corruption errors: 126,134,145
When one thread attempts to lock two (or more) tables and another
thread executes statement that aborts these locks (e.g. REPAIR
TABLE) we may get a table object with wrong lock type in a table
cache.
For example if SELECT FROM t1,t2 was aborted, subsequent INSERT
INTO t1 may be executed under read lock.
As a result we may get various table corruptions and even a server
crash.
This is fixed by resetting lock type in case lock was aborted by
another thread.
I failed to create reasonable test case for this bug.
sql/lock.cc:
If thr_multi_lock was aborted by another thread, it unlocks tables
that were locked before one that was aborted. Lock type for tables
that were after a table that was aborted preserved. Thus we need
to reset lock data in case thr_multi_lock was aborted.
for a query over an empty table right after its creation.
The crash is the result of an attempt made by JOIN::optimize to evaluate
the WHERE condition when no records have been actually read.
The added test case can reproduce the crash only with InnoDB tables and
only with 5.0.x.
mysql-test/r/innodb_mysql.result:
Added a test case for bug #28272.
mysql-test/t/innodb_mysql.test:
Added a test case for bug #28272.
sql/sql_select.cc:
Fixed bug #28272: crash that occurs when running an EXPLAIN command
for a query over an empty table right after its creation.
The crash is the result of an attempt made by JOIN::optimize to evaluate
the WHERE condition when no records have been actually read.
Such attempts could be observed only with EXPLAIN commands.
Now at the optimization phase the WHERE condition is never evaluated if
there is no record previously read from the table.
statement from a UNION query with ORDER BY an expression containing
RAND().
The crash happened because the global order by list in the union query
was not re-initialized for execution.
(Local order by lists were re-initialized though).
mysql-test/r/ps.result:
Added a test case for bug #27937.
mysql-test/t/ps.test:
Added a test case for bug #27937.
sql/sql_union.cc:
Fixed bug #27937: crash for the the second execution of a prepared
statement from a UNION query with ORDER BY an expression containing
RAND().
The crash happened because the global order by list in the union query
was not re-initialized for execution.
(Local order by lists were re-initialized though).
Added re-initialization of the global order by list in the function
st_select_lex_unit::init_prepare_fake_select_lex.
- Queries in the query cache are identified by the individual
characters in the query statement, the current database and
the current environment expressed as a set of system variable
flags.
- Since the set of environment flags didn't properly describe the
current environment unexpected results were returned from the
query cache.
- Query cache is now cleared when the variable ft_boolean_syntax is
updated.
- An identification flag for the variable default_week_format is
added to the query cache record.
Thanks to Martin Friebe who has supplied significant parts of this patch.
mysql-test/r/query_cache.result:
Added test case
mysql-test/t/query_cache.test:
Added test case
sql/mysql_priv.h:
- Added missing flags to reflect the significant local environment
settings of the cached query.
sql/set_var.cc:
- Added query cache flush when system variable ft_boolean_syntax is
updated since this also invalidates all cached result sets using this
variable.
sql/sql_cache.cc:
- Added another local system variable as an identification flag
for cached queries.
Warnings may show up in the error log that complain about a table, yet
don't tell us which table.
Add the database and table names to the message.
sql/ha_myisam.cc:
Add the database and table names to the warning message.
When checking for applicability of join cache
we must disable its usage only if there is no
temp table in use.
When a temp table is used we can use join
cache (and it will not make the result-set
unordered) to fill the temp table. The filesort()
operation is then applied to the data in the temp
table and hence is not affected by join cache
usage.
Fixed by narrowing the condition for disabling
join cache to exclude the case where temp table
is used.
mysql-test/r/join.result:
Bug #27531: test case
mysql-test/t/join.test:
Bug #27531: test case
sql/sql_select.cc:
Bug #27531:
Disable join cache only if not using temp table
The generic string to int conversion was used by the Item_func_signed and
the Item_func_unsigned classes to convert DATE/DATETIME values to the
SIGNED/UNSIGNED type. But this conversion produces wrong results for such
values.
Now if the item which result has to be converted can return its result as
longlong then the item->val_int() method is used to allow the item to carry
out the conversion itself and return the correct result.
This condition is checked in the Item_func_signed::val_int() and the
Item_func_unsigned::val_int() functions.
mysql-test/t/cast.test:
Added a test case for the bug#23656: Wrong conversion result of a DATETIME to integer using CAST function.
mysql-test/r/cast.result:
Added a test case for the bug#23656: Wrong conversion result of a DATETIME to integer using CAST function.
sql/item_func.cc:
Bug#23656: Wrong conversion result of a DATETIME to integer using CAST function.
