/structs.h: Update copyright sql/table.cc: Update copyright sql/table.h: Update copyright sql/thr_malloc.cc: Update copyright sql/time.cc: Update copyright sql/udf_example.cc: Update copyright sql/uniques.cc: Update copyright sql/unireg.cc: Update copyright sql/unireg.h: Update copyright strings/atof.c: Update copyright strings/bchange.c: Update copyright strings/bcmp.c: Update copyright strings/bcopy-duff.c: Update copyright strings/bfill.c: Update copyright strings/bmove.c: Update copyright strings/bmove512.c: Update copyright strings/bmove_upp.c: Update copyright strings/bzero.c: Update copyright strings/conf_to_src.c: Update copyright strings/ctype-big5.c: Update copyright strings/ctype-czech.c: Update copyright strings/ctype-euc_kr.c: Update copyright strings/ctype-gb2312.c: Update copyright strings/ctype-gbk.c: Update copyright strings/ctype-latin1_de.c: Update copyright strings/ctype-sjis.c: Update copyright strings/ctype-tis620.c: Update copyright strings/ctype-ujis.c: Update copyright strings/ctype.c: Update copyright strings/do_ctype.c: Update copyright strings/int2str.c: Update copyright strings/is_prefix.c: Update copyright strings/llstr.c: Update copyright strings/longlong2str.c: Update copyright strings/memcmp.c: Update copyright strings/memcpy.c: Update copyright strings/memset.c: Update copyright strings/r_strinstr.c: Update copyright strings/str2int.c: Update copyright strings/str_test.c: Update copyright strings/strappend.c: Update copyright strings/strcat.c: Update copyright strings/strcend.c: Update copyright strings/strchr.c: Update copyright strings/strcmp.c: Update copyright strings/strcont.c: Update copyright strings/strend.c: Update copyright strings/strfill.c: Update copyright strings/strings-not-used.h: Update copyright strings/strinstr.c: Update copyright strings/strlen.c: Update copyright strings/strmake.c: Update copyright strings/strmov.c: Update copyright strings/strnlen.c: Update copyright strings/strnmov.c: Update copyright strings/strrchr.c: Update copyright strings/strstr.c: Update copyright strings/strto.c: Update copyright strings/strtol.c: Update copyright strings/strtoll.c: Update copyright strings/strtoul.c: Update copyright strings/strtoull.c: Update copyright strings/strxmov.c: Update copyright strings/strxnmov.c: Update copyright strings/t_ctype.h: Update copyright strings/udiv.c: Update copyright tools/mysqlmanager.c: Update copyright vio/test-ssl.c: Update copyright vio/test-sslclient.c: Update copyright vio/test-sslserver.c: Update copyright vio/vio.c: Update copyright vio/viosocket.c: Update copyright vio/viossl.c: Update copyright vio/viosslfactories.c: Update copyright vio/viotest-ssl.c: Update copyright
2001-12-06 13:10:51 +01:00
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
Many files: Changed header to GPL version 2 only BUILD/Makefile.am: Changed header to GPL version 2 only Docs/Makefile.am: Changed header to GPL version 2 only Makefile.am: Changed header to GPL version 2 only SSL/Makefile.am: Changed header to GPL version 2 only bdb/Makefile.in: Changed header to GPL version 2 only client/Makefile.am: Changed header to GPL version 2 only client/client_priv.h: Changed header to GPL version 2 only client/completion_hash.cc: Changed header to GPL version 2 only client/completion_hash.h: Changed header to GPL version 2 only client/get_password.c: Changed header to GPL version 2 only client/my_readline.h: Changed header to GPL version 2 only client/mysql.cc: Changed header to GPL version 2 only client/mysql_upgrade.c: Changed header to GPL version 2 only client/mysqladmin.cc: Changed header to GPL version 2 only client/mysqlbinlog.cc: Changed header to GPL version 2 only client/mysqlcheck.c: Changed header to GPL version 2 only client/mysqldump.c: Changed header to GPL version 2 only client/mysqlimport.c: Changed header to GPL version 2 only client/mysqlmanager-pwgen.c: Changed header to GPL version 2 only client/mysqlmanagerc.c: Changed header to GPL version 2 only client/mysqlshow.c: Changed header to GPL version 2 only client/mysqltest.c: Changed header to GPL version 2 only client/readline.cc: Changed header to GPL version 2 only client/sql_string.cc: Changed header to GPL version 2 only client/sql_string.h: Changed header to GPL version 2 only cmd-line-utils/Makefile.am: Changed header to GPL version 2 only dbug/Makefile.am: Changed header to GPL version 2 only extra/Makefile.am: Changed header to GPL version 2 only extra/charset2html.c: Changed header to GPL version 2 only extra/comp_err.c: Changed header to GPL version 2 only extra/innochecksum.c: Changed header to GPL version 2 only extra/my_print_defaults.c: Changed header to GPL version 2 only extra/mysql_waitpid.c: Changed header to GPL version 2 only extra/perror.c: Changed header to GPL version 2 only extra/replace.c: Changed header to GPL version 2 only extra/resolve_stack_dump.c: Changed header to GPL version 2 only extra/resolveip.c: Changed header to GPL version 2 only heap/Makefile.am: Changed header to GPL version 2 only heap/_check.c: Changed header to GPL version 2 only heap/_rectest.c: Changed header to GPL version 2 only heap/heapdef.h: Changed header to GPL version 2 only heap/hp_block.c: Changed header to GPL version 2 only heap/hp_clear.c: Changed header to GPL version 2 only heap/hp_close.c: Changed header to GPL version 2 only heap/hp_create.c: Changed header to GPL version 2 only heap/hp_delete.c: Changed header to GPL version 2 only heap/hp_extra.c: Changed header to GPL version 2 only heap/hp_hash.c: Changed header to GPL version 2 only heap/hp_info.c: Changed header to GPL version 2 only heap/hp_open.c: Changed header to GPL version 2 only heap/hp_panic.c: Changed header to GPL version 2 only heap/hp_rename.c: Changed header to GPL version 2 only heap/hp_rfirst.c: Changed header to GPL version 2 only heap/hp_rkey.c: Changed header to GPL version 2 only heap/hp_rlast.c: Changed header to GPL version 2 only heap/hp_rnext.c: Changed header to GPL version 2 only heap/hp_rprev.c: Changed header to GPL version 2 only heap/hp_rrnd.c: Changed header to GPL version 2 only heap/hp_rsame.c: Changed header to GPL version 2 only heap/hp_scan.c: Changed header to GPL version 2 only heap/hp_static.c: Changed header to GPL version 2 only heap/hp_test1.c: Changed header to GPL version 2 only heap/hp_test2.c: Changed header to GPL version 2 only heap/hp_update.c: Changed header to GPL version 2 only heap/hp_write.c: Changed header to GPL version 2 only include/Makefile.am: Changed header to GPL version 2 only include/base64.h: Changed header to GPL version 2 only include/config-netware.h: Changed header to GPL version 2 only include/config-os2.h: Changed header to GPL version 2 only include/config-win.h: Changed header to GPL version 2 only include/decimal.h: Changed header to GPL version 2 only include/errmsg.h: Changed header to GPL version 2 only include/ft_global.h: Changed header to GPL version 2 only include/hash.h: Changed header to GPL version 2 only include/heap.h: Changed header to GPL version 2 only include/keycache.h: Changed header to GPL version 2 only include/m_ctype.h: Changed header to GPL version 2 only include/m_string.h: Changed header to GPL version 2 only include/md5.h: Changed header to GPL version 2 only include/my_aes.h: Changed header to GPL version 2 only include/my_alarm.h: Changed header to GPL version 2 only include/my_alloc.h: Changed header to GPL version 2 only include/my_base.h: Changed header to GPL version 2 only include/my_bitmap.h: Changed header to GPL version 2 only include/my_dbug.h: Changed header to GPL version 2 only include/my_dir.h: Changed header to GPL version 2 only include/my_getopt.h: Changed header to GPL version 2 only include/my_global.h: Changed header to GPL version 2 only include/my_handler.h: Changed header to GPL version 2 only include/my_libwrap.h: Changed header to GPL version 2 only include/my_list.h: Changed header to GPL version 2 only include/my_net.h: Changed header to GPL version 2 only include/my_no_pthread.h: Changed header to GPL version 2 only include/my_nosys.h: Changed header to GPL version 2 only include/my_pthread.h: Changed header to GPL version 2 only include/my_sys.h: Changed header to GPL version 2 only include/my_time.h: Changed header to GPL version 2 only include/my_tree.h: Changed header to GPL version 2 only include/my_user.h: Changed header to GPL version 2 only include/my_xml.h: Changed header to GPL version 2 only include/myisam.h: Changed header to GPL version 2 only include/myisammrg.h: Changed header to GPL version 2 only include/myisampack.h: Changed header to GPL version 2 only include/mysql.h: Changed header to GPL version 2 only include/mysql_com.h: Changed header to GPL version 2 only include/mysql_embed.h: Changed header to GPL version 2 only include/mysql_time.h: Changed header to GPL version 2 only include/mysys_err.h: Changed header to GPL version 2 only include/queues.h: Changed header to GPL version 2 only include/raid.h: Changed header to GPL version 2 only include/rijndael.h: Changed header to GPL version 2 only include/sha1.h: Changed header to GPL version 2 only include/sql_common.h: Changed header to GPL version 2 only include/sslopt-case.h: Changed header to GPL version 2 only include/sslopt-longopts.h: Changed header to GPL version 2 only include/sslopt-vars.h: Changed header to GPL version 2 only include/t_ctype.h: Changed header to GPL version 2 only include/thr_alarm.h: Changed header to GPL version 2 only include/thr_lock.h: Changed header to GPL version 2 only include/typelib.h: Changed header to GPL version 2 only include/violite.h: Changed header to GPL version 2 only innobase/Makefile.am: Changed header to GPL version 2 only innobase/btr/Makefile.am: Changed header to GPL version 2 only innobase/buf/Makefile.am: Changed header to GPL version 2 only innobase/data/Makefile.am: Changed header to GPL version 2 only innobase/dict/Makefile.am: Changed header to GPL version 2 only innobase/dyn/Makefile.am: Changed header to GPL version 2 only innobase/eval/Makefile.am: Changed header to GPL version 2 only innobase/fil/Makefile.am: Changed header to GPL version 2 only innobase/fsp/Makefile.am: Changed header to GPL version 2 only innobase/fut/Makefile.am: Changed header to GPL version 2 only innobase/ha/Makefile.am: Changed header to GPL version 2 only innobase/ibuf/Makefile.am: Changed header to GPL version 2 only innobase/include/Makefile.am: Changed header to GPL version 2 only innobase/lock/Makefile.am: Changed header to GPL version 2 only innobase/log/Makefile.am: Changed header to GPL version 2 only innobase/mach/Makefile.am: Changed header to GPL version 2 only innobase/mem/Makefile.am: Changed header to GPL version 2 only innobase/mtr/Makefile.am: Changed header to GPL version 2 only innobase/os/Makefile.am: Changed header to GPL version 2 only innobase/page/Makefile.am: Changed header to GPL version 2 only innobase/pars/Makefile.am: Changed header to GPL version 2 only innobase/que/Makefile.am: Changed header to GPL version 2 only innobase/read/Makefile.am: Changed header to GPL version 2 only innobase/rem/Makefile.am: Changed header to GPL version 2 only innobase/row/Makefile.am: Changed header to GPL version 2 only innobase/srv/Makefile.am: Changed header to GPL version 2 only innobase/sync/Makefile.am: Changed header to GPL version 2 only innobase/thr/Makefile.am: Changed header to GPL version 2 only innobase/trx/Makefile.am: Changed header to GPL version 2 only innobase/usr/Makefile.am: Changed header to GPL version 2 only innobase/ut/Makefile.am: Changed header to GPL version 2 only libmysql/client_settings.h: Changed header to GPL version 2 only libmysqld/Makefile.am: Changed header to GPL version 2 only libmysqld/emb_qcache.cc: Changed header to GPL version 2 only libmysqld/emb_qcache.h: Changed header to GPL version 2 only libmysqld/embedded_priv.h: Changed header to GPL version 2 only libmysqld/examples/Makefile.am: Changed header to GPL version 2 only libmysqld/libmysqld.c: Changed header to GPL version 2 only man/Makefile.am: Changed header to GPL version 2 only myisam/Makefile.am: Changed header to GPL version 2 only myisam/ft_boolean_search.c: Changed header to GPL version 2 only myisam/ft_eval.c: Changed header to GPL version 2 only myisam/ft_eval.h: Changed header to GPL version 2 only myisam/ft_nlq_search.c: Changed header to GPL version 2 only myisam/ft_parser.c: Changed header to GPL version 2 only myisam/ft_static.c: Changed header to GPL version 2 only myisam/ft_stem.c: Changed header to GPL version 2 only myisam/ft_stopwords.c: Changed header to GPL version 2 only myisam/ft_test1.c: Changed header to GPL version 2 only myisam/ft_test1.h: Changed header to GPL version 2 only myisam/ft_update.c: Changed header to GPL version 2 only myisam/ftdefs.h: Changed header to GPL version 2 only myisam/fulltext.h: Changed header to GPL version 2 only myisam/mi_cache.c: Changed header to GPL version 2 only myisam/mi_changed.c: Changed header to GPL version 2 only myisam/mi_check.c: Changed header to GPL version 2 only myisam/mi_checksum.c: Changed header to GPL version 2 only myisam/mi_close.c: Changed header to GPL version 2 only myisam/mi_create.c: Changed header to GPL version 2 only myisam/mi_dbug.c: Changed header to GPL version 2 only myisam/mi_delete.c: Changed header to GPL version 2 only myisam/mi_delete_all.c: Changed header to GPL version 2 only myisam/mi_delete_table.c: Changed header to GPL version 2 only myisam/mi_dynrec.c: Changed header to GPL version 2 only myisam/mi_extra.c: Changed header to GPL version 2 only myisam/mi_info.c: Changed header to GPL version 2 only myisam/mi_key.c: Changed header to GPL version 2 only myisam/mi_keycache.c: Changed header to GPL version 2 only myisam/mi_locking.c: Changed header to GPL version 2 only myisam/mi_log.c: Changed header to GPL version 2 only myisam/mi_open.c: Changed header to GPL version 2 only myisam/mi_packrec.c: Changed header to GPL version 2 only myisam/mi_page.c: Changed header to GPL version 2 only myisam/mi_panic.c: Changed header to GPL version 2 only myisam/mi_preload.c: Changed header to GPL version 2 only myisam/mi_range.c: Changed header to GPL version 2 only myisam/mi_rename.c: Changed header to GPL version 2 only myisam/mi_rfirst.c: Changed header to GPL version 2 only myisam/mi_rkey.c: Changed header to GPL version 2 only myisam/mi_rlast.c: Changed header to GPL version 2 only myisam/mi_rnext.c: Changed header to GPL version 2 only myisam/mi_rnext_same.c: Changed header to GPL version 2 only myisam/mi_rprev.c: Changed header to GPL version 2 only myisam/mi_rrnd.c: Changed header to GPL version 2 only myisam/mi_rsame.c: Changed header to GPL version 2 only myisam/mi_rsamepos.c: Changed header to GPL version 2 only myisam/mi_scan.c: Changed header to GPL version 2 only myisam/mi_search.c: Changed header to GPL version 2 only myisam/mi_static.c: Changed header to GPL version 2 only myisam/mi_statrec.c: Changed header to GPL version 2 only myisam/mi_test1.c: Changed header to GPL version 2 only myisam/mi_test2.c: Changed header to GPL version 2 only myisam/mi_test3.c: Changed header to GPL version 2 only myisam/mi_unique.c: Changed header to GPL version 2 only myisam/mi_update.c: Changed header to GPL version 2 only myisam/mi_write.c: Changed header to GPL version 2 only myisam/myisam_ftdump.c: Changed header to GPL version 2 only myisam/myisamchk.c: Changed header to GPL version 2 only myisam/myisamdef.h: Changed header to GPL version 2 only myisam/myisamlog.c: Changed header to GPL version 2 only myisam/myisampack.c: Changed header to GPL version 2 only myisam/rt_index.c: Changed header to GPL version 2 only myisam/rt_index.h: Changed header to GPL version 2 only myisam/rt_key.c: Changed header to GPL version 2 only myisam/rt_key.h: Changed header to GPL version 2 only myisam/rt_mbr.c: Changed header to GPL version 2 only myisam/rt_mbr.h: Changed header to GPL version 2 only myisam/rt_split.c: Changed header to GPL version 2 only myisam/rt_test.c: Changed header to GPL version 2 only myisam/sort.c: Changed header to GPL version 2 only myisam/sp_defs.h: Changed header to GPL version 2 only myisam/sp_key.c: Changed header to GPL version 2 only myisam/sp_test.c: Changed header to GPL version 2 only myisammrg/Makefile.am: Changed header to GPL version 2 only myisammrg/myrg_close.c: Changed header to GPL version 2 only myisammrg/myrg_create.c: Changed header to GPL version 2 only myisammrg/myrg_def.h: Changed header to GPL version 2 only myisammrg/myrg_delete.c: Changed header to GPL version 2 only myisammrg/myrg_extra.c: Changed header to GPL version 2 only myisammrg/myrg_info.c: Changed header to GPL version 2 only myisammrg/myrg_locking.c: Changed header to GPL version 2 only myisammrg/myrg_open.c: Changed header to GPL version 2 only myisammrg/myrg_panic.c: Changed header to GPL version 2 only myisammrg/myrg_queue.c: Changed header to GPL version 2 only myisammrg/myrg_range.c: Changed header to GPL version 2 only myisammrg/myrg_rfirst.c: Changed header to GPL version 2 only myisammrg/myrg_rkey.c: Changed header to GPL version 2 only myisammrg/myrg_rlast.c: Changed header to GPL version 2 only myisammrg/myrg_rnext.c: Changed header to GPL version 2 only myisammrg/myrg_rnext_same.c: Changed header to GPL version 2 only myisammrg/myrg_rprev.c: Changed header to GPL version 2 only myisammrg/myrg_rrnd.c: Changed header to GPL version 2 only myisammrg/myrg_rsame.c: Changed header to GPL version 2 only myisammrg/myrg_static.c: Changed header to GPL version 2 only myisammrg/myrg_update.c: Changed header to GPL version 2 only myisammrg/myrg_write.c: Changed header to GPL version 2 only mysql-test/Makefile.am: Changed header to GPL version 2 only mysys/Makefile.am: Changed header to GPL version 2 only mysys/array.c: Changed header to GPL version 2 only mysys/base64.c: Changed header to GPL version 2 only mysys/charset-def.c: Changed header to GPL version 2 only mysys/charset.c: Changed header to GPL version 2 only mysys/checksum.c: Changed header to GPL version 2 only mysys/default.c: Changed header to GPL version 2 only mysys/default_modify.c: Changed header to GPL version 2 only mysys/errors.c: Changed header to GPL version 2 only mysys/hash.c: Changed header to GPL version 2 only mysys/list.c: Changed header to GPL version 2 only mysys/make-conf.c: Changed header to GPL version 2 only mysys/md5.c: Changed header to GPL version 2 only mysys/mf_brkhant.c: Changed header to GPL version 2 only mysys/mf_cache.c: Changed header to GPL version 2 only mysys/mf_dirname.c: Changed header to GPL version 2 only mysys/mf_fn_ext.c: Changed header to GPL version 2 only mysys/mf_format.c: Changed header to GPL version 2 only mysys/mf_getdate.c: Changed header to GPL version 2 only mysys/mf_iocache.c: Changed header to GPL version 2 only mysys/mf_iocache2.c: Changed header to GPL version 2 only mysys/mf_keycache.c: Changed header to GPL version 2 only mysys/mf_keycaches.c: Changed header to GPL version 2 only mysys/mf_loadpath.c: Changed header to GPL version 2 only mysys/mf_pack.c: Changed header to GPL version 2 only mysys/mf_path.c: Changed header to GPL version 2 only mysys/mf_qsort.c: Changed header to GPL version 2 only mysys/mf_qsort2.c: Changed header to GPL version 2 only mysys/mf_radix.c: Changed header to GPL version 2 only mysys/mf_same.c: Changed header to GPL version 2 only mysys/mf_sort.c: Changed header to GPL version 2 only mysys/mf_soundex.c: Changed header to GPL version 2 only mysys/mf_strip.c: Changed header to GPL version 2 only mysys/mf_tempdir.c: Changed header to GPL version 2 only mysys/mf_tempfile.c: Changed header to GPL version 2 only mysys/mf_unixpath.c: Changed header to GPL version 2 only mysys/mf_util.c: Changed header to GPL version 2 only mysys/mf_wcomp.c: Changed header to GPL version 2 only mysys/mf_wfile.c: Changed header to GPL version 2 only mysys/mulalloc.c: Changed header to GPL version 2 only mysys/my_access.c: Changed header to GPL version 2 only mysys/my_aes.c: Changed header to GPL version 2 only mysys/my_alarm.c: Changed header to GPL version 2 only mysys/my_alloc.c: Changed header to GPL version 2 only mysys/my_append.c: Changed header to GPL version 2 only mysys/my_bit.c: Changed header to GPL version 2 only mysys/my_bitmap.c: Changed header to GPL version 2 only mysys/my_chsize.c: Changed header to GPL version 2 only mysys/my_clock.c: Changed header to GPL version 2 only mysys/my_compress.c: Changed header to GPL version 2 only mysys/my_conio.c: Changed header to GPL version 2 only mysys/my_copy.c: Changed header to GPL version 2 only mysys/my_crc32.c: Changed header to GPL version 2 only mysys/my_create.c: Changed header to GPL version 2 only mysys/my_delete.c: Changed header to GPL version 2 only mysys/my_div.c: Changed header to GPL version 2 only mysys/my_dup.c: Changed header to GPL version 2 only mysys/my_error.c: Changed header to GPL version 2 only mysys/my_file.c: Changed header to GPL version 2 only mysys/my_fopen.c: Changed header to GPL version 2 only mysys/my_fstream.c: Changed header to GPL version 2 only mysys/my_gethostbyname.c: Changed header to GPL version 2 only mysys/my_gethwaddr.c: Changed header to GPL version 2 only mysys/my_getopt.c: Changed header to GPL version 2 only mysys/my_getpagesize.c: Changed header to GPL version 2 only mysys/my_getsystime.c: Changed header to GPL version 2 only mysys/my_getwd.c: Changed header to GPL version 2 only mysys/my_handler.c: Changed header to GPL version 2 only mysys/my_init.c: Changed header to GPL version 2 only mysys/my_largepage.c: Changed header to GPL version 2 only mysys/my_lib.c: Changed header to GPL version 2 only mysys/my_libwrap.c: Changed header to GPL version 2 only mysys/my_lock.c: Changed header to GPL version 2 only mysys/my_lockmem.c: Changed header to GPL version 2 only mysys/my_lread.c: Changed header to GPL version 2 only mysys/my_lwrite.c: Changed header to GPL version 2 only mysys/my_malloc.c: Changed header to GPL version 2 only mysys/my_messnc.c: Changed header to GPL version 2 only mysys/my_mkdir.c: Changed header to GPL version 2 only mysys/my_mmap.c: Changed header to GPL version 2 only mysys/my_net.c: Changed header to GPL version 2 only mysys/my_netware.c: Changed header to GPL version 2 only mysys/my_new.cc: Changed header to GPL version 2 only mysys/my_once.c: Changed header to GPL version 2 only mysys/my_open.c: Changed header to GPL version 2 only mysys/my_os2cond.c: Changed header to GPL version 2 only mysys/my_os2dirsrch.c: Changed header to GPL version 2 only mysys/my_os2dirsrch.h: Changed header to GPL version 2 only mysys/my_os2dlfcn.c: Changed header to GPL version 2 only mysys/my_os2dlfcn.h0: Changed header to GPL version 2 only mysys/my_os2file64.c: Changed header to GPL version 2 only mysys/my_os2thread.c: Changed header to GPL version 2 only mysys/my_os2tls.c: Changed header to GPL version 2 only mysys/my_port.c: Changed header to GPL version 2 only mysys/my_pread.c: Changed header to GPL version 2 only mysys/my_pthread.c: Changed header to GPL version 2 only mysys/my_quick.c: Changed header to GPL version 2 only mysys/my_read.c: Changed header to GPL version 2 only mysys/my_realloc.c: Changed header to GPL version 2 only mysys/my_redel.c: Changed header to GPL version 2 only mysys/my_rename.c: Changed header to GPL version 2 only mysys/my_seek.c: Changed header to GPL version 2 only mysys/my_semaphore.c: Changed header to GPL version 2 only mysys/my_sleep.c: Changed header to GPL version 2 only mysys/my_static.c: Changed header to GPL version 2 only mysys/my_static.h: Changed header to GPL version 2 only mysys/my_symlink.c: Changed header to GPL version 2 only mysys/my_symlink2.c: Changed header to GPL version 2 only mysys/my_sync.c: Changed header to GPL version 2 only mysys/my_thr_init.c: Changed header to GPL version 2 only mysys/my_wincond.c: Changed header to GPL version 2 only mysys/my_windac.c: Changed header to GPL version 2 only mysys/my_winthread.c: Changed header to GPL version 2 only mysys/my_write.c: Changed header to GPL version 2 only mysys/mysys_priv.h: Changed header to GPL version 2 only mysys/ptr_cmp.c: Changed header to GPL version 2 only mysys/queues.c: Changed header to GPL version 2 only mysys/raid.cc: Changed header to GPL version 2 only mysys/raid2.c: Changed header to GPL version 2 only mysys/rijndael.c: Changed header to GPL version 2 only mysys/safemalloc.c: Changed header to GPL version 2 only mysys/sha1.c: Changed header to GPL version 2 only mysys/string.c: Changed header to GPL version 2 only mysys/test_charset.c: Changed header to GPL version 2 only mysys/test_dir.c: Changed header to GPL version 2 only mysys/test_fn.c: Changed header to GPL version 2 only mysys/test_xml.c: Changed header to GPL version 2 only mysys/testhash.c: Changed header to GPL version 2 only mysys/thr_alarm.c: Changed header to GPL version 2 only mysys/thr_lock.c: Changed header to GPL version 2 only mysys/thr_mutex.c: Changed header to GPL version 2 only mysys/thr_rwlock.c: Changed header to GPL version 2 only mysys/tree.c: Changed header to GPL version 2 only mysys/typelib.c: Changed header to GPL version 2 only ndb/include/debugger/DebuggerNames.hpp: Changed header to GPL version 2 only ndb/include/debugger/EventLogger.hpp: Changed header to GPL version 2 only ndb/include/debugger/GrepError.hpp: Changed header to GPL version 2 only ndb/include/debugger/SignalLoggerManager.hpp: Changed header to GPL version 2 only ndb/include/editline/editline.h: Changed header to GPL version 2 only ndb/include/kernel/AttributeDescriptor.hpp: Changed header to GPL version 2 only ndb/include/kernel/AttributeHeader.hpp: Changed header to GPL version 2 only ndb/include/kernel/AttributeList.hpp: Changed header to GPL version 2 only ndb/include/kernel/BlockNumbers.h: Changed header to GPL version 2 only ndb/include/kernel/GlobalSignalNumbers.h: Changed header to GPL version 2 only ndb/include/kernel/GrepEvent.hpp: Changed header to GPL version 2 only ndb/include/kernel/Interpreter.hpp: Changed header to GPL version 2 only ndb/include/kernel/LogLevel.hpp: Changed header to GPL version 2 only ndb/include/kernel/NodeBitmask.hpp: Changed header to GPL version 2 only ndb/include/kernel/NodeInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/NodeState.hpp: Changed header to GPL version 2 only ndb/include/kernel/RefConvert.hpp: Changed header to GPL version 2 only ndb/include/kernel/kernel_types.h: Changed header to GPL version 2 only ndb/include/kernel/ndb_limits.h: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AbortAll.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AccFrag.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AccLock.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AccScan.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AccSizeAltReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AlterIndx.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AlterTab.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AlterTable.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AlterTrig.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ApiBroadcast.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ApiRegSignalData.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ApiVersion.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ArbitSignalData.hpp: Changed header to GPL version 2 only ndb/include/kernel/trigger_definitions.h: Changed header to GPL version 2 only ndb/include/ndb_constants.h: Changed header to GPL version 2 only ndb/include/ndb_global.h.in: Changed header to GPL version 2 only ndb/include/ndb_init.h: Changed header to GPL version 2 only ndb/include/ndb_types.h.in: Changed header to GPL version 2 only ndb/include/ndb_version.h.in: Changed header to GPL version 2 only ndb/include/kernel/signaldata/AttrInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/BackupContinueB.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/BackupImpl.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/BackupSignalData.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/BlockCommitOrd.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/BuildIndx.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CheckNodeGroups.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CloseComReqConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CmInit.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CmRegSignalData.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CmvmiCfgConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CntrMasterConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CntrMasterReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ConfigParamId.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ContinueFragmented.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CopyActive.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CopyFrag.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CopyGCIReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CreateEvnt.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CreateFrag.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CreateFragmentation.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CreateIndx.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CreateTab.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CreateTable.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/CreateTrig.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DiAddTab.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DiGetNodes.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DictLock.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DictSchemaInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DictSizeAltReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DictStart.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DictTabInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DihAddFrag.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DihContinueB.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DihSizeAltReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DihStartTab.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DihSwitchReplica.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DisconnectRep.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DropIndx.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DropTab.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DropTabFile.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DropTable.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DropTrig.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/DumpStateOrd.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/EmptyLcp.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/EndTo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/EventReport.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/EventSubscribeReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ExecFragReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FailRep.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FireTrigOrd.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FsAppendReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FsCloseReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FsConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FsOpenReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FsReadWriteReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FsRef.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/FsRemoveReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/GCPSave.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/GetTabInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/GetTableId.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/GrepImpl.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/HotSpareRep.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/IndxAttrInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/IndxKeyInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/InvalidateNodeLCPConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/InvalidateNodeLCPReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/KeyInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/LCP.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ListTables.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/LqhFrag.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/LqhKey.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/LqhSizeAltReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/LqhTransConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ManagementServer.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/MasterGCP.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/MasterLCP.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/NFCompleteRep.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/NdbSttor.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/NdbfsContinueB.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/NextScan.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/NodeFailRep.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/NodeStateSignalData.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/PackedSignal.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/PrepDropTab.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/PrepFailReqRef.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ReadNodesConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/RelTabMem.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/RepImpl.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ResumeReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ScanFrag.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/ScanTab.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/SetLogLevelOrd.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/SetVarReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/SignalData.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/SignalDataPrint.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/SignalDroppedRep.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/SrFragidConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StartFragReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StartInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StartMe.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StartOrd.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StartPerm.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StartRec.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StartTo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StopMe.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StopPerm.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/StopReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/SumaImpl.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/SystemError.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TamperOrd.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcCommit.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcContinueB.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcHbRep.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcIndx.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcKeyConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcKeyFailConf.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcKeyRef.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcKeyReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcRollbackRep.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TcSizeAltReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TestOrd.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TransIdAI.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TrigAttrInfo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TupCommit.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TupFrag.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TupKey.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TupSizeAltReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TuxBound.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TuxContinueB.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TuxMaint.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/TuxSizeAltReq.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/UpdateTo.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/UtilDelete.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/UtilExecute.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/UtilLock.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/UtilPrepare.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/UtilRelease.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/UtilSequence.hpp: Changed header to GPL version 2 only ndb/include/kernel/signaldata/WaitGCP.hpp: Changed header to GPL version 2 only ndb/include/logger/ConsoleLogHandler.hpp: Changed header to GPL version 2 only ndb/include/logger/FileLogHandler.hpp: Changed header to GPL version 2 only ndb/include/logger/LogHandler.hpp: Changed header to GPL version 2 only ndb/include/logger/Logger.hpp: Changed header to GPL version 2 only ndb/include/logger/SysLogHandler.hpp: Changed header to GPL version 2 only ndb/include/mgmapi/mgmapi.h: Changed header to GPL version 2 only ndb/include/mgmapi/mgmapi_debug.h: Changed header to GPL version 2 only ndb/include/mgmapi/ndb_logevent.h: Changed header to GPL version 2 only ndb/include/mgmapi/ndbd_exit_codes.h: Changed header to GPL version 2 only ndb/include/mgmcommon/ConfigRetriever.hpp: Changed header to GPL version 2 only ndb/include/mgmcommon/IPCConfig.hpp: Changed header to GPL version 2 only ndb/include/mgmcommon/MgmtErrorReporter.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/Ndb.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbApi.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbBlob.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbDictionary.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbError.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbEventOperation.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbIndexOperation.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbIndexScanOperation.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbOperation.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbPool.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbRecAttr.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbReceiver.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbScanFilter.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbScanOperation.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/NdbTransaction.