mirror of
https://github.com/MariaDB/server.git
synced 2025-01-22 23:04:20 +01:00
67cd330256
------------------------------------------------------------------------ r6130 | marko | 2009-11-02 11:42:56 +0200 (Mon, 02 Nov 2009) | 9 lines Changed paths: M /branches/zip/ChangeLog M /branches/zip/btr/btr0sea.c M /branches/zip/buf/buf0buf.c M /branches/zip/dict/dict0dict.c M /branches/zip/fil/fil0fil.c M /branches/zip/ibuf/ibuf0ibuf.c M /branches/zip/include/btr0sea.h M /branches/zip/include/dict0dict.h M /branches/zip/include/fil0fil.h M /branches/zip/include/ibuf0ibuf.h M /branches/zip/include/lock0lock.h M /branches/zip/include/log0log.h M /branches/zip/include/log0recv.h M /branches/zip/include/mem0mem.h M /branches/zip/include/mem0pool.h M /branches/zip/include/os0file.h M /branches/zip/include/pars0pars.h M /branches/zip/include/srv0srv.h M /branches/zip/include/thr0loc.h M /branches/zip/include/trx0i_s.h M /branches/zip/include/trx0purge.h M /branches/zip/include/trx0rseg.h M /branches/zip/include/trx0sys.h M /branches/zip/include/trx0undo.h M /branches/zip/include/usr0sess.h M /branches/zip/lock/lock0lock.c M /branches/zip/log/log0log.c M /branches/zip/log/log0recv.c M /branches/zip/mem/mem0dbg.c M /branches/zip/mem/mem0pool.c M /branches/zip/os/os0file.c M /branches/zip/os/os0sync.c M /branches/zip/os/os0thread.c M /branches/zip/pars/lexyy.c M /branches/zip/pars/pars0lex.l M /branches/zip/que/que0que.c M /branches/zip/srv/srv0srv.c M /branches/zip/srv/srv0start.c M /branches/zip/sync/sync0arr.c M /branches/zip/sync/sync0sync.c M /branches/zip/thr/thr0loc.c M /branches/zip/trx/trx0i_s.c M /branches/zip/trx/trx0purge.c M /branches/zip/trx/trx0rseg.c M /branches/zip/trx/trx0sys.c M /branches/zip/trx/trx0undo.c M /branches/zip/usr/usr0sess.c M /branches/zip/ut/ut0mem.c branches/zip: Free all resources at shutdown. Set pointers to NULL, so that Valgrind will not complain about freed data structures that are reachable via pointers. This addresses Bug #45992 and Bug #46656. This patch is mostly based on changes copied from branches/embedded-1.0, mainly c5432, c3439, c3134, c2994, c2978, but also some other code was copied. Some added cleanup code is specific to MySQL/InnoDB. rb://199 approved by Sunny Bains ------------------------------------------------------------------------ r6134 | marko | 2009-11-04 09:57:29 +0200 (Wed, 04 Nov 2009) | 5 lines Changed paths: M /branches/zip/ChangeLog M /branches/zip/handler/ha_innodb.cc branches/zip: innobase_convert_identifier(): Convert table names with explain_filename() to address Bug #32430: 'show innodb status' causes errors Invalid (old?) table or database name in logs. rb://134 approved by Sunny Bains ------------------------------------------------------------------------ r6137 | marko | 2009-11-04 15:24:28 +0200 (Wed, 04 Nov 2009) | 1 line Changed paths: M /branches/zip/dict/dict0dict.c branches/zip: dict_index_too_big_for_undo(): Correct a typo. ------------------------------------------------------------------------ r6153 | vasil | 2009-11-10 15:33:22 +0200 (Tue, 10 Nov 2009) | 145 lines Changed paths: M /branches/zip/handler/ha_innodb.cc branches/zip: Merge r6125:6152 from branches/5.1: (everything except the last white-space change was skipped as it is already in branches/zip) ------------------------------------------------------------------------ r6127 | vasil | 2009-10-30 11:18:25 +0200 (Fri, 30 Oct 2009) | 18 lines Changed paths: M /branches/5.1/Makefile.am M /branches/5.1/mysql-test/innodb-autoinc.result M /branches/5.1/mysql-test/innodb-autoinc.test branches/5.1: Backport c6121 from branches/zip: ------------------------------------------------------------------------ r6121 | sunny | 2009-10-30 01:42:11 +0200 (Fri, 30 Oct 2009) | 7 lines Changed paths: M /branches/zip/mysql-test/innodb-autoinc.result branches/zip: This test has been problematic for sometime now. The underlying bug is that the data dictionaries get out of sync. In the AUTOINC code we try and apply salve to the symptoms. In the past MySQL made some unrelated change and the dictionaries stopped getting out of sync and this test started to fail. Now, it seems they have reverted that changed and the test is passing again. I suspect this is not he last time that this test will change. ------------------------------------------------------------------------ ------------------------------------------------------------------------ r6129 | vasil | 2009-10-30 17:14:22 +0200 (Fri, 30 Oct 2009) | 4 lines Changed paths: M /branches/5.1/Makefile.am branches/5.1: Revert a change to Makefile.am that sneaked unnoticed in c6127. ------------------------------------------------------------------------ r6136 | marko | 2009-11-04 12:28:10 +0200 (Wed, 04 Nov 2009) | 15 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc M /branches/5.1/include/ha_prototypes.h M /branches/5.1/ut/ut0ut.c branches/5.1: Port r6134 from branches/zip: ------------------------------------------------------------------------ r6134 | marko | 2009-11-04 07:57:29 +0000 (Wed, 04 Nov 2009) | 5 lines branches/zip: innobase_convert_identifier(): Convert table names with explain_filename() to address Bug #32430: 'show innodb status' causes errors Invalid (old?) table or database name in logs. rb://134 approved by Sunny Bains ------------------------------------------------------------------------ innobase_print_identifier(): Replace with innobase_convert_name(). innobase_convert_identifier(): New function, called by innobase_convert_name(). ------------------------------------------------------------------------ r6149 | vasil | 2009-11-09 11:15:01 +0200 (Mon, 09 Nov 2009) | 5 lines Changed paths: M /branches/5.1/CMakeLists.txt branches/5.1: Followup to r5700: Adjust the changes so they are the same as in the BZR repository. ------------------------------------------------------------------------ r6150 | vasil | 2009-11-09 11:43:31 +0200 (Mon, 09 Nov 2009) | 58 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc branches/5.1: Merge a part of r2911.5.5 from MySQL: (the other part of this was merged in c5700) ------------------------------------------------------------ revno: 2911.5.5 committer: Vladislav Vaintroub <vvaintroub@mysql.com> branch nick: 5.1-innodb_plugin timestamp: Wed 2009-06-10 10:59:49 +0200 message: Backport WL#3653 to 5.1 to enable bundled innodb plugin. Remove custom DLL loader code from innodb plugin code, use symbols exported from mysqld. removed: storage/innodb_plugin/handler/handler0vars.h storage/innodb_plugin/handler/win_delay_loader.cc added: storage/mysql_storage_engine.cmake win/create_def_file.js modified: CMakeLists.txt include/m_ctype.h include/my_global.h include/my_sys.h include/mysql/plugin.h libmysqld/CMakeLists.txt mysql-test/mysql-test-run.pl mysql-test/t/plugin.test mysql-test/t/plugin_load-master.opt mysys/charset.c sql/CMakeLists.txt sql/handler.h sql/mysql_priv.h sql/mysqld.cc sql/sql_class.cc sql/sql_class.h sql/sql_list.h sql/sql_profile.h storage/Makefile.am storage/archive/CMakeLists.txt storage/blackhole/CMakeLists.txt storage/csv/CMakeLists.txt storage/example/CMakeLists.txt storage/federated/CMakeLists.txt storage/heap/CMakeLists.txt storage/innobase/CMakeLists.txt storage/innobase/handler/ha_innodb.cc storage/innodb_plugin/CMakeLists.txt storage/innodb_plugin/handler/ha_innodb.cc storage/innodb_plugin/handler/handler0alter.cc storage/innodb_plugin/handler/i_s.cc storage/innodb_plugin/plug.in storage/myisam/CMakeLists.txt storage/myisammrg/CMakeLists.txt win/Makefile.am win/configure.js ------------------------------------------------------------------------ r6152 | vasil | 2009-11-10 15:30:20 +0200 (Tue, 10 Nov 2009) | 4 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc branches/5.1: White space fixup. ------------------------------------------------------------------------ ------------------------------------------------------------------------ r6157 | jyang | 2009-11-11 14:27:09 +0200 (Wed, 11 Nov 2009) | 10 lines Changed paths: M /branches/zip/handler/ha_innodb.cc A /branches/zip/mysql-test/innodb_bug47167.result A /branches/zip/mysql-test/innodb_bug47167.test M /branches/zip/mysql-test/innodb_file_format.result branches/zip: Fix an issue that a local variable defined in innodb_file_format_check_validate() is being referenced across function in innodb_file_format_check_update(). In addition, fix "set global innodb_file_format_check = DEFAULT" call. Bug #47167: "set global innodb_file_format_check" cannot set value by User-Defined Variable." rb://169 approved by Sunny Bains and Marko. ------------------------------------------------------------------------ r6159 | vasil | 2009-11-11 15:13:01 +0200 (Wed, 11 Nov 2009) | 37 lines Changed paths: M /branches/zip/handler/ha_innodb.cc M /branches/zip/handler/ha_innodb.h branches/zip: Merge a change from MySQL: (this has been reviewed by Calvin and Marko, and Calvin says Luis has incorporated Marko's suggestions) ------------------------------------------------------------ revno: 3092.5.1 committer: Luis Soares <luis.soares@sun.com> branch nick: mysql-5.1-bugteam timestamp: Thu 2009-09-24 15:52:52 +0100 message: BUG#42829: binlogging enabled for all schemas regardless of binlog-db-db / binlog-ignore-db InnoDB will return an error if statement based replication is used along with transaction isolation level READ-COMMITTED (or weaker), even if the statement in question is filtered out according to the binlog-do-db rules set. In this case, an error should not be printed. This patch addresses this issue by extending the existing check in external_lock to take into account the filter rules before deciding to print an error. Furthermore, it also changes decide_logging_format to take into consideration whether the statement is filtered out from binlog before decision is made. added: mysql-test/suite/binlog/r/binlog_stm_do_db.result mysql-test/suite/binlog/t/binlog_stm_do_db-master.opt mysql-test/suite/binlog/t/binlog_stm_do_db.test modified: sql/sql_base.cc sql/sql_class.cc storage/innobase/handler/ha_innodb.cc storage/innobase/handler/ha_innodb.h storage/innodb_plugin/handler/ha_innodb.cc storage/innodb_plugin/handler/ha_innodb.h ------------------------------------------------------------------------ r6160 | vasil | 2009-11-11 15:33:49 +0200 (Wed, 11 Nov 2009) | 72 lines Changed paths: M /branches/zip/include/os0file.h M /branches/zip/os/os0file.c branches/zip: Merge r6152:6159 from branches/5.1: (r6158 was skipped as an equivallent change has already been merged from MySQL) ------------------------------------------------------------------------ r6154 | calvin | 2009-11-11 02:51:17 +0200 (Wed, 11 Nov 2009) | 17 lines Changed paths: M /branches/5.1/include/os0file.h M /branches/5.1/os/os0file.c branches/5.1: fix bug#3139: Mysql crashes: 'windows error 995' after several selects on a large DB During stress environment, Windows AIO may fail with error code ERROR_OPERATION_ABORTED. InnoDB does not handle the error, rather crashes. The cause of the error is unknown, but likely due to faulty hardware or driver. This patch introduces a new error code OS_FILE_OPERATION_ABORTED, which maps to Windows ERROR_OPERATION_ABORTED (995). When the error is detected during AIO, the InnoDB will issue a synchronous retry (read/write). This patch has been extensively tested by MySQL support. Approved by: Marko rb://196 ------------------------------------------------------------------------ r6158 | vasil | 2009-11-11 14:52:14 +0200 (Wed, 11 Nov 2009) | 37 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc M /branches/5.1/handler/ha_innodb.h branches/5.1: Merge a change from MySQL: (this has been reviewed by Calvin and Marko, and Calvin says Luis has incorporated Marko's suggestions) ------------------------------------------------------------ revno: 3092.5.1 committer: Luis Soares <luis.soares@sun.com> branch nick: mysql-5.1-bugteam timestamp: Thu 2009-09-24 15:52:52 +0100 message: BUG#42829: binlogging enabled for all schemas regardless of binlog-db-db / binlog-ignore-db InnoDB will return an error if statement based replication is used along with transaction isolation level READ-COMMITTED (or weaker), even if the statement in question is filtered out according to the binlog-do-db rules set. In this case, an error should not be printed. This patch addresses this issue by extending the existing check in external_lock to take into account the filter rules before deciding to print an error. Furthermore, it also changes decide_logging_format to take into consideration whether the statement is filtered out from binlog before decision is made. added: mysql-test/suite/binlog/r/binlog_stm_do_db.result mysql-test/suite/binlog/t/binlog_stm_do_db-master.opt mysql-test/suite/binlog/t/binlog_stm_do_db.test modified: sql/sql_base.cc sql/sql_class.cc storage/innobase/handler/ha_innodb.cc storage/innobase/handler/ha_innodb.h storage/innodb_plugin/handler/ha_innodb.cc storage/innodb_plugin/handler/ha_innodb.h ------------------------------------------------------------------------ ------------------------------------------------------------------------ r6161 | vasil | 2009-11-11 15:36:16 +0200 (Wed, 11 Nov 2009) | 4 lines Changed paths: M /branches/zip/ChangeLog branches/zip: Add changelog entry for r6160. ------------------------------------------------------------------------ r6162 | vasil | 2009-11-11 16:00:12 +0200 (Wed, 11 Nov 2009) | 4 lines Changed paths: M /branches/zip/ChangeLog branches/zip: Add ChangeLog for r6157. ------------------------------------------------------------------------ r6163 | calvin | 2009-11-11 17:53:20 +0200 (Wed, 11 Nov 2009) | 8 lines Changed paths: M /branches/zip/handler/ha_innodb.cc M /branches/zip/handler/ha_innodb.h branches/zip: Exclude thd_binlog_filter_ok() when building with older version of MySQL. thd_binlog_filter_ok() is introduced in MySQL 5.1.41. But the plugin can be built with MySQL prior to 5.1.41. Approved by Heikki (on IM). ------------------------------------------------------------------------ r6169 | calvin | 2009-11-12 14:40:43 +0200 (Thu, 12 Nov 2009) | 6 lines Changed paths: A /branches/zip/mysql-test/innodb_bug46676.result A /branches/zip/mysql-test/innodb_bug46676.test branches/zip: add test case for bug#46676 This crash is reproducible with InnoDB plugin 1.0.4 + MySQL 5.1.37. But no longer reproducible after MySQL 5.1.38 (with plugin 1.0.5). Add test case to catch future regression. ------------------------------------------------------------------------ r6170 | marko | 2009-11-12 15:49:08 +0200 (Thu, 12 Nov 2009) | 4 lines Changed paths: M /branches/zip/ChangeLog M /branches/zip/handler/ha_innodb.cc M /branches/zip/include/db0err.h M /branches/zip/row/row0merge.c M /branches/zip/row/row0mysql.c branches/zip: Allow CREATE INDEX to be interrupted. (Issue #354) rb://183 approved by Heikki Tuuri ------------------------------------------------------------------------ r6175 | vasil | 2009-11-16 20:07:39 +0200 (Mon, 16 Nov 2009) | 4 lines Changed paths: M /branches/zip/ChangeLog branches/zip: Wrap line at 78th char in the ChangeLog ------------------------------------------------------------------------ r6177 | calvin | 2009-11-16 20:20:38 +0200 (Mon, 16 Nov 2009) | 2 lines Changed paths: M /branches/zip/ChangeLog branches/zip: add an entry to ChangeLog for r6065 ------------------------------------------------------------------------ r6179 | marko | 2009-11-17 10:19:34 +0200 (Tue, 17 Nov 2009) | 2 lines Changed paths: M /branches/zip/handler/ha_innodb.cc branches/zip: ha_innobase::change_active_index(): When the history is missing, report it to the client, not to the error log. ------------------------------------------------------------------------ r6181 | vasil | 2009-11-17 12:21:41 +0200 (Tue, 17 Nov 2009) | 33 lines Changed paths: M /branches/zip/mysql-test/innodb-index.test branches/zip: At the end of innodb-index.test: restore the environment as it was before the test was started to silence this warning: MTR's internal check of the test case 'main.innodb-index' failed. This means that the test case does not preserve the state that existed before the test case was executed. Most likely the test case did not do a proper clean-up. This is the diff of the states of the servers before and after the test case was executed: mysqltest: Logging to '/tmp/autotest.sh-20091117_033000-zip.btyZwu/mysql-5.1/mysql-test/var/tmp/check-mysqld_1.log'. mysqltest: Results saved in '/tmp/autotest.sh-20091117_033000-zip.btyZwu/mysql-5.1/mysql-test/var/tmp/check-mysqld_1.result'. mysqltest: Connecting to server localhost:13000 (socket /tmp/autotest.sh-20091117_033000-zip.btyZwu/mysql-5.1/mysql-test/var/tmp/mysqld.1.sock) as 'root', connection 'default', attempt 0 ... mysqltest: ... Connected. mysqltest: Start processing test commands from './include/check-testcase.test' ... mysqltest: ... Done processing test commands. --- /tmp/autotest.sh-20091117_033000-zip.btyZwu/mysql-5.1/mysql-test/var/tmp/check-mysqld_1.result 2009-11-17 13:10:40.000000000 +0300 +++ /tmp/autotest.sh-20091117_033000-zip.btyZwu/mysql-5.1/mysql-test/var/tmp/check-mysqld_1.reject 2009-11-17 13:10:54.000000000 +0300 @@ -84,7 +84,7 @@ INNODB_DOUBLEWRITE ON INNODB_FAST_SHUTDOWN 1 INNODB_FILE_FORMAT Antelope -INNODB_FILE_FORMAT_CHECK Antelope +INNODB_FILE_FORMAT_CHECK Barracuda INNODB_FILE_PER_TABLE OFF INNODB_FLUSH_LOG_AT_TRX_COMMIT 1 INNODB_FLUSH_METHOD mysqltest: Result content mismatch not ok ------------------------------------------------------------------------ r6182 | marko | 2009-11-17 13:49:15 +0200 (Tue, 17 Nov 2009) | 1 line Changed paths: M /branches/zip/mysql-test/innodb-consistent-master.opt M /branches/zip/mysql-test/innodb-consistent.result M /branches/zip/mysql-test/innodb-consistent.test M /branches/zip/mysql-test/innodb-use-sys-malloc-master.opt M /branches/zip/mysql-test/innodb-use-sys-malloc.result M /branches/zip/mysql-test/innodb-use-sys-malloc.test M /branches/zip/mysql-test/innodb_bug21704.result M /branches/zip/mysql-test/innodb_bug21704.test M /branches/zip/mysql-test/innodb_bug40360.test M /branches/zip/mysql-test/innodb_bug40565.result M /branches/zip/mysql-test/innodb_bug40565.test M /branches/zip/mysql-test/innodb_bug41904.result M /branches/zip/mysql-test/innodb_bug41904.test M /branches/zip/mysql-test/innodb_bug42101-nonzero-master.opt M /branches/zip/mysql-test/innodb_bug42101-nonzero.result M /branches/zip/mysql-test/innodb_bug42101-nonzero.test M /branches/zip/mysql-test/innodb_bug42101.result M /branches/zip/mysql-test/innodb_bug42101.test M /branches/zip/mysql-test/innodb_bug44032.result M /branches/zip/mysql-test/innodb_bug44032.test M /branches/zip/mysql-test/innodb_bug44369.result M /branches/zip/mysql-test/innodb_bug44369.test M /branches/zip/mysql-test/innodb_bug44571.result M /branches/zip/mysql-test/innodb_bug44571.test M /branches/zip/mysql-test/innodb_bug45357.test M /branches/zip/mysql-test/innodb_bug46000.result M /branches/zip/mysql-test/innodb_bug46000.test M /branches/zip/mysql-test/innodb_bug46676.result M /branches/zip/mysql-test/innodb_bug46676.test M /branches/zip/mysql-test/innodb_bug47167.result M /branches/zip/mysql-test/innodb_bug47167.test M /branches/zip/mysql-test/innodb_bug47777.result M /branches/zip/mysql-test/innodb_bug47777.test M /branches/zip/mysql-test/innodb_file_format.result M /branches/zip/mysql-test/innodb_file_format.test branches/zip: Set svn:eol-style on mysql-test files. ------------------------------------------------------------------------ r6183 | marko | 2009-11-17 13:51:16 +0200 (Tue, 17 Nov 2009) | 1 line Changed paths: M /branches/zip/mysql-test/innodb-consistent-master.opt M /branches/zip/mysql-test/innodb-master.opt M /branches/zip/mysql-test/innodb-semi-consistent-master.opt M /branches/zip/mysql-test/innodb-use-sys-malloc-master.opt M /branches/zip/mysql-test/innodb_bug42101-nonzero-master.opt branches/zip: Prepend loose_ to plugin-only mysql-test options. ------------------------------------------------------------------------ r6184 | marko | 2009-11-17 13:52:01 +0200 (Tue, 17 Nov 2009) | 1 line Changed paths: M /branches/zip/mysql-test/innodb-index.result M /branches/zip/mysql-test/innodb-index.test branches/zip: innodb-index.test: Restore innodb_file_format_check. ------------------------------------------------------------------------ r6185 | marko | 2009-11-17 16:44:20 +0200 (Tue, 17 Nov 2009) | 16 lines Changed paths: M /branches/zip/handler/ha_innodb.cc M /branches/zip/mysql-test/innodb.result M /branches/zip/mysql-test/innodb.test M /branches/zip/mysql-test/innodb_bug44369.result M /branches/zip/mysql-test/innodb_bug44369.test D /branches/zip/mysql-test/patches/innodb-index.diff M /branches/zip/row/row0mysql.c branches/zip: Report duplicate table names to the client connection, not to the error log. This change will allow innodb-index.test to be re-enabled. It was previously disabled, because mysql-test-run does not like output in the error log. row_create_table_for_mysql(): Do not output anything to the error log when reporting DB_DUPLICATE_KEY. Let the caller report the error. Add a TODO comment that the dict_table_t object is apparently not freed when an error occurs. create_table_def(): Convert InnoDB table names to the character set of the client connection for reporting. Use my_error(ER_WRONG_COLUMN_NAME) for reporting reserved column names. Report my_error(ER_TABLE_EXISTS_ERROR) when row_create_table_for_mysql() returns DB_DUPLICATE_KEY. rb://206 ------------------------------------------------------------------------ r6186 | vasil | 2009-11-17 16:48:14 +0200 (Tue, 17 Nov 2009) | 4 lines Changed paths: M /branches/zip/ChangeLog branches/zip: Add ChangeLog entry for r6185. ------------------------------------------------------------------------ r6189 | marko | 2009-11-18 11:36:18 +0200 (Wed, 18 Nov 2009) | 5 lines Changed paths: M /branches/zip/ChangeLog M /branches/zip/handler/handler0alter.cc branches/zip: ha_innobase::add_index(): When creating the primary key and the table is being locked by another transaction, do not attempt to drop the table. (Bug #48782) Approved by Sunny Bains over IM ------------------------------------------------------------------------ r6194 | vasil | 2009-11-19 09:24:45 +0200 (Thu, 19 Nov 2009) | 5 lines Changed paths: M /branches/zip/include/univ.i branches/zip: Increment version number from 1.0.5 to 1.0.6 since 1.0.5 was just released by MySQL and we will soon release 1.0.6. ------------------------------------------------------------------------ r6197 | calvin | 2009-11-19 09:32:55 +0200 (Thu, 19 Nov 2009) | 6 lines Changed paths: M /branches/zip/CMakeLists.txt branches/zip: merge the fix of bug#48317 (CMake file) Due to MySQL changes to the CMake, it is no longer able to build InnoDB plugin as a static library on Windows. The fix is proposed by Vlad of MySQL. ------------------------------------------------------------------------ r6198 | vasil | 2009-11-19 09:44:31 +0200 (Thu, 19 Nov 2009) | 4 lines Changed paths: M /branches/zip/ChangeLog branches/zip: Add ChangeLog entry for r6197. ------------------------------------------------------------------------ r6199 | vasil | 2009-11-19 12:10:12 +0200 (Thu, 19 Nov 2009) | 31 lines Changed paths: M /branches/zip/ChangeLog M /branches/zip/btr/btr0btr.c M /branches/zip/data/data0type.c branches/zip: Merge r6159:6198 from branches/5.1: ------------------------------------------------------------------------ r6187 | jyang | 2009-11-18 05:27:30 +0200 (Wed, 18 Nov 2009) | 9 lines Changed paths: M /branches/5.1/btr/btr0btr.c branches/5.1: Fix bug #48469 "when innodb tablespace is configured too small, crash and corruption!". Function btr_create() did not check the return status of fseg_create(), and continue the index creation even there is no sufficient space. rb://205 Approved by Marko ------------------------------------------------------------------------ r6188 | jyang | 2009-11-18 