traversing the index B-tree twice (first in BTR_WATCH_LEAF mode and
then in BTR_DELETE mode), let BTR_DELETE take care of checking that
the record can be purged, and either buffering or performing the
purge.
row_purge_poss_sec(): New function, to check if it is possible to
purge a secondary index record. Refactored from
row_purge_remove_sec_if_poss_low().
row_purge_remove_sec_if_poss_nonbuffered(): Rename to
row_purge_remove_sec_if_poss_tree(). Remove the parameter mode
(always use BTR_MODIFY_TREE). Use row_purge_poss_sec().
row_purge_remove_sec_if_poss_low(): Rename to
row_purge_remove_sec_if_poss_leaf(). Remove the parameter mode
(always use BTR_MODIFY_LEAF). Let row_search_index_entry() do all the
hard work.
btr_cur_t: Add purge_node, which will be needed by
btr_cur_search_to_nth_level() for BTR_DELETE. Replace the flag value
BTR_CUR_ABORTED with BTR_CUR_DELETE_REF and BTR_CUR_DELETE_FAILED.
enum row_search_result, row_search_index_entry(): Replace
ROW_NOT_IN_POOL with ROW_NOT_DELETED_REF and ROW_NOT_DELETED.
btr_cur_search_to_nth_level(): Remove BTR_WATCH_LEAF. As a side
effect, the adaptive hash index can be used in purge as well. If
BTR_DELETE cannot be buffered, attempt btr_cur_optimistic_delete().
Either way, check row_purge_poss_sec(). Move the code to set
cursor->ibuf_count to get rid of another if (height == 0)
check. Eliminate the label loop_end. Do not call ibuf_should_try()
twice.
ibuf_should_try(): Now that the successful calls to this function will
be halved, halve the magic constant that ibuf_flush_count will be
compared to, accordingly.
The changes regarding ibuf_should_try() were merged from branches/zip
r3515.
rb://60 approved by Heikki over IM
------------------------------------------------------------------------
r3328 | marko | 2008-12-02 10:16:05 +0200 (Tue, 02 Dec 2008) | 7 lines
branches/zip: page_cur_insert_rec_zip(): When allocating insert_buf
from the free list, zero out the DB_TRX_ID and DB_ROLL_PTR of the
deleted record if the new record would not overwrite these
fields. This fixes a harmless content mismatch reported by
page_zip_validate() that was reported as Issue #111.
rb://55 approved by Sunny Bains.
------------------------------------------------------------------------
r3329 | vasil | 2008-12-02 12:03:17 +0200 (Tue, 02 Dec 2008) | 4 lines
branches/zip:
Add entry in the ChangeLog for the release of 1.0.2.
------------------------------------------------------------------------
r3331 | vasil | 2008-12-02 12:09:20 +0200 (Tue, 02 Dec 2008) | 5 lines
branches/zip:
Remove an entry from the ChnageLog for a change that was made before
the release of 1.0.2 but was not included in that release.
------------------------------------------------------------------------
r3333 | vasil | 2008-12-02 12:11:54 +0200 (Tue, 02 Dec 2008) | 4 lines
branches/zip:
Now that 1.0.2 is out, the current tree is version 1.0.3.
------------------------------------------------------------------------
r3336 | vasil | 2008-12-02 13:34:36 +0200 (Tue, 02 Dec 2008) | 5 lines
branches/zip:
Resurrect a ChangeLog entry that I removed in c3331 with a fake date
so it does not appear that it has been included in 1.0.2.
------------------------------------------------------------------------
r3361 | vasil | 2008-12-04 18:10:08 +0200 (Thu, 04 Dec 2008) | 9 lines
branches/zip:
Fix Mantis issue#103 mysql_addons.c depends on THD internals - please remove it
Use MySQL's thd_get_thread_id() instead of our own ib_thd_get_thread_id()
since MySQL implemented the former, see http://bugs.mysql.com/30930.
Approved by: Marko (https://svn.innodb.com/rb/r/40, rb://40)
------------------------------------------------------------------------
r3362 | vasil | 2008-12-04 18:49:24 +0200 (Thu, 04 Dec 2008) | 15 lines
branches/zip:
Revert our temporary fix for Bug#40360 Binlog related errors with binlog off
This bug was fixed in MySQL code.
Our fix went into r2944 and r2947, but this patch does not entirely revert
those revisions because we want to leave the test case that was introduced
and also r2944 itself reverted r2935 and r2936. So if we completely revert
r2944 and r2947 then we would loose the test and will restore r2935 and
r2936.
