Problem #1: INSERT...SELECT
INSERT ... SELECT with the same table on both sides (hidden
below a MERGE table) does now work by buffering the select result.
The duplicate detection works now after open_and_lock_tables()
on the locks.
I did not find a test case that failed without the change in
sql_update.cc. I made the change anyway as it should in theory
fix a possible MERGE table problem with multi-table update.
mysql-test/r/create.result:
BUG#5390 - problems with merge tables
Removed a duplicate test.
mysql-test/r/merge.result:
BUG#5390 - problems with merge tables
Problem #1: INSERT...SELECT
Added test results.
mysql-test/t/create.test:
BUG#5390 - problems with merge tables
Removed a duplicate test.
mysql-test/t/merge.test:
BUG#5390 - problems with merge tables
Problem #1: INSERT...SELECT
Added tests.
sql/lock.cc:
BUG#5390 - problems with merge tables
Problem #1: INSERT...SELECT
Added a new function to find a duplicate lock in a list of tables.
sql/mysql_priv.h:
BUG#5390 - problems with merge tables
Problem #1: INSERT...SELECT
Added a declaration for the new function.
sql/sql_parse.cc:
BUG#5390 - problems with merge tables
Problem #1: INSERT...SELECT
Changed the duplicate tables detection for INSERT ... SELECT
to use the new function, which does also work for MERGE tables.
sql/sql_update.cc:
BUG#5390 - problems with merge tables
Changed the duplicate tables detection for UPDATE
to use the new function, which does also work for MERGE tables.
1.) Added a new option to mysql_lock_tables() for ignoring FLUSH TABLES.
Used the new option in create_table_from_items().
It is necessary to prevent the SELECT table from being reopend.
It would get new storage assigned for its fields, while the
SELECT part of the command would still use the old (freed) storage.
2.) Protected the CREATE TABLE and CREATE TABLE ... SELECT commands
against a global read lock. This prevents a deadlock in
CREATE TABLE ... SELECT in conjunction with FLUSH TABLES WITH READ LOCK
and avoids the creation of new tables during a global read lock.
3.) Replaced set_protect_against_global_read_lock() and
unset_protect_against_global_read_lock() by
wait_if_global_read_lock() and start_waiting_global_read_lock()
in the INSERT DELAYED handling.
mysql-test/r/create.result:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Added test results.
mysql-test/t/create.test:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Added tests which do not require concurrency.
sql/lock.cc:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Added a new option to mysql_lock_tables() for ignoring FLUSH TABLES.
Changed the parameter list.
Removed two unnecessary functions. Their functionality is included in
wait_if_global_read_lock() and start_waiting_global_read_lock().
sql/mysql_priv.h:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Changed the declaration of mysql_lock_tables().
Added definitions for the new options.
sql/sql_acl.cc:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Adjusted mysql_lock_tables() calls to the new argument list.
sql/sql_base.cc:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Adjusted mysql_lock_tables() calls to the new argument list.
sql/sql_handler.cc:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Adjusted mysql_lock_tables() calls to the new argument list.
sql/sql_insert.cc:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Replaced set_protect_against_global_read_lock() and
unset_protect_against_global_read_lock() by
wait_if_global_read_lock() and start_waiting_global_read_lock()
in the INSERT DELAYED handling.
Adjusted mysql_lock_tables() calls to the new argument list.
sql/sql_parse.cc:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Protected the CREATE TABLE and CREATE TABLE ... SELECT commands
against a global read lock. This prevents a deadlock in
CREATE TABLE ... SELECT in conjunction with FLUSH TABLES WITH READ LOCK
and avoids the creation of new tables during a global read lock.
sql/sql_table.cc:
Bug#10224 - ANALYZE TABLE crashing with simultaneous CREATE ... SELECT statement.
Adjusted mysql_lock_tables() calls to the new argument list.
Used the new option in create_table_from_items().
This fixes also the reverse lookup bug introduced by the previous patch
mysql-test/t/group_by.test:
Remove empty line
vio/viosocket.c:
Added function comment
Fix replication for multi-update
new test - rpl_multi_update2
sql/mysql_priv.h:
Bug#7011
New function mysql_multi_update_lock()
sql/sql_parse.cc:
Bug#7011
New function check_multi_update_lock()
For multi-update on slave, perform an early open&lock
sql/sql_update.cc:
Bug#7011
Split out multi-update locking into its own function,
mysql_multi_update_lock()
always initialized. (Bug #5569)
vio/viosocket.c:
Remove comment that is no longer correct
sql/sql_parse.cc:
Always call vio_in_addr() on successful connection,
so that thd->remote always gets set
vio/viossl.c:
Remove comment that is no longer correct
case will be added to client_test in 4.1
sql/sql_parse.cc:
A fix for bug #6761: "mysql_list_fields() does not work": init
thd->query_length, which is used next line. Remove a stale comment.
crashes server"
(in 4.0 we fix only connection stalling in case of error, crash itself is
fixed in 4.1, the test case for this code is also there).
sql/sql_parse.cc:
Handling of COM_CREATE_DB, COM_DROP_DB:
We should call send_error() if mysql_create_db or mysql_drop_db
return error (like we do it for SQL versions of these commands).