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/ndb_cluster_connection.hpp: Changed header to GPL version 2 only ndb/include/ndbapi/ndb_opt_defaults.h: Changed header to GPL version 2 only ndb/include/ndbapi/ndbapi_limits.h: Changed header to GPL version 2 only ndb/include/ndbapi/ndberror.h: Changed header to GPL version 2 only ndb/include/newtonapi/dba.h: Changed header to GPL version 2 only ndb/include/newtonapi/defs/pcn_types.h: Changed header to GPL version 2 only ndb/include/portlib/NdbCondition.h: Changed header to GPL version 2 only ndb/include/portlib/NdbConfig.h: Changed header to GPL version 2 only ndb/include/portlib/NdbDaemon.h: Changed header to GPL version 2 only ndb/include/portlib/NdbEnv.h: Changed header to GPL version 2 only ndb/include/portlib/NdbHost.h: Changed header to GPL version 2 only ndb/include/portlib/NdbMain.h: Changed header to GPL version 2 only ndb/include/portlib/NdbMem.h: Changed header to GPL version 2 only ndb/include/portlib/NdbMutex.h: Changed header to GPL version 2 only ndb/include/portlib/NdbSleep.h: Changed header to GPL version 2 only ndb/include/portlib/NdbTCP.h: Changed header to GPL version 2 only ndb/include/portlib/NdbThread.h: Changed header to GPL version 2 only ndb/include/portlib/NdbTick.h: Changed header to GPL version 2 only ndb/include/portlib/PortDefs.h: Changed header to GPL version 2 only ndb/include/portlib/prefetch.h: Changed header to GPL version 2 only ndb/include/transporter/TransporterCallback.hpp: Changed header to GPL version 2 only ndb/include/transporter/TransporterDefinitions.hpp: Changed header to GPL version 2 only ndb/include/transporter/TransporterRegistry.hpp: Changed header to GPL version 2 only ndb/include/util/BaseString.hpp: Changed header to GPL version 2 only ndb/include/util/Bitmask.hpp: Changed header to GPL version 2 only ndb/include/util/File.hpp: Changed header to GPL version 2 only ndb/include/util/InputStream.hpp: Changed header to GPL version 2 only ndb/include/util/NdbAutoPtr.hpp: Changed header to GPL version 2 only ndb/include/util/NdbOut.hpp: Changed header to GPL version 2 only ndb/include/util/NdbSqlUtil.hpp: Changed header to GPL version 2 only ndb/include/util/OutputStream.hpp: Changed header to GPL version 2 only ndb/include/util/Parser.hpp: Changed header to GPL version 2 only ndb/include/util/Properties.hpp: Changed header to GPL version 2 only ndb/include/util/SimpleProperties.hpp: Changed header to GPL version 2 only ndb/include/util/SocketAuthenticator.hpp: Changed header to GPL version 2 only ndb/include/util/SocketClient.hpp: Changed header to GPL version 2 only ndb/include/util/SocketServer.hpp: Changed header to GPL version 2 only ndb/include/util/UtilBuffer.hpp: Changed header to GPL version 2 only ndb/include/util/Vector.hpp: Changed header to GPL version 2 only ndb/include/util/basestring_vsnprintf.h: Changed header to GPL version 2 only ndb/include/util/md5_hash.hpp: Changed header to GPL version 2 only ndb/include/util/ndb_opts.h: Changed header to GPL version 2 only ndb/include/util/random.h: Changed header to GPL version 2 only ndb/include/util/socket_io.h: Changed header to GPL version 2 only ndb/include/util/uucode.h: Changed header to GPL version 2 only ndb/include/util/version.h: Changed header to GPL version 2 only ndb/ndbapi-examples/mgmapi_logevent_example/mgmapi_logevent.cpp: Changed header to GPL version 2 only ndb/ndbapi-examples/ndbapi_async_example/ndbapi_async.cpp: Changed header to GPL version 2 only ndb/ndbapi-examples/ndbapi_async_example1/ndbapi_async1.cpp: Changed header to GPL version 2 only ndb/ndbapi-examples/ndbapi_event_example/ndbapi_event.cpp: Changed header to GPL version 2 only ndb/ndbapi-examples/ndbapi_retries_example/ndbapi_retries.cpp: Changed header to GPL version 2 only ndb/ndbapi-examples/ndbapi_scan_example/ndbapi_scan.cpp: Changed header to GPL version 2 only ndb/ndbapi-examples/ndbapi_simple_example/ndbapi_simple.cpp: Changed header to GPL version 2 only ndb/ndbapi-examples/ndbapi_simple_index_example/ndbapi_simple_index.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/BlockNames.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/DebuggerNames.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/EventLogger.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/GrepError.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/SignalLoggerManager.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/AccLock.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/AlterIndx.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/AlterTab.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/AlterTable.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/AlterTrig.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/BackupImpl.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/BackupSignalData.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/CloseComReqConf.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/ContinueB.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/CopyGCI.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/CreateEvnt.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/CreateFragmentation.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/CreateIndx.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/CreateTrig.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/DictTabInfo.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/DihContinueB.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/DihSwitchReplicaReq.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/DisconnectRep.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/DropIndx.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/DropTab.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/DropTrig.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/FailRep.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/FireTrigOrd.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/FsAppendReq.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/FsCloseReq.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/FsConf.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/FsOpenReq.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/FsReadWriteReq.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/FsRef.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/GCPSave.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/IndxAttrInfo.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/IndxKeyInfo.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/LCP.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/LqhFrag.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/LqhKey.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/LqhTrans.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/MasterLCP.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/NFCompleteRep.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/NdbSttor.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/NdbfsContinueB.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/PackedSignal.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/PrepDropTab.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/PrepFailReqRef.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/ScanFrag.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/ScanTab.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/SignalDataPrint.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/SignalDroppedRep.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/SignalNames.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/StartRec.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/SumaImpl.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/SystemError.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TcIndx.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TcKeyConf.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TcKeyRef.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TcKeyReq.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TcRollbackRep.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TrigAttrInfo.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TupCommit.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TupKey.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/TuxMaint.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/UtilDelete.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/UtilExecute.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/UtilLock.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/UtilPrepare.cpp: Changed header to GPL version 2 only ndb/src/common/debugger/signaldata/UtilSequence.cpp: Changed header to GPL version 2 only ndb/src/common/logger/ConsoleLogHandler.cpp: Changed header to GPL version 2 only ndb/src/common/logger/FileLogHandler.cpp: Changed header to GPL version 2 only ndb/src/common/logger/LogHandler.cpp: Changed header to GPL version 2 only ndb/src/common/logger/LogHandlerList.cpp: Changed header to GPL version 2 only ndb/src/common/logger/LogHandlerList.hpp: Changed header to GPL version 2 only ndb/src/common/logger/Logger.cpp: Changed header to GPL version 2 only ndb/src/common/logger/SysLogHandler.cpp: Changed header to GPL version 2 only ndb/src/common/logger/listtest/LogHandlerListUnitTest.cpp: Changed header to GPL version 2 only ndb/src/common/logger/listtest/LogHandlerListUnitTest.hpp: Changed header to GPL version 2 only ndb/src/common/logger/loggertest/LoggerUnitTest.cpp: Changed header to GPL version 2 only ndb/src/common/logger/loggertest/LoggerUnitTest.hpp: Changed header to GPL version 2 only ndb/src/common/mgmcommon/ConfigRetriever.cpp: Changed header to GPL version 2 only ndb/src/common/mgmcommon/IPCConfig.cpp: Changed header to GPL version 2 only ndb/src/common/mgmcommon/printConfig/printConfig.cpp: Changed header to GPL version 2 only ndb/src/common/portlib/NdbCondition.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbConfig.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbDaemon.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbEnv.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbHost.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbMem.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbMutex.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbPortLibTest.cpp: Changed header to GPL version 2 only ndb/src/common/portlib/NdbSleep.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbTCP.cpp: Changed header to GPL version 2 only ndb/src/common/portlib/NdbThread.c: Changed header to GPL version 2 only ndb/src/common/portlib/NdbTick.c: Changed header to GPL version 2 only ndb/src/common/portlib/memtest.c: Changed header to GPL version 2 only ndb/src/common/portlib/mmslist.cpp: Changed header to GPL version 2 only ndb/src/common/portlib/mmstest.cpp: Changed header to GPL version 2 only ndb/src/common/portlib/munmaptest.cpp: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbCondition.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbConditionOSE.h: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbEnv.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbHost.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbMem.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbMem_SoftOse.cpp: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbMutex.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbOut.cpp: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbSleep.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbTCP.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbThread.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/ose/NdbTick.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbCondition.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbDaemon.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbEnv.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbHost.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbMem.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbMutex.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbSleep.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbTCP.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbThread.c: Changed header to GPL version 2 only ndb/src/common/portlib/old_dirs/win32/NdbTick.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbCondition.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbDaemon.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbEnv.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbHost.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbMem.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbMutex.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbSleep.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbTCP.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbThread.c: Changed header to GPL version 2 only ndb/src/common/portlib/win32/NdbTick.c: Changed header to GPL version 2 only ndb/src/common/transporter/OSE_Receiver.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/OSE_Receiver.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/OSE_Signals.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/OSE_Transporter.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/OSE_Transporter.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/Packer.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/Packer.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/SCI_Transporter.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/SCI_Transporter.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/SHM_Buffer.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/SHM_Transporter.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/SHM_Transporter.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/SHM_Transporter.unix.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/SHM_Transporter.win32.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/SendBuffer.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/SendBuffer.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/TCP_Transporter.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/TCP_Transporter.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/Transporter.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/Transporter.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/TransporterInternalDefinitions.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/TransporterRegistry.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/basictest/basicTransporterTest.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/buddy.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/buddy.hpp: Changed header to GPL version 2 only ndb/src/common/transporter/failoverSCI/failoverSCI.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/perftest/perfTransporterTest.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/priotest/prioSCI/prioSCI.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/priotest/prioSHM/prioSHM.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/priotest/prioTCP/prioTCP.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/priotest/prioTransporterTest.cpp: Changed header to GPL version 2 only ndb/src/common/transporter/priotest/prioTransporterTest.hpp: Changed header to GPL version 2 only ndb/src/common/util/BaseString.cpp: Changed header to GPL version 2 only ndb/src/common/util/File.cpp: Changed header to GPL version 2 only ndb/src/common/util/InputStream.cpp: Changed header to GPL version 2 only ndb/src/common/util/NdbErrHnd.cpp: Changed header to GPL version 2 only ndb/src/common/util/NdbOut.cpp: Changed header to GPL version 2 only ndb/src/common/util/NdbSqlUtil.cpp: Changed header to GPL version 2 only ndb/src/common/util/OutputStream.cpp: Changed header to GPL version 2 only ndb/src/common/util/Parser.cpp: Changed header to GPL version 2 only ndb/src/common/util/Properties.cpp: Changed header to GPL version 2 only ndb/src/common/util/SimpleProperties.cpp: Changed header to GPL version 2 only ndb/src/common/util/SocketAuthenticator.cpp: Changed header to GPL version 2 only ndb/src/common/util/SocketClient.cpp: Changed header to GPL version 2 only ndb/src/common/util/SocketServer.cpp: Changed header to GPL version 2 only ndb/src/common/util/basestring_vsnprintf.c: Changed header to GPL version 2 only ndb/src/common/util/filetest/FileUnitTest.cpp: Changed header to GPL version 2 only ndb/src/common/util/filetest/FileUnitTest.hpp: Changed header to GPL version 2 only ndb/src/common/util/md5_hash.cpp: Changed header to GPL version 2 only ndb/src/common/util/ndb_init.c: Changed header to GPL version 2 only ndb/src/common/util/random.c: Changed header to GPL version 2 only ndb/src/common/util/socket_io.cpp: Changed header to GPL version 2 only ndb/src/common/util/strdup.c: Changed header to GPL version 2 only ndb/src/common/util/testProperties/testProperties.cpp: Changed header to GPL version 2 only ndb/src/common/util/testSimpleProperties/sp_test.cpp: Changed header to GPL version 2 only ndb/src/common/util/uucode.c: Changed header to GPL version 2 only ndb/src/common/util/version.c: Changed header to GPL version 2 only ndb/src/cw/cpcc-win32/C++/CPC_GUI.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcc-win32/C++/CPC_GUI.h: Changed header to GPL version 2 only ndb/src/cw/cpcc-win32/C++/NdbControls.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcc-win32/C++/StdAfx.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcc-win32/C++/StdAfx.h: Changed header to GPL version 2 only ndb/src/cw/cpcc-win32/C++/TreeView.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcc-win32/C++/TreeView.h: Changed header to GPL version 2 only ndb/src/cw/cpcc-win32/C++/resource.h: Changed header to GPL version 2 only ndb/src/cw/cpcd/APIService.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcd/APIService.hpp: Changed header to GPL version 2 only ndb/src/cw/cpcd/CPCD.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcd/CPCD.hpp: Changed header to GPL version 2 only ndb/src/cw/cpcd/Monitor.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcd/Process.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcd/common.cpp: Changed header to GPL version 2 only ndb/src/cw/cpcd/common.hpp: Changed header to GPL version 2 only ndb/src/cw/cpcd/main.cpp: Changed header to GPL version 2 only ndb/src/cw/test/socketclient/socketClientTest.cpp: Changed header to GPL version 2 only ndb/src/cw/util/ClientInterface.cpp: Changed header to GPL version 2 only ndb/src/cw/util/ClientInterface.hpp: Changed header to GPL version 2 only ndb/src/cw/util/SocketRegistry.cpp: Changed header to GPL version 2 only ndb/src/cw/util/SocketRegistry.hpp: Changed header to GPL version 2 only ndb/src/cw/util/SocketService.cpp: Changed header to GPL version 2 only ndb/src/cw/util/SocketService.hpp: Changed header to GPL version 2 only ndb/src/kernel/SimBlockList.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/backup/Backup.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/backup/Backup.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/backup/BackupFormat.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/backup/BackupInit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/backup/FsBuffer.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/backup/read.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/cmvmi/Cmvmi.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/cmvmi/Cmvmi.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbacc/Dbacc.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbacc/DbaccInit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbacc/DbaccMain.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/mutexes.hpp: Changed header to GPL version 2 only ndb/src/kernel/main.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdict/Dbdict.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdict/Dbdict.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdict/SchemaFile.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdict/printSchemaFile.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdih/Dbdih.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdih/DbdihInit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdih/DbdihMain.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdih/Sysfile.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbdih/printSysfile/printSysfile.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dblqh/Dblqh.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dblqh/DblqhInit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dblqh/DblqhMain.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dblqh/redoLogReader/records.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dblqh/redoLogReader/records.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dblqh/redoLogReader/redoLogFileReader.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtc/Dbtc.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtc/DbtcInit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtc/DbtcMain.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/AttributeOffset.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/Dbtup.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupAbort.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupBuffer.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupCommit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupDebug.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupExecQuery.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupFixAlloc.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupGen.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupIndex.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupLCP.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupMeta.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupPagMan.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupPageMap.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupRoutines.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupScan.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupStoredProcDef.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupSystemRestart.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupTabDesMan.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupTrigger.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtup/DbtupUndoLog.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/Dbtux.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxCmp.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxDebug.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxGen.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxMaint.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxMeta.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxNode.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxScan.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxSearch.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbtux/DbtuxTree.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbutil/DbUtil.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/dbutil/DbUtil.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbcntr/Ndbcntr.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbcntr/NdbcntrInit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbcntr/NdbcntrMain.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbcntr/NdbcntrSysTable.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/AsyncFile.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/AsyncFile.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/AsyncFileTest/AsyncFileTest.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/CircularIndex.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/CircularIndex.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/Filename.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/Filename.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/MemoryChannel.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/MemoryChannel.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/MemoryChannelOSE.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/MemoryChannelTest/MemoryChannelTest.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/Ndbfs.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/Ndbfs.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/OpenFiles.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/Pool.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/ndbfs/VoidFs.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/qmgr/Qmgr.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/qmgr/QmgrInit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/qmgr/QmgrMain.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/qmgr/timer.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/suma/Suma.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/suma/Suma.hpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/suma/SumaInit.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/trix/Trix.cpp: Changed header to GPL version 2 only ndb/src/kernel/blocks/trix/Trix.hpp: Changed header to GPL version 2 only ndb/src/kernel/error/ErrorHandlingMacros.hpp: Changed header to GPL version 2 only ndb/src/kernel/error/ErrorReporter.cpp: Changed header to GPL version 2 only ndb/src/kernel/error/ErrorReporter.hpp: Changed header to GPL version 2 only ndb/src/kernel/error/TimeModule.cpp: Changed header to GPL version 2 only ndb/src/kernel/error/TimeModule.hpp: Changed header to GPL version 2 only ndb/src/kernel/error/ndbd_exit_codes.c: Changed header to GPL version 2 only ndb/src/kernel/vm/Array.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ArrayFifoList.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ArrayList.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ArrayPool.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/CArray.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/Callback.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ClusterConfiguration.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ClusterConfiguration.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/Configuration.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/Configuration.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/DLFifoList.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/DLHashTable.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/DLHashTable2.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/DLList.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/DataBuffer.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/Emulator.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/Emulator.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/FastScheduler.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/FastScheduler.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/GlobalData.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/KeyDescriptor.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/KeyTable.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/KeyTable2.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/LongSignal.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/MetaData.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/MetaData.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/Mutex.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/Mutex.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/Prio.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/RequestTracker.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SLList.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SafeCounter.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SafeCounter.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SectionReader.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SectionReader.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SignalCounter.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SimBlockList.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SimplePropertiesSection.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SimulatedBlock.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SimulatedBlock.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SuperPool.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/SuperPool.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ThreadConfig.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ThreadConfig.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/TimeQueue.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/TimeQueue.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/TransporterCallback.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/VMSignal.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/VMSignal.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/WaitQueue.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/WatchDog.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/WatchDog.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/al_test/arrayListTest.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/al_test/arrayPoolTest.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/al_test/main.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ndbd_malloc.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/ndbd_malloc.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/pc.hpp: Changed header to GPL version 2 only ndb/src/kernel/vm/testCopy/rr.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/testCopy/testCopy.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/testDataBuffer/testDataBuffer.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/testLongSig/testLongSig.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/testSimplePropertiesSection/test.cpp: Changed header to GPL version 2 only ndb/src/kernel/vm/testSuperPool.cpp: Changed header to GPL version 2 only ndb/src/mgmapi/LocalConfig.cpp: Changed header to GPL version 2 only ndb/src/mgmapi/LocalConfig.hpp: Changed header to GPL version 2 only ndb/src/mgmapi/mgmapi.cpp: Changed header to GPL version 2 only ndb/src/mgmapi/mgmapi_configuration.hpp: Changed header to GPL version 2 only ndb/src/mgmapi/mgmapi_internal.h: Changed header to GPL version 2 only ndb/src/mgmapi/ndb_logevent.cpp: Changed header to GPL version 2 only ndb/src/mgmapi/ndb_logevent.hpp: Changed header to GPL version 2 only ndb/src/mgmapi/test/keso.c: Changed header to GPL version 2 only ndb/src/mgmapi/test/mgmSrvApi.cpp: Changed header to GPL version 2 only ndb/src/mgmclient/CommandInterpreter.cpp: Changed header to GPL version 2 only ndb/src/mgmclient/main.cpp: Changed header to GPL version 2 only ndb/src/mgmclient/ndb_mgmclient.hpp: Changed header to GPL version 2 only ndb/src/mgmclient/ndb_mgmclient.h: Changed header to GPL version 2 only ndb/src/mgmclient/test_cpcd/test_cpcd.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/Config.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/Config.hpp: Changed header to GPL version 2 only ndb/src/mgmsrv/ConfigInfo.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/ConfigInfo.hpp: Changed header to GPL version 2 only ndb/src/mgmsrv/InitConfigFileParser.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/InitConfigFileParser.hpp: Changed header to GPL version 2 only ndb/src/mgmsrv/MgmtSrvr.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/MgmtSrvr.hpp: Changed header to GPL version 2 only ndb/src/mgmsrv/MgmtSrvrConfig.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/MgmtSrvrGeneralSignalHandling.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/Services.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/Services.hpp: Changed header to GPL version 2 only ndb/src/mgmsrv/SignalQueue.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/SignalQueue.hpp: Changed header to GPL version 2 only ndb/src/mgmsrv/convertStrToInt.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/convertStrToInt.hpp: Changed header to GPL version 2 only ndb/src/mgmsrv/main.cpp: Changed header to GPL version 2 only ndb/src/mgmsrv/mkconfig/mkconfig.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/API.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/ClusterMgr.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/ClusterMgr.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/DictCache.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/DictCache.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/Ndb.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbApiSignal.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbApiSignal.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbBlob.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbBlobImpl.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbDictionary.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbDictionaryImpl.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbDictionaryImpl.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbErrorOut.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbEventOperation.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbEventOperationImpl.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbEventOperationImpl.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbImpl.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbIndexOperation.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbLinHash.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbOperation.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbOperationDefine.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbOperationExec.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbOperationInt.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbOperationScan.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbOperationSearch.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbPool.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbPoolImpl.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbPoolImpl.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbRecAttr.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbReceiver.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbScanFilter.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbScanOperation.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbTransaction.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbTransactionScan.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbUtil.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbUtil.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/NdbWaiter.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/Ndberr.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/Ndbif.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/Ndbinit.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/Ndblist.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/ObjectMap.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/SignalSender.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/SignalSender.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/TransporterFacade.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/TransporterFacade.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/ndb_cluster_connection.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/ndb_cluster_connection_impl.hpp: Changed header to GPL version 2 only ndb/src/ndbapi/ndberror.c: Changed header to GPL version 2 only ndb/src/ndbapi/signal-sender/SignalSender.cpp: Changed header to GPL version 2 only ndb/src/ndbapi/signal-sender/SignalSender.hpp: Changed header to GPL version 2 only ndb/test/include/CpcClient.hpp: Changed header to GPL version 2 only ndb/test/include/HugoAsynchTransactions.hpp: Changed header to GPL version 2 only ndb/test/include/HugoCalculator.hpp: Changed header to GPL version 2 only ndb/test/include/HugoOperations.hpp: Changed header to GPL version 2 only ndb/test/include/HugoTransactions.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_DataSet.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_DataSetTransaction.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_Error.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_Output.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_ResultRow.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_ReturnCodes.h: Changed header to GPL version 2 only ndb/test/include/NDBT_Stats.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_Table.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_Tables.hpp: Changed header to GPL version 2 only ndb/test/include/NDBT_Test.hpp: Changed header to GPL version 2 only ndb/test/include/NdbBackup.hpp: Changed header to GPL version 2 only ndb/test/include/NdbConfig.hpp: Changed header to GPL version 2 only ndb/test/include/NdbGrep.hpp: Changed header to GPL version 2 only ndb/test/include/NdbRestarter.hpp: Changed header to GPL version 2 only ndb/test/include/NdbRestarts.hpp: Changed header to GPL version 2 only ndb/test/include/NdbSchemaCon.hpp: Changed header to GPL version 2 only ndb/test/include/NdbSchemaOp.hpp: Changed header to GPL version 2 only ndb/test/include/NdbTest.hpp: Changed header to GPL version 2 only ndb/test/include/NdbTimer.hpp: Changed header to GPL version 2 only ndb/test/include/TestNdbEventOperation.hpp: Changed header to GPL version 2 only ndb/test/include/UtilTransactions.hpp: Changed header to GPL version 2 only ndb/test/include/getarg.h: Changed header to GPL version 2 only ndb/test/ndbapi/InsertRecs.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ScanFilter.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/ScanFunctions.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/ScanInterpretTest.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/TraceNdbApi.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/VerifyNdbApi.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/acid.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/acid2.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/adoInsertRecs.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/asyncGenerator.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/benchronja.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bulk_copy.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/cdrserver.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/celloDb.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/create_all_tabs.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/create_tab.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/drop_all_tabs.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/flexAsynch.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/flexBench.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/flexHammer.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/flexScan.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/flexTT.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/flexTimedAsynch.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/flex_bench_mysql.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/index.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/index2.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/initronja.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/interpreterInTup.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/mainAsyncGenerator.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/msa.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_async1.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_async2.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_user_populate.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_user_transaction.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_user_transaction2.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_user_transaction3.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_user_transaction4.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_user_transaction5.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/ndb_user_transaction6.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/restarter.