The problem was that maybe_null of Item_row and its componetes was unsynced after update_used_tables() (and so pushed_cond_guards was not initialized but then requested).
Fix updates Item_row::maybe_null on update_used_tables().
Analysis
The reason for the less efficient plan was result of a prior design decision -
to limit the eveluation of constant expressions during optimization to only
non-expensive ones. With this approach all stored procedures were considered
expensive, and were not evaluated during optimization. As a result, SPs didn't
participate in range optimization, which resulted in a plan with table scan
rather than index range scan.
Solution
Instead of considering all SPs expensive, consider expensive only those SPs
that are non-deterministic. If an SP is deterministic, the optimizer will
checj if it is constant, and may eventually evaluate it during optimization.
The original patch with the implementation of virtual columns
did not support INSERT DELAYED into tables with virtual columns.
This patch fixes the problem.
(because it's conceptually wrong. only the user can decide whether the kill is
allowed to leave tables in the inconsistent state, storage engine has no say in that)
The previous fix for MDEV-3992 was incomplete, because it still computed
incorrectly the number of keyparts of the extended secondary key in the
case when columns of the PK participate in the secondary key.
This patch by Monty corrects the above problem.
Analysis:
The crash is a result of incorrect analysis of whether a secondary key
can be extended with a primary in order to compute ORDER BY. The analysis
is done in test_if_order_by_key(). This function doesn't take into account
that the primary key may in fact index the same columns as the secondary
key. For the test query test_if_order_by_key says that there is an extended
key with total 2 keyparts.
At the same time, the condition
if (pkinfo->key_part[i].field->key_start.is_set(nr))
in test_if_cheaper_oredring() becomes true for (i == 0), which results in
an invalid access to rec_per_key[-1].
Solution:
The best solution would be to reuse KEY::ext_key_parts that is already computed
by open_binary_frm(), however after detailed analysis the conclusion is that
the change would be too intrusive for a GA release.
The solution for 5.5 is to add a guard for the case when the 0-th key part is
considered, and to assume that all keys will be scanned in this case.
The bug could cause a crash when several connections needed
persistent statistics for the same table.
Also added a missing call of set_statistics_for_table() in the code
of the function mysql_update.
The bug could lead to a wrong estimate of the number of expected rows
in the output of the EXPLAIN commands for queries with GROUP BY.
This could be observed in the test case for LP bug 934348.
- Added --verbose to BUILD scripts to get make to write out compile commands.
- Detect if AM_EXTRA_MAKEFLAGS=VERBOSE=1 was used with build scripts.
- Don't write warnings about replication variables when doing bootstrap.
- Fixed that mysql_cond_wait() and mysql_cond_timedwait() will report original source file in case of errors.
- Ignore some compiler warnings
BUILD/FINISH.sh:
Detect if AM_EXTRA_MAKEFLAGS=VERBOSE=1 or --verbose was used
BUILD/SETUP.sh:
Added --verbose to print out the full compile lines
Updated help message
client/mysqltest.cc:
Fixed that one can use 'replace' with cat_file
cmake/configure.pl:
If --verbose is used, get make to write out compile commands
debian/dist/Debian/rules:
Added $AM_EXTRA_MAKEFLAGS to get VERBOSE=1 on buildbot builds
debian/dist/Ubuntu/rules:
Added $AM_EXTRA_MAKEFLAGS to get VERBOSE=1 on buildbot builds
include/my_pthread.h:
Made set_timespec_time_nsec() more portable.
include/mysql/psi/mysql_thread.h:
Fixed that mysql_cond_wait() and mysql_cond_timedwait() will report original source file in case of errors.
mysql-test/suite/innodb/r/auto_increment_dup.result:
Fixed wrong DBUG_SYNC
mysql-test/suite/innodb/t/auto_increment_dup.test:
Fixed wrong DBUG_SYNC
mysql-test/suite/perfschema/include/upgrade_check.inc:
Make test more portable for changes in *.sql files
mysql-test/suite/perfschema/r/pfs_upgrade.result:
Updated test results
mysql-test/valgrind.supp:
Ignore running Aria checkpoint thread
scripts/mysqlaccess.sh:
Changed reference of bugs database
Ensure that also client-server group is read.