Ensures that WRITE lock is not obtained on all tables referenced.
mysql-test/r/lock_multi.result:
Bug#4118
New test for multi-update locking
mysql-test/r/multi_update.result:
Bug#4118
Fix test
mysql-test/t/lock_multi.test:
Bug#4118
New test for multi-update locking
mysql-test/t/multi_update.test:
Bug#4118
Fix test
sql/sql_parse.cc:
Bug#4118
Split multi-update to its own case statement in sql_parse.cc
sql/sql_update.cc:
Bug#4118
Overview of locking checking:
1. Open and acquire READ lock
2. Check to see which tables need WRITE lock
3. Unlock tables and relock
sql/sql_yacc.yy:
Bug#4118
Split multi-update to its own case statement in sql_parse.cc
using GROUP BY"
Now we are setting Field_timestamp::field_length to 19 in open_table()
if we are in new mode (and we are restoring it back when we are coming
back to normal mode). This also should solve potential problems with
some of LOAD DATA INFILE and SELECT * INTO in this mode.
mysql-test/r/type_timestamp.result:
Added test for bug #4131 'TIMESTAMP columns missing minutes and seconds
when using GROUP BY' and other --new mode related behavior.
mysql-test/t/type_timestamp.test:
Added test for bug #4131 'TIMESTAMP columns missing minutes and seconds
when using GROUP BY' and other --new mode related behavior.
sql/field.cc:
Added Field_timestamp::orig_field_length member for saving original
field_length value, because this member can be modified if new_mode is
in effect.
Lot of Field_timestamp code simplified and Field_timestamp::make_field()
is no longer needed because we are setting field_length to 19 if we are
in --new mode now.
sql/field.h:
Added Field_timestamp::orig_field_length member for saving original
field_length value, because this member can be modified if new_mode
is in effect.
Field_timestamp::make_field() is no longer needed because we are setting
field_length to 19 if we are in --new mode now.
sql/sql_base.cc:
If --new mode is in effect all TIMESTAMP fields should pretend that they
have length of 19. We are achieving this by setting
Field_timestamp::field_length to 19 (or original value) in open_table().
We are using TABLE::timestamp_mode variable for avoiding of unnecessary
looping over all fields of the table and setting field_length if table
was used with same new_mode value before.
Note: We do not introduce general framework for setting up Field objects
for usage with current THD here because this fix is only needed in 4.0
and Monty said that we will also remove looping over all fields when
updating table_name member at some point. This more general framework
will also complicate nice optimization with avoiding of unneeded looping.
sql/sql_parse.cc:
Now when we are creating TIMESTAMP(19) fields by default in --new mode,
otherwise we will have unaligned behavior between ALTER and CREATE.
sql/table.h:
Added TABLE::timestamp_mode field for saving information whenever we set
field_length members of table's TIMESTAMP fields to 19 (to honor
new_mode) or they have original values.
Update to new valgrind
mysql-test/mysql-test-run.sh:
Update for new valgrind
sql/mysql_priv.h:
Remove default argument to mysql_truncate()
sql/sql_parse.cc:
Remove default argument to mysql_truncate()
in a deadlock-free manner. This splits locking the global read lock in two steps.
This fixes a consequence of this bug, known as:
BUG#4953 'mysqldump --master-data may report incorrect binlog position if using InnoDB'
And a test.
sql/handler.cc:
making COMMIT wait if FLUSH TABLES WITH READ LOCK happened.
sql/lock.cc:
an additional stage so that FLUSH TABLES WITH READ LOCK blocks COMMIT:
make_global_read_lock_block_commit():
taking the global read lock is TWO steps (2nd step is optional; without
it, COMMIT of existing transactions will be allowed):
lock_global_read_lock() THEN make_global_read_lock_block_commit().
sql/mysql_priv.h:
new argument to wait_if_global_read_lock()
sql/sql_class.h:
THD::global_read_lock now an uint to reflect the 2 steps of global read lock (does not block COMMIT / does)
sql/sql_db.cc:
update for new prototype
sql/sql_parse.cc:
implementing the two steps of global read lock so that FLUSH TABLES WITH READ LOCK can block COMMIT without deadlocking with COMMITs.
