bug #57006 "Deadlock between HANDLER and FLUSH TABLES WITH READ
LOCK" and bug #54673 "It takes too long to get readlock for
'FLUSH TABLES WITH READ LOCK'".
The first bug manifested itself as a deadlock which occurred
when a connection, which had some table open through HANDLER
statement, tried to update some data through DML statement
while another connection tried to execute FLUSH TABLES WITH
READ LOCK concurrently.
What happened was that FTWRL in the second connection managed
to perform first step of GRL acquisition and thus blocked all
upcoming DML. After that it started to wait for table open
through HANDLER statement to be flushed. When the first connection
tried to execute DML it has started to wait for GRL/the second
connection creating deadlock.
The second bug manifested itself as starvation of FLUSH TABLES
WITH READ LOCK statements in cases when there was a constant
stream of concurrent DML statements (in two or more
connections).
This has happened because requests for protection against GRL
which were acquired by DML statements were ignoring presence of
pending GRL and thus the latter was starved.
This patch solves both these problems by re-implementing GRL
using metadata locks.
Similar to the old implementation acquisition of GRL in new
implementation is two-step. During the first step we block
all concurrent DML and DDL statements by acquiring global S
metadata lock (each DML and DDL statement acquires global IX
lock for its duration). During the second step we block commits
by acquiring global S lock in COMMIT namespace (commit code
acquires global IX lock in this namespace).
Note that unlike in old implementation acquisition of
protection against GRL in DML and DDL is semi-automatic.
We assume that any statement which should be blocked by GRL
will either open and acquires write-lock on tables or acquires
metadata locks on objects it is going to modify. For any such
statement global IX metadata lock is automatically acquired
for its duration.
The first problem is solved because waits for GRL become
visible to deadlock detector in metadata locking subsystem
and thus deadlocks like one in the first bug become impossible.
The second problem is solved because global S locks which
are used for GRL implementation are given preference over
IX locks which are acquired by concurrent DML (and we can
switch to fair scheduling in future if needed).
Important change:
FTWRL/GRL no longer blocks DML and DDL on temporary tables.
Before this patch behavior was not consistent in this respect:
in some cases DML/DDL statements on temporary tables were
blocked while in others they were not. Since the main use cases
for FTWRL are various forms of backups and temporary tables are
not preserved during backups we have opted for consistently
allowing DML/DDL on temporary tables during FTWRL/GRL.
Important change:
This patch changes thread state names which are used when
DML/DDL of FTWRL is waiting for global read lock. It is now
either "Waiting for global read lock" or "Waiting for commit
lock" depending on the stage on which FTWRL is.
Incompatible change:
To solve deadlock in events code which was exposed by this
patch we have to replace LOCK_event_metadata mutex with
metadata locks on events. As result we have to prohibit
DDL on events under LOCK TABLES.
This patch also adds extensive test coverage for interaction
of DML/DDL and FTWRL.
Performance of new and old global read lock implementations
in sysbench tests were compared. There were no significant
difference between new and old implementations.
Merge 5.3-mwl89 into 5.3 main.
There is one remaining test failure in this merge:
innodb_mysql_lock2. All other tests have been checked to
deliver the same results/explains as 5.3-mwl89, including
the few remaining wrong results.
This is the 5.5 version of the fix. The 5.1 version was too complicated to
merge and was null merged.
This is a regression from the fix for bug no 38999. A storage engine capable
of reading only a subset of a table's columns updates corresponding bits in
the read buffer to signal that it has read NULL values for the corresponding
columns. It cannot, and should not, update any other bits. Bug no 38999
occurred because the implementation of UPDATE statements compare the NULL bits
using memcmp, inadvertently comparing bits that were never requested from the
storage engine. The regression was caused by the storage engine trying to
alleviate the situation by writing to all NULL bits, even those that it had no
knowledge of. This has devastating effects for the index merge algorithm,
which relies on all NULL bits, except those explicitly requested, being left
unchanged.
The fix reverts the fix for bug no 38999 in both InnoDB and InnoDB plugin and
changes the server's method of comparing records. For engines that always read
entire rows, we proceed as usual. For engines capable of reading only select
columns, the record buffers are now compared on a column by column basis. An
assertion was also added so that non comparable buffers are never read. Some
relevant copy-pasted code was also consolidated in a new function.