sql/handler.cc:
Added missing syncpoint
sql/mysqld.cc:
Don't write warnings about replication variables when doing bootstrap
sql/mysqld.h:
Don't write warnings about replication variables when doing bootstrap
sql/rpl_rli.cc:
Don't write warnings about replication variables when doing bootstrap
sql/sql_insert.cc:
Don't mask SERVER_SHUTDOWN in insert_delayed
This is done to be able to distingush between shutdown and interrupt errors
support-files/compiler_warnings.supp:
Ignore some compiler warnings in xtradb,innobase, oqgraph, yassl, string3.h
Fixed MDEV-4002: Server crash or valgrind errors in Item_func_group_concat::setup and Item_func_group_concat::add
mysql-test/r/group_by.result:
Added test case for failing GROUP_CONCAT ... ROLLUP queries
mysql-test/t/group_by.test:
Added test case for failing GROUP_CONCAT ... ROLLUP queries
sql/item_sum.cc:
Fixed issue where field->table pointed to different temporary table than expected.
Ensure that order->next points to the right object (could cause problems with setup_order())
Give error for wrong parameters to CHANGE MASTER
Extend MASTER_PASSWORD and MASTER_HOST lengths
mysql-test/suite/rpl/r/rpl_password_boundaries.result:
Test length of MASTER_PASSWORD, MASTER_HOST and MASTER_USER
mysql-test/suite/rpl/r/rpl_semi_sync.result:
Use different password than user name for better test coverage
mysql-test/suite/rpl/t/rpl_password_boundaries.test:
Test length of MASTER_PASSWORD, MASTER_HOST and MASTER_USER
mysql-test/suite/rpl/t/rpl_semi_sync.test:
Use different password than user name for better test coverage
sql/rpl_mi.h:
Extend MASTER_PASSWORD and MASTER_HOST lengths
sql/sql_repl.cc:
Give error for wrong parameters to CHANGE MASTER
sql/sql_repl.h:
Extend MASTER_PASSWORD and MASTER_HOST lengths
KILL now breaks locks inside InnoDB
Fixed possible deadlock when running INNODB STATUS
Added ha_kill_query() and kill_query() to send kill signal to all storage engines
Added reset_killed() to ensure we don't reset killed state while awake() is getting called
include/mysql/plugin.h:
Added thd_mark_as_hard_kill()
include/mysql/plugin_audit.h.pp:
Added thd_mark_as_hard_kill()
include/mysql/plugin_auth.h.pp:
Added thd_mark_as_hard_kill()
include/mysql/plugin_ftparser.h.pp:
Added thd_mark_as_hard_kill()
sql/handler.cc:
Added ha_kill_query() to send kill signal to all storage engines
sql/handler.h:
Added ha_kill_query() and kill_query() to send kill signal to all storage engines
sql/log_event.cc:
Use reset_killed()
sql/mdl.cc:
use thd->killed instead of thd_killed() to abort on soft kill
sql/sp_rcontext.cc:
Use reset_killed()
sql/sql_class.cc:
Fixed possible deadlock in INNODB STATUS by not getting thd->LOCK_thd_data if it's locked.
Use reset_killed()
Tell storge engines that KILL has been sent
sql/sql_class.h:
Added reset_killed() to ensure we don't reset killed state while awake() is getting called.
Added mark_as_hard_kill()
sql/sql_insert.cc:
Use reset_killed()
sql/sql_parse.cc:
Simplify detection of killed queries.
Use reset_killed()
sql/sql_select.cc:
Use reset_killed()
sql/sql_union.cc:
Use reset_killed()
storage/innobase/handler/ha_innodb.cc:
Added innobase_kill_query()
Fixed error reporting for interrupted queries.
storage/xtradb/handler/ha_innodb.cc:
Added innobase_kill_query()
Fixed error reporting for interrupted queries.
- Fixed broadcast without a proper mutex
- Don't break existing locks if we are just testing if we can get the lock
mysql-test/r/create_delayed.result:
Added test case for failures with INSERT DELAYED with CREATE and DROP TABLE
mysql-test/t/create_delayed.test:
Added test case for failures with INSERT DELAYED with CREATE and DROP TABLE
sql/mdl.cc:
Don't break existing locks for timeout=0 (ie, just check if there are conflicting locks).