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/restarter2.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/restarts.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/size.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testBackup.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testBasic.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testBasicAsynch.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testBlobs.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testDataBuffers.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testDeadlock.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testDict.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testGrepVerify.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testIndex.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testInterpreter.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testMgm.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testNdbApi.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testNodeRestart.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testOIBasic.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testOperations.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testOrderedIndex.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testPartitioning.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testReadPerf.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testRestartGci.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testSRBank.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testScan.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testScanInterpreter.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testScanPerf.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testSystemRestart.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/Bank.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/Bank.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/BankLoad.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/bankCreator.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/bankMakeGL.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/bankSumAccounts.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/bankTimer.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/bankTransactionMaker.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/bankValidateAllGLs.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bank/testBank.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/asyncGenerator.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/dbGenerator.h: Changed header to GPL version 2 only ndb/test/ndbapi/bench/dbPopulate.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/dbPopulate.h: Changed header to GPL version 2 only ndb/test/ndbapi/bench/macros.h: Changed header to GPL version 2 only ndb/test/ndbapi/bench/mainAsyncGenerator.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/mainPopulate.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_async1.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_async2.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_error.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_schema.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_user_transaction.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_user_transaction2.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_user_transaction3.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_user_transaction4.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_user_transaction5.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/ndb_user_transaction6.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/testData.h: Changed header to GPL version 2 only ndb/test/ndbapi/bench/testDefinitions.h: Changed header to GPL version 2 only ndb/test/ndbapi/bench/userInterface.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/bench/userInterface.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/acid2/TraceNdbApi.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/acid2/VerifyNdbApi.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/async-src/include/dbGenerator.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/async-src/include/testData.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/async-src/include/userInterface.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/async-src/user/macros.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/async-src/user/ndb_error.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/include/ndb_schema.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/include/testDefinitions.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/generator/dbGenerator.c: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/generator/dbGenerator.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/generator/mainGenerator.c: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/include/testData.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/include/userInterface.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/populator/dbPopulate.c: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/populator/dbPopulate.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/populator/mainPopulate.c: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/localDbPrepare.c: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/macros.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/ndb_error.hpp: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/old/userHandle.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/old/userInterface.c: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/userHandle.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/userInterface.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/userTransaction.c: Changed header to GPL version 2 only ndb/test/ndbapi/testTimeout.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/testTransactions.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/test_event.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/test_event_merge.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/test_event_multi_table.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/userInterface.cpp: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/lmc-bench/src/user/old/userTransaction.c: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/vw_test/bcd.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/vw_test/utv.h: Changed header to GPL version 2 only ndb/test/ndbapi/old_dirs/vw_test/vcdrfunc.h: Changed header to GPL version 2 only ndb/test/newtonapi/basic_test/basic/basic.cpp: Changed header to GPL version 2 only ndb/test/newtonapi/basic_test/bulk_read/br_test.cpp: Changed header to GPL version 2 only ndb/test/newtonapi/basic_test/common.cpp: Changed header to GPL version 2 only ndb/test/newtonapi/basic_test/common.hpp: Changed header to GPL version 2 only ndb/test/newtonapi/basic_test/ptr_binding/ptr_binding_test.cpp: Changed header to GPL version 2 only ndb/test/newtonapi/basic_test/too_basic.cpp: Changed header to GPL version 2 only ndb/test/newtonapi/perf_test/perf.cpp: Changed header to GPL version 2 only ndb/test/odbc/SQL99_test/SQL99_test.cpp: Changed header to GPL version 2 only ndb/test/odbc/SQL99_test/SQL99_test.h: Changed header to GPL version 2 only ndb/test/odbc/client/NDBT_ALLOCHANDLE.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/NDBT_ALLOCHANDLE_HDBC.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/NDBT_SQLConnect.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/NDBT_SQLPrepare.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLAllocEnvTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLAllocHandleTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLAllocHandleTest_bf.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLBindColTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLBindParameterTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLCancelTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLCloseCursorTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLColAttributeTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLColAttributeTest1.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLColAttributeTest2.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLColAttributeTest3.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLConnectTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLCopyDescTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLDescribeColTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLDisconnectTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLDriverConnectTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLEndTranTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLErrorTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLExecDirectTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLExecuteTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLFetchScrollTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLFetchTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLFreeHandleTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLFreeStmtTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetConnectAttrTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetCursorNameTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetDataTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetDescFieldTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetDescRecTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetDiagFieldTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetDiagRecSimpleTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetDiagRecTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetEnvAttrTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetFunctionsTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetInfoTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetStmtAttrTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLGetTypeInfoTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLMoreResultsTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLNumResultColsTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLParamDataTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLPrepareTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLPutDataTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLRowCountTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLSetConnectAttrTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLSetCursorNameTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLSetDescFieldTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLSetDescRecTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLSetEnvAttrTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLSetStmtAttrTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLTablesTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/SQLTransactTest.cpp: Changed header to GPL version 2 only ndb/test/odbc/client/common.hpp: Changed header to GPL version 2 only ndb/test/odbc/client/main.cpp: Changed header to GPL version 2 only ndb/test/odbc/driver/testOdbcDriver.cpp: Changed header to GPL version 2 only ndb/test/odbc/test_compiler/test_compiler.cpp: Changed header to GPL version 2 only ndb/test/run-test/main.cpp: Changed header to GPL version 2 only ndb/test/run-test/run-test.hpp: Changed header to GPL version 2 only ndb/test/src/CpcClient.cpp: Changed header to GPL version 2 only ndb/test/src/HugoAsynchTransactions.cpp: Changed header to GPL version 2 only ndb/test/src/HugoCalculator.cpp: Changed header to GPL version 2 only ndb/test/src/HugoOperations.cpp: Changed header to GPL version 2 only ndb/test/src/HugoTransactions.cpp: Changed header to GPL version 2 only ndb/test/src/NDBT_Error.cpp: Changed header to GPL version 2 only ndb/test/src/NDBT_Output.cpp: Changed header to GPL version 2 only ndb/test/src/NDBT_ResultRow.cpp: Changed header to GPL version 2 only ndb/test/src/NDBT_ReturnCodes.cpp: Changed header to GPL version 2 only ndb/test/src/NDBT_Table.cpp: Changed header to GPL version 2 only ndb/test/src/NDBT_Tables.cpp: Changed header to GPL version 2 only ndb/test/src/NDBT_Test.cpp: Changed header to GPL version 2 only ndb/test/src/NdbBackup.cpp: Changed header to GPL version 2 only ndb/test/src/NdbConfig.cpp: Changed header to GPL version 2 only ndb/test/src/NdbGrep.cpp: Changed header to GPL version 2 only ndb/test/src/NdbRestarter.cpp: Changed header to GPL version 2 only ndb/test/src/NdbRestarts.cpp: Changed header to GPL version 2 only ndb/test/src/NdbSchemaCon.cpp: Changed header to GPL version 2 only ndb/test/src/NdbSchemaOp.cpp: Changed header to GPL version 2 only ndb/test/src/UtilTransactions.cpp: Changed header to GPL version 2 only ndb/test/tools/copy_tab.cpp: Changed header to GPL version 2 only ndb/test/tools/cpcc.cpp: Changed header to GPL version 2 only ndb/test/tools/create_index.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoCalculator.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoFill.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoLoad.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoLockRecords.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoPkDelete.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoPkRead.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoPkReadRecord.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoPkUpdate.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoScanRead.cpp: Changed header to GPL version 2 only ndb/test/tools/hugoScanUpdate.cpp: Changed header to GPL version 2 only ndb/test/tools/old_dirs/waiter/waiter.cpp: Changed header to GPL version 2 only ndb/test/tools/restart.cpp: Changed header to GPL version 2 only ndb/test/tools/transproxy.cpp: Changed header to GPL version 2 only ndb/test/tools/verify_index.cpp: Changed header to GPL version 2 only ndb/tools/delete_all.cpp: Changed header to GPL version 2 only ndb/tools/desc.cpp: Changed header to GPL version 2 only ndb/tools/drop_index.cpp: Changed header to GPL version 2 only ndb/tools/drop_tab.cpp: Changed header to GPL version 2 only ndb/tools/listTables.cpp: Changed header to GPL version 2 only ndb/tools/ndb_config.cpp: Changed header to GPL version 2 only ndb/tools/ndb_test_platform.cpp: Changed header to GPL version 2 only ndb/tools/ndbsql.cpp: Changed header to GPL version 2 only ndb/tools/restore/Restore.cpp: Changed header to GPL version 2 only ndb/tools/restore/Restore.hpp: Changed header to GPL version 2 only ndb/tools/restore/consumer.cpp: Changed header to GPL version 2 only ndb/tools/restore/consumer.hpp: Changed header to GPL version 2 only ndb/tools/restore/consumer_printer.cpp: Changed header to GPL version 2 only ndb/tools/restore/consumer_printer.hpp: Changed header to GPL version 2 only ndb/tools/restore/consumer_restore.cpp: Changed header to GPL version 2 only ndb/tools/restore/consumer_restore.hpp: Changed header to GPL version 2 only ndb/tools/select_all.cpp: Changed header to GPL version 2 only ndb/tools/select_count.cpp: Changed header to GPL version 2 only ndb/tools/waiter.cpp: Changed header to GPL version 2 only ndb/tools/restore/consumer_restorem.cpp: Changed header to GPL version 2 only ndb/tools/restore/restore_main.cpp: Changed header to GPL version 2 only netware/mysql_fix_privilege_tables.pl: Changed header to GPL version 2 only netware/mysql_secure_installation.pl: Changed header to GPL version 2 only os2/Makefile.am: Changed header to GPL version 2 only os2/include/Makefile.am: Changed header to GPL version 2 only os2/include/sys/Makefile.am: Changed header to GPL version 2 only pstack/Makefile.am: Changed header to GPL version 2 only regex/Makefile.am: Changed header to GPL version 2 only scripts/Makefile.am: Changed header to GPL version 2 only scripts/fill_help_tables.sh: Changed header to GPL version 2 only scripts/mysql_config.sh: Changed header to GPL version 2 only scripts/mysql_secure_installation.sh: Changed header to GPL version 2 only server-tools/instance-manager/Makefile.am: Changed header to GPL version 2 only server-tools/instance-manager/buffer.cc: Changed header to GPL version 2 only server-tools/instance-manager/buffer.h: Changed header to GPL version 2 only server-tools/instance-manager/command.cc: Changed header to GPL version 2 only server-tools/instance-manager/command.h: Changed header to GPL version 2 only server-tools/instance-manager/commands.cc: Changed header to GPL version 2 only server-tools/instance-manager/commands.h: Changed header to GPL version 2 only server-tools/instance-manager/guardian.cc: Changed header to GPL version 2 only server-tools/instance-manager/guardian.h: Changed header to GPL version 2 only server-tools/instance-manager/instance.cc: Changed header to GPL version 2 only server-tools/instance-manager/instance.h: Changed header to GPL version 2 only server-tools/instance-manager/instance_map.cc: Changed header to GPL version 2 only server-tools/instance-manager/instance_map.h: Changed header to GPL version 2 only server-tools/instance-manager/instance_options.cc: Changed header to GPL version 2 only server-tools/instance-manager/instance_options.h: Changed header to GPL version 2 only server-tools/instance-manager/listener.cc: Changed header to GPL version 2 only server-tools/instance-manager/listener.h: Changed header to GPL version 2 only server-tools/instance-manager/log.cc: Changed header to GPL version 2 only server-tools/instance-manager/log.h: Changed header to GPL version 2 only server-tools/instance-manager/manager.cc: Changed header to GPL version 2 only server-tools/instance-manager/manager.h: Changed header to GPL version 2 only server-tools/instance-manager/messages.cc: Changed header to GPL version 2 only server-tools/instance-manager/messages.h: Changed header to GPL version 2 only server-tools/instance-manager/mysql_connection.cc: Changed header to GPL version 2 only server-tools/instance-manager/mysql_connection.h: Changed header to GPL version 2 only server-tools/instance-manager/mysql_manager_error.h: Changed header to GPL version 2 only server-tools/instance-manager/mysqlmanager.cc: Changed header to GPL version 2 only server-tools/instance-manager/options.cc: Changed header to GPL version 2 only server-tools/instance-manager/options.h: Changed header to GPL version 2 only server-tools/instance-manager/parse.cc: Changed header to GPL version 2 only server-tools/instance-manager/parse.h: Changed header to GPL version 2 only server-tools/instance-manager/parse_output.cc: Changed header to GPL version 2 only server-tools/instance-manager/parse_output.h: Changed header to GPL version 2 only server-tools/instance-manager/priv.cc: Changed header to GPL version 2 only server-tools/instance-manager/priv.h: Changed header to GPL version 2 only server-tools/instance-manager/protocol.cc: Changed header to GPL version 2 only server-tools/instance-manager/protocol.h: Changed header to GPL version 2 only server-tools/instance-manager/thread_registry.cc: Changed header to GPL version 2 only server-tools/instance-manager/thread_registry.h: Changed header to GPL version 2 only server-tools/instance-manager/user_map.cc: Changed header to GPL version 2 only server-tools/instance-manager/user_map.h: Changed header to GPL version 2 only sql/Makefile.am: Changed header to GPL version 2 only sql/client_settings.h: Changed header to GPL version 2 only sql/custom_conf.h: Changed header to GPL version 2 only sql/derror.cc: Changed header to GPL version 2 only sql/des_key_file.cc: Changed header to GPL version 2 only sql/discover.cc: Changed header to GPL version 2 only sql/field.cc: Changed header to GPL version 2 only sql/field.h: Changed header to GPL version 2 only sql/field_conv.cc: Changed header to GPL version 2 only sql/filesort.cc: Changed header to GPL version 2 only sql/frm_crypt.cc: Changed header to GPL version 2 only sql/gen_lex_hash.cc: Changed header to GPL version 2 only sql/gstream.cc: Changed header to GPL version 2 only sql/gstream.h: Changed header to GPL version 2 only sql/ha_archive.cc: Changed header to GPL version 2 only sql/ha_archive.h: Changed header to GPL version 2 only sql/ha_berkeley.cc: Changed header to GPL version 2 only sql/ha_berkeley.h: Changed header to GPL version 2 only sql/ha_blackhole.cc: Changed header to GPL version 2 only sql/ha_blackhole.h: Changed header to GPL version 2 only sql/ha_federated.cc: Changed header to GPL version 2 only sql/ha_federated.h: Changed header to GPL version 2 only sql/ha_heap.cc: Changed header to GPL version 2 only sql/ha_heap.h: Changed header to GPL version 2 only sql/ha_innodb.cc: Changed header to GPL version 2 only sql/ha_innodb.h: Changed header to GPL version 2 only sql/ha_myisam.cc: Changed header to GPL version 2 only sql/ha_myisam.h: Changed header to GPL version 2 only sql/ha_myisammrg.cc: Changed header to GPL version 2 only sql/ha_myisammrg.h: Changed header to GPL version 2 only sql/ha_ndbcluster.cc: Changed header to GPL version 2 only sql/ha_ndbcluster.h: Changed header to GPL version 2 only sql/handler.cc: Changed header to GPL version 2 only sql/handler.h: Changed header to GPL version 2 only sql/hash_filo.cc: Changed header to GPL version 2 only sql/hash_filo.h: Changed header to GPL version 2 only sql/hostname.cc: Changed header to GPL version 2 only sql/init.cc: Changed header to GPL version 2 only sql/item.cc: Changed header to GPL version 2 only sql/item.h: Changed header to GPL version 2 only sql/item_buff.cc: Changed header to GPL version 2 only sql/item_cmpfunc.cc: Changed header to GPL version 2 only sql/item_cmpfunc.h: Changed header to GPL version 2 only sql/item_create.cc: Changed header to GPL version 2 only sql/item_create.h: Changed header to GPL version 2 only sql/item_func.cc: Changed header to GPL version 2 only sql/item_func.h: Changed header to GPL version 2 only sql/item_geofunc.cc: Changed header to GPL version 2 only sql/item_geofunc.h: Changed header to GPL version 2 only sql/item_row.cc: Changed header to GPL version 2 only sql/item_row.h: Changed header to GPL version 2 only sql/item_strfunc.cc: Changed header to GPL version 2 only sql/item_strfunc.h: Changed header to GPL version 2 only sql/item_subselect.cc: Changed header to GPL version 2 only sql/item_subselect.h: Changed header to GPL version 2 only sql/item_sum.cc: Changed header to GPL version 2 only sql/item_sum.h: Changed header to GPL version 2 only sql/item_timefunc.cc: Changed header to GPL version 2 only sql/item_timefunc.h: Changed header to GPL version 2 only sql/item_uniq.cc: Changed header to GPL version 2 only sql/item_uniq.h: Changed header to GPL version 2 only sql/key.cc: Changed header to GPL version 2 only sql/lex.h: Changed header to GPL version 2 only sql/lex_symbol.h: Changed header to GPL version 2 only sql/lock.cc: Changed header to GPL version 2 only sql/log.cc: Changed header to GPL version 2 only sql/log_event.cc: Changed header to GPL version 2 only sql/log_event.h: Changed header to GPL version 2 only sql/matherr.c: Changed header to GPL version 2 only sql/mf_iocache.cc: Changed header to GPL version 2 only sql/my_decimal.cc: Changed header to GPL version 2 only sql/my_decimal.h: Changed header to GPL version 2 only sql/my_lock.c: Changed header to GPL version 2 only sql/mysql_priv.h: Changed header to GPL version 2 only sql/mysqld.cc: Changed header to GPL version 2 only sql/mysqld_suffix.h: Changed header to GPL version 2 only sql/net_serv.cc: Changed header to GPL version 2 only sql/opt_range.cc: Changed header to GPL version 2 only sql/opt_range.h: Changed header to GPL version 2 only sql/opt_sum.cc: Changed header to GPL version 2 only sql/parse_file.cc: Changed header to GPL version 2 only sql/parse_file.h: Changed header to GPL version 2 only sql/password.c: Changed header to GPL version 2 only sql/procedure.cc: Changed header to GPL version 2 only sql/procedure.h: Changed header to GPL version 2 only sql/protocol.cc: Changed header to GPL version 2 only sql/protocol.h: Changed header to GPL version 2 only sql/records.cc: Changed header to GPL version 2 only sql/repl_failsafe.cc: Changed header to GPL version 2 only sql/repl_failsafe.h: Changed header to GPL version 2 only sql/set_var.cc: Changed header to GPL version 2 only sql/set_var.h: Changed header to GPL version 2 only sql/slave.cc: Changed header to GPL version 2 only sql/slave.h: Changed header to GPL version 2 only sql/sp.cc: Changed header to GPL version 2 only sql/sp.h: Changed header to GPL version 2 only sql/sp_cache.cc: Changed header to GPL version 2 only sql/sp_cache.h: Changed header to GPL version 2 only sql/sp_head.cc: Changed header to GPL version 2 only sql/sp_head.h: Changed header to GPL version 2 only sql/sp_pcontext.cc: Changed header to GPL version 2 only sql/sp_pcontext.h: Changed header to GPL version 2 only sql/sp_rcontext.cc: Changed header to GPL version 2 only sql/sp_rcontext.h: Changed header to GPL version 2 only sql/spatial.cc: Changed header to GPL version 2 only sql/spatial.h: Changed header to GPL version 2 only sql/sql_acl.cc: Changed header to GPL version 2 only sql/sql_acl.h: Changed header to GPL version 2 only sql/sql_analyse.cc: Changed header to GPL version 2 only sql/sql_analyse.h: Changed header to GPL version 2 only sql/sql_array.h: Changed header to GPL version 2 only sql/sql_base.cc: Changed header to GPL version 2 only sql/sql_bitmap.h: Changed header to GPL version 2 only sql/sql_cache.cc: Changed header to GPL version 2 only sql/sql_cache.h: Changed header to GPL version 2 only sql/sql_class.cc: Changed header to GPL version 2 only sql/sql_class.h: Changed header to GPL version 2 only sql/sql_client.cc: Changed header to GPL version 2 only sql/sql_crypt.cc: Changed header to GPL version 2 only sql/sql_crypt.h: Changed header to GPL version 2 only sql/sql_cursor.cc: Changed header to GPL version 2 only sql/sql_cursor.h: Changed header to GPL version 2 only sql/sql_db.cc: Changed header to GPL version 2 only sql/sql_delete.cc: Changed header to GPL version 2 only sql/sql_derived.cc: Changed header to GPL version 2 only sql/sql_do.cc: Changed header to GPL version 2 only sql/sql_error.cc: Changed header to GPL version 2 only sql/sql_error.h: Changed header to GPL version 2 only sql/sql_handler.cc: Changed header to GPL version 2 only sql/sql_help.cc: Changed header to GPL version 2 only sql/sql_insert.cc: Changed header to GPL version 2 only sql/sql_lex.cc: Changed header to GPL version 2 only sql/sql_lex.h: Changed header to GPL version 2 only sql/sql_list.cc: Changed header to GPL version 2 only sql/sql_list.h: Changed header to GPL version 2 only sql/sql_load.cc: Changed header to GPL version 2 only sql/sql_locale.cc: Changed header to GPL version 2 only sql/sql_manager.cc: Changed header to GPL version 2 only sql/sql_manager.h: Changed header to GPL version 2 only sql/sql_map.cc: Changed header to GPL version 2 only sql/sql_map.h: Changed header to GPL version 2 only sql/sql_olap.cc: Changed header to GPL version 2 only sql/sql_parse.cc: Changed header to GPL version 2 only sql/sql_prepare.cc: Changed header to GPL version 2 only sql/sql_rename.cc: Changed header to GPL version 2 only sql/sql_repl.cc: Changed header to GPL version 2 only sql/sql_repl.h: Changed header to GPL version 2 only sql/sql_select.cc: Changed header to GPL version 2 only sql/sql_select.h: Changed header to GPL version 2 only sql/sql_show.cc: Changed header to GPL version 2 only sql/sql_sort.h: Changed header to GPL version 2 only sql/sql_state.c: Changed header to GPL version 2 only sql/sql_string.cc: Changed header to GPL version 2 only sql/sql_string.h: Changed header to GPL version 2 only sql/sql_table.cc: Changed header to GPL version 2 only sql/sql_test.cc: Changed header to GPL version 2 only sql/sql_trigger.cc: Changed header to GPL version 2 only sql/sql_trigger.h: Changed header to GPL version 2 only sql/sql_udf.cc: Changed header to GPL version 2 only sql/sql_udf.h: Changed header to GPL version 2 only sql/sql_union.cc: Changed header to GPL version 2 only sql/sql_update.cc: Changed header to GPL version 2 only sql-bench/Makefile.am: Changed header to GPL version 2 only sql-bench/as3ap.sh: Changed header to GPL version 2 only sql-bench/bench-count-distinct.sh: Changed header to GPL version 2 only sql-bench/bench-init.pl.sh: Changed header to GPL version 2 only sql-bench/compare-results.sh: Changed header to GPL version 2 only sql-bench/copy-db.sh: Changed header to GPL version 2 only sql-bench/crash-me.sh: Changed header to GPL version 2 only sql-bench/print-limit-table: Changed header to GPL version 2 only sql-bench/run-all-tests.sh: Changed header to GPL version 2 only sql/examples/ha_example.cc: Changed header to GPL version 2 only sql/examples/ha_example.h: Changed header to GPL version 2 only sql/examples/ha_tina.cc: Changed header to GPL version 2 only sql/examples/ha_tina.h: Changed header to GPL version 2 only sql/share/Makefile.am: Changed header to GPL version 2 only sql/share/charsets/Index.xml: Changed header to GPL version 2 only sql/share/charsets/armscii8.xml: Changed header to GPL version 2 only sql/share/charsets/ascii.xml: Changed header to GPL version 2 only sql/share/charsets/cp1250.xml: Changed header to GPL version 2 only sql/share/charsets/cp1251.xml: Changed header to GPL version 2 only sql/share/charsets/cp1256.xml: Changed header to GPL version 2 only sql/share/charsets/cp1257.xml: Changed header to GPL version 2 only sql/share/charsets/cp850.xml: Changed header to GPL version 2 only sql/share/charsets/cp852.xml: Changed header to GPL version 2 only sql/share/charsets/cp866.xml: Changed header to GPL version 2 only sql/share/charsets/dec8.xml: Changed header to GPL version 2 only sql/share/charsets/geostd8.xml: Changed header to GPL version 2 only sql/share/charsets/greek.xml: Changed header to GPL version 2 only sql/share/charsets/hebrew.xml: Changed header to GPL version 2 only sql/share/charsets/hp8.xml: Changed header to GPL version 2 only sql/share/charsets/keybcs2.xml: Changed header to GPL version 2 only sql/share/charsets/koi8r.xml: Changed header to GPL version 2 only sql/share/charsets/koi8u.xml: Changed header to GPL version 2 only sql/share/charsets/latin1.xml: Changed header to GPL version 2 only sql/share/charsets/latin2.xml: Changed header to GPL version 2 only sql/share/charsets/latin5.xml: Changed header to GPL version 2 only sql/share/charsets/latin7.xml: Changed header to GPL version 2 only sql/share/charsets/macce.xml: Changed header to GPL version 2 only sql/share/charsets/macroman.xml: Changed header to GPL version 2 only sql/share/charsets/swe7.xml: Changed header to GPL version 2 only sql/sql_view.cc: Changed header to GPL version 2 only sql/sql_view.h: Changed header to GPL version 2 only sql/sql_yacc.yy: Changed header to GPL version 2 only sql/stacktrace.c: Changed header to GPL version 2 only sql/stacktrace.h: Changed header to GPL version 2 only sql/strfunc.cc: Changed header to GPL version 2 only sql/structs.h: Changed header to GPL version 2 only sql/table.cc: Changed header to GPL version 2 only sql/table.h: Changed header to GPL version 2 only sql/thr_malloc.cc: Changed header to GPL version 2 only sql/time.cc: Changed header to GPL version 2 only sql/tzfile.h: Changed header to GPL version 2 only sql/tztime.cc: Changed header to GPL version 2 only sql/tztime.h: Changed header to GPL version 2 only sql/udf_example.c: Changed header to GPL version 2 only sql/uniques.cc: Changed header to GPL version 2 only sql/unireg.cc: Changed header to GPL version 2 only sql/unireg.h: Changed header to GPL version 2 only sql-bench/server-cfg.sh: Changed header to GPL version 2 only sql-bench/test-ATIS.sh: Changed header to GPL version 2 only sql-bench/test-alter-table.sh: Changed header to GPL version 2 only sql-bench/test-big-tables.sh: Changed header to GPL version 2 only sql-bench/test-connect.sh: Changed header to GPL version 2 only sql-bench/test-create.sh: Changed header to GPL version 2 only sql-bench/test-insert.sh: Changed header to GPL version 2 only sql-bench/test-select.sh: Changed header to GPL version 2 only sql-bench/test-transactions.sh: Changed header to GPL version 2 only sql-bench/test-wisconsin.sh: Changed header to GPL version 2 only sql-common/Makefile.am: Changed header to GPL version 2 only sql-common/client.c: Changed header to GPL version 2 only sql-common/my_time.c: Changed header to GPL version 2 only sql-common/my_user.c: Changed header to GPL version 2 only sql-common/pack.c: Changed header to GPL version 2 only strings/Makefile.am: Changed header to GPL version 2 only strings/bchange.c: Changed header to GPL version 2 only strings/bcmp.c: Changed header to GPL version 2 only strings/bcopy-duff.c: Changed header to GPL version 2 only strings/bfill.c: Changed header to GPL version 2 only strings/bmove.c: Changed header to GPL version 2 only strings/bmove512.c: Changed header to GPL version 2 only strings/bmove_upp-sparc.s: Changed header to GPL version 2 only strings/bmove_upp.c: Changed header to GPL version 2 only strings/bzero.c: Changed header to GPL version 2 only strings/conf_to_src.c: Changed header to GPL version 2 only strings/ctype-big5.c: Changed header to GPL version 2 only strings/ctype-bin.c: Changed header to GPL version 2 only strings/ctype-cp932.c: Changed header to GPL version 2 only strings/ctype-czech.c: Changed header to GPL version 2 only strings/ctype-euc_kr.c: Changed header to GPL version 2 only strings/ctype-eucjpms.c: Changed header to GPL version 2 only strings/ctype-gb2312.c: Changed header to GPL version 2 only strings/ctype-gbk.c: Changed header to GPL version 2 only strings/ctype-latin1.c: Changed header to GPL version 2 only strings/ctype-mb.c: Changed header to GPL version 2 only strings/ctype-simple.c: Changed header to GPL version 2 only strings/ctype-sjis.c: Changed header to GPL version 2 only strings/ctype-tis620.c: Changed header to GPL version 2 only strings/ctype-uca.c: Changed header to GPL version 2 only strings/ctype-ucs2.c: Changed header to GPL version 2 only strings/ctype-ujis.c: Changed header to GPL version 2 only strings/ctype-utf8.c: Changed header to GPL version 2 only strings/ctype-win1250ch.c: Changed header to GPL version 2 only strings/ctype.c: Changed header to GPL version 2 only strings/decimal.c: Changed header to GPL version 2 only strings/do_ctype.c: Changed header to GPL version 2 only strings/int2str.c: Changed header to GPL version 2 only strings/is_prefix.c: Changed header to GPL version 2 only strings/llstr.c: Changed header to GPL version 2 only strings/longlong2str-x86.s: Changed header to GPL version 2 only strings/longlong2str.c: Changed header to GPL version 2 only strings/longlong2str_asm.c: Changed header to GPL version 2 only strings/macros.asm: Changed header to GPL version 2 only strings/memcmp.c: Changed header to GPL version 2 only strings/memcpy.c: Changed header to GPL version 2 only strings/memset.c: Changed header to GPL version 2 only strings/my_strtoll10-x86.s: Changed header to GPL version 2 only strings/my_strtoll10.c: Changed header to GPL version 2 only strings/my_vsnprintf.c: Changed header to GPL version 2 only strings/ptr_cmp.asm: Changed header to GPL version 2 only strings/r_strinstr.c: Changed header to GPL version 2 only strings/str2int.c: Changed header to GPL version 2 only strings/str_alloc.c: Changed header to GPL version 2 only strings/str_test.c: Changed header to GPL version 2 only strings/strappend-sparc.s: Changed header to GPL version 2 only strings/strappend.c: Changed header to GPL version 2 only strings/strcat.c: Changed header to GPL version 2 only strings/strcend.c: Changed header to GPL version 2 only strings/strchr.c: Changed header to GPL version 2 only strings/strcmp.c: Changed header to GPL version 2 only strings/strcont.c: Changed header to GPL version 2 only strings/strend-sparc.s: Changed header to GPL version 2 only strings/strend.c: Changed header to GPL version 2 only strings/strfill.c: Changed header to GPL version 2 only strings/strings-not-used.h: Changed header to GPL version 2 only strings/strings-x86.s: Changed header to GPL version 2 only strings/strings.asm: Changed header to GPL version 2 only strings/strinstr-sparc.s: Changed header to GPL version 2 only strings/strinstr.c: Changed header to GPL version 2 only strings/strlen.c: Changed header to GPL version 2 only strings/strmake-sparc.s: Changed header to GPL version 2 only strings/strmake.c: Changed header to GPL version 2 only strings/strmov-sparc.s: Changed header to GPL version 2 only strings/strmov.c: Changed header to GPL version 2 only strings/strnlen.c: Changed header to GPL version 2 only strings/strnmov-sparc.s: Changed header to GPL version 2 only strings/strnmov.c: Changed header to GPL version 2 only strings/strrchr.c: Changed header to GPL version 2 only strings/strstr-sparc.s: Changed header to GPL version 2 only strings/strstr.c: Changed header to GPL version 2 only strings/strto.c: Changed header to GPL version 2 only strings/strtol.c: Changed header to GPL version 2 only strings/strtoll.c: Changed header to GPL version 2 only strings/strtoul.c: Changed header to GPL version 2 only strings/strtoull.c: Changed header to GPL version 2 only strings/strxmov-sparc.s: Changed header to GPL version 2 only strings/strxmov.asm: Changed header to GPL version 2 only strings/strxmov.c: Changed header to GPL version 2 only strings/strxnmov.c: Changed header to GPL version 2 only strings/t_ctype.h: Changed header to GPL version 2 only strings/udiv.c: Changed header to GPL version 2 only strings/xml.c: Changed header to GPL version 2 only support-files/MacOSX/Makefile.am: Changed header to GPL version 2 only support-files/Makefile.am: Changed header to GPL version 2 only support-files/MySQL-shared-compat.spec.sh: Changed header to GPL version 2 only tests/Makefile.am: Changed header to GPL version 2 only tests/connect_test.c: Changed header to GPL version 2 only tests/deadlock_test.c: Changed header to GPL version 2 only tests/insert_test.c: Changed header to GPL version 2 only tests/list_test.c: Changed header to GPL version 2 only tests/mysql_client_test.c: Changed header to GPL version 2 only tests/select_test.c: Changed header to GPL version 2 only tests/showdb_test.c: Changed header to GPL version 2 only tests/ssl_test.c: Changed header to GPL version 2 only tests/thread_test.c: Changed header to GPL version 2 only tools/Makefile.am: Changed header to GPL version 2 only tools/mysqlmanager.c: Changed header to GPL version 2 only vio/Makefile.am: Changed header to GPL version 2 only vio/test-ssl.c: Changed header to GPL version 2 only vio/test-sslclient.c: Changed header to GPL version 2 only vio/test-sslserver.c: Changed header to GPL version 2 only vio/vio.c: Changed header to GPL version 2 only vio/vio_priv.h: Changed header to GPL version 2 only vio/viosocket.c: Changed header to GPL version 2 only vio/viossl.c: Changed header to GPL version 2 only vio/viosslfactories.c: Changed header to GPL version 2 only vio/viotest-ssl.c: Changed header to GPL version 2 only win/Makefile.am: Changed header to GPL version 2 only zlib/Makefile.am: Changed header to GPL version 2 only
2006-12-23 20:17:15 +01:00
the Free Software Foundation; version 2 of the License.