This is the 5.5 version of the fix. The 5.1 version was too complicated to
merge and was null merged.
This is a regression from the fix for bug no 38999. A storage engine capable
of reading only a subset of a table's columns updates corresponding bits in
the read buffer to signal that it has read NULL values for the corresponding
columns. It cannot, and should not, update any other bits. Bug no 38999
occurred because the implementation of UPDATE statements compare the NULL bits
using memcmp, inadvertently comparing bits that were never requested from the
storage engine. The regression was caused by the storage engine trying to
alleviate the situation by writing to all NULL bits, even those that it had no
knowledge of. This has devastating effects for the index merge algorithm,
which relies on all NULL bits, except those explicitly requested, being left
unchanged.
The fix reverts the fix for bug no 38999 in both InnoDB and InnoDB plugin and
changes the server's method of comparing records. For engines that always read
entire rows, we proceed as usual. For engines capable of reading only select
columns, the record buffers are now compared on a column by column basis. An
assertion was also added so that non comparable buffers are never read. Some
relevant copy-pasted code was also consolidated in a new function.
This is a regression from the fix for bug no 38999. A storage engine capable
of reading only a subset of a table's columns updates corresponding bits in
the read buffer to signal that it has read NULL values for the corresponding
columns. It cannot, and should not, update any other bits. Bug no 38999
occurred because the implementation of UPDATE statements compare the NULL bits
using memcmp, inadvertently comparing bits that were never requested from the
storage engine. The regression was caused by the storage engine trying to
alleviate the situation by writing to all NULL bits, even those that it had no
knowledge of. This has devastating effects for the index merge algorithm,
which relies on all NULL bits, except those explicitly requested, being left
unchanged.
The fix reverts the fix for bug no 38999 in both InnoDB and InnoDB plugin and
changes the server's method of comparing records. For engines that always read
entire rows, we proceed as usual. For engines capable of reading only select
columns, the record buffers are now compared on a column by column basis. An
assertion was also added so that non comparable buffers are never read. Some
relevant copy-pasted code was also consolidated in a new function.
This is a regression from the fix for bug no 38999. A storage engine capable
of reading only a subset of a table's columns updates corresponding bits in
the read buffer to signal that it has read NULL values for the corresponding
columns. It cannot, and should not, update any other bits. Bug no 38999
occurred because the implementation of UPDATE statements compare the NULL bits
using memcmp, inadvertently comparing bits that were never requested from the
storage engine. The regression was caused by the storage engine trying to
alleviate the situation by writing to all NULL bits, even those that it had no
knowledge of. This has devastating effects for the index merge algorithm,
which relies on all NULL bits, except those explicitly requested, being left
unchanged.
The fix reverts the fix for bug no 38999 in both InnoDB and InnoDB plugin and
changes the server's method of comparing records. For engines that always read
entire rows, we proceed as usual. For engines capable of reading only select
columns, the record buffers are now compared on a column by column basis. An
assertion was also added so that non comparable buffers are never read. Some
relevant copy-pasted code was also consolidated in a new function.
- Changed to still use bcmp() in certain cases becasue
- Faster for short unaligneed strings than memcmp()
- Bettern when using valgrind
- Changed to use my_sprintf() instead of sprintf() to get higher portability for old systems
- Changed code to use MariaDB version of select->skip_record()
- Removed -%::SCCS/s.% from Makefile.am:s to remove automake warnings
INSERT IGNORE ... SELECT ... UNION SELECT ...
This assert was triggered by INSERT IGNORE ... SELECT. The assert checks that a
statement either sends OK or an error to the client. If the bug was triggered
on release builds, it caused OK to be sent to the client instead of the correct
error message (in this case ER_FIELD_SPECIFIED_TWICE).
The reason the assert was triggered, was that lex->no_error was set to TRUE
during JOIN::optimize() because of IGNORE. This causes all errors to be ignored.
However, not all errors can be ignored. Some, such as ER_FIELD_SPECIFIED_TWICE
will cause the INSERT to fail no matter what. But since lex->no_error was set,
the critical errors were ignored, the INSERT failed and neither OK nor the
error message was sent to the client.
This patch fixes the problem by temporarily turning off lex->no_error in
places where errors cannot be ignored during processing of INSERT ... SELECT.