07:14:23 +0200 (Wed, 18 Nov 2009) | 8 lines Changed paths: M /branches/5.1/data/data0type.c branches/5.1: Fix bug #48526 "Data type for float and double is incorrectly reported in InnoDB table monitor". Certain datatypes are not printed correctly in dtype_print(). rb://204 Approved by Marko. ------------------------------------------------------------------------ ------------------------------------------------------------------------ r6201 | marko | 2009-11-19 14:09:11 +0200 (Thu, 19 Nov 2009) | 2 lines Changed paths: M /branches/zip/handler/handler0alter.cc branches/zip: ha_innobase::add_index(): Clarify the comment on orphaned tables when creating a primary key. ------------------------------------------------------------------------ r6202 | jyang | 2009-11-19 15:01:00 +0200 (Thu, 19 Nov 2009) | 8 lines Changed paths: M /branches/zip/btr/btr0btr.c branches/zip: Function fseg_free() is no longer defined in branches/zip. To port fix for bug #48469 to zip, we can use btr_free_root() which frees the page, and also does not require mini-transaction. Approved by Marko. ------------------------------------------------------------------------ r6207 | vasil | 2009-11-20 10:19:14 +0200 (Fri, 20 Nov 2009) | 54 lines Changed paths: M /branches/zip/handler/ha_innodb.cc branches/zip: Merge r6198:6206 from branches/5.1: (r6203 was skipped as it is already in branches/zip) ------------------------------------------------------------------------ r6200 | vasil | 2009-11-19 12:14:23 +0200 (Thu, 19 Nov 2009) | 4 lines Changed paths: M /branches/5.1/btr/btr0btr.c branches/5.1: White space fixup - indent under the opening ( ------------------------------------------------------------------------ r6203 | jyang | 2009-11-19 15:12:22 +0200 (Thu, 19 Nov 2009) | 8 lines Changed paths: M /branches/5.1/btr/btr0btr.c branches/5.1: Use btr_free_root() instead of fseg_free() for the fix of bug #48469, because fseg_free() is not defined in the zip branch. And we could save one mini-trasaction started by fseg_free(). Approved by Marko. ------------------------------------------------------------------------ r6205 | jyang | 2009-11-20 07:55:48 +0200 (Fri, 20 Nov 2009) | 11 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc branches/5.1: Add a special case to handle the Duplicated Key error and return DB_ERROR instead. This is to avoid a possible SIGSEGV by mysql error handling re-entering the storage layer for dup key info without proper table handle. This is to prevent a server crash when error situation in bug #45961 "DDL on partitioned innodb tables leaves data dictionary in an inconsistent state" happens. rb://157 approved by Sunny Bains. ------------------------------------------------------------------------ r6206 | jyang | 2009-11-20 09:38:43 +0200 (Fri, 20 Nov 2009) | 5 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc branches/5.1: Fix a minor code formating issue for the parenthesis iplacement of the if condition in rename_table(). ------------------------------------------------------------------------ ------------------------------------------------------------------------ r6208 | vasil | 2009-11-20 10:49:24 +0200 (Fri, 20 Nov 2009) | 4 lines Changed paths: M /branches/zip/ChangeLog branches/zip: Add ChangeLog entry for c6207. ------------------------------------------------------------------------ r6210 | vasil | 2009-11-20 23:39:48 +0200 (Fri, 20 Nov 2009) | 3 lines Changed paths: M /branches/zip/trx/trx0i_s.c branches/zip: Whitespace fixup. ------------------------------------------------------------------------ r6248 | marko | 2009-11-30 12:19:50 +0200 (Mon, 30 Nov 2009) | 1 line Changed paths: M /branches/zip/ChangeLog branches/zip: ChangeLog: Document r4922 that was forgotten. ------------------------------------------------------------------------ r6252 | marko | 2009-11-30 12:50:11 +0200 (Mon, 30 Nov 2009) | 23 lines Changed paths: M /branches/zip/ChangeLog M /branches/zip/dict/dict0boot.c M /branches/zip/dict/dict0crea.c M /branches/zip/dict/dict0load.c M /branches/zip/dict/dict0mem.c M /branches/zip/fil/fil0fil.c M /branches/zip/handler/ha_innodb.cc M /branches/zip/include/dict0mem.h M /branches/zip/row/row0mysql.c branches/zip: Suppress errors about non-found temporary tables. Write the is_temp flag to SYS_TABLES.MIX_LEN. dict_table_t:🎏 Add a flag for is_temporary, DICT_TF2_TEMPORARY. Unlike other flags, this will not be written to the tablespace flags or SYS_TABLES.TYPE, but only to SYS_TABLES.MIX_LEN. dict_build_table_def_step(): Only pass DICT_TF_BITS to tablespaces. dict_check_tablespaces_and_store_max_id(), dict_load_table(): Suppress errors about temporary tables not being found. dict_create_sys_tables_tuple(): Write the DICT_TF2_TEMPORARY flag to SYS_TABLES.MIX_LEN. fil_space_create(), fil_create_new_single_table_tablespace(): Add assertions about space->flags. row_drop_table_for_mysql(): Do not complain about non-found temporary tables. rb://160 approved by Heikki Tuuri. This addresses the second part of Bug #41609 Crash recovery does not work for InnoDB temporary tables. ------------------------------------------------------------------------ r6263 | vasil | 2009-12-01 14:49:05 +0200 (Tue, 01 Dec 2009) | 4 lines Changed paths: M /branches/zip/include/univ.i branches/zip: Increment version number from 1.0.6 to 1.0.7 1.0.6 has been released ------------------------------------------------------------------------ r6264 | vasil | 2009-12-01 16:19:44 +0200 (Tue, 01 Dec 2009) | 1 line Changed paths: M /branches/zip/ChangeLog branches/zip: Add ChangeLog entry for the release of 1.0.6. ------------------------------------------------------------------------ r6269 | marko | 2009-12-02 11:35:22 +0200 (Wed, 02 Dec 2009) | 2 lines Changed paths: M /branches/zip/srv/srv0start.c branches/zip: innobase_start_or_create_for_mysql(): UNIV_IBUF_DEBUG should not break crash recovery, but UNIV_IBUF_COUNT_DEBUG will. ------------------------------------------------------------------------ r6270 | marko | 2009-12-02 11:36:47 +0200 (Wed, 02 Dec 2009) | 1 line Changed paths: M /branches/zip/srv/srv0start.c branches/zip: innobase_start_or_create_for_mysql(): Log the zlib version. ------------------------------------------------------------------------ r6271 | marko | 2009-12-02 11:43:49 +0200 (Wed, 02 Dec 2009) | 2 lines Changed paths: M /branches/zip/ChangeLog M /branches/zip/Makefile.am M /branches/zip/include/univ.i M /branches/zip/plug.in branches/zip: ChangeLog: Document that since r6270, the zlib version number will be displayed at start-up. ------------------------------------------------------------------------ r6272 | marko | 2009-12-02 11:46:05 +0200 (Wed, 02 Dec 2009) | 1 line Changed paths: M /branches/zip/Makefile.am M /branches/zip/include/univ.i M /branches/zip/plug.in branches/zip: Revert changes that were accidentally committed in r6271. ------------------------------------------------------------------------ r6274 | marko | 2009-12-03 14:47:12 +0200 (Thu, 03 Dec 2009) | 6 lines Changed paths: M /branches/zip/dict/dict0dict.c branches/zip: dict_table_check_for_dup_indexes(): Assert that the data dictionary mutex is being held while table->indexes is accessed. This is already the case. Currently, only dict_table_get_next_index() and dict_table_get_first_index() are being invoked without holding dict_sys->mutex. ------------------------------------------------------------------------ r6275 | pekka | 2009-12-03 18:32:47 +0200 (Thu, 03 Dec 2009) | 10 lines Changed paths: M /branches/zip/include/log0recv.h M /branches/zip/include/trx0sys.h M /branches/zip/log/log0recv.c M /branches/zip/trx/trx0sys.c branches/zip: Minor changes which allow build with UNIV_HOTBACKUP defined to succeed: include/trx0sys.h: Allow Hot Backup build to see some TRX_SYS_DOUBLEWRITE_... macros. trx/trx0sys.c: Exclude trx_sys_close() function from Hot Backup build. log/log0recv.[ch]: Exclude recv_sys_var_init() function from Hot Backup build. This change should not affect !UNIV_HOTBACKUP build. ------------------------------------------------------------------------ r6277 | marko | 2009-12-08 11:13:36 +0200 (Tue, 08 Dec 2009) | 1 line Changed paths: M /branches/zip/fsp/fsp0fsp.c branches/zip: fsp0fsp.c: Add some missing in/out and const qualifiers. ------------------------------------------------------------------------ r6285 | marko | 2009-12-09 09:24:50 +0200 (Wed, 09 Dec 2009) | 13 lines Changed paths: M /branches/zip/row/row0sel.c branches/zip: row_sel_fetch_columns(): Remove redundant code that was accidentally added in r1591, which introduced dfield_t::ext in order to make the merge sort of fast index creation support externally stored columns, Initially, I tried to allocate the bit for dfield_t::ext from dfield_t::len by making the length 31 bits and mapping UNIV_SQL_NULL to something that would fit in it. Then I decided that it would be too risky. The redundant check was part of the mapping. The condition may have been dfield_is_null() initially. This redundant code was noticed by Sergey Petrunya on the MySQL internals list. ------------------------------------------------------------------------ r6288 | marko | 2009-12-09 09:51:00 +0200 (Wed, 09 Dec 2009) | 15 lines Changed paths: M /branches/zip/row/row0upd.c branches/zip: row_upd_copy_columns(): Remove redundant code that was accidentally added in r1591, which introduced dfield_t::ext in order to make the merge sort of fast index creation support externally stored columns. Initially, I tried to allocate the bit for dfield_t::ext from dfield_t::len by making the length 31 bits and mapping UNIV_SQL_NULL to something that would fit in it. Then I decided that it would be too risky. The redundant check was part of the mapping. The condition may have been dfield_is_null() initially. This is similar to the redundant code in row_sel_fetch_columns() that was noticed by Sergey Petrunya on the MySQL internals list and removed in r6285. As far as I can tell, there are no redundant UNIV_SQL_NULL assignments remaining after this change. ------------------------------------------------------------------------ r6305 | marko | 2009-12-14 13:03:57 +0200 (Mon, 14 Dec 2009) | 2 lines Changed paths: M /branches/zip/row/row0umod.c branches/zip: row_undo_mod_del_unmark_sec_and_undo_update(): Add a missing const qualifier. ------------------------------------------------------------------------ r6309 | marko | 2009-12-15 14:05:50 +0200 (Tue, 15 Dec 2009) | 3 lines Changed paths: M /branches/zip/lock/lock0lock.c branches/zip: lock_rec_insert_check_and_lock(): Avoid casting away constness. Use page_rec_get_next_const() instead. This silences a gcc 4.2.4 warning. Reported by Sunny Bains. ------------------------------------------------------------------------ r6312 | marko | 2009-12-16 10:10:36 +0200 (Wed, 16 Dec 2009) | 6 lines Changed paths: M /branches/zip/fil/fil0fil.c branches/zip: fil_close(): Add #ifndef UNIV_HOTBACKUP around a debug assertion on mutex.magic_n. InnoDB Hot Backup is a single-threaded program and does not contain mutexes. This change allows InnoDB Hot Backup to be compiled with UNIV_DEBUG. Suggested by Michael Izioumtchenko. ------------------------------------------------------------------------ r6321 | marko | 2009-12-16 16:16:33 +0200 (Wed, 16 Dec 2009) | 4 lines Changed paths: M /branches/zip/row/row0merge.c branches/zip: row_merge_drop_temp_indexes(): Revert a hack to transaction isolation level that was made unnecessary by r5826 (Issue #337). When this function is called, any active data dictionary transaction should have been rolled back. ------------------------------------------------------------------------ r6345 | marko | 2009-12-21 10:46:14 +0200 (Mon, 21 Dec 2009) | 7 lines Changed paths: M /branches/zip/log/log0recv.c branches/zip: recv_scan_log_recs(): Non-functional change: Replace a debug assertion ut_ad(len > 0) with ut_ad(len >= OS_FILE_LOG_BLOCK_SIZE). This change is only for readability, for Issue #428. Another assertion on len being an integer multiple of OS_FILE_LOG_BLOCK_SIZE already ensured together with the old ut_ad(len > 0) that actually len must be at least OS_FILE_LOG_BLOCK_SIZE. ------------------------------------------------------------------------ r6346 | marko | 2009-12-21 12:03:25 +0200 (Mon, 21 Dec 2009) | 2 lines Changed paths: M /branches/zip/log/log0recv.c branches/zip: recv_recovery_from_checkpoint_finish(): Revert a change that was accidentally committed in r6345. ------------------------------------------------------------------------ r6348 | marko | 2009-12-22 11:04:34 +0200 (Tue, 22 Dec 2009) | 37 lines Changed paths: M /branches/zip/handler/ha_innodb.cc M /branches/zip/include/ha_prototypes.h M /branches/zip/include/trx0trx.h M /branches/zip/lock/lock0lock.c M /branches/zip/trx/trx0i_s.c M /branches/zip/trx/trx0trx.c branches/zip: Merge a change from MySQL: ------------------------------------------------------------ revno: 3236 committer: Satya B <satya.bn@sun.com> branch nick: mysql-5.1-bugteam timestamp: Tue 2009-12-01 17:48:57 +0530 message: merge to mysql-5.1-bugteam ------------------------------------------------------------ revno: 3234.1.1 committer: Gleb Shchepa <gshchepa@mysql.com> branch nick: mysql-5.1-bugteam timestamp: Tue 2009-12-01 14:38:40 +0400 message: Bug #38883 (reopened): thd_security_context is not thread safe, crashes? manual merge 5.0-->5.1, updating InnoDB plugin. ------------------------------------------------------------ revno: 1810.3968.13 committer: Gleb Shchepa <gshchepa@mysql.com> branch nick: mysql-5.0-bugteam timestamp: Tue 2009-12-01 14:24:44 +0400 message: Bug #38883 (reopened): thd_security_context is not thread safe, crashes? The bug 38816 changed the lock that protects THD::query from LOCK_thread_count to LOCK_thd_data, but didn't update the associated InnoDB functions. 1. The innobase_mysql_prepare_print_arbitrary_thd and the innobase_mysql_end_print_arbitrary_thd InnoDB functions have been removed, since now we have a per-thread mutex: now we don't need to wrap several inter-thread access tries to THD::query with a single global LOCK_thread_count lock, so we can simplify the code. 2. The innobase_mysql_print_thd function has been modified to lock LOCK_thd_data in direct way. ------------------------------------------------------------------------ r6351 | marko | 2009-12-22 11:11:18 +0200 (Tue, 22 Dec 2009) | 1 line Changed paths: M /branches/zip/handler/ha_innodb.cc branches/zip: Remove an obsolete declaration of LOCK_thread_count. ------------------------------------------------------------------------ r6352 | marko | 2009-12-22 12:33:01 +0200 (Tue, 22 Dec 2009) | 104 lines Changed paths: M /branches/zip/handler/ha_innodb.cc M /branches/zip/include/lock0lock.h M /branches/zip/include/srv0srv.h M /branches/zip/lock/lock0lock.c M /branches/zip/log/log0log.c M /branches/zip/mysql-test/innodb-autoinc.result M /branches/zip/mysql-test/innodb-autoinc.test M /branches/zip/row/row0sel.c M /branches/zip/srv/srv0srv.c M /branches/zip/srv/srv0start.c branches/zip: Merge revisions 6206:6350 from branches/5.1, except r6347, r6349, r6350 which were committed separately to both branches, and r6310, which was backported from zip to 5.1. ------------------------------------------------------------------------ r6206 | jyang | 2009-11-20 09:38:43 +0200 (Fri, 20 Nov 2009) | 3 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc branches/5.1: Non-functional change, fix formatting. ------------------------------------------------------------------------ r6230 | sunny | 2009-11-24 23:52:43 +0200 (Tue, 24 Nov 2009) | 3 lines Changed paths: M /branches/5.1/mysql-test/innodb-autoinc.result branches/5.1: Fix autoinc failing test results. (this should be skipped when merging 5.1 into zip) ------------------------------------------------------------------------ r6231 | sunny | 2009-11-25 10:26:27 +0200 (Wed, 25 Nov 2009) | 7 lines Changed paths: M /branches/5.1/mysql-test/innodb-autoinc.result M /branches/5.1/mysql-test/innodb-autoinc.test M /branches/5.1/row/row0sel.c branches/5.1: Fix BUG#49032 - auto_increment field does not initialize to last value in InnoDB Storage Engine. We use the appropriate function to read the column value for non-integer autoinc column types, namely float and double. rb://208. Approved by Marko. ------------------------------------------------------------------------ r6232 | sunny | 2009-11-25 10:27:39 +0200 (Wed, 25 Nov 2009) | 2 lines Changed paths: M /branches/5.1/row/row0sel.c branches/5.1: This is an interim fix, fix white space errors. ------------------------------------------------------------------------ r6233 | sunny | 2009-11-25 10:28:35 +0200 (Wed, 25 Nov 2009) | 2 lines Changed paths: M /branches/5.1/include/mach0data.h M /branches/5.1/include/mach0data.ic M /branches/5.1/mysql-test/innodb-autoinc.result M /branches/5.1/mysql-test/innodb-autoinc.test M /branches/5.1/row/row0sel.c branches/5.1: This is an interim fix, fix tests and make read float/double arg const. ------------------------------------------------------------------------ r6234 | sunny | 2009-11-25 10:29:03 +0200 (Wed, 25 Nov 2009) | 2 lines Changed paths: M /branches/5.1/row/row0sel.c branches/5.1: This is an interim fix, fix whitepsace issues. ------------------------------------------------------------------------ r6235 | sunny | 2009-11-26 01:14:42 +0200 (Thu, 26 Nov 2009) | 9 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc M /branches/5.1/mysql-test/innodb-autoinc.result M /branches/5.1/mysql-test/innodb-autoinc.test branches/5.1: Fix Bug#47720 - REPLACE INTO Autoincrement column with negative values. This bug is similiar to the negative autoinc filter patch from earlier, with the additional handling of filtering out the negative column values set explicitly by the user. rb://184 Approved by Heikki. ------------------------------------------------------------------------ r6242 | vasil | 2009-11-27 22:07:12 +0200 (Fri, 27 Nov 2009) | 4 lines Changed paths: M /branches/5.1/export.sh branches/5.1: Minor changes to support plugin snapshots. ------------------------------------------------------------------------ r6306 | calvin | 2009-12-14 15:12:46 +0200 (Mon, 14 Dec 2009) | 5 lines Changed paths: M /branches/5.1/mysql-test/innodb-autoinc.result M /branches/5.1/mysql-test/innodb-autoinc.test branches/5.1: fix bug#49267: innodb-autoinc.test fails on windows because of different case mode There is no change to the InnoDB code, only to fix test case by changing "T1" to "t1". ------------------------------------------------------------------------ r6324 | jyang | 2009-12-17 06:54:24 +0200 (Thu, 17 Dec 2009) | 8 lines Changed paths: M /branches/5.1/handler/ha_innodb.cc M /branches/5.1/include/lock0lock.h M /branches/5.1/include/srv0srv.h M /branches/5.1/lock/lock0lock.c M /branches/5.1/log/log0log.c M /branches/5.1/srv/srv0srv.c M /branches/5.1/srv/srv0start.c branches/5.1: Fix bug #47814 - Diagnostics are frequently not printed after a long lock wait in InnoDB. Separate out the lock wait timeout check thread from monitor information printing thread. rb://200 Approved by Marko. ------------------------------------------------------------------------ ------------------------------------------------------------------------ r6364 | marko | 2009-12-26 21:06:31 +0200 (Sat, 26 Dec 2009) | 4 lines Changed paths: M /branches/zip/ibuf/ibuf0ibuf.c branches/zip: ibuf_bitmap_get_map_page(): Define a wrapper macro that passes __FILE__, __LINE__ of the caller to buf_page_get_gen(). This will ease the diagnosis of the likes of Issue #135. ------------------------------------------------------------------------
4808 lines
126 KiB
C
4808 lines
126 KiB
C
/*****************************************************************************
|
|
|
|
Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
|
|
|
|
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 the Free Software
|
|
Foundation; version 2 of the License.
|
|
|
|
This program is distributed in the hope that it will be useful, but WITHOUT
|
|
ANY WARRANTY; without even the implied warranty of 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 Foundation, Inc., 59 Temple
|
|
Place, Suite 330, Boston, MA 02111-1307 USA
|
|
|
|
*****************************************************************************/
|
|
|
|
/**************************************************//**
|
|
@file fil/fil0fil.c
|
|
The tablespace memory cache
|
|
|
|
Created 10/25/1995 Heikki Tuuri
|
|
*******************************************************/
|
|
|
|
#include "fil0fil.h"
|
|
|
|
#include "mem0mem.h"
|
|
#include "hash0hash.h"
|
|
#include "os0file.h"
|
|
#include "mach0data.h"
|
|
#include "buf0buf.h"
|
|
#include "buf0flu.h"
|
|
#include "log0recv.h"
|
|
#include "fsp0fsp.h"
|
|
#include "srv0srv.h"
|
|
#include "srv0start.h"
|
|
#include "mtr0mtr.h"
|
|
#include "mtr0log.h"
|
|
#include "dict0dict.h"
|
|
#include "page0zip.h"
|
|
#ifndef UNIV_HOTBACKUP
|
|
# include "buf0lru.h"
|
|
# include "ibuf0ibuf.h"
|
|
# include "sync0sync.h"
|
|
# include "os0sync.h"
|
|
#else /* !UNIV_HOTBACKUP */
|
|
static ulint srv_data_read, srv_data_written;
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
|
|
/*
|
|
IMPLEMENTATION OF THE TABLESPACE MEMORY CACHE
|
|
=============================================
|
|
|
|
The tablespace cache is responsible for providing fast read/write access to
|
|
tablespaces and logs of the database. File creation and deletion is done
|
|
in other modules which know more of the logic of the operation, however.
|
|
|
|
A tablespace consists of a chain of files. The size of the files does not
|
|
have to be divisible by the database block size, because we may just leave
|
|
the last incomplete block unused. When a new file is appended to the
|
|
tablespace, the maximum size of the file is also specified. At the moment,
|
|
we think that it is best to extend the file to its maximum size already at
|
|
the creation of the file, because then we can avoid dynamically extending
|
|
the file when more space is needed for the tablespace.
|
|
|
|
A block's position in the tablespace is specified with a 32-bit unsigned
|
|
integer. The files in the chain are thought to be catenated, and the block
|
|
corresponding to an address n is the nth block in the catenated file (where
|
|
the first block is named the 0th block, and the incomplete block fragments
|
|
at the end of files are not taken into account). A tablespace can be extended
|
|
by appending a new file at the end of the chain.
|
|
|
|
Our tablespace concept is similar to the one of Oracle.
|
|
|
|
To acquire more speed in disk transfers, a technique called disk striping is
|
|
sometimes used. This means that logical block addresses are divided in a
|
|
round-robin fashion across several disks. Windows NT supports disk striping,
|
|
so there we do not need to support it in the database. Disk striping is
|
|
implemented in hardware in RAID disks. We conclude that it is not necessary
|
|
to implement it in the database. Oracle 7 does not support disk striping,
|
|
either.
|
|
|
|
Another trick used at some database sites is replacing tablespace files by
|
|
raw disks, that is, the whole physical disk drive, or a partition of it, is
|
|
opened as a single file, and it is accessed through byte offsets calculated
|
|
from the start of the disk or the partition. This is recommended in some
|
|
books on database tuning to achieve more speed in i/o. Using raw disk
|
|
certainly prevents the OS from fragmenting disk space, but it is not clear
|
|
if it really adds speed. We measured on the Pentium 100 MHz + NT + NTFS file
|
|
system + EIDE Conner disk only a negligible difference in speed when reading
|
|
from a file, versus reading from a raw disk.
|
|
|
|
To have fast access to a tablespace or a log file, we put the data structures
|
|
to a hash table. Each tablespace and log file is given an unique 32-bit
|
|
identifier.
