SPORADICALLY ON PB2-5.5 FOR LINUX-VALGRIND
Description: Sporadic failure of variables-bug21503595 test
on pb2-5.5 for linux-valgrind platform.
Fix: This is a issue related to libc and not related to
MySQL code. During dlclose few blocks of memory left
unfreed. This is a known issue in libc and needs to be
suppressed.
Fix: Added a valgrind suppression.
Problem: CREATE TABLE using a fully qualified name with INDEX DIR/DATA DIR
option reports an error when the current database is not SET.
check_access() was incorrectly called with NULL as the database
argument in a situation where the database name was not needed for
the particular privilege being checked. This will cause the current
database to be used, or an error to be reported if there is no current
database.
Fix: Call check_access() with any_db as the database argument in this situation.
Fix of Bug#25088048 caused paths to be relative, not absolute, this
proved to be problematic.
Fix is to still ignore current working directory, however switch to
using full path of basedir, which is set to parent directory of bin/
directory where mysqld_safe is located.
References to legacy tool mysql_print_defaults are removed, only
my_print_defaults is used these days.
This will also fix:
Bug#11745176 (11192) MYSQLD_SAFE ONLY EVALUATES --DEFAULTS-FILE OPTION WHEN IT IS THE FIRST OP
Bug#23013510 (80866) MYSQLD_SAFE SHOULD NOT SEARCH $MY_BASEDIR_VERSION/VAR AS DATADIR
Bug#25244898 (84173) MYSQLD_SAFE --NO-DEFAULTS & SILENTLY DOES NOT WORK ANY MORE
Bug#25261472 (84219) INITSCRIPT ERRORS WHEN LAUCHING MYSQLD_SAFE IN NON DEFAULT BASEDIR
Bug#25319392 (84263) MYSQL.SERVER (MYSQL SERVER STARTUP SCRIPT) CAN NOT WORK,AND EXPORT SOME ERROR.
Bug#25319457 MYSQLD_SAFE MIGHT FAIL IF $DATADIR HAS TRAILING /
Bug#25341981 MYSQLD_SAFE ASSUMES INCORRECT BASEDIR WHEN EXECUTED WITH ABSOLUTE PATH
Bug#25356221 (84427) MYSQLD_SAFE FAILS TO START WHEN USING A FIFO FOR LOG-ERROR (REGRESSION)
Bug#25365194 (84447) MYSQLD_SAFE DOESN'T CHECK EXISTENCE OF GIVEN BASEDIR PARAMETER
Bug#25377815 ERRORS WHILE STARTING MYSQLD_SAFE WITH SYM LINK ENABLED
MYSQLD_SAFE
DESCRIPTION
===========
Starting a mysql server by running init script:
/etc/init.d/mysqld start
is failing. This is happening after the changes done in
script 'mysqld_safe' as a patch to Bug#24464380.
ANALYSIS
========
Say customer's /etc/my.cnf has following content:
[mysqld_safe]
.
.
ledir = /mysqld_ledir
mysqld = mysqld_wrapper
Patch to Bug#24464380 prohibits using "mysqld" (and few
other variables) in config file due to privilege reasons.
Since mysqld init scripts internally calls 'mysqld_safe'
script, the existing configuration has started failing.
FIX
===
In the init script, we now pass MYSQLD_OPTS as the first
argument (expected to be read from /etc/sysconfig/mysqld)
to mysqld_safe command. This new variable can have all the
mysqld_safe's special options as a string containing command
line arguments. For example:
MYSQLD_OPTS=" --ledir=/mysqld_ledir --mysqld=my_wrapper "
NOTE TO THE DOCUMENTATION TEAM
==============================
As mentioned above, the prohibited variables have to be
moved from /etc/my.cnf to /etc/sysconfig/mysqld as a string
containing command-line arguments in the form of variable
MYSQLD_OPTS. We can pass mysqld options as well in this new
variable which would be further passed to mysqld process.
This is backport of WL#7195 to MySQL-5.5. In 5.5, we
offload connection authentication from the acceptor
thread to tp worker threads.
Connection authentication happens in the acceptor thread that
accepts the connection for thread pool plugin. Connection authentication
involves exchanging packets with client and disk I/O
which is time consuming. This can cause other client
connections to starve and wait in the queue possibly increasing the
connect latency and decreasing throughput. In the worst case, some
connections could be dropped. n addition, SSL handshakes are quite
expensive and can stall connections in the accept queue.
This patch offloads connection authentication when thread pool
plugin is used for client connection. Each thread group
shall have a queue of connection_context objects, which represents
new connections that need to be processed by thread group threads.
