MDEV-8938 Server Crash on Update with joins
Make unique table check after setup_fields of update because unique table can materialize table and we do not need field resolving after materialization.
Make unique table check after setup_fields of update because unique table can materialize table and we do not need field resolving after materialization.
INCORRECT RESULTS
Issue:
-----
Updating varchar and text fields in the same update
statement can produce incorrect results. When a varchar
field is assigned to the text field and the varchar field
is then set to a different value, the text field's result
contains the varchar field's new value.
SOLUTION:
---------
Currently the blob type does not allocate space for the
string to be stored. Instead it contains a pointer to the
varchar string. So when the varchar field is changed as
part of the update statement, the value contained in the
blob also changes.
The fix would be to actually store the value by allocating
space for the blob's string. We can avoid allocating this
space when the varchar field is not being written into.
- If an UPDATE 1) modifies the key it is using, and 2) has ORDER BY ... LIMIT
which matches the key it is using, Then we should use "Using buffer", not
"Using filesort".
from MariaDB 10.0.
The bug in mdev-3948 was an instance of the problem fixed by Sergey's patch
in 10.0 - namely that the range optimizer could change table->[read | write]_set,
and not restore it.
revno: 3471
committer: Sergey Petrunya <psergey@askmonty.org>
branch nick: 10.0-serg-fix-imerge
timestamp: Sat 2012-11-03 12:24:36 +0400
message:
# MDEV-3817: Wrong result with index_merge+index_merge_intersection, InnoDB table, join, AND and OR conditions
Reconcile the fixes from:
#
# guilhem.bichot@oracle.com-20110805143029-ywrzuz15uzgontr0
# Fix for BUG#12698916 - "JOIN QUERY GIVES WRONG RESULT AT 2ND EXEC. OR
# AFTER FLUSH TABLES [-INT VS NULL]"
#
# guilhem.bichot@oracle.com-20111209150650-tzx3ldzxe1yfwji6
# Fix for BUG#12912171 - ASSERTION FAILED: QUICK->HEAD->READ_SET == SAVE_READ_SET
# and
#
and related fixes from: BUG#1006164, MDEV-376:
Now, ROR-merged QUICK_RANGE_SELECT objects make no assumptions about the values
of table->read_set and table->write_set.
Each QUICK_ROR_SELECT has (and had before) its own column bitmap, but now, all
QUICK_ROR_SELECT's functions that care: reset(), init_ror_merged_scan(), and
get_next() will set table->read_set when invoked and restore it back to what
it was before the call before they return.
This allows to avoid the mess when somebody else modifies table->read_set for
some reason.
Reconcile the fixes from:
#
# guilhem.bichot@oracle.com-20110805143029-ywrzuz15uzgontr0
# Fix for BUG#12698916 - "JOIN QUERY GIVES WRONG RESULT AT 2ND EXEC. OR
# AFTER FLUSH TABLES [-INT VS NULL]"
#
# guilhem.bichot@oracle.com-20111209150650-tzx3ldzxe1yfwji6
# Fix for BUG#12912171 - ASSERTION FAILED: QUICK->HEAD->READ_SET == SAVE_READ_SET
# and
#
and related fixes from: BUG#1006164, MDEV-376:
Now, ROR-merged QUICK_RANGE_SELECT objects make no assumptions about the values
of table->read_set and table->write_set.
Each QUICK_ROR_SELECT has (and had before) its own column bitmap, but now, all
QUICK_ROR_SELECT's functions that care: reset(), init_ror_merged_scan(), and
get_next() will set table->read_set when invoked and restore it back to what
it was before the call before they return.
This allows to avoid the mess when somebody else modifies table->read_set for
some reason.
Before this fix, the ha_read_last_count status variable was defined and
updated internally, for never exposed as a system variable.
This fix exposes the system variable as "Handler_read_last",
for completness of the Handler_read_* system variables interface.
Adjusted tests results accordingly.
Before this fix, the ha_read_last_count status variable was defined and
updated internally, for never exposed as a system variable.
This fix exposes the system variable as "Handler_read_last",
for completness of the Handler_read_* system variables interface.
Adjusted tests results accordingly.
This assert checks that the server does not try to send OK to the
client if there has been some error during processing. This is done
to make sure that the error is in fact sent to the client.
The problem was that view errors during processing of WHERE conditions
in UPDATE statements where not detected by the update code. It therefore
tried to send OK to the client, triggering the assert.
The bug was only noticeable in debug builds.
This patch fixes the problem by making sure that the update code
checks for errors during condition processing and acts accordingly.
This assert checks that the server does not try to send OK to the
client if there has been some error during processing. This is done
to make sure that the error is in fact sent to the client.
The problem was that view errors during processing of WHERE conditions
in UPDATE statements where not detected by the update code. It therefore
tried to send OK to the client, triggering the assert.
