Date: Mon, 01 Nov 2010 15:15:25 -0000
3272 Roy Lyseng 2010-11-01
Bug#52068: Optimizer generates invalid semijoin materialization plan
When MaterializeScan semijoin strategy was used and there were one
or more outer dependent tables before the semijoin tables, the scan
over the materialized table was not properly reset for each row of
the prefix outer tables.
Example: suppose we have a join order:
ot1 SJ-Mat-Scan(it2 it3) ot4
Notice that this is called a MaterializeScan, even though there is an
outer table ahead of the materialized tables. Usually a MaterializeScan
has the outer tables after the materialized table, but this is
a special (but legal) case with outer dependent tables both before and
after the materialized table.
For each qualifying row from ot1, a new scan over the materialized
table must be set up. The code failed to do that, so all scans after
the first one returned zero rows from the materialized table.
This is a merge of percona patch mysqlbinlog_fix.patch.
Apparently, MySQL 5.0 does not flag BEGIN/SAVEPOINT/COMMIT/ROLLBACK statements
in its binlogs with LOG_EVENT_SUPPRESS_USE_F like 5.1+ does. This causes
unnecessary `use` statements around such statements when the binlog is dumped
by mysqlbinlog.
Fix by always suppressing the output of `use` for these statements.
other crashes
Some string manipulating SQL functions use a shared string object intended to
contain an immutable empty string. This object was used by the SQL function
SUBSTRING_INDEX() to return an empty string when one argument was of the wrong
datatype. If the string object was then modified by the sql function INSERT(),
undefined behavior ensued.
Fixed by instead modifying the string object representing the function's
result value whenever string manipulating SQL functions return an empty
string.
Relevant code has also been documented.
Changed handler.inc to make test row order independent
mysql-test/r/table_elim.result:
Updated test results for 5.3
mysql-test/suite/handler/aria.result:
Updated result
mysql-test/suite/handler/handler.inc:
Changed handler.inc to make test row order independent
mysql-test/suite/handler/heap.result:
Updated result
mysql-test/suite/handler/innodb.result:
Updated result
mysql-test/suite/handler/myisam.result:
Updated result
The test case fails with out of memory while updating a table
with several multi-megabytes sized rows. This can probably be too
exhausting for PB2 env.
The quick fix here is to reduce the size of the biggest
row (256MB) so that it becomes a little smaller (64MB).
mysql-test/suite/handler/interface.result:
Test case
mysql-test/suite/handler/interface.test:
Test case
sql/sql_prepare.cc:
Fix for LP#697622 Assertion `! is_set()' failed when preparing a HANDLER statement
Return TRUE if mysql_test_handler_read() fails.
mysql-test/suite/handler/innodb.result:
Added test case
mysql-test/suite/handler/innodb.test:
Added test case
sql/handler.h:
Move setting/resetting of active_index to ha_index_init()/ha_index_end() to simplify handler functions index_init()/index_end()
Fixed that get_index() returns MAX_KEY if index is not inited (this fixed LP#697610)
storage/federated/ha_federated.cc:
Settting of active_index is not needed anymore
storage/maria/ma_pagecache.c:
Added error message if we have too little memory for Maria page cache
INVOKER-security view access check wrong".
When privilege checks were done for tables used from an
INVOKER-security view which in its turn was used from
a DEFINER-security view connection's active security
context was incorrectly used instead of security context
with privileges of the second view's creator.
This meant that users which had enough rights to access
the DEFINER-security view and as result were supposed to
be able successfully access it were unable to do so in
cases when they didn't have privileges on underlying tables
of the INVOKER-security view.
This problem was caused by the fact that for INVOKER-security
views TABLE_LIST::security_ctx member for underlying tables
were set to 0 even in cases when particular view was used from
another DEFINER-security view. This meant that when checks of
privileges on these underlying tables was done in
setup_tables_and_check_access() active connection security
context was used instead of context corresponding to the
creator of caller view.
This fix addresses the problem by ensuring that underlying
tables of an INVOKER-security view inherit security context
from the view and thus correct security context is used for
privilege checks on underlying tables in cases when such view
is used from another view with DEFINER-security.
mysql-test/r/view_grant.result:
Added coverage for various combinations of DEFINER and
INVOKER-security views, including test for bug #58499
"DEFINER-security view selecting from INVOKER-security
view access check wrong".
mysql-test/t/view_grant.test:
Added coverage for various combinations of DEFINER and
INVOKER-security views, including test for bug #58499
"DEFINER-security view selecting from INVOKER-security
view access check wrong".
sql/sql_view.cc:
When opening a non-suid view ensure that its underlying
tables will get the same security context as use for
checking privileges on the view, i.e. security context
of view invoker. This context can be different from the
security context which is currently active for connection
in cases when this non-suid view is used from a view with
suid security. Inheriting security context in such situation
allows correctly apply privileges of creator of suid view
in checks for tables of non-suid view (since in this
situation creator/definer of suid view serves as invoker
for non-suid view).