The connection context is composed of THD object & list pointers
for intrusive queue implementation. Whenever a new connection
arrives, connection context object is created and added to the
queue. A new connect handler thread is created or woken up to handle
the authentication task. The worker thread loop is modified to
process connection events on connect handler threads in addition to
checking for query processing events. The initial number of connect
handler threads is one per thread group and it is restricted to
a maximum of 4 threads per thread group.
MY_THREAD_INIT IN BACKGROUND THREAD
Description:
===========
Add my_thread_init() and my_thread_exit() for background threads which
initializes and frees the st_my_thread_var structure.
Reviewed-by: Jimmy Yang<jimmy.yang@oracle.com>
RB: 15003
IS STARTING: CONFUSING ERROR
DESCRIPTION
===========
When mysql server processes transactions but has not yet
committed and shuts down abnormally (due to crash, external
killing etc.), a recovery is due from Storage engine side
which takes place the next time mysql server (either
through mysqld or mysqld_safe) is run.
While the 1st server is in mid of recovery, if another
instance of mysqld_safe is made to run, it may result into
2nd instance killing the 1st one after a moment.
ANALYSIS
========
In the "while true" loop, we've a check (which is done
after the server stops) for the existence of pid file to
enquire if it was a normal shutdown or not. If the file is
absent, it means that the graceful exit of server had
removed this file.
However if the file is present, the scripts makes a plain
assumption that this file is leftover of the "current"
server. It misses to consider that it could be a valid pid
file belonging to another running mysql server.
We need to add more checks in the latter case. The script
should extract the PID from this existing file and check if
its running or not. If yes, it means an older instance of
mysql server is running and hence the script should abort.
FIX
===
Checking the status of process (alive or not) by adding a
@CHECK_PID@ in such a case. Aborting if its alive. Detailed
logic is as follows:
- The mysqld_safe script would quit at start only as soon
as it finds that there is an active PID i.e. a mysql server
is already running.
- The PID file creation takes place after InnoDb recovery,
which means in rare case (when PID file isn't created yet)
it may happen that more than 1 server can come up but even
in that case others will have to wait till the 1st server
has released the acquired InnoDb lock. In this case all
these servers will either TIMEOUT waiting for InnoDb lock
or after this they would find that the 1st server is
already running (by reading $pid_file) and would abort.
- Our core fix is that we now check the status of mysql
server process (alive or not) after the server stops
running within the loop of "run -> shutdown/kill/abort ->
run ... ", so that only the script who owns the mysql
server would be able to bring it down if required.
NOTE
====
Removed the deletion of pid file and socket file from entry
of the loop, as it may result in 2nd instance deleting
these files created by 1st instance in RACE condition.
Compensated this by deleting these files at end of the loop
Reverted the changes made in patch to Bug#16776528. So
after this patch is pushed, the concept of mysqld_safe.pid
would go altogether. This was required as the script was
deleting other instance's mysqld_safe.pid allowing multiple
mysqld_safe instances to run in parallel. This patch would
fix Bug#16776528 as well as the resources would be guarded
anyway by InnoDb lock + our planned 5.7 patch.
Issue:
------
While using the LOAD statement to insert data into an
updateable view, the check to verify whether a column
is actually updatable is missing.
Solution for 5.5 and 5.6:
-------------------------
For a view whose column-list in specified in the LOAD
command, this check is not performed. This fix adds the
check.
This is a partial backport of Bug#21097485.
Solution for 5.7 and trunk:
---------------------------
For a view whose column-list is specified in the LOAD
command, this check is already performed. This fix adds the
same check when no column-list is specified.
The XML parser position stack for each level is with a fixed depth.
So a bounds check was done to ensure that this depth is not exceeded.
But it was off by one (i.e. the size of the array was a valid index).
Fixed by decreasing the allowable depth by one to match the maximum
number of elements in the position stack.
Description :
=============
When a MTR test run is started, it initializes the server and creates
the datadir under '$MYSQL_TEST_DIR/var'('/tmp/var' or '/dev/shm/var'
if --mem option is used) location and then copies it to the datadir
location of server(s).
If $parallel == 1, datadir location of the server is
'$MYSQL_TEST_DIR/var/data'. If $parallel > 1, datadir location of any
server is '$MYSQL_TEST_DIR/var/<thread_num>/data'.
This is the reason MTR searches for the initialized datadir in 2
locations('$opt_vardir' and '$opt_vardir/..') from the current vardir
location..
But this can cause few problems. If a directory with the name 'data'
already exists under '$MYSQL_TEST_DIR' and if the MTR run is started
with parallel value 1, then
1. copytree($install_db, '$opt_vardir/..') command will fail if the
user doesn't have the access permission to '$MYSQL_TEST_DIR/data'
directory.
2. Unnecessary contents from '$MYSQL_TEST_DIR/data' directory will be
copied to server datadir location and this might affect the server
startup.