This fixed the bug that INSERT DELAYED didn't work properly with CREATE TABLE
sql/sql_base.cc:
One neads to hold the mutex before doing a mysql_cond_broadcast()
This fixed the bug that INSERT DELAYED didn't work properly with DROP TABLE
sql/sql_insert.cc:
Protect setting of mysys_var->current_mutex.
Backport the fix from 10.0 tree
- The problem was that thd_killed() may be called by innodb from
an internal innodb thread.
- Fixed by not processing APC requests when we're not in the thread
that owns the APC target.
Made allocation of memory for statistical data in a table share to be thread safe.
This memory is now allocated in a special MEM_ROOT that is created for each
table share.
from a MERGE view.
The problem was in the lost ability to be null for the table of a left join if it
is a view/derived table.
It hapenned because setup_table_map(), was called earlier then we merged
the view or derived.
Fixed by propagating new maybe_null flag during Item::update_used_tables().
Change in join_outer.test and join_outer_jcl6.test appeared because
IS NULL reported no used tables (i.e. constant) for argument which could not be
NULL and new maybe_null flag was propagated for IS NULL argument (Item_field)
because table the Item_field belonged to changed its maybe_null status.
The patch lifts the limitation of the current implementation
of ALTER TABLE that does not allow to build unique/primary
indexes by sort for MyISAM and Aria engines.
Analysis:
The following call stack shows that it is possible to set Item_cache::value_cached, and the relevant value
without setting Item_cache::example.
#0 Item_cache_temporal::store_packed at item.cc:8395
#1 get_datetime_value at item_cmpfunc.cc:915
#2 resolve_const_item at item.cc:7987
#3 propagate_cond_constants at sql_select.cc:12264
#4 propagate_cond_constants at sql_select.cc:12227
#5 optimize_cond at sql_select.cc:13026
#6 JOIN::optimize at sql_select.cc:1016
#7 st_select_lex::optimize_unflattened_subqueries at sql_lex.cc:3161
#8 JOIN::optimize_unflattened_subqueries at opt_subselect.cc:4880
#9 JOIN::optimize at sql_select.cc:1554
The fix is to set Item_cache_temporal::example even when the value is
set directly by Item_cache_temporal::store_packed. This makes the
Item_cache_temporal object consistent.
have_openssl variable was ON even when OpenSSL was not used (but YaSSL was).
fix that, so that have_openssl really corresponds to OpenSSL
rename not_openssl.inc to not_ssl.inc and fix the test accordingly.
backport improved bootstrap error handling from 5.6
Was:
revno: 3768.1.1
committer: Christopher Powers <chris.powers@oracle.com>
timestamp: Wed 2012-05-02 22:16:40 -0500
message:
Bug#11766342 INITIAL DB CREATION FAILS ON WINDOWS WITH AN ASSERT IN SQL_ERROR.CC
Improved bootstrap error handling:
- Detect and report file i/o errors
- Report query size errors with nearest query text
This was failing not only for P_S, but for any engine that had
HA_PRIMARY_KEY_REQUIRED_FOR_DELETE flag set (in the tree - only P_S and federated).
Because of this flag, read_set and write_set were (possibly) changed
on update. But later the code modified these bitmaps and restored them to the default
state, losing HA_PRIMARY_KEY_REQUIRED_FOR_DELETE related changes.
sql/handler.cc:
small optimization.
don't change the *write* set only because all columns has to be *read*
revno: 3383
revision-id: georgi.kodinov@oracle.com-20110818083108-qa3h3ufqu4zne80a
committer: Georgi Kodinov <Georgi.Kodinov@Oracle.com>
timestamp: Thu 2011-08-18 11:31:08 +0300
message:
Bug #11766001: 59026: ALLOW MULTIPLE --PLUGIN-LOAD OPTIONS
Implemented support for a new command line option :
--plugin-load-add=<comma-separated-name-equals-value-list>
This option takes the same type of arguments that --plugin-load does
and complements --plugin-load (that continues to operate as before) by
appending its argument to the list specified by --plugin-load.
So --plugin-load can be considered a composite option consisting of
resetting the plugin load list and then calling --plugin-load-add to process
the argument.
