Bug#54678: InnoDB, TRUNCATE, ALTER, I_S SELECT, crash or deadlock
- Incompatible change: truncate no longer resorts to a row by
row delete if the storage engine does not support the truncate
method. Consequently, the count of affected rows does not, in
any case, reflect the actual number of rows.
- Incompatible change: it is no longer possible to truncate a
table that participates as a parent in a foreign key constraint,
unless it is a self-referencing constraint (both parent and child
are in the same table). To work around this incompatible change
and still be able to truncate such tables, disable foreign checks
with SET foreign_key_checks=0 before truncate. Alternatively, if
foreign key checks are necessary, please use a DELETE statement
without a WHERE condition.
Problem description:
The problem was that for storage engines that do not support
truncate table via a external drop and recreate, such as InnoDB
which implements truncate via a internal drop and recreate, the
delete_all_rows method could be invoked with a shared metadata
lock, causing problems if the engine needed exclusive access
to some internal metadata. This problem originated with the
fact that there is no truncate specific handler method, which
ended up leading to a abuse of the delete_all_rows method that
is primarily used for delete operations without a condition.
Solution:
The solution is to introduce a truncate handler method that is
invoked when the engine does not support truncation via a table
drop and recreate. This method is invoked under a exclusive
metadata lock, so that there is only a single instance of the
table when the method is invoked.
Also, the method is not invoked and a error is thrown if
the table is a parent in a non-self-referencing foreign key
relationship. This was necessary to avoid inconsistency as
some integrity checks are bypassed. This is inline with the
fact that truncate is primarily a DDL operation that was
designed to quickly remove all data from a table.
to allow temp table operations) -- prerequisite patch #1.
Move a piece of code that initialiazes TABLE instance
after it was successfully opened into a separate function.
This function will be reused in the following patches.
'CREATE TABLE IF NOT EXISTS ... SELECT' behaviour
BUG#47132, BUG#47442, BUG49494, BUG#23992 and BUG#48814 will disappear
automatically after the this patch.
BUG#55617 is fixed by this patch too.
This is the 5.5 part.
It implements:
- 'CREATE TABLE IF NOT EXISTS ... SELECT' statement will not insert
anything and binlog anything if the table already exists.
It only generate a warning that table already exists.
- A couple of test cases for the behavior changing.
corruption on ADD PARTITION and LOCK TABLE
Bug#53770: Server crash at handler.cc:2076 on
LOAD DATA after timed out COALESCE PARTITION
5.5 fix for:
Bug#51042: REORGANIZE PARTITION can leave table in an
inconsistent state in case of crash
Needs to be back-ported to 5.1
5.5 fix for:
Bug#50418: DROP PARTITION does not interact with
transactions
Main problem was non-persistent operations done
before meta-data lock was taken (53770+53676).
And 53676 needed to keep the table/partitions opened and locked
while copying the data to the new partitions.
Also added thorough tests to spot some additional bugs
in the ddl_log code, which could result in bad state
between the .frm and partitions.
Collapsed patch, includes all fixes required from the reviewers.
/*![:version:] Query Code */, where [:version:] is a sequence of 5
digits representing the mysql server version(e.g /*!50200 ... */),
is a special comment that the query in it can be executed on those
servers whose versions are larger than the version appearing in the
comment. It leads to a security issue when slave's version is larger
than master's. A malicious user can improve his privileges on slaves.
Because slave SQL thread is running with SUPER privileges, so it can
execute queries that he/she does not have privileges on master.
This bug is fixed with the logic below:
- To replace '!' with ' ' in the magic comments which are not applied on
master. So they become common comments and will not be applied on slave.
- Example:
'INSERT INTO t1 VALUES (1) /*!10000, (2)*/ /*!99999 ,(3)*/
will be binlogged as
'INSERT INTO t1 VALUES (1) /*!10000, (2)*/ /* 99999 ,(3)*/
TABLES <list> WITH READ LOCK are incompatible".
The problem was that FLUSH TABLES <list> WITH READ LOCK
which was issued when other connection has acquired global
read lock using FLUSH TABLES WITH READ LOCK was blocked
and has to wait until global read lock is released.
This issue stemmed from the fact that FLUSH TABLES <list>
WITH READ LOCK implementation has acquired X metadata locks
on tables to be flushed. Since these locks required acquiring
of global IX lock this statement was incompatible with global
read lock.
