The reason for the bug was that replaying of a query on slave could not be possible since its event
was recorded with the killed error. Due to the specific of handling INSERT, which per-row-while-loop is
unbreakable to killing, the query on transactional table should have not appeared in binlog unless
there was a call to a stored routine that got interrupted with killing (and then there must be an error
returned out of the loop).
The offered solution added the following rule for binlogging of INSERT that accounts the above
specifics:
For INSERT on transactional-table if the error was not set the only raised flag
is harmless and is ignored via masking out on time of creation of binlog event.
For both table types the combination of raised error and KILLED flag indicates that there
was potentially partial execution on master and consistency is under the question.
In that case the code continues to binlog an event with an appropriate killed error.
The fix relies on the specified behaviour of stored routine that must propagate the error
to the top level query handling if the thd->killed flag was raised in the routine execution.
The patch adds an arg with the default killed-status-unset value to Query_log_event::Query_log_event.
Problem: Memory overrun happened in attempts to generate
error messages (e.g. in case of incorrect XPath syntax).
Reason: set_if_bigger() was used instead of set_if_smaller().
Change: replacing wrong set_if_bigger() to set_if_smaller(),
and making minor additional code clean-ups.
This patch corrects a problem encountered when reading the binlog from a remote
host. The application was crashing because the buffer variable (temp_buf) in
log_event was not pointing to the incoming data. For a normal file read, this
buffer is allocated by a previous call of read_log_event. However, when reading
from a remote host, the first call to read_log_event is not executed therefore
no buffer is allocated. Furthermore, there is no need to allocate a new buffer
because the incoming stream is what needs to be read.
This patch adds the call to initialize the temp_buf variable if reading from a
remote host. It also adds a check at destroy time to ensure the temp_buf is not
freed if reading from a remote host.
- Added suppressions for dlopen to make plugin test pass.
- Do not pass empty string to mysqld, since my_getopt is
not capable to handle it.
- Re-enabled trailing UNINSTALL PLUGIN statement of plugin.test.
The memory leak described in the bug report happens in libdl, not
in mysqld. On some valgrind installations this error is suppressed
by default, no idea why it isn't suppressed on pb-valgrind.
<observation>
If library remains open after thread has finished, and is closed by
another thread, we get memory leak. But in case library is opened and
closed by the same thread no leak occurs.
</observation>
- The SQL commands used by mysql_upgrade are written to be run
with sql_mode set to '' - thus the scripts should change sql_mode
for the session to make sure the SQL is legal.
UDF can be created from any library in any part of the server
LD_LIBRARY_PATH.
Allow to load udfs only from plugin_dir.
On windows, refuse to open udf in case it's path contains a slash.
No good test case for this bug because of imperfect error message
that includes error code and error string when it fails to dlopen a
library.
is involved.
The Arg_comparator::compare_datetime() comparator caches its arguments if
they are constants i.e. const_item() returns true. The
Item_func_get_user_var::const_item() returns true or false based on
the current query_id and the query_id where the variable was created.
Thus even if a query can change its value its const_item() still will return
true. All this leads to a wrong comparison result when an object of the
Item_func_get_user_var class is involved.
Now the Arg_comparator::can_compare_as_dates() and the
get_datetime_value() functions never cache result of the GET_USER_VAR()
function (the Item_func_get_user_var class).
- Problem was reported as a SP variable using itself as
right value inside SUBSTR caused corruption of data.
- This bug could not be verified in either 5.0bk or 5.1bk
- Added test case to prevent future regressions.
