transaction
BUG#52616 Temp table prevents switch binlog format from STATEMENT to ROW
Before the WL#2687 and BUG#46364, every non-transactional change that happened
after a transactional change was written to trx-cache and flushed upon
committing the transaction. WL#2687 and BUG#46364 changed this behavior and
non-transactional changes are now written to the binary log upon committing
the statement.
A binary log event is identified as transactional or non-transactional through
a flag in the Log_event which is set taking into account the underlie storage
engine on what it is stems from. In the current bug, this flag was not being
set properly when the DROP TEMPORARY TABLE was executed.
However, while fixing this bug we figured out that changes to temporary tables
should be always written to the trx-cache if there is an on-going transaction.
Otherwise, binlog events in the reversed order would be produced.
Regarding concurrency, keeping changes to temporary tables in the trx-cache is
also safe as temporary tables are only visible to the owner connection.
In this patch, we classify the following statements as unsafe:
1 - INSERT INTO t_myisam SELECT * FROM t_myisam_temp
2 - INSERT INTO t_myisam_temp SELECT * FROM t_myisam
3 - CREATE TEMPORARY TABLE t_myisam_temp SELECT * FROM t_myisam
On the other hand, the following statements are classified as safe:
1 - INSERT INTO t_innodb SELECT * FROM t_myisam_temp
2 - INSERT INTO t_myisam_temp SELECT * FROM t_innodb
The patch also guarantees that transactions that have a DROP TEMPORARY are
always written to the binary log regardless of the mode and the outcome:
commit or rollback. In particular, the DROP TEMPORARY is extended with the
IF EXISTS clause when the current statement logging format is set to row.
Finally, the patch allows to switch from STATEMENT to MIXED/ROW when there
are temporary tables but the contrary is not possible.
The server variable opt_secure_file_priv wasn't
normalized properly and caused the operations
LOAD DATA INFILE .. INTO TABLE ..
and
SELECT load_file(..)
to do different interpretations of the
--secure-file-priv option.
The patch moves code to the server initialization
routines so that the path always is normalized
once and only once.
It was also intended that setting the option
to an empty string should be equal to
lifting all previously set restrictions. This
is also fixed by this patch.
Arg_comparator initializes 'comparators' array in case of
ROW comparison and does not free this array on destruction.
It leads to memory leaks.
The fix:
-added Arg_comparator::cleanup() method which frees
'comparators' array.
-added Item_bool_func2::cleanup() method which calls
Arg_comparator::cleanup() method
When re-setting (SET GLOBAL debug='') the GLOBAL debug settings the
server was not freeing the data elements from the top (initial) frame
before setting them to 0 without freeing the underlying memory. As these
are global settings there's a chance that something is there already.
Fixed by :
1. making sure the allocated data are cleaned up before re-setting them
while parsing a debug string
2. making sure the stuff allocated in the global settings is freed on
shutdown.
This assertion could be triggered during execution of OPTIMIZE TABLE for
InnoDB tables. As part of optimize for InnoDB tables, the table is recreated
and then opened again. If the reopen failed for any reason, the assertion
would be triggered. This could for example be caused by a concurrent DROP
TABLE executed by a different connection. The reason for the assertion was
that any failures during reopening were ignored.
This patch fixes the problem by making sure that the result of reopening the
table is checked and that any error messages are sent to the client.
Test case added to innodb_mysql_sync.test.
This was a deadlock between CREATE/ALTER/DROP EVENT and a query
accessing both the mysql.event table and I_S.GLOBAL_VARIABLES.
The root of the problem was that the LOCK_event_metadata mutex was
used to both protect the "event_scheduler" global system variable
and the internal event data structures used by CREATE/ALTER/DROP EVENT.
The deadlock would occur if CREATE/ALTER/DROP EVENT held
LOCK_event_metadata while trying to open the mysql.event table,
at the same time as the query had mysql.event open, trying to
lock LOCK_event_metadata to access "event_scheduler".
This bug was fixed in the scope of Bug#51160 by using only
LOCK_global_system_variables to protect "event_scheduler".
This makes it so that the query above won't lock LOCK_event_metadata,
thereby preventing this deadlock from occuring.
This patch contains no code changes.
Test case added to lock_sync.test.
even if myisam-recover is OFF
The problem was that a corrupted MyISAM table was auto repaired
even if the myisam_recover_options server variable (or the
myisam_recover option) was set to OFF.
The reason was that the auto_repair() function, which is supposed
to say if auto repair is to be used, did not use the server variable
setting correctly. This bug was a regression introduced by WL#4738.
This patch fixes the problem by making sure auto_repair() returns
FALSE if myisam_recover_options is set to OFF.