This patch addresses problem by using SNW metadata type of
lock for tables to be flushed by FLUSH TABLES <list> WITH
READ LOCK. It is OK to acquire them without global IX lock
as long as we won't try to upgrade those locks. Since SNW
locks allow concurrent statements using same table FLUSH
TABLE <list> WITH READ LOCK now has to wait until old
versions of tables to be flushed go away after acquiring
metadata locks. Since such waiting can lead to deadlock
MDL deadlock detector was extended to take into account
waits for flush and resolve such deadlocks.
As a bonus code in open_tables() which was responsible for
waiting old versions of tables to go away was refactored.
Now when we encounter old version of table in open_table()
we don't back-off and wait for all old version to go away,
but instead wait for this particular table to be flushed.
Such approach supported by deadlock detection should reduce
number of scenarios in which FLUSH TABLES aborts concurrent
multi-statement transactions.
Note that active FLUSH TABLES <list> WITH READ LOCK still
blocks concurrent FLUSH TABLES WITH READ LOCK statement
as the former keeps tables open and thus prevents the
latter statement from doing flush.
prepared statements
Using GROUP_CONCAT() together with the WITH ROLLUP modifier
could crash the server.
The reason was a combination of several facts:
1. The Item_func_group_concat class stores pointers to ORDER
objects representing the columns in the ORDER BY clause of
GROUP_CONCAT().
2. find_order_in_list() called from
Item_func_group_concat::setup() modifies the ORDER objects so
that their 'item' member points to the arguments list
allocated in the Item_func_group_concat constructor.
3. In some cases (e.g. in JOIN::rollup_make_fields) a copy of
the original Item_func_group_concat object could be created by
using the Item_func_group_concat::Item_func_group_concat(THD
*thd, Item_func_group_concat *item) copy constructor. The
latter essentially creates a shallow copy of the source
object. Memory for the arguments array is allocated on
thd->mem_root, but the pointers for arguments and ORDER are
copied verbatim.
What happens in the test case is that when executing the query
for the first time, after a copy of the original
Item_func_group_concat object has been created by
JOIN::rollup_make_fields(), find_order_in_list() is called for
this new object. It then resolves ORDER BY by modifying the
ORDER objects so that they point to elements of the arguments
array which is local to the cloned object. When thd->mem_root
is freed upon completing the execution, pointers in the ORDER
objects become invalid. Those ORDER objects, however, are also
shared with the original Item_func_group_concat object which is
preserved between executions of a prepared statement. So the
first call to find_order_in_list() for the original object on
the second execution tries to dereference an invalid pointer.
The solution is to create copies of the ORDER objects when
copying Item_func_group_concat to not leave any stale pointers
in other instances with different lifecycles.
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()).
an atomic counter"
Split the large LOCK_open section in open_table().
Do not call open_table_from_share() under LOCK_open.
Remove thd->version.
This fixes
Bug#50589 "Server hang on a query evaluated using a temporary
table"
Bug#51557 "LOCK_open and kernel_mutex are not happy together"
Bug#49463 "LOCK_table and innodb are not nice when handler
instances are created".
This patch has effect on storage engines that rely on
ha_open() PSEA method being called under LOCK_open.
In particular:
1) NDB is broken and left unfixed. NDB relies on LOCK_open
being kept as part of ha_open(), since it uses auto-discovery.
While previously the NDB open code was race-prone, now
it simply fails on asserts.
2) HEAP engine had a race in ha_heap::open() when
a share for the same table could be added twice
to the list of shares, or a dangling reference to a share
stored in HEAP handler. This patch aims to address this
problem by 'pinning' the newly created share in the
internal HEAP engine share list until at least one
handler instance is created using that share.
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.
Conflicts:
Text conflict in mysql-test/r/archive.result
Contents conflict in mysql-test/r/innodb_bug38231.result
Text conflict in mysql-test/r/mdl_sync.result
Text conflict in mysql-test/suite/binlog/t/disabled.def
Text conflict in mysql-test/suite/rpl_ndb/r/rpl_ndb_binlog_format_errors.result
Text conflict in mysql-test/t/archive.test
Contents conflict in mysql-test/t/innodb_bug38231.test
Text conflict in mysql-test/t/mdl_sync.test
Text conflict in sql/sp_head.cc
Text conflict in sql/sql_show.cc
Text conflict in sql/table.cc
Text conflict in sql/table.h
The problem was that TRUNCATE TABLE didn't take a exclusive
lock on a table if it resorted to truncating via delete of
all rows in the table. Specifically for InnoDB tables, this
could break proper isolation as InnoDB ends up aborting some
granted locks when truncating a table.