|
|
|
|
Some operating systems do not support many open files at the same time,
|
|
though NT seems to tolerate at least 900 open files. Therefore, we put the
|
|
open files in an LRU-list. If we need to open another file, we may close the
|
|
file at the end of the LRU-list. When an i/o-operation is pending on a file,
|
|
the file cannot be closed. We take the file nodes with pending i/o-operations
|
|
out of the LRU-list and keep a count of pending operations. When an operation
|
|
completes, we decrement the count and return the file node to the LRU-list if
|
|
the count drops to zero. */
|
|
|
|
/** When mysqld is run, the default directory "." is the mysqld datadir,
|
|
but in the MySQL Embedded Server Library and ibbackup it is not the default
|
|
directory, and we must set the base file path explicitly */
|
|
UNIV_INTERN const char* fil_path_to_mysql_datadir = ".";
|
|
|
|
/** The number of fsyncs done to the log */
|
|
UNIV_INTERN ulint fil_n_log_flushes = 0;
|
|
|
|
/** Number of pending redo log flushes */
|
|
UNIV_INTERN ulint fil_n_pending_log_flushes = 0;
|
|
/** Number of pending tablespace flushes */
|
|
UNIV_INTERN ulint fil_n_pending_tablespace_flushes = 0;
|
|
|
|
/** The null file address */
|
|
UNIV_INTERN fil_addr_t fil_addr_null = {FIL_NULL, 0};
|
|
|
|
/** File node of a tablespace or the log data space */
|
|
struct fil_node_struct {
|
|
fil_space_t* space; /*!< backpointer to the space where this node
|
|
belongs */
|
|
char* name; /*!< path to the file */
|
|
ibool open; /*!< TRUE if file open */
|
|
os_file_t handle; /*!< OS handle to the file, if file open */
|
|
ibool is_raw_disk;/*!< TRUE if the 'file' is actually a raw
|
|
device or a raw disk partition */
|
|
ulint size; /*!< size of the file in database pages, 0 if
|
|
not known yet; the possible last incomplete
|
|
megabyte may be ignored if space == 0 */
|
|
ulint n_pending;
|
|
/*!< count of pending i/o's on this file;
|
|
closing of the file is not allowed if
|
|
this is > 0 */
|
|
ulint n_pending_flushes;
|
|
/*!< count of pending flushes on this file;
|
|
closing of the file is not allowed if
|
|
this is > 0 */
|
|
ib_int64_t modification_counter;/*!< when we write to the file we
|
|
increment this by one */
|
|
ib_int64_t flush_counter;/*!< up to what
|
|
modification_counter value we have
|
|
flushed the modifications to disk */
|
|
UT_LIST_NODE_T(fil_node_t) chain;
|
|
/*!< link field for the file chain */
|
|
UT_LIST_NODE_T(fil_node_t) LRU;
|
|
/*!< link field for the LRU list */
|
|
ulint magic_n;/*!< FIL_NODE_MAGIC_N */
|
|
};
|
|
|
|
/** Value of fil_node_struct::magic_n */
|
|
#define FIL_NODE_MAGIC_N 89389
|
|
|
|
/** Tablespace or log data space: let us call them by a common name space */
|
|
struct fil_space_struct {
|
|
char* name; /*!< space name = the path to the first file in
|
|
it */
|
|
ulint id; /*!< space id */
|
|
ib_int64_t tablespace_version;
|
|
/*!< in DISCARD/IMPORT this timestamp
|
|
is used to check if we should ignore
|
|
an insert buffer merge request for a
|
|
page because it actually was for the
|
|
previous incarnation of the space */
|
|
ibool mark; /*!< this is set to TRUE at database startup if
|
|
the space corresponds to a table in the InnoDB
|
|
data dictionary; so we can print a warning of
|
|
orphaned tablespaces */
|
|
ibool stop_ios;/*!< TRUE if we want to rename the
|
|
.ibd file of tablespace and want to
|
|
stop temporarily posting of new i/o
|
|
requests on the file */
|
|
ibool stop_ibuf_merges;
|
|
/*!< we set this TRUE when we start
|
|
deleting a single-table tablespace */
|
|
ibool is_being_deleted;
|
|
/*!< this is set to TRUE when we start
|
|
deleting a single-table tablespace and its
|
|
file; when this flag is set no further i/o
|
|
or flush requests can be placed on this space,
|
|
though there may be such requests still being
|
|
processed on this space */
|
|
ulint purpose;/*!< FIL_TABLESPACE, FIL_LOG, or
|
|
FIL_ARCH_LOG */
|
|
UT_LIST_BASE_NODE_T(fil_node_t) chain;
|
|
/*!< base node for the file chain */
|
|
ulint size; /*!< space size in pages; 0 if a single-table
|
|
tablespace whose size we do not know yet;
|
|
last incomplete megabytes in data files may be
|
|
ignored if space == 0 */
|
|
ulint flags; /*!< compressed page size and file format, or 0 */
|
|
ulint n_reserved_extents;
|
|
/*!< number of reserved free extents for
|
|
ongoing operations like B-tree page split */
|
|
ulint n_pending_flushes; /*!< this is positive when flushing
|
|
the tablespace to disk; dropping of the
|
|
tablespace is forbidden if this is positive */
|
|
ulint n_pending_ibuf_merges;/*!< this is positive
|
|
when merging insert buffer entries to
|
|
a page so that we may need to access
|
|
the ibuf bitmap page in the
|
|
tablespade: dropping of the tablespace
|
|
is forbidden if this is positive */
|
|
hash_node_t hash; /*!< hash chain node */
|
|
hash_node_t name_hash;/*!< hash chain the name_hash table */
|
|
#ifndef UNIV_HOTBACKUP
|
|
rw_lock_t latch; /*!< latch protecting the file space storage
|
|
allocation */
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
UT_LIST_NODE_T(fil_space_t) unflushed_spaces;
|
|
/*!< list of spaces with at least one unflushed
|
|
file we have written to */
|
|
ibool is_in_unflushed_spaces; /*!< TRUE if this space is
|
|
currently in unflushed_spaces */
|
|
UT_LIST_NODE_T(fil_space_t) space_list;
|
|
/*!< list of all spaces */
|
|
ulint magic_n;/*!< FIL_SPACE_MAGIC_N */
|
|
};
|
|
|
|
/** Value of fil_space_struct::magic_n */
|
|
#define FIL_SPACE_MAGIC_N 89472
|
|
|
|
/** The tablespace memory cache */
|
|
typedef struct fil_system_struct fil_system_t;
|
|
|
|
/** The tablespace memory cache; also the totality of logs (the log
|
|
data space) is stored here; below we talk about tablespaces, but also
|
|
the ib_logfiles form a 'space' and it is handled here */
|
|
|
|
struct fil_system_struct {
|
|
#ifndef UNIV_HOTBACKUP
|
|
mutex_t mutex; /*!< The mutex protecting the cache */
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
hash_table_t* spaces; /*!< The hash table of spaces in the
|
|
system; they are hashed on the space
|
|
id */
|
|
hash_table_t* name_hash; /*!< hash table based on the space
|
|
name */
|
|
UT_LIST_BASE_NODE_T(fil_node_t) LRU;
|
|
/*!< base node for the LRU list of the
|
|
most recently used open files with no
|
|
pending i/o's; if we start an i/o on
|
|
the file, we first remove it from this
|
|
list, and return it to the start of
|
|
the list when the i/o ends;
|
|
log files and the system tablespace are
|
|
not put to this list: they are opened
|
|
after the startup, and kept open until
|
|
shutdown */
|
|
UT_LIST_BASE_NODE_T(fil_space_t) unflushed_spaces;
|
|
/*!< base node for the list of those
|
|
tablespaces whose files contain
|
|
unflushed writes; those spaces have
|
|
at least one file node where
|
|
modification_counter > flush_counter */
|
|
ulint n_open; /*!< number of files currently open */
|
|
ulint max_n_open; /*!< n_open is not allowed to exceed
|
|
this */
|
|
ib_int64_t modification_counter;/*!< when we write to a file we
|
|
increment this by one */
|
|
ulint max_assigned_id;/*!< maximum space id in the existing
|
|
tables, or assigned during the time
|
|
mysqld has been up; at an InnoDB
|
|
startup we scan the data dictionary
|
|
and set here the maximum of the
|
|
space id's of the tables there */
|
|
ib_int64_t tablespace_version;
|
|
/*!< a counter which is incremented for
|
|
every space object memory creation;
|
|
every space mem object gets a
|
|
'timestamp' from this; in DISCARD/
|
|
IMPORT this is used to check if we
|
|
should ignore an insert buffer merge
|
|
request */
|
|
UT_LIST_BASE_NODE_T(fil_space_t) space_list;
|
|
/*!< list of all file spaces */
|
|
};
|
|
|
|
/** The tablespace memory cache. This variable is NULL before the module is
|
|
initialized. */
|
|
static fil_system_t* fil_system = NULL;
|
|
|
|
|
|
/********************************************************************//**
|
|
NOTE: you must call fil_mutex_enter_and_prepare_for_io() first!
|
|
|
|
Prepares a file node for i/o. Opens the file if it is closed. Updates the
|
|
pending i/o's field in the node and the system appropriately. Takes the node
|
|
off the LRU list if it is in the LRU list. The caller must hold the fil_sys
|
|
mutex. */
|
|
static
|
|
void
|
|
fil_node_prepare_for_io(
|
|
/*====================*/
|
|
fil_node_t* node, /*!< in: file node */
|
|
fil_system_t* system, /*!< in: tablespace memory cache */
|
|
fil_space_t* space); /*!< in: space */
|
|
/********************************************************************//**
|
|
Updates the data structures when an i/o operation finishes. Updates the
|
|
pending i/o's field in the node appropriately. */
|
|
static
|
|
void
|
|
fil_node_complete_io(
|
|
/*=================*/
|
|
fil_node_t* node, /*!< in: file node */
|
|
fil_system_t* system, /*!< in: tablespace memory cache */
|
|
ulint type); /*!< in: OS_FILE_WRITE or OS_FILE_READ; marks
|
|
the node as modified if
|
|
type == OS_FILE_WRITE */
|
|
/*******************************************************************//**
|
|
Checks if a single-table tablespace for a given table name exists in the
|
|
tablespace memory cache.
|
|
@return space id, ULINT_UNDEFINED if not found */
|
|
static
|
|
ulint
|
|
fil_get_space_id_for_table(
|
|
/*=======================*/
|
|
const char* name); /*!< in: table name in the standard
|
|
'databasename/tablename' format */
|
|
/*******************************************************************//**
|
|
Frees a space object from the tablespace memory cache. Closes the files in
|
|
the chain but does not delete them. There must not be any pending i/o's or
|
|
flushes on the files. */
|
|
static
|
|
ibool
|
|
fil_space_free(
|
|
/*===========*/
|
|
/* out: TRUE if success */
|
|
ulint id, /* in: space id */
|
|
ibool own_mutex);/* in: TRUE if own system->mutex */
|
|
/********************************************************************//**
|
|
Reads data from a space to a buffer. Remember that the possible incomplete
|
|
blocks at the end of file are ignored: they are not taken into account when
|
|
calculating the byte offset within a space.
|
|
@return DB_SUCCESS, or DB_TABLESPACE_DELETED if we are trying to do
|
|
i/o on a tablespace which does not exist */
|
|
UNIV_INLINE
|
|
ulint
|
|
fil_read(
|
|
/*=====*/
|
|
ibool sync, /*!< in: TRUE if synchronous aio is desired */
|
|
ulint space_id, /*!< in: space id */
|
|
ulint zip_size, /*!< in: compressed page size in bytes;
|
|
0 for uncompressed pages */
|
|
ulint block_offset, /*!< in: offset in number of blocks */
|
|
ulint byte_offset, /*!< in: remainder of offset in bytes; in aio
|
|
this must be divisible by the OS block size */
|
|
ulint len, /*!< in: how many bytes to read; this must not
|
|
cross a file boundary; in aio this must be a
|
|
block size multiple */
|
|
void* buf, /*!< in/out: buffer where to store data read;
|
|
in aio this must be appropriately aligned */
|
|
void* message) /*!< in: message for aio handler if non-sync
|
|
aio used, else ignored */
|
|
{
|
|
return(fil_io(OS_FILE_READ, sync, space_id, zip_size, block_offset,
|
|
byte_offset, len, buf, message));
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Writes data to a space from a buffer. Remember that the possible incomplete
|
|
blocks at the end of file are ignored: they are not taken into account when
|
|
calculating the byte offset within a space.
|
|
@return DB_SUCCESS, or DB_TABLESPACE_DELETED if we are trying to do
|
|
i/o on a tablespace which does not exist */
|
|
UNIV_INLINE
|
|
ulint
|
|
fil_write(
|
|
/*======*/
|
|
ibool sync, /*!< in: TRUE if synchronous aio is desired */
|
|
ulint space_id, /*!< in: space id */
|
|
ulint zip_size, /*!< in: compressed page size in bytes;
|
|
0 for uncompressed pages */
|
|
ulint block_offset, /*!< in: offset in number of blocks */
|
|
ulint byte_offset, /*!< in: remainder of offset in bytes; in aio
|
|
this must be divisible by the OS block size */
|
|
ulint len, /*!< in: how many bytes to write; this must
|
|
not cross a file boundary; in aio this must
|
|
be a block size multiple */
|
|
void* buf, /*!< in: buffer from which to write; in aio
|
|
this must be appropriately aligned */
|
|
void* message) /*!< in: message for aio handler if non-sync
|
|
aio used, else ignored */
|
|
{
|
|
return(fil_io(OS_FILE_WRITE, sync, space_id, zip_size, block_offset,
|
|
byte_offset, len, buf, message));
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns the table space by a given id, NULL if not found. */
|
|
UNIV_INLINE
|
|
fil_space_t*
|
|
fil_space_get_by_id(
|
|
/*================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
ut_ad(mutex_own(&fil_system->mutex));
|
|
|
|
HASH_SEARCH(hash, fil_system->spaces, id,
|
|
fil_space_t*, space,
|
|
ut_ad(space->magic_n == FIL_SPACE_MAGIC_N),
|
|
space->id == id);
|
|
|
|
return(space);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns the table space by a given name, NULL if not found. */
|
|
UNIV_INLINE
|
|
fil_space_t*
|
|
fil_space_get_by_name(
|
|
/*==================*/
|
|
const char* name) /*!< in: space name */
|
|
{
|
|
fil_space_t* space;
|
|
ulint fold;
|
|
|
|
ut_ad(mutex_own(&fil_system->mutex));
|
|
|
|
fold = ut_fold_string(name);
|
|
|
|
HASH_SEARCH(name_hash, fil_system->name_hash, fold,
|
|
fil_space_t*, space,
|
|
ut_ad(space->magic_n == FIL_SPACE_MAGIC_N),
|
|
!strcmp(name, space->name));
|
|
|
|
return(space);
|
|
}
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
/*******************************************************************//**
|
|
Returns the version number of a tablespace, -1 if not found.
|
|
@return version number, -1 if the tablespace does not exist in the
|
|
memory cache */
|
|
UNIV_INTERN
|
|
ib_int64_t
|
|
fil_space_get_version(
|
|
/*==================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_space_t* space;
|
|
ib_int64_t version = -1;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space) {
|
|
version = space->tablespace_version;
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(version);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns the latch of a file space.
|
|
@return latch protecting storage allocation */
|
|
UNIV_INTERN
|
|
rw_lock_t*
|
|
fil_space_get_latch(
|
|
/*================*/
|
|
ulint id, /*!< in: space id */
|
|
ulint* flags) /*!< out: tablespace flags */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
ut_a(space);
|
|
|
|
if (flags) {
|
|
*flags = space->flags;
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(&(space->latch));
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns the type of a file space.
|
|
@return FIL_TABLESPACE or FIL_LOG */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_space_get_type(
|
|
/*===============*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
ut_a(space);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(space->purpose);
|
|
}
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
|
|
/**********************************************************************//**
|
|
Checks if all the file nodes in a space are flushed. The caller must hold
|
|
the fil_system mutex.
|
|
@return TRUE if all are flushed */
|
|
static
|
|
ibool
|
|
fil_space_is_flushed(
|
|
/*=================*/
|
|
fil_space_t* space) /*!< in: space */
|
|
{
|
|
fil_node_t* node;
|
|
|
|
ut_ad(mutex_own(&fil_system->mutex));
|
|
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
while (node) {
|
|
if (node->modification_counter > node->flush_counter) {
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
node = UT_LIST_GET_NEXT(chain, node);
|
|
}
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Appends a new file to the chain of files of a space. File must be closed. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_node_create(
|
|
/*============*/
|
|
const char* name, /*!< in: file name (file must be closed) */
|
|
ulint size, /*!< in: file size in database blocks, rounded
|
|
downwards to an integer */
|
|
ulint id, /*!< in: space id where to append */
|
|
ibool is_raw) /*!< in: TRUE if a raw device or
|
|
a raw disk partition */
|
|
{
|
|
fil_node_t* node;
|
|
fil_space_t* space;
|
|
|
|
ut_a(fil_system);
|
|
ut_a(name);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
node = mem_alloc(sizeof(fil_node_t));
|
|
|
|
node->name = mem_strdup(name);
|
|
node->open = FALSE;
|
|
|
|
ut_a(!is_raw || srv_start_raw_disk_in_use);
|
|
|
|
node->is_raw_disk = is_raw;
|
|
node->size = size;
|
|
node->magic_n = FIL_NODE_MAGIC_N;
|
|
node->n_pending = 0;
|
|
node->n_pending_flushes = 0;
|
|
|
|
node->modification_counter = 0;
|
|
node->flush_counter = 0;
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (!space) {
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
" InnoDB: Error: Could not find tablespace %lu for\n"
|
|
"InnoDB: file ", (ulong) id);
|
|
ut_print_filename(stderr, name);
|
|
fputs(" in the tablespace memory cache.\n", stderr);
|
|
mem_free(node->name);
|
|
|
|
mem_free(node);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return;
|
|
}
|
|
|
|
space->size += size;
|
|
|
|
node->space = space;
|
|
|
|
UT_LIST_ADD_LAST(chain, space->chain, node);
|
|
|
|
if (id < SRV_LOG_SPACE_FIRST_ID && fil_system->max_assigned_id < id) {
|
|
|
|
fil_system->max_assigned_id = id;
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Opens a the file of a node of a tablespace. The caller must own the fil_system
|
|
mutex. */
|
|
static
|
|
void
|
|
fil_node_open_file(
|
|
/*===============*/
|
|
fil_node_t* node, /*!< in: file node */
|
|
fil_system_t* system, /*!< in: tablespace memory cache */
|
|
fil_space_t* space) /*!< in: space */
|
|
{
|
|
ib_int64_t size_bytes;
|
|
ulint size_low;
|
|
ulint size_high;
|
|
ibool ret;
|
|
ibool success;
|
|
byte* buf2;
|
|
byte* page;
|
|
ulint space_id;
|
|
ulint flags;
|
|
|
|
ut_ad(mutex_own(&(system->mutex)));
|
|
ut_a(node->n_pending == 0);
|
|
ut_a(node->open == FALSE);
|
|
|
|
if (node->size == 0) {
|
|
/* It must be a single-table tablespace and we do not know the
|
|
size of the file yet. First we open the file in the normal
|
|
mode, no async I/O here, for simplicity. Then do some checks,
|
|
and close the file again.
|
|
NOTE that we could not use the simple file read function
|
|
os_file_read() in Windows to read from a file opened for
|
|
async I/O! */
|
|
|
|
node->handle = os_file_create_simple_no_error_handling(
|
|
node->name, OS_FILE_OPEN, OS_FILE_READ_ONLY, &success);
|
|
if (!success) {
|
|
/* The following call prints an error message */
|
|
os_file_get_last_error(TRUE);
|
|
|
|
ut_print_timestamp(stderr);
|
|
|
|
fprintf(stderr,
|
|
" InnoDB: Fatal error: cannot open %s\n."
|
|
"InnoDB: Have you deleted .ibd files"
|
|
" under a running mysqld server?\n",
|
|
node->name);
|
|
ut_a(0);
|
|
}
|
|
|
|
os_file_get_size(node->handle, &size_low, &size_high);
|
|
|
|
size_bytes = (((ib_int64_t)size_high) << 32)
|
|
+ (ib_int64_t)size_low;
|
|
#ifdef UNIV_HOTBACKUP
|
|
if (space->id == 0) {
|
|
node->size = (ulint) (size_bytes / UNIV_PAGE_SIZE);
|
|
os_file_close(node->handle);
|
|
goto add_size;
|
|
}
|
|
#endif /* UNIV_HOTBACKUP */
|
|
ut_a(space->purpose != FIL_LOG);
|
|
ut_a(space->id != 0);
|
|
|
|
if (size_bytes < FIL_IBD_FILE_INITIAL_SIZE * UNIV_PAGE_SIZE) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: the size of single-table"
|
|
" tablespace file %s\n"
|
|
"InnoDB: is only %lu %lu,"
|
|
" should be at least %lu!\n",
|
|
node->name,
|
|
(ulong) size_high,
|
|
(ulong) size_low,
|
|
(ulong) (FIL_IBD_FILE_INITIAL_SIZE
|
|
* UNIV_PAGE_SIZE));
|
|
|
|
ut_a(0);
|
|
}
|
|
|
|
/* Read the first page of the tablespace */
|
|
|
|
buf2 = ut_malloc(2 * UNIV_PAGE_SIZE);
|
|
/* Align the memory for file i/o if we might have O_DIRECT
|
|
set */
|
|
page = ut_align(buf2, UNIV_PAGE_SIZE);
|
|
|
|
success = os_file_read(node->handle, page, 0, 0,
|
|
UNIV_PAGE_SIZE);
|
|
space_id = fsp_header_get_space_id(page);
|
|
flags = fsp_header_get_flags(page);
|
|
|
|
ut_free(buf2);
|
|
|
|
/* Close the file now that we have read the space id from it */
|
|
|
|
os_file_close(node->handle);
|
|
|
|
if (UNIV_UNLIKELY(space_id != space->id)) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: tablespace id is %lu"
|
|
" in the data dictionary\n"
|
|
"InnoDB: but in file %s it is %lu!\n",
|
|
space->id, node->name, space_id);
|
|
|
|
ut_error;
|
|
}
|
|
|
|
if (UNIV_UNLIKELY(space_id == ULINT_UNDEFINED
|
|
|| space_id == 0)) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: tablespace id %lu"
|
|
" in file %s is not sensible\n",
|
|
(ulong) space_id, node->name);
|
|
|
|
ut_error;
|
|
}
|
|
|
|
if (UNIV_UNLIKELY(space->flags != flags)) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: table flags are %lx"
|
|
" in the data dictionary\n"
|
|
"InnoDB: but the flags in file %s are %lx!\n",
|
|
space->flags, node->name, flags);
|
|
|
|
ut_error;
|
|
}
|
|
|
|
if (size_bytes >= 1024 * 1024) {
|
|
/* Truncate the size to whole megabytes. */
|
|
size_bytes = ut_2pow_round(size_bytes, 1024 * 1024);
|
|
}
|
|
|
|
if (!(flags & DICT_TF_ZSSIZE_MASK)) {
|
|
node->size = (ulint) (size_bytes / UNIV_PAGE_SIZE);
|
|
} else {
|
|
node->size = (ulint)
|
|
(size_bytes
|
|
/ dict_table_flags_to_zip_size(flags));
|
|
}
|
|
|
|
#ifdef UNIV_HOTBACKUP
|
|
add_size:
|
|
#endif /* UNIV_HOTBACKUP */
|
|
space->size += node->size;
|
|
}
|
|
|
|
/* printf("Opening file %s\n", node->name); */
|
|
|
|
/* Open the file for reading and writing, in Windows normally in the
|
|
unbuffered async I/O mode, though global variables may make
|
|
os_file_create() to fall back to the normal file I/O mode. */
|
|
|
|
if (space->purpose == FIL_LOG) {
|
|
node->handle = os_file_create(node->name, OS_FILE_OPEN,
|
|
OS_FILE_AIO, OS_LOG_FILE, &ret);
|
|
} else if (node->is_raw_disk) {
|
|
node->handle = os_file_create(node->name,
|
|
OS_FILE_OPEN_RAW,
|
|
OS_FILE_AIO, OS_DATA_FILE, &ret);
|
|
} else {
|
|
node->handle = os_file_create(node->name, OS_FILE_OPEN,
|
|
OS_FILE_AIO, OS_DATA_FILE, &ret);
|
|
}
|
|
|
|
ut_a(ret);
|
|
|
|
node->open = TRUE;
|
|
|
|
system->n_open++;
|
|
|
|
if (space->purpose == FIL_TABLESPACE && space->id != 0) {
|
|
/* Put the node to the LRU list */
|
|
UT_LIST_ADD_FIRST(LRU, system->LRU, node);
|
|
}
|
|
}
|
|
|
|
/**********************************************************************//**
|
|
Closes a file. */
|
|
static
|
|
void
|
|
fil_node_close_file(
|
|
/*================*/
|
|
fil_node_t* node, /*!< in: file node */
|
|
fil_system_t* system) /*!< in: tablespace memory cache */
|
|
{
|
|
ibool ret;
|
|
|
|
ut_ad(node && system);
|
|
ut_ad(mutex_own(&(system->mutex)));
|
|
ut_a(node->open);
|
|
ut_a(node->n_pending == 0);
|
|
ut_a(node->n_pending_flushes == 0);
|
|
ut_a(node->modification_counter == node->flush_counter);
|
|
|
|
ret = os_file_close(node->handle);
|
|
ut_a(ret);
|
|
|
|
/* printf("Closing file %s\n", node->name); */
|
|
|
|
node->open = FALSE;
|
|
ut_a(system->n_open > 0);
|
|
system->n_open--;
|
|
|
|
if (node->space->purpose == FIL_TABLESPACE && node->space->id != 0) {
|
|
ut_a(UT_LIST_GET_LEN(system->LRU) > 0);
|
|
|
|
/* The node is in the LRU list, remove it */
|
|
UT_LIST_REMOVE(LRU, system->LRU, node);
|
|
}
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Tries to close a file in the LRU list. The caller must hold the fil_sys
|
|
mutex.