Update copyright Fixed memory leak on shutdown (Affects the embedded version & MyODBC) client/client_priv.h: Update copyright client/completion_hash.cc: Update copyright client/completion_hash.h: Update copyright client/connect_test.c: Update copyright client/errmsg.c: Update copyright client/get_password.c: Update copyright client/insert_test.c: Update copyright client/list_test.c: Update copyright client/my_readline.h: Update copyright client/mysql.cc: Update copyright client/mysqladmin.c: Update copyright client/mysqlbinlog.cc: Update copyright client/mysqlcheck.c: Update copyright client/mysqldump.c: Update copyright client/mysqlimport.c: Update copyright client/mysqlmanager-pwgen.c: Update copyright client/mysqlmanagerc.c: Update copyright client/mysqlshow.c: Update copyright client/mysqltest.c: Update copyright client/password.c: Update copyright client/readline.cc: Update copyright client/select_test.c: Update copyright client/showdb_test.c: Update copyright client/sql_string.cc: Update copyright client/sql_string.h: Update copyright client/ssl_test.c: Update copyright client/thimble.cc: Update copyright client/thread_test.c: Update copyright div/deadlock_test.c: Update copyright extra/comp_err.c: Update copyright extra/my_print_defaults.c: Update copyright extra/perror.c: Update copyright extra/replace.c: Update copyright extra/resolve_stack_dump.c: Update copyright extra/resolveip.c: Update copyright fs/database.c: Update copyright fs/libmysqlfs.c: Update copyright fs/mysqlcorbafs.c: Update copyright fs/mysqlcorbafs.h: Update copyright fs/mysqlcorbafs_test.c: Update copyright heap/_check.c: Update copyright heap/_rectest.c: Update copyright heap/heapdef.h: Update copyright heap/hp_block.c: Update copyright heap/hp_clear.c: Update copyright heap/hp_close.c: Update copyright heap/hp_create.c: Update copyright heap/hp_delete.c: Update copyright heap/hp_extra.c: Update copyright heap/hp_hash.c: Update copyright heap/hp_info.c: Update copyright heap/hp_open.c: Update copyright heap/hp_panic.c: Update copyright heap/hp_rename.c: Update copyright heap/hp_rfirst.c: Update copyright heap/hp_rkey.c: Update copyright heap/hp_rlast.c: Update copyright heap/hp_rnext.c: Update copyright heap/hp_rprev.c: Update copyright heap/hp_rrnd.c: Update copyright heap/hp_rsame.c: Update copyright heap/hp_scan.c: Update copyright heap/hp_static.c: Update copyright heap/hp_test1.c: Update copyright heap/hp_test2.c: Update copyright heap/hp_update.c: Update copyright heap/hp_write.c: Update copyright include/config-win.h: Update copyright include/dbug.h: Update copyright include/errmsg.h: Update copyright include/ft_global.h: Update copyright include/getopt.h: Update copyright include/hash.h: Update copyright include/heap.h: Update copyright include/m_ctype.h: Update copyright include/m_string.h: Update copyright include/md5.h: Update copyright include/merge.h: Update copyright include/my_alarm.h: Update copyright include/my_base.h: Update copyright include/my_bitmap.h: Update copyright include/my_dir.h: Update copyright include/my_global.h: Update copyright include/my_list.h: Update copyright include/my_net.h: Update copyright include/my_no_pthread.h: Update copyright include/my_nosys.h: Update copyright include/my_pthread.h: Update copyright include/my_sys.h: Update copyright include/my_tree.h: Update copyright include/myisam.h: Update copyright include/myisammrg.h: Update copyright include/myisampack.h: Update copyright include/mysql.h: Update copyright include/mysql_com.h: Update copyright include/mysql_embed.h: Update copyright include/mysqld_error.h: Update copyright include/mysys_err.h: Update copyright include/nisam.h: Update copyright include/queues.h: Update copyright include/raid.h: Update copyright include/sslopt-case.h: Update copyright include/sslopt-longopts.h: Update copyright include/sslopt-usage.h: Update copyright include/sslopt-vars.h: Update copyright include/t_ctype.h: Update copyright include/thr_alarm.h: Update copyright include/thr_lock.h: Update copyright include/violite.h: Update copyright isam/_cache.c: Update copyright isam/_dbug.c: Update copyright isam/_key.c: Update copyright isam/_locking.c: Update copyright isam/_packrec.c: Update copyright isam/_page.c: Update copyright isam/_search.c: Update copyright isam/_statrec.c: Update copyright isam/changed.c: Update copyright isam/close.c: Update copyright isam/create.c: Update copyright isam/delete.c: Update copyright isam/extra.c: Update copyright isam/info.c: Update copyright isam/isamchk.c: Update copyright isam/isamdef.h: Update copyright isam/log.c: Update copyright isam/open.c: Update copyright isam/panic.c: Update copyright isam/range.c: Update copyright isam/rfirst.c: Update copyright isam/rkey.c: Update copyright isam/rlast.c: Update copyright isam/rnext.c: Update copyright isam/rprev.c: Update copyright isam/rrnd.c: Update copyright isam/rsame.c: Update copyright isam/rsamepos.c: Update copyright isam/sort.c: Update copyright isam/static.c: Update copyright isam/test1.c: Update copyright isam/test2.c: Update copyright isam/test3.c: Update copyright isam/update.c: Update copyright isam/write.c: Update copyright libmysql/conf_to_src.c: Update copyright libmysql/dll.c: Update copyright libmysql/errmsg.c: Update copyright libmysql/get_password.c: Update copyright libmysql/libmysql.c: Update copyright libmysql/manager.c: Update copyright libmysql/net.c: Update copyright libmysql/password.c: Update copyright libmysqld/lib_sql.cc: Update copyright libmysqld/lib_vio.c: Update copyright libmysqld/libmysqld.c: Update copyright merge/mrg_close.c: Update copyright merge/mrg_create.c: Update copyright merge/mrg_def.h: Update copyright merge/mrg_delete.c: Update copyright merge/mrg_extra.c: Update copyright merge/mrg_info.c: Update copyright merge/mrg_locking.c: Update copyright merge/mrg_open.c: Update copyright merge/mrg_panic.c: Update copyright merge/mrg_rrnd.c: Update copyright merge/mrg_rsame.c: Update copyright merge/mrg_static.c: Update copyright merge/mrg_update.c: Update copyright myisam/ft_boolean_search.c: Update copyright myisam/ft_dump.c: Update copyright myisam/ft_eval.h: Update copyright myisam/ft_static.c: Update copyright myisam/ft_stem.c: Update copyright myisam/ft_stopwords.c: Update copyright myisam/ft_test1.h: Update copyright myisam/mi_cache.c: Update copyright myisam/mi_changed.c: Update copyright myisam/mi_check.c: Update copyright myisam/mi_checksum.c: Update copyright myisam/mi_close.c: Update copyright myisam/mi_create.c: Update copyright myisam/mi_dbug.c: Update copyright myisam/mi_delete.c: Update copyright myisam/mi_delete_all.c: Update copyright myisam/mi_delete_table.c: Update copyright myisam/mi_dynrec.c: Update copyright myisam/mi_extra.c: Update copyright myisam/mi_info.c: Update copyright myisam/mi_key.c: Update copyright myisam/mi_locking.c: Update copyright myisam/mi_log.c: Update copyright myisam/mi_open.c: Update copyright myisam/mi_packrec.c: Update copyright myisam/mi_page.c: Update copyright myisam/mi_panic.c: Update copyright myisam/mi_range.c: Update copyright myisam/mi_rename.c: Update copyright myisam/mi_rfirst.c: Update copyright myisam/mi_rlast.c: Update copyright myisam/mi_rnext_same.c: Update copyright myisam/mi_rrnd.c: Update copyright myisam/mi_rsame.c: Update copyright myisam/mi_rsamepos.c: Update copyright myisam/mi_scan.c: Update copyright myisam/mi_search.c: Update copyright myisam/mi_static.c: Update copyright myisam/mi_statrec.c: Update copyright myisam/mi_test1.c: Update copyright myisam/mi_test2.c: Update copyright myisam/mi_test3.c: Update copyright myisam/mi_unique.c: Update copyright myisam/mi_update.c: Update copyright myisam/mi_write.c: Update copyright myisam/myisamchk.c: Update copyright myisam/myisampack.c: Update copyright myisammrg/myrg_close.c: Update copyright myisammrg/myrg_create.c: Update copyright myisammrg/myrg_def.h: Update copyright myisammrg/myrg_delete.c: Update copyright myisammrg/myrg_locking.c: Update copyright myisammrg/myrg_open.c: Update copyright myisammrg/myrg_panic.c: Update copyright myisammrg/myrg_rsame.c: Update copyright myisammrg/myrg_static.c: Update copyright myisammrg/myrg_update.c: Update copyright myisammrg/myrg_write.c: Update copyright mysql-test/r/gcc296.result: Update of benchmark results mysql-test/r/innodb.result: Update of benchmark results mysql-test/r/join_outer.result: Update of benchmark results mysql-test/r/myisam.result: Update of benchmark results mysys/array.c: Update copyright mysys/charset.c: Fix for restart of character sets mysys/checksum.c: Update copyright mysys/default.c: Update copyright mysys/errors.c: Update copyright mysys/getopt.c: Cleanup mysys/getvar.c: Update copyright mysys/hash.c: Update copyright mysys/list.c: Update copyright mysys/make-conf.c: Update copyright mysys/md5.c: Update copyright mysys/mf_brkhant.c: Update copyright mysys/mf_cache.c: Update copyright mysys/mf_casecnv.c: Update copyright mysys/mf_dirname.c: Update copyright mysys/mf_fn_ext.c: Update copyright mysys/mf_format.c: Update copyright mysys/mf_getdate.c: Update copyright mysys/mf_iocache.c: Update copyright mysys/mf_iocache2.c: Update copyright mysys/mf_keycache.c: Update copyright mysys/mf_loadpath.c: Update copyright mysys/mf_pack.c: Update copyright mysys/mf_path.c: Update copyright mysys/mf_qsort.c: Update copyright mysys/mf_qsort2.c: Update copyright mysys/mf_radix.c: Update copyright mysys/mf_same.c: Update copyright mysys/mf_sleep.c: Update copyright mysys/mf_sort.c: Update copyright mysys/mf_soundex.c: Update copyright mysys/mf_stripp.c: Update copyright mysys/mf_tempfile.c: Update copyright mysys/mf_unixpath.c: Update copyright mysys/mf_util.c: Update copyright mysys/mf_wcomp.c: Update copyright mysys/mf_wfile.c: Update copyright mysys/mulalloc.c: Update copyright mysys/my_alarm.c: Update copyright mysys/my_alloc.c: Update copyright mysys/my_append.c: Update copyright mysys/my_bit.c: Update copyright mysys/my_bitmap.c: Update copyright mysys/my_chsize.c: Update copyright mysys/my_clock.c: Update copyright mysys/my_compress.c: Update copyright mysys/my_copy.c: Update copyright mysys/my_create.c: Update copyright mysys/my_delete.c: Update copyright mysys/my_div.c: Update copyright mysys/my_dup.c: Update copyright mysys/my_error.c: Update copyright mysys/my_fopen.c: Update copyright mysys/my_fstream.c: Update copyright mysys/my_getwd.c: Update copyright mysys/my_init.c: Free 'once_alloc' memory at shutdown. mysys/my_lib.c: Update copyright mysys/my_lock.c: Update copyright mysys/my_lockmem.c: Update copyright mysys/my_lread.c: Update copyright mysys/my_lwrite.c: Update copyright mysys/my_malloc.c: Update copyright mysys/my_messnc.c: Update copyright mysys/my_mkdir.c: Update copyright mysys/my_net.c: Update copyright mysys/my_once.c: Update copyright mysys/my_open.c: Update copyright mysys/my_pread.c: Update copyright mysys/my_pthread.c: Update copyright mysys/my_quick.c: Update copyright mysys/my_read.c: Update copyright mysys/my_realloc.c: Update copyright mysys/my_redel.c: Update copyright mysys/my_rename.c: Update copyright mysys/my_seek.c: Update copyright mysys/my_static.c: Update copyright mysys/my_static.h: Update copyright mysys/my_symlink.c: Update copyright mysys/my_symlink2.c: Update copyright mysys/my_tempnam.c: Update copyright mysys/my_thr_init.c: Update copyright mysys/my_vsnprintf.c: Update copyright mysys/my_wincond.c: Update copyright mysys/my_winthread.c: Update copyright mysys/my_write.c: Update copyright mysys/mysys_priv.h: Update copyright mysys/ptr_cmp.c: Update copyright mysys/queues.c: Update copyright mysys/raid.cc: Update copyright mysys/safemalloc.c: Update copyright mysys/string.c: Update copyright mysys/test_charset.c: Update copyright mysys/test_dir.c: Update copyright mysys/test_fn.c: Update copyright mysys/testhash.c: Update copyright mysys/thr_alarm.c: Update copyright mysys/thr_lock.c: Update copyright mysys/thr_mutex.c: Update copyright mysys/thr_rwlock.c: Update copyright mysys/tree.c: Update copyright mysys/typelib.c: Update copyright pstack/debug.c: Update copyright pstack/debug.h: Update copyright pstack/demangle.h: Update copyright pstack/ieee.c: Update copyright pstack/ieee.h: Update copyright pstack/pstack.c: Update copyright readline/bind.c: Cleanup empty lines readline/complete.c: Cleanup empty lines readline/display.c: Cleanup empty lines readline/funmap.c: Cleanup empty lines readline/histexpand.c: Cleanup empty lines readline/histfile.c: Cleanup empty lines readline/history.c: Cleanup empty lines readline/history.h: Cleanup empty lines readline/input.c: Cleanup empty lines readline/kill.c: Cleanup empty lines readline/readline.c: Cleanup empty lines readline/readline.h: Cleanup empty lines readline/vi_mode.c: Cleanup empty lines sql/cache_manager.cc: Update copyright sql/cache_manager.h: Update copyright sql/convert.cc: Update copyright sql/custom_conf.h: Update copyright sql/derror.cc: Update copyright sql/field.cc: Update copyright sql/field.h: Update copyright sql/field_conv.cc: Update copyright sql/filesort.cc: Update copyright sql/frm_crypt.cc: Update copyright sql/ha_berkeley.cc: Update copyright sql/ha_heap.cc: Update copyright sql/ha_heap.h: Update copyright sql/ha_innobase.cc: Update copyright sql/ha_isam.cc: Update copyright sql/ha_isam.h: Update copyright sql/ha_isammrg.cc: Update copyright sql/ha_isammrg.h: Update copyright sql/ha_myisam.cc: Update copyright sql/handler.cc: Update copyright sql/hash_filo.cc: Update copyright sql/hash_filo.h: Update copyright sql/hostname.cc: Update copyright sql/init.cc: Update copyright sql/item.cc: Update copyright sql/item.h: Update copyright sql/item_buff.cc: Update copyright sql/item_cmpfunc.cc: Update copyright sql/item_cmpfunc.h: Update copyright sql/item_create.cc: Update copyright sql/item_create.h: Update copyright sql/item_func.cc: Update copyright sql/item_strfunc.cc: Update copyright sql/item_sum.cc: Update copyright sql/item_sum.h: Update copyright sql/item_timefunc.cc: Update copyright sql/item_timefunc.h: Update copyright sql/item_uniq.cc: Update copyright sql/item_uniq.h: Update copyright sql/key.cc: Update copyright sql/lex_symbol.h: Update copyright sql/lock.cc: Update copyright sql/log.cc: Update copyright sql/log_event.cc: Update copyright sql/log_event.h: Update copyright sql/matherr.c: Update copyright sql/mf_iocache.cc: Update copyright sql/mini_client.cc: Update copyright sql/mini_client.h: Update copyright sql/my_lock.c: Update copyright sql/mysqld.cc: Update copyright sql/net_pkg.cc: Update copyright sql/net_serv.cc: Update copyright sql/opt_sum.cc: Update copyright sql/password.c: Update copyright sql/procedure.cc: Update copyright sql/procedure.h: Update copyright sql/records.cc: Update copyright sql/repl_failsafe.cc: Update copyright sql/slave.cc: Update copyright sql/slave.h: Update copyright sql/sql_acl.cc: Update copyright sql/sql_acl.h: Update copyright sql/sql_analyse.cc: Update copyright sql/sql_analyse.h: Update copyright sql/sql_base.cc: Update copyright sql/sql_cache.cc: Update copyright sql/sql_class.cc: Update copyright sql/sql_class.h: Update copyright sql/sql_crypt.cc: Update copyright sql/sql_crypt.h: Update copyright sql/sql_db.cc: Update copyright sql/sql_delete.cc: Update copyright sql/sql_handler.cc: Update copyright sql/sql_insert.cc: Update copyright sql/sql_lex.cc: Update copyright sql/sql_lex.h: Update copyright sql/sql_list.cc: Update copyright sql/sql_list.h: Update copyright sql/sql_load.cc: Update copyright sql/sql_map.cc: Update copyright sql/sql_map.h: Update copyright sql/sql_parse.cc: Update copyright sql/sql_rename.cc: Update copyright sql/sql_repl.cc: Update copyright sql/sql_select.h: Update copyright sql/sql_string.cc: Update copyright sql/sql_string.h: Update copyright sql/sql_table.cc: Update copyright sql/sql_test.cc: Update copyright sql/sql_udf.cc: Update copyright sql/sql_udf.h: Update copyright sql/stacktrace.c: Update copyright sql/structs.h: Update copyright sql/table.cc: Update copyright sql/table.h: Update copyright sql/thr_malloc.cc: Update copyright sql/time.cc: Update copyright sql/udf_example.cc: Update copyright sql/uniques.cc: Update copyright sql/unireg.cc: Update copyright sql/unireg.h: Update copyright strings/atof.c: Update copyright strings/bchange.c: Update copyright strings/bcmp.c: Update copyright strings/bcopy-duff.c: Update copyright strings/bfill.c: Update copyright strings/bmove.c: Update copyright strings/bmove512.c: Update copyright strings/bmove_upp.c: Update copyright strings/bzero.c: Update copyright strings/conf_to_src.c: Update copyright strings/ctype-big5.c: Update copyright strings/ctype-czech.c: Update copyright strings/ctype-euc_kr.c: Update copyright strings/ctype-gb2312.c: Update copyright strings/ctype-gbk.c: Update copyright strings/ctype-latin1_de.c: Update copyright strings/ctype-sjis.c: Update copyright strings/ctype-tis620.c: Update copyright strings/ctype-ujis.c: Update copyright strings/ctype.c: Update copyright strings/do_ctype.c: Update copyright strings/int2str.c: Update copyright strings/is_prefix.c: Update copyright strings/llstr.c: Update copyright strings/longlong2str.c: Update copyright strings/memcmp.c: Update copyright strings/memcpy.c: Update copyright strings/memset.c: Update copyright strings/r_strinstr.c: Update copyright strings/str2int.c: Update copyright strings/str_test.c: Update copyright strings/strappend.c: Update copyright strings/strcat.c: Update copyright strings/strcend.c: Update copyright strings/strchr.c: Update copyright strings/strcmp.c: Update copyright strings/strcont.c: Update copyright strings/strend.c: Update copyright strings/strfill.c: Update copyright strings/strings-not-used.h: Update copyright strings/strinstr.c: Update copyright strings/strlen.c: Update copyright strings/strmake.c: Update copyright strings/strmov.c: Update copyright strings/strnlen.c: Update copyright strings/strnmov.c: Update copyright strings/strrchr.c: Update copyright strings/strstr.c: Update copyright strings/strto.c: Update copyright strings/strtol.c: Update copyright strings/strtoll.c: Update copyright strings/strtoul.c: Update copyright strings/strtoull.c: Update copyright strings/strxmov.c: Update copyright strings/strxnmov.c: Update copyright strings/t_ctype.h: Update copyright strings/udiv.c: Update copyright tools/mysqlmanager.c: Update copyright vio/test-ssl.c: Update copyright vio/test-sslclient.c: Update copyright vio/test-sslserver.c: Update copyright vio/vio.c: Update copyright vio/viosocket.c: Update copyright vio/viossl.c: Update copyright vio/viosslfactories.c: Update copyright vio/viotest-ssl.c: Update copyright
2001-12-06 13:10:51 +01:00
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
Update copyright Fixed memory leak on shutdown (Affects the embedded version & MyODBC) client/client_priv.h: Update copyright client/completion_hash.cc: Update copyright client/completion_hash.h: Update copyright client/connect_test.c: Update copyright client/errmsg.c: Update copyright client/get_password.c: Update copyright client/insert_test.c: Update copyright client/list_test.c: Update copyright client/my_readline.h: Update copyright client/mysql.cc: Update copyright client/mysqladmin.c: Update copyright client/mysqlbinlog.cc: Update copyright client/mysqlcheck.c: Update copyright client/mysqldump.c: Update copyright client/mysqlimport.c: Update copyright client/mysqlmanager-pwgen.c: Update copyright client/mysqlmanagerc.c: Update copyright client/mysqlshow.c: Update copyright client/mysqltest.c: Update copyright client/password.c: Update copyright client/readline.cc: Update copyright client/select_test.c: Update copyright client/showdb_test.c: Update copyright client/sql_string.cc: Update copyright client/sql_string.h: Update copyright client/ssl_test.c: Update copyright client/thimble.cc: Update copyright client/thread_test.c: Update copyright div/deadlock_test.c: Update copyright extra/comp_err.c: Update copyright extra/my_print_defaults.c: Update copyright extra/perror.c: Update copyright extra/replace.c: Update copyright extra/resolve_stack_dump.c: Update copyright extra/resolveip.c: Update copyright fs/database.c: Update copyright fs/libmysqlfs.c: Update copyright fs/mysqlcorbafs.c: Update copyright fs/mysqlcorbafs.h: Update copyright fs/mysqlcorbafs_test.c: Update copyright heap/_check.c: Update copyright heap/_rectest.c: Update copyright heap/heapdef.h: Update copyright heap/hp_block.c: Update copyright heap/hp_clear.c: Update copyright heap/hp_close.c: Update copyright heap/hp_create.c: Update copyright heap/hp_delete.c: Update copyright heap/hp_extra.c: Update copyright heap/hp_hash.c: Update copyright heap/hp_info.c: Update copyright heap/hp_open.c: Update copyright heap/hp_panic.c: Update copyright heap/hp_rename.c: Update copyright heap/hp_rfirst.c: Update copyright heap/hp_rkey.c: Update copyright heap/hp_rlast.c: Update copyright heap/hp_rnext.c: Update copyright heap/hp_rprev.c: Update copyright heap/hp_rrnd.c: Update copyright heap/hp_rsame.c: Update copyright heap/hp_scan.c: Update copyright heap/hp_static.c: Update copyright heap/hp_test1.c: Update copyright heap/hp_test2.c: Update copyright heap/hp_update.c: Update copyright heap/hp_write.c: Update copyright include/config-win.h: Update copyright include/dbug.h: Update copyright include/errmsg.h: Update copyright include/ft_global.h: Update copyright include/getopt.h: Update copyright include/hash.h: Update copyright include/heap.h: Update copyright include/m_ctype.h: Update copyright include/m_string.h: Update copyright include/md5.h: Update copyright include/merge.h: Update copyright include/my_alarm.h: Update copyright include/my_base.h: Update copyright include/my_bitmap.h: Update copyright include/my_dir.h: Update copyright include/my_global.h: Update copyright include/my_list.h: Update copyright include/my_net.h: Update copyright include/my_no_pthread.h: Update copyright include/my_nosys.h: Update copyright include/my_pthread.h: Update copyright include/my_sys.h: Update copyright include/my_tree.h: Update copyright include/myisam.h: Update copyright include/myisammrg.h: Update copyright include/myisampack.h: Update copyright include/mysql.h: Update copyright include/mysql_com.h: Update copyright include/mysql_embed.h: Update copyright include/mysqld_error.h: Update copyright include/mysys_err.h: Update copyright include/nisam.h: Update copyright include/queues.h: Update copyright include/raid.h: Update copyright include/sslopt-case.h: Update copyright include/sslopt-longopts.h: Update copyright include/sslopt-usage.h: Update copyright include/sslopt-vars.h: Update copyright include/t_ctype.h: Update copyright include/thr_alarm.h: Update copyright include/thr_lock.h: Update copyright include/violite.h: Update copyright isam/_cache.c: Update copyright isam/_dbug.c: Update copyright isam/_key.c: Update copyright isam/_locking.c: Update copyright isam/_packrec.c: Update copyright isam/_page.c: Update copyright isam/_search.c: Update copyright isam/_statrec.c: Update copyright isam/changed.c: Update copyright isam/close.c: Update copyright isam/create.c: Update copyright isam/delete.c: Update copyright isam/extra.c: Update copyright isam/info.c: Update copyright isam/isamchk.c: Update copyright isam/isamdef.h: Update copyright isam/log.c: Update copyright isam/open.c: Update copyright isam/panic.c: Update copyright isam/range.c: Update copyright isam/rfirst.c: Update copyright isam/rkey.c: Update copyright isam/rlast.c: Update copyright isam/rnext.c: Update copyright isam/rprev.c: Update copyright isam/rrnd.c: Update copyright isam/rsame.c: Update copyright isam/rsamepos.c: Update copyright isam/sort.c: Update copyright isam/static.c: Update copyright isam/test1.c: Update copyright isam/test2.c: Update copyright isam/test3.c: Update copyright isam/update.c: Update copyright isam/write.c: Update copyright libmysql/conf_to_src.c: Update copyright libmysql/dll.c: Update copyright libmysql/errmsg.c: Update copyright libmysql/get_password.c: Update copyright libmysql/libmysql.c: Update copyright libmysql/manager.c: Update copyright libmysql/net.c: Update copyright libmysql/password.c: Update copyright libmysqld/lib_sql.cc: Update copyright libmysqld/lib_vio.c: Update copyright libmysqld/libmysqld.c: Update copyright merge/mrg_close.c: Update copyright merge/mrg_create.c: Update copyright merge/mrg_def.h: Update copyright merge/mrg_delete.c: Update copyright merge/mrg_extra.c: Update copyright merge/mrg_info.c: Update copyright merge/mrg_locking.c: Update copyright merge/mrg_open.c: Update copyright merge/mrg_panic.c: Update copyright merge/mrg_rrnd.c: Update copyright merge/mrg_rsame.c: Update copyright merge/mrg_static.c: Update copyright merge/mrg_update.c: Update copyright myisam/ft_boolean_search.c: Update copyright myisam/ft_dump.c: Update copyright myisam/ft_eval.h: Update copyright myisam/ft_static.c: Update copyright myisam/ft_stem.c: Update copyright myisam/ft_stopwords.c: Update copyright myisam/ft_test1.h: Update copyright myisam/mi_cache.c: Update copyright myisam/mi_changed.c: Update copyright myisam/mi_check.c: Update copyright myisam/mi_checksum.c: Update copyright myisam/mi_close.c: Update copyright myisam/mi_create.c: Update copyright myisam/mi_dbug.c: Update copyright myisam/mi_delete.c: Update copyright myisam/mi_delete_all.c: Update copyright myisam/mi_delete_table.c: Update copyright myisam/mi_dynrec.c: Update copyright myisam/mi_extra.c: Update copyright myisam/mi_info.c: Update copyright myisam/mi_key.c: Update copyright myisam/mi_locking.c: Update copyright myisam/mi_log.c: Update copyright myisam/mi_open.c: Update copyright myisam/mi_packrec.c: Update copyright myisam/mi_page.c: Update copyright myisam/mi_panic.c: Update copyright myisam/mi_range.c: Update copyright myisam/mi_rename.c: Update copyright myisam/mi_rfirst.c: Update copyright myisam/mi_rlast.c: Update copyright myisam/mi_rnext_same.c: Update copyright myisam/mi_rrnd.c: Update copyright myisam/mi_rsame.c: Update copyright myisam/mi_rsamepos.c: Update copyright myisam/mi_scan.c: Update copyright myisam/mi_search.c: Update copyright myisam/mi_static.c: Update copyright myisam/mi_statrec.c: Update copyright myisam/mi_test1.c: Update copyright myisam/mi_test2.c: Update copyright myisam/mi_test3.c: Update copyright myisam/mi_unique.c: Update copyright myisam/mi_update.c: Update copyright myisam/mi_write.c: Update copyright myisam/myisamchk.c: Update copyright myisam/myisampack.c: Update copyright myisammrg/myrg_close.c: Update copyright myisammrg/myrg_create.c: Update copyright myisammrg/myrg_def.h: Update copyright myisammrg/myrg_delete.c: Update copyright myisammrg/myrg_locking.c: Update copyright myisammrg/myrg_open.c: Update copyright myisammrg/myrg_panic.c: Update copyright myisammrg/myrg_rsame.c: Update copyright myisammrg/myrg_static.c: Update copyright myisammrg/myrg_update.c: Update copyright myisammrg/myrg_write.c: Update copyright mysql-test/r/gcc296.result: Update of benchmark results mysql-test/r/innodb.result: Update of benchmark results mysql-test/r/join_outer.result: Update of benchmark results mysql-test/r/myisam.result: Update of benchmark results mysys/array.c: Update copyright mysys/charset.c: Fix for restart of character sets mysys/checksum.c: Update copyright mysys/default.c: Update copyright mysys/errors.c: Update copyright mysys/getopt.c: Cleanup mysys/getvar.c: Update copyright mysys/hash.c: Update copyright mysys/list.c: Update copyright mysys/make-conf.c: Update copyright mysys/md5.c: Update copyright mysys/mf_brkhant.c: Update copyright mysys/mf_cache.c: Update copyright mysys/mf_casecnv.c: Update copyright mysys/mf_dirname.c: Update copyright mysys/mf_fn_ext.c: Update copyright mysys/mf_format.c: Update copyright mysys/mf_getdate.c: Update copyright mysys/mf_iocache.c: Update copyright mysys/mf_iocache2.c: Update copyright mysys/mf_keycache.c: Update copyright mysys/mf_loadpath.c: Update copyright mysys/mf_pack.c: Update copyright mysys/mf_path.c: Update copyright mysys/mf_qsort.c: Update copyright mysys/mf_qsort2.c: Update copyright mysys/mf_radix.c: Update copyright mysys/mf_same.c: Update copyright mysys/mf_sleep.c: Update copyright mysys/mf_sort.c: Update copyright mysys/mf_soundex.c: Update copyright mysys/mf_stripp.c: Update copyright mysys/mf_tempfile.c: Update copyright mysys/mf_unixpath.c: Update copyright mysys/mf_util.c: Update copyright mysys/mf_wcomp.c: Update copyright mysys/mf_wfile.c: Update copyright mysys/mulalloc.c: Update copyright mysys/my_alarm.c: Update copyright mysys/my_alloc.c: Update copyright mysys/my_append.c: Update copyright mysys/my_bit.c: Update copyright mysys/my_bitmap.c: Update copyright mysys/my_chsize.c: Update copyright mysys/my_clock.c: Update copyright mysys/my_compress.c: Update copyright mysys/my_copy.c: Update copyright mysys/my_create.c: Update copyright mysys/my_delete.c: Update copyright mysys/my_div.c: Update copyright mysys/my_dup.c: Update copyright mysys/my_error.c: Update copyright mysys/my_fopen.c: Update copyright mysys/my_fstream.c: Update copyright mysys/my_getwd.c: Update copyright mysys/my_init.c: Free 'once_alloc' memory at shutdown. mysys/my_lib.c: Update copyright mysys/my_lock.c: Update copyright mysys/my_lockmem.c: Update copyright mysys/my_lread.c: Update copyright mysys/my_lwrite.c: Update copyright mysys/my_malloc.c: Update copyright mysys/my_messnc.c: Update copyright mysys/my_mkdir.c: Update copyright mysys/my_net.c: Update copyright mysys/my_once.c: Update copyright mysys/my_open.c: Update copyright mysys/my_pread.c: Update copyright mysys/my_pthread.c: Update copyright mysys/my_quick.c: Update copyright mysys/my_read.c: Update copyright mysys/my_realloc.c: Update copyright mysys/my_redel.c: Update copyright mysys/my_rename.c: Update copyright mysys/my_seek.c: Update copyright mysys/my_static.c: Update copyright mysys/my_static.h: Update copyright mysys/my_symlink.c: Update copyright mysys/my_symlink2.c: Update copyright mysys/my_tempnam.c: Update copyright mysys/my_thr_init.c: Update copyright mysys/my_vsnprintf.c: Update copyright mysys/my_wincond.c: Update copyright mysys/my_winthread.c: Update copyright mysys/my_write.c: Update copyright mysys/mysys_priv.h: Update copyright mysys/ptr_cmp.c: Update copyright mysys/queues.c: Update copyright mysys/raid.cc: Update copyright mysys/safemalloc.c: Update copyright mysys/string.c: Update copyright mysys/test_charset.c: Update copyright mysys/test_dir.c: Update copyright mysys/test_fn.c: Update copyright mysys/testhash.c: Update copyright mysys/thr_alarm.c: Update copyright mysys/thr_lock.c: Update copyright mysys/thr_mutex.c: Update copyright mysys/thr_rwlock.c: Update copyright mysys/tree.c: Update copyright mysys/typelib.c: Update copyright pstack/debug.c: Update copyright pstack/debug.h: Update copyright pstack/demangle.h: Update copyright pstack/ieee.c: Update copyright pstack/ieee.h: Update copyright pstack/pstack.c: Update copyright readline/bind.c: Cleanup empty lines readline/complete.c: Cleanup empty lines readline/display.c: Cleanup empty lines readline/funmap.c: Cleanup empty lines readline/histexpand.c: Cleanup empty lines readline/histfile.c: Cleanup empty lines readline/history.c: Cleanup empty lines readline/history.h: Cleanup empty lines readline/input.c: Cleanup empty lines readline/kill.c: Cleanup empty lines readline/readline.c: Cleanup empty lines readline/readline.h: Cleanup empty lines readline/vi_mode.c: Cleanup empty lines sql/cache_manager.cc: Update copyright sql/cache_manager.h: Update copyright sql/convert.cc: Update copyright sql/custom_conf.h: Update copyright sql/derror.cc: Update copyright sql/field.cc: Update copyright sql/field.h: Update copyright sql/field_conv.cc: Update copyright sql/filesort.cc: Update copyright sql/frm_crypt.cc: Update copyright sql/ha_berkeley.cc: Update copyright sql/ha_heap.cc: Update copyright sql/ha_heap.h: Update copyright sql/ha_innobase.cc: Update copyright sql/ha_isam.cc: Update copyright sql/ha_isam.h: Update copyright sql/ha_isammrg.cc: Update copyright sql/ha_isammrg.h: Update copyright sql/ha_myisam.cc: Update copyright sql/handler.cc: Update copyright sql/hash_filo.cc: Update copyright sql/hash_filo.h: Update copyright sql/hostname.cc: Update copyright sql/init.cc: Update copyright sql/item.cc: Update copyright sql/item.h: Update copyright sql/item_buff.cc: Update copyright sql/item_cmpfunc.cc: Update copyright sql/item_cmpfunc.h: Update copyright sql/item_create.cc: Update copyright sql/item_create.h: Update copyright sql/item_func.cc: Update copyright sql/item_strfunc.cc: Update copyright sql/item_sum.cc: Update copyright sql/item_sum.h: Update copyright sql/item_timefunc.cc: Update copyright sql/item_timefunc.h: Update copyright sql/item_uniq.cc: Update copyright sql/item_uniq.h: Update copyright sql/key.cc: Update copyright sql/lex_symbol.h: Update copyright sql/lock.cc: Update copyright sql/log.cc: Update copyright sql/log_event.cc: Update copyright sql/log_event.h: Update copyright sql/matherr.c: Update copyright sql/mf_iocache.cc: Update copyright sql/mini_client.cc: Update copyright sql/mini_client.h: Update copyright sql/my_lock.c: Update copyright sql/mysqld.cc: Update copyright sql/net_pkg.cc: Update copyright sql/net_serv.cc: Update copyright sql/opt_sum.cc: Update copyright sql/password.c: Update copyright sql/procedure.cc: Update copyright sql/procedure.h: Update copyright sql/records.cc: Update copyright sql/repl_failsafe.cc: Update copyright sql/slave.cc: Update copyright sql/slave.h: Update copyright sql/sql_acl.cc: Update copyright sql/sql_acl.h: Update copyright sql/sql_analyse.cc: Update copyright sql/sql_analyse.h: Update copyright sql/sql_base.cc: Update copyright sql/sql_cache.cc: Update copyright sql/sql_class.cc: Update copyright sql/sql_class.h: Update copyright sql/sql_crypt.cc: Update copyright sql/sql_crypt.h: Update copyright sql/sql_db.cc: Update copyright sql/sql_delete.cc: Update copyright sql/sql_handler.cc: Update copyright sql/sql_insert.cc: Update copyright sql/sql_lex.cc: Update copyright sql/sql_lex.h: Update copyright sql/sql_list.cc: Update copyright sql/sql_list.h: Update copyright sql/sql_load.cc: Update copyright sql/sql_map.cc: Update copyright sql/sql_map.h: Update copyright sql/sql_parse.cc: Update copyright sql/sql_rename.cc: Update copyright sql/sql_repl.cc: Update copyright sql/sql_select.h: Update copyright sql/sql_string.cc: Update copyright sql/sql_string.h: Update copyright sql/sql_table.cc: Update copyright sql/sql_test.cc: Update copyright sql/sql_udf.cc: Update copyright sql/sql_udf.h: Update copyright sql/stacktrace.c: Update copyright sql/structs.h: Update copyright sql/table.cc: Update copyright sql/table.h: Update copyright sql/thr_malloc.cc: Update copyright sql/time.cc: Update copyright sql/udf_example.cc: Update copyright sql/uniques.cc: Update copyright sql/unireg.cc: Update copyright sql/unireg.h: Update copyright strings/atof.c: Update copyright strings/bchange.c: Update copyright strings/bcmp.c: Update copyright strings/bcopy-duff.c: Update copyright strings/bfill.c: Update copyright strings/bmove.c: Update copyright strings/bmove512.c: Update copyright strings/bmove_upp.c: Update copyright strings/bzero.c: Update copyright strings/conf_to_src.c: Update copyright strings/ctype-big5.c: Update copyright strings/ctype-czech.c: Update copyright strings/ctype-euc_kr.c: Update copyright strings/ctype-gb2312.c: Update copyright strings/ctype-gbk.c: Update copyright strings/ctype-latin1_de.c: Update copyright strings/ctype-sjis.c: Update copyright strings/ctype-tis620.c: Update copyright strings/ctype-ujis.c: Update copyright strings/ctype.c: Update copyright strings/do_ctype.c: Update copyright strings/int2str.c: Update copyright strings/is_prefix.c: Update copyright strings/llstr.c: Update copyright strings/longlong2str.c: Update copyright strings/memcmp.c: Update copyright strings/memcpy.c: Update copyright strings/memset.c: Update copyright strings/r_strinstr.c: Update copyright strings/str2int.c: Update copyright strings/str_test.c: Update copyright strings/strappend.c: Update copyright strings/strcat.c: Update copyright strings/strcend.c: Update copyright strings/strchr.c: Update copyright strings/strcmp.c: Update copyright strings/strcont.c: Update copyright strings/strend.c: Update copyright strings/strfill.c: Update copyright strings/strings-not-used.h: Update copyright strings/strinstr.c: Update copyright strings/strlen.c: Update copyright strings/strmake.c: Update copyright strings/strmov.c: Update copyright strings/strnlen.c: Update copyright strings/strnmov.c: Update copyright strings/strrchr.c: Update copyright strings/strstr.c: Update copyright strings/strto.c: Update copyright strings/strtol.c: Update copyright strings/strtoll.c: Update copyright strings/strtoul.c: Update copyright strings/strtoull.c: Update copyright strings/strxmov.c: Update copyright strings/strxnmov.c: Update copyright strings/t_ctype.h: Update copyright strings/udiv.c: Update copyright tools/mysqlmanager.c: Update copyright vio/test-ssl.c: Update copyright vio/test-sslclient.c: Update copyright vio/test-sslserver.c: Update copyright vio/vio.c: Update copyright vio/viosocket.c: Update copyright vio/viossl.c: Update copyright vio/viosslfactories.c: Update copyright vio/viotest-ssl.c: Update copyright
2001-12-06 13:10:51 +01:00
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1335
USA
*/
/*
Handling of my_bitmap_map (ulonglong) arrays as large bitmaps.