Now if the item which result has to be converted can return its result as
longlong then the item->val_int() method is used to allow the item to carry
out the conversion itself and return the correct result.
This condition is checked in the Item_func_signed::val_int() and the
Item_func_unsigned::val_int() functions.
some rollup rows (rows with NULLs for grouping attributes) if GROUP BY
list contained constant expressions.
This happened because the results of constant expressions were not put
in the temporary table used for duplicate elimination. In fact a constant
item from the GROUP BY list of a ROLLUP query can be replaced for an
Item_null_result object when a rollup row is produced .
Now the JOIN::rollup_init function wraps any constant item referenced in
the GROYP BY list of a ROLLUP query into an Item_func object of a special
class that is never detected as constant item. This ensures creation of
fields for such constant items in temporary tables and guarantees right
results when the result of the rollup operation first has to be written
into a temporary table, e.g. in the cases when duplicate elimination is
required.
mysql-test/r/olap.result:
Added a test case for bug #24856.
mysql-test/t/olap.test:
Added a test case for bug #24856.
sql/item_func.h:
Fixed bug #24856: the result set of a ROLLUP query with DISTINCT could lack
some rollup rows (rows with NULLs for grouping attributes) if GROUP BY
list contained constant expressions.
Itroduced class Item_func_rollup_const derived from Item_func. The object of
this class are never detected as constant items.
We use them for wrapping constant items from the GROUP BY list of any ROLLUP
query. This wrapping allows us to ensure writing constant items into temporary
tables whenever the result of the ROLLUP operation has to be written into a
temporary table, e.g. when ROLLUP is used together with DISTINCT in the SELECT
list.
sql/sql_select.cc:
Fixed bug #24856: the result set of a ROLLUP query with DISTINCT could lack
some rollup rows (rows with NULLs for grouping attributes) if GROUP BY
list contained constant expressions.
Now the JOIN::rollup_init function wraps any constant item referenced in
the GROYP BY list of a ROLLUP query into an Item_func object of a special
class that is never detected as constant item. This ensures creation of
fields for such constant items in temporary tables and guarantees right
results when the result of the rollup operation first has to be written
into a temporary table, e.g. in the cases when duplicate elimination is
required.
INSERT...ON DUPLICATE KEY UPDATE may cause error 1032:
"Can't find record in ..." if we are inserting into
InnoDB table unique index of partial key with
underlying UTF-8 string field.
This error occurs because INSERT...ON DUPLICATE uses a wrong
procedure to copy string fields of multi-byte character sets
for index search.
mysql-test/t/innodb_mysql.test:
Added test case for bug #13191.
mysql-test/r/innodb_mysql.result:
Added test case for bug #13191.
sql/field.h:
Fixed bug #13191.
Field_string::get_key_image() virtual function was overloaded
to implement copying of variable length character (UTF-8) fields.
Field::get_key_image() function prototype has been changed to
return byte size of copied data.
sql/field.cc:
Fixed bug #13191.
Field_string::get_key_image() virtual function was overloaded
to implement copying of variable length character (UTF-8) fields.
Field::get_key_image() function prototype has been changed to
return byte size of copied data.
sql/key.cc:
Fixed bug #13191.
INSERT...ON DUPLICATE KEY UPDATE may cause error 1032:
"Can't find record in ...".
This error occurs because INSERT...ON DUPLICATE uses
a wrong procedure to copy field parts for index search.
key_copy() function has been fixed.
on a BLACKHOLE table
Using INSERT DELAYED on BLACKHOLE tables could lead to server
crash.
This happens because delayed thread wants to upgrade a lock,
but BLACKHOLE tables do not have locks at all.
This patch rejects attempts to use INSERT DELAYED on MERGE
tables.
mysql-test/r/blackhole.result:
A test case for BUG#27998.
mysql-test/t/blackhole.test:
A test case for BUG#27998.
sql/ha_blackhole.h:
Removed HA_CAN_INSERT_DELAYED flag from table_flags().
The insert delayed thread upgrades the lock. Hence it is
incapable to handle BLACKHOLE tables, which do not have
locks at all.