Test case added to insert.test.
INSERT IGNORE ... SELECT ... UNION SELECT ...
This assert was triggered by INSERT IGNORE ... SELECT. The assert checks that a
statement either sends OK or an error to the client. If the bug was triggered
on release builds, it caused OK to be sent to the client instead of the correct
error message (in this case ER_FIELD_SPECIFIED_TWICE).
The reason the assert was triggered, was that lex->no_error was set to TRUE
during JOIN::optimize() because of IGNORE. This causes all errors to be ignored.
However, not all errors can be ignored. Some, such as ER_FIELD_SPECIFIED_TWICE
will cause the INSERT to fail no matter what. But since lex->no_error was set,
the critical errors were ignored, the INSERT failed and neither OK nor the
error message was sent to the client.
This patch fixes the problem by temporarily turning off lex->no_error in
places where errors cannot be ignored during processing of INSERT ... SELECT.
Test case added to insert.test.
Fixed some wrong test cases
Fixed bug in null handling in XtraDB
extra/comp_err.c:
Fixed compiler warnings
extra/my_print_defaults.c:
Fixed compiler warnings
mysql-test/suite/binlog/t/binlog_killed.test:
Added support for timeouts
mysql-test/suite/funcs_1/r/is_columns_is.result:
Updated results (INNODB_SYS_TABLES had got new column)
scripts/mysql_install_db.sh:
Fixed typo
sql/mysql_priv.h:
Removed not needed argument for compare_record()
sql/sql_insert.cc:
Removed not needed argument for compare_record()
sql/sql_update.cc:
Removed not needed argument for compare_record()
The argument is not needed becasue we copy the full record[0] to record[1] and the comparison should work even if all columns are not read
sql/table.cc:
The comparison of rows is independent of HA_PARTIAL_COLUMN_READ
storage/maria/maria_chk.c:
Fixed compiler warnings
storage/maria/maria_read_log.c:
Fixed compiler warnings
storage/myisam/myisamchk.c:
Fixed compiler warnings
storage/myisam/myisampack.c:
Fixed compiler warnings
storage/xtradb/dict/dict0load.c:
Fixed compiler warnings
storage/xtradb/row/row0sel.c:
Fixed null handling in XtraDB. (See comment)
storage/xtradb/trx/trx0sys.c:
Fixed compiler warnings
support-files/compiler_warnings.supp:
Fixed compiler warnings
This was triggered by innodb.innodb_multi_update, where we had a static length row without nulls and xtradb didn't fill in the delete-marker byte
include/my_bitmap.h:
Added prototype for bitmap_union_is_set_all()
mysys/my_bitmap.c:
Added function to check if union of two bit maps covers all bits.
sql/mysql_priv.h:
Updated protype for compare_record()
sql/sql_insert.cc:
Send to compare_record() flag if all fields are used.
sql/sql_select.cc:
Set share->null_bytes_for_compare.
sql/sql_update.cc:
In compare_record() don't use the fast cmp_record() (which is basically memcmp) if we don't know that all fields exists.
Don't compare the null_bytes if there is no data there.
sql/table.cc:
Store in share->null_bytes_for_compare the number of bytes that has null or bit fields (but not delete marker)
Store in can_cmp_whole_record if we can use memcmp() (assuming all rows are read) to compare rows in compare_record()
sql/table.h:
Added two elements in table->share to speed up checking how updated rows can be compared.
This assert checks that the server does not try to send OK to the
client if there has been some error during processing. This is done
to make sure that the error is in fact sent to the client.
The problem was that view errors during processing of WHERE conditions
in UPDATE statements where not detected by the update code. It therefore
tried to send OK to the client, triggering the assert.
The bug was only noticeable in debug builds.
This patch fixes the problem by making sure that the update code
checks for errors during condition processing and acts accordingly.
This assert checks that the server does not try to send OK to the
client if there has been some error during processing. This is done
to make sure that the error is in fact sent to the client.
The problem was that view errors during processing of WHERE conditions
in UPDATE statements where not detected by the update code. It therefore
tried to send OK to the client, triggering the assert.
The bug was only noticeable in debug builds.
This patch fixes the problem by making sure that the update code
checks for errors during condition processing and acts accordingly.