Item_func_spatial_collection::fix_length_and_dec didn't call parent's method, so
the maybe_null was set to '0' after it. But in this case the result was
just NULL, that caused wrong behaviour.
per-file comments:
mysql-test/r/gis.result
Bug #57321 crashes and valgrind errors from spatial types
test result updated.
mysql-test/t/gis.test
Bug #57321 crashes and valgrind errors from spatial types
test case added.
sql/item_geofunc.h
Bug #57321 crashes and valgrind errors from spatial types
Item_func_geometry::fix_length_and_dec() called in
Item_func_spatial_collection::fix_length_and_dec().
- Cause: handler::in_range_check_pushed_down was not reset when a
command would call handler->idx_cond_push() without later calling
handler->index_end().
- Fix: reset the variable in handler->reset(), too (like we do with other
Index Condition Pushdown members).
get_year_value() contains code to convert 2-digits year to
4-digits. The fix for Bug#49910 added a check on the size of
the underlying field so that this conversion is not done for
YEAR(4) values. (Since otherwise one would convert invalid
YEAR(4) values to valid ones.)
The existing check does not work when Item_cache is used, since
it is not detected when the cache is based on a Field. The
reported change in behavior is due to Bug#58030 which added
extra cached items in min/max computations.
The elegant solution would be to implement
Item_cache::real_item() to return the underlying Item.
However, some side effects are observed (change in explain
output) that indicates that such a change is not straight-
forward, and definitely not appropriate for an MRU.
Instead, a Item_cache::field() method has been added in order
to get access to the underlying field. (This field() method
eliminates the need for Item_cache::eq_def() used in
test_if_ref(), but in order to limit the scope of this fix,
that code has been left as is.)
mysql-test/r/type_year.result:
Added test case for Bug#59211.
mysql-test/t/type_year.test:
Added test case for Bug#59211.
sql/item.h:
Added function Item_cache::field() to get access to the
underlying Field of a cached field Value.
sql/item_cmpfunc.cc:
Also check underlying fields of Item_cache, not just Item_Field,
when checking whether the value is of type YEAR(4) or not.
tmptable needed
The function DEFAULT() works by modifying the the data buffer pointers (often
referred to as 'record' or 'table record') of its argument. This modification
is done during name resolution (fix_fields().) Unfortunately, the same
modification is done when creating a temporary table, because default values
need to propagate to the new table.
Fixed by skipping the pointer modification for fields that are arguments to
the DEFAULT function.
Added key and file version numbers to MEMORY tables so that we can detect if someone has changed them between HANDLER calls.
mysql-test/suite/handler/aria.result:
Fixed result after test changes
mysql-test/suite/handler/handler.inc:
Changed test to use combined key to ensure rows are returned in a pre-determinated order.
mysql-test/suite/handler/heap.result:
New result
mysql-test/suite/handler/heap.test:
Added test for HANDLER + HEAP
mysql-test/suite/handler/innodb.result:
Fixed result after test changes
mysql-test/suite/handler/myisam.result:
Fixed result after test changes
sql/sql_handler.cc:
Fixed wrong parameter to ha_index_next_same()
storage/heap/ha_heap.cc:
Abort key scan if table has changed.
Abort table scan if table has been recreated.
storage/heap/ha_heap.h:
Added support for HANDLER
storage/heap/hp_clear.c:
Increase version number so that we can notice changes if using HANDLER
storage/heap/hp_delete.c:
Increase key data version number on key changes.
storage/heap/hp_rfirst.c:
Remember version of key data
Give error if using read-first on hash key.
storage/heap/hp_rkey.c:
Remember version of key data
storage/heap/hp_rlast.c:
Remember version of key data
Give error if using read-last on hash key.
storage/heap/hp_rnext.c:
Fixed that we get next key from last search.
storage/heap/hp_rprev.c:
Fixed that we get previous key from last search.
storage/heap/hp_scan.c:
Remember version of key and file data
storage/heap/hp_update.c:
Increase key data version number on key changes.
storage/heap/hp_write.c:
Increase key data version number on key changes.