Support for NULL components was incomplete for row comparison,
fixed. Added support for abort_on_null at compare_row() like
in 5.x
sql/item_cmpfunc.h:
Bug#27704: incorrect comparison of rows with NULL components
Added support for abort_on_null at Item_bool_func2
like in 5.x
sql/item_cmpfunc.cc:
Bug#27704: incorrect comparison of rows with NULL components
Support for NULL components was incomplete for row comparison,
fixed. Added support for abort_on_null at compare_row() like
in 5.x
mysql-test/t/row.test:
Test case updated for Bug#27704 (incorrect comparison
of rows with NULL components)
mysql-test/r/row.result:
Test case updated for Bug#27704 (incorrect comparison
of rows with NULL components)
mysql-test/r/subselect.result:
Test case updated for Bug#27704 (incorrect comparison
of rows with NULL components)
Added missed DROP privilege check on the original table for RENAME TABLE command.
mysql-test/r/grant.result:
Fix for bug #27515: DROP previlege is not required anymore for RENAME TABLE
- test result.
mysql-test/t/grant.test:
Fix for bug #27515: DROP previlege is not required anymore for RENAME TABLE
- test case.
sql/sql_parse.cc:
Fix for bug #27515: DROP previlege is not required anymore for RENAME TABLE
- added DROP privilege check on the original table for RENAME TABLE command.
IGNORE/USE/FORCE INDEX hints were honored when choosing FULLTEXT
index.
With this fix these hints are ignored. For regular indexes we may
perform table scan instead of index lookup when IGNORE INDEX was
specified. We cannot do this for FULLTEXT in NLQ mode.
mysql-test/r/fulltext.result:
A test case for bug#25951.
mysql-test/t/fulltext.test:
A test case for bug#25951.
sql/item_func.cc:
IGNOR/USE/FORCE INDEX hints should not be honored when choosing FULLTEXT
index.
Use proper bitmap, that is not modified by IGNORE/USE/FORCE INDEX hints.
are used as arguments of the IN predicate.
Added a function to check compatibility of row expressions. Made sure that this
function to be called for Item_func_in objects by fix_length_and_dec().
mysql-test/r/row.result:
Added a test case for bug #27484.
mysql-test/t/row.test:
Added a test case for bug #27484.
IN/BETWEEN predicates in sorting expressions.
Wrong results may occur when the select list contains an expression
with IN/BETWEEN predicate that differs from a sorting expression by
an additional NOT only.
Added the method Item_func_opt_neg::eq to compare correctly expressions
containing [NOT] IN/BETWEEN.
The eq method inherited from the Item_func returns TRUE when comparing
'a IN (1,2)' with 'a NOT IN (1,2)' that is not, of course, correct.
mysql-test/r/order_by.result:
Added a test case for bug #27532.
mysql-test/t/order_by.test:
Added a test case for bug #27532.
sql/item_cmpfunc.cc:
Fixed bug #27532.
Added the method Item_func_opt_neg::eq to compare correctly expressions
containing [NOT] IN/BETWEEN.
The eq method inherited from the Item_func returns TRUE when comparing
'a IN (1,2)' with 'a NOT IN (1,2)' that is not, of course, correct.
sql/item_cmpfunc.h:
Added the method Item_func_opt_neg::eq to compare correctly expressions
containing [NOT] IN/BETWEEN.
The eq method inherited from the Item_func returns TRUE when comparing
'a IN (1,2)' with 'a NOT IN (1,2)' that is not, of course, correct.
Pushbuild fixes:
- Make MAX_SEL_ARGS smaller (even 16K records_in_range() calls is
more than it makes sense to do in typical cases)
- Don't call sel_arg->test_use_count() if we've already allocated
more than MAX_SEL_ARGs elements. The test will succeed but will take
too much time for the test suite (and not provide much value).
mysql-test/r/range.result:
BUG#26624: high mem usage (crash) in range optimizer
Pushbuild fixes: make the test go faster
mysql-test/t/range.test:
BUG#26624: high mem usage (crash) in range optimizer
Pushbuild fixes: make the test go faster
- GRANT and REVOKE statments didn't have the "updating" flag set and
thus statements with a table specified would not replicate if
slave filtering rules where turned on.
For example "GRANT ... ON test.t1 TO ..." would not replicate.
mysql-test/r/rpl_ignore_table.result:
Add test results
mysql-test/t/rpl_ignore_table.test:
Add tests
sql/sql_yacc.yy:
Pass option TL_OPTION_UPDATING to 'add_table_to_list' when parsing a
GRANT or REVOKE and a table specifier is found. This will set the
property "updating" on the table and thus the slave filtering rules will
be applied.
