GRADUALLY IF A TRIGGER EXISTS".
This bug manifested itself in two ways:
- Firstly execution of any data-changing statement which
required prelocking (i.e. involved stored function or
trigger) as part of transaction slowed down a bit all
subsequent statements in this transaction. So performance
in transaction which periodically involved such statements
gradually degraded over time.
- Secondly execution of any data-changing statement which
required prelocking as part of transaction prevented
concurrent FLUSH TABLES WITH READ LOCK from proceeding
until the end of transaction instead of end of particular
statement.
The problem was caused by incorrect handling of metadata lock
used in FTWRL implementation for statements requiring prelocked
mode.
Each statement which changes data acquires global IX lock
with STATEMENT duration. This lock is supposed to block
concurrent FTWRL from proceeding until the statement ends.
When entering prelocked mode, durations of all metadata locks
acquired so far were changed to EXPLICIT, to prevent
substatements from releasing these locks. When prelocked mode
was left, durations of metadata locks were changed to
TRANSACTIONAL (with a few exceptions) so they can be properly
released at the end of transaction.
Unfortunately, this meant that the global IX lock blocking
FTWRL with STATEMENT duration was moved to TRANSACTIONAL
duration after execution of statement requiring prelocking.
Since each subsequent statement that required prelocking and
tried to acquire global IX lock with STATEMENT duration got
a new instance of MDL_ticket, which was later moved to
TRANSACTIONAL duration, this led to unwarranted growth of
number of tickets with TRANSACITONAL duration in this
connection's MDL_context. As result searching for other
tickets in it became slow and acquisition of other metadata
locks by this transaction started to hog CPU.
Moreover, this also meant that after execution of statement
requiring prelocking concurrent FTWRL was blocked
until the end of transaction instead of end of statement.
This patch solves this problem by not moving locks to EXPLICIT
duration when thread enters prelocked mode (unless it is a real
LOCK TABLES mode). This step turned out to be not really
necessary as substatements don't try to release metadata locks.
Consequently, the global IX lock blocking FTWRL keeps its
STATEMENT duration and is properly released at the end of
statement and the above issue goes away.
mysql-test/r/flush.result:
Added test for bug #12641342 - "61401: UPDATE PERFORMANCE
DEGRADES GRADUALLY IF A TRIGGER EXISTS".
mysql-test/t/flush.test:
Added test for bug #12641342 - "61401: UPDATE PERFORMANCE
DEGRADES GRADUALLY IF A TRIGGER EXISTS".
sql/mdl.h:
Added comment describing various types of metadata lock
duration.
sql/sql_class.cc:
Since we no longer change duration of metadata locks to EXPLICIT
when entering prelocked mode (unless it is a real LOCK TABLES)
there is no need to restore proper duration of the locks when
leaving prelocked mode.
sql/sql_class.h:
Do not change duration of metadata locks to EXPLICIT when
entering prelocking mode (unless it is a real LOCK TABLES).
This allows to avoid problems with restoring correct duration
when leaving this mode. It is possible to do this as
substatements won't release metadata locks in any case.
sql/sql_parse.cc:
Added assert checking that we won't release metadata locks
when in substatement.
Issue:
------
New test case mysql_embedded.test was failing on pb2.
Description:
------------
To run this test case executable libmysqld/examples/mysql_embedded is required.
But as per /libmysqld/examples/cmake_install.cmake this executable doesn't get
copied to <install_dir> when mysql is installed at <install_dir>.That is the
reason it was passing in my local branch and failed on pb2 when pushed.
Solution;
---------
Added code in mysql-test-run.pl, which will try to see if this file exists.If
It doesn't exist, test case will be skipped with a skip message. New code in
mysql-test-run.pl looks only for directory libmysqld/examples/mysql_embedded
because this is the only place where this file could/does exist.
mysql-test/mysql-test-run.pl:
Added new variable for mysql_embedded executable.
mysql-test/t/disabled.def:
enabled mysql_embedded.test which was disabled earlier.
mysql-test/t/mysql_embedded.test:
Modified test case to first verify if mysql_embedded executable exists. If
it does not, skip the test.
FAIL IN EMBEDDED SERVER
FreeBSD 64 bit needs the FP_X_DNML to fpsetmask() to prevent exceptions from
propagating into mysql (as a threaded application).
