corruption and crash results
An index creation statement where the index key
is larger/wider than the column it references
should throw an error.
A statement like:
CREATE TABLE t1 (a CHAR(1), PRIMARY KEY (A(255)))
did not error, but a segmentation fault followed when
an insertion was attempted on the table
The partial key validiation clause has been
restructured to (hopefully) better document which
uses of partial keys are valid.
This patch introduces timeouts for metadata locks.
The timeout is specified in seconds using the new dynamic system
variable "lock_wait_timeout" which has both GLOBAL and SESSION
scopes. Allowed values range from 1 to 31536000 seconds (= 1 year).
The default value is 1 year.
The new server parameter "lock-wait-timeout" can be used to set
the default value parameter upon server startup.
"lock_wait_timeout" applies to all statements that use metadata locks.
These include DML and DDL operations on tables, views, stored procedures
and stored functions. They also include LOCK TABLES, FLUSH TABLES WITH
READ LOCK and HANDLER statements.
The patch also changes thr_lock.c code (table data locks used by MyISAM
and other simplistic engines) to use the same system variable.
InnoDB row locks are unaffected.
One exception to the handling of the "lock_wait_timeout" variable
is delayed inserts. All delayed inserts are executed with a timeout
of 1 year regardless of the setting for the global variable. As the
connection issuing the delayed insert gets no notification of
delayed insert timeouts, we want to avoid unnecessary timeouts.
It's important to note that the timeout value is used for each lock
acquired and that one statement can take more than one lock.
A statement can therefore block for longer than the lock_wait_timeout
value before reporting a timeout error. When lock timeout occurs,
ER_LOCK_WAIT_TIMEOUT is reported.
Test case added to lock_multi.test.
added:
include/ctype_numconv.inc
mysql-test/include/ctype_numconv.inc
mysql-test/r/ctype_binary.result
mysql-test/t/ctype_binary.test
Adding tests
modified:
mysql-test/r/bigint.result
mysql-test/r/case.result
mysql-test/r/create.result
mysql-test/r/ctype_cp1251.result
mysql-test/r/ctype_latin1.result
mysql-test/r/ctype_ucs.result
mysql-test/r/func_gconcat.result
mysql-test/r/func_str.result
mysql-test/r/metadata.result
mysql-test/r/ps_1general.result
mysql-test/r/ps_2myisam.result
mysql-test/r/ps_3innodb.result
mysql-test/r/ps_4heap.result
mysql-test/r/ps_5merge.result
mysql-test/r/show_check.result
mysql-test/r/type_datetime.result
mysql-test/r/type_ranges.result
mysql-test/r/union.result
mysql-test/suite/ndb/r/ps_7ndb.result
mysql-test/t/ctype_cp1251.test
mysql-test/t/ctype_latin1.test
mysql-test/t/ctype_ucs.test
mysql-test/t/func_str.test
Fixing tests
@ sql/field.cc
- Return str result using my_charset_numeric.
- Using real multi-byte aware str_to_XXX functions
to handle tricky charset values propely (e.g. UCS2)
@ sql/field.h
- Changing derivation of non-string field types to DERIVATION_NUMERIC.
- Changing binary() for numeric/datetime fields to always
return TRUE even if charset is not my_charset_bin. We need
this to keep ha_base_keytype() return HA_KEYTYPE_BINARY.
- Adding BINARY_FLAG into some fields, because it's not
being set automatically anymore with
"my_charset_bin to my_charset_numeric" change.
- Changing derivation for numeric/datetime datatypes to a weaker
value, to make "SELECT concat('string', field)" use character
set of the string literal for the result of the function.
@ sql/item.cc
- Implementing generic val_str_ascii().
- Using max_char_length() instead of direct read of max_length
to make "tricky" charsets like UCS2 work.
NOTE: in the future we'll possibly remove all direct reads of max_length
- Fixing Item_num::safe_charset_converter().
Previously it alligned binary string to
character string (for example by adding leading 0x00
when doing binary->UCS2 conversion). Now it just
converts from my_charset_numbner to "tocs".