The bug was only noticeable in debug builds.
This patch fixes the problem by making sure that the update code
checks for errors during condition processing and acts accordingly.
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()).
mysql-test/r/log_state.result:
Updated result for optimized query, bug #36569.
mysql-test/r/single_delete_update.result:
Test case for bug #30584, bug #36569 and bug #53742.
mysql-test/r/update.result:
Updated result for optimized query, bug #30584.
Note:
"Handler_read_last 1" omitted, see bug 52312:
lost Handler_read_last status variable.
mysql-test/t/single_delete_update.test:
Test case for bug #30584, bug #36569 and bug #53742.
sql/opt_range.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
* get_index_for_order() has been rewritten entirely and moved
to sql_select.cc
New QUICK_RANGE_SELECT::make_reverse method has been added.
sql/opt_range.h:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
* get_index_for_order() has been rewritten entirely and moved
to sql_select.cc
New functions:
* QUICK_SELECT_I::make_reverse()
* SQL_SELECT::set_quick()
sql/records.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
* init_read_record_idx() has been modified to allow reverse index scan
New functions:
* rr_index_last()
* rr_index_desc()
sql/records.h:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
init_read_record_idx() has been modified to allow reverse index scan
sql/sql_delete.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
mysql_delete: an optimization has been added to skip
unnecessary sorting with ORDER BY clause where select
result ordering is acceptable.
sql/sql_select.cc:
Bug #30584, bug #36569, bug #53737, bug #53742:
UPDATE/DELETE ... WHERE ... ORDER BY... always does a filesort
even if not required
The const_expression_in_where function has been modified
to accept both Item and Field pointers.
New functions:
* get_index_for_order()
* test_if_cheaper_ordering() has been extracted from
test_if_skip_sort_order() to share with get_index_for_order()
* simple_remove_const()
sql/sql_select.h:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
New functions:
* test_if_cheaper_ordering()
* simple_remove_const()
* get_index_for_order()
sql/sql_update.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
mysql_update: an optimization has been added to skip
unnecessary sorting with ORDER BY clause where a select
result ordering is acceptable.
sql/table.cc:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
New functions:
* TABLE::update_const_key_parts()
* is_simple_order()
sql/table.h:
Bug #30584, bug #36569: UPDATE/DELETE ... WHERE ... ORDER BY...
always does a filesort even if not required
New functions:
* TABLE::update_const_key_parts()
* is_simple_order()
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()).
This assertion would occur if UPDATE was used to update multiple
tables containing an AUTO_INCREMENT column and if the inserted
row had a user-supplied value for that column. The assertion
could then be triggered by the next statement.
The problem was only noticeable on debug builds of the server.
The cause of the problem was that the code for multi update did
not properly reset the TABLE->auto_increment_if_null flag after update.
The flag is used to indicate that a non-null value of an auto_increment field
has been provided by the user or retrieved from a current record.
Open_tables() contains an assertion that tests this flag, and this
was triggered in this case by ALTER TABLE.
This patch fixes the problem by resetting the auto_increment_if_null
field to FALSE once a row has been updated.
This bug is similar to Bug#47274, but for multi update rather
than INSERT DELAYED.
Test case added to update.test.
This assertion would occur if UPDATE was used to update multiple
tables containing an AUTO_INCREMENT column and if the inserted
row had a user-supplied value for that column. The assertion
could then be triggered by the next statement.
The problem was only noticeable on debug builds of the server.
The cause of the problem was that the code for multi update did
not properly reset the TABLE->auto_increment_if_null flag after update.
The flag is used to indicate that a non-null value of an auto_increment field
has been provided by the user or retrieved from a current record.
Open_tables() contains an assertion that tests this flag, and this
was triggered in this case by ALTER TABLE.
This patch fixes the problem by resetting the auto_increment_if_null
field to FALSE once a row has been updated.
This bug is similar to Bug#47274, but for multi update rather
than INSERT DELAYED.
Test case added to update.test.
leads to an assertion failure
Any run-time error in stored function (like recursive function
call or update of table that is already updating by statement
which invoked this stored function etc.) that was used in some
expression of the single-table UPDATE statement caused an
assertion failure.
Multiple-table UPDATE (as well as INSERT and both single- and
multiple-table DELETE) are not affected.
mysql-test/r/update.result:
Added test case for bug #40745.
mysql-test/t/update.test:
Added test case for bug #40745.
sql/sql_update.cc:
Bug #40745: Error during WHERE clause calculation in UPDATE
leads to an assertion failure
The mysql_update function has been updated to take into account
the status of invoked stored functions before setting the status
of whole UPDATE query to OK.
leads to an assertion failure
Any run-time error in stored function (like recursive function
call or update of table that is already updating by statement
which invoked this stored function etc.) that was used in some
expression of the single-table UPDATE statement caused an
assertion failure.