2003-10-30 19:17:57 +01:00
API limitations (or, rather asserted safety assumptions,
to encourage correct programming)
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
* the internal storage is a set of 64 bit words
2005-07-18 13:31:02 +02:00
* the number of bits specified in creation can be any number > 0
New myisamchk option --sort-recover Allow delete of crashed MyISAM tables Fixed bug when BLOB was first part of key Fixed bug when using result from CASE in GROUP BY Fixed core-dump bug in monthname() Optimized calling of check_db_name() Docs/manual.texi: Added more information about myisamchk client/mysqladmin.c: Added error message for CREATE database and fixed possible overflow bug include/myisam.h: New myisamchk option --sort-recover libmysql/libmysql.c: Removed commented code Don't define getpwuid (breaks on SCO 3.2) myisam/mi_check.c: Fixed (new) bug when using --recover --optimize myisam/mi_delete_table.c: Allow delete of crashed tables myisam/mi_key.c: Fixed bug when BLOB was first part of key myisam/myisamchk.c: New myisamchk option --sort-recover mysql-test/r/case.result: New test cases to check for reported bugs mysql-test/r/func_time.result: New test cases to check for reported bugs mysql-test/r/type_blob.result: New test cases to check for reported bugs mysql-test/r/type_datetime.result: New test cases to check for reported bugs mysql-test/t/case.test: New test cases to check for reported bugs mysql-test/t/func_time.test: New test cases to check for reported bugs mysql-test/t/type_blob.test: New test cases to check for reported bugs mysql-test/t/type_datetime.test: New test cases to check for reported bugs mysys/my_bitmap.c: Optimize sql-bench/limits/ms-sql.cfg: Updated limits sql/item_cmpfunc.cc: Fixed bug when using result from CASE in GROUP BY sql/item_cmpfunc.h: Fixed bug when using result from CASE in GROUP BY sql/item_timefunc.cc: Fixed core-dump bug in monthname() sql/sql_db.cc: Optimized calling of check_db_name() sql/sql_parse.cc: Optimized calling of check_db_name() sql/table.cc: Fixed typo
2001-01-31 03:47:25 +01:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
Implementation notes:
* MY_BITMAP includes a pointer, last_word_ptr, to the last word.
The implication is that if one copies bitmaps to another memory
location, one has to call create_last_bit_mask() on the bitmap to
fix the internal pointer.
* The not used part of a the last word should always be 0.
This avoids special handling of the last bitmap in several cases.
This is checked for most calls to bitmap functions.
New myisamchk option --sort-recover Allow delete of crashed MyISAM tables Fixed bug when BLOB was first part of key Fixed bug when using result from CASE in GROUP BY Fixed core-dump bug in monthname() Optimized calling of check_db_name() Docs/manual.texi: Added more information about myisamchk client/mysqladmin.c: Added error message for CREATE database and fixed possible overflow bug include/myisam.h: New myisamchk option --sort-recover libmysql/libmysql.c: Removed commented code Don't define getpwuid (breaks on SCO 3.2) myisam/mi_check.c: Fixed (new) bug when using --recover --optimize myisam/mi_delete_table.c: Allow delete of crashed tables myisam/mi_key.c: Fixed bug when BLOB was first part of key myisam/myisamchk.c: New myisamchk option --sort-recover mysql-test/r/case.result: New test cases to check for reported bugs mysql-test/r/func_time.result: New test cases to check for reported bugs mysql-test/r/type_blob.result: New test cases to check for reported bugs mysql-test/r/type_datetime.result: New test cases to check for reported bugs mysql-test/t/case.test: New test cases to check for reported bugs mysql-test/t/func_time.test: New test cases to check for reported bugs mysql-test/t/type_blob.test: New test cases to check for reported bugs mysql-test/t/type_datetime.test: New test cases to check for reported bugs mysys/my_bitmap.c: Optimize sql-bench/limits/ms-sql.cfg: Updated limits sql/item_cmpfunc.cc: Fixed bug when using result from CASE in GROUP BY sql/item_cmpfunc.h: Fixed bug when using result from CASE in GROUP BY sql/item_timefunc.cc: Fixed core-dump bug in monthname() sql/sql_db.cc: Optimized calling of check_db_name() sql/sql_parse.cc: Optimized calling of check_db_name() sql/table.cc: Fixed typo
2001-01-31 03:47:25 +01:00
TODO:
2011-07-02 22:08:51 +02:00
Make assembler thread safe versions of these using test-and-set instructions
2005-07-18 13:31:02 +02:00
Original version created by Sergei Golubchik 2001 - 2004.
New version written and test program added and some changes to the interface
2011-07-02 22:08:51 +02:00
was made by Mikael Ronstrom 2005, with assistance of Tomas Ulin and Mats
2005-07-18 13:31:02 +02:00
Kindahl.
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
Updated to 64 bits and use my_find_first_bit() to speed up
bitmap_get_next_set() by Monty in 2024
*/
#include "mysys_priv.h"
#include <my_bitmap.h>
#include <m_string.h>
Moved a lot of old bug fixes and safe cleanups from Maria 5.1 tree to 5.1 - Reserver namespace and place in frm for TABLE_CHECKSUM and PAGE_CHECKSUM create options - Added syncing of directory when creating .frm files - Portability fixes - Added missing cast that could cause bugs - Code cleanups - Made some bit functions inline - Moved things out of myisam.h to my_handler.h to make them more accessable - Renamed some myisam variables and defines to make them more globaly usable (as they are used outside of MyISAM) - Fixed bugs in error conditions - Use compiler time asserts instead of run time - Fixed indentation HA_EXTRA_PREPARE_FOR_DELETE -> HA_EXTRA_PREPARE_FOR_DROP as the old name was wrong (Added a define for old value to ensure we don't break any old code) Added HA_EXTRA_PREPARE_FOR_RENAME as a signal for rename (before we used a DROP signal which is wrong) - Initialize error messages early to get better errors when mysqld or an engine fails to start - Fix windows bug that query_performance_frequency was not initialized if registry code failed - thread_stack -> my_thread_stack_size BitKeeper/etc/ignore: added libmysqld/scheduler.cc libmysqld/sql_connect.cc libmysqld/sql_tablespace.cc include/Makefile.am: Added my_bit.h include/m_string.h: Added bzero_if_purify() to simplify code include/my_base.h: Reserve options for the future Added HA_OPTION_NULL_FIELDS, HA_OPTION_PAGE_CHECKSUM, HA_CREATE_PAGE_CHECKSUM Added new error message HA_ERR_NEW_FILE Added optional new row type BLOCK_RECORD Renamed HA_EXTRA_PREPARE_FOR_DELETE to HA_EXTRA_PREPARE_FOR_DROP Added HA_EXTRA_PREARE_FOR_RENAME to inform handler we will do a rename (Added define to make things compatible until 6.0) Moved invalidator_by_filename form myisam.h include/my_dbug.h: Poirtablity fix include/my_global.h: Added helper macros STATIC_INLINE and MY_ERRPTR Added NEED_EXPLICIT_SYNC_DIR include/my_handler.h: Added missing casts Moved some constants and macros out from myisam.h to make these generally available Renamed mi_compare_text() to ha_compare_text() as this function is not myisam specific Renamed mi_portable_sizeof_char_ptr to portable_sizeof_char_ptr Added registering of handler messages for better error reporting during startup include/my_sys.h: Added my_sync_dir() and my_sync_dir_by_file() More comments Some indentation fixes Moved bit functions to my_bit.h Added prototype for crc32() include/myisam.h: Moved things from here to my_handler.h to make them more accessable libmysql/Makefile.shared: Added my_sync mysys/array.c: Fixed indentation and spelling errors Split set_dynamic() to two functions Added allocate_dynamic() as a new visiable function (no new code, only refactoring) mysys/mf_iocache.c: More DBUG mysys/mf_keycache.c: More explicite ASSERT Removed some casts Fixed indentation mysys/mf_tempfile.c: Fixed bug with possible dangling file descriptor mysys/my_atomic.c: Use compile time asserts instead of run time mysys/my_bit.c: Make most bit functions inline mysys/my_bitmap.c: Added my_bit.h mysys/my_compress.c: Fixed indentation mysys/my_create.c: Added my_sync_by_dir() mysys/my_delete.c: Added my_sync_by_dir() mysys/my_error.c: init_glob_errs() is now done in my_init() mysys/my_handler.c: mi_compare_text() -> ha_compare_text() as this is not MyISAM specific Added functions to initialize handler error messages Fixed indentation More clear usage of include files mysys/my_init.c: Added my_thread_stack_size to be used by other programs Ensure that global error messages are always initialized Fix windows bug that query_performance_frequency was not initialized if registry code failed mysys/my_open.c: More comments Removed duplicate code mysys/my_pread.c: Ensure that my_errno is set even if errno is 0 mysys/my_realloc.c: Added comment mysys/my_rename.c: Added syncing of directories mysys/my_symlink.c: Added my_sync_by_dir() mysys/my_sync.c: Added my_sync_dir() On recent Mac OS X, fcntl(F_FULLFSYNC) is recommended over fsync() (see "man fsync" on Mac OS X 10.3). my_sync_dir(): to sync a directory after a file creation/deletion/ renaming; can be called directly or via MY_SYNC_DIR in my_create/ my_delete/my_rename(). No-op except on Linux (see "man fsync" on Linux). my_sync_dir_from_file(): same as above, just more practical when the caller has a file name but no directory name ready. Should the #warning even be a #error? I mean do we want to release binaries which don't guarantee any durability? mysys/safemalloc.c: Added sf_malloc_report_allocated() (Debugging aid) sql/gen_lex_hash.cc: Remove inline for big function sql/ha_partition.cc: HA_EXTRA_PREPARE_FOR_DELETE -> HA_EXTRA_PREPARE_FOR_DROP prepare_for_delete -> prepare_for_rename() as this is the the time this function is called sql/ha_partition.h: prepare_for_delete -> prepare_for_rename() as this is the the time this function is called sql/handler.cc: ha_init_errors() is now called at startup before plugins This allows us to get better error messages sql/handler.h: Reserve enum value for Maria Add future proof enum for page checksums sql/item_func.cc: Include my_bit.h sql/lex.h: Added future proof CREATE table options sql/log.cc: Added comment sql/mysql_priv.h: thread_stack moved to mysys sql/mysqld.cc: thread_stack moved to mysys thread_stack -> my_thread_stack_size Initialize myisam key caches before plugins starts Initialize error to allow storage engine to give better error messages if init failes. Fixed indentation Group all MyISAM options together Added new status variable 'Opened_table_definitions' to allow one to monitor if table definition cache is too small Clarified some option help messages sql/opt_range.cc: Removed wrong usage of SAFE_MODE (this disabled key usage for UPDATES, which was never the intention) Removed print if total cost in a place where it didn't have any usable value sql/set_var.cc: thread_stack -> my_thread_stack sql/sql_class.cc: Intialize transaction object properly sql/sql_parse.cc: thread_stack -> my_thread_stack sql/sql_select.cc: Include my_bit.h mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr sql/sql_show.cc: Simplify handling of ha_choice variables Added future safe PAGE_CHECKSUM option Addid missing 'transactional=#' in information schema sql/sql_table.cc: HA_EXTRA_PREPARE_FOR_DELETE -> HA_EXTRA_FORCE_REOPEN when doing reopen HA_EXTRA_PREPARE_FOR_DELETE -> HA_EXTRA_PREPARE_FOR_RENAME when doing rename Removed not needed initialization sql/sql_test.cc: thread_stack -> my_thread_stack sql/sql_yacc.yy: Simplify handling of ha_choice variables Added future proof create table options TABLE_CHECKSUM=# & PAGE_CHECKSUM=# sql/table.cc: Save page_checksum in .frm sql/table.h: Added variable to hold create table option PAGE_CHECKSUM sql/unireg.cc: Added syncing of directories storage/myisam/ft_boolean_search.c: mi_compare_text() -> ha_compare_text() storage/myisam/ft_eval.c: mi_compare_text() -> ha_compare_text() storage/myisam/ft_nlq_search.c: mi_compare_text() -> ha_compare_text() storage/myisam/ft_parser.c: mi_compare_text() -> ha_compare_text() storage/myisam/ft_stopwords.c: mi_compare_text() -> ha_compare_text() storage/myisam/ft_test1.c: mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/ft_update.c: mi_compare_text() -> ha_compare_text() storage/myisam/ha_myisam.cc: Include my_bit.h storage/myisam/mi_check.c: MI_MAX_POSSIBLE_KEY_BUFF -> HA_MAX_POSSIBLE_KEY_BUFF mi_compare_text() -> ha_compare_text() Added BLOCK_RECORD to avoid compiler warnings storage/myisam/mi_checksum.c: mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/mi_create.c: MI_MAX_POSSIBLE_KEY -> HA_MAX_POSSIBLE_KEY MI_MAX_KEY_BLOCK_SIZE -> HA_MAX_KEY_BLOCK_SIZE mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/mi_dynrec.c: mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/mi_extra.c: HA_EXTRA_PREPARE_FOR_DELETE -> HA_EXTRA_PREPARE_FOR_DROP storage/myisam/mi_open.c: MI_MAX_POSSIBLE_KEY -> HA_MAX_POSSIBLE_KEY mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/mi_packrec.c: mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/mi_range.c: mi_compare_text -> ha_compare_text storage/myisam/mi_test1.c: mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/mi_test2.c: mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/mi_unique.c: mi_compare_text() -> ha_compare_text() storage/myisam/mi_write.c: mi_compare_text() -> ha_compare_text() storage/myisam/myisamchk.c: Include my_bit.h storage/myisam/myisamdef.h: Moved store_key_length_inc to handler.h storage/myisam/myisampack.c: mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisam/sp_test.c: mi_portable_sizeof_char_ptr -> portable_sizeof_char_ptr storage/myisammrg/ha_myisammrg.cc: HA_EXTRA_PREPARE_FOR_DELETE -> HA_EXTRA_PREPARE_FOR_DROP include/my_bit.h: New BitKeeper file ``include/my_bit.h''
2007-10-11 17:07:40 +02:00
#include <my_bit.h>
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
#include <my_byteorder.h>
/* Defines to check bitmaps */
#define DBUG_ASSERT_BITMAP(M) \
DBUG_ASSERT((M)->bitmap); \
DBUG_ASSERT((M)->n_bits > 0); \
DBUG_ASSERT((M)->last_word_ptr == (M)->bitmap + no_words_in_map(M)-1); \
DBUG_ASSERT((*(M)->last_word_ptr & (M)->last_bit_mask) == 0);
#define DBUG_ASSERT_BITMAP_AND_BIT(M,B) \
DBUG_ASSERT_BITMAP(M); \
DBUG_ASSERT((B) < (M)->n_bits);
#define DBUG_ASSERT_DIFFERENT_BITMAPS(M,N) \
DBUG_ASSERT_BITMAP(M); \
DBUG_ASSERT_BITMAP(N);
#define DBUG_ASSERT_IDENTICAL_BITMAPS(M,N) \
DBUG_ASSERT_BITMAP(M); \
DBUG_ASSERT_BITMAP(N); \
DBUG_ASSERT((M)->n_bits == (N)->n_bits);
/*
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
Create a mask for the usable bits on the LAST my_bitmap_map position for
a bitmap with 'bits' number of bits.
The lowest 'bits' bits are set to zero and the rest bits are set to 1.
For (bits & 63) == 0 , 0 is returned as in this case all bits in the
my_bitmap_position are significant. (This example assumes the
storage is ulonglong).
For 'bits & 63' it will return values from the series
0, 0xfffffffffffffffe,.... 0x8000000000000000
*/
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
static inline my_bitmap_map last_bit_mask(uint bits)
{
uint bits_in_last_map= (bits & (my_bitmap_map_bits-1));
return bits_in_last_map ? ~((1ULL << bits_in_last_map)-1) : 0ULL;
2005-05-12 11:20:50 +02:00
}
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/*
Get a mask of the bits that are to be considered as 'on' at location
starting with 'bits'.
This function has _inv in it's name as it's usage is invers compared
to last_bit_mask().
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
For (bits & 63) it will return values from the series
0xffffffffffffffff, 0xfffffffffffffffe,.... 0x8000000000000000
*/
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
static inline my_bitmap_map first_bit_mask_inv(uint bits)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
uint bits_in_last_map= (bits & (my_bitmap_map_bits-1));
return ~((1ULL << bits_in_last_map)-1);
2005-05-12 11:20:50 +02:00
}
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/*
Update the bitmap's last_word_ptr and last_bit_mask
Also ensure that the last world is all zero to make it
easy to find the next set bit.
Note that if n_bits is 0, then last_word_ptr will point to
bitmap (safely). The bitmap will not be usable for almost any operation.
*/
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
void create_last_bit_mask(MY_BITMAP *map)
{
my_bitmap_map mask= last_bit_mask(map->n_bits);
map->last_bit_mask= mask;
map->last_word_ptr= map->bitmap + MY_MAX(no_words_in_map(map),1) -1;
if (map->n_bits > 0)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
*map->last_word_ptr&= ~mask; /* Set not used bits to 0 */
DBUG_ASSERT_BITMAP(map);
}
}
/*
Initialize a bitmap object. All bits will be set to zero
*/
my_bool my_bitmap_init(MY_BITMAP *map, my_bitmap_map *buf, uint n_bits)
{
Fixes to get valgrind to work with jemalloc - Added MALLOC_LIBRARY variable to hold name of malloc library - Back ported valgrind related fixes from jemalloc 3.4.1 to the included jemalloc 3.3.1 - Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() to avoid clash with jemalloc 3.4.1 - Use option --soname-synonyms=somalloc=NON to valgrind when using jemalloc - Show version related variables in mysqld --help -- Added SHOW_VALUE_IN_HELP marker Increased back_log to 150 as the original value was a bit too small CMakeLists.txt: Added MALLOC_LIBRARY variable to hold name of malloc library cmake/jemalloc.cmake: Added MALLOC_LIBRARY variable to hold name of malloc library config.h.cmake: Added MALLOC_LIBRARY variable to hold name of malloc library extra/jemalloc/ChangeLog: Updates changelog extra/jemalloc/include/jemalloc/internal/arena.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/jemalloc_internal.h.in: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/private_namespace.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/tcache.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/src/arena.c: Backported valgrind fixes from jemalloc 3.4.1 include/my_bitmap.h: Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() to avoid clash with jemalloc 3.4.1 mysql-test/mysql-test-run.pl: Use option --soname-synonyms=somalloc=NON to valgrind when using jemalloc mysql-test/valgrind.supp: Supression of memory leak in OpenSuse 12.3 mysys/my_bitmap.c: Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() sql/ha_ndbcluster_binlog.cc: Renames sql/ha_ndbcluster_cond.h: Renames sql/ha_partition.cc: Renames sql/handler.cc: Renames sql/item_subselect.cc: Renames sql/log_event.cc: Renames sql/log_event_old.cc: Renames sql/mysqld.cc: Renames Show version related variables in mysqld --help sql/opt_range.cc: Renames sql/opt_table_elimination.cc: Renames sql/partition_info.cc: Renames sql/rpl_injector.h: Renames sql/set_var.h: Renames sql/slave.cc: Renames sql/sql_bitmap.h: Renames sql/sql_insert.cc: Renames sql/sql_lex.h: Renames sql/sql_parse.cc: Renames sql/sql_partition.cc: Renames sql/sql_select.cc: Renames sql/sql_show.cc: Renames sql/sql_update.cc: Renames sql/sys_vars.cc: Show version related variables in mysqld --help sql/sys_vars.h: Added SHOW_VALUE_IN_HELP marker for variables that should be shown in --help sql/table.cc: Renames sql/table.h: Removed not used bitmap_init_value storage/connect/ha_connect.cc: Removed compiler warning storage/maria/ma_open.c: Renames unittest/mysys/bitmap-t.c: Renames
2014-01-02 10:19:19 +01:00
DBUG_ENTER("my_bitmap_init");
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
2005-06-02 18:40:25 +02:00
if (!buf)
{
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
uint size_in_bytes= bitmap_buffer_size(n_bits);
if (!(buf= (my_bitmap_map*) my_malloc(key_memory_MY_BITMAP_bitmap,
size_in_bytes, MYF(MY_WME))))
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
{
map->bitmap= 0;
DBUG_RETURN(1);
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
}
map->bitmap_allocated= 1;
2003-10-24 22:44:48 +02:00
}
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
else
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
map->bitmap_allocated= 0;
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
2005-06-02 18:40:25 +02:00
map->bitmap= buf;
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
map->n_bits= n_bits;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
create_last_bit_mask(map);
2005-05-12 11:20:50 +02:00
bitmap_clear_all(map);
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
DBUG_RETURN(0);
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
Fixes to get valgrind to work with jemalloc - Added MALLOC_LIBRARY variable to hold name of malloc library - Back ported valgrind related fixes from jemalloc 3.4.1 to the included jemalloc 3.3.1 - Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() to avoid clash with jemalloc 3.4.1 - Use option --soname-synonyms=somalloc=NON to valgrind when using jemalloc - Show version related variables in mysqld --help -- Added SHOW_VALUE_IN_HELP marker Increased back_log to 150 as the original value was a bit too small CMakeLists.txt: Added MALLOC_LIBRARY variable to hold name of malloc library cmake/jemalloc.cmake: Added MALLOC_LIBRARY variable to hold name of malloc library config.h.cmake: Added MALLOC_LIBRARY variable to hold name of malloc library extra/jemalloc/ChangeLog: Updates changelog extra/jemalloc/include/jemalloc/internal/arena.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/jemalloc_internal.h.in: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/private_namespace.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/tcache.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/src/arena.c: Backported valgrind fixes from jemalloc 3.4.1 include/my_bitmap.h: Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() to avoid clash with jemalloc 3.4.1 mysql-test/mysql-test-run.pl: Use option --soname-synonyms=somalloc=NON to valgrind when using jemalloc mysql-test/valgrind.supp: Supression of memory leak in OpenSuse 12.3 mysys/my_bitmap.c: Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() sql/ha_ndbcluster_binlog.cc: Renames sql/ha_ndbcluster_cond.h: Renames sql/ha_partition.cc: Renames sql/handler.cc: Renames sql/item_subselect.cc: Renames sql/log_event.cc: Renames sql/log_event_old.cc: Renames sql/mysqld.cc: Renames Show version related variables in mysqld --help sql/opt_range.cc: Renames sql/opt_table_elimination.cc: Renames sql/partition_info.cc: Renames sql/rpl_injector.h: Renames sql/set_var.h: Renames sql/slave.cc: Renames sql/sql_bitmap.h: Renames sql/sql_insert.cc: Renames sql/sql_lex.h: Renames sql/sql_parse.cc: Renames sql/sql_partition.cc: Renames sql/sql_select.cc: Renames sql/sql_show.cc: Renames sql/sql_update.cc: Renames sql/sys_vars.cc: Show version related variables in mysqld --help sql/sys_vars.h: Added SHOW_VALUE_IN_HELP marker for variables that should be shown in --help sql/table.cc: Renames sql/table.h: Removed not used bitmap_init_value storage/connect/ha_connect.cc: Removed compiler warning storage/maria/ma_open.c: Renames unittest/mysys/bitmap-t.c: Renames
2014-01-02 10:19:19 +01:00
void my_bitmap_free(MY_BITMAP *map)
{
Fixes to get valgrind to work with jemalloc - Added MALLOC_LIBRARY variable to hold name of malloc library - Back ported valgrind related fixes from jemalloc 3.4.1 to the included jemalloc 3.3.1 - Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() to avoid clash with jemalloc 3.4.1 - Use option --soname-synonyms=somalloc=NON to valgrind when using jemalloc - Show version related variables in mysqld --help -- Added SHOW_VALUE_IN_HELP marker Increased back_log to 150 as the original value was a bit too small CMakeLists.txt: Added MALLOC_LIBRARY variable to hold name of malloc library cmake/jemalloc.cmake: Added MALLOC_LIBRARY variable to hold name of malloc library config.h.cmake: Added MALLOC_LIBRARY variable to hold name of malloc library extra/jemalloc/ChangeLog: Updates changelog extra/jemalloc/include/jemalloc/internal/arena.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/jemalloc_internal.h.in: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/private_namespace.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/include/jemalloc/internal/tcache.h: Backported valgrind fixes from jemalloc 3.4.1 extra/jemalloc/src/arena.c: Backported valgrind fixes from jemalloc 3.4.1 include/my_bitmap.h: Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() to avoid clash with jemalloc 3.4.1 mysql-test/mysql-test-run.pl: Use option --soname-synonyms=somalloc=NON to valgrind when using jemalloc mysql-test/valgrind.supp: Supression of memory leak in OpenSuse 12.3 mysys/my_bitmap.c: Renamed bitmap_init() and bitmap_free() to my_bitmap_init() and my_bitmap_free() sql/ha_ndbcluster_binlog.cc: Renames sql/ha_ndbcluster_cond.h: Renames sql/ha_partition.cc: Renames sql/handler.cc: Renames sql/item_subselect.cc: Renames sql/log_event.cc: Renames sql/log_event_old.cc: Renames sql/mysqld.cc: Renames Show version related variables in mysqld --help sql/opt_range.cc: Renames sql/opt_table_elimination.cc: Renames sql/partition_info.cc: Renames sql/rpl_injector.h: Renames sql/set_var.h: Renames sql/slave.cc: Renames sql/sql_bitmap.h: Renames sql/sql_insert.cc: Renames sql/sql_lex.h: Renames sql/sql_parse.cc: Renames sql/sql_partition.cc: Renames sql/sql_select.cc: Renames sql/sql_show.cc: Renames sql/sql_update.cc: Renames sql/sys_vars.cc: Show version related variables in mysqld --help sql/sys_vars.h: Added SHOW_VALUE_IN_HELP marker for variables that should be shown in --help sql/table.cc: Renames sql/table.h: Removed not used bitmap_init_value storage/connect/ha_connect.cc: Removed compiler warning storage/maria/ma_open.c: Renames unittest/mysys/bitmap-t.c: Renames
2014-01-02 10:19:19 +01:00
DBUG_ENTER("my_bitmap_free");
if (map->bitmap)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
if (map->bitmap_allocated)
my_free(map->bitmap);
map->bitmap=0;
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
DBUG_VOID_RETURN;
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
Name resolution context added (BUG#6443) include/my_bitmap.h: new bitmap operation mysql-test/r/view.result: added warnings Correct inserting data check (absence of default value) for view underlying tables (BUG#6443) mysql-test/t/view.test: Correct inserting data check (absence of default value) for view underlying tables (BUG#6443) mysys/my_bitmap.c: new bitmap operation sql/field.h: index of field in table added sql/item.cc: Name resolution context added table list removed from fix_fields() arguments sql/item.h: Name resolution context added table list removed from fix_fields() arguments sql/item_cmpfunc.cc: table list removed from fix_fields() arguments sql/item_cmpfunc.h: table list removed from fix_fields() arguments sql/item_func.cc: table list removed from fix_fields() arguments sql/item_func.h: table list removed from fix_fields() arguments sql/item_row.cc: table list removed from fix_fields() arguments sql/item_row.h: table list removed from fix_fields() arguments sql/item_strfunc.cc: fixed server crash on NULL argument sql/item_strfunc.h: table list removed from fix_fields() arguments sql/item_subselect.cc: table list removed from fix_fields() arguments sql/item_subselect.h: table list removed from fix_fields() arguments sql/item_sum.cc: table list removed from fix_fields() arguments sql/item_sum.h: table list removed from fix_fields() arguments sql/item_timefunc.cc: table list removed from fix_fields() arguments sql/item_timefunc.h: table list removed from fix_fields() arguments sql/item_uniq.h: table list removed from fix_fields() arguments sql/log_event.cc: Name resolution context added sql/log_event.h: Name resolution context added sql/mysql_priv.h: Name resolution context added sql/set_var.cc: table list removed from fix_fields() arguments sql/share/errmsg.txt: new error message sql/sp.cc: Name resolution context added sql/sp_head.cc: table list removed from fix_fields() arguments sql/sp_head.h: Name resolution context added sql/sql_base.cc: table list removed from fix_fields() arguments Name resolution context added sql/sql_class.cc: renamed variable sql/sql_delete.cc: Name resolution context added sql/sql_derived.cc: Name resolution context added sql/sql_do.cc: table list removed from fix_fields() arguments sql/sql_handler.cc: Name resolution context added sql/sql_help.cc: Name resolution context added sql/sql_insert.cc: Name resolution context added table list removed from fix_fields() arguments sql/sql_lex.cc: Name resolution context added sql/sql_lex.h: removed resolve mode (information stored into name resolution context) sql/sql_load.cc: table list removed from fix_fields() arguments sql/sql_olap.cc: Name resolution context added sql/sql_parse.cc: Name resolution context added sql/sql_prepare.cc: table list removed from fix_fields() arguments sql/sql_select.cc: table list removed from fix_fields() arguments sql/sql_show.cc: Name resolution context added sql/sql_trigger.cc: table list removed from fix_fields() arguments sql/sql_udf.h: table list removed from fix_fields() arguments sql/sql_union.cc: Name resolution context added sql/sql_update.cc: Name resolution context added sql/sql_view.cc: Name resolution context added sql/sql_view.h: table list removed from fix_fields() arguments sql/sql_yacc.yy: Name resolution context added sql/table.cc: Name resolution context added merged view processing moved sql/table.h: merged view processing moved
2005-07-01 06:05:42 +02:00
/*
test if bit already set and set it if it was not (thread unsafe method)
SYNOPSIS
bitmap_fast_test_and_set()
MAP bit map struct
BIT bit number
RETURN
0 bit was not set
!=0 bit was set
*/
my_bool bitmap_fast_test_and_set(MY_BITMAP *map, uint bitmap_bit)
{
my_bitmap_map *value, bit, res;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP_AND_BIT(map, bitmap_bit);
value= map->bitmap + (bitmap_bit/my_bitmap_map_bits);
bit= 1ULL << (bitmap_bit & (my_bitmap_map_bits-1));
res= *value & bit;