- Using val_str_ascii() in Item::get_time() to make UCS2 arguments work.
- Other misc changes
@ sql/item.h
- Changing MY_COLL_CMP_CONV and MY_COLL_ALLOW_CONV to
bit operations instead of hard-coded bit masks.
- Addding new method DTCollation.set_numeric().
- Adding new methods to Item.
- Adding helper functions to make code look nicer:
agg_item_charsets_for_string_result()
agg_item_charsets_for_comparison()
- Changing charset for Item_num-derived items
from my_charset_bin to my_charset_numeric
(which is an alias for latin1).
@ sql/item_cmpfunc.cc
- Using new helper functions
- Other misc changes
@ sql/item_cmpfunc.h
- Fixing strcmp() to return max_length=2.
Previously it returned 1, which was wrong,
because it did not fit '-1'.
@ sql/item_func.cc
- Using new helper functions
- Other minor changes
@ sql/item_func.h
- Removing unused functions
- Adding helper functions
agg_arg_charsets_for_string_result()
agg_arg_charsets_for_comparison()
- Adding set_numeric() into constructors of numeric items.
- Using fix_length_and_charset() and fix_char_length()
instead of direct write to max_length.
@ sql/item_geofunc.cc
- Changing class for Item_func_geometry_type and
Item_func_as_wkt from Item_str_func to
Item_str_ascii_func, to make them return UCS2 result
properly (when character_set_connection=ucs2).
@ sql/item_geofunc.h
- Changing class for Item_func_geometry_type and
Item_func_as_wkt from Item_str_func to
Item_str_ascii_func, to make them return UCS2 result
properly (when @@character_set_connection=ucs2).
@ sql/item_strfunc.cc
- Implementing Item_str_func::val_str().
- Renaming val_str to val_str_ascii for some items,
to make them work with UCS2 properly.
- Using new helper functions
- All single-argument functions that expect string
result now call this method:
agg_arg_charsets_for_string_result(collation, args, 1);
This enables character set conversion to @@character_set_connection
in case of pure numeric input.
@ sql/item_strfunc.h
- Introducing Item_str_ascii_func - for functions
which return pure ASCII data, for performance purposes,
as well as for the cases when the old implementation
of val_str() was heavily 8-bit oriented and implementing
a UCS2-aware version is tricky.
@ sql/item_sum.cc
- Using new helper functions.
@ sql/item_timefunc.cc
- Using my_charset_numeric instead of my_charset_bin.
- Using fix_char_length(), fix_length_and_charset()
and fix_length_and_charset_datetime()
instead of direct write to max_length.
- Using tricky-charset aware function str_to_time_with_warn()
@ sql/item_timefunc.h
- Using new helper functions for charset and length initialization.
- Changing base class for Item_func_get_format() to make
it return UCS2 properly (when character_set_connection=ucs2).
@ sql/item_xmlfunc.cc
- Using new helper function
@ sql/my_decimal.cc
- Adding a new DECIMAL to CHAR converter
with real multibyte support (e.g. UCS2)
@ sql/mysql_priv.h
- Introducing a new derivation level for numeric/datetime data types.
- Adding macros for my_charset_numeric and MY_REPERTOIRE_NUMERIC.
- Adding prototypes for str_set_decimal()
- Adding prototypes for character-set aware str_to_xxx() functions.
@ sql/protocol.cc
- Changing charsetnr to "binary" client-side metadata for
numeric/datetime data types.
@ sql/time.cc
- Adding to_ascii() helper function, to convert a string
in any character set to ascii representation. In the
future can be extended to understand digits written
in various non-Latin word scripts.
- Adding real multy-byte character set aware versions for str_to_XXXX,
to make these these type of queries work correct:
INSERT INTO t1 SET datetime_column=ucs2_expression;
@ strings/ctype-ucs2.c
- endptr was not calculated correctly. INSERTing of UCS2
values into numeric columns returned warnings about
truncated wrong data.
rqg_mdl_stability".
When start of statement's waiting on a metadata lock
created more than one loop in waiters graph server might
have entered deadlock condition.