If the expression for a virtual column of table contained datetime
comparison then the execution of the second query that used this
virtual column caused a crash. It happened because the execution
of the first query that used this virtual column inserted a cached
item into the expression tree. The cached tree was allocated in
the statement memory while the expression tree was allocated in
the table memory.
Now the cached items that are inserted into expressions for virtual
columns with datetime comparisons are always allocated in the same
mem_root as the expressions for virtual columns. So now the inserted
cached items are valid for any queries that use these virtual columns.
The issue was that we didn't always check result of ha_rnd_init() which caused a problem for handlers that returned an error in this code.
- Changed prototype of ha_rnd_init() to ensure that we get a compile warning if result is not checked.
- Added ha_rnd_init_with_error() that prints error on failure.
- Checked all usage of ha_rnd_init() and ensure we generate an error message on failures.
- Changed init_read_record() to return 1 on failure.
sql/create_options.cc:
Fixed wrong printf
sql/event_db_repository.cc:
Check result from init_read_record()
sql/events.cc:
Check result from init_read_record()
sql/filesort.cc:
Check result from ha_rnd_init()
sql/ha_partition.cc:
Check result from ha_rnd_init()
sql/ha_partition.h:
Fixed compiler warning
sql/handler.cc:
Added ha_rnd_init_with_error()
Check result from ha_rnd_init()
sql/handler.h:
Added ha_rnd_init_with_error()
Changed prototype of ha_rnd_init() to ensure that we get a compile warning if result is not checked
sql/item_subselect.cc:
Check result from ha_rnd_init()
sql/log.cc:
Check result from ha_rnd_init()
sql/log_event.cc:
Check result from ha_rnd_init()
sql/log_event_old.cc:
Check result from ha_rnd_init()
sql/mysql_priv.h:
init_read_record() now returns error code on failure
sql/opt_range.cc:
Check result from ha_rnd_init()
sql/records.cc:
init_read_record() now returns error code on failure
Check result from ha_rnd_init()
sql/sql_acl.cc:
Check result from init_read_record()
sql/sql_cursor.cc:
Print error if ha_rnd_init() fails
sql/sql_delete.cc:
Check result from init_read_record()
sql/sql_help.cc:
Check result from init_read_record()
sql/sql_plugin.cc:
Check result from init_read_record()
sql/sql_select.cc:
Check result from ha_rnd_init()
Print error if ha_rnd_init() fails.
sql/sql_servers.cc:
Check result from init_read_record()
sql/sql_table.cc:
Check result from init_read_record()
sql/sql_udf.cc:
Check result from init_read_record()
sql/sql_update.cc:
Check result from init_read_record()
storage/example/ha_example.cc:
Don't return error on rnd_init()
storage/ibmdb2i/ha_ibmdb2i.cc:
Removed not relevant comment
libmysqld/Makefile.am:
The new file added.
mysql-test/r/index_merge_myisam.result:
subquery_cache optimization option added.
mysql-test/r/myisam_mrr.result:
subquery_cache optimization option added.
mysql-test/r/subquery_cache.result:
The subquery cache tests added.
mysql-test/r/subselect3.result:
Subquery cache switched off to avoid changing read statistics.
mysql-test/r/subselect3_jcl6.result:
Subquery cache switched off to avoid changing read statistics.
mysql-test/r/subselect_no_mat.result:
subquery_cache optimization option added.
mysql-test/r/subselect_no_opts.result:
subquery_cache optimization option added.
mysql-test/r/subselect_no_semijoin.result:
subquery_cache optimization option added.
mysql-test/r/subselect_sj.result:
subquery_cache optimization option added.
mysql-test/r/subselect_sj_jcl6.result:
subquery_cache optimization option added.
mysql-test/t/subquery_cache.test:
The subquery cache tests added.
mysql-test/t/subselect3.test:
Subquery cache switched off to avoid changing read statistics.
sql/CMakeLists.txt:
The new file added.
sql/Makefile.am:
The new files added.
sql/item.cc:
Expression cache item (Item_cache_wrapper) added.
Item_ref and Item_field fixed for correct usage of result field and fast resolwing in SP.
sql/item.h:
Expression cache item (Item_cache_wrapper) added.