Test case added to myisam.test.
for write by another connection
The problem was that if a table was locked in one connection by
LOCK TABLES ... WRITE, REPAIR TABLE or OPTIMIZE TABLE, SHOW CREATE
TABLE from another connection would be blocked. As SHOW CREATE TABLE
only reads metadata about the table, such blocking is not needed.
The problem was that when SHOW CREATE TABLE tried to get a metadata
lock on the table in order to open it, it used the wrong type of
metadata lock request. It used MDL_SHARED_READ which is used when
the intent is to read both table metadata and table data. Instead
it should have used MDL_SHARED_HIGH_PRIO which signifies an intent
to only read metadata.
This patch fixes the problem by making sure SHOW CREATE TABLE uses
the MDL_SHARED_HIGH_PRIO metadata lock request type when trying to
open the table. The patch also fixes a similar problem with the
mysql_list_fields API call.
Test case added to show_check.test.
Allow stored procedure variables in LIMIT clause.
Only allow variables of INTEGER types.
Handle negative values by means of an implicit cast to UNSIGNED
(similarly to prepared statement placeholders).
Add tests.
Make sure replication works by not doing NAME_CONST substitution
for variables in LIMIT clause.
Add replication tests.
ChangeSet@1.2703, 2007-12-07 09:35:28-05:00, cmiller@zippy.cornsilk.net +40 -0
Bug#13174: SHA2 function
Patch contributed from Bill Karwin, paper unnumbered CLA in Seattle
Implement SHA2 functions.
Chad added code to make it work with YaSSL. Also, he removed the
(probable) bug of embedded server never using SSL-dependent
functions. (libmysqld/Makefile.am didn't read ANY autoconf defs.)
Function specification:
SHA2( string cleartext, integer hash_length )
-> string hash, or NULL
where hash_length is one of 224, 256, 384, or 512. If either is
NULL or a length is unsupported, then the result is NULL. The
resulting string is always the length of the hash_length parameter
or is NULL.
Include the canonical hash examples from the NIST in the test
results.
---
Polish and address concerns of reviewers.
Problem: Segmentation fault in add_group_and_distinct_keys() when accessing
field of what is assumed to be an Item_field object.
Cause: In case of views, the item added to list by is_indexed_agg_distinct()
was not of type Item_field, but Item_ref.
Resolution: Add the real Item_field object, the one referred to by
Item_ref object, to the list, instead.
they are ignored to a new test suite "innodb_plugin".
Remove a hack in mtr that was deployed to run the builtin InnoDB tests against
the InnoDB Plugin. Also detect if a test is an 'innodb plugin test' and if so
then transparently replace the builtin InnoDB with the InnoDB Plugin.
The failing assertion was written with the assumption that a NULL
string can never be passed to my_strtod(). However, an empty string
may be passed under some circumstances by passing str == NULL and
*end == NULL.
Fixed the assertion to take the above case into account.
We should disable const subselect item evaluation because
subselect transformation does not happen in view_prepare_mode
and thus val_...() methods can not be called.
The problem is that we can not use make_cond_for_table().
This function relies on used_tables() condition
which is not set properly for subqueries.
As result subquery is not filtered out.
The fix is to use remove_eq_conds() function instead
of make_cond_for_table() func. 'remove_eq_conds()'
algorithm relies on const_item() value and it allows
to handle subqueries in right way.
Conflicts:
Text conflict in mysql-test/suite/binlog/r/binlog_row_mix_innodb_myisam.result
Text conflict in sql/log.cc
Text conflict in sql/set_var.cc
Text conflict in sql/sql_class.cc
Procedure, while DECIMAL works
Selecting of the CONCAT(...<SP variable>...) result into
a user variable may return wrong data.
Item_func_concat::val_str contains a number of memory
allocation-saving tricks. One of them concatenates
strings inplace inserting the value of one string
at the beginning of the other string. However,
this trick didn't care about strings those points
to the same data buffer: this is possible when
a CONCAT() parameter is a stored procedure variable -
Item_sp_variable::val_str() uses the intermediate
Item_sp_variable::str_value field, where it may
store a reference to an external buffer.
The Item_func_concat::val_str function has been
modified to take into account val_str functions
(such as Item_sp_variable::val_str) that return
a pointer to an internal Item member variable
that may reference to a buffer provided.
Conflicts:
Text conflict in mysql-test/r/func_str.result
Text conflict in mysql-test/suite/sys_vars/r/myisam_sort_buffer_size_basic_32.result
Text conflict in mysql-test/suite/sys_vars/r/myisam_sort_buffer_size_basic_64.result
Text conflict in mysql-test/t/func_str.test
Text conflict in sql/mysqld.cc
Text conflict in sql/protocol.cc
Text conflict in storage/myisam/mi_open.c
on index
'my_decimal' class has two members which can be used to access the
value. The member variable buf (inherited from parent class decimal_t)
is set to member variable buffer so that both are pointing to same value.