Fixed compiler warnings detected by option -Wshadow and -Wunused: - Removed not used variables and functions - Added #ifdef around code that is not used - Renamed variables and functions to avoid conflicts - Removed some not used arguments Fixed some class/struct warnings in ndb Added define IS_LONGDATA() to simplify code in libmysql.c I did run gcov on the changes and added 'purecov' comments on almost all lines that was not just variable name changes BUILD/SETUP.sh: Added printing of unused functions and variables. Made it easy to test compiling with -Wshadow BUILD/compile-pentium-gcov: Added warnings Mark binary with -gcov client/mysql.cc: Fixed warnings found with gcc -Wshadow client/mysql_upgrade.c: Fixed warnings found with gcc -Wshadow client/mysqlbinlog.cc: Fixed warnings found with gcc -Wshadow client/mysqldump.c: Fixed warnings found with gcc -Wshadow client/mysqltest.c: Fixed warnings found with gcc -Wshadow client/sql_string.cc: Fixed warnings found with gcc -Wshadow Merged with sql/sql_string.cc client/sql_string.h: Fixed warnings found with gcc -Wshadow Merged with sql/sql_string.h cmd-line-utils/readline/display.c: Fixed compiler warning cmd-line-utils/readline/histexpand.c: Fixed warnings found with gcc -Wshadow cmd-line-utils/readline/input.c: Fixed warnings found with gcc -Wshadow cmd-line-utils/readline/text.c: Fixed warnings found with gcc -Wshadow cmd-line-utils/readline/vi_mode.c: Fixed warnings found with gcc -Wshadow dbug/dbug_analyze.c: Fixed warnings found with gcc -Wshadow extra/my_print_defaults.c: Prefixed defaults_extra_file and defaults_group_suffix with 'my' to avoid conflicts with similar named local variables extra/yassl/include/buffer.hpp: Fixed compiler warnings extra/yassl/include/crypto_wrapper.hpp: Fixed compiler warnings extra/yassl/include/yassl_imp.hpp: Fixed compiler warnings extra/yassl/include/yassl_int.hpp: Fixed compiler warnings extra/yassl/src/crypto_wrapper.cpp: Fixed compiler warnings extra/yassl/taocrypt/benchmark/benchmark.cpp: Fixed warnings found with gcc -Wshadow extra/yassl/taocrypt/include/algebra.hpp: Fixed compiler warnings extra/yassl/taocrypt/include/des.hpp: Fixed compiler warnings extra/yassl/taocrypt/include/hash.hpp: Fixed compiler warnings extra/yassl/taocrypt/include/hmac.hpp: Fixed compiler warnings extra/yassl/taocrypt/include/modarith.hpp: Fixed compiler warnings extra/yassl/taocrypt/include/modes.hpp: Fixed compiler warnings extra/yassl/taocrypt/include/rsa.hpp: Fixed compiler warnings extra/yassl/taocrypt/mySTL/list.hpp: Fixed compiler warnings extra/yassl/taocrypt/src/aes.cpp: Fixed compiler warnings extra/yassl/taocrypt/src/algebra.cpp: Fixed compiler warnings extra/yassl/taocrypt/src/asn.cpp: Fixed compiler warnings extra/yassl/taocrypt/test/test.cpp: Fixed compiler warnings extra/yassl/testsuite/testsuite.cpp: Fixed compiler warnings include/m_ctype.h: Fixed warnings found with gcc -Wshadow include/my_pthread.h: Fixed warnings found with gcc -Wshadow include/my_sys.h: Fixed warnings found with gcc -Wshadow include/my_time.h: Fixed warnings found with gcc -Wshadow include/mysql.h: Fixed warnings found with gcc -Wshadow Added define IS_LONGDATA() to simplify code in libmysql.c libmysql/libmysql.c: Fixed warnings found with gcc -Wshadow (Mostly replaced bind -> my_bind and time -> my_time) libmysqld/lib_sql.cc: Removed not used variables and labels myisam/ft_boolean_search.c: Fixed warnings found with gcc -Wshadow myisam/mi_open.c: Fixed warnings found with gcc -Wshadow myisam/mi_search.c: Fixed warnings found with gcc -Wshadow myisam/mi_unique.c: Fixed compiler warning myisam/myisampack.c: Fixed warnings found with gcc -Wshadow myisam/rt_index.c: Remove not used variables myisam/sort.c: Fixed warnings found with gcc -Wshadow mysql-test/r/mysqlcheck.result: Remove databases and tables possible left by previous test mysql-test/r/mysqltest.result: New test results mysql-test/t/mysql.test: Coverage tests mysql-test/t/mysqlbinlog.test: Coverage tests mysql-test/t/mysqlcheck.test: Remove databases and tables possible left by previous test mysql-test/t/mysqltest.test: Coverage tests mysys/default.c: Prefixed defaults_file, defaults_group_suffix and defaults_extra_file with 'my' to avoid conflicts with local variables in some functions mysys/mf_iocache2.c: Fixed warnings found with gcc -Wshadow mysys/mf_keycache.c: Fixed warnings found with gcc -Wshadow mysys/my_bitmap.c: Fixed warnings found with gcc -Wshadow mysys/sha1.c: Fixed warnings found with gcc -Wshadow ndb/include/kernel/signaldata/ArbitSignalData.hpp: Fixed compiler warning ndb/include/kernel/signaldata/DictTabInfo.hpp: Fixed compiler warnings ndb/include/ndbapi/NdbReceiver.hpp: Fixed warnings found with gcc -Wshadow ndb/include/transporter/TransporterDefinitions.hpp: Fixed compiler warning ndb/include/util/InputStream.hpp: Fixed compiler warning ndb/include/util/OutputStream.hpp: Fixed compiler warning ndb/include/util/SimpleProperties.hpp: Fixed compiler warning ndb/include/util/SocketAuthenticator.hpp: Fixed compiler warning ndb/include/util/SocketServer.hpp: Fixed compiler warning ndb/src/common/mgmcommon/ConfigRetriever.cpp: Fixed warnings found with gcc -Wshadow ndb/src/common/portlib/NdbTick.c: Fixed warnings found with gcc -Wshadow ndb/src/common/transporter/SHM_Transporter.cpp: Fixed warnings found with gcc -Wshadow ndb/src/common/transporter/TCP_Transporter.cpp: Fixed warnings found with gcc -Wshadow ndb/src/common/transporter/TCP_Transporter.hpp: Fixed compiler warning ndb/src/common/transporter/Transporter.cpp: Removed not used variable ndb/src/common/transporter/TransporterRegistry.cpp: Removed not used variable ndb/src/common/util/Bitmask.cpp: Moved function to avoid warnings of not used function ndb/src/common/util/ConfigValues.cpp: Fixed warnings found with gcc -Wshadow ndb/src/common/util/File.cpp: Fixed warnings found with gcc -Wshadow ndb/src/common/util/Properties.cpp: Fixed warnings found with gcc -Wshadow ndb/src/common/util/SocketClient.cpp: Fixed wrong return value ndb/src/common/util/random.c: Fixed warnings found with gcc -Wshadow ndb/src/common/util/socket_io.cpp: Fixed warnings found with gcc -Wshadow ndb/src/cw/cpcd/APIService.cpp: Removed not used variable ndb/src/cw/cpcd/main.cpp: Removed not used variables ndb/src/kernel/blocks/cmvmi/Cmvmi.cpp: Fixed compiler warnings ndb/src/kernel/blocks/dbdict/Dbdict.cpp: Removed not used variables ndb/src/kernel/blocks/dbdict/Dbdict.hpp: Fixed compiler warnings ndb/src/kernel/blocks/dbdih/Dbdih.hpp: Fixed compiler warnings ndb/src/kernel/blocks/dblqh/Dblqh.hpp: Fixed compiler warnings ndb/src/kernel/blocks/dblqh/DblqhMain.cpp: Removed not used variables ndb/src/kernel/blocks/dbtc/Dbtc.hpp: Fixed compiler warnings ndb/src/kernel/blocks/dbtup/Dbtup.hpp: Fixed compiler warnings ndb/src/kernel/blocks/dbtup/DbtupScan.cpp: Removed not used variable ndb/src/kernel/blocks/dbtux/DbtuxNode.cpp: Removed not used variables ndb/src/kernel/blocks/dbtux/DbtuxScan.cpp: Removed not used variables ndb/src/kernel/blocks/dbtux/DbtuxTree.cpp: Removed not used variables ndb/src/kernel/blocks/ndbcntr/Ndbcntr.hpp: Fixed compiler warnings ndb/src/kernel/blocks/ndbcntr/NdbcntrMain.cpp: Removed not used variables ndb/src/kernel/blocks/ndbfs/AsyncFile.hpp: Fixed compiler warnings ndb/src/kernel/blocks/qmgr/Qmgr.hpp: Fixed compiler warnings ndb/src/kernel/blocks/qmgr/QmgrMain.cpp: Removed not used variables ndb/src/kernel/blocks/suma/Suma.cpp: Removed not used variables ndb/src/kernel/blocks/suma/Suma.hpp: Fixed compiler warnings ndb/src/kernel/vm/MetaData.hpp: Fixed compiler warnings ndb/src/mgmapi/LocalConfig.cpp: Fixed warnings found with gcc -Wshadow ndb/src/mgmapi/mgmapi.cpp: Fixed warnings found with gcc -Wshadow ndb/src/mgmclient/CommandInterpreter.cpp: Removed not used variables ndb/src/mgmsrv/ConfigInfo.cpp: Fixed warnings found with gcc -Wshadow Removed not used variables ndb/src/mgmsrv/ConfigInfo.hpp: Fixed warnings found with gcc -Wshadow ndb/src/mgmsrv/InitConfigFileParser.cpp: Prefixed defaults_file, defaults_group_suffix and defaults_extra_file with 'my' to avoid conflicts with local variables in some functions ndb/src/mgmsrv/MgmtSrvr.cpp: Removed not used variables and functions ndb/src/mgmsrv/MgmtSrvr.hpp: Fixed compiler warnings ndb/src/mgmsrv/Services.cpp: Removed not used variables and functions ndb/src/mgmsrv/main.cpp: Removed not used variable ndb/src/ndbapi/ClusterMgr.hpp: Fixed compiler warnings ndb/src/ndbapi/Ndb.cpp: Removed not used variables ndb/src/ndbapi/NdbBlob.cpp: Removed not used variables ndb/src/ndbapi/NdbDictionaryImpl.cpp: Removed not used variables ndb/src/ndbapi/NdbIndexOperation.cpp: Removed not used variables ndb/src/ndbapi/NdbOperationDefine.cpp: Removed not used variables ndb/src/ndbapi/NdbOperationExec.cpp: Removed not used variables ndb/src/ndbapi/NdbOperationSearch.cpp: Removed not used variables ndb/src/ndbapi/NdbScanFilter.cpp: Fixed compiler warning ndb/src/ndbapi/NdbScanOperation.cpp: Removed not used variables ndb/src/ndbapi/SignalSender.cpp: Removed not used variables ndb/src/ndbapi/ndb_cluster_connection.cpp: Removed not used variable ndb/tools/delete_all.cpp: Removed not used variable ndb/tools/desc.cpp: Removed not used variable ndb/tools/drop_index.cpp: Removed not used variable ndb/tools/drop_tab.cpp: Removed not used variable ndb/tools/listTables.cpp: Removed not used variable ndb/tools/ndb_config.cpp: Fixed warnings found with gcc -Wshadow Added missing puts(desc) ndb/tools/restore/Restore.hpp: Changed delimiter to define instead of static variable, as the static variable caused a LOT of compiler warnings Fixed compiler warning ndb/tools/restore/consumer.hpp: Fixed compiler warning ndb/tools/restore/restore_main.cpp: Fixed compiler warnings ndb/tools/select_all.cpp: Removed not used variables ndb/tools/select_count.cpp: Removed not used variable server-tools/instance-manager/commands.h: Fixed compiler warnings server-tools/instance-manager/guardian.cc: Fixed compiler warnings server-tools/instance-manager/instance_options.cc: Removed not used variables server-tools/instance-manager/mysql_connection.cc: Fixed compiler warnings server-tools/instance-manager/options.cc: Fixed compiler warnings server-tools/instance-manager/options.h: Fixed compiler warnings server-tools/instance-manager/parse.cc: Removed not used variable server-tools/instance-manager/user_map.cc: Fixed compiler warnings server-tools/instance-manager/user_map.h: Fixed compiler warnings sql/field.cc: Fixed compiler warnings sql/field.h: Fixed compiler warnings sql/filesort.cc: Fixed compiler warnings sql/ha_archive.cc: Removed table and share arguments from get_share() / free_share() to get rid of compiler warnings sql/ha_archive.h: Removed table and share arguments from get_share() / free_share() to get rid of compiler warnings sql/ha_federated.cc: Fixed compiler warnings sql/ha_heap.cc: Fixed compiler warnings sql/ha_myisam.cc: Fixed compiler warnings sql/ha_myisammrg.cc: Fixed compiler warnings sql/ha_ndbcluster.cc: Fixed compiler warnings sql/handler.cc: Fixed compiler warnings sql/item.cc: Fixed compiler warnings sql/item.h: Fixed compiler warnings new_item() -> clone_item(), to avoid a lot of warnings with variable 'new_item' el() -> element_index() sql/item_cmpfunc.cc: Fixed compiler warnings sql/item_cmpfunc.h: Fixed compiler warnings sql/item_func.cc: Fixed compiler warnings sql/item_geofunc.cc: Fixed compiler warnings sql/item_row.h: Fixed compiler warnings sql/item_strfunc.cc: Fixed compiler warnings sql/item_subselect.cc: Fixed compiler warnings sql/item_subselect.h: Fixed compiler warnings sql/item_sum.cc: Fixed compiler warnings sql/item_timefunc.cc: Fixed compiler warnings sql/log.cc: Fixed compiler warnings More comments Added #ifdef HAVE_REPLICATION sql/log_event.cc: Fixed compiler warnings sql/log_event.h: Fixed compiler warnings sql/mysql_priv.h: query_id -> global_query_id, to avoid a lot of clashes with function and class variables start_time -> server_start_time sql/mysqld.cc: Fixed compiler warnings: - Removed not used variables - Added #ifndef EMBEDDED_LIBRARY - Fixed shadow warnings sql/net_serv.cc: Fixed compiler warnings sql/opt_range.cc: range -> last_range to avoid shadow warnings Removed not used function print_rowid() sql/opt_range.h: range -> last_range to avoid shadow warnings sql/password.c: Fixed compiler warnings sql/protocol.cc: Fixed compiler warnings sql/repl_failsafe.cc: Fixed compiler warnings sql/set_var.cc: Fixed compiler warnings sql/set_var.h: type() -> show_type() Fixed compiler warnings sql/slave.cc: Fixed compiler warnings sql/sp_head.cc: Fixed compiler warnings sql/sp_head.h: Fixed compiler warnings sql/spatial.cc: Fixed compiler warnings sql/spatial.h: length() -> geom_length() to avoid compiler warnings wkb_end -> wkb_last to avoid compiler warnings with local variables named 'wkb_end' sql/sql_cache.h: Fixed compiler warnings sql/sql_class.cc: Fixed compiler warnings sql/sql_class.h: log -> log_xid() to avoid compiler warnings Fixed shadow compiler warnings sql/sql_derived.cc: Removed not used variable sql/sql_insert.cc: Fixed compiler warnings sql/sql_lex.cc: Fixed compiler warnings sql/sql_lex.h: res -> saved_error to make the meaning of the variable clear and avoid shadow warnings sql/sql_load.cc: Fixed compiler warnings sql/sql_parse.cc: Fixed compiler warnings sql/sql_prepare.cc: Fixed compiler warnings sql/sql_select.cc: Fixed compiler warnings sql/sql_show.cc: Fixed compiler warnings sql/sql_string.cc: Fixed compiler warnings sql/sql_string.h: Fixed compiler warnings sql/sql_table.cc: Fixed compiler warnings sql/sql_trigger.cc: Fixed compiler warnings sql/sql_trigger.h: table -> trigger_table to avoid warnings from local variables sql/sql_union.cc: Fixed compiler warnings (mainly res -> saved_error) sql-common/client.c: Removed not used variable sql-common/my_time.c: Removed not used variable time -> my_time sql/sql_update.cc: Removed not used variable sql/sql_view.cc: Removed not used variable sql/sql_yacc.yy: Removed not used variable sql/table.cc: Removed not used variable sql/tztime.cc: Removed not used variable sql/unireg.cc: Removed not used variable strings/ctype-bin.c: mblen -> mb_len to avoid compiler warnings with local variable mblen strings/ctype-cp932.c: Fixed compiler warnings strings/ctype-eucjpms.c: Fixed compiler warnings strings/ctype-mb.c: mblen -> mb_len to avoid compiler warnings with local variable mblen strings/ctype-simple.c: mblen -> mb_len to avoid compiler warnings with local variable mblen exp -> exponent strings/ctype-sjis.c: Fixed compiler warnings strings/ctype-uca.c: mblen -> mb_len to avoid compiler warnings with local variable mblen strings/ctype-ujis.c: Fixed compiler warnings strings/ctype-utf8.c: Fixed compiler warnings strings/decimal.c: Fixed compiler warnings strings/my_vsnprintf.c: Added comment strings/strtod.c: Fixed compiler warnings tests/mysql_client_test.c: Fixed compiler warnings (Biggest part of patch is to not get a conflict with global function 'bind')
2006-12-14 23:51:37 +01:00
*value|= bit;
return MY_TEST(res);
Name resolution context added (BUG#6443) include/my_bitmap.h: new bitmap operation mysql-test/r/view.result: added warnings Correct inserting data check (absence of default value) for view underlying tables (BUG#6443) mysql-test/t/view.test: Correct inserting data check (absence of default value) for view underlying tables (BUG#6443) mysys/my_bitmap.c: new bitmap operation sql/field.h: index of field in table added sql/item.cc: Name resolution context added table list removed from fix_fields() arguments sql/item.h: Name resolution context added table list removed from fix_fields() arguments sql/item_cmpfunc.cc: table list removed from fix_fields() arguments sql/item_cmpfunc.h: table list removed from fix_fields() arguments sql/item_func.cc: table list removed from fix_fields() arguments sql/item_func.h: table list removed from fix_fields() arguments sql/item_row.cc: table list removed from fix_fields() arguments sql/item_row.h: table list removed from fix_fields() arguments sql/item_strfunc.cc: fixed server crash on NULL argument sql/item_strfunc.h: table list removed from fix_fields() arguments sql/item_subselect.cc: table list removed from fix_fields() arguments sql/item_subselect.h: table list removed from fix_fields() arguments sql/item_sum.cc: table list removed from fix_fields() arguments sql/item_sum.h: table list removed from fix_fields() arguments sql/item_timefunc.cc: table list removed from fix_fields() arguments sql/item_timefunc.h: table list removed from fix_fields() arguments sql/item_uniq.h: table list removed from fix_fields() arguments sql/log_event.cc: Name resolution context added sql/log_event.h: Name resolution context added sql/mysql_priv.h: Name resolution context added sql/set_var.cc: table list removed from fix_fields() arguments sql/share/errmsg.txt: new error message sql/sp.cc: Name resolution context added sql/sp_head.cc: table list removed from fix_fields() arguments sql/sp_head.h: Name resolution context added sql/sql_base.cc: table list removed from fix_fields() arguments Name resolution context added sql/sql_class.cc: renamed variable sql/sql_delete.cc: Name resolution context added sql/sql_derived.cc: Name resolution context added sql/sql_do.cc: table list removed from fix_fields() arguments sql/sql_handler.cc: Name resolution context added sql/sql_help.cc: Name resolution context added sql/sql_insert.cc: Name resolution context added table list removed from fix_fields() arguments sql/sql_lex.cc: Name resolution context added sql/sql_lex.h: removed resolve mode (information stored into name resolution context) sql/sql_load.cc: table list removed from fix_fields() arguments sql/sql_olap.cc: Name resolution context added sql/sql_parse.cc: Name resolution context added sql/sql_prepare.cc: table list removed from fix_fields() arguments sql/sql_select.cc: table list removed from fix_fields() arguments sql/sql_show.cc: Name resolution context added sql/sql_trigger.cc: table list removed from fix_fields() arguments sql/sql_udf.h: table list removed from fix_fields() arguments sql/sql_union.cc: Name resolution context added sql/sql_update.cc: Name resolution context added sql/sql_view.cc: Name resolution context added sql/sql_view.h: table list removed from fix_fields() arguments sql/sql_yacc.yy: Name resolution context added sql/table.cc: Name resolution context added merged view processing moved sql/table.h: merged view processing moved
2005-07-01 06:05:42 +02:00
}
/*
test if bit already set and set it if it was not (thread safe method)
SYNOPSIS
bitmap_fast_test_and_set()
map bit map struct
bitmap_bit bit number
RETURN
0 bit was not set
!=0 bit was set
*/
my_bool bitmap_test_and_set(MY_BITMAP *map, uint bitmap_bit)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP_AND_BIT(map, bitmap_bit);
return bitmap_fast_test_and_set(map, bitmap_bit);
Name resolution context added (BUG#6443) include/my_bitmap.h: new bitmap operation mysql-test/r/view.result: added warnings Correct inserting data check (absence of default value) for view underlying tables (BUG#6443) mysql-test/t/view.test: Correct inserting data check (absence of default value) for view underlying tables (BUG#6443) mysys/my_bitmap.c: new bitmap operation sql/field.h: index of field in table added sql/item.cc: Name resolution context added table list removed from fix_fields() arguments sql/item.h: Name resolution context added table list removed from fix_fields() arguments sql/item_cmpfunc.cc: table list removed from fix_fields() arguments sql/item_cmpfunc.h: table list removed from fix_fields() arguments sql/item_func.cc: table list removed from fix_fields() arguments sql/item_func.h: table list removed from fix_fields() arguments sql/item_row.cc: table list removed from fix_fields() arguments sql/item_row.h: table list removed from fix_fields() arguments sql/item_strfunc.cc: fixed server crash on NULL argument sql/item_strfunc.h: table list removed from fix_fields() arguments sql/item_subselect.cc: table list removed from fix_fields() arguments sql/item_subselect.h: table list removed from fix_fields() arguments sql/item_sum.cc: table list removed from fix_fields() arguments sql/item_sum.h: table list removed from fix_fields() arguments sql/item_timefunc.cc: table list removed from fix_fields() arguments sql/item_timefunc.h: table list removed from fix_fields() arguments sql/item_uniq.h: table list removed from fix_fields() arguments sql/log_event.cc: Name resolution context added sql/log_event.h: Name resolution context added sql/mysql_priv.h: Name resolution context added sql/set_var.cc: table list removed from fix_fields() arguments sql/share/errmsg.txt: new error message sql/sp.cc: Name resolution context added sql/sp_head.cc: table list removed from fix_fields() arguments sql/sp_head.h: Name resolution context added sql/sql_base.cc: table list removed from fix_fields() arguments Name resolution context added sql/sql_class.cc: renamed variable sql/sql_delete.cc: Name resolution context added sql/sql_derived.cc: Name resolution context added sql/sql_do.cc: table list removed from fix_fields() arguments sql/sql_handler.cc: Name resolution context added sql/sql_help.cc: Name resolution context added sql/sql_insert.cc: Name resolution context added table list removed from fix_fields() arguments sql/sql_lex.cc: Name resolution context added sql/sql_lex.h: removed resolve mode (information stored into name resolution context) sql/sql_load.cc: table list removed from fix_fields() arguments sql/sql_olap.cc: Name resolution context added sql/sql_parse.cc: Name resolution context added sql/sql_prepare.cc: table list removed from fix_fields() arguments sql/sql_select.cc: table list removed from fix_fields() arguments sql/sql_show.cc: Name resolution context added sql/sql_trigger.cc: table list removed from fix_fields() arguments sql/sql_udf.h: table list removed from fix_fields() arguments sql/sql_union.cc: Name resolution context added sql/sql_update.cc: Name resolution context added sql/sql_view.cc: Name resolution context added sql/sql_view.h: table list removed from fix_fields() arguments sql/sql_yacc.yy: Name resolution context added sql/table.cc: Name resolution context added merged view processing moved sql/table.h: merged view processing moved
2005-07-01 06:05:42 +02:00
}
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
/*
test if bit already set and clear it if it was set(thread unsafe method)
SYNOPSIS
bitmap_fast_test_and_set()
MAP bit map struct
BIT bit number
RETURN
0 bit was not set
!=0 bit was set
*/
my_bool bitmap_fast_test_and_clear(MY_BITMAP *map, uint bitmap_bit)
{
my_bitmap_map *value, bit, res;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP_AND_BIT(map, bitmap_bit);
value= map->bitmap + (bitmap_bit/my_bitmap_map_bits);
bit= 1ULL << (bitmap_bit & (my_bitmap_map_bits-1));
res= *value & bit;
*value&= ~bit;
return MY_TEST(res);
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
}
my_bool bitmap_test_and_clear(MY_BITMAP *map, uint bitmap_bit)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP_AND_BIT(map, bitmap_bit);
return bitmap_fast_test_and_clear(map, bitmap_bit);
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
}
uint bitmap_set_next(MY_BITMAP *map)
New myisamchk option --sort-recover Allow delete of crashed MyISAM tables Fixed bug when BLOB was first part of key Fixed bug when using result from CASE in GROUP BY Fixed core-dump bug in monthname() Optimized calling of check_db_name() Docs/manual.texi: Added more information about myisamchk client/mysqladmin.c: Added error message for CREATE database and fixed possible overflow bug include/myisam.h: New myisamchk option --sort-recover libmysql/libmysql.c: Removed commented code Don't define getpwuid (breaks on SCO 3.2) myisam/mi_check.c: Fixed (new) bug when using --recover --optimize myisam/mi_delete_table.c: Allow delete of crashed tables myisam/mi_key.c: Fixed bug when BLOB was first part of key myisam/myisamchk.c: New myisamchk option --sort-recover mysql-test/r/case.result: New test cases to check for reported bugs mysql-test/r/func_time.result: New test cases to check for reported bugs mysql-test/r/type_blob.result: New test cases to check for reported bugs mysql-test/r/type_datetime.result: New test cases to check for reported bugs mysql-test/t/case.test: New test cases to check for reported bugs mysql-test/t/func_time.test: New test cases to check for reported bugs mysql-test/t/type_blob.test: New test cases to check for reported bugs mysql-test/t/type_datetime.test: New test cases to check for reported bugs mysys/my_bitmap.c: Optimize sql-bench/limits/ms-sql.cfg: Updated limits sql/item_cmpfunc.cc: Fixed bug when using result from CASE in GROUP BY sql/item_cmpfunc.h: Fixed bug when using result from CASE in GROUP BY sql/item_timefunc.cc: Fixed core-dump bug in monthname() sql/sql_db.cc: Optimized calling of check_db_name() sql/sql_parse.cc: Optimized calling of check_db_name() sql/table.cc: Fixed typo
2001-01-31 03:47:25 +01:00
{
2005-05-12 11:20:50 +02:00
uint bit_found;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP(map);
if ((bit_found= bitmap_get_first_clear(map)) != MY_BIT_NONE)
2005-05-12 11:20:50 +02:00
bitmap_set_bit(map, bit_found);
return bit_found;
New myisamchk option --sort-recover Allow delete of crashed MyISAM tables Fixed bug when BLOB was first part of key Fixed bug when using result from CASE in GROUP BY Fixed core-dump bug in monthname() Optimized calling of check_db_name() Docs/manual.texi: Added more information about myisamchk client/mysqladmin.c: Added error message for CREATE database and fixed possible overflow bug include/myisam.h: New myisamchk option --sort-recover libmysql/libmysql.c: Removed commented code Don't define getpwuid (breaks on SCO 3.2) myisam/mi_check.c: Fixed (new) bug when using --recover --optimize myisam/mi_delete_table.c: Allow delete of crashed tables myisam/mi_key.c: Fixed bug when BLOB was first part of key myisam/myisamchk.c: New myisamchk option --sort-recover mysql-test/r/case.result: New test cases to check for reported bugs mysql-test/r/func_time.result: New test cases to check for reported bugs mysql-test/r/type_blob.result: New test cases to check for reported bugs mysql-test/r/type_datetime.result: New test cases to check for reported bugs mysql-test/t/case.test: New test cases to check for reported bugs mysql-test/t/func_time.test: New test cases to check for reported bugs mysql-test/t/type_blob.test: New test cases to check for reported bugs mysql-test/t/type_datetime.test: New test cases to check for reported bugs mysys/my_bitmap.c: Optimize sql-bench/limits/ms-sql.cfg: Updated limits sql/item_cmpfunc.cc: Fixed bug when using result from CASE in GROUP BY sql/item_cmpfunc.h: Fixed bug when using result from CASE in GROUP BY sql/item_timefunc.cc: Fixed core-dump bug in monthname() sql/sql_db.cc: Optimized calling of check_db_name() sql/sql_parse.cc: Optimized calling of check_db_name() sql/table.cc: Fixed typo
2001-01-31 03:47:25 +01:00
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
MDEV-9021: MYSQLD SEGFAULTS WHEN BUILT USING --WITH-MAX-INDEXES=128 The bitmap implementation defines two template Bitmap classes. One optimized for 64-bit (default) wide bitmaps while the other is used for all other widths. In order to optimize the computations, Bitmap<64> class has defined its own member functions for bitmap operations, the other one, however, relies on mysys' bitmap implementation (mysys/my_bitmap.c). Issue 1: In case of non 64-bit Bitmap class, intersect() wrongly reset the received bitmap while initialising a new local bitmap structure (bitmap_init() clears the bitmap buffer) thus, the received bitmap was getting cleared. Fixed by initializing the local bitmap structure by using a temporary buffer and later copying the received bitmap to the initialised bitmap structure. Issue 2: The non 64-bit Bitmap class had the Iterator missing which caused compilation failure. Also added a cmake variable to hold the MAX_INDEXES value when supplied from the command prompt. (eg. cmake .. -DMAX_INDEXES=128U). Checks have been put in place to trigger build failure if MAX_INDEXES value is greater than 128. Test modifications: * Introduced include/have_max_indexes_[64|128].inc to facilitate skipping of tests for which the output differs with different MAX_INDEXES. * Introduced include/max_indexes.inc which would get modified by cmake to reflect the MAX_INDEXES value used to build the server. This file simply sets an mtr variable '$max_indexes' to show the MAX_INDEXES value, which will then be consumed by the above introduced include file. * Some tests (portions), dependent on MAX_INDEXES value, have been moved to separate test files.
2015-11-06 20:38:03 +01:00
/**
Set the specified number of bits in the bitmap buffer.
@param map [IN] Bitmap
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
@param prefix_size [IN] Number of bits to be set or (uint) ~0 for all
MDEV-9021: MYSQLD SEGFAULTS WHEN BUILT USING --WITH-MAX-INDEXES=128 The bitmap implementation defines two template Bitmap classes. One optimized for 64-bit (default) wide bitmaps while the other is used for all other widths. In order to optimize the computations, Bitmap<64> class has defined its own member functions for bitmap operations, the other one, however, relies on mysys' bitmap implementation (mysys/my_bitmap.c). Issue 1: In case of non 64-bit Bitmap class, intersect() wrongly reset the received bitmap while initialising a new local bitmap structure (bitmap_init() clears the bitmap buffer) thus, the received bitmap was getting cleared. Fixed by initializing the local bitmap structure by using a temporary buffer and later copying the received bitmap to the initialised bitmap structure. Issue 2: The non 64-bit Bitmap class had the Iterator missing which caused compilation failure. Also added a cmake variable to hold the MAX_INDEXES value when supplied from the command prompt. (eg. cmake .. -DMAX_INDEXES=128U). Checks have been put in place to trigger build failure if MAX_INDEXES value is greater than 128. Test modifications: * Introduced include/have_max_indexes_[64|128].inc to facilitate skipping of tests for which the output differs with different MAX_INDEXES. * Introduced include/max_indexes.inc which would get modified by cmake to reflect the MAX_INDEXES value used to build the server. This file simply sets an mtr variable '$max_indexes' to show the MAX_INDEXES value, which will then be consumed by the above introduced include file. * Some tests (portions), dependent on MAX_INDEXES value, have been moved to separate test files.