This resolves Issue#108 We should remove ib_bin_log_is_engaged() once mysql
add an equivallent, see Bug#40360
------------------------------------------------------------------------
r3404 | marko | 2008-12-05 10:02:54 +0200 (Fri, 05 Dec 2008) | 5 lines
branches/zip: row_undo_mod_del_mark_or_remove_sec_low(): Do not
complain if the record is not found. Explain that this is possible
when a deadlock occurs during an update. Heikki investigated this
in Issue #134.
------------------------------------------------------------------------
r3432 | marko | 2008-12-08 14:14:01 +0200 (Mon, 08 Dec 2008) | 4 lines
branches/zip: ibuf_merge_or_delete_for_page(): Remove the redundant
checks for ibuf_fixed_addr_page() || fsp_descr_page(). The one at
the beginning of the function is enough.
------------------------------------------------------------------------
r3456 | marko | 2008-12-09 10:31:26 +0200 (Tue, 09 Dec 2008) | 3 lines
branches/zip: row_purge_remove_sec_if_poss_low(): Allocate mtr_vers
from the stack, not with mem_alloc().
------------------------------------------------------------------------
r3459 | vasil | 2008-12-09 11:49:03 +0200 (Tue, 09 Dec 2008) | 152 lines
branches/zip:
Merge 2929:3458 from branches/5.1 (resolving conflict in c3257,
note also that r3363 reverted r2933 so there are not changes in
mysql-test/innodb-autoinc.result with the current merge):
------------------------------------------------------------------------
r2933 | calvin | 2008-10-30 02:57:31 +0200 (Thu, 30 Oct 2008) | 10 lines
Changed paths:
M /branches/5.1/mysql-test/innodb-autoinc.result
branches/5.1: correct the result file innodb-autoinc.result
Change the followings:
auto_increment_increment
auto_increment_offset
to
auto-increment-increment
auto-increment-offset
------------------------------------------------------------------------
r2981 | marko | 2008-11-07 14:54:10 +0200 (Fri, 07 Nov 2008) | 5 lines
Changed paths:
M /branches/5.1/row/row0mysql.c
branches/5.0: row_mysql_store_col_in_innobase_format(): Correct a misleading
comment. In the UTF-8 encoding, ASCII takes 1 byte per character, while
the "latin1" character set (normally ISO-8859-1, but in MySQL it actually
refers to the Windows Code Page 1252 a.k.a. CP1252, WinLatin1)
takes 1 to 3 bytes (1 to 2 bytes for the ISO-8859-1 subset).
------------------------------------------------------------------------
r3114 | calvin | 2008-11-14 20:31:48 +0200 (Fri, 14 Nov 2008) | 8 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
branches/5.1: fix bug#40386: Not flushing query cache after truncate
ha_statistics.records can not be 0 unless the table is empty, set to
1 instead. The original problem of bug#29507 is fixed in the server.
Additional test was done with the fix of bug#29507 in the server.
Approved by: Heikki (on IM)
------------------------------------------------------------------------
r3257 | inaam | 2008-11-24 22:06:50 +0200 (Mon, 24 Nov 2008) | 13 lines
Changed paths:
M /branches/5.1/handler/ha_innodb.cc
M /branches/5.1/srv/srv0srv.c
M /branches/5.1/trx/trx0trx.c
branches/5.1 bug#40760
The config param innodb_thread_concurrency is dynamically set and is
read when a thread enters/exits innodb. If the value is changed between
the enter and exit time the behaviour becomes erratic.
The fix is not to use srv_thread_concurrency when exiting, instead use
the flag trx->declared_to_be_inside_innodb.
rb://57
Approved by: Marko
------------------------------------------------------------------------
r3363 | calvin | 2008-12-04 19:00:20 +0200 (Thu, 04 Dec 2008) | 13 lines
Changed paths:
M /branches/5.1/mysql-test/innodb-autoinc.result
branches/5.1: revert the changes in r2933
The changes in r2933 causes test failure on Linux.
More investigation is needed for Windows.
Change the followings in innodb-autoinc.result:
auto-increment-increment
auto-increment-offset
back to:
auto_increment_increment
auto_increment_offset
------------------------------------------------------------------------
r3412 | vasil | 2008-12-05 10:46:18 +0200 (Fri, 05 Dec 2008) | 7 lines
Changed paths:
M /branches/5.1/trx/trx0undo.c
branches/5.1:
Add the traditional 2 spaces after the timestamp so the message does
not look like:
070223 13:26:01InnoDB: Warning: canno....