However fpsetmask() itself is deprecated in favor of fedisableexcept().
1. Fixed the #ifdef to check for FP_X_DNML instead of i386.
2. Added a configure.in check for fedisableexcept() and, if present,
this function is called insted of the fpsetmask().
No need for new tests, as the existing tests cover this already.
Removed the affected tests from the experimental list.
Since the Windows authentication support has been added to libmysql, this library depends on the system Secur32 library. Consequently, clients which are linked against libmysql should be also linked with Secur32 (in addition to ws2_32).
In MS VC++ it is possible to embed information about required libraries into object file using #pragma directive. This patch adds such directive when the Windows authentiaction support is compiled. This is similar to analogous #pragma for ws2_32 library in my_init.c
The types mysql_event_general/mysql_event_connection are
being cast to the incompatible type mysql_event. The way
mysql_event and the other types are designed are prone to
strict aliasing violations and can break things depending
on how compilers optimizes this code.
This patch fixes audit interface, so it confirms to strict-
aliasing rules. It introduces incompatible changes to audit
interface:
- mysql_event type has been removed;
- event_class has been removed from mysql_event_generic and
mysql_event_connection types;
- st_mysql_audit::event_notify() second argument is event_class;
- st_mysql_audit::event_notify() third argument is event of type
(const void *).
"Writing Audit Plugins" section of manual should be updated:
http://dev.mysql.com/doc/refman/5.5/en/writing-audit-plugins.html
include/mysql/plugin_audit.h:
event_class has been moved out of mysql_event types.
include/mysql/plugin_audit.h.pp:
event_class has been moved out of mysql_event types.
plugin/audit_null/audit_null.c:
event_class has been moved out of mysql_event types.
sql/sql_audit.cc:
event_class has been moved out of mysql_event types.
The check for empty password in the user account was checking the wrong field.
Fixed to check the proper password hash.
Test case added.
Fixed native_password and old_password plugins that suffered from the same
problems.
Unambuguated the auth_string ACL_USER member : previously it was used for
both password and the authentication string (depending on the plugin). Now
fixed to contain either the authentication string specified or empty string.
SECONDARY INDEX IN INNODB
This is a follow-up patch.
This patch moves part of the new test coverage to a test
file that is only run on debug builds since it used debug-
only features and therefore broke the test case on
release builds.
SECONDARY INDEX IN INNODB
The patches for Bug#11751388 and Bug#11784056 enabled concurrent
reads while creating secondary indexes in InnoDB. However, they
introduced a regression. This regression occured if ALTER TABLE
failed after the index had been added, for example during the
lock upgrade needed to update .FRM. If this happened, InnoDB
and the server got out of sync with regards to which indexes
actually existed. Therefore the patch for Bug#11815600 again
disabled concurrent reads.
This patch re-enables concurrent reads. The original regression
is fixed by splitting the ADD INDEX operation into two parts.
First the new index is created but not made active. This is
done while concurrent reads are allowed. The second part of
the operation makes the index active (or reverts the change).
This is done after lock upgrade, which prevents the original
regression.
In order to implement this change, the patch changes the storage
API for in-place index creation. handler::add_index() is split
into two functions, handler_add_index() and
handler::final_add_index(). The former for creating indexes without
making them visible and the latter for commiting (i.e. making
visible) new indexes or reverting the changes.
Large parts of this patch were written by Marko Mäkelä.
Test case added to innodb_mysql_lock.test.
With this change, the index prefix column length lifted from 767 bytes
to 3072 bytes if "innodb_large_prefix" is set to "true".
rb://603 approved by Marko
mtr_start(): Declare the mtr memory area uninitialized in Valgrind
before initializing the fields.
mtr_commit(): Declare everything uninitialized except
mtr->start_lsn, mtr->end_lsn and mtr->state.
The test case problem stemmed from the fact that a debug sync
signal is a global variable that persists until overwritten
by a new signal. This means that if two different signals
are raised in sequence, a thread waiting for the first signal
might miss it if the second signal sets the global variable
before the thread wakes up.
The solution is to deliver a subsequent signal only after the
waiting thread has received it.
mysql-test/t/query_cache_debug.test:
Wait for signal to be delivered.
The problem is that clients implementing the 4.0 version of the
protocol (that is, mysql-4.0) do not null terminate a string
at the end of the authentication packet. These clients denote
the end of the string with the end of the packet.
