and for bug #31070: crash during conversion of charsets
Problem: passing a 0 byte length string to some my_mb_wc_XXX()
functions leads to server crash due to improper argument check.
Fix: properly check arguments passed to my_mb_wc_XXX() functions.
mysql-test/include/ctype_common.inc:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- test case.
mysql-test/r/ctype_big5.result:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- test result.
mysql-test/r/ctype_euckr.result:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- test result.
mysql-test/r/ctype_gb2312.result:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- test result.
mysql-test/r/ctype_gbk.result:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- test result.
mysql-test/r/ctype_uca.result:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- test result.
strings/ctype-big5.c:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- check the string length before testing its first byte.
strings/ctype-cp932.c:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- check the string length before testing its first byte.
strings/ctype-euc_kr.c:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- check the string length before testing its first byte.
strings/ctype-gb2312.c:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- check the string length before testing its first byte.
strings/ctype-sjis.c:
Fix for bug #31069: crash in 'sounds like'
and bug #31070: crash during conversion of charsets
- check the string length before testing its first byte.
in get_index_for_order(), don't walk over the end of the index key parts
when matching index description and needed ordering.
mysql-test/r/delete.result:
BUG#30385: Testcase
mysql-test/t/delete.test:
BUG#30385: Testcase
mysql-test/lib/mtr_misc.pl:
Add function 'mtr_rmtree' it will try 'rmtree' and if that fails (most likely
due to permission problems we will fun File::find to chmod all files and dirs
to 0777 and then delete.
mysql-test/mysql-test-run.pl:
Use 'mtr_rmtree' in favour of 'rmtree'
which does not work. Removing these attempted privileges makes
this identical to option 5 so remove it completely. The spirit
of the program appears to be aimed at database privileges, so do
not add another option for granting global privileges as it may
be unexpected. Fixes bug#14618 (same as previous patch, this
time applied to -maint tree).
scripts/mysql_setpermission.sh:
Option 6 tries to apply global privileges at the database
level which does not work - remove it.
When using concurrent insert with parallel index reads, it could
happen that reading sessions found keys that pointed to records
yet to be written to the data file. The result was a report of
a corrupted table. But it was false alert.
When inserting a record in a table with indexes, the keys are
inserted into the indexes before the record is written to the data
file. When the insert happens concurrently to selects, an
index read can find a key that references the record that is not
yet written to the data file. To avoid any access to such record,
the select saves the current end of file position when it starts.
Since concurrent inserts are always appended at end of the data
file, the select can easily ignore any concurrently inserted record.
The problem was that the ignore was only done for non-exact key
searches (partial key or using >, >=, < or <=).
The fix is to ignore concurrently inserted records also for
exact key searches.
No test case. Concurrent inserts cannot be tested with the test
suite. Test cases are attached to the bug report.
myisam/mi_rkey.c:
Bug#29838 - myisam corruption using concurrent select ... and update
Fixed mi_rkey() to always ignore records beyond saved eof.
SELECT statement itself returns empty.
As a result of this bug 'SELECT AGGREGATE_FUNCTION(fld) ... GROUP BY'
can return one row instead of an empty result set.
When GROUP BY only has fields of constant tables
(with a single row), the optimizer deletes the group_list.
After that we lose the information about whether we had an
GROUP BY statement. Though it's important
as SELECT min(x) from empty_table; and
SELECT min(x) from empty_table GROUP BY y; have to return
different results - the first query should return one row,
second - an empty result set.
So here we add the 'group_optimized_away' flag to remember this case
when GROUP BY exists in the query and is removed
by the optimizer, and check this flag in end_send_group()
mysql-test/r/group_by.result:
Bug #29717 INSERT INTO SELECT inserts values even if
SELECT statement itself returns empty.
test result
mysql-test/r/insert_select.result:
Bug #29717 INSERT INTO SELECT inserts values even if
SELECT statement itself returns empty.
test result
mysql-test/t/group_by.test:
Bug #29717 INSERT INTO SELECT inserts values even if
SELECT statement itself returns empty.
This is additional testcase that is more basic than the
original bug's testcase and has the same reason.
mysql-test/t/insert_select.test:
Bug #29717 INSERT INTO SELECT inserts values even if
SELECT statement itself returns empty.
test case
sql/sql_select.cc:
Bug #29717 INSERT INTO SELECT inserts values even if
SELECT statement itself returns empty.