|
|
@return TRUE if success, FALSE if should retry later; since i/o's
|
|
generally complete in < 100 ms, and as InnoDB writes at most 128 pages
|
|
from the buffer pool in a batch, and then immediately flushes the
|
|
files, there is a good chance that the next time we find a suitable
|
|
node from the LRU list */
|
|
static
|
|
ibool
|
|
fil_try_to_close_file_in_LRU(
|
|
/*=========================*/
|
|
ibool print_info) /*!< in: if TRUE, prints information why it
|
|
cannot close a file */
|
|
{
|
|
fil_node_t* node;
|
|
|
|
ut_ad(mutex_own(&fil_system->mutex));
|
|
|
|
node = UT_LIST_GET_LAST(fil_system->LRU);
|
|
|
|
if (print_info) {
|
|
fprintf(stderr,
|
|
"InnoDB: fil_sys open file LRU len %lu\n",
|
|
(ulong) UT_LIST_GET_LEN(fil_system->LRU));
|
|
}
|
|
|
|
while (node != NULL) {
|
|
if (node->modification_counter == node->flush_counter
|
|
&& node->n_pending_flushes == 0) {
|
|
|
|
fil_node_close_file(node, fil_system);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
if (print_info && node->n_pending_flushes > 0) {
|
|
fputs("InnoDB: cannot close file ", stderr);
|
|
ut_print_filename(stderr, node->name);
|
|
fprintf(stderr, ", because n_pending_flushes %lu\n",
|
|
(ulong) node->n_pending_flushes);
|
|
}
|
|
|
|
if (print_info
|
|
&& node->modification_counter != node->flush_counter) {
|
|
fputs("InnoDB: cannot close file ", stderr);
|
|
ut_print_filename(stderr, node->name);
|
|
fprintf(stderr,
|
|
", because mod_count %ld != fl_count %ld\n",
|
|
(long) node->modification_counter,
|
|
(long) node->flush_counter);
|
|
}
|
|
|
|
node = UT_LIST_GET_PREV(LRU, node);
|
|
}
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Reserves the fil_system mutex and tries to make sure we can open at least one
|
|
file while holding it. This should be called before calling
|
|
fil_node_prepare_for_io(), because that function may need to open a file. */
|
|
static
|
|
void
|
|
fil_mutex_enter_and_prepare_for_io(
|
|
/*===============================*/
|
|
ulint space_id) /*!< in: space id */
|
|
{
|
|
fil_space_t* space;
|
|
ibool success;
|
|
ibool print_info = FALSE;
|
|
ulint count = 0;
|
|
ulint count2 = 0;
|
|
|
|
retry:
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
if (space_id == 0 || space_id >= SRV_LOG_SPACE_FIRST_ID) {
|
|
/* We keep log files and system tablespace files always open;
|
|
this is important in preventing deadlocks in this module, as
|
|
a page read completion often performs another read from the
|
|
insert buffer. The insert buffer is in tablespace 0, and we
|
|
cannot end up waiting in this function. */
|
|
|
|
return;
|
|
}
|
|
|
|
if (fil_system->n_open < fil_system->max_n_open) {
|
|
|
|
return;
|
|
}
|
|
|
|
space = fil_space_get_by_id(space_id);
|
|
|
|
if (space != NULL && space->stop_ios) {
|
|
/* We are going to do a rename file and want to stop new i/o's
|
|
for a while */
|
|
|
|
if (count2 > 20000) {
|
|
fputs("InnoDB: Warning: tablespace ", stderr);
|
|
ut_print_filename(stderr, space->name);
|
|
fprintf(stderr,
|
|
" has i/o ops stopped for a long time %lu\n",
|
|
(ulong) count2);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
os_thread_sleep(20000);
|
|
|
|
count2++;
|
|
|
|
goto retry;
|
|
}
|
|
|
|
/* If the file is already open, no need to do anything; if the space
|
|
does not exist, we handle the situation in the function which called
|
|
this function */
|
|
|
|
if (!space || UT_LIST_GET_FIRST(space->chain)->open) {
|
|
|
|
return;
|
|
}
|
|
|
|
if (count > 1) {
|
|
print_info = TRUE;
|
|
}
|
|
|
|
/* Too many files are open, try to close some */
|
|
close_more:
|
|
success = fil_try_to_close_file_in_LRU(print_info);
|
|
|
|
if (success && fil_system->n_open >= fil_system->max_n_open) {
|
|
|
|
goto close_more;
|
|
}
|
|
|
|
if (fil_system->n_open < fil_system->max_n_open) {
|
|
/* Ok */
|
|
|
|
return;
|
|
}
|
|
|
|
if (count >= 2) {
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
" InnoDB: Warning: too many (%lu) files stay open"
|
|
" while the maximum\n"
|
|
"InnoDB: allowed value would be %lu.\n"
|
|
"InnoDB: You may need to raise the value of"
|
|
" innodb_open_files in\n"
|
|
"InnoDB: my.cnf.\n",
|
|
(ulong) fil_system->n_open,
|
|
(ulong) fil_system->max_n_open);
|
|
|
|
return;
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
/* Wake the i/o-handler threads to make sure pending i/o's are
|
|
performed */
|
|
os_aio_simulated_wake_handler_threads();
|
|
|
|
os_thread_sleep(20000);
|
|
#endif
|
|
/* Flush tablespaces so that we can close modified files in the LRU
|
|
list */
|
|
|
|
fil_flush_file_spaces(FIL_TABLESPACE);
|
|
|
|
count++;
|
|
|
|
goto retry;
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Frees a file node object from a tablespace memory cache. */
|
|
static
|
|
void
|
|
fil_node_free(
|
|
/*==========*/
|
|
fil_node_t* node, /*!< in, own: file node */
|
|
fil_system_t* system, /*!< in: tablespace memory cache */
|
|
fil_space_t* space) /*!< in: space where the file node is chained */
|
|
{
|
|
ut_ad(node && system && space);
|
|
ut_ad(mutex_own(&(system->mutex)));
|
|
ut_a(node->magic_n == FIL_NODE_MAGIC_N);
|
|
ut_a(node->n_pending == 0);
|
|
|
|
if (node->open) {
|
|
/* We fool the assertion in fil_node_close_file() to think
|
|
there are no unflushed modifications in the file */
|
|
|
|
node->modification_counter = node->flush_counter;
|
|
|
|
if (space->is_in_unflushed_spaces
|
|
&& fil_space_is_flushed(space)) {
|
|
|
|
space->is_in_unflushed_spaces = FALSE;
|
|
|
|
UT_LIST_REMOVE(unflushed_spaces,
|
|
system->unflushed_spaces,
|
|
space);
|
|
}
|
|
|
|
fil_node_close_file(node, system);
|
|
}
|
|
|
|
space->size -= node->size;
|
|
|
|
UT_LIST_REMOVE(chain, space->chain, node);
|
|
|
|
mem_free(node->name);
|
|
mem_free(node);
|
|
}
|
|
|
|
#ifdef UNIV_LOG_ARCHIVE
|
|
/****************************************************************//**
|
|
Drops files from the start of a file space, so that its size is cut by
|
|
the amount given. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_space_truncate_start(
|
|
/*=====================*/
|
|
ulint id, /*!< in: space id */
|
|
ulint trunc_len) /*!< in: truncate by this much; it is an error
|
|
if this does not equal to the combined size of
|
|
some initial files in the space */
|
|
{
|
|
fil_node_t* node;
|
|
fil_space_t* space;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
ut_a(space);
|
|
|
|
while (trunc_len > 0) {
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
ut_a(node->size * UNIV_PAGE_SIZE <= trunc_len);
|
|
|
|
trunc_len -= node->size * UNIV_PAGE_SIZE;
|
|
|
|
fil_node_free(node, fil_system, space);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
#endif /* UNIV_LOG_ARCHIVE */
|
|
|
|
/*******************************************************************//**
|
|
Creates a space memory object and puts it to the tablespace memory cache. If
|
|
there is an error, prints an error message to the .err log.
|
|
@return TRUE if success */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_space_create(
|
|
/*=============*/
|
|
const char* name, /*!< in: space name */
|
|
ulint id, /*!< in: space id */
|
|
ulint flags, /*!< in: compressed page size
|
|
and file format, or 0 */
|
|
ulint purpose)/*!< in: FIL_TABLESPACE, or FIL_LOG if log */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
/* The tablespace flags (FSP_SPACE_FLAGS) should be 0 for
|
|
ROW_FORMAT=COMPACT (table->flags == DICT_TF_COMPACT) and
|
|
ROW_FORMAT=REDUNDANT (table->flags == 0). For any other
|
|
format, the tablespace flags should equal table->flags. */
|
|
ut_a(flags != DICT_TF_COMPACT);
|
|
ut_a(!(flags & (~0UL << DICT_TF_BITS)));
|
|
|
|
try_again:
|
|
/*printf(
|
|
"InnoDB: Adding tablespace %lu of name %s, purpose %lu\n", id, name,
|
|
purpose);*/
|
|
|
|
ut_a(fil_system);
|
|
ut_a(name);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_name(name);
|
|
|
|
if (UNIV_LIKELY_NULL(space)) {
|
|
ulint namesake_id;
|
|
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
" InnoDB: Warning: trying to init to the"
|
|
" tablespace memory cache\n"
|
|
"InnoDB: a tablespace %lu of name ", (ulong) id);
|
|
ut_print_filename(stderr, name);
|
|
fprintf(stderr, ",\n"
|
|
"InnoDB: but a tablespace %lu of the same name\n"
|
|
"InnoDB: already exists in the"
|
|
" tablespace memory cache!\n",
|
|
(ulong) space->id);
|
|
|
|
if (id == 0 || purpose != FIL_TABLESPACE) {
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
fprintf(stderr,
|
|
"InnoDB: We assume that InnoDB did a crash recovery,"
|
|
" and you had\n"
|
|
"InnoDB: an .ibd file for which the table"
|
|
" did not exist in the\n"
|
|
"InnoDB: InnoDB internal data dictionary in the"
|
|
" ibdata files.\n"
|
|
"InnoDB: We assume that you later removed the"
|
|
" .ibd and .frm files,\n"
|
|
"InnoDB: and are now trying to recreate the table."
|
|
" We now remove the\n"
|
|
"InnoDB: conflicting tablespace object"
|
|
" from the memory cache and try\n"
|
|
"InnoDB: the init again.\n");
|
|
|
|
namesake_id = space->id;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
fil_space_free(namesake_id, FALSE);
|
|
|
|
goto try_again;
|
|
}
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (UNIV_LIKELY_NULL(space)) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: trying to add tablespace %lu"
|
|
" of name ", (ulong) id);
|
|
ut_print_filename(stderr, name);
|
|
fprintf(stderr, "\n"
|
|
"InnoDB: to the tablespace memory cache,"
|
|
" but tablespace\n"
|
|
"InnoDB: %lu of name ", (ulong) space->id);
|
|
ut_print_filename(stderr, space->name);
|
|
fputs(" already exists in the tablespace\n"
|
|
"InnoDB: memory cache!\n", stderr);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
space = mem_alloc(sizeof(fil_space_t));
|
|
|
|
space->name = mem_strdup(name);
|
|
space->id = id;
|
|
|
|
fil_system->tablespace_version++;
|
|
space->tablespace_version = fil_system->tablespace_version;
|
|
space->mark = FALSE;
|
|
|
|
if (purpose == FIL_TABLESPACE && id > fil_system->max_assigned_id) {
|
|
fil_system->max_assigned_id = id;
|
|
}
|
|
|
|
space->stop_ios = FALSE;
|
|
space->stop_ibuf_merges = FALSE;
|
|
space->is_being_deleted = FALSE;
|
|
space->purpose = purpose;
|
|
space->size = 0;
|
|
space->flags = flags;
|
|
|
|
space->n_reserved_extents = 0;
|
|
|
|
space->n_pending_flushes = 0;
|
|
space->n_pending_ibuf_merges = 0;
|
|
|
|
UT_LIST_INIT(space->chain);
|
|
space->magic_n = FIL_SPACE_MAGIC_N;
|
|
|
|
rw_lock_create(&space->latch, SYNC_FSP);
|
|
|
|
HASH_INSERT(fil_space_t, hash, fil_system->spaces, id, space);
|
|
|
|
HASH_INSERT(fil_space_t, name_hash, fil_system->name_hash,
|
|
ut_fold_string(name), space);
|
|
space->is_in_unflushed_spaces = FALSE;
|
|
|
|
UT_LIST_ADD_LAST(space_list, fil_system->space_list, space);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Assigns a new space id for a new single-table tablespace. This works simply by
|
|
incrementing the global counter. If 4 billion id's is not enough, we may need
|
|
to recycle id's.
|
|
@return new tablespace id; ULINT_UNDEFINED if could not assign an id */
|
|
static
|
|
ulint
|
|
fil_assign_new_space_id(void)
|
|
/*=========================*/
|
|
{
|
|
ulint id;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
fil_system->max_assigned_id++;
|
|
|
|
id = fil_system->max_assigned_id;
|
|
|
|
if (id > (SRV_LOG_SPACE_FIRST_ID / 2) && (id % 1000000UL == 0)) {
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
"InnoDB: Warning: you are running out of new"
|
|
" single-table tablespace id's.\n"
|
|
"InnoDB: Current counter is %lu and it"
|
|
" must not exceed %lu!\n"
|
|
"InnoDB: To reset the counter to zero"
|
|
" you have to dump all your tables and\n"
|
|
"InnoDB: recreate the whole InnoDB installation.\n",
|
|
(ulong) id,
|
|
(ulong) SRV_LOG_SPACE_FIRST_ID);
|
|
}
|
|
|
|
if (id >= SRV_LOG_SPACE_FIRST_ID) {
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
"InnoDB: You have run out of single-table"
|
|
" tablespace id's!\n"
|
|
"InnoDB: Current counter is %lu.\n"
|
|
"InnoDB: To reset the counter to zero you"
|
|
" have to dump all your tables and\n"
|
|
"InnoDB: recreate the whole InnoDB installation.\n",
|
|
(ulong) id);
|
|
fil_system->max_assigned_id--;
|
|
|
|
id = ULINT_UNDEFINED;
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(id);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Frees a space object from the tablespace memory cache. Closes the files in
|
|
the chain but does not delete them. There must not be any pending i/o's or
|
|
flushes on the files.
|
|
@return TRUE if success */
|
|
static
|
|
ibool
|
|
fil_space_free(
|
|
/*===========*/
|
|
/* out: TRUE if success */
|
|
ulint id, /* in: space id */
|
|
ibool own_mutex) /* in: TRUE if own system->mutex */
|
|
{
|
|
fil_space_t* space;
|
|
fil_space_t* namespace;
|
|
fil_node_t* fil_node;
|
|
|
|
if (!own_mutex) {
|
|
mutex_enter(&fil_system->mutex);
|
|
}
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (!space) {
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
" InnoDB: Error: trying to remove tablespace %lu"
|
|
" from the cache but\n"
|
|
"InnoDB: it is not there.\n", (ulong) id);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
HASH_DELETE(fil_space_t, hash, fil_system->spaces, id, space);
|
|
|
|
namespace = fil_space_get_by_name(space->name);
|
|
ut_a(namespace);
|
|
ut_a(space == namespace);
|
|
|
|
HASH_DELETE(fil_space_t, name_hash, fil_system->name_hash,
|
|
ut_fold_string(space->name), space);
|
|
|
|
if (space->is_in_unflushed_spaces) {
|
|
space->is_in_unflushed_spaces = FALSE;
|
|
|
|
UT_LIST_REMOVE(unflushed_spaces, fil_system->unflushed_spaces,
|
|
space);
|
|
}
|
|
|
|
UT_LIST_REMOVE(space_list, fil_system->space_list, space);
|
|
|
|
ut_a(space->magic_n == FIL_SPACE_MAGIC_N);
|
|
ut_a(0 == space->n_pending_flushes);
|
|
|
|
fil_node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
while (fil_node != NULL) {
|
|
fil_node_free(fil_node, fil_system, space);
|
|
|
|
fil_node = UT_LIST_GET_FIRST(space->chain);
|
|
}
|
|
|
|
ut_a(0 == UT_LIST_GET_LEN(space->chain));
|
|
|
|
if (!own_mutex) {
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
|
|
rw_lock_free(&(space->latch));
|
|
|
|
mem_free(space->name);
|
|
mem_free(space);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns the size of the space in pages. The tablespace must be cached in the
|
|
memory cache.
|
|
@return space size, 0 if space not found */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_space_get_size(
|
|
/*===============*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_node_t* node;
|
|
fil_space_t* space;
|
|
ulint size;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
fil_mutex_enter_and_prepare_for_io(id);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space == NULL) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(0);
|
|
}
|
|
|
|
if (space->size == 0 && space->purpose == FIL_TABLESPACE) {
|
|
ut_a(id != 0);
|
|
|
|
ut_a(1 == UT_LIST_GET_LEN(space->chain));
|
|
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
/* It must be a single-table tablespace and we have not opened
|
|
the file yet; the following calls will open it and update the
|
|
size fields */
|
|
|
|
fil_node_prepare_for_io(node, fil_system, space);
|
|
fil_node_complete_io(node, fil_system, OS_FILE_READ);
|
|
}
|
|
|
|
size = space->size;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(size);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns the flags of the space. The tablespace must be cached
|
|
in the memory cache.
|
|
@return flags, ULINT_UNDEFINED if space not found */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_space_get_flags(
|
|
/*================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_node_t* node;
|
|
fil_space_t* space;
|
|
ulint flags;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
if (UNIV_UNLIKELY(!id)) {
|
|
return(0);
|
|
}
|
|
|
|
fil_mutex_enter_and_prepare_for_io(id);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space == NULL) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(ULINT_UNDEFINED);
|
|
}
|
|
|
|
if (space->size == 0 && space->purpose == FIL_TABLESPACE) {
|
|
ut_a(id != 0);
|
|
|
|
ut_a(1 == UT_LIST_GET_LEN(space->chain));
|
|
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
/* It must be a single-table tablespace and we have not opened
|
|
the file yet; the following calls will open it and update the
|
|
size fields */
|
|
|
|
fil_node_prepare_for_io(node, fil_system, space);
|
|
fil_node_complete_io(node, fil_system, OS_FILE_READ);
|
|
}
|
|
|
|
flags = space->flags;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(flags);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns the compressed page size of the space, or 0 if the space
|
|
is not compressed. The tablespace must be cached in the memory cache.
|
|
@return compressed page size, ULINT_UNDEFINED if space not found */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_space_get_zip_size(
|
|
/*===================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
ulint flags;
|
|
|
|
flags = fil_space_get_flags(id);
|
|
|
|
if (flags && flags != ULINT_UNDEFINED) {
|
|
|
|
return(dict_table_flags_to_zip_size(flags));
|
|
}
|
|
|
|
return(flags);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Checks if the pair space, page_no refers to an existing page in a tablespace
|
|
file space. The tablespace must be cached in the memory cache.
|
|
@return TRUE if the address is meaningful */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_check_adress_in_tablespace(
|
|
/*===========================*/
|
|
ulint id, /*!< in: space id */
|
|
ulint page_no)/*!< in: page number */
|
|
{
|
|
if (fil_space_get_size(id) > page_no) {
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
/****************************************************************//**
|
|
Initializes the tablespace memory cache. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_init(
|
|
/*=====*/
|
|
ulint hash_size, /*!< in: hash table size */
|
|
ulint max_n_open) /*!< in: max number of open files */
|
|
{
|
|
ut_a(fil_system == NULL);
|
|
|
|
ut_a(hash_size > 0);
|
|
ut_a(max_n_open > 0);
|
|
|
|
fil_system = mem_alloc(sizeof(fil_system_t));
|
|
|
|
mutex_create(&fil_system->mutex, SYNC_ANY_LATCH);
|
|
|
|
fil_system->spaces = hash_create(hash_size);
|
|
fil_system->name_hash = hash_create(hash_size);
|
|
|
|
UT_LIST_INIT(fil_system->LRU);
|
|
|
|
fil_system->n_open = 0;
|
|
fil_system->max_n_open = max_n_open;
|
|
|
|
fil_system->modification_counter = 0;
|
|
fil_system->max_assigned_id = 0;
|
|
|
|
fil_system->tablespace_version = 0;
|
|
|
|
UT_LIST_INIT(fil_system->unflushed_spaces);
|
|
UT_LIST_INIT(fil_system->space_list);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Opens all log files and system tablespace data files. They stay open until the
|
|
database server shutdown. This should be called at a server startup after the
|
|
space objects for the log and the system tablespace have been created. The
|
|
purpose of this operation is to make sure we never run out of file descriptors
|
|
if we need to read from the insert buffer or to write to the log. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_open_log_and_system_tablespace_files(void)
|
|
/*==========================================*/
|
|
{
|
|
fil_space_t* space;
|
|
fil_node_t* node;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = UT_LIST_GET_FIRST(fil_system->space_list);
|
|
|
|
while (space != NULL) {
|
|
if (space->purpose != FIL_TABLESPACE || space->id == 0) {
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
while (node != NULL) {
|
|
if (!node->open) {
|
|
fil_node_open_file(node, fil_system,
|
|
space);
|
|
}
|
|
if (fil_system->max_n_open
|
|
< 10 + fil_system->n_open) {
|
|
fprintf(stderr,
|
|
"InnoDB: Warning: you must"
|
|
" raise the value of"
|
|
" innodb_open_files in\n"
|
|
"InnoDB: my.cnf! Remember that"
|
|
" InnoDB keeps all log files"
|
|
" and all system\n"
|
|
"InnoDB: tablespace files open"
|
|
" for the whole time mysqld is"
|
|
" running, and\n"
|
|
"InnoDB: needs to open also"
|
|
" some .ibd files if the"
|
|
" file-per-table storage\n"
|
|
"InnoDB: model is used."
|
|
" Current open files %lu,"
|
|
" max allowed"
|
|
" open files %lu.\n",
|
|
(ulong) fil_system->n_open,
|
|
(ulong) fil_system->max_n_open);
|
|
}
|
|
node = UT_LIST_GET_NEXT(chain, node);
|
|
}
|
|
}
|
|
space = UT_LIST_GET_NEXT(space_list, space);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Closes all open files. There must not be any pending i/o's or not flushed
|
|
modifications in the files. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_close_all_files(void)
|
|
/*=====================*/
|
|
{
|
|
fil_space_t* space;
|
|
fil_node_t* node;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = UT_LIST_GET_FIRST(fil_system->space_list);
|
|
|
|
while (space != NULL) {
|
|
fil_space_t* prev_space = space;
|
|
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
while (node != NULL) {
|
|
if (node->open) {
|
|
fil_node_close_file(node, fil_system);
|
|
}
|
|
node = UT_LIST_GET_NEXT(chain, node);
|
|
}
|
|
space = UT_LIST_GET_NEXT(space_list, space);
|
|
fil_space_free(prev_space->id, TRUE);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Sets the max tablespace id counter if the given number is bigger than the
|
|
previous value. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_set_max_space_id_if_bigger(
|
|
/*===========================*/
|
|
ulint max_id) /*!< in: maximum known id */
|
|
{
|
|
if (max_id >= SRV_LOG_SPACE_FIRST_ID) {
|
|
fprintf(stderr,
|
|
"InnoDB: Fatal error: max tablespace id"
|
|
" is too high, %lu\n", (ulong) max_id);
|
|
ut_error;
|
|
}
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
if (fil_system->max_assigned_id < max_id) {
|
|
|
|
fil_system->max_assigned_id = max_id;
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
|
|
/****************************************************************//**
|
|
Writes the flushed lsn and the latest archived log number to the page header
|
|
of the first page of a data file of the system tablespace (space 0),
|
|
which is uncompressed. */
|
|
static
|
|
ulint
|
|
fil_write_lsn_and_arch_no_to_file(
|
|
/*==============================*/
|
|
ulint sum_of_sizes, /*!< in: combined size of previous files
|
|
in space, in database pages */
|
|
ib_uint64_t lsn, /*!< in: lsn to write */
|
|
ulint arch_log_no __attribute__((unused)))
|
|
/*!< in: archived log number to write */
|
|
{
|
|
byte* buf1;
|
|
byte* buf;
|
|
|
|
buf1 = mem_alloc(2 * UNIV_PAGE_SIZE);
|
|
buf = ut_align(buf1, UNIV_PAGE_SIZE);
|
|
|
|
fil_read(TRUE, 0, 0, sum_of_sizes, 0, UNIV_PAGE_SIZE, buf, NULL);
|
|
|
|
mach_write_ull(buf + FIL_PAGE_FILE_FLUSH_LSN, lsn);
|
|
|
|
fil_write(TRUE, 0, 0, sum_of_sizes, 0, UNIV_PAGE_SIZE, buf, NULL);
|
|
|
|
mem_free(buf1);
|
|
|
|
return(DB_SUCCESS);
|
|
}
|
|
|
|
/****************************************************************//**
|
|
Writes the flushed lsn and the latest archived log number to the page
|
|
header of the first page of each data file in the system tablespace.
