The problem is that the Table_locks_waited was incremented only
when the lock request succeed. If a thread waiting for the lock
gets killed or the lock request is aborted, the variable would
not be incremented, leading to inaccurate values in the variable.
The solution is to increment the Table_locks_waited whenever the
lock request is queued. This reflects better the intended behavior
of the variable -- show how many times a lock was waited.
Two disjuncts containing equalities of the form key=const1 and key=const2 can
be merged into one if const1 is equal to const2. To check it the common
collation of the constants were used rather than the collation of the field key.
For example when the default collation of the constants was cases insensitive
while the collation of the field was case sensitive, then two or-ed equality
predicates key='b' and key='B' incorrectly were merged into one f='b'. As a
result ref access was used instead of range access and wrong result sets were
returned in many cases.
Fixed the problem by comparing constant in the or-ed predicate with collation of
the key field.
The problem is when create/rename/drop users, the statement was logged regardless of error, even if no data has been changed, the statement was logged.
After this patch, create/rename/drop users don't write the binlog if the statement makes no changes, if the statement does make any changes, log the statement with possible error code.
This patch is based on the patch for BUG#29749, which is not pushed
Bug 33983 (Stored Procedures: wrong end <label> syntax is accepted)
The server used to crash when REPEAT or another control instruction
was used in conjunction with labels and a LEAVE instruction.
The crash was caused by a missing "pop" of handlers or cursors in the
code representing the stored program. When executing the code in a loop,
this missing "pop" would result in a stack overflow, corrupting memory.
Code generation has been fixed to produce the missing h_pop/c_pop
instructions.
Also, the logic checking that labels at the beginning and the end of a
statement are matched was incorrect, causing Bug 33983.
End labels, when used, must match the label used at the beginning of a block.
Bug#25347: mysqlcheck -A -r doesn't repair table marked as crashed
mysqlcheck tests nullness of the engine type to know whether the
"table" is a view or not. That also falsely catches tables that
are severly damaged.
Instead, use SHOW FULL TABLES to test whether a "table" is a view
or not.
(Don't add new function. Instead, get original data a smarter way.)
Make it safe for use against databases before when views appeared.
Add new variable m_highest_seen when only peeking at auto_increment NEXTID and not retrieving to cache. Add new method to check tupleId before calling data node
ndb_restore.result, ndb_restore.test:
Changed test to use information_schema to check auto_increment
DictCache.cpp, Ndb.cpp:
Add new variable m_highest_seen when only peeking at auto_increment NEXTID and not retrieving to cache. Add new method to check tupleId before calling data node. When setting the auto_increment value we'll also read up the new value, this is useful if we use the table the first time in this MySQL Server and haven't yet seen the NEXTID value. The kernel will avoid updating since it already has the value but will also read up the NEXTID value to ensure we don't need to do this any more time.
ndb_auto_increment.result:
Updated result file since it was incorrect
The problem occurred when one had a subquery that had an equality X=Y where
Y referred to a named select list expression from the parent select. MySQL
crashed when trying to use the X=Y equality for ref-based access.
Fixed by allowing non-Item_field items in the described case.
Fixes:
Bug #18942: DROP DATABASE does not drop an orphan FOREIGN KEY constraint
Fix Bug#18942 by dropping all foreign key constraints at the end of
DROP DATABASE. Usually, by then, there are no foreign constraints
left because all of them are dropped when the relevant tables are
dropped. This code is to ensure that any orphaned FKs are wiped too.
Bug #29157: UPDATE, changed rows incorrect
Return HA_ERR_RECORD_IS_THE_SAME from ha_innobase::update_row() if no
columns were updated.
Bug #32440: InnoDB free space info does not appear in SHOW TABLE STATUS or I_S
Put information about the free space in a tablespace in
INFORMATION_SCHEMA.TABLES.DATA_FREE. This information was previously
available in INFORMATION_SCHEMA.TABLES.TABLE_COMMENT, but MySQL has
removed it from there recently.