Without setting updating the statement will be not
replicated - since "it's not updating anything" - an optimization
to quickly skip SELECT's and similar.
Thanks to Martin Friebe for finding and submitting a fix for this bug!
A table with maximum number of key segments and maximum length key name
would have a corrupted .frm file, due to an incorrect calculation of the
complete key length. Now the key length is computed correctly (I hope) :-)
MyISAM would reject a table with the maximum number of keys and the maximum
number of key segments in all keys. It would allow one less than this total
maximum. Now MyISAM accepts a table defined with the maximum. (This is a
very minor issue.)
myisam/mi_open.c:
change >= to > in a comparison (i.e., error only if key_parts_in_table
really is greater than MAX_KEY * MAX_KEY_SEG)
mysql-test/r/create.result:
Add test results for bug #26642 (create index corrupts table definition in .frm)
mysql-test/t/create.test:
Add test case for bug #26642 (create index corrupts table definition in .frm)
sql/table.cc:
In create_frm(), fix formula for key_length; it was too small by (keys * 2) bytes
- Added PARAM::alloced_sel_args where we count the # of SEL_ARGs
created by SEL_ARG tree cloning operations.
- Made the range analyzer to shortcut and not do any more cloning
if we've already created MAX_SEL_ARGS SEL_ARG objects in cloning.
- Added comments about space complexity of SEL_ARG-graph
representation.
mysql-test/r/range.result:
BUG#26624: Testcase
mysql-test/t/range.test:
BUG#26624: Testcase
- Define Sql_alloc::operator new() as thow() so that C++ compiler
handles NULL return values
(there is no testcase as there is no portable way to set limit on the
amount of memory that a process can allocate)
sql/sql_list.h:
BUG#26625: crash in range optimizer (out of mem)
- Define Sql_alloc::operator new() as thow() so that C++ compiler
handles NULL return values
causes incorrect duplicate entries
Keys for BTREE indexes on ENUM and SET columns of MEMORY tables
with character set UTF8 were computed incorrectly. Many
different column values got the same key value.
Apart of possible performance problems, it made unique indexes
of this type unusable because it rejected many different
values as duplicates.
The problem was that multibyte character detection was tried
on the internal numeric column value. Many values were not
identified as characters. Their key value became blank filled.
Thanks to Alexander Barkov and Ramil Kalimullin for the patch,
which sets the character set of ENUM and SET key segments to
the pseudo binary character set.
mysql-test/r/heap_btree.result:
Bug#24985 - UTF8 ENUM primary key on MEMORY using BTREE
causes incorrect duplicate entries
Added test result.
mysql-test/t/heap_btree.test:
Bug#24985 - UTF8 ENUM primary key on MEMORY using BTREE
causes incorrect duplicate entries
Added test.
sql/ha_heap.cc:
Bug#24985 - UTF8 ENUM primary key on MEMORY using BTREE
causes incorrect duplicate entries
Set key segment charset to my_charset_bin for ENUM and SET
columns.
to 0 causes wrong (large) length to be read
from the row in _mi_calc_blob_length() when
storing NULL values in (e.g) POINT columns.
This large length is then used to allocate
a block of memory that (on some OSes) causes
trouble.
Fixed by calling the base class's
Field_blob::reset() from Field_geom::reset()
that is called when storing a NULL value into
the column.
mysql-test/r/gis.result:
Bug #27164: test case
mysql-test/t/gis.test:
Bug #27164: test case
sql/field.h:
Bug #27164: not reseting the data pointer
to 0 causes wrong (large) length to be read
from the row in _mi_calc_blob_length() when
storing NULL values in (e.g) POINT columns.
This large length is then used to allocate
a block of memory that (on some OSes) causes
trouble.
fixed differently: wake up select_thread with THR_SERVER_ALARM instead
mysys/thr_alarm.c:
reverted linuxthreads thr_client_alarm fix (not future-proof)
(in thr_alarm.cc it happened too late).
mysys/thr_alarm.c:
move thr_client_alarm initialization to mysqld.cc
(here it happened too late)
sql/mysqld.cc:
move thr_client_alarm initialization to mysqld.cc
(in thr_alarm.cc it happened too late).
moved thr_kill_signal initialization to init_signals()