|
|
@return DB_SUCCESS or error number */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_write_flushed_lsn_to_data_files(
|
|
/*================================*/
|
|
ib_uint64_t lsn, /*!< in: lsn to write */
|
|
ulint arch_log_no) /*!< in: latest archived log
|
|
file number */
|
|
{
|
|
fil_space_t* space;
|
|
fil_node_t* node;
|
|
ulint sum_of_sizes;
|
|
ulint err;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = UT_LIST_GET_FIRST(fil_system->space_list);
|
|
|
|
while (space) {
|
|
/* We only write the lsn to all existing data files which have
|
|
been open during the lifetime of the mysqld process; they are
|
|
represented by the space objects in the tablespace memory
|
|
cache. Note that all data files in the system tablespace 0 are
|
|
always open. */
|
|
|
|
if (space->purpose == FIL_TABLESPACE
|
|
&& space->id == 0) {
|
|
sum_of_sizes = 0;
|
|
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
while (node) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
err = fil_write_lsn_and_arch_no_to_file(
|
|
sum_of_sizes, lsn, arch_log_no);
|
|
if (err != DB_SUCCESS) {
|
|
|
|
return(err);
|
|
}
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
sum_of_sizes += node->size;
|
|
node = UT_LIST_GET_NEXT(chain, node);
|
|
}
|
|
}
|
|
space = UT_LIST_GET_NEXT(space_list, space);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(DB_SUCCESS);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Reads the flushed lsn and arch no fields from a data file at database
|
|
startup. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_read_flushed_lsn_and_arch_log_no(
|
|
/*=================================*/
|
|
os_file_t data_file, /*!< in: open data file */
|
|
ibool one_read_already, /*!< in: TRUE if min and max
|
|
parameters below already
|
|
contain sensible data */
|
|
#ifdef UNIV_LOG_ARCHIVE
|
|
ulint* min_arch_log_no, /*!< in/out: */
|
|
ulint* max_arch_log_no, /*!< in/out: */
|
|
#endif /* UNIV_LOG_ARCHIVE */
|
|
ib_uint64_t* min_flushed_lsn, /*!< in/out: */
|
|
ib_uint64_t* max_flushed_lsn) /*!< in/out: */
|
|
{
|
|
byte* buf;
|
|
byte* buf2;
|
|
ib_uint64_t flushed_lsn;
|
|
|
|
buf2 = ut_malloc(2 * UNIV_PAGE_SIZE);
|
|
/* Align the memory for a possible read from a raw device */
|
|
buf = ut_align(buf2, UNIV_PAGE_SIZE);
|
|
|
|
os_file_read(data_file, buf, 0, 0, UNIV_PAGE_SIZE);
|
|
|
|
flushed_lsn = mach_read_ull(buf + FIL_PAGE_FILE_FLUSH_LSN);
|
|
|
|
ut_free(buf2);
|
|
|
|
if (!one_read_already) {
|
|
*min_flushed_lsn = flushed_lsn;
|
|
*max_flushed_lsn = flushed_lsn;
|
|
#ifdef UNIV_LOG_ARCHIVE
|
|
*min_arch_log_no = arch_log_no;
|
|
*max_arch_log_no = arch_log_no;
|
|
#endif /* UNIV_LOG_ARCHIVE */
|
|
return;
|
|
}
|
|
|
|
if (*min_flushed_lsn > flushed_lsn) {
|
|
*min_flushed_lsn = flushed_lsn;
|
|
}
|
|
if (*max_flushed_lsn < flushed_lsn) {
|
|
*max_flushed_lsn = flushed_lsn;
|
|
}
|
|
#ifdef UNIV_LOG_ARCHIVE
|
|
if (*min_arch_log_no > arch_log_no) {
|
|
*min_arch_log_no = arch_log_no;
|
|
}
|
|
if (*max_arch_log_no < arch_log_no) {
|
|
*max_arch_log_no = arch_log_no;
|
|
}
|
|
#endif /* UNIV_LOG_ARCHIVE */
|
|
}
|
|
|
|
/*================ SINGLE-TABLE TABLESPACES ==========================*/
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
/*******************************************************************//**
|
|
Increments the count of pending insert buffer page merges, if space is not
|
|
being deleted.
|
|
@return TRUE if being deleted, and ibuf merges should be skipped */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_inc_pending_ibuf_merges(
|
|
/*========================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space == NULL) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: trying to do ibuf merge to a"
|
|
" dropped tablespace %lu\n",
|
|
(ulong) id);
|
|
}
|
|
|
|
if (space == NULL || space->stop_ibuf_merges) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
space->n_pending_ibuf_merges++;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Decrements the count of pending insert buffer page merges. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_decr_pending_ibuf_merges(
|
|
/*=========================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space == NULL) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: decrementing ibuf merge of a"
|
|
" dropped tablespace %lu\n",
|
|
(ulong) id);
|
|
}
|
|
|
|
if (space != NULL) {
|
|
space->n_pending_ibuf_merges--;
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
|
|
/********************************************************//**
|
|
Creates the database directory for a table if it does not exist yet. */
|
|
static
|
|
void
|
|
fil_create_directory_for_tablename(
|
|
/*===============================*/
|
|
const char* name) /*!< in: name in the standard
|
|
'databasename/tablename' format */
|
|
{
|
|
const char* namend;
|
|
char* path;
|
|
ulint len;
|
|
|
|
len = strlen(fil_path_to_mysql_datadir);
|
|
namend = strchr(name, '/');
|
|
ut_a(namend);
|
|
path = mem_alloc(len + (namend - name) + 2);
|
|
|
|
memcpy(path, fil_path_to_mysql_datadir, len);
|
|
path[len] = '/';
|
|
memcpy(path + len + 1, name, namend - name);
|
|
path[len + (namend - name) + 1] = 0;
|
|
|
|
srv_normalize_path_for_win(path);
|
|
|
|
ut_a(os_file_create_directory(path, FALSE));
|
|
mem_free(path);
|
|
}
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
/********************************************************//**
|
|
Writes a log record about an .ibd file create/rename/delete. */
|
|
static
|
|
void
|
|
fil_op_write_log(
|
|
/*=============*/
|
|
ulint type, /*!< in: MLOG_FILE_CREATE,
|
|
MLOG_FILE_CREATE2,
|
|
MLOG_FILE_DELETE, or
|
|
MLOG_FILE_RENAME */
|
|
ulint space_id, /*!< in: space id */
|
|
ulint log_flags, /*!< in: redo log flags (stored
|
|
in the page number field) */
|
|
ulint flags, /*!< in: compressed page size
|
|
and file format
|
|
if type==MLOG_FILE_CREATE2, or 0 */
|
|
const char* name, /*!< in: table name in the familiar
|
|
'databasename/tablename' format, or
|
|
the file path in the case of
|
|
MLOG_FILE_DELETE */
|
|
const char* new_name, /*!< in: if type is MLOG_FILE_RENAME,
|
|
the new table name in the
|
|
'databasename/tablename' format */
|
|
mtr_t* mtr) /*!< in: mini-transaction handle */
|
|
{
|
|
byte* log_ptr;
|
|
ulint len;
|
|
|
|
log_ptr = mlog_open(mtr, 11 + 2 + 1);
|
|
|
|
if (!log_ptr) {
|
|
/* Logging in mtr is switched off during crash recovery:
|
|
in that case mlog_open returns NULL */
|
|
return;
|
|
}
|
|
|
|
log_ptr = mlog_write_initial_log_record_for_file_op(
|
|
type, space_id, log_flags, log_ptr, mtr);
|
|
if (type == MLOG_FILE_CREATE2) {
|
|
mach_write_to_4(log_ptr, flags);
|
|
log_ptr += 4;
|
|
}
|
|
/* Let us store the strings as null-terminated for easier readability
|
|
and handling */
|
|
|
|
len = strlen(name) + 1;
|
|
|
|
mach_write_to_2(log_ptr, len);
|
|
log_ptr += 2;
|
|
mlog_close(mtr, log_ptr);
|
|
|
|
mlog_catenate_string(mtr, (byte*) name, len);
|
|
|
|
if (type == MLOG_FILE_RENAME) {
|
|
len = strlen(new_name) + 1;
|
|
log_ptr = mlog_open(mtr, 2 + len);
|
|
ut_a(log_ptr);
|
|
mach_write_to_2(log_ptr, len);
|
|
log_ptr += 2;
|
|
mlog_close(mtr, log_ptr);
|
|
|
|
mlog_catenate_string(mtr, (byte*) new_name, len);
|
|
}
|
|
}
|
|
#endif
|
|
|
|
/*******************************************************************//**
|
|
Parses the body of a log record written about an .ibd file operation. That is,
|
|
the log record part after the standard (type, space id, page no) header of the
|
|
log record.
|
|
|
|
If desired, also replays the delete or rename operation if the .ibd file
|
|
exists and the space id in it matches. Replays the create operation if a file
|
|
at that path does not exist yet. If the database directory for the file to be
|
|
created does not exist, then we create the directory, too.
|
|
|
|
Note that ibbackup --apply-log sets fil_path_to_mysql_datadir to point to the
|
|
datadir that we should use in replaying the file operations.
|
|
@return end of log record, or NULL if the record was not completely
|
|
contained between ptr and end_ptr */
|
|
UNIV_INTERN
|
|
byte*
|
|
fil_op_log_parse_or_replay(
|
|
/*=======================*/
|
|
byte* ptr, /*!< in: buffer containing the log record body,
|
|
or an initial segment of it, if the record does
|
|
not fir completely between ptr and end_ptr */
|
|
byte* end_ptr, /*!< in: buffer end */
|
|
ulint type, /*!< in: the type of this log record */
|
|
ulint space_id, /*!< in: the space id of the tablespace in
|
|
question, or 0 if the log record should
|
|
only be parsed but not replayed */
|
|
ulint log_flags) /*!< in: redo log flags
|
|
(stored in the page number parameter) */
|
|
{
|
|
ulint name_len;
|
|
ulint new_name_len;
|
|
const char* name;
|
|
const char* new_name = NULL;
|
|
ulint flags = 0;
|
|
|
|
if (type == MLOG_FILE_CREATE2) {
|
|
if (end_ptr < ptr + 4) {
|
|
|
|
return(NULL);
|
|
}
|
|
|
|
flags = mach_read_from_4(ptr);
|
|
ptr += 4;
|
|
}
|
|
|
|
if (end_ptr < ptr + 2) {
|
|
|
|
return(NULL);
|
|
}
|
|
|
|
name_len = mach_read_from_2(ptr);
|
|
|
|
ptr += 2;
|
|
|
|
if (end_ptr < ptr + name_len) {
|
|
|
|
return(NULL);
|
|
}
|
|
|
|
name = (const char*) ptr;
|
|
|
|
ptr += name_len;
|
|
|
|
if (type == MLOG_FILE_RENAME) {
|
|
if (end_ptr < ptr + 2) {
|
|
|
|
return(NULL);
|
|
}
|
|
|
|
new_name_len = mach_read_from_2(ptr);
|
|
|
|
ptr += 2;
|
|
|
|
if (end_ptr < ptr + new_name_len) {
|
|
|
|
return(NULL);
|
|
}
|
|
|
|
new_name = (const char*) ptr;
|
|
|
|
ptr += new_name_len;
|
|
}
|
|
|
|
/* We managed to parse a full log record body */
|
|
/*
|
|
printf("Parsed log rec of type %lu space %lu\n"
|
|
"name %s\n", type, space_id, name);
|
|
|
|
if (type == MLOG_FILE_RENAME) {
|
|
printf("new name %s\n", new_name);
|
|
}
|
|
*/
|
|
if (!space_id) {
|
|
|
|
return(ptr);
|
|
}
|
|
|
|
/* Let us try to perform the file operation, if sensible. Note that
|
|
ibbackup has at this stage already read in all space id info to the
|
|
fil0fil.c data structures.
|
|
|
|
NOTE that our algorithm is not guaranteed to work correctly if there
|
|
were renames of tables during the backup. See ibbackup code for more
|
|
on the problem. */
|
|
|
|
switch (type) {
|
|
case MLOG_FILE_DELETE:
|
|
if (fil_tablespace_exists_in_mem(space_id)) {
|
|
ut_a(fil_delete_tablespace(space_id));
|
|
}
|
|
|
|
break;
|
|
|
|
case MLOG_FILE_RENAME:
|
|
/* We do the rename based on space id, not old file name;
|
|
this should guarantee that after the log replay each .ibd file
|
|
has the correct name for the latest log sequence number; the
|
|
proof is left as an exercise :) */
|
|
|
|
if (fil_tablespace_exists_in_mem(space_id)) {
|
|
/* Create the database directory for the new name, if
|
|
it does not exist yet */
|
|
fil_create_directory_for_tablename(new_name);
|
|
|
|
/* Rename the table if there is not yet a tablespace
|
|
with the same name */
|
|
|
|
if (fil_get_space_id_for_table(new_name)
|
|
== ULINT_UNDEFINED) {
|
|
/* We do not care of the old name, that is
|
|
why we pass NULL as the first argument */
|
|
if (!fil_rename_tablespace(NULL, space_id,
|
|
new_name)) {
|
|
ut_error;
|
|
}
|
|
}
|
|
}
|
|
|
|
break;
|
|
|
|
case MLOG_FILE_CREATE:
|
|
case MLOG_FILE_CREATE2:
|
|
if (fil_tablespace_exists_in_mem(space_id)) {
|
|
/* Do nothing */
|
|
} else if (fil_get_space_id_for_table(name)
|
|
!= ULINT_UNDEFINED) {
|
|
/* Do nothing */
|
|
} else if (log_flags & MLOG_FILE_FLAG_TEMP) {
|
|
/* Temporary table, do nothing */
|
|
} else {
|
|
/* Create the database directory for name, if it does
|
|
not exist yet */
|
|
fil_create_directory_for_tablename(name);
|
|
|
|
if (fil_create_new_single_table_tablespace(
|
|
&space_id, name, FALSE, flags,
|
|
FIL_IBD_FILE_INITIAL_SIZE) != DB_SUCCESS) {
|
|
ut_error;
|
|
}
|
|
}
|
|
|
|
break;
|
|
|
|
default:
|
|
ut_error;
|
|
}
|
|
|
|
return(ptr);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Deletes a single-table tablespace. The tablespace must be cached in the
|
|
memory cache.
|
|
@return TRUE if success */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_delete_tablespace(
|
|
/*==================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
ibool success;
|
|
fil_space_t* space;
|
|
fil_node_t* node;
|
|
ulint count = 0;
|
|
char* path;
|
|
|
|
ut_a(id != 0);
|
|
stop_ibuf_merges:
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space != NULL) {
|
|
space->stop_ibuf_merges = TRUE;
|
|
|
|
if (space->n_pending_ibuf_merges == 0) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
count = 0;
|
|
|
|
goto try_again;
|
|
} else {
|
|
if (count > 5000) {
|
|
ut_print_timestamp(stderr);
|
|
fputs(" InnoDB: Warning: trying to"
|
|
" delete tablespace ", stderr);
|
|
ut_print_filename(stderr, space->name);
|
|
fprintf(stderr, ",\n"
|
|
"InnoDB: but there are %lu pending"
|
|
" ibuf merges on it.\n"
|
|
"InnoDB: Loop %lu.\n",
|
|
(ulong) space->n_pending_ibuf_merges,
|
|
(ulong) count);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
os_thread_sleep(20000);
|
|
count++;
|
|
|
|
goto stop_ibuf_merges;
|
|
}
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
count = 0;
|
|
|
|
try_again:
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space == NULL) {
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
" InnoDB: Error: cannot delete tablespace %lu\n"
|
|
"InnoDB: because it is not found in the"
|
|
" tablespace memory cache.\n",
|
|
(ulong) id);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
ut_a(space);
|
|
ut_a(space->n_pending_ibuf_merges == 0);
|
|
|
|
space->is_being_deleted = TRUE;
|
|
|
|
ut_a(UT_LIST_GET_LEN(space->chain) == 1);
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
if (space->n_pending_flushes > 0 || node->n_pending > 0) {
|
|
if (count > 1000) {
|
|
ut_print_timestamp(stderr);
|
|
fputs(" InnoDB: Warning: trying to"
|
|
" delete tablespace ", stderr);
|
|
ut_print_filename(stderr, space->name);
|
|
fprintf(stderr, ",\n"
|
|
"InnoDB: but there are %lu flushes"
|
|
" and %lu pending i/o's on it\n"
|
|
"InnoDB: Loop %lu.\n",
|
|
(ulong) space->n_pending_flushes,
|
|
(ulong) node->n_pending,
|
|
(ulong) count);
|
|
}
|
|
mutex_exit(&fil_system->mutex);
|
|
os_thread_sleep(20000);
|
|
|
|
count++;
|
|
|
|
goto try_again;
|
|
}
|
|
|
|
path = mem_strdup(space->name);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
#ifndef UNIV_HOTBACKUP
|
|
/* Invalidate in the buffer pool all pages belonging to the
|
|
tablespace. Since we have set space->is_being_deleted = TRUE, readahead
|
|
or ibuf merge can no longer read more pages of this tablespace to the
|
|
buffer pool. Thus we can clean the tablespace out of the buffer pool
|
|
completely and permanently. The flag is_being_deleted also prevents
|
|
fil_flush() from being applied to this tablespace. */
|
|
|
|
buf_LRU_invalidate_tablespace(id);
|
|
#endif
|
|
/* printf("Deleting tablespace %s id %lu\n", space->name, id); */
|
|
|
|
success = fil_space_free(id, FALSE);
|
|
|
|
if (success) {
|
|
success = os_file_delete(path);
|
|
|
|
if (!success) {
|
|
success = os_file_delete_if_exists(path);
|
|
}
|
|
}
|
|
|
|
if (success) {
|
|
#ifndef UNIV_HOTBACKUP
|
|
/* Write a log record about the deletion of the .ibd
|
|
file, so that ibbackup can replay it in the
|
|
--apply-log phase. We use a dummy mtr and the familiar
|
|
log write mechanism. */
|
|
mtr_t mtr;
|
|
|
|
/* When replaying the operation in ibbackup, do not try
|
|
to write any log record */
|
|
mtr_start(&mtr);
|
|
|
|
fil_op_write_log(MLOG_FILE_DELETE, id, 0, 0, path, NULL, &mtr);
|
|
mtr_commit(&mtr);
|
|
#endif
|
|
mem_free(path);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
mem_free(path);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
/*******************************************************************//**
|
|
Discards a single-table tablespace. The tablespace must be cached in the
|
|
memory cache. Discarding is like deleting a tablespace, but
|
|
1) we do not drop the table from the data dictionary;
|
|
2) we remove all insert buffer entries for the tablespace immediately; in DROP
|
|
TABLE they are only removed gradually in the background;
|
|
3) when the user does IMPORT TABLESPACE, the tablespace will have the same id
|
|
as it originally had.
|
|
@return TRUE if success */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_discard_tablespace(
|
|
/*===================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
ibool success;
|
|
|
|
success = fil_delete_tablespace(id);
|
|
|
|
if (!success) {
|
|
fprintf(stderr,
|
|
"InnoDB: Warning: cannot delete tablespace %lu"
|
|
" in DISCARD TABLESPACE.\n"
|
|
"InnoDB: But let us remove the"
|
|
" insert buffer entries for this tablespace.\n",
|
|
(ulong) id);
|
|
}
|
|
|
|
/* Remove all insert buffer entries for the tablespace */
|
|
|
|
ibuf_delete_for_discarded_space(id);
|
|
|
|
return(success);
|
|
}
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
|
|
/*******************************************************************//**
|
|
Renames the memory cache structures of a single-table tablespace.
|
|
@return TRUE if success */
|
|
static
|
|
ibool
|
|
fil_rename_tablespace_in_mem(
|
|
/*=========================*/
|
|
fil_space_t* space, /*!< in: tablespace memory object */
|
|
fil_node_t* node, /*!< in: file node of that tablespace */
|
|
const char* path) /*!< in: new name */
|
|
{
|
|
fil_space_t* space2;
|
|
const char* old_name = space->name;
|
|
|
|
ut_ad(mutex_own(&fil_system->mutex));
|
|
|
|
space2 = fil_space_get_by_name(old_name);
|
|
if (space != space2) {
|
|
fputs("InnoDB: Error: cannot find ", stderr);
|
|
ut_print_filename(stderr, old_name);
|
|
fputs(" in tablespace memory cache\n", stderr);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
space2 = fil_space_get_by_name(path);
|
|
if (space2 != NULL) {
|
|
fputs("InnoDB: Error: ", stderr);
|
|
ut_print_filename(stderr, path);
|
|
fputs(" is already in tablespace memory cache\n", stderr);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
HASH_DELETE(fil_space_t, name_hash, fil_system->name_hash,
|
|
ut_fold_string(space->name), space);
|
|
mem_free(space->name);
|
|
mem_free(node->name);
|
|
|
|
space->name = mem_strdup(path);
|
|
node->name = mem_strdup(path);
|
|
|
|
HASH_INSERT(fil_space_t, name_hash, fil_system->name_hash,
|
|
ut_fold_string(path), space);
|
|
return(TRUE);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Allocates a file name for a single-table tablespace. The string must be freed
|
|
by caller with mem_free().
|
|
@return own: file name */
|
|
static
|
|
char*
|
|
fil_make_ibd_name(
|
|
/*==============*/
|
|
const char* name, /*!< in: table name or a dir path of a
|
|
TEMPORARY table */
|
|
ibool is_temp) /*!< in: TRUE if it is a dir path */
|
|
{
|
|
ulint namelen = strlen(name);
|
|
ulint dirlen = strlen(fil_path_to_mysql_datadir);
|
|
char* filename = mem_alloc(namelen + dirlen + sizeof "/.ibd");
|
|
|
|
if (is_temp) {
|
|
memcpy(filename, name, namelen);
|
|
memcpy(filename + namelen, ".ibd", sizeof ".ibd");
|
|
} else {
|
|
memcpy(filename, fil_path_to_mysql_datadir, dirlen);
|
|
filename[dirlen] = '/';
|
|
|
|
memcpy(filename + dirlen + 1, name, namelen);
|
|
memcpy(filename + dirlen + namelen + 1, ".ibd", sizeof ".ibd");
|
|
}
|
|
|
|
srv_normalize_path_for_win(filename);
|
|
|
|
return(filename);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Renames a single-table tablespace. The tablespace must be cached in the
|
|
tablespace memory cache.
|
|
@return TRUE if success */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_rename_tablespace(
|
|
/*==================*/
|
|
const char* old_name, /*!< in: old table name in the standard
|
|
databasename/tablename format of
|
|
InnoDB, or NULL if we do the rename
|
|
based on the space id only */
|
|
ulint id, /*!< in: space id */
|
|
const char* new_name) /*!< in: new table name in the standard
|
|
databasename/tablename format
|
|
of InnoDB */
|
|
{
|
|
ibool success;
|
|
fil_space_t* space;
|
|
fil_node_t* node;
|
|
ulint count = 0;
|
|
char* path;
|
|
ibool old_name_was_specified = TRUE;
|
|
char* old_path;
|
|
|
|
ut_a(id != 0);
|
|
|
|
if (old_name == NULL) {
|
|
old_name = "(name not specified)";
|
|
old_name_was_specified = FALSE;
|
|
}
|
|
retry:
|
|
count++;
|
|
|
|
if (count > 1000) {
|
|
ut_print_timestamp(stderr);
|
|
fputs(" InnoDB: Warning: problems renaming ", stderr);
|
|
ut_print_filename(stderr, old_name);
|
|
fputs(" to ", stderr);
|
|
ut_print_filename(stderr, new_name);
|
|
fprintf(stderr, ", %lu iterations\n", (ulong) count);
|
|
}
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space == NULL) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: cannot find space id %lu"
|
|
" in the tablespace memory cache\n"
|
|
"InnoDB: though the table ", (ulong) id);
|
|
ut_print_filename(stderr, old_name);
|
|
fputs(" in a rename operation should have that id\n", stderr);
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
if (count > 25000) {
|
|
space->stop_ios = FALSE;
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
/* We temporarily close the .ibd file because we do not trust that
|
|
operating systems can rename an open file. For the closing we have to
|
|
wait until there are no pending i/o's or flushes on the file. */
|
|
|
|
space->stop_ios = TRUE;
|
|
|
|
ut_a(UT_LIST_GET_LEN(space->chain) == 1);
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
if (node->n_pending > 0 || node->n_pending_flushes > 0) {
|
|
/* There are pending i/o's or flushes, sleep for a while and
|
|
retry */
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
os_thread_sleep(20000);
|
|
|
|
goto retry;
|
|
|
|
} else if (node->modification_counter > node->flush_counter) {
|
|
/* Flush the space */
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
os_thread_sleep(20000);
|
|
|
|
fil_flush(id);
|
|
|
|
goto retry;
|
|
|
|
} else if (node->open) {
|
|
/* Close the file */
|
|
|
|
fil_node_close_file(node, fil_system);
|
|
}
|
|
|
|
/* Check that the old name in the space is right */
|
|
|
|
if (old_name_was_specified) {
|
|
old_path = fil_make_ibd_name(old_name, FALSE);
|
|
|
|
ut_a(strcmp(space->name, old_path) == 0);
|
|
ut_a(strcmp(node->name, old_path) == 0);
|
|
} else {
|
|
old_path = mem_strdup(space->name);
|
|
}
|
|
|
|
/* Rename the tablespace and the node in the memory cache */
|
|
path = fil_make_ibd_name(new_name, FALSE);
|
|
success = fil_rename_tablespace_in_mem(space, node, path);
|
|
|
|
if (success) {
|
|
success = os_file_rename(old_path, path);
|
|
|
|
if (!success) {
|
|
/* We have to revert the changes we made
|
|
to the tablespace memory cache */
|
|
|
|
ut_a(fil_rename_tablespace_in_mem(space, node,
|
|
old_path));
|
|
}
|
|
}
|
|
|
|
mem_free(path);
|
|
mem_free(old_path);
|
|
|
|
space->stop_ios = FALSE;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
if (success) {
|
|
mtr_t mtr;
|
|
|
|
mtr_start(&mtr);
|
|
|
|
fil_op_write_log(MLOG_FILE_RENAME, id, 0, 0, old_name, new_name,
|
|
&mtr);
|
|
mtr_commit(&mtr);
|
|
}
|
|
#endif
|
|
return(success);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Creates a new single-table tablespace to a database directory of MySQL.
|
|
Database directories are under the 'datadir' of MySQL. The datadir is the
|
|
directory of a running mysqld program. We can refer to it by simply the
|
|
path '.'. Tables created with CREATE TEMPORARY TABLE we place in the temp
|
|
dir of the mysqld server.