The stored value is in kilobytes.
This can be considered as a permanent workaround to
http://bugs.mysql.com/32440. "Workaround" becasue that bug is about the
data missing from TABLE_COMMENT and this is actually not solved.
The ROUND(X, D) function would change the Item::decimals field during
execution to achieve the effect of a dynamic number of decimal digits.
This caused a series of bugs:
Bug #30617:Round() function not working under some circumstances in InnoDB
Bug #33402:ROUND with decimal and non-constant cannot round to 0 decimal places
Bug #30889:filesort and order by with float/numeric crashes server
Fixed by never changing the number of shown digits for DECIMAL when
used with a nonconstant number of decimal digits.
The name resolution for correlated subqueries and HAVING clauses
failed to distinguish which of two was being performed when there
was a reference to an outer aliased field.
Fixed by adding the condition that HAVING clause name resulotion
is being performed.
value when inserting into a view.
The mysql_prepare_insert function checks all fields of the target table that
directly or indirectly (through a view) are specified in the INSERT
statement to have a default value. This check can be skipped if the INSERT
statement doesn't mention any insert fields. In case of a view this allows
fields that aren't mentioned in the view to bypass the check.
Now fields of the target table are always checked to have a default value
when insert goes into a view.
columns (default datatype value is assigned).
The mysql_update function has been modified to generate
an error when trying to set a NOT NULL field to NULL rather than a warning
in the set_field_to_null_with_conversions function.
When resolving references we need to take into consideration
the view "fields" and allow qualified access to them.
Fixed by extending the reference resolution to process view
fields correctly.
server crash.
The filesort implementation has an optimization for subquery execution which
consists of reusing previously allocated buffers. In particular the call to
the read_buffpek_from_file function might be skipped when a big enough buffer
for buffer descriptors (buffpeks) is already allocated. Beside allocating
memory for buffpeks this function fills allocated buffer with data read from
disk. Skipping it might led to using an arbitrary memory as fields' data and
finally to a crash.
Now the read_buffpek_from_file function is always called. It allocates
new buffer only when necessary, but always fill it with correct data.
Reseting the query cache by issuing a SET GLOBAL query_cache_size=0 caused the server
to crash if a the server concurrently was saving a new result set to the query cache. The
reason for this was that the invalidation wasn't waiting on the result writers to
release the block level locks on the query cache.
to be compiled in
The problem was that on a statically built server an attempt to create
a UDF resulted in a different, but reasonable error ("Can't open shared
library" instead of "UDFs are unavailable with the --skip-grant-tables
option"), which caused a failure for the test case for bug #32020.
Fixed by moving the test case for bug #32020 from skip_grants.test to a
separate test to ensure that it is only run when the server is built
with support for dynamically loaded libraries.
read_buffer_size set on master
BUG#33413 show binlog events fails if binlog has event size of close
to max_allowed_packet
The size of Append_block replication event was determined solely by
read_buffer_size whereas the rest of replication code deals with
max_allowed_packet.
When the former parameter was set to larger than the latter there were
two artifacts: the master could not read events from binlog;
show master events did not show.
Fixed with
- fragmenting the used io-cached buffer into pieces each size of less
than max_allowed_packet (bug#30435)
- incrementing show-binlog-events handling thread's max_allowed_packet
with the max estimated for the replication header size
Now, every transaction (including autocommit transactions) start with
a BEGIN and end with a COMMIT/ROLLBACK in the binlog.
Added a test case, and updated lots of test case result files.
w/ Field_date instead of Field_newdate
Field_date was still used in temp table creation.
Fixed by using Field_newdate consistently throughout the server
except when reading tables defined with older MySQL version.
No test suite is possible because both Field_date and Field_newdate
return the same values in all the metadata calls.
When set the server-id dynamically, the server_id member of current thread is not updated.
Update the server_id member of current thread after updated the global variable value.
Complementary patch since LOAD DATA INFILE was not covered in
the previous patch.