Although this goes against the documented (see MySQL Internals
ClientServer Protocol wiki) description of the protocol, these
old clients still need to be supported.
The solution is to support the documented and actual behavior
of the clients. If a client is using the pre-4.1 version of
the protocol, the end of a string in the authentication packet
can either be denoted with a null character or by the end of
the packet. This restores backwards compatibility with old
clients implementing either the documented or actual behavior.
sql/password.c:
The scrambled message, as provided by the user, might not be
properly null terminated. If this is the case, uninitialized
memory past the end of the buffer could theoretically be
accessed. To ensure that this is never the case, copy the
scrambled message over to a null terminated auxiliar buffer.
sql/sql_connect.cc:
Use different execution paths to read strings depending on the
protocol being used. If version 4.0 of the protocol is used,
end of string can be denoted with a NUL character or by the
end of the packet.
If there are not enough bytes left after the current position
of the buffer to satisfy the current string, the string is
considered to be empty. This is required because old clients
do not send the password string field if the password is empty.
The problem is that although AIX implements bzero, its prototype
is not declared by default. Since AC_CHECK_FUNC(bzero) succeeds
even though a prototype is not declared, this breaks compilation
in C++ files where a prototype is required.
The solution is to only use bzero if a prototype is also declared.
configure.in:
Check if bzero is declared. No need to specify the includes,
unisted.h and strings.h are already part of AC_INCLUDES_DEFAULT.
will create multiple running events.
A CREATE IF NOT EXIST on an event that existed and was enabled caused
multiple instances of the event to run. Disabling the event didn't help.
If the event was dropped, the event stopped running, but when created
again, multiple instances of the event were still running. The only way
to get out of this situation was to restart the server.
The problem was that Event_db_repository::create_event() didn't return
enough information to discriminate between situation when event didn't
exist and was created and when event did exist and was not created
(but a warning was emitted). As result in the latter case event
was added to in-memory queue of events second time. And this led to
unwarranted multiple executions of the same event.
The solution is to add out-parameter to Event_db_repository::create_event()
method which will signal that event was not created because it already
exists and so it should not be added to the in-memory queue.
mysql-test/r/events_bugs.result:
Added results for test for Bug#12546938.
mysql-test/t/events_bugs.test:
Added test for Bug#12546938.
sql/event_db_repository.cc:
Event_db_repository::create_event was modified: set newly added out-parameter
event_already_exists to true value if event wasn't created because event
already existed and IF NOT EXIST clause was present.
sql/event_db_repository.h:
Added out-parameter 'event_already_exists' to create_event() method.
sql/events.cc:
Events::create_event was modified: insert new element into
event queue only if event was actually created.
HA_INNOBASE::UPDATE_ROW, TEMPORARY TABLE, TABLE LOCK".
Attempt to update an InnoDB temporary table under LOCK TABLES
led to assertion failure in both debug and production builds
if this temporary table was explicitly locked for READ. The
same scenario works fine for MyISAM temporary tables.
The assertion failure was caused by discrepancy between lock
that was requested on the rows of temporary table at LOCK TABLES
time and by update operation. Since SQL-layer requested a
read-lock at LOCK TABLES time InnoDB engine assumed that upcoming
statements which are going to be executed under LOCK TABLES will
only read table and therefore should acquire only S-lock.
An update operation broken this assumption by requesting X-lock.
Possible approaches to fixing this problem are:
1) Skip locking of temporary tables as locking doesn't make any
sense for connection-local objects.
2) Prohibit changing of temporary table locked by LOCK TABLES ...
READ.
Unfortunately both of these approaches have drawbacks which make
them unviable for stable versions of server.
So this patch takes another approach and changes code in such way
that LOCK TABLES for a temporary table will always request write
lock. In 5.5 version of this patch switch from read lock to write
lock is done on SQL-layer.
mysql-test/suite/innodb/r/innodb_mysql.result:
Added test for bug #11762012 - "54553: INNODB ASSERTS IN
HA_INNOBASE::UPDATE_ROW, TEMPORARY TABLE, TABLE LOCK".
mysql-test/suite/innodb/t/innodb_mysql.test:
Added test for bug #11762012 - "54553: INNODB ASSERTS IN
HA_INNOBASE::UPDATE_ROW, TEMPORARY TABLE, TABLE LOCK".
sql/sql_parse.cc:
Since a temporary table locked by LOCK TABLES can be updated even
if it was only locked for read we always request TL_WRITE locks
for such tables at LOCK TABLES time. This allows to avoid
discrepancy between locks acquired at LOCK TABLES time and by
a statement executed under LOCK TABLES. Such a discrepancy has
caused problems for InnoDB storage engine.