Item_copy_decimal::copy() uses memcpy to clone 'my_decimal'. The member
buffer is declared as an array and memcpy results in copying the values
of the array, but the inherited member buf, which should be pointing at
the begining of the array 'buffer' starts pointing to the begining of
buffer in original object (which is being cloned). Further updates on
'my_decimal' updates only the inherited member 'buf' but leaves
buffer unchanged.
Later when the new object (which now holds a inconsistent value) is cloned
again using proper cloning function 'my_decimal2decimal' the buf pointer
is fixed resulting in loss of the current value.
Using my_decimal2decimal instead of memcpy in Item_copy_decimal::copy()
fixed this problem.
data and index files
It was possible if DATA/INDEX DIRECTORY is pointing to
symlinked MySQL data home directory.
Do not allow to drop data/index files implicitly symlinked
to data home directory. For such tables remove symlink only.
The problem was that a syntactically invalid trigger could cause
the server to crash when trying to list triggers. The crash would
happen due to a mishap in the backup/restore procedure that should
protect parser items which are not associated with the trigger. The
backup/restore is used to isolate the parse tree (and context) of
a statement from the load (and parsing) of a trigger. In this case,
a error during the parsing of a trigger could cause the improper
backup/restore sequence.
The solution is to properly restore the original statement context
before the parser is exited due to syntax errors in the trigger body.
Detailed revision comments:
r6800 | marko | 2010-03-11 12:02:57 +0200 (Thu, 11 Mar 2010) | 1 line
branches/zip: Add ut_ad(mtr->state == MTR_ACTIVE) to various places.
Detailed revision comments:
r6799 | jyang | 2010-03-11 09:59:42 +0200 (Thu, 11 Mar 2010) | 5 lines
branches/zip: Once change in bug #47621 merges into zip branch,
zip only test innodb_bug44571 needs to be updated to reflect the
column name change would be successful be done in InnoDB as well.
Detailed revision comments:
r6790 | jyang | 2010-03-10 13:09:41 +0200 (Wed, 10 Mar 2010) | 7 lines
branches/zip: Fix bug #51356: "many valgrind errors in error messages
with concurrent ddl". Null terminate the name string returned
from innobase_convert_identifier() call when reporting DB_DUPLICATE_KEY
error in create_table_def().
rb://266 approved by Marko
Detailed revision comments:
r6770 | marko | 2010-03-03 12:52:55 +0200 (Wed, 03 Mar 2010) | 12 lines
branches/zip: Disallow duplicate index name when creating an index.
This should fix Mantis Issue #461.
innodb.test, innodb.result, innodb-index.test, innodb-index.result:
Adjust the test result and mention that the introduced restriction
has been reported as MySQL Bug #51451.
innobase_check_index_keys(): Add a parameter for the InnoDB table and
check that no duplicate index name is added. Report errors by
my_error() instead of sql_print_error().
rb://260 approved by Sunny Bains
Detailed revision comments:
r6526 | jyang | 2010-01-28 18:12:40 +0200 (Thu, 28 Jan 2010) | 8 lines
branches/zip: Add index translation table to map mysql index
number to InnoDB index structure directly. Fix Bug #47622:
"the new index is added before the existing ones in MySQL,
but after one in SE".
rb://215, approved by Marko
Detailed revision comments:
r6472 | calvin | 2010-01-16 01:53:47 +0200 (Sat, 16 Jan 2010) | 12 lines
branches/zip: Merge revisions 6425:6471 from branches/5.1
to pick up the first part fix of bug49396.
------------------------------------------------------------------------
r6471 | calvin | 2010-01-15 17:43:27 -0600 (Fri, 15 Jan 2010) | 4 lines
branches/5.1: fix bug#49396: main.innodb test fails in embedded mode
Change replace_result by using $MYSQLD_DATADIR. Tested in both embedded
mode and normal server mode.
------------------------------------------------------------------------
r6473 | calvin | 2010-01-16 01:58:16 +0200 (Sat, 16 Jan 2010) | 6 lines
branches/zip: fix bug#49396: innodb.innodb-index test fails in
embedded mode
This is 2nd part of the fix for bug#49396. The 1st part is
innodb.test. Tested in both embedded mode and normal server mode.