replacing the no_log argument of mysql_create_table() by some safer method
(temporarily setting OPTION_BIN_LOG to 0) which guarantees that even the automatic
DELETE FROM heap_table does not get into the binlog when a not-yet-existing HEAP table
is opened by mysql_create_table().
mysql-test/r/rpl_heap.result:
result update
mysql-test/t/rpl_heap.test:
testing a bug
sql/log.cc:
new class Disable_binlog used to temporarily disable binlogging for one thread.
sql/mysql_priv.h:
removing argument no_log from mysql_create_table(); no_log was perfect as some
binlogging could still be done by open_unireg_entry() for a HEAP table.
sql/sql_class.h:
new class Disable_binlog used to temporarily disable binlogging for one thread.
sql/sql_parse.cc:
removing no_log
sql/sql_table.cc:
removing no_log from mysql_create_table(); instead using new class Disable_binlog.
Disabling binlogging in some cases, where the binlogging is done later by some other code
(case of CREATE SELECT and ALTER).
We should allow 19 as length of newly created TIMESTAMP fields.
mysql-test/r/type_timestamp.result:
Added test of TIMESTAMP(19) support.
mysql-test/t/type_timestamp.test:
Added test of TIMESTAMP(19) support.
sql/sql_parse.cc:
add_field_to_list(): TIMESTAMP columns should also support 19 as length since it is
length of 4.1 compatible representation.
BitKeeper/etc/logging_ok:
Logging to logging@openlogging.org accepted
will disable warnings, --log-warnings will increment warning level by one, or the
level can be given as an optional argument. Default level is 1.
Changed aborted connection warning to be logged only if the level is > 1.
sql/sql_class.h:
Changed boolean into ulong.
sql/sql_parse.cc:
Changed aborted connection warning to be logged only if the level is > 1.
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
sql/sql_class.h:
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
sql/ha_innodb.cc:
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
sql/log_event.cc:
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
sql/slave.cc:
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
sql/sql_db.cc:
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
sql/sql_parse.cc:
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
sql/sql_show.cc:
Fix remaining cases of Bug #3596: fix possible races caused by an obsolete value of thd->query_length in SHOW PROCESSLIST and SHOW INNODB STATUS; this fix depends on the fact that thd->query is always set to NULL before setting it to point to a new query
BitKeeper/etc/logging_ok:
auto-union
scripts/mysqlbug.sh:
Auto merged
sql/sql_parse.cc:
Auto merged
mysql-test/t/alter_table.test:
Merge code with 3.23
sql/nt_servc.cc:
use original code
mysql-test/r/func_math.result:
Updated tests after merge with 3.23
mysql-test/r/multi_update.result:
Fixed not portable test
mysql-test/r/mysqlbinlog.result:
Fixed test after push without doing 'make test'
mysql-test/t/multi_update.test:
Fixed not portable test
sql/sql_parse.cc:
Don't use static inline, as this is not portable
We introduce a new function mysql_test_parse_for_slave().
If the slave sees that the query got a really bad error on master
(killed e.g.), then it calls this function to know if this query
can be ignored because of replicate-*-table rules (do not worry
about replicate-*-db rules: they are checked so early that they have
no bug). If the answer is yes, it skips the query and continues. If
it's no, then it stops and say "fix your slave data manually" (like it
did before this change).
mysql-test/r/rpl_error_ignored_table.result:
result update
mysql-test/t/rpl_error_ignored_table-slave.opt:
ignore more tables
mysql-test/t/rpl_error_ignored_table.test:
we test if a killed query on the master, is ignored on the slave
if the tables it involves should be excluded because of replicate-*-table
rules.
sql/log_event.cc:
If the query got a really bad error on the master (thread killed etc),
parse it to test the table names: if the replicate-*-do|ignore-table
rules say "this query must be ignored" then we exit gracefully;
otherwise we warn about the bad error and tell DBA to check/fix it.
Before this change, we always warned and stopped.
sql/mysql_priv.h:
new function
sql/slave.cc:
don't print error immediately as we need to do one more test to
be sure.
sql/sql_parse.cc:
we add a function mysql_test_parse_for_slave(), to be used only
by the slave if it wants to know if the query should be ignored or not;
so this function only parses the query, does not execute it.
re-using unused LOCK_active_mi to serialize all administrative
commands related to replication:
START SLAVE, STOP SLAVE, RESET SLAVE, CHANGE MASTER, init_slave()
(replication autostart at server startup), end_slave() (replication
autostop at server shutdown), LOAD DATA FROM MASTER.