------------------------------------------------------------------------
r3458 | vasil | 2008-12-09 11:21:08 +0200 (Tue, 09 Dec 2008) | 51 lines
Changed paths:
M /branches/5.1/mysql-test/innodb_bug34300.test
branches/5.1:
Merge a change from MySQL to fix the failing innodb_bug34300 mysql-test:
main.innodb_bug34300 [ fail ]
mysqltest: At line 11: query 'SET @@max_allowed_packet=16777216' failed: 1621: SESSION variable 'max_allowed_packet' is read-only. Use SET GLOBAL to assign the value
Aborting: main.innodb_bug34300 failed in default mode.
The changeset is this:
------------------------------------------------------------
revno: 2709.1.10
committer: Staale Smedseng <staale.smedseng@sun.com>
branch nick: b22891-51-bugteam
timestamp: Thu 2008-11-20 08:51:48 +0100
message:
A fix for Bug#22891 "session level max_allowed_packet can be
set but is ignored".
This patch makes @@session.max_allowed_packed and
@@session.net_buffer_length read-only as suggested in the bug
report. The user will have to use SET GLOBAL (and reconnect)
to alter the session values of these variables.
The error string ER_VARIABLE_IS_READONLY is introduced.
Tests are modified accordingly.
modified:
mysql-test/r/func_compress.result
mysql-test/r/max_allowed_packet_basic.result
mysql-test/r/max_allowed_packet_func.result
mysql-test/r/net_buffer_length_basic.result
mysql-test/r/packet.result
mysql-test/r/union.result
mysql-test/r/variables.result
mysql-test/t/func_compress.test
mysql-test/t/innodb_bug34300.test
mysql-test/t/max_allowed_packet_basic.test
mysql-test/t/max_allowed_packet_func.test
mysql-test/t/net_buffer_length_basic.test
mysql-test/t/packet.test
mysql-test/t/union.test
mysql-test/t/variables.test
sql/set_var.cc
sql/set_var.h
sql/share/errmsg.txt
------------------------------------------------------------
------------------------------------------------------------------------
------------------------------------------------------------------------
determined from flag == BTR_CUR_ABORTED. This flag value was previously
never tested for. It was only assigned to, in the only place where
leaf_in_buf_pool would be set FALSE.
btr_cur_search_to_nth_level(): Do not initialize cursor->leaf_in_buf_pool.
btr_pcur_was_buffered(): Remove. The only caller, row_search_index_entry(),
will switch (cursor->flag) and handle BTR_CUR_ABORTED as well.
row_purge_remove_sec_if_poss_low(): Remove the assertions on leaf_in_buf_pool.
------------------------------------------------------------------------
r3254 | marko | 2008-11-24 18:01:42 +0200 (Mon, 24 Nov 2008) | 4 lines
branches/zip: Note that it is legitimate for a secondary index record not
to be found during purge. This tries to address Issue #129. The comments
were supplied by Heikki.
------------------------------------------------------------------------
r3286 | marko | 2008-11-26 10:00:28 +0200 (Wed, 26 Nov 2008) | 18 lines
branches/zip: row_merge_drop_temp_indexes(): Replace the WHILE 1 with
WHILE 1=1 in the SQL procedure, so that the loop will actually be
entered and temporary indexes be dropped during crash recovery.
Thanks to Sunny Bains for pointing this out.
Tested as follows:
Set a breakpoint in row_merge_rename_indexes.
CREATE TABLE t(a INT)ENGINE=InnoDB;
CREATE INDEX a ON t(a);
-- The breakpoint will be reached. Kill and restart mysqld.
SHOW CREATE TABLE t;
-- This shows the MySQL .frm file, without and index.
CREATE TABLE innodb_table_monitor(a INT)ENGINE=InnoDB;
-- This will dump the InnoDB dictionary to the error log, without the index.
------------------------------------------------------------------------
r3302 | vasil | 2008-11-27 23:26:39 +0200 (Thu, 27 Nov 2008) | 12 lines
branches/zip:
Fix Mantis issue#130 wdl: does not handle 64-bit address
- Change the call from strtoul() to strtoull()
- Change "%16X" to "%16llx" when scanning preferred load address
rb://58
Submitted by: Calvin
Approved by: Marko
------------------------------------------------------------------------
r3303 | vasil | 2008-11-27 23:31:18 +0200 (Thu, 27 Nov 2008) | 10 lines
branches/zip:
* Remove a change from win-plugin/win-plugin.diff about time_t because
MySQL has used VS2005 for building 5.1.30.
* Adjust the line numbers so the patch applies cleanly without fuzz and
offset messages.