Note that the order in which you specify --plugin-load and --plugin-load-add
is important : "--plugin-load=x --plugin-load-add=y" will be equivalent to
"--plugin-load=x,y" whereas "--plugin-load-add=y --plugin-load=x" will be
equivalent to "plugin-load=x".
Incompatible change : the --help --verbose command will no longer print the
--plugin-load variable's values (as it doesn't have one). Otherwise both --plugin-load
and --plugin-load-add are mentioned in it.
Make the commit checkpoint inside InnoDB be asynchroneous.
Implement a background thread in binlog to do the writing and flushing of
binlog checkpoint events to disk.
that introduced engine independent persistent statistics.
In particular:
- added an enumeration type for possible values of the system
variable use_stat_tables
- renamed KEY::real_rec_per_key to KEY::actual_rec_per_key
- optimized the collection of statistical data for any primary
key defined only on one column.
If a query referenced some system statistical tables, but not all of them,
then executing an ANALYZE command simultaneously with this query could
lead to a deadlock.
The fix prohibited reading statistics from system statistical tables
for such queries.
Removed the function unlock_tables_n_open_system_tables_for_write()
as not used anymore.
Performed some minor refactoring of the code in sql_statistics.cc.
Using too long table aliases in stored routines might
have caused server crashes.
Code in sp_head::merge_table_list() which is responsible
for collecting information about tables used in stored
routine was not aware of the fact that table alias might
have arbitrary length. I.e. it assumed that table alias
can't be longer than NAME_LEN bytes and allocated buffer
for a key identifying table accordingly.
This patch fixes the issue by ensuring that we use
dynamically allocated buffer for table key when table
alias is too long. By default stack based buffer is used
in which NAME_LEN bytes are reserved for table alias.
engine-independent statistics.
If a table was created for InnoDB then the execution of the
ANALYZE command over this table blocked any INSERT/DELETE/UPDATE
of the table.
engine-independent statistics.
When the primary key was dropped or changed statistics on secondary
indexes for the prefixes that included components of the primary
key was not removed from the table mysql.index_stats.
Also fixed: in the some cases when a column was changed statistics
on the indexes that included this column was not removed from the
table mysql.index_stats.
Also disabled the test mdev-504 for --ps-protocol.
taking a change done to main 5.1 by Dmitri Lenev.
This is the original comment:
> committer: Dmitry Lenev <Dmitry.Lenev@oracle.com>
> branch nick: mysql-5.1-15954896
> timestamp: Wed 2012-12-05 19:26:56 +0400
> message:
> Bug #15954896 "SP, MULTI-TABLE DELETE AND LONG ALIAS".
Using too long table aliases in stored routines might
have caused server crashes.
Code in sp_head::merge_table_list() which is responsible
for collecting information about tables used in stored
routine was not aware of the fact that table alias might
have arbitrary length. I.e. it assumed that table alias
can't be longer than NAME_LEN bytes and allocated buffer
for a key identifying table accordingly.
This patch fixes the issue by ensuring that we use
dynamically allocated buffer for table key when table
alias is too long. By default stack based buffer is used
in which NAME_LEN bytes are reserved for table alias.
The patch decreases the duration of LOCK_thread_count, so it is not hold during THD destructor and freeing memory.
This mutex now only protects the integrity of threads list, when removing THD from it, and thread_count variable.
The add_to_status() function that updates global status during client disconnect, is now correctly protected by the LOCK_status mutex.
Benchmark : in a "non-persistent" sysbench test (oltp_ro with reconnect after each query), ~ 25% more connects/disconnects were measured
Analysys:
In the beginning of JOIN::cleanup there is code that is supposed to
free all filesort buffers. The code assumes that the table being sorted
is the first non-constant table. To get this table it calls:
first_top_level_tab(this, WITHOUT_CONST_TABLES)
However, first_top_level_tab() instead returned the wrong table - the first
one in the plan, instead of the first non-constant table. There is no other
place outside filesort() where sort buffers may be freed. As a result, the
sort buffer was not freed, and there was a memory leak.