|
|
@return DB_SUCCESS or error code */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_create_new_single_table_tablespace(
|
|
/*===================================*/
|
|
ulint* space_id, /*!< in/out: space id; if this is != 0,
|
|
then this is an input parameter,
|
|
otherwise output */
|
|
const char* tablename, /*!< in: the table name in the usual
|
|
databasename/tablename format
|
|
of InnoDB, or a dir path to a temp
|
|
table */
|
|
ibool is_temp, /*!< in: TRUE if a table created with
|
|
CREATE TEMPORARY TABLE */
|
|
ulint flags, /*!< in: tablespace flags */
|
|
ulint size) /*!< in: the initial size of the
|
|
tablespace file in pages,
|
|
must be >= FIL_IBD_FILE_INITIAL_SIZE */
|
|
{
|
|
os_file_t file;
|
|
ibool ret;
|
|
ulint err;
|
|
byte* buf2;
|
|
byte* page;
|
|
ibool success;
|
|
char* path;
|
|
|
|
ut_a(size >= FIL_IBD_FILE_INITIAL_SIZE);
|
|
/* The tablespace flags (FSP_SPACE_FLAGS) should be 0 for
|
|
ROW_FORMAT=COMPACT (table->flags == DICT_TF_COMPACT) and
|
|
ROW_FORMAT=REDUNDANT (table->flags == 0). For any other
|
|
format, the tablespace flags should equal table->flags. */
|
|
ut_a(flags != DICT_TF_COMPACT);
|
|
ut_a(!(flags & (~0UL << DICT_TF_BITS)));
|
|
|
|
path = fil_make_ibd_name(tablename, is_temp);
|
|
|
|
file = os_file_create(path, OS_FILE_CREATE, OS_FILE_NORMAL,
|
|
OS_DATA_FILE, &ret);
|
|
if (ret == FALSE) {
|
|
ut_print_timestamp(stderr);
|
|
fputs(" InnoDB: Error creating file ", stderr);
|
|
ut_print_filename(stderr, path);
|
|
fputs(".\n", stderr);
|
|
|
|
/* The following call will print an error message */
|
|
|
|
err = os_file_get_last_error(TRUE);
|
|
|
|
if (err == OS_FILE_ALREADY_EXISTS) {
|
|
fputs("InnoDB: The file already exists though"
|
|
" the corresponding table did not\n"
|
|
"InnoDB: exist in the InnoDB data dictionary."
|
|
" Have you moved InnoDB\n"
|
|
"InnoDB: .ibd files around without using the"
|
|
" SQL commands\n"
|
|
"InnoDB: DISCARD TABLESPACE and"
|
|
" IMPORT TABLESPACE, or did\n"
|
|
"InnoDB: mysqld crash in the middle of"
|
|
" CREATE TABLE? You can\n"
|
|
"InnoDB: resolve the problem by"
|
|
" removing the file ", stderr);
|
|
ut_print_filename(stderr, path);
|
|
fputs("\n"
|
|
"InnoDB: under the 'datadir' of MySQL.\n",
|
|
stderr);
|
|
|
|
mem_free(path);
|
|
return(DB_TABLESPACE_ALREADY_EXISTS);
|
|
}
|
|
|
|
if (err == OS_FILE_DISK_FULL) {
|
|
|
|
mem_free(path);
|
|
return(DB_OUT_OF_FILE_SPACE);
|
|
}
|
|
|
|
mem_free(path);
|
|
return(DB_ERROR);
|
|
}
|
|
|
|
buf2 = ut_malloc(3 * UNIV_PAGE_SIZE);
|
|
/* Align the memory for file i/o if we might have O_DIRECT set */
|
|
page = ut_align(buf2, UNIV_PAGE_SIZE);
|
|
|
|
ret = os_file_set_size(path, file, size * UNIV_PAGE_SIZE, 0);
|
|
|
|
if (!ret) {
|
|
ut_free(buf2);
|
|
os_file_close(file);
|
|
os_file_delete(path);
|
|
|
|
mem_free(path);
|
|
return(DB_OUT_OF_FILE_SPACE);
|
|
}
|
|
|
|
if (*space_id == 0) {
|
|
*space_id = fil_assign_new_space_id();
|
|
}
|
|
|
|
/* printf("Creating tablespace %s id %lu\n", path, *space_id); */
|
|
|
|
if (*space_id == ULINT_UNDEFINED) {
|
|
ut_free(buf2);
|
|
error_exit:
|
|
os_file_close(file);
|
|
error_exit2:
|
|
os_file_delete(path);
|
|
|
|
mem_free(path);
|
|
return(DB_ERROR);
|
|
}
|
|
|
|
/* We have to write the space id to the file immediately and flush the
|
|
file to disk. This is because in crash recovery we must be aware what
|
|
tablespaces exist and what are their space id's, so that we can apply
|
|
the log records to the right file. It may take quite a while until
|
|
buffer pool flush algorithms write anything to the file and flush it to
|
|
disk. If we would not write here anything, the file would be filled
|
|
with zeros from the call of os_file_set_size(), until a buffer pool
|
|
flush would write to it. */
|
|
|
|
memset(page, '\0', UNIV_PAGE_SIZE);
|
|
|
|
fsp_header_init_fields(page, *space_id, flags);
|
|
mach_write_to_4(page + FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID, *space_id);
|
|
|
|
if (!(flags & DICT_TF_ZSSIZE_MASK)) {
|
|
buf_flush_init_for_writing(page, NULL, 0);
|
|
ret = os_file_write(path, file, page, 0, 0, UNIV_PAGE_SIZE);
|
|
} else {
|
|
page_zip_des_t page_zip;
|
|
ulint zip_size;
|
|
|
|
zip_size = ((PAGE_ZIP_MIN_SIZE >> 1)
|
|
<< ((flags & DICT_TF_ZSSIZE_MASK)
|
|
>> DICT_TF_ZSSIZE_SHIFT));
|
|
|
|
page_zip_set_size(&page_zip, zip_size);
|
|
page_zip.data = page + UNIV_PAGE_SIZE;
|
|
#ifdef UNIV_DEBUG
|
|
page_zip.m_start =
|
|
#endif /* UNIV_DEBUG */
|
|
page_zip.m_end = page_zip.m_nonempty =
|
|
page_zip.n_blobs = 0;
|
|
buf_flush_init_for_writing(page, &page_zip, 0);
|
|
ret = os_file_write(path, file, page_zip.data, 0, 0, zip_size);
|
|
}
|
|
|
|
ut_free(buf2);
|
|
|
|
if (!ret) {
|
|
fputs("InnoDB: Error: could not write the first page"
|
|
" to tablespace ", stderr);
|
|
ut_print_filename(stderr, path);
|
|
putc('\n', stderr);
|
|
goto error_exit;
|
|
}
|
|
|
|
ret = os_file_flush(file);
|
|
|
|
if (!ret) {
|
|
fputs("InnoDB: Error: file flush of tablespace ", stderr);
|
|
ut_print_filename(stderr, path);
|
|
fputs(" failed\n", stderr);
|
|
goto error_exit;
|
|
}
|
|
|
|
os_file_close(file);
|
|
|
|
if (*space_id == ULINT_UNDEFINED) {
|
|
goto error_exit2;
|
|
}
|
|
|
|
success = fil_space_create(path, *space_id, flags, FIL_TABLESPACE);
|
|
|
|
if (!success) {
|
|
goto error_exit2;
|
|
}
|
|
|
|
fil_node_create(path, size, *space_id, FALSE);
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
{
|
|
mtr_t mtr;
|
|
|
|
mtr_start(&mtr);
|
|
|
|
fil_op_write_log(flags
|
|
? MLOG_FILE_CREATE2
|
|
: MLOG_FILE_CREATE,
|
|
*space_id,
|
|
is_temp ? MLOG_FILE_FLAG_TEMP : 0,
|
|
flags,
|
|
tablename, NULL, &mtr);
|
|
|
|
mtr_commit(&mtr);
|
|
}
|
|
#endif
|
|
mem_free(path);
|
|
return(DB_SUCCESS);
|
|
}
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
/********************************************************************//**
|
|
It is possible, though very improbable, that the lsn's in the tablespace to be
|
|
imported have risen above the current system lsn, if a lengthy purge, ibuf
|
|
merge, or rollback was performed on a backup taken with ibbackup. If that is
|
|
the case, reset page lsn's in the file. We assume that mysqld was shut down
|
|
after it performed these cleanup operations on the .ibd file, so that it at
|
|
the shutdown stamped the latest lsn to the FIL_PAGE_FILE_FLUSH_LSN in the
|
|
first page of the .ibd file, and we can determine whether we need to reset the
|
|
lsn's just by looking at that flush lsn.
|
|
@return TRUE if success */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_reset_too_high_lsns(
|
|
/*====================*/
|
|
const char* name, /*!< in: table name in the
|
|
databasename/tablename format */
|
|
ib_uint64_t current_lsn) /*!< in: reset lsn's if the lsn stamped
|
|
to FIL_PAGE_FILE_FLUSH_LSN in the
|
|
first page is too high */
|
|
{
|
|
os_file_t file;
|
|
char* filepath;
|
|
byte* page;
|
|
byte* buf2;
|
|
ib_uint64_t flush_lsn;
|
|
ulint space_id;
|
|
ib_int64_t file_size;
|
|
ib_int64_t offset;
|
|
ulint zip_size;
|
|
ibool success;
|
|
|
|
filepath = fil_make_ibd_name(name, FALSE);
|
|
|
|
file = os_file_create_simple_no_error_handling(
|
|
filepath, OS_FILE_OPEN, OS_FILE_READ_WRITE, &success);
|
|
if (!success) {
|
|
/* The following call prints an error message */
|
|
os_file_get_last_error(TRUE);
|
|
|
|
ut_print_timestamp(stderr);
|
|
|
|
fputs(" InnoDB: Error: trying to open a table,"
|
|
" but could not\n"
|
|
"InnoDB: open the tablespace file ", stderr);
|
|
ut_print_filename(stderr, filepath);
|
|
fputs("!\n", stderr);
|
|
mem_free(filepath);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
/* Read the first page of the tablespace */
|
|
|
|
buf2 = ut_malloc(3 * UNIV_PAGE_SIZE);
|
|
/* Align the memory for file i/o if we might have O_DIRECT set */
|
|
page = ut_align(buf2, UNIV_PAGE_SIZE);
|
|
|
|
success = os_file_read(file, page, 0, 0, UNIV_PAGE_SIZE);
|
|
if (!success) {
|
|
|
|
goto func_exit;
|
|
}
|
|
|
|
/* We have to read the file flush lsn from the header of the file */
|
|
|
|
flush_lsn = mach_read_ull(page + FIL_PAGE_FILE_FLUSH_LSN);
|
|
|
|
if (current_lsn >= flush_lsn) {
|
|
/* Ok */
|
|
success = TRUE;
|
|
|
|
goto func_exit;
|
|
}
|
|
|
|
space_id = fsp_header_get_space_id(page);
|
|
zip_size = fsp_header_get_zip_size(page);
|
|
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
" InnoDB: Flush lsn in the tablespace file %lu"
|
|
" to be imported\n"
|
|
"InnoDB: is %llu, which exceeds current"
|
|
" system lsn %llu.\n"
|
|
"InnoDB: We reset the lsn's in the file ",
|
|
(ulong) space_id,
|
|
flush_lsn, current_lsn);
|
|
ut_print_filename(stderr, filepath);
|
|
fputs(".\n", stderr);
|
|
|
|
ut_a(ut_is_2pow(zip_size));
|
|
ut_a(zip_size <= UNIV_PAGE_SIZE);
|
|
|
|
/* Loop through all the pages in the tablespace and reset the lsn and
|
|
the page checksum if necessary */
|
|
|
|
file_size = os_file_get_size_as_iblonglong(file);
|
|
|
|
for (offset = 0; offset < file_size;
|
|
offset += zip_size ? zip_size : UNIV_PAGE_SIZE) {
|
|
success = os_file_read(file, page,
|
|
(ulint)(offset & 0xFFFFFFFFUL),
|
|
(ulint)(offset >> 32),
|
|
zip_size ? zip_size : UNIV_PAGE_SIZE);
|
|
if (!success) {
|
|
|
|
goto func_exit;
|
|
}
|
|
if (mach_read_ull(page + FIL_PAGE_LSN) > current_lsn) {
|
|
/* We have to reset the lsn */
|
|
|
|
if (zip_size) {
|
|
memcpy(page + UNIV_PAGE_SIZE, page, zip_size);
|
|
buf_flush_init_for_writing(
|
|
page, page + UNIV_PAGE_SIZE,
|
|
current_lsn);
|
|
} else {
|
|
buf_flush_init_for_writing(
|
|
page, NULL, current_lsn);
|
|
}
|
|
success = os_file_write(filepath, file, page,
|
|
(ulint)(offset & 0xFFFFFFFFUL),
|
|
(ulint)(offset >> 32),
|
|
zip_size
|
|
? zip_size
|
|
: UNIV_PAGE_SIZE);
|
|
if (!success) {
|
|
|
|
goto func_exit;
|
|
}
|
|
}
|
|
}
|
|
|
|
success = os_file_flush(file);
|
|
if (!success) {
|
|
|
|
goto func_exit;
|
|
}
|
|
|
|
/* We now update the flush_lsn stamp at the start of the file */
|
|
success = os_file_read(file, page, 0, 0,
|
|
zip_size ? zip_size : UNIV_PAGE_SIZE);
|
|
if (!success) {
|
|
|
|
goto func_exit;
|
|
}
|
|
|
|
mach_write_ull(page + FIL_PAGE_FILE_FLUSH_LSN, current_lsn);
|
|
|
|
success = os_file_write(filepath, file, page, 0, 0,
|
|
zip_size ? zip_size : UNIV_PAGE_SIZE);
|
|
if (!success) {
|
|
|
|
goto func_exit;
|
|
}
|
|
success = os_file_flush(file);
|
|
func_exit:
|
|
os_file_close(file);
|
|
ut_free(buf2);
|
|
mem_free(filepath);
|
|
|
|
return(success);
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Tries to open a single-table tablespace and optionally checks the space id is
|
|
right in it. If does not succeed, prints an error message to the .err log. This
|
|
function is used to open a tablespace when we start up mysqld, and also in
|
|
IMPORT TABLESPACE.
|
|
NOTE that we assume this operation is used either at the database startup
|
|
or under the protection of the dictionary mutex, so that two users cannot
|
|
race here. This operation does not leave the file associated with the
|
|
tablespace open, but closes it after we have looked at the space id in it.
|
|
@return TRUE if success */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_open_single_table_tablespace(
|
|
/*=============================*/
|
|
ibool check_space_id, /*!< in: should we check that the space
|
|
id in the file is right; we assume
|
|
that this function runs much faster
|
|
if no check is made, since accessing
|
|
the file inode probably is much
|
|
faster (the OS caches them) than
|
|
accessing the first page of the file */
|
|
ulint id, /*!< in: space id */
|
|
ulint flags, /*!< in: tablespace flags */
|
|
const char* name) /*!< in: table name in the
|
|
databasename/tablename format */
|
|
{
|
|
os_file_t file;
|
|
char* filepath;
|
|
ibool success;
|
|
byte* buf2;
|
|
byte* page;
|
|
ulint space_id;
|
|
ulint space_flags;
|
|
|
|
filepath = fil_make_ibd_name(name, FALSE);
|
|
|
|
/* The tablespace flags (FSP_SPACE_FLAGS) should be 0 for
|
|
ROW_FORMAT=COMPACT (table->flags == DICT_TF_COMPACT) and
|
|
ROW_FORMAT=REDUNDANT (table->flags == 0). For any other
|
|
format, the tablespace flags should be equal to
|
|
table->flags & ~(~0 << DICT_TF_BITS). */
|
|
ut_a(flags != DICT_TF_COMPACT);
|
|
ut_a(!(flags & (~0UL << DICT_TF_BITS)));
|
|
|
|
file = os_file_create_simple_no_error_handling(
|
|
filepath, OS_FILE_OPEN, OS_FILE_READ_ONLY, &success);
|
|
if (!success) {
|
|
/* The following call prints an error message */
|
|
os_file_get_last_error(TRUE);
|
|
|
|
ut_print_timestamp(stderr);
|
|
|
|
fputs(" InnoDB: Error: trying to open a table,"
|
|
" but could not\n"
|
|
"InnoDB: open the tablespace file ", stderr);
|
|
ut_print_filename(stderr, filepath);
|
|
fputs("!\n"
|
|
"InnoDB: Have you moved InnoDB .ibd files around"
|
|
" without using the\n"
|
|
"InnoDB: commands DISCARD TABLESPACE and"
|
|
" IMPORT TABLESPACE?\n"
|
|
"InnoDB: It is also possible that this is"
|
|
" a temporary table #sql...,\n"
|
|
"InnoDB: and MySQL removed the .ibd file for this.\n"
|
|
"InnoDB: Please refer to\n"
|
|
"InnoDB: " REFMAN "innodb-troubleshooting-datadict.html\n"
|
|
"InnoDB: for how to resolve the issue.\n", stderr);
|
|
|
|
mem_free(filepath);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
if (!check_space_id) {
|
|
space_id = id;
|
|
|
|
goto skip_check;
|
|
}
|
|
|
|
/* Read the first page of the tablespace */
|
|
|
|
buf2 = ut_malloc(2 * UNIV_PAGE_SIZE);
|
|
/* Align the memory for file i/o if we might have O_DIRECT set */
|
|
page = ut_align(buf2, UNIV_PAGE_SIZE);
|
|
|
|
success = os_file_read(file, page, 0, 0, UNIV_PAGE_SIZE);
|
|
|
|
/* We have to read the tablespace id and flags from the file. */
|
|
|
|
space_id = fsp_header_get_space_id(page);
|
|
space_flags = fsp_header_get_flags(page);
|
|
|
|
ut_free(buf2);
|
|
|
|
if (UNIV_UNLIKELY(space_id != id
|
|
|| space_flags != (flags & ~(~0 << DICT_TF_BITS)))) {
|
|
ut_print_timestamp(stderr);
|
|
|
|
fputs(" InnoDB: Error: tablespace id and flags in file ",
|
|
stderr);
|
|
ut_print_filename(stderr, filepath);
|
|
fprintf(stderr, " are %lu and %lu, but in the InnoDB\n"
|
|
"InnoDB: data dictionary they are %lu and %lu.\n"
|
|
"InnoDB: Have you moved InnoDB .ibd files"
|
|
" around without using the\n"
|
|
"InnoDB: commands DISCARD TABLESPACE and"
|
|
" IMPORT TABLESPACE?\n"
|
|
"InnoDB: Please refer to\n"
|
|
"InnoDB: " REFMAN "innodb-troubleshooting-datadict.html\n"
|
|
"InnoDB: for how to resolve the issue.\n",
|
|
(ulong) space_id, (ulong) space_flags,
|
|
(ulong) id, (ulong) flags);
|
|
|
|
success = FALSE;
|
|
|
|
goto func_exit;
|
|
}
|
|
|
|
skip_check:
|
|
success = fil_space_create(filepath, space_id, flags, FIL_TABLESPACE);
|
|
|
|
if (!success) {
|
|
goto func_exit;
|
|
}
|
|
|
|
/* We do not measure the size of the file, that is why we pass the 0
|
|
below */
|
|
|
|
fil_node_create(filepath, 0, space_id, FALSE);
|
|
func_exit:
|
|
os_file_close(file);
|
|
mem_free(filepath);
|
|
|
|
return(success);
|
|
}
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
|
|
#ifdef UNIV_HOTBACKUP
|
|
/*******************************************************************//**
|
|
Allocates a file name for an old version of a single-table tablespace.
|
|
The string must be freed by caller with mem_free()!
|
|
@return own: file name */
|
|
static
|
|
char*
|
|
fil_make_ibbackup_old_name(
|
|
/*=======================*/
|
|
const char* name) /*!< in: original file name */
|
|
{
|
|
static const char suffix[] = "_ibbackup_old_vers_";
|
|
ulint len = strlen(name);
|
|
char* path = mem_alloc(len + (15 + sizeof suffix));
|
|
|
|
memcpy(path, name, len);
|
|
memcpy(path + len, suffix, (sizeof suffix) - 1);
|
|
ut_sprintf_timestamp_without_extra_chars(path + len + sizeof suffix);
|
|
return(path);
|
|
}
|
|
#endif /* UNIV_HOTBACKUP */
|
|
|
|
/********************************************************************//**
|
|
Opens an .ibd file and adds the associated single-table tablespace to the
|
|
InnoDB fil0fil.c data structures. */
|
|
static
|
|
void
|
|
fil_load_single_table_tablespace(
|
|
/*=============================*/
|
|
const char* dbname, /*!< in: database name */
|
|
const char* filename) /*!< in: file name (not a path),
|
|
including the .ibd extension */
|
|
{
|
|
os_file_t file;
|
|
char* filepath;
|
|
ibool success;
|
|
byte* buf2;
|
|
byte* page;
|
|
ulint space_id;
|
|
ulint flags;
|
|
ulint size_low;
|
|
ulint size_high;
|
|
ib_int64_t size;
|
|
#ifdef UNIV_HOTBACKUP
|
|
fil_space_t* space;
|
|
#endif
|
|
filepath = mem_alloc(strlen(dbname) + strlen(filename)
|
|
+ strlen(fil_path_to_mysql_datadir) + 3);
|
|
|
|
sprintf(filepath, "%s/%s/%s", fil_path_to_mysql_datadir, dbname,
|
|
filename);
|
|
srv_normalize_path_for_win(filepath);
|
|
#ifdef __WIN__
|
|
# ifndef UNIV_HOTBACKUP
|
|
/* If lower_case_table_names is 0 or 2, then MySQL allows database
|
|
directory names with upper case letters. On Windows, all table and
|
|
database names in InnoDB are internally always in lower case. Put the
|
|
file path to lower case, so that we are consistent with InnoDB's
|
|
internal data dictionary. */
|
|
|
|
dict_casedn_str(filepath);
|
|
# endif /* !UNIV_HOTBACKUP */
|
|
#endif
|
|
file = os_file_create_simple_no_error_handling(
|
|
filepath, OS_FILE_OPEN, OS_FILE_READ_ONLY, &success);
|
|
if (!success) {
|
|
/* The following call prints an error message */
|
|
os_file_get_last_error(TRUE);
|
|
|
|
fprintf(stderr,
|
|
"InnoDB: Error: could not open single-table tablespace"
|
|
" file\n"
|
|
"InnoDB: %s!\n"
|
|
"InnoDB: We do not continue the crash recovery,"
|
|
" because the table may become\n"
|
|
"InnoDB: corrupt if we cannot apply the log records"
|
|
" in the InnoDB log to it.\n"
|
|
"InnoDB: To fix the problem and start mysqld:\n"
|
|
"InnoDB: 1) If there is a permission problem"
|
|
" in the file and mysqld cannot\n"
|
|
"InnoDB: open the file, you should"
|
|
" modify the permissions.\n"
|
|
"InnoDB: 2) If the table is not needed, or you can"
|
|
" restore it from a backup,\n"
|
|
"InnoDB: then you can remove the .ibd file,"
|
|
" and InnoDB will do a normal\n"
|
|
"InnoDB: crash recovery and ignore that table.\n"
|
|
"InnoDB: 3) If the file system or the"
|
|
" disk is broken, and you cannot remove\n"
|
|
"InnoDB: the .ibd file, you can set"
|
|
" innodb_force_recovery > 0 in my.cnf\n"
|
|
"InnoDB: and force InnoDB to continue crash"
|
|
" recovery here.\n", filepath);
|
|
|
|
mem_free(filepath);
|
|
|
|
if (srv_force_recovery > 0) {
|
|
fprintf(stderr,
|
|
"InnoDB: innodb_force_recovery"
|
|
" was set to %lu. Continuing crash recovery\n"
|
|
"InnoDB: even though we cannot access"
|
|
" the .ibd file of this table.\n",
|
|
srv_force_recovery);
|
|
return;
|
|
}
|
|
|
|
exit(1);
|
|
}
|
|
|
|
success = os_file_get_size(file, &size_low, &size_high);
|
|
|
|
if (!success) {
|
|
/* The following call prints an error message */
|
|
os_file_get_last_error(TRUE);
|
|
|
|
fprintf(stderr,
|
|
"InnoDB: Error: could not measure the size"
|
|
" of single-table tablespace file\n"
|
|
"InnoDB: %s!\n"
|
|
"InnoDB: We do not continue crash recovery,"
|
|
" because the table will become\n"
|
|
"InnoDB: corrupt if we cannot apply the log records"
|
|
" in the InnoDB log to it.\n"
|
|
"InnoDB: To fix the problem and start mysqld:\n"
|
|
"InnoDB: 1) If there is a permission problem"
|
|
" in the file and mysqld cannot\n"
|
|
"InnoDB: access the file, you should"
|
|
" modify the permissions.\n"
|
|
"InnoDB: 2) If the table is not needed,"
|
|
" or you can restore it from a backup,\n"
|
|
"InnoDB: then you can remove the .ibd file,"
|
|
" and InnoDB will do a normal\n"
|
|
"InnoDB: crash recovery and ignore that table.\n"
|
|
"InnoDB: 3) If the file system or the disk is broken,"
|
|
" and you cannot remove\n"
|
|
"InnoDB: the .ibd file, you can set"
|
|
" innodb_force_recovery > 0 in my.cnf\n"
|
|
"InnoDB: and force InnoDB to continue"
|
|
" crash recovery here.\n", filepath);
|
|
|
|
os_file_close(file);
|
|
mem_free(filepath);
|
|
|
|
if (srv_force_recovery > 0) {
|
|
fprintf(stderr,
|
|
"InnoDB: innodb_force_recovery"
|
|
" was set to %lu. Continuing crash recovery\n"
|
|
"InnoDB: even though we cannot access"
|
|
" the .ibd file of this table.\n",
|
|
srv_force_recovery);
|
|
return;
|
|
}
|
|
|
|
exit(1);
|
|
}
|
|
|
|
/* TODO: What to do in other cases where we cannot access an .ibd
|
|
file during a crash recovery? */
|
|
|
|
/* Every .ibd file is created >= 4 pages in size. Smaller files
|
|
cannot be ok. */
|
|
|
|
size = (((ib_int64_t)size_high) << 32) + (ib_int64_t)size_low;
|
|
#ifndef UNIV_HOTBACKUP
|
|
if (size < FIL_IBD_FILE_INITIAL_SIZE * UNIV_PAGE_SIZE) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: the size of single-table tablespace"
|
|
" file %s\n"
|
|
"InnoDB: is only %lu %lu, should be at least %lu!",
|
|
filepath,
|
|
(ulong) size_high,
|
|
(ulong) size_low, (ulong) (4 * UNIV_PAGE_SIZE));
|
|
os_file_close(file);
|
|
mem_free(filepath);
|
|
|
|
return;
|
|
}
|
|
#endif
|
|
/* Read the first page of the tablespace if the size big enough */
|
|
|
|
buf2 = ut_malloc(2 * UNIV_PAGE_SIZE);
|
|
/* Align the memory for file i/o if we might have O_DIRECT set */
|
|
page = ut_align(buf2, UNIV_PAGE_SIZE);
|
|
|
|
if (size >= FIL_IBD_FILE_INITIAL_SIZE * UNIV_PAGE_SIZE) {
|
|
success = os_file_read(file, page, 0, 0, UNIV_PAGE_SIZE);
|
|
|
|
/* We have to read the tablespace id from the file */
|
|
|
|
space_id = fsp_header_get_space_id(page);
|
|
flags = fsp_header_get_flags(page);
|
|
} else {
|
|
space_id = ULINT_UNDEFINED;
|
|
flags = 0;
|
|
}
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
if (space_id == ULINT_UNDEFINED || space_id == 0) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: tablespace id %lu in file %s"
|
|
" is not sensible\n",
|
|
(ulong) space_id,
|
|
filepath);
|
|
goto func_exit;
|
|
}
|
|
#else
|
|
if (space_id == ULINT_UNDEFINED || space_id == 0) {
|
|
char* new_path;
|
|
|
|
fprintf(stderr,
|
|
"InnoDB: Renaming tablespace %s of id %lu,\n"
|
|
"InnoDB: to %s_ibbackup_old_vers_<timestamp>\n"
|
|
"InnoDB: because its size %" PRId64 " is too small"
|
|
" (< 4 pages 16 kB each),\n"
|
|
"InnoDB: or the space id in the file header"
|
|
" is not sensible.\n"
|
|
"InnoDB: This can happen in an ibbackup run,"
|
|
" and is not dangerous.\n",
|
|
filepath, space_id, filepath, size);
|
|
os_file_close(file);
|
|
|
|
new_path = fil_make_ibbackup_old_name(filepath);
|
|
ut_a(os_file_rename(filepath, new_path));
|
|
|
|
ut_free(buf2);
|
|
mem_free(filepath);
|
|
mem_free(new_path);
|
|
|
|
return;
|
|
}
|
|
|
|
/* A backup may contain the same space several times, if the space got
|
|
renamed at a sensitive time. Since it is enough to have one version of
|
|
the space, we rename the file if a space with the same space id
|
|
already exists in the tablespace memory cache. We rather rename the
|
|
file than delete it, because if there is a bug, we do not want to
|
|
destroy valuable data. */
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(space_id);
|
|
|
|
if (space) {
|
|
char* new_path;
|
|
|
|
fprintf(stderr,
|
|
"InnoDB: Renaming tablespace %s of id %lu,\n"
|
|
"InnoDB: to %s_ibbackup_old_vers_<timestamp>\n"
|
|
"InnoDB: because space %s with the same id\n"
|
|
"InnoDB: was scanned earlier. This can happen"
|
|
" if you have renamed tables\n"
|
|
"InnoDB: during an ibbackup run.\n",
|
|
filepath, space_id, filepath,
|
|
space->name);
|
|
os_file_close(file);
|
|
|
|
new_path = fil_make_ibbackup_old_name(filepath);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
ut_a(os_file_rename(filepath, new_path));
|
|
|
|
ut_free(buf2);
|
|
mem_free(filepath);
|
|
mem_free(new_path);
|
|
|
|
return;
|
|
}
|
|
mutex_exit(&fil_system->mutex);
|
|
#endif
|
|
success = fil_space_create(filepath, space_id, flags, FIL_TABLESPACE);
|
|
|
|
if (!success) {
|
|
|
|
if (srv_force_recovery > 0) {
|
|
fprintf(stderr,
|
|
"InnoDB: innodb_force_recovery"
|
|
" was set to %lu. Continuing crash recovery\n"
|
|
"InnoDB: even though the tablespace creation"
|
|
" of this table failed.\n",
|
|
srv_force_recovery);
|
|
goto func_exit;
|
|
}
|
|
|
|
exit(1);
|
|
}
|
|
|
|
/* We do not use the size information we have about the file, because
|
|
the rounding formula for extents and pages is somewhat complex; we
|
|
let fil_node_open() do that task. */
|
|
|
|
fil_node_create(filepath, 0, space_id, FALSE);
|
|
func_exit:
|
|
os_file_close(file);
|
|
ut_free(buf2);
|
|
mem_free(filepath);
|
|
}
|
|
|
|
/***********************************************************************//**
|
|
A fault-tolerant function that tries to read the next file name in the
|
|
directory. We retry 100 times if os_file_readdir_next_file() returns -1. The
|
|
idea is to read as much good data as we can and jump over bad data.