Item_ref and Item_field fixed for correct usage of result field and fast resolwing in SP.
sql/item_cmpfunc.cc:
Subquery cache added.
sql/item_cmpfunc.h:
Subquery cache added.
sql/item_subselect.cc:
Subquery cache added.
sql/item_subselect.h:
Subquery cache added.
sql/item_sum.cc:
Registration of subquery parameters added.
sql/mysql_priv.h:
subquery_cache optimization option added.
sql/mysqld.cc:
subquery_cache optimization option added.
sql/opt_range.cc:
Fix due to subquery cache.
sql/opt_subselect.cc:
Parameters of the function cahnged.
sql/procedure.h:
.h file guard added.
sql/sql_base.cc:
Registration of subquery parameters added.
sql/sql_class.cc:
Option to allow add indeces to temporary table.
sql/sql_class.h:
Item iterators added.
Option to allow add indeces to temporary table.
sql/sql_expression_cache.cc:
Expression cache for caching subqueries added.
sql/sql_expression_cache.h:
Expression cache for caching subqueries added.
sql/sql_lex.cc:
Registration of subquery parameters added.
sql/sql_lex.h:
Registration of subqueries and subquery parameters added.
sql/sql_select.cc:
Subquery cache added.
sql/sql_select.h:
Subquery cache added.
sql/sql_union.cc:
A new parameter to the function added.
sql/sql_update.cc:
A new parameter to the function added.
sql/table.cc:
Procedures to manage temporarty tables index added.
sql/table.h:
Procedures to manage temporarty tables index added.
storage/maria/ha_maria.cc:
Fix of handler to allow destoy a table in case of error during the table creation.
storage/maria/ha_maria.h:
.h file guard added.
storage/myisam/ha_myisam.cc:
Fix of handler to allow destoy a table in case of error during the table creation.
The functions mysql_delete and mysql_update lacked calls of
updated_virtual_fields(). This caused wrong results for
some DELETEs/UPDATEs.
Added test cases for this bug.
use limit efficiently
Bug #36569: UPDATE ... WHERE ... ORDER BY... always does a
filesort even if not required
Also two bugs reported after QA review (before the commit
of bugs above to public trees, no documentation needed):
Bug #53737: Performance regressions after applying patch
for bug 36569
Bug #53742: UPDATEs have no effect after applying patch
for bug 36569
Execution of single-table UPDATE and DELETE statements did not use the
same optimizer as was used in the compilation of SELECT statements.
Instead, it had an optimizer of its own that did not take into account
that you can omit sorting by retrieving rows using an index.
Extra optimization has been added: when applicable, single-table
UPDATE/DELETE statements use an existing index instead of filesort. A
corresponding SELECT query would do the former.
Also handling of the DESC ordering expression has been added when
reverse index scan is applicable.
From now on most single table UPDATE and DELETE statements show the
same disk access patterns as the corresponding SELECT query. We verify
this by comparing the result of SHOW STATUS LIKE 'Sort%
Currently the get_index_for_order function
a) checks quick select index (if any) for compatibility with the
ORDER expression list or
b) chooses the cheapest available compatible index, but only if
the index scan is cheaper than filesort.
Second way is implemented by the new test_if_cheaper_ordering
function (extracted part the test_if_skip_sort_order()).
mysql-test/r/log_state.result:
Updated result for optimized query, bug #36569.
mysql-test/r/single_delete_update.result:
Test case for bug #30584, bug #36569 and bug #53742.
mysql-test/r/update.result:
Updated result for optimized query, bug #30584.
Note:
"Handler_read_last 1" omitted, see bug 52312:
lost Handler_read_last status variable.
mysql-test/t/single_delete_update.test:
Test case for bug #30584, bug #36569 and bug #53742.
sql/opt_range.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
* get_index_for_order() has been rewritten entirely and moved
to sql_select.cc
New QUICK_RANGE_SELECT::make_reverse method has been added.
sql/opt_range.h:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
* get_index_for_order() has been rewritten entirely and moved
to sql_select.cc
New functions:
* QUICK_SELECT_I::make_reverse()
* SQL_SELECT::set_quick()
sql/records.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
* init_read_record_idx() has been modified to allow reverse index scan
New functions:
* rr_index_last()
* rr_index_desc()
sql/records.h:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
init_read_record_idx() has been modified to allow reverse index scan
sql/sql_delete.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
mysql_delete: an optimization has been added to skip
unnecessary sorting with ORDER BY clause where select
result ordering is acceptable.
sql/sql_select.cc:
Bug #30584, bug #36569, bug #53737, bug #53742:
UPDATE/DELETE ... WHERE ... ORDER BY... always does a filesort
even if not required
The const_expression_in_where function has been modified
to accept both Item and Field pointers.