BUG#46364 introduced the flag binlog_direct_non_transactional_updates which
would make N-changes to be written to the binary log upon committing the
statement when "ON". On the other hand, when "OFF" the option was supposed
to mimic the behavior in 5.1. However, the implementation was not mimicking
the behavior correctly and the following bugs popped up:
Case #1: N-changes executed within a transaction would go into
the S-cache. When later in the same transaction a
T-change occurs, N-changes following it were written
to the T-cache instead of the S-cache. In some cases,
this raises problems. For example, a
Table_map_log_event being written initially into the
S-cache, together with the initial N-changes, would be
absent from the T-cache. This would log N-changes
orphaned from a Table_map_log_event (thence discarded
at the slave). (MIXED and ROW)
Case #2: When rolling back a transaction, the N-changes that
might be in the T-cache were disregarded and
truncated along with the T-changes. (MIXED and ROW)
Case #3: When a MIXED statement (TN) is ahead of any other
T-changes in the transaction and it fails, it is kept
in the T-cache until the transaction ends. This is
not the case in 5.1 or Betony (5.5.2). In these, the
failed TN statement would be written to the binlog at
the same instant it had failed and not deferred until
transaction end. (SBR)
To fix these problems, we have decided to do what follows:
For Case #1 and #2, we circumvent them:
1. by not letting binlog_direct_non_transactional_updates
affect MIXED and RBR. These modes will keep the behavior
provided by WL#2687. Although this will make Celosia to
behave differently from 5.1, an execution will be always
safe under such modes in the sense that slaves will never
go out sync. In 5.1, using either MIXED or ROW while
mixing N-statements and T-statements was not safe.
For Case #3, we don't actually fix it. We:
1. keep it and make all MIXED statements whether they end
up failing or not or whether they are up front in the
transaction or after some transactional change to always
be stored in the T-cache. This means that it is written
to the binary log on transaction commit/rollback only.
2. We make the warning message even more specific about the
MIXED statement and SBR.
Problem: EXPLAIN EXTENDED was trying to resolve references to
freed temporary table fields for GROUP_CONCAT()'s ORDER BY arguments.
Fix: use stored original GROUP_CONCAT()'s arguments in such a case.
------------------------------------------------------------------------
r6921 | mmakela | 2010-03-31 14:33:04 +0300 (Wed, 31 Mar 2010) | 2 lines
Changed paths:
M /branches/5.1/mysql-test/innodb_bug51920.result
M /branches/5.1/mysql-test/innodb_bug51920.test
branches/5.1: innodb_bug51920.test: Make the test quicker and more
deterministic. Suggested by Vasil Dimov.
------------------------------------------------------------------------
function on windows
When making sure that the directory path ends up with a
slash/backslash we need to check for the correct length of
the buffer and trim at the appropriate location so we don't
write past the end of the buffer.
When mysqlbinlog was given the --database=X flag, it always printed
'ROLLBACK TO', but the corresponding 'SAVEPOINT' statement was not
printed. The replicated filter(replicated-do/ignore-db) and binlog
filter (binlog-do/ignore-db) has the same problem. They are solved
in this patch together.
After this patch, We always check whether the query is 'SAVEPOINT'
statement or not. Because this is a literal check, 'SAVEPOINT' and
'ROLLBACK TO' statements are also binlogged in uppercase with no
any comments.
The binlog before this patch can be handled correctly except one case
that any comments are in front of the keywords. for example:
/* bla bla */ SAVEPOINT a;
/* bla bla */ ROLLBACK TO a;
The log event of 'CREATE EVENT' was being binlogged with garbage
at the end of the query if 'CREATE EVENT' is followed by another SQL statement
and they were executed as one command.
for example:
DELIMITER |;
CREATE EVENT e1 ON EVERY DAY DO SELECT 1; SELECT 'a';
DELIMITER ;|
When binlogging 'CREATE EVENT', we always create a new statement with definer
and write it into the log event. The new statement is made from cpp_buf(preprocessed buffer).
which is not a c string(end with '\0'), but it is copied as a c string.
In this patch, cpp_buf is copied with its length.
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 fix is to remove erroneous 'outer_join' variable check.
The crash happens because of incorrect max_length calculation
in QUOTE function(due to overflow). max_length is set
to 0 and it leads to assert failure.
The fix is to cast expression result to
ulonglong variable and adjust it if the
result exceeds MAX_BLOB_WIDTH.
There was no way to repair corrupt ARCHIVE data file,
when unrecoverable data loss is inevitable.
With this fix REPAIR ... EXTENDED attempts to restore
as much rows as possible, ignoring unrecoverable data.
Normal REPAIR is still able to repair meta-data file
only.
Repairing MyISAM table with fulltext indexes and low
myisam_sort_buffer_size may crash the server.
Estimation of number of index entries was done incorrectly,
causing further assertion failure or server crash.
Docs note: min value for myisam_sort_buffer_size has been
changed from 4 to 4096.