|
|
@return 0 if ok, -1 if error even after the retries, 1 if at the end
|
|
of the directory */
|
|
static
|
|
int
|
|
fil_file_readdir_next_file(
|
|
/*=======================*/
|
|
ulint* err, /*!< out: this is set to DB_ERROR if an error
|
|
was encountered, otherwise not changed */
|
|
const char* dirname,/*!< in: directory name or path */
|
|
os_file_dir_t dir, /*!< in: directory stream */
|
|
os_file_stat_t* info) /*!< in/out: buffer where the info is returned */
|
|
{
|
|
ulint i;
|
|
int ret;
|
|
|
|
for (i = 0; i < 100; i++) {
|
|
ret = os_file_readdir_next_file(dirname, dir, info);
|
|
|
|
if (ret != -1) {
|
|
|
|
return(ret);
|
|
}
|
|
|
|
fprintf(stderr,
|
|
"InnoDB: Error: os_file_readdir_next_file()"
|
|
" returned -1 in\n"
|
|
"InnoDB: directory %s\n"
|
|
"InnoDB: Crash recovery may have failed"
|
|
" for some .ibd files!\n", dirname);
|
|
|
|
*err = DB_ERROR;
|
|
}
|
|
|
|
return(-1);
|
|
}
|
|
|
|
/********************************************************************//**
|
|
At the server startup, if we need crash recovery, scans the database
|
|
directories under the MySQL datadir, looking for .ibd files. Those files are
|
|
single-table tablespaces. We need to know the space id in each of them so that
|
|
we know into which file we should look to check the contents of a page stored
|
|
in the doublewrite buffer, also to know where to apply log records where the
|
|
space id is != 0.
|
|
@return DB_SUCCESS or error number */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_load_single_table_tablespaces(void)
|
|
/*===================================*/
|
|
{
|
|
int ret;
|
|
char* dbpath = NULL;
|
|
ulint dbpath_len = 100;
|
|
os_file_dir_t dir;
|
|
os_file_dir_t dbdir;
|
|
os_file_stat_t dbinfo;
|
|
os_file_stat_t fileinfo;
|
|
ulint err = DB_SUCCESS;
|
|
|
|
/* The datadir of MySQL is always the default directory of mysqld */
|
|
|
|
dir = os_file_opendir(fil_path_to_mysql_datadir, TRUE);
|
|
|
|
if (dir == NULL) {
|
|
|
|
return(DB_ERROR);
|
|
}
|
|
|
|
dbpath = mem_alloc(dbpath_len);
|
|
|
|
/* Scan all directories under the datadir. They are the database
|
|
directories of MySQL. */
|
|
|
|
ret = fil_file_readdir_next_file(&err, fil_path_to_mysql_datadir, dir,
|
|
&dbinfo);
|
|
while (ret == 0) {
|
|
ulint len;
|
|
/* printf("Looking at %s in datadir\n", dbinfo.name); */
|
|
|
|
if (dbinfo.type == OS_FILE_TYPE_FILE
|
|
|| dbinfo.type == OS_FILE_TYPE_UNKNOWN) {
|
|
|
|
goto next_datadir_item;
|
|
}
|
|
|
|
/* We found a symlink or a directory; try opening it to see
|
|
if a symlink is a directory */
|
|
|
|
len = strlen(fil_path_to_mysql_datadir)
|
|
+ strlen (dbinfo.name) + 2;
|
|
if (len > dbpath_len) {
|
|
dbpath_len = len;
|
|
|
|
if (dbpath) {
|
|
mem_free(dbpath);
|
|
}
|
|
|
|
dbpath = mem_alloc(dbpath_len);
|
|
}
|
|
sprintf(dbpath, "%s/%s", fil_path_to_mysql_datadir,
|
|
dbinfo.name);
|
|
srv_normalize_path_for_win(dbpath);
|
|
|
|
dbdir = os_file_opendir(dbpath, FALSE);
|
|
|
|
if (dbdir != NULL) {
|
|
/* printf("Opened dir %s\n", dbinfo.name); */
|
|
|
|
/* We found a database directory; loop through it,
|
|
looking for possible .ibd files in it */
|
|
|
|
ret = fil_file_readdir_next_file(&err, dbpath, dbdir,
|
|
&fileinfo);
|
|
while (ret == 0) {
|
|
/* printf(
|
|
" Looking at file %s\n", fileinfo.name); */
|
|
|
|
if (fileinfo.type == OS_FILE_TYPE_DIR) {
|
|
|
|
goto next_file_item;
|
|
}
|
|
|
|
/* We found a symlink or a file */
|
|
if (strlen(fileinfo.name) > 4
|
|
&& 0 == strcmp(fileinfo.name
|
|
+ strlen(fileinfo.name) - 4,
|
|
".ibd")) {
|
|
/* The name ends in .ibd; try opening
|
|
the file */
|
|
fil_load_single_table_tablespace(
|
|
dbinfo.name, fileinfo.name);
|
|
}
|
|
next_file_item:
|
|
ret = fil_file_readdir_next_file(&err,
|
|
dbpath, dbdir,
|
|
&fileinfo);
|
|
}
|
|
|
|
if (0 != os_file_closedir(dbdir)) {
|
|
fputs("InnoDB: Warning: could not"
|
|
" close database directory ", stderr);
|
|
ut_print_filename(stderr, dbpath);
|
|
putc('\n', stderr);
|
|
|
|
err = DB_ERROR;
|
|
}
|
|
}
|
|
|
|
next_datadir_item:
|
|
ret = fil_file_readdir_next_file(&err,
|
|
fil_path_to_mysql_datadir,
|
|
dir, &dbinfo);
|
|
}
|
|
|
|
mem_free(dbpath);
|
|
|
|
if (0 != os_file_closedir(dir)) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: could not close MySQL datadir\n");
|
|
|
|
return(DB_ERROR);
|
|
}
|
|
|
|
return(err);
|
|
}
|
|
|
|
/********************************************************************//**
|
|
If we need crash recovery, and we have called
|
|
fil_load_single_table_tablespaces() and dict_load_single_table_tablespaces(),
|
|
we can call this function to print an error message of orphaned .ibd files
|
|
for which there is not a data dictionary entry with a matching table name
|
|
and space id. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_print_orphaned_tablespaces(void)
|
|
/*================================*/
|
|
{
|
|
fil_space_t* space;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = UT_LIST_GET_FIRST(fil_system->space_list);
|
|
|
|
while (space) {
|
|
if (space->purpose == FIL_TABLESPACE && space->id != 0
|
|
&& !space->mark) {
|
|
fputs("InnoDB: Warning: tablespace ", stderr);
|
|
ut_print_filename(stderr, space->name);
|
|
fprintf(stderr, " of id %lu has no matching table in\n"
|
|
"InnoDB: the InnoDB data dictionary.\n",
|
|
(ulong) space->id);
|
|
}
|
|
|
|
space = UT_LIST_GET_NEXT(space_list, space);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns TRUE if a single-table tablespace does not exist in the memory cache,
|
|
or is being deleted there.
|
|
@return TRUE if does not exist or is being\ deleted */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_tablespace_deleted_or_being_deleted_in_mem(
|
|
/*===========================================*/
|
|
ulint id, /*!< in: space id */
|
|
ib_int64_t version)/*!< in: tablespace_version should be this; if
|
|
you pass -1 as the value of this, then this
|
|
parameter is ignored */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
if (space == NULL || space->is_being_deleted) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
if (version != ((ib_int64_t)-1)
|
|
&& space->tablespace_version != version) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns TRUE if a single-table tablespace exists in the memory cache.
|
|
@return TRUE if exists */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_tablespace_exists_in_mem(
|
|
/*=========================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(space != NULL);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Returns TRUE if a matching tablespace exists in the InnoDB tablespace memory
|
|
cache. Note that if we have not done a crash recovery at the database startup,
|
|
there may be many tablespaces which are not yet in the memory cache.
|
|
@return TRUE if a matching tablespace exists in the memory cache */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_space_for_table_exists_in_mem(
|
|
/*==============================*/
|
|
ulint id, /*!< in: space id */
|
|
const char* name, /*!< in: table name in the standard
|
|
'databasename/tablename' format or
|
|
the dir path to a temp table */
|
|
ibool is_temp, /*!< in: TRUE if created with CREATE
|
|
TEMPORARY TABLE */
|
|
ibool mark_space, /*!< in: in crash recovery, at database
|
|
startup we mark all spaces which have
|
|
an associated table in the InnoDB
|
|
data dictionary, so that
|
|
we can print a warning about orphaned
|
|
tablespaces */
|
|
ibool print_error_if_does_not_exist)
|
|
/*!< in: print detailed error
|
|
information to the .err log if a
|
|
matching tablespace is not found from
|
|
memory */
|
|
{
|
|
fil_space_t* namespace;
|
|
fil_space_t* space;
|
|
char* path;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
path = fil_make_ibd_name(name, is_temp);
|
|
|
|
/* Look if there is a space with the same id */
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
/* Look if there is a space with the same name; the name is the
|
|
directory path from the datadir to the file */
|
|
|
|
namespace = fil_space_get_by_name(path);
|
|
if (space && space == namespace) {
|
|
/* Found */
|
|
|
|
if (mark_space) {
|
|
space->mark = TRUE;
|
|
}
|
|
|
|
mem_free(path);
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
if (!print_error_if_does_not_exist) {
|
|
|
|
mem_free(path);
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
if (space == NULL) {
|
|
if (namespace == NULL) {
|
|
ut_print_timestamp(stderr);
|
|
fputs(" InnoDB: Error: table ", stderr);
|
|
ut_print_filename(stderr, name);
|
|
fprintf(stderr, "\n"
|
|
"InnoDB: in InnoDB data dictionary"
|
|
" has tablespace id %lu,\n"
|
|
"InnoDB: but tablespace with that id"
|
|
" or name does not exist. Have\n"
|
|
"InnoDB: you deleted or moved .ibd files?\n"
|
|
"InnoDB: This may also be a table created with"
|
|
" CREATE TEMPORARY TABLE\n"
|
|
"InnoDB: whose .ibd and .frm files"
|
|
" MySQL automatically removed, but the\n"
|
|
"InnoDB: table still exists in the"
|
|
" InnoDB internal data dictionary.\n",
|
|
(ulong) id);
|
|
} else {
|
|
ut_print_timestamp(stderr);
|
|
fputs(" InnoDB: Error: table ", stderr);
|
|
ut_print_filename(stderr, name);
|
|
fprintf(stderr, "\n"
|
|
"InnoDB: in InnoDB data dictionary has"
|
|
" tablespace id %lu,\n"
|
|
"InnoDB: but a tablespace with that id"
|
|
" does not exist. There is\n"
|
|
"InnoDB: a tablespace of name %s and id %lu,"
|
|
" though. Have\n"
|
|
"InnoDB: you deleted or moved .ibd files?\n",
|
|
(ulong) id, namespace->name,
|
|
(ulong) namespace->id);
|
|
}
|
|
error_exit:
|
|
fputs("InnoDB: Please refer to\n"
|
|
"InnoDB: " REFMAN "innodb-troubleshooting-datadict.html\n"
|
|
"InnoDB: for how to resolve the issue.\n", stderr);
|
|
|
|
mem_free(path);
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
if (0 != strcmp(space->name, path)) {
|
|
ut_print_timestamp(stderr);
|
|
fputs(" InnoDB: Error: table ", stderr);
|
|
ut_print_filename(stderr, name);
|
|
fprintf(stderr, "\n"
|
|
"InnoDB: in InnoDB data dictionary has"
|
|
" tablespace id %lu,\n"
|
|
"InnoDB: but the tablespace with that id"
|
|
" has name %s.\n"
|
|
"InnoDB: Have you deleted or moved .ibd files?\n",
|
|
(ulong) id, space->name);
|
|
|
|
if (namespace != NULL) {
|
|
fputs("InnoDB: There is a tablespace"
|
|
" with the right name\n"
|
|
"InnoDB: ", stderr);
|
|
ut_print_filename(stderr, namespace->name);
|
|
fprintf(stderr, ", but its id is %lu.\n",
|
|
(ulong) namespace->id);
|
|
}
|
|
|
|
goto error_exit;
|
|
}
|
|
|
|
mem_free(path);
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(FALSE);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Checks if a single-table tablespace for a given table name exists in the
|
|
tablespace memory cache.
|
|
@return space id, ULINT_UNDEFINED if not found */
|
|
static
|
|
ulint
|
|
fil_get_space_id_for_table(
|
|
/*=======================*/
|
|
const char* name) /*!< in: table name in the standard
|
|
'databasename/tablename' format */
|
|
{
|
|
fil_space_t* namespace;
|
|
ulint id = ULINT_UNDEFINED;
|
|
char* path;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
path = fil_make_ibd_name(name, FALSE);
|
|
|
|
/* Look if there is a space with the same name; the name is the
|
|
directory path to the file */
|
|
|
|
namespace = fil_space_get_by_name(path);
|
|
|
|
if (namespace) {
|
|
id = namespace->id;
|
|
}
|
|
|
|
mem_free(path);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(id);
|
|
}
|
|
|
|
/**********************************************************************//**
|
|
Tries to extend a data file so that it would accommodate the number of pages
|
|
given. The tablespace must be cached in the memory cache. If the space is big
|
|
enough already, does nothing.
|
|
@return TRUE if success */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_extend_space_to_desired_size(
|
|
/*=============================*/
|
|
ulint* actual_size, /*!< out: size of the space after extension;
|
|
if we ran out of disk space this may be lower
|
|
than the desired size */
|
|
ulint space_id, /*!< in: space id */
|
|
ulint size_after_extend)/*!< in: desired size in pages after the
|
|
extension; if the current space size is bigger
|
|
than this already, the function does nothing */
|
|
{
|
|
fil_node_t* node;
|
|
fil_space_t* space;
|
|
byte* buf2;
|
|
byte* buf;
|
|
ulint buf_size;
|
|
ulint start_page_no;
|
|
ulint file_start_page_no;
|
|
ulint offset_high;
|
|
ulint offset_low;
|
|
ulint page_size;
|
|
ibool success = TRUE;
|
|
|
|
fil_mutex_enter_and_prepare_for_io(space_id);
|
|
|
|
space = fil_space_get_by_id(space_id);
|
|
ut_a(space);
|
|
|
|
if (space->size >= size_after_extend) {
|
|
/* Space already big enough */
|
|
|
|
*actual_size = space->size;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
page_size = dict_table_flags_to_zip_size(space->flags);
|
|
if (!page_size) {
|
|
page_size = UNIV_PAGE_SIZE;
|
|
}
|
|
|
|
node = UT_LIST_GET_LAST(space->chain);
|
|
|
|
fil_node_prepare_for_io(node, fil_system, space);
|
|
|
|
start_page_no = space->size;
|
|
file_start_page_no = space->size - node->size;
|
|
|
|
/* Extend at most 64 pages at a time */
|
|
buf_size = ut_min(64, size_after_extend - start_page_no) * page_size;
|
|
buf2 = mem_alloc(buf_size + page_size);
|
|
buf = ut_align(buf2, page_size);
|
|
|
|
memset(buf, 0, buf_size);
|
|
|
|
while (start_page_no < size_after_extend) {
|
|
ulint n_pages = ut_min(buf_size / page_size,
|
|
size_after_extend - start_page_no);
|
|
|
|
offset_high = (start_page_no - file_start_page_no)
|
|
/ (4096 * ((1024 * 1024) / page_size));
|
|
offset_low = ((start_page_no - file_start_page_no)
|
|
% (4096 * ((1024 * 1024) / page_size)))
|
|
* page_size;
|
|
#ifdef UNIV_HOTBACKUP
|
|
success = os_file_write(node->name, node->handle, buf,
|
|
offset_low, offset_high,
|
|
page_size * n_pages);
|
|
#else
|
|
success = os_aio(OS_FILE_WRITE, OS_AIO_SYNC,
|
|
node->name, node->handle, buf,
|
|
offset_low, offset_high,
|
|
page_size * n_pages,
|
|
NULL, NULL);
|
|
#endif
|
|
if (success) {
|
|
node->size += n_pages;
|
|
space->size += n_pages;
|
|
|
|
os_has_said_disk_full = FALSE;
|
|
} else {
|
|
/* Let us measure the size of the file to determine
|
|
how much we were able to extend it */
|
|
|
|
n_pages = ((ulint)
|
|
(os_file_get_size_as_iblonglong(
|
|
node->handle)
|
|
/ page_size)) - node->size;
|
|
|
|
node->size += n_pages;
|
|
space->size += n_pages;
|
|
|
|
break;
|
|
}
|
|
|
|
start_page_no += n_pages;
|
|
}
|
|
|
|
mem_free(buf2);
|
|
|
|
fil_node_complete_io(node, fil_system, OS_FILE_WRITE);
|
|
|
|
*actual_size = space->size;
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
if (space_id == 0) {
|
|
ulint pages_per_mb = (1024 * 1024) / page_size;
|
|
|
|
/* Keep the last data file size info up to date, rounded to
|
|
full megabytes */
|
|
|
|
srv_data_file_sizes[srv_n_data_files - 1]
|
|
= (node->size / pages_per_mb) * pages_per_mb;
|
|
}
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
|
|
/*
|
|
printf("Extended %s to %lu, actual size %lu pages\n", space->name,
|
|
size_after_extend, *actual_size); */
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
fil_flush(space_id);
|
|
|
|
return(success);
|
|
}
|
|
|
|
#ifdef UNIV_HOTBACKUP
|
|
/********************************************************************//**
|
|
Extends all tablespaces to the size stored in the space header. During the
|
|
ibbackup --apply-log phase we extended the spaces on-demand so that log records
|
|
could be applied, but that may have left spaces still too small compared to
|
|
the size stored in the space header. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_extend_tablespaces_to_stored_len(void)
|
|
/*======================================*/
|
|
{
|
|
fil_space_t* space;
|
|
byte* buf;
|
|
ulint actual_size;
|
|
ulint size_in_header;
|
|
ulint error;
|
|
ibool success;
|
|
|
|
buf = mem_alloc(UNIV_PAGE_SIZE);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = UT_LIST_GET_FIRST(fil_system->space_list);
|
|
|
|
while (space) {
|
|
ut_a(space->purpose == FIL_TABLESPACE);
|
|
|
|
mutex_exit(&fil_system->mutex); /* no need to protect with a
|
|
mutex, because this is a
|
|
single-threaded operation */
|
|
error = fil_read(TRUE, space->id,
|
|
dict_table_flags_to_zip_size(space->flags),
|
|
0, 0, UNIV_PAGE_SIZE, buf, NULL);
|
|
ut_a(error == DB_SUCCESS);
|
|
|
|
size_in_header = fsp_get_size_low(buf);
|
|
|
|
success = fil_extend_space_to_desired_size(
|
|
&actual_size, space->id, size_in_header);
|
|
if (!success) {
|
|
fprintf(stderr,
|
|
"InnoDB: Error: could not extend the"
|
|
" tablespace of %s\n"
|
|
"InnoDB: to the size stored in header,"
|
|
" %lu pages;\n"
|
|
"InnoDB: size after extension %lu pages\n"
|
|
"InnoDB: Check that you have free disk space"
|
|
" and retry!\n",
|
|
space->name, size_in_header, actual_size);
|
|
exit(1);
|
|
}
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = UT_LIST_GET_NEXT(space_list, space);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
mem_free(buf);
|
|
}
|
|
#endif
|
|
|
|
/*========== RESERVE FREE EXTENTS (for a B-tree split, for example) ===*/
|
|
|
|
/*******************************************************************//**
|
|
Tries to reserve free extents in a file space.
|
|
@return TRUE if succeed */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_space_reserve_free_extents(
|
|
/*===========================*/
|
|
ulint id, /*!< in: space id */
|
|
ulint n_free_now, /*!< in: number of free extents now */
|
|
ulint n_to_reserve) /*!< in: how many one wants to reserve */
|
|
{
|
|
fil_space_t* space;
|
|
ibool success;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
ut_a(space);
|
|
|
|
if (space->n_reserved_extents + n_to_reserve > n_free_now) {
|
|
success = FALSE;
|
|
} else {
|
|
space->n_reserved_extents += n_to_reserve;
|
|
success = TRUE;
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(success);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Releases free extents in a file space. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_space_release_free_extents(
|
|
/*===========================*/
|
|
ulint id, /*!< in: space id */
|
|
ulint n_reserved) /*!< in: how many one reserved */
|
|
{
|
|
fil_space_t* space;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
ut_a(space);
|
|
ut_a(space->n_reserved_extents >= n_reserved);
|
|
|
|
space->n_reserved_extents -= n_reserved;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
|
|
/*******************************************************************//**
|
|
Gets the number of reserved extents. If the database is silent, this number
|
|
should be zero. */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_space_get_n_reserved_extents(
|
|
/*=============================*/
|
|
ulint id) /*!< in: space id */
|
|
{
|
|
fil_space_t* space;
|
|
ulint n;
|
|
|
|
ut_ad(fil_system);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(id);
|
|
|
|
ut_a(space);
|
|
|
|
n = space->n_reserved_extents;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(n);
|
|
}
|
|
|
|
/*============================ FILE I/O ================================*/
|
|
|
|
/********************************************************************//**
|
|
NOTE: you must call fil_mutex_enter_and_prepare_for_io() first!