New functions:
* get_index_for_order()
* test_if_cheaper_ordering() has been extracted from
test_if_skip_sort_order() to share with get_index_for_order()
* simple_remove_const()
sql/sql_select.h:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
New functions:
* test_if_cheaper_ordering()
* simple_remove_const()
* get_index_for_order()
sql/sql_update.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
mysql_update: an optimization has been added to skip
unnecessary sorting with ORDER BY clause where a select
result ordering is acceptable.
sql/table.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
New functions:
* TABLE::update_const_key_parts()
* is_simple_order()
sql/table.h:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
New functions:
* TABLE::update_const_key_parts()
* is_simple_order()
use limit efficiently
Bug #36569: UPDATE ... WHERE ... ORDER BY... always does a
filesort even if not required
Also two bugs reported after QA review (before the commit
of bugs above to public trees, no documentation needed):
Bug #53737: Performance regressions after applying patch
for bug 36569
Bug #53742: UPDATEs have no effect after applying patch
for bug 36569
Execution of single-table UPDATE and DELETE statements did not use the
same optimizer as was used in the compilation of SELECT statements.
Instead, it had an optimizer of its own that did not take into account
that you can omit sorting by retrieving rows using an index.
Extra optimization has been added: when applicable, single-table
UPDATE/DELETE statements use an existing index instead of filesort. A
corresponding SELECT query would do the former.
Also handling of the DESC ordering expression has been added when
reverse index scan is applicable.
From now on most single table UPDATE and DELETE statements show the
same disk access patterns as the corresponding SELECT query. We verify
this by comparing the result of SHOW STATUS LIKE 'Sort%
Currently the get_index_for_order function
a) checks quick select index (if any) for compatibility with the
ORDER expression list or
b) chooses the cheapest available compatible index, but only if
the index scan is cheaper than filesort.
Second way is implemented by the new test_if_cheaper_ordering
function (extracted part the test_if_skip_sort_order()).
strict aliasing violations.
One somewhat major source of strict-aliasing violations and
related warnings is the SQL_LIST structure. For example,
consider its member function `link_in_list` which takes
a pointer to pointer of type T (any type) as a pointer to
pointer to unsigned char. Dereferencing this pointer, which
is done to reset the next field, violates strict-aliasing
rules and might cause problems for surrounding code that
uses the next field of the object being added to the list.
The solution is to use templates to parametrize the SQL_LIST
structure in order to deference the pointers with compatible
types. As a side bonus, it becomes possible to remove quite
a few casts related to acessing data members of SQL_LIST.
sql/handler.h:
Use the appropriate template type argument.
sql/item.cc:
Remove now-unnecessary cast.
sql/item_subselect.cc:
Remove now-unnecessary casts.
sql/item_sum.cc:
Use the appropriate template type argument.
Remove now-unnecessary cast.
sql/mysql_priv.h:
Move SQL_LIST structure to sql_list.h
Use the appropriate template type argument.
sql/sp.cc:
Remove now-unnecessary casts.
sql/sql_delete.cc:
Use the appropriate template type argument.
Remove now-unnecessary casts.
sql/sql_derived.cc:
Remove now-unnecessary casts.
sql/sql_lex.cc:
Remove now-unnecessary casts.
sql/sql_lex.h:
SQL_LIST now takes a template type argument which must
match the type of the elements of the list. Use forward
declaration when the type is not available, it is used
in pointers anyway.
sql/sql_list.h:
Rename SQL_LIST to SQL_I_List. The template parameter is
the type of object that is stored in the list.
sql/sql_olap.cc:
Remove now-unnecessary casts.
sql/sql_parse.cc:
Remove now-unnecessary casts.
sql/sql_prepare.cc:
Remove now-unnecessary casts.
sql/sql_select.cc:
Remove now-unnecessary casts.
sql/sql_show.cc:
Remove now-unnecessary casts.
sql/sql_table.cc:
Remove now-unnecessary casts.
sql/sql_trigger.cc:
Remove now-unnecessary casts.
sql/sql_union.cc:
Remove now-unnecessary casts.
sql/sql_update.cc:
Remove now-unnecessary casts.
sql/sql_view.cc:
Remove now-unnecessary casts.
sql/sql_yacc.yy:
Remove now-unnecessary casts.
storage/myisammrg/ha_myisammrg.cc:
Remove now-unnecessary casts.
strict aliasing violations.