Multiple-table UPDATE (as well as INSERT and both single- and
multiple-table DELETE) are not affected.
In multi_update::send_data(), the counter of matched rows was not correctly incremented, when during insertion of a new row to a temporay table it had to be converted from HEAP to MyISAM.
This fix changes the logic to increment the counter of matched rows in the following cases:
1. If the error returned from write_row() is zero.
2. If the error returned from write_row() is non-zero, is neither HA_ERR_FOUND_DUPP_KEY nor HA_ERR_FOUND_DUPP_UNIQUE, and a call to create_myisam_from_heap() succeeds.
mysql-test/r/update.result:
Added a test case for bug #22364 "Inconsistent "matched rows" when executing UPDATE"
mysql-test/t/update.test:
Added a test case for bug #22364 "Inconsistent "matched rows" when executing UPDATE"
sql/sql_update.cc:
In multi_update::send_data(), the counter of matched rows was not correctly incremented, when during insertion of a new row to a temporay table it had to be converted from HEAP to MyISAM.
This fix changes the logic to increment the counter of matched rows in the following cases:
1. If the error returned from write_row() is zero.
2. If the error returned from write_row() is non-zero, is neither HA_ERR_FOUND_DUPP_KEY nor HA_ERR_FOUND_DUPP_UNIQUE, and a call to create_myisam_from_heap() succeeds.
In multi_update::send_data(), the counter of matched rows was not correctly incremented, when during insertion of a new row to a temporay table it had to be converted from HEAP to MyISAM.
This fix changes the logic to increment the counter of matched rows in the following cases:
1. If the error returned from write_row() is zero.
2. If the error returned from write_row() is non-zero, is neither HA_ERR_FOUND_DUPP_KEY nor HA_ERR_FOUND_DUPP_UNIQUE, and a call to create_myisam_from_heap() succeeds.
into mysql.com:/home/hf/work/mrg/mysql-5.1-opt
BUILD/check-cpu:
Auto merged
mysql-test/r/explain.result:
Auto merged
mysql-test/r/func_str.result:
Auto merged
mysql-test/r/func_test.result:
Auto merged
mysql-test/r/subselect.result:
Auto merged
mysql-test/r/update.result:
Auto merged
mysql-test/r/view.result:
Auto merged
mysql-test/t/explain.test:
Auto merged
mysql-test/t/func_str.test:
Auto merged
mysql-test/t/sp.test:
Auto merged
sql/item.cc:
Auto merged
sql/item.h:
Auto merged
sql/item_cmpfunc.cc:
Auto merged
sql/item_func.cc:
Auto merged
sql/item_strfunc.h:
Auto merged
sql/sql_base.cc:
Auto merged
sql/sql_delete.cc:
Auto merged
sql/sql_insert.cc:
Auto merged
sql/sql_lex.cc:
Auto merged
sql/sql_lex.h:
Auto merged
sql/sql_load.cc:
Auto merged
sql/sql_parse.cc:
Auto merged
sql/sql_select.h:
Auto merged
sql/sql_update.cc:
Auto merged
mysql-test/r/func_in.result:
SCCS merged
mysql-test/r/order_by.result:
SCCS merged
mysql-test/r/sp.result:
merging
mysql-test/t/func_in.test:
SCCS merged
mysql-test/t/order_by.test:
SCCS merged
mysql-test/t/view.test:
merging
sql/item_cmpfunc.h:
merging
sql/mysql_priv.h:
merging
sql/sql_select.cc:
SCCS merged
When the ORDER BY clause gets fixed it's allowed to search in the current
item_list in order to find aliased fields and expressions. This is ok for a
SELECT but wrong for an UPDATE statement. If the ORDER BY clause will
contain a non-existing field which is mentioned in the UPDATE set list
then the server will crash due to using of non-existing (0x0) field.
When an Item_field is getting fixed it's allowed to search item list for
aliased expressions and fields only for selects.
sql/sql_base.cc:
Bug#25126: Wrongly resolved field leads to a crash.
When an Item_field is getting fixed it's allowed to search item list for
aliased expressions and fields only for selects.
sql/sql_select.cc:
Bug#25126: Wrongly resolved field leads to a crash.
When an Item_field is getting fixed it's allowed to search item list for
aliased expressions and fields only for selects.
mysql-test/r/update.result:
Added a test case for bug#25126: Wrongly resolved field leads to a crash.
mysql-test/t/update.test:
Added a test case for bug#25126: Wrongly resolved field leads to a crash.
When the ORDER BY clause gets fixed it's allowed to search in the current
item_list in order to find aliased fields and expressions. This is ok for a
SELECT but wrong for an UPDATE statement. If the ORDER BY clause will
contain a non-existing field which is mentioned in the UPDATE set list
then the server will crash due to using of non-existing (0x0) field.