Remember the 'GROUP BY was optimized away' case in the JOIN::group_optimized
and check this in the end_send_group()
sql/sql_select.h:
Bug #29717 INSERT INTO SELECT inserts values even if
SELECT statement itself returns empty.
JOIN::group_optimized member added to remember the 'GROUP BY optimied away'
case
Backport of correction for Mac OS X build problem, global variable not
initiated is "common" and can't be used in shared libraries, unless
special flags are used (bug#26218)
mysys/my_pthread.c:
Backport of correction for Mac OS X build problem, global variable not
initiated is "common" and can't be used in shared libraries, unless
special flags are used (bug#26218)
to 150 or 107 characters for those messages which are generated
by the embedded server during release builds.
This fixes bug#16635:
Error messages wrong: absolute path names, "%s" format code
See the bug report or the changelog for "sql/share/english/errmsg.txt"
for instructions how to do that with other languages,
even at the customer site, and for the restrictions to keep.
sql/share/english/errmsg.txt:
The embedded server uses absolute path names in its error messages,
in the release build environment these exceed the 64 character limit
which the format strings for the error messages impose (bug#16635).
But when the messages are output, the server does the "printf()"
internally in a 256 character buffer; the constant text and the
expanded variables (strings, error number) must fit into this.
(If the buffer would overflow, a format specification will not be
expanded but just copied with its code, and the message output
will just contain '%s' or '%d' where a value is expected.)
So the string lengths are increased to 150 characters in those messages
which are issued by the embedded server during release tests
and contain 1 (one) path name,
but only to 107 in the "rename" message which contains 2 (two).
This solves bug#16635 for the release builds.
For other languages used by OEM customers, similar fixes may be needed,
but we cannot test them.
These fixes can be done even in a binary installation at the customer site
by following these steps:
cd <<install-root>>/share
$EDITOR <<lang>>/errmsg.txt
../../bin/comp_err -C./charsets/ <<lang>>/errmsg.txt <<lang>>/errmsg.sys
and then restarting the server.
This bug manifested itself for join queries with GROUP BY and HAVING clauses
whose SELECT lists contained DISTINCT. It occurred when the optimizer could
deduce that the result set would have not more than one row.
The bug could lead to wrong result sets for queries of this type because
HAVING conditions were erroneously ignored in some cases in the function
remove_duplicates.
mysql-test/r/having.result:
Added a test case for bug #29911.
mysql-test/t/having.test:
Added a test case for bug #29911.
unpack_fields() didn't expect NULL_LENGHT in the field's descriptions.
In this case we get NULL in the resulting string so cannot use
strdup_root to make a copy of it.
strdup_root changed with strmake_root as it's NULL-safe
sql-common/client.c:
Bug #29494 Field packet with NULL fields crashes libmysqlclient
strdup_root changed with strmake_root in unpack_fields()
into sin.intern.azundris.com:/home/tnurnberg/27198/41-27198
sql/mysql_priv.h:
Bug #27198: Error returns from time() are ignored
manual merge
sql/sql_class.h:
Bug #27198: Error returns from time() are ignored
manual merge
gettimeofday() can fail and presumably, so can time().
Keep an eye on it.
Since we have no data on this at all so far, we just
retry on failure (and log the event), assuming that
this is just an intermittant failure. This might of
course hang the threat until we succeed. Once we know
more about these failures, an appropriate more clever
scheme may be picked (only try so many times per thread,
etc., if that fails, return last "good" time() we got or
some such). Using sql_print_information() to log as this
probably only occurs in high load scenarios where the debug-
trace likely is disabled (or might interfere with testing
the effect). No test-case as this is a non-deterministic
issue.
sql/mysql_priv.h:
Bug#27198: Error returns from time() are ignored
move declarations for log.cc to before inclusion of
sql_class.h as we now use sql_print_information() in
there.
sql/sql_class.h:
Bug#27198: Error returns from time() are ignored
gettimeofday() can fail and presumably, so can time().
Keep an eye on it.
Dropping an user defined function may cause server crash in
case this function is still in use by another thread.
The problem was that our hash implementation didn't update
hash link list properly when hash_update() was called.
mysys/hash.c:
The following requirement wasn't met by hash_update() function
causing corruption of hash links list:
After a record was unlinked from the old chain during update, it
holds random position. By the chance this position is equal to
position for the first element in the new chain. That means
updated record is the only record in the new chain.
Test case update for bug #29294.
mysql-test/t/loaddata.test:
Test case update for bug #29294.
mysql-test/r/loaddata.result:
Test case update for bug #29294.