This patch adds a check so that the slave skip counter is not
decreased to zero if seeing a BEGIN_LOAD_QUERY_EVENT,
APPEND_BLOCK_EVENT, or CREATE_FILE_EVENT since these cannot
end a group. The group is terminated by an EXECUTE_LOAD_QUERY_
EVENT or DELETE_FILE_EVENT.
Parser rejects ODBC's escape sequences for outer joins other
than left outer join, yet the escape sequence BNF specifies
that this syntax can be used for left, right, and full outer
join syntax.
The problem is that although the MySQL Connector/ODBC advertises
"Outer Join Escape Sequence" capabilities, the parsing is done
in the server and historically it only supported this syntax
for left outer joins and applications such as Crystal Reports
11 tries to use this syntax for inner joins.
The chosen solution is to reorganize a couple of parser rules
to ignore any kind of SQL escape sequence. Ignoring the escape
sequences is harmless because the various SQL join clauses
are supported by the server.
that the entire server uses their public ha_* counterparts instead,
since only then we can ensure proper tracing of these calls that
is necessary for Bug#12713.
A pre-requisite for Bug#12713 "Error in a stored function called from
a SELECT doesn't cause ROLLBACK of statem"
The checks in the test for bug #12480 were too wide and
made the test to depend on the procedures and triggers
present in the server.
Corrected the test to check only for the procedure and
trigger it creates.
fine
The reason of this bug is that when mysqlbinlog dumps a query, the query is written to
output with a delimeter appended right after it, if the query string ends with a '--'
comment, then the delimeter would be considered as part of the comment, if there are any
statements after this query, then it will cause a syntax error.
Start a newline before appending delimiter after a query string
The reason of this bug is that when mysqlbinlog dumps a query, the query is written to
output with a delimeter appended right after it, if the query string ends with a '--'
comment, then the delimeter would be considered as part of the comment, if there are any
statements after this query, then it will cause a syntax error.
Start a newline before appending delimiter after a query string
In a union without braces, the order by at the end is applied to the
overall union. It therefore should not interfere with the individual
select parts of the union.
Fixed by changing our parser rules appropriately.
Problem: it is unsafe to read base64-printed events without first
reading the Format_description_log_event (FD). Currently, mysqlbinlog
cannot print the FD.
As a side effect, another bug has also been fixed: When mysqlbinlog
--start-position=X was specified, no ROLLBACK was printed. I changed
this, so that ROLLBACK is always printed.
This patch does several things:
- Format_description_log_event (FD) now print themselves in base64
format.
- mysqlbinlog is now able to print FD events. It has three modes:
--base64-output=auto Print row events in base64 output, and print
FD event. The FD event is printed even if
it is outside the range specified with
--start-position, because it would not be
safe to read row events otherwise. This is
the default.
--base64-output=always Like --base64-output=auto, but also print
base64 output for query events. This is
like the old --base64-output flag, which
is also a shorthand for
--base64-output=always
--base64-output=never Never print base64 output, generate error if
row events occur in binlog. This is
useful to suppress the FD event in binlogs
known not to contain row events (e.g.,
because BINLOG statement is unsafe,
requires root privileges, is not SQL, etc)
- the BINLOG statement now handles FD events correctly, by setting
the thread's rli's relay log's description_event_for_exec to the
loaded event.
In fact, executing a BINLOG statement is almost the same as reading
an event from a relay log. Before my patch, the code for this was
separated (exec_relay_log_event in slave.cc executes events from
the relay log, mysql_client_binlog_statement in sql_binlog.cc
executes BINLOG statements). I needed to augment
mysql_client_binlog_statement to do parts of what
exec_relay_log_event does. Hence, I did a small refactoring and
moved parts of exec_relay_log_event to a new function, which I
named apply_event_and_update_pos. apply_event_and_update_pos is
called both from exec_relay_log_event and from
mysql_client_binlog_statement.