One somewhat major source of strict-aliasing violations and
related warnings is the SQL_LIST structure. For example,
consider its member function `link_in_list` which takes
a pointer to pointer of type T (any type) as a pointer to
pointer to unsigned char. Dereferencing this pointer, which
is done to reset the next field, violates strict-aliasing
rules and might cause problems for surrounding code that
uses the next field of the object being added to the list.
The solution is to use templates to parametrize the SQL_LIST
structure in order to deference the pointers with compatible
types. As a side bonus, it becomes possible to remove quite
a few casts related to acessing data members of SQL_LIST.
without FOR UPDATE is causing a lock".
SELECT statements with subqueries referencing InnoDB tables
were acquiring shared locks on rows in these tables when they
were executed in REPEATABLE-READ mode and with statement or
mixed mode binary logging turned on.
This was a regression which were introduced when fixing
bug 39843.
The problem was that for tables belonging to subqueries
parser set TL_READ_DEFAULT as a lock type. In cases when
statement/mixed binary logging at open_tables() time this
type of lock was converted to TL_READ_NO_INSERT lock at
open_tables() time and caused InnoDB engine to acquire
shared locks on reads from these tables. Although in some
cases such behavior was correct (e.g. for subqueries in
DELETE) in case of SELECT it has caused unnecessary locking.
This patch implements minimal version of the fix for the
specific problem described in the bug-report which supposed
to be not too risky for pushing into 5.1 tree.
The 5.5 tree already contains a more appropriate solution
which also addresses other related issues like bug 53921
"Wrong locks for SELECTs used stored functions may lead
to broken SBR".
This patch tries to solve the problem by ensuring that
TL_READ_DEFAULT lock which is set in the parser for
tables participating in subqueries at open_tables()
time is interpreted as TL_READ_NO_INSERT or TL_READ.
TL_READ is used only if we know that this is a SELECT
and that this particular table is not used by a stored
function.
Test coverage is added for both InnoDB and MyISAM.
This patch introduces an "incompatible" change in locking
scheme for subqueries used in SELECT ... FOR UPDATE and
SELECT .. IN SHARE MODE.
In 4.1 (as well as in 5.0 and 5.1 before fix for bug 39843)
the server would use a snapshot InnoDB read for subqueries
in SELECT FOR UPDATE and SELECT .. IN SHARE MODE statements,
regardless of whether the binary log is on or off.
If the user required a different type of read (i.e. locking
read), he/she could request so explicitly by providing FOR
UPDATE/IN SHARE MODE clause for each individual subquery.
The patch for bug 39843 broke this behaviour (which was not
documented or tested), and started to use locking reads for
all subqueries in SELECT ... FOR UPDATE/IN SHARE MODE.
This patch restores 4.1 behaviour.
This patch should be mostly null-merged into 5.5 tree.
mysql-test/include/check_concurrent_insert.inc:
Added auxiliary script which allows to check if statement
reading table allows concurrent inserts in it.
mysql-test/include/check_no_concurrent_insert.inc:
Added auxiliary script which allows to check that statement
reading table doesn't allow concurrent inserts in it.
mysql-test/include/check_no_row_lock.inc:
Added auxiliary script which allows to check if statement
reading table doesn't take locks on its rows.
mysql-test/include/check_shared_row_lock.inc:
Added auxiliary script which allows to check if statement
reading table takes shared locks on some of its rows.
mysql-test/r/bug39022.result:
After bug #46947 'Embedded SELECT without FOR UPDATE is
causing a lock' was fixed test case for bug 39022 has to
be adjusted in order to trigger execution path on which
original problem was encountered.
mysql-test/r/innodb_mysql_lock2.result:
Added coverage for handling of locking in various cases when
we read data from InnoDB tables (includes test case for
bug #46947 'Embedded SELECT without FOR UPDATE is causing a
lock').