Fix :
=====
Depending on the $parallel value decide whether the path for the
initialize datadir is "$opt_vardir"(i.e $parallel = 1) or
"$opt_vardir/.."(i.e $parallel > 1).
Reviewed-by: Deepa Dixit <deepa.dixit@oracle.com>
Reviewed-by: Srikanth B R <srikanth.b.r@oracle.com>
RB: 14773
DESCRIPTION
===========
Performing a pattern match of a Regex resulting into a very
large string, leads to crash due to integer wraparound.
ANALYSIS
========
doinsert() - The length calculated here (to copy the
number of bytes) comes out to be too large to be stored in
the "int" variable 'length'. We need to ensure that the
variable can accommodate large lengths.
FIX
===
'length' in doinsert() is now defined as of type "size_t"
instead of "int"
DESCRIPTION
===========
Performing a pattern match of a Regex resulting into a very
large string, leads to crash due to failed realloc().
ANALYSIS
========
dupl() calls enlarge(). It in turn calls realloc() for
pointer p->strip. This eventually fails due to OOM.
However we are still using the same pointer in memcpy()
causing a SEGFAULT!
FIX
===
1) In dupl(), checking for error code (which would be set
if realloc fails) immediately after call to enlarge().
Returning now with this error code.
2) Handling the same in the caller functions.
Don't read --ledir option from config file.
Ignore current working for finding location of mysqld
Remove use of chown/chmod in scripts.
Be helpful only when basedir is /var/log or /var/lib.
Removed unused systemd files for SLES.
Set explicit basedir in scripts.
Don't read --ledir option from config file.
Ignore current working for finding location of mysqld
Remove use of chown/chmod in scripts.
Be helpful only when basedir is /var/log or /var/lib.
Removed unused systemd files for SLES.
Set explicit basedir in scripts.
ANALYSIS:
=========
'CREATE TABLE' query with a large value for 'CONNECTION'
string reports an incorrect error.
The length of connection string is stored in .frm in two
bytes (max value= 65535). When the string length exceeds
the max value, the length is truncated to fit the two
bytes limit. Further processing leads to reading only a
part of the string as the length stored is incorrect. The
remaining part of the string is treated as engine type and
hence results in an error.
FIX:
====
We are now restricting the connection string length to 1024.
An appropriate error is reported if the length crosses this
limit.
NOTE:
=====
The 'PASSWORD' table option is documented as unused and
processed within a dead code. Hence it will not cause
similar issue with large strings.
LOAD DATA AT MASTER.
Revert "BUG#23080148 - BACKPORT BUG 14653594 AND BUG 20683959 TO"
This reverts commit 1d31f5b3090d129382b50b95512f2f79305715a1.
The commit causes replication incompatibility between minor revisions
and based on discussion with Srinivasarao, the patch is reverted.
Wrapper for mysql_config used in multilib installs modified to work as
intended, added more archs (aarch64, ppc64le, s390x, s390, sparc and
sparc64) to lists in fallback mode and use same script for EL and
Fedora.
Thanks to Alexey Kopytov for report and fix.
700101
ANALYSIS:
=========
To set the time 'start_time' of query in THD, current time
is obtained by calling 'gettimeofday()'. On Solaris
platform, due to some system level issues, time obtained is
invalid i.e. its either greater than 2038 (max signed value
to hold microseconds since 1970) or 1970 (0 microseconds
since 1970). In these cases, validation checks infer that
the 'start_time' is invalid and mysql server initiates the
shutdown process. But the reason for shutdown is not logged.
FIX:
====
We are now logging appropriate message when shutdown is
triggered in the above mentioned scenarios. Now, even if
the initial validation checks infer that the 'start_time'
is invalid, server shutdown is not initiated immediately.
Before initiating the server shutdown, the process of
setting 'start_time' and validating it is reiterated (for
max 5 times). If correct time is obtained in these 5
iterations then server continues to run.
- Remove use of touch and chmod.
- Restrict usage of chown to cases where target directory is /var/log.
- Due to limited feature set in /bin/sh on Solaris, /bin/bash will be
used on this platform.
- Give error if directory for UNIX socket file is missing.
- Privileged user should not log to files owned by different user
(mysqld will log as before).
Prevent GCC from moving a mach_read_from_4() before we have checked that
we have 4 bytes to read. The pointer may only point to a 1, 2 or 3
bytes in which case the code should not read 4 bytes. This is a
workaround to a GCC bug:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77673
Patch submitted by: Laurynas Biveinis <laurynas.biveinis@gmail.com>
RB: 14135
Reviewed by: Pawel Olchawa <pawel.olchawa@oracle.com>
RESTRICTED IN ALL GA RELEASES
Back port of WL#6782 to 5.5 and 5.6. This also includes
back port of Bug#20771331, Bug#20741572 and Bug#20770671.
Bug#24695274 and Bug#24679907 are also handled along with
this.