The solution is to take a exclusive metadata lock before
TRUNCATE TABLE can proceed. This guarantees that no other
transaction is using the table.
Incompatible change: Truncate via delete no longer fails
if sql_safe_updates is activated (this was a undocumented
side effect).
bitmap_is_set(table->read_set, field_index))
UPDATE on an InnoDB table modifying the same index that is used
to satisfy the WHERE condition could trigger a debug assertion
under some circumstances.
Since for engines with the HA_PRIMARY_KEY_IN_READ_INDEX flag
set results of an index scan on a secondary index are appended
by the primary key value, if a query involves only columns from
the primary key and a secondary index, the latter is considered
to be covering.
That tricks mysql_update() to mark for reading only columns
from the secondary index when it does an index scan to retrieve
rows to update in case a part of that key is also being
updated. However, there may be other columns in WHERE that are
part of the primary key, but not the secondary one.
What we actually want to do in this case is to add index
columns to the existing WHERE columns bitmap rather than
replace it.
transactional SELECT and ALTER TABLE ... REBUILD PARTITION".
The goal of this patch is to decouple type of metadata
lock acquired for table by open_tables() from type of
table-level lock to be acquired on it.
To achieve this we change approach to how we determine what
type of metadata lock should be acquired on table to be open.
Now instead of inferring it at open_tables() time from flags
and type of table-level lock we rely on that type of metadata
lock is properly set at parsing time and is not changed
further.
Conflicts:
Text conflict in mysql-test/r/grant.result
Text conflict in mysql-test/t/grant.test
Text conflict in mysys/mf_loadpath.c
Text conflict in sql/slave.cc
Text conflict in sql/sql_priv.h
MYSQL_BIN_LOG m_table_map_version member and it's associated
functions were not used in the logic of binlogging and replication,
this patch removed all related code.
greedy_search optimizer_search_depth=0
The algorithm inside restore_prev_nj_state failed to
properly update the counters within the NESTED_JOIN
tree. The counter was decremented each time a table in the
node was removed from the QEP, the correct thing to do being
only to decrement it when the last table in the child node
was removed from the plan. This lead to node counters
getting negative values and the plan thus appeared
impossible. An assertion caught this.
Fixed by not recursing up the tree unless the last table in
the join nest node is removed from the plan
Adding my_global.h first in all files using
NO_EMBEDDED_ACCESS_CHECKS.
Correcting a merge problem resulting from a changed definition
of check_some_access compared to the original patches.
This patch:
- Moves all definitions from the mysql_priv.h file into
header files for the component where the variable is
defined
- Creates header files if the component lacks one
- Eliminates all include directives from mysql_priv.h
- Eliminates all circular include cycles
- Rename time.cc to sql_time.cc
- Rename mysql_priv.h to sql_priv.h
into partitioned MyISAM table
Problem was that the ha_data structure was introduced in 5.1
and only used for partitioning first, but with the intention
of be of use for others engines as well, and when used by other
engines it would clash if it also was partitioned.
Solution is to move the partitioning specific data to a separate
structure, with its own mutex (which is used for auto_increment).
Also did rename PARTITION_INFO to PARTITION_STATS since there
already exist a class named partition_info, also cleaned up
some related variables.
Add a wait-for graph based deadlock detector to the
MDL subsystem.
Fixes bug #46272 "MySQL 5.4.4, new MDL: unnecessary deadlock" and
bug #37346 "innodb does not detect deadlock between update and
alter table".
The first bug manifested itself as an unwarranted abort of a
transaction with ER_LOCK_DEADLOCK error by a concurrent ALTER
statement, when this transaction tried to repeat use of a
table, which it has already used in a similar fashion before
ALTER started.
The second bug showed up as a deadlock between table-level
locks and InnoDB row locks, which was "detected" only after
innodb_lock_wait_timeout timeout.
A transaction would start using the table and modify a few
rows.