Submitted by: Calvin
------------------------------------------------------------------------
r3304 | vasil | 2008-11-27 23:33:48 +0200 (Thu, 27 Nov 2008) | 6 lines
branches/zip:
Non-functional change in win-plugin/win-plugin.diff: fix the file name
before the diff, this is irrelevant but it is nice to be the same as
the file name on the following line.
------------------------------------------------------------------------
r3312 | marko | 2008-11-28 16:18:43 +0200 (Fri, 28 Nov 2008) | 5 lines
branches/zip: row_undo_mod_del_mark_or_remove_sec_low(): Complain if
the secondary index entry cannot be found, and this is not an incomplete
transaction that is being rolled back in crash recovery. The source code
comments were suggested by Heikki.
------------------------------------------------------------------------
and add assertions to callers. Remove the added ibool* parameter and
return enum row_search_result instead of ibool, to reflect the four
different outcomes: record found, not found, buffered, and status unknown
(record not in the buffer pool).
rb://6
buf_pool->watch_ fields to start with the common prefix buf_pool_watch,
so that they can be searched easily:
buf_pool_watch_set(): Renamed from buf_pool_add_watch(). The "add" was
misleading, because only one watch can be active at a time.
buf_pool_watch_clear(): Renamed from buf_pool_remove_watch().
buf_pool_watch_occurred(): Renamed from buf_pool_watch_happened().
The followin mysql-tests failed (and they are known to fail):
main.information_schema [ fail ]
main.innodb_file_per_table_basic[ fail ]
main.type_bit_innodb [ fail ]
Tested against : MYSQL_SERVER_VERSION "5.1.28"
symbols. Use it for all definitions of non-static variables and functions.
lexyy.c, make_flex.sh: Declare yylex as UNIV_INTERN, not static. It is
referenced from pars0grm.c.
Actually, according to
nm .libs/ha_innodb.so|grep -w '[ABCE-TVXYZ]'
the following symbols are still global:
* The vtable for class ha_innodb
* pars0grm.c: The function yyparse() and the variables yychar, yylval, yynerrs
The required changes to the Bison-generated file pars0grm.c will be addressed
in a separate commit, which will add a script similar to make_flex.sh.
The class ha_innodb is renamed from class ha_innobase by a #define. Thus,
there will be no clash with the builtin InnoDB. However, there will be some
overhead for invoking virtual methods of class ha_innodb. Ideas for making
the vtable hidden are welcome. -fvisibility=hidden is not available in GCC 3.
enough prefixes of externally stored columns, so that purge will not have
to dereference any BLOB pointers, which may be invalid. This will not be
necessary for logging inserts, because inserts are no-ops in purge, and
the record will remain locked during transaction rollback.
TODO: in dict_build_table_def_step() or dict_build_index_def_step(),
prevent the creation of tables with too many columns for which a
prefix index is defined. This is because there is a size limit of undo
log records, and for each prefix-indexed column, the log must store
REC_MAX_INDEX_COL_LEN + BTR_EXTERN_FIELD_REF_SIZE bytes.
trx_undo_page_report_insert(): Assert that the index is clustered.
trx_undo_page_fetch_ext(): New function, for fetching the BLOB prefix
in trx_undo_page_report_modify().
trx_undo_page_report_modify(): Write long enough prefixes of the externally
stored columns to the undo log.
trx_undo_rec_get_partial_row(): Remove the parameter "ext". Assert that
the undo log contains long enough prefixes of the externally stored columns.
purge_node_t: Remove the field "ext".
inserted, uncommitted clustered index records when determining if a
secondary index record that contains a column prefix of an externally
stored column is referencing the clustered index record.
field_ref_zero[]: A BLOB pointer full of zero, for use in comparisons.
btr_copy_externally_stored_field_prefix(): Assert that the BLOB pointer is set.
row_ext_lookup_ith(), row_ext_lookup(), row_ext_lookup_low(): Document
that field_ref_zero is returned when the BLOB cannot be fetched.
row_ext_lookup_low(): Return field_ref_zero and *len = 0 when the
BLOB pointer is unset.
row_build_index_entry(): Return NULL when a needed BLOB pointer cannot
be dereferenced (row_ext_lookup returns field_ref_zero). Check the
return value for NULL in callers.
row_vers_impl_x_locked_off_kernel(): Avoid comparisons when
row_build_index_entry() returns NULL.
row_vers_old_has_index_entry(): Ignore records for which
row_build_index_entry() returns NULL. The entry should never be NULL
in rollback, but it may be NULL in purge.