|
|
|
|
Prepares a file node for i/o. Opens the file if it is closed. Updates the
|
|
pending i/o's field in the node and the system appropriately. Takes the node
|
|
off the LRU list if it is in the LRU list. The caller must hold the fil_sys
|
|
mutex. */
|
|
static
|
|
void
|
|
fil_node_prepare_for_io(
|
|
/*====================*/
|
|
fil_node_t* node, /*!< in: file node */
|
|
fil_system_t* system, /*!< in: tablespace memory cache */
|
|
fil_space_t* space) /*!< in: space */
|
|
{
|
|
ut_ad(node && system && space);
|
|
ut_ad(mutex_own(&(system->mutex)));
|
|
|
|
if (system->n_open > system->max_n_open + 5) {
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
" InnoDB: Warning: open files %lu"
|
|
" exceeds the limit %lu\n",
|
|
(ulong) system->n_open,
|
|
(ulong) system->max_n_open);
|
|
}
|
|
|
|
if (node->open == FALSE) {
|
|
/* File is closed: open it */
|
|
ut_a(node->n_pending == 0);
|
|
|
|
fil_node_open_file(node, system, space);
|
|
}
|
|
|
|
if (node->n_pending == 0 && space->purpose == FIL_TABLESPACE
|
|
&& space->id != 0) {
|
|
/* The node is in the LRU list, remove it */
|
|
|
|
ut_a(UT_LIST_GET_LEN(system->LRU) > 0);
|
|
|
|
UT_LIST_REMOVE(LRU, system->LRU, node);
|
|
}
|
|
|
|
node->n_pending++;
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Updates the data structures when an i/o operation finishes. Updates the
|
|
pending i/o's field in the node appropriately. */
|
|
static
|
|
void
|
|
fil_node_complete_io(
|
|
/*=================*/
|
|
fil_node_t* node, /*!< in: file node */
|
|
fil_system_t* system, /*!< in: tablespace memory cache */
|
|
ulint type) /*!< in: OS_FILE_WRITE or OS_FILE_READ; marks
|
|
the node as modified if
|
|
type == OS_FILE_WRITE */
|
|
{
|
|
ut_ad(node);
|
|
ut_ad(system);
|
|
ut_ad(mutex_own(&(system->mutex)));
|
|
|
|
ut_a(node->n_pending > 0);
|
|
|
|
node->n_pending--;
|
|
|
|
if (type == OS_FILE_WRITE) {
|
|
system->modification_counter++;
|
|
node->modification_counter = system->modification_counter;
|
|
|
|
if (!node->space->is_in_unflushed_spaces) {
|
|
|
|
node->space->is_in_unflushed_spaces = TRUE;
|
|
UT_LIST_ADD_FIRST(unflushed_spaces,
|
|
system->unflushed_spaces,
|
|
node->space);
|
|
}
|
|
}
|
|
|
|
if (node->n_pending == 0 && node->space->purpose == FIL_TABLESPACE
|
|
&& node->space->id != 0) {
|
|
/* The node must be put back to the LRU list */
|
|
UT_LIST_ADD_FIRST(LRU, system->LRU, node);
|
|
}
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Report information about an invalid page access. */
|
|
static
|
|
void
|
|
fil_report_invalid_page_access(
|
|
/*===========================*/
|
|
ulint block_offset, /*!< in: block offset */
|
|
ulint space_id, /*!< in: space id */
|
|
const char* space_name, /*!< in: space name */
|
|
ulint byte_offset, /*!< in: byte offset */
|
|
ulint len, /*!< in: I/O length */
|
|
ulint type) /*!< in: I/O type */
|
|
{
|
|
fprintf(stderr,
|
|
"InnoDB: Error: trying to access page number %lu"
|
|
" in space %lu,\n"
|
|
"InnoDB: space name %s,\n"
|
|
"InnoDB: which is outside the tablespace bounds.\n"
|
|
"InnoDB: Byte offset %lu, len %lu, i/o type %lu.\n"
|
|
"InnoDB: If you get this error at mysqld startup,"
|
|
" please check that\n"
|
|
"InnoDB: your my.cnf matches the ibdata files"
|
|
" that you have in the\n"
|
|
"InnoDB: MySQL server.\n",
|
|
(ulong) block_offset, (ulong) space_id, space_name,
|
|
(ulong) byte_offset, (ulong) len, (ulong) type);
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Reads or writes data. This operation is asynchronous (aio).
|
|
@return DB_SUCCESS, or DB_TABLESPACE_DELETED if we are trying to do
|
|
i/o on a tablespace which does not exist */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_io(
|
|
/*===*/
|
|
ulint type, /*!< in: OS_FILE_READ or OS_FILE_WRITE,
|
|
ORed to OS_FILE_LOG, if a log i/o
|
|
and ORed to OS_AIO_SIMULATED_WAKE_LATER
|
|
if simulated aio and we want to post a
|
|
batch of i/os; NOTE that a simulated batch
|
|
may introduce hidden chances of deadlocks,
|
|
because i/os are not actually handled until
|
|
all have been posted: use with great
|
|
caution! */
|
|
ibool sync, /*!< in: TRUE if synchronous aio is desired */
|
|
ulint space_id, /*!< in: space id */
|
|
ulint zip_size, /*!< in: compressed page size in bytes;
|
|
0 for uncompressed pages */
|
|
ulint block_offset, /*!< in: offset in number of blocks */
|
|
ulint byte_offset, /*!< in: remainder of offset in bytes; in
|
|
aio this must be divisible by the OS block
|
|
size */
|
|
ulint len, /*!< in: how many bytes to read or write; this
|
|
must not cross a file boundary; in aio this
|
|
must be a block size multiple */
|
|
void* buf, /*!< in/out: buffer where to store read data
|
|
or from where to write; in aio this must be
|
|
appropriately aligned */
|
|
void* message) /*!< in: message for aio handler if non-sync
|
|
aio used, else ignored */
|
|
{
|
|
ulint mode;
|
|
fil_space_t* space;
|
|
fil_node_t* node;
|
|
ulint offset_high;
|
|
ulint offset_low;
|
|
ibool ret;
|
|
ulint is_log;
|
|
ulint wake_later;
|
|
|
|
is_log = type & OS_FILE_LOG;
|
|
type = type & ~OS_FILE_LOG;
|
|
|
|
wake_later = type & OS_AIO_SIMULATED_WAKE_LATER;
|
|
type = type & ~OS_AIO_SIMULATED_WAKE_LATER;
|
|
|
|
ut_ad(byte_offset < UNIV_PAGE_SIZE);
|
|
ut_ad(!zip_size || !byte_offset);
|
|
ut_ad(ut_is_2pow(zip_size));
|
|
ut_ad(buf);
|
|
ut_ad(len > 0);
|
|
#if (1 << UNIV_PAGE_SIZE_SHIFT) != UNIV_PAGE_SIZE
|
|
# error "(1 << UNIV_PAGE_SIZE_SHIFT) != UNIV_PAGE_SIZE"
|
|
#endif
|
|
ut_ad(fil_validate());
|
|
#ifndef UNIV_HOTBACKUP
|
|
# ifndef UNIV_LOG_DEBUG
|
|
/* ibuf bitmap pages must be read in the sync aio mode: */
|
|
ut_ad(recv_no_ibuf_operations || (type == OS_FILE_WRITE)
|
|
|| !ibuf_bitmap_page(zip_size, block_offset)
|
|
|| sync || is_log);
|
|
ut_ad(!ibuf_inside() || is_log || (type == OS_FILE_WRITE)
|
|
|| ibuf_page(space_id, zip_size, block_offset, NULL));
|
|
# endif /* UNIV_LOG_DEBUG */
|
|
if (sync) {
|
|
mode = OS_AIO_SYNC;
|
|
} else if (is_log) {
|
|
mode = OS_AIO_LOG;
|
|
} else if (type == OS_FILE_READ
|
|
&& !recv_no_ibuf_operations
|
|
&& ibuf_page(space_id, zip_size, block_offset, NULL)) {
|
|
mode = OS_AIO_IBUF;
|
|
} else {
|
|
mode = OS_AIO_NORMAL;
|
|
}
|
|
#else /* !UNIV_HOTBACKUP */
|
|
ut_a(sync);
|
|
mode = OS_AIO_SYNC;
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
|
|
if (type == OS_FILE_READ) {
|
|
srv_data_read+= len;
|
|
} else if (type == OS_FILE_WRITE) {
|
|
srv_data_written+= len;
|
|
}
|
|
|
|
/* Reserve the fil_system mutex and make sure that we can open at
|
|
least one file while holding it, if the file is not already open */
|
|
|
|
fil_mutex_enter_and_prepare_for_io(space_id);
|
|
|
|
space = fil_space_get_by_id(space_id);
|
|
|
|
if (!space) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
ut_print_timestamp(stderr);
|
|
fprintf(stderr,
|
|
" InnoDB: Error: trying to do i/o"
|
|
" to a tablespace which does not exist.\n"
|
|
"InnoDB: i/o type %lu, space id %lu,"
|
|
" page no. %lu, i/o length %lu bytes\n",
|
|
(ulong) type, (ulong) space_id, (ulong) block_offset,
|
|
(ulong) len);
|
|
|
|
return(DB_TABLESPACE_DELETED);
|
|
}
|
|
|
|
ut_ad((mode != OS_AIO_IBUF) || (space->purpose == FIL_TABLESPACE));
|
|
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
for (;;) {
|
|
if (UNIV_UNLIKELY(node == NULL)) {
|
|
fil_report_invalid_page_access(
|
|
block_offset, space_id, space->name,
|
|
byte_offset, len, type);
|
|
|
|
ut_error;
|
|
}
|
|
|
|
if (space->id != 0 && node->size == 0) {
|
|
/* We do not know the size of a single-table tablespace
|
|
before we open the file */
|
|
|
|
break;
|
|
}
|
|
|
|
if (node->size > block_offset) {
|
|
/* Found! */
|
|
break;
|
|
} else {
|
|
block_offset -= node->size;
|
|
node = UT_LIST_GET_NEXT(chain, node);
|
|
}
|
|
}
|
|
|
|
/* Open file if closed */
|
|
fil_node_prepare_for_io(node, fil_system, space);
|
|
|
|
/* Check that at least the start offset is within the bounds of a
|
|
single-table tablespace */
|
|
if (UNIV_UNLIKELY(node->size <= block_offset)
|
|
&& space->id != 0 && space->purpose == FIL_TABLESPACE) {
|
|
|
|
fil_report_invalid_page_access(
|
|
block_offset, space_id, space->name, byte_offset,
|
|
len, type);
|
|
|
|
ut_error;
|
|
}
|
|
|
|
/* Now we have made the changes in the data structures of fil_system */
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
/* Calculate the low 32 bits and the high 32 bits of the file offset */
|
|
|
|
if (!zip_size) {
|
|
offset_high = (block_offset >> (32 - UNIV_PAGE_SIZE_SHIFT));
|
|
offset_low = ((block_offset << UNIV_PAGE_SIZE_SHIFT)
|
|
& 0xFFFFFFFFUL) + byte_offset;
|
|
|
|
ut_a(node->size - block_offset
|
|
>= ((byte_offset + len + (UNIV_PAGE_SIZE - 1))
|
|
/ UNIV_PAGE_SIZE));
|
|
} else {
|
|
ulint zip_size_shift;
|
|
switch (zip_size) {
|
|
case 1024: zip_size_shift = 10; break;
|
|
case 2048: zip_size_shift = 11; break;
|
|
case 4096: zip_size_shift = 12; break;
|
|
case 8192: zip_size_shift = 13; break;
|
|
case 16384: zip_size_shift = 14; break;
|
|
default: ut_error;
|
|
}
|
|
offset_high = block_offset >> (32 - zip_size_shift);
|
|
offset_low = (block_offset << zip_size_shift & 0xFFFFFFFFUL)
|
|
+ byte_offset;
|
|
ut_a(node->size - block_offset
|
|
>= (len + (zip_size - 1)) / zip_size);
|
|
}
|
|
|
|
/* Do aio */
|
|
|
|
ut_a(byte_offset % OS_FILE_LOG_BLOCK_SIZE == 0);
|
|
ut_a((len % OS_FILE_LOG_BLOCK_SIZE) == 0);
|
|
|
|
#ifdef UNIV_HOTBACKUP
|
|
/* In ibbackup do normal i/o, not aio */
|
|
if (type == OS_FILE_READ) {
|
|
ret = os_file_read(node->handle, buf, offset_low, offset_high,
|
|
len);
|
|
} else {
|
|
ret = os_file_write(node->name, node->handle, buf,
|
|
offset_low, offset_high, len);
|
|
}
|
|
#else
|
|
/* Queue the aio request */
|
|
ret = os_aio(type, mode | wake_later, node->name, node->handle, buf,
|
|
offset_low, offset_high, len, node, message);
|
|
#endif
|
|
ut_a(ret);
|
|
|
|
if (mode == OS_AIO_SYNC) {
|
|
/* The i/o operation is already completed when we return from
|
|
os_aio: */
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
fil_node_complete_io(node, fil_system, type);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
ut_ad(fil_validate());
|
|
}
|
|
|
|
return(DB_SUCCESS);
|
|
}
|
|
|
|
#ifndef UNIV_HOTBACKUP
|
|
/**********************************************************************//**
|
|
Waits for an aio operation to complete. This function is used to write the
|
|
handler for completed requests. The aio array of pending requests is divided
|
|
into segments (see os0file.c for more info). The thread specifies which
|
|
segment it wants to wait for. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_aio_wait(
|
|
/*=========*/
|
|
ulint segment) /*!< in: the number of the segment in the aio
|
|
array to wait for */
|
|
{
|
|
ibool ret;
|
|
fil_node_t* fil_node;
|
|
void* message;
|
|
ulint type;
|
|
|
|
ut_ad(fil_validate());
|
|
|
|
if (srv_use_native_aio) {
|
|
srv_set_io_thread_op_info(segment, "native aio handle");
|
|
#ifdef WIN_ASYNC_IO
|
|
ret = os_aio_windows_handle(segment, 0, &fil_node,
|
|
&message, &type);
|
|
#elif defined(LINUX_NATIVE_AIO)
|
|
ret = os_aio_linux_handle(segment, &fil_node,
|
|
&message, &type);
|
|
#else
|
|
ret = 0; /* Eliminate compiler warning */
|
|
ut_error;
|
|
#endif
|
|
} else {
|
|
srv_set_io_thread_op_info(segment, "simulated aio handle");
|
|
|
|
ret = os_aio_simulated_handle(segment, &fil_node,
|
|
&message, &type);
|
|
}
|
|
|
|
ut_a(ret);
|
|
|
|
srv_set_io_thread_op_info(segment, "complete io for fil node");
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
fil_node_complete_io(fil_node, fil_system, type);
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
ut_ad(fil_validate());
|
|
|
|
/* Do the i/o handling */
|
|
/* IMPORTANT: since i/o handling for reads will read also the insert
|
|
buffer in tablespace 0, you have to be very careful not to introduce
|
|
deadlocks in the i/o system. We keep tablespace 0 data files always
|
|
open, and use a special i/o thread to serve insert buffer requests. */
|
|
|
|
if (fil_node->space->purpose == FIL_TABLESPACE) {
|
|
srv_set_io_thread_op_info(segment, "complete io for buf page");
|
|
buf_page_io_complete(message);
|
|
} else {
|
|
srv_set_io_thread_op_info(segment, "complete io for log");
|
|
log_io_complete(message);
|
|
}
|
|
}
|
|
#endif /* UNIV_HOTBACKUP */
|
|
|
|
/**********************************************************************//**
|
|
Flushes to disk possible writes cached by the OS. If the space does not exist
|
|
or is being dropped, does not do anything. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_flush(
|
|
/*======*/
|
|
ulint space_id) /*!< in: file space id (this can be a group of
|
|
log files or a tablespace of the database) */
|
|
{
|
|
fil_space_t* space;
|
|
fil_node_t* node;
|
|
os_file_t file;
|
|
ib_int64_t old_mod_counter;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
space = fil_space_get_by_id(space_id);
|
|
|
|
if (!space || space->is_being_deleted) {
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return;
|
|
}
|
|
|
|
space->n_pending_flushes++; /*!< prevent dropping of the space while
|
|
we are flushing */
|
|
node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
while (node) {
|
|
if (node->modification_counter > node->flush_counter) {
|
|
ut_a(node->open);
|
|
|
|
/* We want to flush the changes at least up to
|
|
old_mod_counter */
|
|
old_mod_counter = node->modification_counter;
|
|
|
|
if (space->purpose == FIL_TABLESPACE) {
|
|
fil_n_pending_tablespace_flushes++;
|
|
} else {
|
|
fil_n_pending_log_flushes++;
|
|
fil_n_log_flushes++;
|
|
}
|
|
#ifdef __WIN__
|
|
if (node->is_raw_disk) {
|
|
|
|
goto skip_flush;
|
|
}
|
|
#endif
|
|
retry:
|
|
if (node->n_pending_flushes > 0) {
|
|
/* We want to avoid calling os_file_flush() on
|
|
the file twice at the same time, because we do
|
|
not know what bugs OS's may contain in file
|
|
i/o; sleep for a while */
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
os_thread_sleep(20000);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
if (node->flush_counter >= old_mod_counter) {
|
|
|
|
goto skip_flush;
|
|
}
|
|
|
|
goto retry;
|
|
}
|
|
|
|
ut_a(node->open);
|
|
file = node->handle;
|
|
node->n_pending_flushes++;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
/* fprintf(stderr, "Flushing to file %s\n",
|
|
node->name); */
|
|
|
|
os_file_flush(file);
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
node->n_pending_flushes--;
|
|
skip_flush:
|
|
if (node->flush_counter < old_mod_counter) {
|
|
node->flush_counter = old_mod_counter;
|
|
|
|
if (space->is_in_unflushed_spaces
|
|
&& fil_space_is_flushed(space)) {
|
|
|
|
space->is_in_unflushed_spaces = FALSE;
|
|
|
|
UT_LIST_REMOVE(
|
|
unflushed_spaces,
|
|
fil_system->unflushed_spaces,
|
|
space);
|
|
}
|
|
}
|
|
|
|
if (space->purpose == FIL_TABLESPACE) {
|
|
fil_n_pending_tablespace_flushes--;
|
|
} else {
|
|
fil_n_pending_log_flushes--;
|
|
}
|
|
}
|
|
|
|
node = UT_LIST_GET_NEXT(chain, node);
|
|
}
|
|
|
|
space->n_pending_flushes--;
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
}
|
|
|
|
/**********************************************************************//**
|
|
Flushes to disk the writes in file spaces of the given type possibly cached by
|
|
the OS. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_flush_file_spaces(
|
|
/*==================*/
|
|
ulint purpose) /*!< in: FIL_TABLESPACE, FIL_LOG */
|
|
{
|
|
fil_space_t* space;
|
|
ulint* space_ids;
|
|
ulint n_space_ids;
|
|
ulint i;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
n_space_ids = UT_LIST_GET_LEN(fil_system->unflushed_spaces);
|
|
if (n_space_ids == 0) {
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
return;
|
|
}
|
|
|
|
/* Assemble a list of space ids to flush. Previously, we
|
|
traversed fil_system->unflushed_spaces and called UT_LIST_GET_NEXT()
|
|
on a space that was just removed from the list by fil_flush().
|
|
Thus, the space could be dropped and the memory overwritten. */
|
|
space_ids = mem_alloc(n_space_ids * sizeof *space_ids);
|
|
|
|
n_space_ids = 0;
|
|
|
|
for (space = UT_LIST_GET_FIRST(fil_system->unflushed_spaces);
|
|
space;
|
|
space = UT_LIST_GET_NEXT(unflushed_spaces, space)) {
|
|
|
|
if (space->purpose == purpose && !space->is_being_deleted) {
|
|
|
|
space_ids[n_space_ids++] = space->id;
|
|
}
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
/* Flush the spaces. It will not hurt to call fil_flush() on
|
|
a non-existing space id. */
|
|
for (i = 0; i < n_space_ids; i++) {
|
|
|
|
fil_flush(space_ids[i]);
|
|
}
|
|
|
|
mem_free(space_ids);
|
|
}
|
|
|
|
/******************************************************************//**
|
|
Checks the consistency of the tablespace cache.
|
|
@return TRUE if ok */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_validate(void)
|
|
/*==============*/
|
|
{
|
|
fil_space_t* space;
|
|
fil_node_t* fil_node;
|
|
ulint n_open = 0;
|
|
ulint i;
|
|
|
|
mutex_enter(&fil_system->mutex);
|
|
|
|
/* Look for spaces in the hash table */
|
|
|
|
for (i = 0; i < hash_get_n_cells(fil_system->spaces); i++) {
|
|
|
|
space = HASH_GET_FIRST(fil_system->spaces, i);
|
|
|
|
while (space != NULL) {
|
|
UT_LIST_VALIDATE(chain, fil_node_t, space->chain,
|
|
ut_a(ut_list_node_313->open
|
|
|| !ut_list_node_313->n_pending));
|
|
|
|
fil_node = UT_LIST_GET_FIRST(space->chain);
|
|
|
|
while (fil_node != NULL) {
|
|
if (fil_node->n_pending > 0) {
|
|
ut_a(fil_node->open);
|
|
}
|
|
|
|
if (fil_node->open) {
|
|
n_open++;
|
|
}
|
|
fil_node = UT_LIST_GET_NEXT(chain, fil_node);
|
|
}
|
|
space = HASH_GET_NEXT(hash, space);
|
|
}
|
|
}
|
|
|
|
ut_a(fil_system->n_open == n_open);
|
|
|
|
UT_LIST_VALIDATE(LRU, fil_node_t, fil_system->LRU, (void) 0);
|
|
|
|
fil_node = UT_LIST_GET_FIRST(fil_system->LRU);
|
|
|
|
while (fil_node != NULL) {
|
|
ut_a(fil_node->n_pending == 0);
|
|
ut_a(fil_node->open);
|
|
ut_a(fil_node->space->purpose == FIL_TABLESPACE);
|
|
ut_a(fil_node->space->id != 0);
|
|
|
|
fil_node = UT_LIST_GET_NEXT(LRU, fil_node);
|
|
}
|
|
|
|
mutex_exit(&fil_system->mutex);
|
|
|
|
return(TRUE);
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Returns TRUE if file address is undefined.
|
|
@return TRUE if undefined */
|
|
UNIV_INTERN
|
|
ibool
|
|
fil_addr_is_null(
|
|
/*=============*/
|
|
fil_addr_t addr) /*!< in: address */
|
|
{
|
|
return(addr.page == FIL_NULL);
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Get the predecessor of a file page.
|
|
@return FIL_PAGE_PREV */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_page_get_prev(
|
|
/*==============*/
|
|
const byte* page) /*!< in: file page */
|
|
{
|
|
return(mach_read_from_4(page + FIL_PAGE_PREV));
|
|
}
|
|
|
|
/********************************************************************//**
|
|
Get the successor of a file page.
|
|
@return FIL_PAGE_NEXT */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_page_get_next(
|
|
/*==============*/
|
|
const byte* page) /*!< in: file page */
|
|
{
|
|
return(mach_read_from_4(page + FIL_PAGE_NEXT));
|
|
}
|
|
|
|
/*********************************************************************//**
|
|
Sets the file page type. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_page_set_type(
|
|
/*==============*/
|
|
byte* page, /*!< in/out: file page */
|
|
ulint type) /*!< in: type */
|
|
{
|
|
ut_ad(page);
|
|
|
|
mach_write_to_2(page + FIL_PAGE_TYPE, type);
|
|
}
|
|
|
|
/*********************************************************************//**
|
|
Gets the file page type.
|
|
@return type; NOTE that if the type has not been written to page, the
|
|
return value not defined */
|
|
UNIV_INTERN
|
|
ulint
|
|
fil_page_get_type(
|
|
/*==============*/
|
|
const byte* page) /*!< in: file page */
|
|
{
|
|
ut_ad(page);
|
|
|
|
return(mach_read_from_2(page + FIL_PAGE_TYPE));
|
|
}
|
|
|
|
/********************************************************************
|
|
Initializes the tablespace memory cache. */
|
|
UNIV_INTERN
|
|
void
|
|
fil_close(void)
|
|
/*===========*/
|
|
{
|
|
#ifndef UNIV_HOTBACKUP
|
|
/* The mutex should already have been freed. */
|
|
ut_ad(fil_system->mutex.magic_n == 0);
|
|
#endif /* !UNIV_HOTBACKUP */
|
|
|
|
hash_table_free(fil_system->spaces);
|
|
|
|
hash_table_free(fil_system->name_hash);
|
|
|
|
ut_a(UT_LIST_GET_LEN(fil_system->LRU) == 0);
|
|
ut_a(UT_LIST_GET_LEN(fil_system->unflushed_spaces) == 0);
|
|
ut_a(UT_LIST_GET_LEN(fil_system->space_list) == 0);
|
|
|
|
mem_free(fil_system);
|
|
|
|
fil_system = NULL;
|
|
}
|