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.
multi-table UPDATE IGNORE.
The problem was that if there was an active SELECT statement
during trigger execution, an error risen during the execution
may cause a crash. The fix is to temporary reset LEX::current_select
before trigger execution and restore it afterwards. This way
errors risen during the trigger execution are processed as
if there was no active SELECT.
FLUSH TABLES <list> WITH READ LOCK are incompatible" to
be pushed as separate patch.
Replaced thread state name "Waiting for table", which was
used by threads waiting for a metadata lock or table flush,
with a set of names which better reflect types of resources
being waited for.
Also replaced "Table lock" thread state name, which was used
by threads waiting on thr_lock.c table level lock, with more
elaborate "Waiting for table level lock", to make it
more consistent with other thread state names.
Updated test cases and their results according to these
changes.
Fixed sys_vars.query_cache_wlock_invalidate_func test to not
to wait for timeout of wait_condition.inc script.
Backport of:
------------------------------------------------------------
revno: 2630.4.1
committer: Dmitry Lenev <dlenev@mysql.com>
branch nick: mysql-6.0-3726-w
timestamp: Fri 2008-05-23 17:54:03 +0400
message:
WL#3726 "DDL locking for all metadata objects".
After review fixes in progress.
------------------------------------------------------------
This is the first patch in series. It transforms the metadata
locking subsystem to use a dedicated module (mdl.h,cc). No
significant changes in the locking protocol.
The import passes the test suite with the exception of
deprecated/removed 6.0 features, and MERGE tables. The latter
are subject to a fix by WL#4144.
Unfortunately, the original changeset comments got lost in a merge,
thus this import has its own (largely insufficient) comments.
This patch fixes Bug#25144 "replication / binlog with view breaks".
Warning: this patch introduces an incompatible change:
Under LOCK TABLES, it's no longer possible to FLUSH a table that
was not locked for WRITE.
Under LOCK TABLES, it's no longer possible to DROP a table or
VIEW that was not locked for WRITE.
******
Backport of:
------------------------------------------------------------
revno: 2630.4.2
committer: Dmitry Lenev <dlenev@mysql.com>
branch nick: mysql-6.0-3726-w
timestamp: Sat 2008-05-24 14:03:45 +0400
message:
WL#3726 "DDL locking for all metadata objects".
After review fixes in progress.
******
Backport of:
------------------------------------------------------------
revno: 2630.4.3
committer: Dmitry Lenev <dlenev@mysql.com>
branch nick: mysql-6.0-3726-w
timestamp: Sat 2008-05-24 14:08:51 +0400
message:
WL#3726 "DDL locking for all metadata objects"
Fixed failing Windows builds by adding mdl.cc to the lists
of files needed to build server/libmysqld on Windows.
******
Backport of:
------------------------------------------------------------
revno: 2630.4.4
committer: Dmitry Lenev <dlenev@mysql.com>
branch nick: mysql-6.0-3726-w
timestamp: Sat 2008-05-24 21:57:58 +0400
message:
WL#3726 "DDL locking for all metadata objects".
Fix for assert failures in kill.test which occured when one
tried to kill ALTER TABLE statement on merge table while it
was waiting in wait_while_table_is_used() for other connections
to close this table.
These assert failures stemmed from the fact that cleanup code
in this case assumed that temporary table representing new
version of table was open with adding to THD::temporary_tables
list while code which were opening this temporary table wasn't
always fulfilling this.
This patch changes code that opens new version of table to
always do this linking in. It also streamlines cleanup process
for cases when error occurs while we have new version of table
open.
******
WL#3726 "DDL locking for all metadata objects"
Add libmysqld/mdl.cc to .bzrignore.
******
Backport of:
------------------------------------------------------------
revno: 2630.4.6
committer: Dmitry Lenev <dlenev@mysql.com>
branch nick: mysql-6.0-3726-w
timestamp: Sun 2008-05-25 00:33:22 +0400
message:
WL#3726 "DDL locking for all metadata objects".
Addition to the fix of assert failures in kill.test caused by
changes for this worklog.
Make sure we close the new table only once.
The problem is that some DDL statements (ALTER TABLE, CREATE
TRIGGER, FLUSH TABLES, ...) when under LOCK TABLES need to
momentarily drop the lock, reopen the table and grab the write
lock again (using reopen_tables). When grabbing the lock again,
reopen_tables doesn't pass a flag to mysql_lock_tables in
order to ignore the impending global read lock, which causes a
assertion because LOCK_open is being hold. Also dropping the
lock must not signal to any threads that the table has been
relinquished (related to the locking/flushing protocol).
The solution is to correct the way the table is reopenned
and the locks grabbed. When reopening the table and under
LOCK TABLES, the table version should be set to 0 so other
threads have to wait for the table. When grabbing the lock,
any other flush should be ignored because it's theoretically
a atomic operation. The chosen solution also fixes a potential
discrepancy between binlog and GRL (global read lock) because
table placeholders were being ignored, now a FLUSH TABLES WITH
READ LOCK will properly for table with open placeholders.
It's also important to mention that this patch doesn't fix
a potential deadlock if one uses two GRLs under LOCK TABLES
concurrently.
This bug is actually two bugs in one, one of which is CREATE TRIGGER under
LOCK TABLES and the other is CREATE TRIGGER under LOCK TABLES simultaneous
to a FLUSH TABLES WITH READ LOCK (global read lock). Both situations could
lead to a server crash or deadlock.
The first problem arises from the fact that when under LOCK TABLES, if the
table is in the set of locked tables, the table is already open and it doesn't
need to be reopened (not a placeholder). Also in this case, if the table is
not write locked, a exclusive lock can't be acquired because of a possible
deadlock with another thread also holding a (read) lock on the table. The
second issue arises from the fact that one should never wait for a global
read lock if it's holding any locked tables, because the global read lock
is waiting for these tables and this leads to a circular wait deadlock.
The solution for the first case is to check if the table is write locked
and upgraded the write lock to a exclusive lock and fail otherwise for non
write locked tables. Grabbin the exclusive lock in this case also means
to ensure that the table is opened only by the calling thread. The second
issue is partly fixed by not waiting for the global read lock if the thread
is holding any locked tables.
The second issue is only partly addressed in this patch because it turned
out to be much wider and also affects other DDL statements. Reported as
Bug#32395