row_merge_buf_add(): Assert that row_ext_lookup() does not return
field_ref_zero. The table will be locked during index creation.
offsets_[] arrays, as suggested by Vasil.
rec_offs_set_n_alloc(): Declare as a public function. Assert that
n_alloc > REC_OFFS_HEADER_SIZE.
rec_offs_get_n_alloc(): Assert that n_alloc > REC_OFFS_HEADER_SIZE.
row_upd_index_is_referenced(), row_upd_check_references_constraints(),
row_purge_parse_undo_rec(): Remove duplicated code that calls
row_mysql_unfreeze_data_dictionary().
row_undo_ins_remove_clust_rec():
Assert that the data dictionary is X-latched. Do not attempt to re-latch
the data dictionary, because doing so introduces race conditions. This
re-latching was probably made unnecessary already in r1676, which keeps
the data dictionary X-latched in ha_innobase::add_index() for the complete
duration of the data dictionary operations.
row_undo_ins_parse_undo_rec(): Do not attempt to acquire the data dictionary
latch. The data dictionary should be already X-latched when a transaction
that modifies system tables is rolled back. This also holds in the rollback
during crash recovery: trx_rollback_active() will X-latch the data dictionary
when needed.
Some things still fail in innodb-index.test, and there seems to be
a race condition (data dictionary lock wait) when running with --valgrind.
dfield_t: Add an "external storage" flag, dfield->ext.
dfield_is_null(), dfield_is_ext(), dfield_set_ext(), dfield_set_null():
New functions.
dfield_copy(), dfield_copy_data(): Add const qualifiers, fix in/out comments.
data_write_sql_null(): Use memset().
big_rec_field_t: Replace byte* data with const void* data.
ut_ulint_sort(): Remove.
upd_field_t: Remove extern_storage.
upd_node_t: Replace ext_vec, n_ext_vec with n_ext.
row_merge_copy_blobs(): New function.
row_ins_index_entry(): Add the parameter "ibool foreign" for suppressing
foreign key checks during fast index creation or when inserting into
secondary indexes.
btr_page_insert_fits(): Add const qualifiers.
btr_cur_add_ext(), upd_ext_vec_contains(): Remove.
dfield_print_also_hex(), dfield_print(): Replace if...else if with switch.
Observe dfield_is_ext().
buf_page_get_gen(). This saves one mutex operation per block request.
buf_page_get_gen(), various macros and functions: Add parameter zip_size.
btr_node_ptr_get_child(): Add parameter index.
fil_space_get_latch(): Add optional output parameter zip_size.
fil_space_get_zip_size(): Return 0 for space id==0, because the
system tablespace is never compressed.
fsp_header_init(): Remove the parameter zip_size.
ibuf_free_excess_pages(): Remove the parameter zip_size.
trx_rseg_t, trx_undo_t: Add field zip_size.
xdes_lst_get_next(): Remove, unused.
in the clustered index to be smaller than the indexed prefix in secondary
indexes.
row_ext_lookup(): Return NULL if the column is not stored externally.
trx_undo_rec_get_partial_row(): row_build(): Add parameter row_ext_t** ext.
row_build_index_entry(): Add the parameter row_ext_t* ext.
Invoke row_ext_lookup() to fetch prefixes of externally stored columns.
upd_node_t, undo_node_t, purge_node_t: Add the field row_ext_t* ext.
btr_compress(): Defer any tree modifications until the records
have successfully been copied.
row_purge_remove_sec_if_poss_low(): Simplify the setting of
the return status.
Replace btr_page_get_level() with page_is_leaf() where possible.
row_purge_upd_exist_or_extern(): Remove obsolete TODO comment.
dtuple_convert_big_rec(): Replace a flag variable with goto.
Fix the way how btr_free_externally_stored_field() is called in purge.
btr_free_externally_stored_field(): Add parameter field_ref that points
directly to the BLOB reference. Use rec, offsets, page_zip, and i
only for the page_zip_write_blob_ptr() call.
row_purge_upd_exist_or_extern(): Do not assume that the undo log contains
the entire record. Only pass the BLOB reference to
btr_free_externally_stored_field().
BLOB pointers, trx_id, and roll_ptr.
btr_empty(), btr_create(), page_create(): Add parameter "index", as some
index information will be encoded on the compressed page.
Define REC_NODE_PTR_SIZE as 4.
Allow btr_page_reorganize() and btr_page_reorganize_low() to fail.
Define the error code DB_ZIP_OVERFLOW.
Make row_ins_index_entry_low() static.
page0zip: Encode the index, log reorganized records, and store uncompressed
fields separately from the compressed data stream.