This protects us against a handful of deadlocks (like BUG#2921
when two START SLAVE, but when two STOP SLAVE too).
Removing unused variables.
sql/item_func.cc:
We don't need LOCK_active_mi just to MASTER_POS_WAIT().
sql/repl_failsafe.cc:
no need for macro
sql/set_var.cc:
no need for macro
sql/slave.cc:
Re-using unused LOCK_active_mi to serialize all administrative
commands related to replication:
START SLAVE, STOP SLAVE, RESET SLAVE, CHANGE MASTER, init_slave()
(replication autostart at server startup), end_slave() (replication
autostop at server shutdown), LOAD DATA FROM MASTER.
This protects us against a handful of deadlocks.
Removing unused variables.
sql/slave.h:
Re-using LOCK_active_mi to serialize administrative replication commands.
Macros unneeded. Removing unneeded variables.
sql/sql_parse.cc:
found unused variable.
Replacing macros.
sql/sql_show.cc:
replacing macros
"Table truncated when creating another table name with Spaces"
added to check_db_name, check_table_name and check_column_name
test for end space
mysql-test/r/create.result:
added test for Bug #2985
"Table truncated when creating another table name with Spaces"
mysql-test/t/create.test:
added test for Bug #2985
"Table truncated when creating another table name with Spaces"
sql/sql_db.cc:
cancel strip end spaces for database name
sql/sql_parse.cc:
cancel strip end spaces for database name
sql/table.cc:
added to check_db_name, check_table_name and check_column_name
test for end space
fixed linking query_prealloc_size to query cache presence
mysql-test/r/query_cache.result:
test of QC invalidation by LOCK command
mysql-test/t/query_cache.test:
test of QC invalidation by LOCK command
sql/mysqld.cc:
new variable query_cache_wlock_invalidate
fixed query_prealloc_size with QC absence
sql/set_var.cc:
new variable query_cache_wlock_invalidate
sql/sql_cache.cc:
new method for table invalidation
sql/sql_cache.h:
new method for table invalidation
sql/sql_class.h:
new variable query_cache_wlock_invalidate
sql/sql_parse.cc:
layout fixed
Now one need only SELECT privilege for tables that are only read in
UPDATE statements with many tables. (Bug #2377).
sql/sql_acl.cc:
Comment cleanup
sql/sql_parse.cc:
Merged duplicate code.
Removed some outdated 'tables->db' tests.
Check privileges for multi-updates properly (Bug #2377)
sql/sql_show.cc:
Remove disabled code
sql/sql_update.cc:
Ensure that privileges are tested properly for multi-table-updates
tests/grant.pl:
Added more tests
tests/grant.res:
updated results
values for TIMESTAMP columns. The solution is not perfect since
we just silently ignoring default value for first TIMESTAMP
column and properly reflecting this fact in SHOW CREATE TABLE.
We can't give a warning or simply support standard syntax
(niladic functions as legal value for default) for first field
since it is 4.0 tree.
mysql-test/r/type_timestamp.result:
Added test for bugs #1885, #2464, #2539
(proper support of default values for TIMESTAMP columns)
mysql-test/t/type_timestamp.test:
Added test for bugs #1885, #2464, #2539
(proper support of default values for TIMESTAMP columns)
sql/field.cc:
Enabled copying of defaults for TIMESTAMP fields when we are
creating table with CREATE TABLE x (SELECT ...)
sql/field.h:
Set proper DEFAULT value for non-first TIMESTAMP column.
sql/sql_parse.cc:
Allowed default values for TIMESTAMP column.
sql/sql_show.cc:
Enabled printing of default values in SHOW CREATE TABLE and
SHOW COLUMNS for all TIMESTAMP columns except first one.
(as long as replicate-*-table rules were defined, multi-table DELETE was never
replicated by the slave).
sql/sql_parse.cc:
For a multi-table DELETE, there are two lists of tables:
- 'tables' contains the tables in the FROM. They have 'updating==0'.
- 'thd->lex.auxilliary_table_list.first' contains the tables between the DELETE and the FROM. They have 'updating==1'.
So the bug was that the slave only tests 'tables' and as all its elements have updating==0,
tables_ok() always returns 0.
So for a multi DELETE the slave now tests the 2nd list too.
I started with the other tip of the sausage: I thought of changing multi DELETE to use
only one list (given that 'updating' can be used to separate it in two when needed)
(with one list we wouldn't need to change the slave code), but finally hit the unicity
check in add_table_to_list() which started to return ER_NONUNIQ_TABLE (logical).