2015-11-06 20:38:03 +01:00
*/
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
void bitmap_set_prefix(MY_BITMAP *map, uint prefix_size)
{
uint prefix, prefix_bits;
my_bitmap_map *value= map->bitmap;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP(map);
DBUG_ASSERT(prefix_size <= map->n_bits || prefix_size == (uint) ~0);
set_if_smaller(prefix_size, map->n_bits);
if ((prefix= prefix_size / my_bitmap_map_bits))
2011-02-24 01:36:32 +01:00
{
my_bitmap_map *end= value+prefix;
do
{
*value++= ~(my_bitmap_map) 0;
} while (value < end);
2011-02-24 01:36:32 +01:00
}
if ((prefix_bits= prefix_size & (my_bitmap_map_bits-1)))
*value++= (1ULL << prefix_bits)-1;
while (value <= map->last_word_ptr)
*value++= 0;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP(map);
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/**
Check if bitmap is a bitmap of prefix bits set in the beginning
@param map bitmap
@param prefix_size number of bits that should be set. 0 is allowed.
@return 1 Yes, prefix bits where set or prefix_size == 0.
@return 0 No
*/
2003-10-30 19:17:57 +01:00
my_bool bitmap_is_prefix(const MY_BITMAP *map, uint prefix_size)
{
my_bitmap_map *value= map->bitmap;
my_bitmap_map *end= value+ (prefix_size/my_bitmap_map_bits);
uint prefix_bits;
/* Empty prefix is always true */
if (!prefix_size)
return 1;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP_AND_BIT(map, prefix_size-1);
while (value < end)
if (*value++ != ~(my_bitmap_map) 0)
2005-05-12 11:20:50 +02:00
return 0;
if ((prefix_bits= prefix_size & (my_bitmap_map_bits-1)))
{
if (*value++ != (1ULL << prefix_bits)-1)
return 0;
}
end= map->last_word_ptr;
while (value <= end)
if (*value++ != 0)
return 0;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
return 1;
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
2003-10-30 19:17:57 +01:00
my_bool bitmap_is_set_all(const MY_BITMAP *map)
{
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
my_bitmap_map *data_ptr= map->bitmap;
my_bitmap_map *end= map->last_word_ptr;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP(map);
for (; data_ptr < end; data_ptr++)
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
if (*data_ptr != ~(my_bitmap_map)0)
2005-05-12 11:20:50 +02:00
return FALSE;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
return (*data_ptr | map->last_bit_mask) == ~(my_bitmap_map)0;
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
2005-05-12 11:20:50 +02:00
my_bool bitmap_is_clear_all(const MY_BITMAP *map)
{
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
my_bitmap_map *data_ptr= map->bitmap;
Bug #11752069 (former bug 43152) Assertion `bitmap_is_set_all(&table->s->all_set)' failed in handler::ha_reset This assertion could be triggered if two connections simultaneously executed two bitmap test functions on the same bitmap. For example, the assertion could be triggered if one connection executed UPDATE while a second connection executed SELECT on the same table. Even if bitmap test functions have read-only semantics and have const bitmaps as parameter, several of them modified the internal state of the bitmap. With interleaved execution of two such functions it was possible for one function to modify the state of the same bitmap that the other function had just modified. This lead to an inconsistent state and could trigger the assert. Internally the bitmap uses 32 bit words for storage. Since bitmaps can contain any number of bits, the last word in the bitmap may not be fully used. A 32 bit mask is maintained where a bit is set if the corresponding bit in the last bitmap word is unused. The problem was that several test functions applied this mask to the last word. Sometimes the mask was negated and used to zero out the remainder of the last word and sometimes the mask was used as-is to fill the remainder of the last word with 1's. This meant that if a function first used the negated mask and another function then used the mask as-is (or vice-versa), the first function would then get the wrong result. This patch fixes the problem by changing the implementation of 9 bitmap functions that modified the bitmap state even if the bitmap was declared const. These functions now preserve the internal state of the bitmap. This makes it possible for two connections to concurrently execute two of these functions on the same bitmap without issues. The patch also removes dead testing code from my_bitmap.c. These tests have already been moved to unittest/mysys/bitmap-t.c. Existing test coverage of my_bitmap has been extended. No MTR test case added as this would require adding several sync points to the bitmap functions. The patch has been tested with a non-deterministic test case posted on the bug report. include/my_bit.h: Removed my_count_bits_ushort() which is not needed anymore. Added my_count_bits_uint32(). unittest/mysys/bitmap-t.c: Extended test coverage of my_bitmap.
2011-02-16 16:26:19 +01:00
my_bitmap_map *end= map->last_word_ptr;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP(map);
Bug #11752069 (former bug 43152) Assertion `bitmap_is_set_all(&table->s->all_set)' failed in handler::ha_reset This assertion could be triggered if two connections simultaneously executed two bitmap test functions on the same bitmap. For example, the assertion could be triggered if one connection executed UPDATE while a second connection executed SELECT on the same table. Even if bitmap test functions have read-only semantics and have const bitmaps as parameter, several of them modified the internal state of the bitmap. With interleaved execution of two such functions it was possible for one function to modify the state of the same bitmap that the other function had just modified. This lead to an inconsistent state and could trigger the assert. Internally the bitmap uses 32 bit words for storage. Since bitmaps can contain any number of bits, the last word in the bitmap may not be fully used. A 32 bit mask is maintained where a bit is set if the corresponding bit in the last bitmap word is unused. The problem was that several test functions applied this mask to the last word. Sometimes the mask was negated and used to zero out the remainder of the last word and sometimes the mask was used as-is to fill the remainder of the last word with 1's. This meant that if a function first used the negated mask and another function then used the mask as-is (or vice-versa), the first function would then get the wrong result. This patch fixes the problem by changing the implementation of 9 bitmap functions that modified the bitmap state even if the bitmap was declared const. These functions now preserve the internal state of the bitmap. This makes it possible for two connections to concurrently execute two of these functions on the same bitmap without issues. The patch also removes dead testing code from my_bitmap.c. These tests have already been moved to unittest/mysys/bitmap-t.c. Existing test coverage of my_bitmap has been extended. No MTR test case added as this would require adding several sync points to the bitmap functions. The patch has been tested with a non-deterministic test case posted on the bug report. include/my_bit.h: Removed my_count_bits_ushort() which is not needed anymore. Added my_count_bits_uint32(). unittest/mysys/bitmap-t.c: Extended test coverage of my_bitmap.
2011-02-16 16:26:19 +01:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
for (; data_ptr <= end; data_ptr++)
2005-05-12 11:20:50 +02:00
if (*data_ptr)
return FALSE;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
return TRUE;
}
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
/* Return TRUE if map1 is a subset of map2 */
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
2003-10-30 19:17:57 +01:00
my_bool bitmap_is_subset(const MY_BITMAP *map1, const MY_BITMAP *map2)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
my_bitmap_map *m1= map1->bitmap, *m2= map2->bitmap, *end= map1->last_word_ptr;
DBUG_ASSERT_IDENTICAL_BITMAPS(map1,map2);
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
while (m1 <= end)
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
{
if ((*m1++) & ~(*m2++))
2005-05-12 11:20:50 +02:00
return 0;
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
}
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
return 1;
}
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
/* True if bitmaps has any common bits */
my_bool bitmap_is_overlapping(const MY_BITMAP *map1, const MY_BITMAP *map2)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
my_bitmap_map *m1= map1->bitmap, *m2= map2->bitmap, *end= map1->last_word_ptr;
DBUG_ASSERT_IDENTICAL_BITMAPS(map1,map2);
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
while (m1 <= end)
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
{
if ((*m1++) & (*m2++))
return 1;
}
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
return 0;
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/*
Create intersection of two bitmaps
@param map map1. Result is stored here
@param map2 map2
*/
2003-10-30 19:17:57 +01:00
void bitmap_intersect(MY_BITMAP *map, const MY_BITMAP *map2)
{
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
my_bitmap_map *to= map->bitmap, *from= map2->bitmap, *end;
2005-05-12 11:20:50 +02:00
uint len= no_words_in_map(map), len2 = no_words_in_map(map2);
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_DIFFERENT_BITMAPS(map,map2);
2013-03-25 23:03:13 +01:00
end= to+MY_MIN(len,len2);
2003-10-24 22:44:48 +02:00
while (to < end)
*to++ &= *from++;
if (len2 <= len)
2003-10-30 21:04:09 +01:00
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
to[-1]&= ~map2->last_bit_mask; /* Clear last not relevant bits */
end+= len-len2;
2003-10-30 21:04:09 +01:00
while (to < end)
*to++= 0;
2003-10-30 21:04:09 +01:00
}
}
/*
Check if there is some bit index between start_bit and end_bit, such that
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
this is at least on bit that set for all bitmaps in bitmap_list.
SYNOPSIS
bitmap_exists_intersection()
bitmpap_array [in] a set of MY_BITMAPs
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
bitmap_count [in] number of elements in bitmap_array
start_bit [in] beginning (inclusive) of the range of bits to search
end_bit [in] end (inclusive) of the range of bits to search, must be
no bigger than the bits of the shortest bitmap.
RETURN
TRUE if an intersecion exists
FALSE no intersection
*/
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
my_bool bitmap_exists_intersection(MY_BITMAP **bitmap_array,
uint bitmap_count,
uint start_bit, uint end_bit)
{
uint i, j, start_idx, end_idx;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
my_bitmap_map cur_res, first_map;
DBUG_ASSERT(bitmap_count);
DBUG_ASSERT(end_bit >= start_bit);
for (j= 0; j < bitmap_count; j++)
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
{
DBUG_ASSERT_BITMAP_AND_BIT(bitmap_array[j], end_bit);
}
start_idx= start_bit/8/sizeof(my_bitmap_map);
end_idx= end_bit/8/sizeof(my_bitmap_map);
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
first_map= first_bit_mask_inv(start_bit);
cur_res= first_map;
for (i= start_idx; i < end_idx; i++)
{
for (j= 0; cur_res && j < bitmap_count; j++)
cur_res &= bitmap_array[j]->bitmap[i];
if (cur_res)
return TRUE;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
cur_res= ~(my_bitmap_map) 0;
}
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
cur_res= ~last_bit_mask(end_bit+1);
if (start_idx == end_idx)
cur_res&= first_map;
for (j= 0; cur_res && j < bitmap_count; j++)
cur_res &= bitmap_array[j]->bitmap[end_idx];
return cur_res != 0;
}
/* True if union of bitmaps have all bits set */
my_bool bitmap_union_is_set_all(const MY_BITMAP *map1, const MY_BITMAP *map2)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
my_bitmap_map *m1= map1->bitmap, *m2= map2->bitmap, *end= map1->last_word_ptr;
DBUG_ASSERT_IDENTICAL_BITMAPS(map1,map2);
while ( m1 < end)
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
if ((*m1++ | *m2++) != ~(my_bitmap_map)0)
return FALSE;
/* here both maps have the same number of bits - see assert above */
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
return ((*m1 | *m2 | map1->last_bit_mask) != ~(my_bitmap_map)0);
Bug#10932 - Building server with key limit of 128, makes test cases fail This patch allows to configure MyISAM for 128 indexes per table. The main problem is the key_map, wich is implemented as an ulonglong. To get rid of the limit and keep the efficient and flexible implementation, the highest bit is now used for all upper keys. This means that the lower keys can be disabled and enabled individually as usual and the high keys can only be disabled and enabled as a block. That way the existing test suite is still applicable, while more keys work, though slightly less efficient. To really get more than 64 keys, some defines need to be changed. Another patch will address this. include/my_bitmap.h: Bug#10932 - Building server with key limit of 128, makes test cases fail Added the declaration for a function that extends the highest bit value to all upper bits of a bigger bitmap. include/myisam.h: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed MI_MAX_POSSIBLE_KEY to what it was meant for. Added a bunch of macros to handle the MyISAM key_map. myisam/mi_check.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_create.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_delete.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_extra.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_open.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. Changed pointer types from signed char* to unsigned char*. myisam/mi_preload.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_rsame.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_rsamepos.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_search.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_update.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/mi_write.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/myisamchk.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/myisamdef.h: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed pointer types from signed char* to unsigned char*. myisam/myisamlog.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/myisampack.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. myisam/sort.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. mysys/my_bitmap.c: Bug#10932 - Building server with key limit of 128, makes test cases fail Added a function that extends the highest bit value to all upper bits of a bigger bitmap. sql/ha_myisam.cc: Bug#10932 - Building server with key limit of 128, makes test cases fail Changed key_map access to use the new macros. sql/sql_bitmap.h: Bug#10932 - Building server with key limit of 128, makes test cases fail Added a method that extends the highest bit value to all upper bits of a bigger bitmap.
2005-07-19 14:13:56 +02:00
}
2003-10-30 19:17:57 +01:00
void bitmap_subtract(MY_BITMAP *map, const MY_BITMAP *map2)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
my_bitmap_map *to= map->bitmap, *from= map2->bitmap, *end= map->last_word_ptr;
DBUG_ASSERT_IDENTICAL_BITMAPS(map,map2);
2005-05-12 11:20:50 +02:00
while (to <= end)
2003-10-24 22:44:48 +02:00
*to++ &= ~(*from++);
}
CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS New multi-key-cache handling. This was needed becasue the old one didn't work reliable with MERGE tables. ALTER TABLE table_name ... CHARACTER SET ... now changes all char/varchar/text columns to the given character set (One must use ALTER TABLE ... DEFAULT CHARACTER SET ... to change the default character set) Fixed that have_compress is detected properly (fixes problems with func_compress.test on platforms without zlib) New syntax for CACHE INDEX ('keys' is optional if no index name is given and one mentions the key cache name only ones) Removed compiler warnings Added mysql_set_server_option() to allow clients like PHP to easaily set/reset the multi-statement flag. BUILD/compile-pentium-valgrind-max: Add test of isam client/mysql.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS include/my_base.h: Remove HA_EXTRA_SET_KEY_CACHE include/my_no_pthread.h: Add defines to ignore rw-locks when running without threads include/my_sys.h: Added function for multi-key-caches include/myisam.h: Added function to handle multi-key-caches include/mysql.h: Added mysql_set_server_option include/mysql_com.h: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Added enum_mysql_set_option include/mysqld_error.h: Added error message for unknown key cache innobase/srv/srv0start.c: Removed warning that is confused for MySQL users libmysql/libmysql.c: Added mysql_set_server_option() libmysql/libmysql.def: Added mysql_set_server_option() myisam/ft_nlq_search.c: Removed compiler warning myisam/ft_static.c: Removed compiler warning and fixed wrong return value myisam/mi_check.c: Clean up multi-key-cache usage myisam/mi_checksum.c: Removed not used variable myisam/mi_close.c: keycache -> key_cache myisam/mi_delete_all.c: keycache -> key_cache myisam/mi_extra.c: keycache -> key_cache Removed HA_EXTRA_SET_KEY_CACHE myisam/mi_keycache.c: Changed logic so that it's MyISAM that is responsible for assign tables to different key caches instead of the upper level myisam/mi_locking.c: Don't change key cache on unlock (must be done before) myisam/mi_open.c: Fetch key cache to use from multi_key_cache_search() myisam/mi_page.c: keycache -> key_cache myisam/mi_panic.c: keycache -> key_cache myisam/mi_preload.c: keycache -> key_cache myisam/mi_test1.c: Use KEY_CACHE_BLOCK_SIZE myisam/mi_test2.c: Always test resize_key_cache() myisam/mi_test3.c: Use KEY_CACHE_BLOCK_SIZE instead of 512 myisam/myisamchk.c: update for multiple key caches myisam/myisamdef.h: Remove reg_keycache Add unique_name_length for storing length of unique_file_name myisam/myisamlog.c: Change how end_key_cache() is called mysql-test/mysql-test-run.sh: Fixed web link Added name of failed test to abort row. mysql-test/r/alter_table.result: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/r/case.result: Update result for DEFAULT CHARSET... mysql-test/r/cast.result: Update result for DEFAULT CHARSET... mysql-test/r/create.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_collate.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_latin1_de.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_many.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_mb.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_recoding.result: Update result for DEFAULT CHARSET... mysql-test/r/ctype_ucs.result: Update result for DEFAULT CHARSET... mysql-test/r/derived.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/r/fulltext.result: Update result for DEFAULT CHARSET... mysql-test/r/func_str.result: Update result for DEFAULT CHARSET... mysql-test/r/func_system.result: Update result for DEFAULT CHARSET... mysql-test/r/gis-rtree.result: Update result for DEFAULT CHARSET... mysql-test/r/innodb.result: Update result for DEFAULT CHARSET... mysql-test/r/key_cache.result: Update test for new key cache syntax. Added more tests mysql-test/r/merge.result: Update result for DEFAULT CHARSET... mysql-test/r/preload.result: New syntax mysql-test/r/show_check.result: Update result for DEFAULT CHARSET... mysql-test/r/sql_mode.result: Update result for DEFAULT CHARSET... mysql-test/r/subselect.result: Update result for DEFAULT CHARSET... mysql-test/r/type_blob.result: Update result for DEFAULT CHARSET... mysql-test/r/type_enum.result: Update result for DEFAULT CHARSET... mysql-test/r/type_nchar.result: Update result for DEFAULT CHARSET... mysql-test/r/type_set.result: Update result for DEFAULT CHARSET... mysql-test/r/union.result: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/alter_table.test: Testing of ALTER TABLE ... [DEFAULT] CHARACTER SET mysql-test/t/ctype_many.test: Update result for DEFAULT CHARSET... mysql-test/t/derived.test: Use STRAIGHT_JOIN to make join order predictable mysql-test/t/isam.test: Use disable warnings for test loop mysql-test/t/join.test: Update test now when we only support 61 tables in join mysql-test/t/key_cache.test: Update test for new key cache syntax. Added more tests mysql-test/t/preload.test: Update for new syntax mysql-test/t/union.test: Use STRAIGHT_JOIN to make join order predictable mysys/Makefile.am: Added mf_keycaches.c mysys/hash.c: TRUE -> 1 mysys/mf_keycache.c: Removed compiler warnings Striped end space Fixed indentation and improved function comments TRUE -> 1 Changed parameters to end_key_cache() to make it easer to use Fixed bug when using key blocks size > 1024 bytes (First part of index file could be overwritten with wrong data) Split function flush_key_blocks into two functions to not get mutex used twice when called from flush_all_key_blocks() mysys/my_bitmap.c: More debugging Safe bitmap_free() Fixed indentation mysys/my_getopt.c: Ensure that we initialize option->value, option->max_value and value from GET_ASK_ADDR mysys/my_thr_init.c: Remove not used mutex THR_LOCK_keycache mysys/typelib.c: Fixed function comments sql-common/client.c: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS Fixed the multi_result flag is set also on SELECT;s sql/ha_myisam.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/ha_myisammrg.cc: Fixed multiple key_cache handling (Now done on MyISAM level) sql/handler.cc: New multi key cache handling sql/handler.h: New multi key cache handling Added support for default character set sql/item.h: Added function cleanup() (Needed for prepared statements / cursors) sql/item_cmpfunc.h: Added cleanup function sql/item_func.cc: Indentation cleanup sql/mysql_priv.h: New multi-key-cache functions Removed LOCK_assign sql/mysqld.cc: New multi-key-cache handling Fixed that variable have_compress is set correctly sql/protocol.cc: SELECT didn't work reliable in multi-statements sql/set_var.cc: Support for new key cache variables sql/set_var.h: Support for new key cache variables sql/share/czech/errmsg.txt: New error messages sql/share/danish/errmsg.txt: New error messages sql/share/dutch/errmsg.txt: New error messages sql/share/english/errmsg.txt: New error messages sql/share/estonian/errmsg.txt: New error messages sql/share/french/errmsg.txt: New error messages sql/share/german/errmsg.txt: New error messages sql/share/greek/errmsg.txt: New error messages sql/share/hungarian/errmsg.txt: New error messages sql/share/italian/errmsg.txt: New error messages sql/share/japanese/errmsg.txt: New error messages sql/share/korean/errmsg.txt: New error messages sql/share/norwegian-ny/errmsg.txt: New error messages sql/share/norwegian/errmsg.txt: New error messages sql/share/polish/errmsg.txt: New error messages sql/share/portuguese/errmsg.txt: New error messages sql/share/romanian/errmsg.txt: New error messages sql/share/russian/errmsg.txt: New error messages sql/share/serbian/errmsg.txt: New error messages sql/share/slovak/errmsg.txt: New error messages sql/share/spanish/errmsg.txt: New error messages sql/share/swedish/errmsg.txt: New error messages sql/share/ukrainian/errmsg.txt: New error messages sql/sql_base.cc: Removed all key_cache handling (this is now done on MyISAM level) Change table_charset -> default_table_charset sql/sql_db.cc: table_charset -> default_table_charset sql/sql_delete.cc: table_charset -> default_table_charset sql/sql_lex.cc: CLIENT_MULTI_QUERIES -> CLIENT_MULTI_STATEMENTS sql/sql_lex.h: New option to store a name and length sql/sql_parse.cc: Support for mysql_set_server_option() Reset "default" keycache status variables in 'FLUSH STATUS' (Need to be improved later) sql/sql_show.cc: Add DEFAULT before CHARSET (for table character sets) Fetch key cache variables from 'sql_key_cache' sql/sql_table.cc: table_charset -> default_table_charset New multi-key-cache handling sql/sql_test.cc: Write information from all key caches sql/sql_yacc.yy: Changed syntax for CACHE INDEX ... Force user to use DEFAULT before database/table level character sets sql/structs.h: Added SHOW_KEY_CACHE_LONG (to get values from sql_key_cache) sql/table.cc: table_charset -> default_table_charset sql/table.h: New key cache handling (this is now done in mysys/mf_keycaches.c) sql/unireg.h: A
2003-11-18 12:47:27 +01:00
2003-10-30 19:17:57 +01:00
void bitmap_union(MY_BITMAP *map, const MY_BITMAP *map2)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
my_bitmap_map *to= map->bitmap, *from= map2->bitmap, *end= map->last_word_ptr;
DBUG_ASSERT_IDENTICAL_BITMAPS(map,map2);
2005-05-12 11:20:50 +02:00
while (to <= end)
*to++ |= *from++;
}
void bitmap_xor(MY_BITMAP *map, const MY_BITMAP *map2)
{
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
my_bitmap_map *to= map->bitmap, *from= map2->bitmap, *end= map->last_word_ptr;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_IDENTICAL_BITMAPS(map,map2);
2005-05-12 11:20:50 +02:00
while (to <= end)
*to++ ^= *from++;
}
void bitmap_invert(MY_BITMAP *map)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
my_bitmap_map *to= map->bitmap, *end= map->last_word_ptr;
DBUG_ASSERT_BITMAP(map);
2005-05-12 11:20:50 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
while (to < end)
*to++ ^= ~(my_bitmap_map)0;
2005-05-12 11:20:50 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
*to ^= (~(my_bitmap_map)0 & ~map->last_bit_mask);
DBUG_ASSERT_BITMAP(map);
2005-05-12 11:20:50 +02:00
}
uint bitmap_bits_set(const MY_BITMAP *map)
{
my_bitmap_map *data_ptr= map->bitmap;
my_bitmap_map *end= map->last_word_ptr;
2005-05-12 11:20:50 +02:00
uint res= 0;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP(map);
2005-05-12 11:20:50 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
for (; data_ptr <= end; data_ptr++)
res+= my_count_bits(*data_ptr);
return res;
}
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/**
Copy bitmaps
Bug #11752069 (former bug 43152) Assertion `bitmap_is_set_all(&table->s->all_set)' failed in handler::ha_reset This assertion could be triggered if two connections simultaneously executed two bitmap test functions on the same bitmap. For example, the assertion could be triggered if one connection executed UPDATE while a second connection executed SELECT on the same table. Even if bitmap test functions have read-only semantics and have const bitmaps as parameter, several of them modified the internal state of the bitmap. With interleaved execution of two such functions it was possible for one function to modify the state of the same bitmap that the other function had just modified. This lead to an inconsistent state and could trigger the assert. Internally the bitmap uses 32 bit words for storage. Since bitmaps can contain any number of bits, the last word in the bitmap may not be fully used. A 32 bit mask is maintained where a bit is set if the corresponding bit in the last bitmap word is unused. The problem was that several test functions applied this mask to the last word. Sometimes the mask was negated and used to zero out the remainder of the last word and sometimes the mask was used as-is to fill the remainder of the last word with 1's. This meant that if a function first used the negated mask and another function then used the mask as-is (or vice-versa), the first function would then get the wrong result. This patch fixes the problem by changing the implementation of 9 bitmap functions that modified the bitmap state even if the bitmap was declared const. These functions now preserve the internal state of the bitmap. This makes it possible for two connections to concurrently execute two of these functions on the same bitmap without issues. The patch also removes dead testing code from my_bitmap.c. These tests have already been moved to unittest/mysys/bitmap-t.c. Existing test coverage of my_bitmap has been extended. No MTR test case added as this would require adding several sync points to the bitmap functions. The patch has been tested with a non-deterministic test case posted on the bug report. include/my_bit.h: Removed my_count_bits_ushort() which is not needed anymore. Added my_count_bits_uint32(). unittest/mysys/bitmap-t.c: Extended test coverage of my_bitmap.
2011-02-16 16:26:19 +01:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
@param map1 to-bitmap
@param map2 from-bitmap
@notes
Code will work even of the bitmaps are of different size.
In this case, only up to to->n_bits will be copied.
*/
void bitmap_copy(MY_BITMAP *map1, const MY_BITMAP *map2)
{
my_bitmap_map *to= map1->bitmap, *from= map2->bitmap;
uint map1_length= no_words_in_map(map1)*sizeof(my_bitmap_map);
uint map2_length= no_words_in_map(map2)*sizeof(my_bitmap_map);
uint length= MY_MIN(map1_length, map2_length);
DBUG_ASSERT_DIFFERENT_BITMAPS(map1,map2);
memcpy(to, from, length);
if (length < map1_length)
bzero(to + length, map1_length - length);
*map1->last_word_ptr&= ~map1->last_bit_mask;
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
}
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/*
Find first set bit in the bitmap
*/
2005-05-12 11:20:50 +02:00
uint bitmap_get_first_set(const MY_BITMAP *map)
{
my_bitmap_map *data_ptr= map->bitmap, *end= map->last_word_ptr;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
DBUG_ASSERT_BITMAP(map);
2005-05-12 11:20:50 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
for (uint i=0; data_ptr <= end; data_ptr++, i++)
2005-05-12 11:20:50 +02:00
if (*data_ptr)
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
return my_find_first_bit(*data_ptr) + i * sizeof(my_bitmap_map)*8;
return MY_BIT_NONE;
}
/**
Get the next set bit.