To support this change a part of code implementing LOCK TABLES
has been moved to a helper function.
HA_INNOBASE::UPDATE_ROW, TEMPORARY TABLE, TABLE LOCK".
Attempt to update an InnoDB temporary table under LOCK TABLES
led to assertion failure in both debug and production builds
if this temporary table was explicitly locked for READ. The
same scenario works fine for MyISAM temporary tables.
The assertion failure was caused by discrepancy between lock
that was requested on the rows of temporary table at LOCK TABLES
time and by update operation. Since SQL-layer requested a
read-lock at LOCK TABLES time InnoDB engine assumed that upcoming
statements which are going to be executed under LOCK TABLES will
only read table and therefore should acquire only S-lock.
An update operation broken this assumption by requesting X-lock.
Possible approaches to fixing this problem are:
1) Skip locking of temporary tables as locking doesn't make any
sense for connection-local objects.
2) Prohibit changing of temporary table locked by LOCK TABLES ...
READ.
Unfortunately both of these approaches have drawbacks which make
them unviable for stable versions of server.
So this patch takes another approach and changes code in such way
that LOCK TABLES for a temporary table will always request write
lock. In 5.1 version of this patch switch from read lock to write
lock is done inside of InnoDBs handler methods as doing it on
SQL-layer causes compatibility troubles with FLUSH TABLES WITH
READ LOCK.
mysql-test/suite/innodb/r/innodb_mysql.result:
Added test for bug #11762012 - "54553: INNODB ASSERTS IN
HA_INNOBASE::UPDATE_ROW, TEMPORARY TABLE, TABLE LOCK".
mysql-test/suite/innodb/t/innodb_mysql.test:
Added test for bug #11762012 - "54553: INNODB ASSERTS IN
HA_INNOBASE::UPDATE_ROW, TEMPORARY TABLE, TABLE LOCK".
mysql-test/suite/innodb_plugin/r/innodb_mysql.result:
Added test for bug #11762012 - "54553: INNODB ASSERTS IN
HA_INNOBASE::UPDATE_ROW, TEMPORARY TABLE, TABLE LOCK".
mysql-test/suite/innodb_plugin/t/innodb_mysql.test:
Added test for bug #11762012 - "54553: INNODB ASSERTS IN
HA_INNOBASE::UPDATE_ROW, TEMPORARY TABLE, TABLE LOCK".
storage/innobase/handler/ha_innodb.cc:
Assume that a temporary table locked by LOCK TABLES can be updated
even if it was only locked for read and therefore an X-lock should
be always requested for such tables.
storage/innodb_plugin/handler/ha_innodb.cc:
Assume that a temporary table locked by LOCK TABLES can be updated
even if it was only locked for read and therefore an X-lock should
be always requested for such tables.
Problem: MYSQL_BIN_LOG::reset_logs acquires mutexes in wrong order.
The correct order is first LOCK_thread_count and then LOCK_log. This function
does it the other way around. This leads to deadlock when run in parallel
with a thread that takes the two locks in correct order. For example, a thread
that disconnects will take the locks in the correct order.
Fix: change order of the locks in MYSQL_BIN_LOG::reset_logs:
first LOCK_thread_count and then LOCK_log.
mysql-test/suite/binlog/r/binlog_reset_master.result:
added result file
mysql-test/suite/binlog/t/binlog_reset_master.test:
Added test case that demonstrates deadlock because of wrong mutex order.
The deadlock is between two threads:
- RESET MASTER acquires mutexes in wrong order.
- client thread shutdown code acquires mutexes in right order.
Actually, this test case does not produce deadlock in 5.1, probably
the client thread shutdown code does not hold both mutexes at the same
time. However, the bug existed in 5.1 (mutexes are taken in the wrong
order) so we push the test case to 5.1 too, to prevent future
regressions.
sql/log.cc:
Change mutex acquisition to the correct order:
first LOCK_thread_count, then LOCK_log.
sql/mysqld.cc:
Add debug code to synchronize test case.