The problem was that in the case described above MDL deadlock
detector had to perform several searches for deadlock but
forgot to reset Deadlock_detection_context before performing
new search.
Failure to do so has broken assumption in code resposible for
choosing victim that if Deadlock_detection_context::victim
is set we also have read lock on m_waiting_for_lock for this
context. As result this lock could have been unlocked more
times than it was acquired which corrupted rwlock's state
which led to server deadlock.
This fix ensures that such reset is done before each attempt
to find a deadlock.
The problem becomes apparent only if HAVE_purify is undefined.
It related to the part of code placed in open_table_from_share() fuction
where we initialize record buffer only if HAVE_purify is enabled.
So in case of HAVE_purify=OFF record buffer is not initialized
on open table stage.
Next we read key, find NULL value and update appropriate null bit
but do not update record buffer. After that the record is stored
in the join cache(store_record_in_cache). For CHAR fields we
strip trailing spaces and in our case this procedure uses
uninitialized record buffer.
The fix is to skip stripping space procedure in case of null values
for CHAR fields(partially based on 6.0 JOIN_CACHE implementation).
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).
Conflicts:
Text conflict in .bzr-mysql/default.conf
Text conflict in mysql-test/suite/rpl/r/rpl_slow_query_log.result
Text conflict in mysql-test/suite/rpl/t/rpl_slow_query_log.test
Conflict adding files to server-tools. Created directory.
Conflict because server-tools is not versioned, but has versioned children. Versioned directory.
Conflict adding files to server-tools/instance-manager. Created directory.
Conflict because server-tools/instance-manager is not versioned, but has versioned children. Versioned directory.
Contents conflict in server-tools/instance-manager/options.cc
Text conflict in sql/mysqld.cc
and MDL".
Concurrent execution of a multi-DELETE statement and ALTER
TABLE statement which affected one of the tables used in
the multi-DELETE sometimes led to deadlock.
Similar deadlocks might have occured when one performed
INSERT/UPDATE/DELETE on a view and concurrently executed
ALTER TABLE for the view's underlying table, or when one
concurrently executed TRUNCATE TABLE for InnoDB table and
ALTER TABLE for the same table.
These deadlocks were caused by a discrepancy between types of
metadata and thr_lock.cc locks acquired by those statements.
What happened was that multi-DELETE/TRUNCATE/DML-through-the-
view statement in the first connection acquired SR lock on a
table, then ALTER TABLE would come in in the second connection
and acquire SNW metadata lock and TL_WRITE_ALLOW_READ
thr_lock.c lock and then would start waiting for the first
connection during lock upgrade. After that the statement in
the first connection would try to acquire TL_WRITE lock on
table and would start waiting for the second connection,
creating a deadlock.
This patch solves this problem by ensuring that we acquire
SW metadata lock in all cases in which we acquiring write
thr_lock.c lock. This guarantees that deadlocks like the
one described above won't occur since all lock conflicts
in such situation are resolved within MDL subsystem.
This patch also adds assert which should guarantee that
such situations won't arise in future.
Grouping by a subquery in a query with a distinct aggregate
function lead to a wrong result (wrong and unordered
grouping values).
There are two related problems:
1) The query like this:
SELECT (SELECT t1.a) aa, COUNT(DISTINCT b) c
FROM t1 GROUP BY aa
returned wrong result, because the outer reference "t1.a"
in the subquery was substituted with the Item_ref item.
The Item_ref item obtains data from the result_field object
that refreshes once after the end of each group. This data
is not applicable to filesort since filesort() doesn't care
about groups (and doesn't update result_field objects with
copy_fields() and so on). Also that data is not applicable
to group separation algorithm: end_send_group() checks every
record with test_if_group_changed() that evaluates Item_ref
items, but it refreshes those Item_ref-s only after the end
of group, that is a vicious circle and the grouped column
values in the output are shifted.
Fix: if
a) we grouping by a subquery and
b) that subquery has outer references to FROM list
of the grouping query,
then we substitute these outer references with
Item_direct_ref like references under aggregate
functions: Item_direct_ref obtains data directly
from the current record.