@param map Bitmap
@param bitmap_bit Bit to start search from
@return Index to first bit set after bitmap_bit
*/
uint bitmap_get_next_set(const MY_BITMAP *map, uint bitmap_bit)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
uint word_pos;
my_bitmap_map first_word, *data_ptr, *end= map->last_word_ptr;
DBUG_ASSERT_BITMAP(map);
/* Look for the next bit */
bitmap_bit++;
if (bitmap_bit >= map->n_bits)
return MY_BIT_NONE;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
word_pos= bitmap_bit / 64;
data_ptr= map->bitmap + word_pos;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
first_word= *data_ptr & first_bit_mask_inv(bitmap_bit);
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
if (first_word)
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/* Optimize common case when most bits are set */
if (first_word & (1ULL << ((bitmap_bit & (my_bitmap_map_bits-1)))))
return bitmap_bit;
return my_find_first_bit(first_word) + (bitmap_bit & ~(my_bitmap_map_bits-1));
2005-05-12 11:20:50 +02:00
}
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
for (data_ptr++; data_ptr <= end; data_ptr++)
{
bitmap_bit+= 64;
if (*data_ptr)
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
return my_find_first_bit(*data_ptr) + (bitmap_bit & ~(my_bitmap_map_bits-1));
}
return MY_BIT_NONE;
2005-05-12 11:20:50 +02:00
}
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/* Get first clear bit */
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
uint bitmap_get_first_clear(const MY_BITMAP *map)
2005-05-12 11:20:50 +02:00
{
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
uint i;
my_bitmap_map *data_ptr= map->bitmap, *end= map->last_word_ptr;
DBUG_ASSERT_BITMAP(map);
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes Changes that requires code changes in other code of other storage engines. (Note that all changes are very straightforward and one should find all issues by compiling a --debug build and fixing all compiler errors and all asserts in field.cc while running the test suite), - New optional handler function introduced: reset() This is called after every DML statement to make it easy for a handler to statement specific cleanups. (The only case it's not called is if force the file to be closed) - handler::extra(HA_EXTRA_RESET) is removed. Code that was there before should be moved to handler::reset() - table->read_set contains a bitmap over all columns that are needed in the query. read_row() and similar functions only needs to read these columns - table->write_set contains a bitmap over all columns that will be updated in the query. write_row() and update_row() only needs to update these columns. The above bitmaps should now be up to date in all context (including ALTER TABLE, filesort()). The handler is informed of any changes to the bitmap after fix_fields() by calling the virtual function handler::column_bitmaps_signal(). If the handler does caching of these bitmaps (instead of using table->read_set, table->write_set), it should redo the caching in this code. as the signal() may be sent several times, it's probably best to set a variable in the signal and redo the caching on read_row() / write_row() if the variable was set. - Removed the read_set and write_set bitmap objects from the handler class - Removed all column bit handling functions from the handler class. (Now one instead uses the normal bitmap functions in my_bitmap.c instead of handler dedicated bitmap functions) - field->query_id is removed. One should instead instead check table->read_set and table->write_set if a field is used in the query. - handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now instead use table->read_set to check for which columns to retrieve. - If a handler needs to call Field->val() or Field->store() on columns that are not used in the query, one should install a temporary all-columns-used map while doing so. For this, we provide the following functions: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set); field->val(); dbug_tmp_restore_column_map(table->read_set, old_map); and similar for the write map: my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set); field->val(); dbug_tmp_restore_column_map(table->write_set, old_map); If this is not done, you will sooner or later hit a DBUG_ASSERT in the field store() / val() functions. (For not DBUG binaries, the dbug_tmp_restore_column_map() and dbug_tmp_restore_column_map() are inline dummy functions and should be optimized away be the compiler). - If one needs to temporary set the column map for all binaries (and not just to avoid the DBUG_ASSERT() in the Field::store() / Field::val() methods) one should use the functions tmp_use_all_columns() and tmp_restore_column_map() instead of the above dbug_ variants. - All 'status' fields in the handler base class (like records, data_file_length etc) are now stored in a 'stats' struct. This makes it easier to know what status variables are provided by the base handler. This requires some trivial variable names in the extra() function. - New virtual function handler::records(). This is called to optimize COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true. (stats.records is not supposed to be an exact value. It's only has to be 'reasonable enough' for the optimizer to be able to choose a good optimization path). - Non virtual handler::init() function added for caching of virtual constants from engine. - Removed has_transactions() virtual method. Now one should instead return HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support transactions. - The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument that is to be used with 'new handler_name()' to allocate the handler in the right area. The xxxx_create_handler() function is also responsible for any initialization of the object before returning. For example, one should change: static handler *myisam_create_handler(TABLE_SHARE *table) { return new ha_myisam(table); } -> static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root) { return new (mem_root) ha_myisam(table); } - New optional virtual function: use_hidden_primary_key(). This is called in case of an update/delete when (table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined but we don't have a primary key. This allows the handler to take precisions in remembering any hidden primary key to able to update/delete any found row. The default handler marks all columns to be read. - handler::table_flags() now returns a ulonglong (to allow for more flags). - New/changed table_flags() - HA_HAS_RECORDS Set if ::records() is supported - HA_NO_TRANSACTIONS Set if engine doesn't support transactions - HA_PRIMARY_KEY_REQUIRED_FOR_DELETE Set if we should mark all primary key columns for read when reading rows as part of a DELETE statement. If there is no primary key, all columns are marked for read. - HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some cases (based on table->read_set) - HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - HA_DUPP_POS Renamed to HA_DUPLICATE_POS - HA_REQUIRES_KEY_COLUMNS_FOR_DELETE Set this if we should mark ALL key columns for read when when reading rows as part of a DELETE statement. In case of an update we will mark all keys for read for which key part changed value. - HA_STATS_RECORDS_IS_EXACT Set this if stats.records is exact. (This saves us some extra records() calls when optimizing COUNT(*)) - Removed table_flags() - HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if handler::records() gives an exact count() and HA_STATS_RECORDS_IS_EXACT if stats.records is exact. - HA_READ_RND_SAME Removed (no one supported this one) - Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk() - Renamed handler::dupp_pos to handler::dup_pos - Removed not used variable handler::sortkey Upper level handler changes: - ha_reset() now does some overall checks and calls ::reset() - ha_table_flags() added. This is a cached version of table_flags(). The cache is updated on engine creation time and updated on open. MySQL level changes (not obvious from the above): - DBUG_ASSERT() added to check that column usage matches what is set in the column usage bit maps. (This found a LOT of bugs in current column marking code). - In 5.1 before, all used columns was marked in read_set and only updated columns was marked in write_set. Now we only mark columns for which we need a value in read_set. - Column bitmaps are created in open_binary_frm() and open_table_from_share(). (Before this was in table.cc) - handler::table_flags() calls are replaced with handler::ha_table_flags() - For calling field->val() you must have the corresponding bit set in table->read_set. For calling field->store() you must have the corresponding bit set in table->write_set. (There are asserts in all store()/val() functions to catch wrong usage) - thd->set_query_id is renamed to thd->mark_used_columns and instead of setting this to an integer value, this has now the values: MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE Changed also all variables named 'set_query_id' to mark_used_columns. - In filesort() we now inform the handler of exactly which columns are needed doing the sort and choosing the rows. - The TABLE_SHARE object has a 'all_set' column bitmap one can use when one needs a column bitmap with all columns set. (This is used for table->use_all_columns() and other places) - The TABLE object has 3 column bitmaps: - def_read_set Default bitmap for columns to be read - def_write_set Default bitmap for columns to be written - tmp_set Can be used as a temporary bitmap when needed. The table object has also two pointer to bitmaps read_set and write_set that the handler should use to find out which columns are used in which way. - count() optimization now calls handler::records() instead of using handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true). - Added extra argument to Item::walk() to indicate if we should also traverse sub queries. - Added TABLE parameter to cp_buffer_from_ref() - Don't close tables created with CREATE ... SELECT but keep them in the table cache. (Faster usage of newly created tables). New interfaces: - table->clear_column_bitmaps() to initialize the bitmaps for tables at start of new statements. - table->column_bitmaps_set() to set up new column bitmaps and signal the handler about this. - table->column_bitmaps_set_no_signal() for some few cases where we need to setup new column bitmaps but don't signal the handler (as the handler has already been signaled about these before). Used for the momement only in opt_range.cc when doing ROR scans. - table->use_all_columns() to install a bitmap where all columns are marked as use in the read and the write set. - table->default_column_bitmaps() to install the normal read and write column bitmaps, but not signaling the handler about this. This is mainly used when creating TABLE instances. - table->mark_columns_needed_for_delete(), table->mark_columns_needed_for_delete() and table->mark_columns_needed_for_insert() to allow us to put additional columns in column usage maps if handler so requires. (The handler indicates what it neads in handler->table_flags()) - table->prepare_for_position() to allow us to tell handler that it needs to read primary key parts to be able to store them in future table->position() calls. (This replaces the table->file->ha_retrieve_all_pk function) - table->mark_auto_increment_column() to tell handler are going to update columns part of any auto_increment key. - table->mark_columns_used_by_index() to mark all columns that is part of an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow it to quickly know that it only needs to read colums that are part of the key. (The handler can also use the column map for detecting this, but simpler/faster handler can just monitor the extra() call). - table->mark_columns_used_by_index_no_reset() to in addition to other columns, also mark all columns that is used by the given key. - table->restore_column_maps_after_mark_index() to restore to default column maps after a call to table->mark_columns_used_by_index(). - New item function register_field_in_read_map(), for marking used columns in table->read_map. Used by filesort() to mark all used columns - Maintain in TABLE->merge_keys set of all keys that are used in query. (Simplices some optimization loops) - Maintain Field->part_of_key_not_clustered which is like Field->part_of_key but the field in the clustered key is not assumed to be part of all index. (used in opt_range.cc for faster loops) - dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map() tmp_use_all_columns() and tmp_restore_column_map() functions to temporally mark all columns as usable. The 'dbug_' version is primarily intended inside a handler when it wants to just call Field:store() & Field::val() functions, but don't need the column maps set for any other usage. (ie:: bitmap_is_set() is never called) - We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set. The reason for this is that if we have a column marked only for write we can't in the MySQL level know if the value changed or not. The reason this worked before was that MySQL marked all to be written columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden bug'. - open_table_from_share() does not anymore setup temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler(). (Simpler, faster code) Bugs fixed: - Column marking was not done correctly in a lot of cases. (ALTER TABLE, when using triggers, auto_increment fields etc) (Could potentially result in wrong values inserted in table handlers relying on that the old column maps or field->set_query_id was correct) Especially when it comes to triggers, there may be cases where the old code would cause lost/wrong values for NDB and/or InnoDB tables. - Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags: OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG. This allowed me to remove some wrong warnings about: "Some non-transactional changed tables couldn't be rolled back" - Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose some warnings about "Some non-transactional changed tables couldn't be rolled back") - Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table() which could cause delete_table to report random failures. - Fixed core dumps for some tests when running with --debug - Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) - slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. - If we get an duplicate row on insert, change column map to read and write all columns while retrying the operation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. - For table handler that doesn't support NULL in keys, we would give an error when creating a primary key with NULL fields, even after the fields has been automaticly converted to NOT NULL. - Creating a primary key on a SPATIAL key, would fail if field was not declared as NOT NULL. Cleanups: - Removed not used condition argument to setup_tables - Removed not needed item function reset_query_id_processor(). - Field->add_index is removed. Now this is instead maintained in (field->flags & FIELD_IN_ADD_INDEX) - Field->fieldnr is removed (use field->field_index instead) - New argument to filesort() to indicate that it should return a set of row pointers (not used columns). This allowed me to remove some references to sql_command in filesort and should also enable us to return column results in some cases where we couldn't before. - Changed column bitmap handling in opt_range.cc to be aligned with TABLE bitmap, which allowed me to use bitmap functions instead of looping over all fields to create some needed bitmaps. (Faster and smaller code) - Broke up found too long lines - Moved some variable declaration at start of function for better code readability. - Removed some not used arguments from functions. (setup_fields(), mysql_prepare_insert_check_table()) - setup_fields() now takes an enum instead of an int for marking columns usage. - For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. - Changed some constants to enum's and define's. - Using separate column read and write sets allows for easier checking of timestamp field was set by statement. - Remove calls to free_io_cache() as this is now done automaticly in ha_reset() - Don't build table->normalized_path as this is now identical to table->path (after bar's fixes to convert filenames) - Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to do comparision with the 'convert-dbug-for-diff' tool. Things left to do in 5.1: - We wrongly log failed CREATE TABLE ... SELECT in some cases when using row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result) Mats has promised to look into this. - Test that my fix for CREATE TABLE ... SELECT is indeed correct. (I added several test cases for this, but in this case it's better that someone else also tests this throughly). Lars has promosed to do this. BitKeeper/etc/ignore: added mysys/test_bitmap include/base64.h: Removed my_global.h, as this must be included first in any program include/heap.h: Added heap_reset() (Required by new handler interface) include/my_base.h: Removed HA_EXTRA_RESET. MySQL will now call ::reset() instead of ::extra(HA_EXTRA_RESET). HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIVE_PRIMARY key are deleted as the column bitmaps makes these unnecessary include/my_bitmap.h: Remove my_pthread.h (should be included at upper level) Introduced my_bitmap_map typedef to make it the bitmap handling more like a black box Added bitmap_is_overlapping(), bitmap_test_and_clear(), bitmap_copy() and bitmap_cmp() Made bitmap_set_bit(), bitmap_flip_bit(), bitmap_clear_bit() return void include/myisam.h: Added mi_reset() (Required by new handler interface) include/myisammrg.h: Added myrg_reset() (Required by new handler interface) include/mysql_com.h: Added flag FIELD_IN_ADD_INDEX to be able to remove Field->add_index mysql-test/extra/binlog_tests/mix_innodb_myisam_binlog.test: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/install_test_db.sh: Simplify ldata usage Added --tmpdir=. option to mysqld bootstrap (Removed some warnings when TMPDIR was wrongly set) mysql-test/mysql-test-run.pl: Added --tmpdir=. to bootstrap mysql-test/mysql-test-run.sh: Use copy instead of INSTALL_DB for master and slave databases. (Speeds up startup time a lot!) Remove snapshot directories at startup (removes some strange warnings) mysql-test/r/binlog_row_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment (This found some bugs that Mats is going to fix shortly) mysql-test/r/binlog_stm_mix_innodb_myisam.result: Added testing of CREATE ... SELECT in a mixed environment mysql-test/r/create.result: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/r/federated.result: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/r/func_gconcat.result: Enable some disabled tests (converted them slightly to be predictable) mysql-test/r/func_time.result: Added drop of test function mysql-test/r/innodb_mysql.result: Added tests for CREATE ... SELECT mysql-test/r/insert.result: More tests Added testing of duplicate columns in insert mysql-test/r/loaddata.result: Added testing LOAD DATA ... SET ... mysql-test/r/multi_update.result: Test multi updates and deletes using primary key and without mysql-test/r/ndb_index_unique.result: Better error message mysql-test/r/ndb_replace.result: New correct result after fixing REPLACE handling with NDB mysql-test/r/rpl_ddl.result: Now we don't get these (wrong) warnings anymore mysql-test/r/view_grant.result: Drop used views mysql-test/t/create.test: Some extra tests of warnings and number of tables opened by CREATE ... SELECT mysql-test/t/federated.test: Drop some left over tables Added testing of multiple table update and multiple table delete (with and without keys) mysql-test/t/func_gconcat.test: Enable some disabled tests (converted them slightly to be predictable) mysql-test/t/func_time.test: Added drop of test function mysql-test/t/innodb_mysql.test: Added tests for CREATE ... SELECT mysql-test/t/insert.test: More tests Added testing of duplicate columns in insert mysql-test/t/loaddata.test: Added testing LOAD DATA ... SET ... mysql-test/t/multi_update.test: Test multi updates and deletes using primary key and without mysql-test/t/view_grant.test: Drop used views mysql-test/valgrind.supp: Added supression of not needed warnings when printing stack trace mysys/base64.c: Include my_global.h first mysys/my_bitmap.c: Added bitmap_is_overlapping(), bitmap_test_and_clear() and bitmap_copy() Changed logic of bitmap handling to be a bit more efficent (Did this together with Mikael Ronström) Now the 'extra, not used bits' in the bitmap are assumed to have a 'random value' and the bitmap functions are free to change them whenever needed. Changed how mutex is allocated to make 'bitmap_free()' function simpler. mysys/thr_lock.c: Added 0x before thread pointers (for easier comparison of DBUG traces) sql/event.cc: Ensure 'use_all_columns()' is used for event tables Don't print warning that event table is damaged if it doesn't exists. sql/field.cc: Added ASSERT_COLUMN_MARKED_FOR_WRITE in all store() methods and ASSERT_COLUMN_MARKED_FOR_READ in all val() methods to catch wrong setting if table->read_set and table->write_set (Rest of changes are only indentation cleanups) sql/field.h: Removed Field->query_id (replaced by table->read_set and table->write_set) Removed Field->fieldnr (use Field->field_index instead) Removed Field->add_index (Use Field->flags instead) Add Field->part_of_key_not_clustered (for usage in opt_range.cc) sql/filesort.cc: Added paramater sort_postion to filesort() to force sorting by position instead of storing all fields in the result set. This allowed me to remove checking of sql_command. Create a temporary column bitmap for fields that are used by the sorting process. Use column bitmaps instead of query_id sql/ha_berkeley.cc: Update to 'newer' table handler interface sql/ha_berkeley.h: Update to 'newer' table handler interface sql/ha_federated.cc: Update to 'newer' table handler interface Only read columns that are needed from remote server. In case of eq ranges, don't generate two conditions in the WHERE clause (this can still be optimized, but would require a bigger code change) Use 'simpler to use' XXXX_LEN' macros A bit simpler logic in ::write_row() when creating statements. In update, only include test of fields actually read. (This greatly simplifies the queries sent by the federated engine) Similar changes done for delete_row() sql/ha_federated.h: Update to 'newer' table handler interface Changed XXX_LEN macros to use sizeof(...)-1, to simplify usage in ha_federated.cc Added HA_PRIMARY_KEY_REQUIRED_FOR_DELETE to tell MySQL to read all primary key columns in case of DELETE sql/ha_heap.cc: Update to 'newer' table handler interface sql/ha_heap.h: Update to 'newer' table handler interface sql/ha_innodb.cc: Update to 'newer' table handler interface - Update innobase_create_handler() to new interface - Removed HA_NOT_EXACT_COUNT (not needed) - Renamed HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION. - Prefixed base status variables with 'stats' - Use table column bitmaps instead of ha_get_bit_in_read_set() - Added ::reset(), with code from ::extra(HA_EXTRA_RESET) - Removed HA_EXTRA_RETRIVE_ALL_COLS and HA_EXTRA_RETRIEVE_PRIMARY_KEY as the table->read_set and table->write_set bitmaps now are accurate sql/ha_innodb.h: Update to 'newer' table handler interface - table_flags are now ulonglong - Added reset() method - Removed not needed ha_retrieve_all_cols() and ha_retrieve_all_pk() columns. - Made build_template() a class function to be able to easier access class variables sql/ha_myisam.cc: Update to 'newer' table handler interface sql/ha_myisam.h: Update to 'newer' table handler interface sql/ha_myisammrg.cc: Update to 'newer' table handler interface sql/ha_myisammrg.h: Update to 'newer' table handler interface sql/ha_ndbcluster.cc: Update to 'newer' table handler interface Fixed use_blob_value() to be accurate In ::complemented_read() we have to check both the read and write bitmap as the old code did mark all changed columns also in the read map Correct dumping of field data with DBUG_DUMP Prefix addresses in DBUG_PRINT with 0x Fixed usage of not initialized memory Update to use field->flags & FIELD_IN_ADD_INDEX instead of field->add_index. sql/ha_ndbcluster.h: Update to 'newer' table handler interface sql/ha_ndbcluster_binlog.cc: Mark usage of all columns in ndbcluster binlog tables false -> FALSE, true -> TRUE Use table->s->all_set instead of creating a temporary bitmap. sql/ha_partition.cc: Update to 'newer' table handler interface Added memroot to initialise_partitions() and related functions to get faster memory allocation. partition_create_handler() is now responsible for initialisation of the partition object Some trivial optimizations and indentation fixes Ensure that table_flags() are up to date Removed documentation for removed HA_EXTRA flags Fixed 'strange' usage of m_file[i] in new_handlers_from_part_info()that worked in current code 'by chance' sql/ha_partition.h: Update to 'newer' table handler interface sql/handler.cc: create_xxx handler now takes MEMROOT as an argument to simplify memory allocation. Much simpler get_new_handler() (Initialization of the object is now handled by the create method for the engine) Moved all allocation of bitmap handling to the TABLE object (in table.cc) Added column_bitmaps_signal() to signal column usage changes. Changed binlog_log_row() to use the exiusting all_set bitmap in the table object. Added ha_reset() function to test that the file object is ok at end of statement and call handler::reset() Added use_hidden_primary_key() to signal handler that we we are going to read and update + delete the row and the handler should thus remember the position for the row sql/handler.h: Added HA_NO_TRANSACTIONS, HA_PARTIAL_COLUMN_READ, HA_REQUIRES_KEY_COLUMNS_FOR_DELETE,HA_PRIMARY_KEY_REQUIRED_FOR_DELETE and HA_HAS_RECORDS Removed HA_NOT_EXACT_COUNT, HA_READ_RND_SAME HA_DUPP_POS -> HA_DUPLICATE_POS HA_NOT_EXACT_COUNT replaced by HA_STATS_RECORDS_IS_EXACT, HA_HAS_RECORDS and records() HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION Added future row type 'ROW_TYPE_PAGES' Added MEM_ROOT to handlerton 'create' function Added ha_statistics, a structure for all status variable in the base handler class. Moved all status variables in the handler class into a stats structs to improve readability. ha_table_flags() is now a cached (not virtual) version of table_flags() reset() doesn't anymore call extra(HA_EXTRA_RESET) but is a function of it's own. Renamed dupp_ref to dup_ref Renamed not used handler::sortkey Moved read_set and write_set to TABLE structure handler::init() function added for cacheing of virtual constants from engine. sql/item.cc: Added register_field_in_read_map() for marking used columns in expression. This is used by filesort() for creating an optimal column bitmap while retrieving columns for sorting. Initalize value.cs_info.character_set_client to fix core dump bug with --debug set_query_id -> mark_used_columns Mark used columns in read_set OR write_set. sql/item.h: Removed reset_query_id_processor() as it's not needed anymore. Added register_field_in_read_map() Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_cmpfunc.cc: Temporary mark used columns to be read/writable Update Item::walk to new interface sql/item_cmpfunc.h: Added extra argument to Item::walk() to indicate if we should also traverse sub queries. sql/item_func.cc: Update Item::walk() to new interface table_flags() -> ha_table_flags() sql/item_func.h: Update Item::walk() to new interface sql/item_row.cc: Update Item::walk() to new interface sql/item_row.h: Update Item::walk() to new interface sql/item_strfunc.h: Update Item::walk() to new interface sql/item_subselect.cc: Added Item_subselect::walk() (It was a bug it was missing before. Not sure what kind of bugs this could have caused) sql/item_subselect.h: Update Item::walk() to new interface sql/item_sum.cc: Update Item::walk() to new interface Updates for new handler interace sql/item_sum.h: Update Item::walk() to new interface sql/key.cc: Updates for new handler interace sql/log.cc: Mark all columns used for log tables Split options flag Ensured that second argument to trans_register_ha is a bool sql/log_event.cc: Fixed comments to be withing 79 characters Use OPTION_KEEP_LOG instead of OPTION_STATUS_NO_TRANS_UPDATE to remove wrong warnings Updates for new handler interface Use 0x%lx instead of %p (portability problem) sql/mysql_priv.h: Added OPTION_KEEP_LOG to indicate that we should replicate the binlog even on rollback Removed not used 'conds' argument to setup_tables sql/mysqld.cc: Indentation fixes and removed old comment sql/opt_range.cc: Update to new handler and bitmap interface. Fixed calls to cp_buffer_from_ref() and walk() (new argument). Create new temporary bitmaps for ror scans. (Needed because of handler changes and to get more accurate column bitmaps than before) Remove not needed file->ha_reset() call before file->close(). Some trivial optimization and indentation fixes. Use Field->part_of_key_not_clustered() to check if field is part of a key, instead of looping over all key parts. Added flag 'in_ror_merged_scan' to allow ::get_next() to know that we need a special column bitmap to only fetch pointer to record. This is needed because ror scan uses the same TABLE object but different file objects, which creates problem for the column bitmap handling. (This is a temporary solution. A better one would be to allocate an own TABLE object for ROR scans) Optimized bitmap handling in ror scans: - Start bitmap at position 0, not 1 - Use same bitmap size as in TABLE - Use table->read_set and table->write_set to create column bitmaps instead of looping over all fields in table sql/opt_range.h: Added 'in_ror_merged_scan' to indicate if we are doing a ROR scan Added temporary column bitmaps used in ROR scans sql/opt_sum.cc: Added get_ext_record_count() which is used in COUNT() optimization if handler has HA_HAS_RECORDS Note that we don't call this if handler has HA_STATS_RECORDS_IS_EXACT set. sql/protocol.cc: We need to mark columns as readable in ::store() as we sometimes return default value for fields to the user sql/records.cc: Updates for new handler interface sql/set_var.cc: Handle splitting OPTION_STATUS_NO_TRANS_UPDATE to two flags sql/share/errmsg.txt: Fixed wrong sql/sp.cc: Mark that we are using all columns for the proc table Update call to setup_tables() to use new prototype sql/sp_head.cc: Removed QQ comment sql/spatial.cc: Removed wrong QQ comment sql/sql_acl.cc: Mark that we need all columns for acl tables Supply memroot to some 'new' calls. Indentation fixes sql/sql_base.cc: set_query_id removed Ensure we call ha_reset() at end of each statement Mark read columns in read_set and changed columns in write_set (Before all columns was marked in read set) Fixed marking of some columns that was not proplerly marked before Maintain in TABLE->merge_keys set of all keys that are used in some way Removed not used 'conds' argument from setup_tables() Remove not used setting of 'dupp_field' in insert_fields() Added missing FN_LIBCHAR in mysql_rm_tmp_tables() (This has probably caused us to not properly remove temporary files after crash) sql/sql_bitmap.h: Added is_overlapping() sql/sql_class.cc: Slow_logs was not properly initialized, which could maybe cause extra/lost entries in slow log. set_query_id -> mark_used_columns Simpler variable usage in pack_row() (cleanup) Moved some variable declartion at start of function for better code readability sql/sql_class.h: Added enum_mark_columns Updated comments Renamed dupp_field -> dup_field Added virtual function 'can_rollback_data()' to select_insert() to be used in CREATE ... SELECT to optimize use of OPTION_STATUS_NO_TRANS_UPDATE. (This fixes a bug in CREATE ... SELECT where we did give wrong warnings when using non transacational tables) sql/sql_delete.cc: Updates to new handler interface Call table->mark_columns_needed_for_delete() to allow us to put additional columns in column usage maps if handler so requires. Call table->prepare_for_position() to tell handler that we are going to call ha_position(). Removed call to free_io_cache(). (io_cache is now removed in ha_reset()). Fixed calls to setup_tables() sql/sql_do.cc: Update call to setup_fields() sql/sql_handler.cc: Tell handler tables to always read all columns. Use temporary column map when storing value in field for later index usage sql/sql_help.cc: Makr all used fields to be read Update call to setup_fields() sql/sql_insert.cc: Tell handler we are going to update the auto_increment column dupp_field -> dup_field Set column usage bits for timestamp field. Call table->mark_columns_needed_for_insert() and table->mark_auto_increment_column() Removed not used argument from mysql_prepare_insert_check_table(). If we get an duplicate row on insert, change column map to read and write all columns while retrying the operatation. This is required by the definition of REPLACE and also ensures that fields that are only part of UPDATE are properly handled. This fixed a bug in NDB and REPLACE where REPLACE wrongly copied some column values from the replaced row. Setup new bitmaps for delayed insert rows Remove reseting of next_number_fields as it will be reset on next call to handler_insert() Fixed usage of thd->options and OPTION_STATUS_NO_TRANS_UPDATE. The issue was that one should not to reset this flag as it may be set by a previous statement. The way it was now used caused us to loose some warnings and get other wrong warnings when using non transactional tables mixed with transactional. I fixed it by introducing 'select_insert::can_rollback_data' to inform send_error() that the given statement can be rolled back (which in case of CREATE TABLE can always be done) Don't close tables created with CREATE ... SELECT but keep them in the table cache. Moved out MY_HOOKS from inside function (better readability) sql/sql_load.cc: Update to use new handler and column marking interface Update using setup_tables() sql/sql_olap.cc: Update calls to setup_tables Use enums instead of constants to setup_fields() sql/sql_parse.cc: Handle OPTION_KEEP_LOG: - Set it on CREATE TEMPORARY TABLE / DROP TABLE - Reset it when OPTION_STATUS_NO_TRANS_UPDATE is reset - Don't set it for CREATE ... SELECT (this is handled in select_create class) Remove reseting of OPTION_STATUS_NO_TRANS_UPDATE in begin_trans() as this should already be reset. If in autocommit mode, reset OPTION_KEEP_LOG and OPTION_STATUS_NO_TRANS_UPDATE to not give warnings in future commands sql/sql_partition.cc: Update walk() usage Trivial indentation fixes sql/sql_plugin.cc: Mark all columns as used for plugins sql/sql_prepare.cc: Added assert to find out hidden bugs in character_set_client (got an error in debug binary when this not set correctly) Updates for new handler interface Update calls to setup_fields() sql/sql_repl.cc: Indentation fixes sql/sql_select.cc: Update call to setup_tables() and setup_fields() Remove some old disabled code Update to new hadler interface Indentation cleanups Added column bitmaps for temporary tables. Remove updating of the removed slots in the Field class Added TABLE argument to cp_buffer_from_ref() (To be able to install temporary column maps) For internal temporary tables, use handler::write_row(), handler::delete_row() and handler::update_row() instead of handler::ha_xxxx() for faster execution. sql/sql_select.h: Indentaition fixes. Install temporary column usage maps when needed Added TABLE element to cp_buffer_from_ref() sql/sql_show.cc: Update to new handler interface Mark all columns used for internal tables. Style fixes. Added support for 'future' ROW_TYPE_PAGES. Don't allocate TMP_TABLE_PARAM with calloc. The 'init()' function will initialize the structure properly. sql/sql_table.cc: Update to new handler interface Simple my_snprintf -> strmake() Changed some constants to defines Don't test for NULL in primary key (as we a couple of line above force the PRIMARY KEY to be NOT NULL) Change field->add_index to use field->flags & FIELD_IN_ADD_INDEX Mark all columns as used for ALTER TABLE Style fixes Update call to filesort() sql/sql_trigger.h: Added friend functions to be able to test if triggers exists for table we are going to insert/update or delete in. sql/sql_udf.cc: Mark all columns as used for udf system table. sql/sql_union.cc: Update call to walk() Update to new handler interface sql/sql_update.cc: Remove query_id argument from compare_record() Use column bitmaps instead of query_id. We can't use compare_records() to skip updates for handlers that returns a partial column set and the read_set doesn't cover all columns in the write set, because compare_record() can't in this case know if a not read column changed value. Update call to setup_fields() Using separate column read and write sets allows for easier checking of timestamp field was set by statement. Removed call to free_io_cache() as this is now done in ha_reset() Call table->mark_columns_needed_for_update() and table->prepare_for_position() Style fixes sql/sql_view.cc: Style fixes sql/table.cc: Remove implicitely include 'errno.h' Remove code for building normalized path, as this is now identical to 'path' Remove field->fieldnr Added update of field->part_of_key_not_clustered() Create column bitmaps in TABLE and TABLE_SHARE Don't setup a temporary MEM_ROOT object as a thread specific variable for the handler. Instead we send the to-be-used MEMROOT to get_new_handler() Update to new handler interface Update call to walk() Added new functions: - st_table::clear_column_bitmaps() - st_table::prepare_for_position() - st_table::mark_columns_used_by_index() - st_table::restore_column_maps_after_mark_index() - st_table::mark_columns_used_by_index_no_reset() - st_table::mark_auto_increment_column() - st_table::mark_columns_needed_for_delete() - st_table::mark_columns_needed_for_update() - st_table::mark_columns_needed_for_insert() sql/table.h: Moved column usage bitmaps from handler to TABLE Added to TABLE_SHARE all_set and column_bitmap_size Added to TABLE merge_keys, bitmap_init_values, def_read_set, def_write_set, tmp_set, read_set and write_set. Declared all new table column bitmap functions Added TABLE functions column_bitmaps_set(), column_bitmaps_set_no_signal(), use_all_columns() and default_column_bitmaps() Added functions: tmp_use_all_columns() and tmp_restore_column_map() to temporarly switch column bitmaps Added functions: dbug_tmp_use_all_columns() and dbug_tmp_restore_column_map() to temporarly switch column bitmaps to avoid asserts in Field::store() and Field::val(). sql/tztime.cc: Mark all columns as used for timezone tables storage/archive/ha_archive.cc: Update to new handler interface storage/archive/ha_archive.h: Update to new handler interface storage/blackhole/ha_blackhole.cc: Update to new handler interface storage/blackhole/ha_blackhole.h: Update to new handler interface removed not needed flag HA_DUPP_POS storage/csv/ha_tina.cc: Update to new handler interface storage/csv/ha_tina.h: Update to new handler interface storage/example/ha_example.cc: Update to new handler interface storage/example/ha_example.h: Update to new handler interface storage/heap/hp_extra.c: Added heap_reset() (Required by new handler interface) storage/heap/hp_test2.c: Use heap_reset() storage/myisam/ft_boolean_search.c: Fixed compiler warning storage/myisam/mi_extra.c: Added mi_reset() (Required by new handler interface) storage/myisam/mi_search.c: Fixed DBUG_PRINT messages to use 0x%lx instead of %lx storage/myisam/mi_test2.c: Use mi_reset() storage/myisam/myisampack.c: Use mi_reset() storage/myisammrg/myrg_extra.c: Added myrg_reset() (Required by new handler interface) unittest/mysys/base64.t.c: Include my_global.h Don't include implictely include file 'stdlib.h'
2006-06-04 17:52:22 +02:00
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
for (i= 0; data_ptr < end; data_ptr++, i++)
if (*data_ptr != ~(my_bitmap_map)0)
goto found;
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
if ((*data_ptr | map->last_bit_mask) == ~(my_bitmap_map)0)
return MY_BIT_NONE;
found:
Optimize performance of my_bitmap MDEV-33502 Slowdown when running nested statement with many partitions This change was triggered to help some MariaDB users with close to 10000 bits in their bitmaps. - Change underlaying storage to be 64 bit instead of 32bit. - This reduses number of loops to scan bitmaps. - This can cause some bitmaps to be 4 byte large. - Ensure that all not used top-bits are always 0 (simplifes code as the last 64 bit storage is not a special case anymore). - Use my_find_first_bit() to find the first set bit which is much faster than scanning trough things byte by byte and then bit by bit. Other things: - Added a bool to remember if my_bitmap_init() did allocate the bitmap array. my_bitmap_free() will only free arrays it did allocate. This allowed me to remove setting 'bitmap=0' before calling my_bitmap_free() for cases where the bitmap's where allocated externally. - my_bitmap_init() sets bitmap to 0 in case of failure. - Added 'universal' asserts to most bitmap functions. - Change all remaining calls to bitmap_init() to my_bitmap_init(). - To finish the change from 2014. - Changed all usage of uint32 in my_bitmap.h to my_bitmap_map. - Updated bitmap_copy() to handle bitmaps of different size. - Removed const from bitmap_exists_intersection() as this caused casts on all usage. - Removed not used function bitmap_set_above(). - Renamed create_last_word_mask() to create_last_bit_mask() (to match name changes in my_bitmap.cc) - Extended bitmap-t with test for more bitmap functions.
2024-02-18 16:30:01 +01:00
/* find first zero bit by reverting all bits and find first bit */
return my_find_first_bit(~*data_ptr) + i * sizeof(my_bitmap_map)*8;
2005-05-12 11:20:50 +02:00
}
/*
Functions to export/import bitmaps to an architecture independent format
(low_byte_first)
*/
#ifdef WORDS_BIGENDIAN
/* Big endian machines, like powerpc or s390x */
void bitmap_export(uchar *to, MY_BITMAP *map)
{
my_bitmap_map *value;
uint length;
uchar buff[my_bitmap_map_bytes];
for (value= map->bitmap ; value < map->last_word_ptr ; value++)
{
int8store(to, *value);
to+= 8;
2005-05-12 11:20:50 +02:00
}
int8store(buff, *value);
/* We want length & 7 to return a serie 8,2,3,4,5,6,7, 8,2,3,... */
length= 1+ ((no_bytes_in_export_map(map) + 7) & 7);
memcpy(to, buff, length);
}
void bitmap_import(MY_BITMAP *map, uchar *from)
{
my_bitmap_map *value;
uint length;
uchar buff[my_bitmap_map_bytes];
for (value= map->bitmap ; value < map->last_word_ptr ; value++)
{
*value= uint8korr(from);
from+= 8;
}
bzero(buff, sizeof(buff));
/* We want length & 7 to return a serie 8,2,3,4,5,6,7, 8,2,3,... */
length= 1+ ((no_bytes_in_export_map(map) + 7) & 7);
memcpy(buff, from, length);
*value= uint8korr(buff) & ~map->last_bit_mask;
}
#else
/* Little endian machines, like intel and amd */
void bitmap_export(uchar *to, MY_BITMAP *map)
{
memcpy(to, (uchar*) map->bitmap, no_bytes_in_export_map(map));
}
void bitmap_import(MY_BITMAP *map, uchar *from)
{
memcpy((uchar*) map->bitmap, from, no_bytes_in_export_map(map));
*map->last_word_ptr&= ~map->last_bit_mask;
2005-05-12 11:20:50 +02:00
}
#endif /* WORDS_BIGENDIAN */