- Fix for MySQL BUG#52357 added NESTED_JOIN::is_fully_covered() which would
not take into account that MariaDB's table elimination could eliminate tables
from join plan (and so, from join nest).
Fixed the check in the function to compare post-table-elimination numbers.
- Added test case for Aria
- Tested HANDLER with HEAP (changes to HEAP code will be pushed in 5.3)
- Moved all HANDLER test to suite/handler.
mysql-test/Makefile.am:
Added suite/handler
mysql-test/mysql-test-run.pl:
Added suite/handler
mysql-test/r/lock_multi.result:
Remove test that is already in handler test suite
mysql-test/suite/handler/aria.result:
Test for HANDLER with Aria storage engine
mysql-test/suite/handler/aria.test:
Test for HANDLER with Aria storage engine
mysql-test/suite/handler/handler.inc:
Extended the general handler test
Moved interface testing to 'interface.test'
mysql-test/suite/handler/init.inc:
Common init for handler tests.
mysql-test/suite/handler/innodb.result:
New results
mysql-test/suite/handler/innodb.test:
Update to use new include files
mysql-test/suite/handler/interface.result:
Test of HANDLER interface (not storage engine dependent parts)
mysql-test/suite/handler/interface.test:
Test of HANDLER interface (not storage engine dependent parts)
mysql-test/suite/handler/myisam.result:
New results
mysql-test/suite/handler/myisam.test:
Update to use new include files
mysql-test/t/lock_multi.test:
Remove test that is already in handler test suite
mysys/tree.c:
Added missing handling of read previous (showed up in HEAP testing)
sql/handler.cc:
Don't marka 'HA_ERR_RECORD_CHANGED' as fatal (can be used with HANDLER READ, especially with MEMORY ENGINE)
sql/handler.h:
Added prototype for can_continue_handler_scan()
sql/sql_handler.cc:
Re-initialize search if we switch from key to table search.
Check if handler can continue searching between calls (via can_continue_handler_scan())
Don't write common not fatal errors to log
storage/maria/ma_extra.c:
Don't set index 0 as default. This forces call to ma_check_index() to set up index variables.
storage/maria/ma_ft_boolean_search.c:
Ensure that info->last_key.keyinfo is set
storage/maria/ma_open.c:
Don't set index 0 as default. This forces call to ma_check_index() to set up index variables.
storage/maria/ma_rkey.c:
Trivial optimization
storage/maria/ma_rnext.c:
Added missing code from mi_rnext.c to ensure that handler next/prev works.
storage/maria/ma_rsame.c:
Simple optimizations
storage/maria/ma_search.c:
Initialize info->last_key once and for all when we change keys.
storage/maria/ma_unique.c:
Ensure that info->last_key.keyinfo is up to date.
multiple columns in the partition key
ndb crash if duplicate columns in the partitioning key.
Backport from mysql-5.1-telco-7.0, see bug#53354.
Changed from case sensitive field name comparision
to non case sensitive too.
mysql-test/r/partition_error.result:
updated result
mysql-test/t/partition_error.test:
Added test for the error in non-ndb partitioned table.
sql/sql_partition.cc:
Added check for duplicated field names in the
partitioning key.
This patch adds options to annotate the binlog (and the mysqlbinlog
output) with the original SQL query for queries that are logged
using row-based replication.
This assert could be triggered if -1 was inserted into
an auto increment column by a statement writing more than
one row.
Unless explicitly given, an interval of auto increment values
is generated when a statement first needs an auto increment
value. The triggered assert checks that the auto increment
counter is equal to or higher than the lower bound of this
interval.
Generally, the auto increment counter starts at 1 and is
incremented by 1 each time it is used. However, inserting an
explicit value into the auto increment column, sets the auto
increment counter to this value + 1 if this value is higher
than the current value of the auto increment counter.
This bug was triggered if the explicit value was -1. Since the
value was converted to unsigned before any comparisons were made,
it was found to be higher than the current vale of the auto
increment counter and the counter was set to -1 + 1. This value
was below the reserved interval and caused the assert to be
triggered the next time the statement tried to write a row.
With the patch for Bug#39828, this bug is no longer repeatable.
Now, -1 + 1 is detected as an "overflow" which causes the auto
increment counter to be set to ULONGLONG_MAX. This avoids hitting
the assert for the next insert and causes a new interval of
auto increment values to be generated. This resolves the issue.
This patch therefore only contains a regression test and no code
changes. Test case added to auto_increment.test.