Then ALTER TABLE would come in, and start copying rows
into a temporary table. Eventually it would stumble on
the modified records and get blocked on a row lock.
The first transaction would try to do more updates, and get
blocked on thr_lock.c lock.
This situation of circular wait would only get resolved
by a timeout.
Both these bugs stemmed from inadequate solutions to the
problem of deadlocks occurring between different
locking subsystems.
In the first case we tried to avoid deadlocks between metadata
locking and table-level locking subsystems, when upgrading shared
metadata lock to exclusive one.
Transactions holding the shared lock on the table and waiting for
some table-level lock used to be aborted too aggressively.
We also allowed ALTER TABLE to start in presence of transactions
that modify the subject table. ALTER TABLE acquires
TL_WRITE_ALLOW_READ lock at start, and that block all writes
against the table (naturally, we don't want any writes to be lost
when switching the old and the new table). TL_WRITE_ALLOW_READ
lock, in turn, would block the started transaction on thr_lock.c
lock, should they do more updates. This, again, lead to the need
to abort such transactions.
The second bug occurred simply because we didn't have any
mechanism to detect deadlocks between the table-level locks
in thr_lock.c and row-level locks in InnoDB, other than
innodb_lock_wait_timeout.
This patch solves both these problems by moving lock conflicts
which are causing these deadlocks into the metadata locking
subsystem, thus making it possible to avoid or detect such
deadlocks inside MDL.
To do this we introduce new type-of-operation-aware metadata
locks, which allow MDL subsystem to know not only the fact that
transaction has used or is going to use some object but also what
kind of operation it has carried out or going to carry out on the
object.
This, along with the addition of a special kind of upgradable
metadata lock, allows ALTER TABLE to wait until all
transactions which has updated the table to go away.
This solves the second issue.
Another special type of upgradable metadata lock is acquired
by LOCK TABLE WRITE. This second lock type allows to solve the
first issue, since abortion of table-level locks in event of
DDL under LOCK TABLES becomes also unnecessary.
Below follows the list of incompatible changes introduced by
this patch:
- From now on, ALTER TABLE and CREATE/DROP TRIGGER SQL (i.e. those
statements that acquire TL_WRITE_ALLOW_READ lock)
wait for all transactions which has *updated* the table to
complete.
- From now on, LOCK TABLES ... WRITE, REPAIR/OPTIMIZE TABLE
(i.e. all statements which acquire TL_WRITE table-level lock) wait
for all transaction which *updated or read* from the table
to complete.
As a consequence, innodb_table_locks=0 option no longer applies
to LOCK TABLES ... WRITE.
- DROP DATABASE, DROP TABLE, RENAME TABLE no longer abort
statements or transactions which use tables being dropped or
renamed, and instead wait for these transactions to complete.
- Since LOCK TABLES WRITE now takes a special metadata lock,
not compatible with with reads or writes against the subject table
and transaction-wide, thr_lock.c deadlock avoidance algorithm
that used to ensure absence of deadlocks between LOCK TABLES
WRITE and other statements is no longer sufficient, even for
MyISAM. The wait-for graph based deadlock detector of MDL
subsystem may sometimes be necessary and is involved. This may
lead to ER_LOCK_DEADLOCK error produced for multi-statement
transactions even if these only use MyISAM:
session 1: session 2:
begin;
update t1 ... lock table t2 write, t1 write;
-- gets a lock on t2, blocks on t1
update t2 ...
(ER_LOCK_DEADLOCK)
- Finally, support of LOW_PRIORITY option for LOCK TABLES ... WRITE
was abandoned.
LOCK TABLE ... LOW_PRIORITY WRITE from now on has the same
priority as the usual LOCK TABLE ... WRITE.
SELECT HIGH PRIORITY no longer trumps LOCK TABLE ... WRITE in
the wait queue.
- We do not take upgradable metadata locks on implicitly
locked tables. So if one has, say, a view v1 that uses
table t1, and issues:
LOCK TABLE v1 WRITE;
FLUSH TABLE t1; -- (or just 'FLUSH TABLES'),
an error is produced.
In order to be able to perform DDL on a table under LOCK TABLES,
the table must be locked explicitly in the LOCK TABLES list.
Queries optimized with GROUP_MIN_MAX didn't cleanup KEYREAD
optimization properly. As a result subsequent queries may
return incomplete rows (fields are initialized to default
values).