- When a non-FD event is executed in a BINLOG statement, without
previously executing a FD event in a BINLOG statement, it generates
an error, because that's unsafe. I took a new error code for that:
ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS.
In order to get a decent error message containing the name of the
event, I added the class method char*
Log_event::get_type_str(Log_event_type type), which returns a
string name for the given Log_event_type. This is just like the
existing char* Log_event::get_type_str(), except it is a class
method that takes the log event type as parameter.
I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(),
so that names of old rows event are properly printed.
- When reading an event, I added a check that the event type is known
by the current Format_description_log_event. Without this, it may
crash on bad input (and I was struck by this several times).
- I patched the following test cases, which all contain BINLOG
statements for row events which must be preceded by BINLOG
statements for FD events:
- rpl_bug31076
While I was here, I fixed some small things in log_event.cc:
- replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places
- replaced return by DBUG_VOID_RETURN in one place
- The name of the logfile can be '-' to indicate stdin. Before my
patch, the code just checked if the first character is '-'; now it
does a full strcmp(). Probably, all arguments that begin with a -
are already handled somewhere else as flags, but I still think it
is better that the code reflects what it is supposed to do, with as
little dependencies as possible on other parts of the code. If we
one day implement that all command line arguments after -- are
files (as most unix tools do), then we need this.
I also fixed the following in slave.cc:
- next_event() was declared twice, and queue_event was not static but
should be static (not used outside the file).
Now, every transaction (including autocommit transactions) starts with
a BEGIN and ends with a COMMIT/ROLLBACK in the binlog.
Added a test case, and updated lots of test case result files.
with null values
For queries containing GROUP_CONCAT(DISTINCT fields ORDER BY fields), there
was a limitation that the DISTINCT fields had to be the same as ORDER BY
fields, owing to the fact that one single sorted tree was used for keeping
track of tuples, ordering and uniqueness. Fixed by introducing a second
structure to handle uniqueness so that the original structure has only to
order the result.
subselects into account
It is forbidden to use the SELECT INTO construction inside UNION statements
unless on the last SELECT of the union. The parser records whether it
has seen INTO or not when parsing a UNION statement. But if the INTO was
legally used in an outer query, an error is thrown if UNION is seen in a
subquery. Fixed in 5.0 by remembering the nesting level of INTO tokens and
mitigate the error unless it collides with the UNION.
The problem is that some DDL statements (ALTER TABLE, CREATE
TRIGGER, FLUSH TABLES, ...) when under LOCK TABLES need to
momentarily drop the lock, reopen the table and grab the write
lock again (using reopen_tables). When grabbing the lock again,
reopen_tables doesn't pass a flag to mysql_lock_tables in
order to ignore the impending global read lock, which causes a
assertion because LOCK_open is being hold. Also dropping the
lock must not signal to any threads that the table has been
relinquished (related to the locking/flushing protocol).
The solution is to correct the way the table is reopenned
and the locks grabbed. When reopening the table and under
LOCK TABLES, the table version should be set to 0 so other
threads have to wait for the table. When grabbing the lock,
any other flush should be ignored because it's theoretically
a atomic operation. The chosen solution also fixes a potential
discrepancy between binlog and GRL (global read lock) because
table placeholders were being ignored, now a FLUSH TABLES WITH
READ LOCK will properly for table with open placeholders.
It's also important to mention that this patch doesn't fix
a potential deadlock if one uses two GRLs under LOCK TABLES
concurrently.
cause ROLLBACK of statement", part 1. Review fixes.
Do not send OK/EOF packets to the client until we reached the end of
the current statement.
This is a consolidation, to keep the functionality that is shared by all
SQL statements in one place in the server.
Currently this functionality includes:
- close_thread_tables()
- log_slow_statement().
After this patch and the subsequent patch for Bug#12713, it shall also include:
- ha_autocommit_or_rollback()
- net_end_statement()
- query_cache_end_of_result().
In future it may also include:
- mysql_reset_thd_for_next_command().