mysql-test/r/lock_sync.result:
Added coverage for handling of locking in various cases when
we read data from MyISAM tables.
mysql-test/t/bug39022.test:
After bug #46947 'Embedded SELECT without FOR UPDATE is
causing a lock' was fixed test case for bug 39022 has to
be adjusted in order to trigger execution path on which
original problem was encountered.
mysql-test/t/innodb_mysql_lock2.test:
Added coverage for handling of locking in various cases when
we read data from InnoDB tables (includes test case for
bug #46947 'Embedded SELECT without FOR UPDATE is causing a
lock').
mysql-test/t/lock_sync.test:
Added coverage for handling of locking in various cases when
we read data from MyISAM tables.
sql/mysql_priv.h:
Function read_lock_type_for_table() now takes pointers to
LEX and TABLE_LIST elements as its arguments since to
correctly determine lock type it needs to know what
statement is being performed and whether table element for
which lock type to be determined belongs to prelocking list.
sql/sql_base.cc:
Changed read_lock_type_for_table() to return a weak TL_READ
type of lock in cases when we are executing SELECT (and so
won't update tables directly) and table doesn't belong to
statement's prelocking list and thus can't be used by a
stored function. It is OK to do so since in this case table
won't be used by statement or function call which will be
written to the binary log, so serializability requirements
for it can be relaxed.
One of results from this change is that SELECTs on InnoDB
tables no longer takes shared row locks for tables which
are used in subqueries (i.e. bug #46947 is fixed).
Another result is that for similar SELECTs on MyISAM tables
concurrent inserts are allowed.
In order to implement this change signature of
read_lock_type_for_table() function was changed to
take pointers to LEX and TABLE_LIST objects.
sql/sql_update.cc:
Function read_lock_type_for_table() now takes pointers to
LEX and TABLE_LIST elements as its arguments since to
correctly determine lock type it needs to know what
statement is being performed and whether table element for
which lock type to be determined belongs to prelocking list.
without FOR UPDATE is causing a lock".
SELECT statements with subqueries referencing InnoDB tables
were acquiring shared locks on rows in these tables when they
were executed in REPEATABLE-READ mode and with statement or
mixed mode binary logging turned on.
This was a regression which were introduced when fixing
bug 39843.
The problem was that for tables belonging to subqueries
parser set TL_READ_DEFAULT as a lock type. In cases when
statement/mixed binary logging at open_tables() time this
type of lock was converted to TL_READ_NO_INSERT lock at
open_tables() time and caused InnoDB engine to acquire
shared locks on reads from these tables. Although in some
cases such behavior was correct (e.g. for subqueries in
DELETE) in case of SELECT it has caused unnecessary locking.
This patch implements minimal version of the fix for the
specific problem described in the bug-report which supposed
to be not too risky for pushing into 5.1 tree.
The 5.5 tree already contains a more appropriate solution
which also addresses other related issues like bug 53921
"Wrong locks for SELECTs used stored functions may lead
to broken SBR".
This patch tries to solve the problem by ensuring that
TL_READ_DEFAULT lock which is set in the parser for
tables participating in subqueries at open_tables()
time is interpreted as TL_READ_NO_INSERT or TL_READ.
TL_READ is used only if we know that this is a SELECT
and that this particular table is not used by a stored
function.
Test coverage is added for both InnoDB and MyISAM.
This patch introduces an "incompatible" change in locking
scheme for subqueries used in SELECT ... FOR UPDATE and
SELECT .. IN SHARE MODE.
In 4.1 (as well as in 5.0 and 5.1 before fix for bug 39843)
the server would use a snapshot InnoDB read for subqueries
in SELECT FOR UPDATE and SELECT .. IN SHARE MODE statements,
regardless of whether the binary log is on or off.
If the user required a different type of read (i.e. locking
read), he/she could request so explicitly by providing FOR
UPDATE/IN SHARE MODE clause for each individual subquery.
The patch for bug 39843 broke this behaviour (which was not
documented or tested), and started to use locking reads for
all subqueries in SELECT ... FOR UPDATE/IN SHARE MODE.
This patch restores 4.1 behaviour.
This patch should be mostly null-merged into 5.5 tree.