Solution:
Change first_top_level_tab(), to test for WITH_CONST_TABLES instead of
WITHOUT_CONST_TABLES.
mysql-test/r/create.result:
Updated test results
mysql-test/t/create.test:
Updated test
sql/sql_base.cc:
Use push_internal_handler/pop_internal_handler to avoid errors & warnings instead of clear_error
Give a warnings instead of an error for CREATE TABLE IF EXISTS
sql/sql_parse.cc:
Check if we failed because of table exists (can only happen from create)
sql/sql_table.cc:
Check if we failed because of table exists (can only happen from create)
Analysis:
The reason for the suboptimal plan when querying IS tables through a view
was that the view columns that participate in an equality are wrapped by
an Item_direct_view_ref and were not recognized as being direct column
references.
Solution:
Use the original Item_field objects via the real_item() method.
mysql-test/r/create.result:
Added test case to show that CREATE TABLE also is not waiting if table exists.
mysql-test/t/create.test:
Added test case to show that CREATE TABLE also is not waiting if table exists.
sql/sql_base.cc:
Clear also warnings from acquire_locks if we retry.
- Added option to check_if_table_exists() to quickly check if table exists (either SHARE or .FRM)
- Extended lock_table_names() to not wait for meta data locks if CREATE IF NOT EXISTS is used.
mysql-test/r/create.result:
New test case
mysql-test/t/create.test:
New test case
sql/sql_base.cc:
Added option to check_if_table_exists() to quickly check if table exists (either SHARE or .FRM)
Extended lock_table_names() to not wait for meta data locks if CREATE IF NOT EXISTS is used.
sql/sql_base.h:
Updated prototype
sql/sql_db.cc:
Added extra argument to call to check_if_table_exists()
Description: A very large database name causes buffer
overflow in functions acl_get() and
check_grant_db() in sql_acl.cc. It happens
due to an unguarded string copy operation.
This puts required sanity checks before
copying db string to destination buffer.
1. The PERSISTENT FOR clause of the ANALYZE command overrides
the setting of the system variable use_stat_tables:
with this clause ANALYZE unconditionally collects persistent
statistics.
2. ANALYZE collects persistent statistics only for tables of
the USER category. So it never collects persistent statistics
for system tables.
Code in MDL subsystem assumes that identifiers of objects can't
be longer than NAME_LEN characters. This assumption was broken
when one tried to construct MDL_key based on table alias, which
can have arbitrary length. Since MDL_key's (and MDL locks) are
not really used for table aliases this patch changes code to
not initialize MDL_key object for table list element representing
aliases.
table_stat -> table_stats
column_stat -> column_stats
index_stat -> index_stats
to be in line with the names of innodb statistical tables
from mysql-5.6: innodb_table_stats and innodb_index_stats.
When inserting a record with update on duplicate keys the server calls
the ha_index_read_idx_map handler function to look for the record
that violates unique key constraints. The third parameter of this call
should mark only the base components of the index where the server is
searched for the record. Possible hidden components of the primary key
are to be unmarked.
Assertion happened because sql_kill did not set OK status in diagnostic area
in the case of connection suicide (id to kill == thd->thread_id), issued
via COM_PROCESS_KILL , e.g using mysql_kill()
This patch ensures that diagnostic area is initialized in this specific case.
The problem was that in debugging binaries it try to print item to assign human readable name to the item.
But subquery item was already freed (join_free/cleanup with full cleanup) so Item_field refers to temporary
table which memory had been already freed.
If the setting of system variables does not allow to use join buffer
for a join query with GROUP BY <f1,...> / ORDER BY <f1,...> then
filesort is not needed if the first joined table is scanned in
the order compatible with order specified by the list <f1,...>.
Root cause was that number of entries in commit checkpoint buffer
was bigger than total available entries in the mmap()'ed score
file. This causes TC_LOG_MMAP to run out of entries before even
the first checkpoint is started, which causes a hang.
Fixed by making sure we have fewer entries within one commit
checkpoint than total available scorefile entries.
Another part of this bug was discovery of severel unrelated bugs
in TC_LOG_MMAP dating back to 5.1. These were fixed in 5.1 and
will be merged up (the problem this patch fixes exists only in
10.0).
Fix some problems in the TC_LOG_MMAP commit processing, which could
lead to assertions in some cases.
Problems are mostly reproducible in MariaDB 10.0 with asynchroneous
commit checkpoints, but most of the problems were present in earlier
versions also.
The invalid implementation of the method Field_bit::cmp_max could
trigger a valgrind complain or could lead to incorrect statistical
data when collecting engine-independent statistics on BIT fields.