When an Item_field is getting fixed it's allowed to search item list for
aliased expressions and fields only for selects.
into mysql.com:/home/gluh/MySQL/Merge/5.1-opt
mysql-test/r/innodb.result:
Auto merged
mysql-test/r/insert_update.result:
Auto merged
mysql-test/r/myisam.result:
Auto merged
mysql-test/r/select.result:
Auto merged
mysql-test/r/subselect3.result:
Auto merged
mysql-test/r/type_blob.result:
Auto merged
mysql-test/t/insert_update.test:
Auto merged
mysql-test/t/select.test:
Auto merged
BitKeeper/deleted/.del-bdb.result:
Auto merged
mysql-test/extra/binlog_tests/blackhole.test:
Auto merged
sql/item.cc:
Auto merged
sql/item.h:
Auto merged
sql/mysql_priv.h:
Auto merged
sql/sql_base.cc:
Auto merged
sql/sql_delete.cc:
Auto merged
sql/sql_lex.cc:
Auto merged
sql/sql_lex.h:
Auto merged
sql/sql_prepare.cc:
Auto merged
sql/sql_select.cc:
Auto merged
sql/sql_union.cc:
Auto merged
sql/sql_yacc.yy:
Auto merged
storage/blackhole/ha_blackhole.cc:
Auto merged
mysql-test/r/binlog_stm_blackhole.result:
manual merge
mysql-test/r/insert_select.result:
manual merge
mysql-test/r/subselect.result:
manual merge
mysql-test/r/update.result:
manual merge
mysql-test/t/insert_select.test:
manual merge
mysql-test/t/subselect.test:
manual merge
sql/sql_help.cc:
manual merge
sql/sql_insert.cc:
manual merge
sql/sql_update.cc:
manual merge
it doesn't select.
This bug was fixed along with bug #16861: User defined variable can
have a wrong value if a tmp table was used.
There the fix consisted of Item_func_set_user_var overloading the method
Item::save_in_field. Consider the query from the test case:
INSERT INTO foo( bar, baz )
SELECT
bar,
@newBaz := 1 + baz
FROM
foo
WHERE
quux <= 0.1;
Here the assignment expression '@newBaz := 1 + baz' is represented by an
Item_func_set_user_var. Its member method save_in_field, which writes the
value of this assignment into the result field, writes the val_xxx() value,
which is not updated at this point. In the fix introduced by the patch,
the save_in_field method reads the actual variable value instead.
See also comment for
ChangeSet@1.2368.1.3, 2007-01-09 23:24:56+03:00, evgen@moonbone.local +4 -0
and comment for
Item_func_set_user_var::save_in_field (item_func.cc)
mysql-test/r/update.result:
BUG#24010
The correct, and expected, result
mysql-test/t/update.test:
BUG#24010
The test case for this bug. When the bug is active, the INSERT ... SELECT
statement will try to insert the values <1, 2> which gives an error
it doesn't select.
This bug was fixed along with bug #16861: User defined variable can
have a wrong value if a tmp table was used.
There the fix consisted of Item_func_set_user_var overloading the method
Item::save_in_field. Consider the query from the test case:
INSERT INTO foo( bar, baz )
SELECT
bar,
@newBaz := 1 + baz
FROM
foo
WHERE
quux <= 0.1;
Here the assignment expression '@newBaz := 1 + baz' is represented by an
Item_func_set_user_var. Its member method save_in_field, which writes the
value of this assignment into the result field, writes the val_xxx() value,
which is not updated at this point. In the fix introduced by the patch,
the save_in_field method reads the actual variable value instead.
See also comment for
ChangeSet@1.2368.1.3, 2007-01-09 23:24:56+03:00, evgen@moonbone.local +4 -0
and comment for
Item_func_set_user_var::save_in_field (item_func.cc)
This performance degradation for UPDATEs could be observed in the update
statements for which the search key cannot be converted to any valid
value of the type of the search column, like for a the condition
int_fld=99999999999999999999999999, though it can be guaranteed here
that there is no row with such a key value.
mysql-test/r/update.result:
Added a test case for bug #24035.
mysql-test/t/update.test:
Added a test case for bug #24035.
sql/opt_range.cc:
Fix bug #24035.
This performance degradation for could be observed in the update
statements for which the search key cannot be converted to any valid
value of the type of the search column, like for a the condition
int_fld=99999999999999999999999999, though it can be guaranteed here
that there is no row with such a key value.
Now the function get_mm_leaf creates trees of the type SEL_ARG::IMPOSSIBLE
for such conditions that tells the range scan not to perform any search
at all.
This performance degradation for UPDATEs could be observed in the update
statements for which the search key cannot be converted to any valid
value of the type of the search column, like for a the condition
int_fld=99999999999999999999999999, though it can be guaranteed here
that there is no row with such a key value.