2) The query with a non-trivial grouping expression like:
SELECT (SELECT t1.a) aa, COUNT(DISTINCT b) c
FROM t1 GROUP BY aa+0
also returned wrong result, since JOIN::exec() substitutes
references to top-level aliases in SELECT list with Item_copy
caching items. Item_copy items have same refreshing policy
as Item_ref items, so the whole groping expression with
Item_copy inside returns wrong result in filesort() and
end_send_group().
Fix: include aliased items into GROUP BY item tree instead
of Item_ref references to them.
logging is disabled
The server would hit an assertion because of a DBUG violation.
There was a missing DBUG_RETURN and instead a plain return
was used.
This patch replaces the return with DBUG_RETURN.
This was a deadlock between ALTER TABLE and another DML statement
(or LOCK TABLES ... READ). ALTER TABLE would wait trying to upgrade
its lock to MDL_EXCLUSIVE and the DML statement would wait trying
to acquire a TL_READ_NO_INSERT table level lock.
This could happen if one connection first acquired a MDL_SHARED_READ
lock on a table. In another connection ALTER TABLE is then started.
ALTER TABLE eventually blocks trying to upgrade to MDL_EXCLUSIVE,
but while holding a TL_WRITE_ALLOW_READ table level lock.
If the first connection then tries to acquire TL_READ_NO_INSERT,
it will block and we have a deadlock since neither connection can
proceed.
This patch fixes the problem by allowing TL_READ_NO_INSERT
locks to be granted if another connection holds TL_WRITE_ALLOW_READ
on the same table. This will allow the DML statement to proceed
such that it eventually can release its MDL lock which in turn
makes ALTER TABLE able to proceed.
Note that TL_READ_NO_INSERT was already partially compatible with
TL_WRITE_ALLOW_READ as the latter would be granted if the former
lock was held. This patch just makes the opposite true as well.
Also note that since ALTER TABLE takes an upgradable MDL lock,
there will be no starvation of ALTER TABLE statements by
statements acquiring TL_READ or TL_READ_NO_INSERT.
Test case added to lock_sync.test.
failed in open_ltable()
The problem was too restrictive asserts that enforced that
open_ltable() was called without any active HANDLERs, LOCK TABLES
or global read locks.
However, this can happen in several cases when opening system
tables. The assert would, for example, be triggered when drop
function was called from a connection with active HANDLERs as
this would cause open_ltable() to be called for mysql.proc.
The assert could also be triggered when using table-based
general log (mysql.general_log).
This patch removes the asserts since they will be triggered in
several legitimate cases and because the asserts are no longer
relevant due to changes in how locks are released.
The patch also fixes set_needs_thr_lock_abort() that before
ignored its parameter and always set the member variable to TRUE.
Test case added to mdl_sync.test.
Thanks to Dmitry Lenev for help with this bug!
m_tickets.front() == m_trans_sentinel'".
Debug build of server crashed due to assert failure in MDL
subsystem when one tried to execute multi-table REPAIR or
OPTIMIZE in autocommit=0 mode.
The assert failure occured when multi-table REPAIR or OPTIMIZE
started processing of second table from its table list and
tried to acquire upgradable metadata lock on this table.
The cause of the assert failure were MDL locks left over from
processing of previous table. It turned out that in autocommit=0
mode close_thread_tables() which happens at the end of table
processing doesn't release metadata locks.
This fix solves problem by releasing locks explicitly using
MDL_context::release_trans_locks() call.
fulltext search and row op.
The search for fulltext indexes is searching for some special
predicate layouts. While doing so it's not checking for the number
of columns of the expressions it tries to calculate.
And since row expressions can't return a single scalar value there
was a crash.
Fixed by checking if the expressions are scalar (in addition to
being constant) before calling Item::val_xxx() methods.
Fix Bug#50555 "handler commands crash server in my_hash_first()"
as a post-merge fix (the new handler tests are not passing
otherwise).
- in hash.c, don't call calc_hash if ! my_hash_inited().
- add tests and results for the test case for Bug#50555