There were two problems when inferring the correct field types resulting from
UNION queries.
- If the type is NULL for all corresponding fields in the UNION, the resulting
type would be NULL, while the type is BINARY(0) if there is just a single
SELECT NULL.
- If one SELECT in the UNION uses a subselect, a temporary table is created
to represent the subselect, and the result type defaults to a STRING type,
hiding the fact that the type was unknown(just a NULL value).
Fixed by remembering whenever a field was created from a NULL value and pass
type NULL to the type coercion if that is the case, and creating a string field
as result of UNION only if the type would otherwise be NULL.
HOUR(), MINUTE(), ... returned spurious results when used on a DATE-cast.
This happened because DATE-cast object did not overload get_time() method
in superclass Item. The default method was inappropriate here and
misinterpreted the data.
Patch adds missing method; get_time() on DATE-casts now returns SQL-NULL
on NULL input, 0 otherwise. This coincides with the way DATE-columns
behave.
Also fixes similar bug in Date-Field now.
The problem was that when convert_constant_item is called for subqueries,
this happens when we already started executing the top-level query, and
the field argument of convert_constant_item pointed to a valid table row.
In turn convert_constant_item used the field buffer to compute the value
of its item argument. This copied the item's value into the field,
and made equalities with outer references always true.
The fix saves/restores the original field's value when it belongs to an
outer table.
Both arguments of the function NAME_CONST must be constant expressions.
This constraint is checked in the Item_name_const::fix_fields method.
Yet if the argument of the function was not a constant expression no
error message was reported. As a result the client hanged waiting for a
response.
Now the function Item_name_const::fix_fields reports an error message
when any of the additional context conditions imposed on the function
NAME_CONST is not satisfied.
The index (key_part_1, key_part-2) was erroneously considered as compatible
with the required ordering in the function test_test_if_order_by_key when
a query with an ORDER BY clause contained a condition of the form
key_part_1=const OR key_part_1 IS NULL
and the order list contained only key_part_2. This happened because the value
of the const_key_parts field in the KEYUSE structure was not formed correctly
for the keys that could be used for ref_or_null access.
This was fixed in the code of the update_ref_and_keys function.
The problem could not manifest itself for MyISAM databases because the
implementation of the keys_to_use_for_scanning() handler function always
returns an empty bitmap for the MyISAM engine.
Change LAST_EXECUTED time the execution start time, instead of the execution completion time. This ensures the END time always the same or later than the LAST_EXECUTED time.
The Item_func_set_user_var::register_field_in_read_map() did not check
that the result_field was null.This caused server crashes for queries that
required order by such a field and were executed without using a temporary
table.
The Item_func_set_user_var::register_field_in_read_map() now checks the
result_field to be not null.
When read_only option was enabled, a user without SUPER privilege could
perform CREATE DATABASE and DROP DATABASE operations.
This patch adds a check to make sure this isn't possible. It also attempts to
simplify the logic used to determine if relevant tables are updated,
making it more human readable.
Problem: when alter to partitioned table,
it does not see it as change of engine.
Solution: If alter includes partitioning, check if it is possible
to change engines (eg. is the table referenced by a FK)
ha_partition::update_create_info() just calls update_create_info
of a first partition, so only get the autoincrement maximum
of the first partition, so SHOW CREATE TABLE can show
small AUTO_INCREMENT parameters.
Fixed by implementing ha_partition::update_create_info() in a way
other handlers work.
HA_ARCHIVE:stats.auto_increment handling made consistent with other engines
Anti-patch. This patch undoes the previously pushed patch. It is
null-merged in versions 5.1 and above since there the original
patch is still desired.
When executing drop view statement on the master, the statement is written
into bin-log without checking for possible errors, so the statement would
always be bin-logged with error code cleared even if some error might occur,
for example, some of the views being dropped does not exist. This would cause
failure on the slave.
Writing bin-log after check for errors, if at least one view has been dropped
the query is bin-logged possible with an error.
