MySQL Bugs: #46347: Segmentation fault in main.subselect

Bug #46347	Segmentation fault in main.subselect_gis
Submitted:	23 Jul 2009 10:22	Modified:	21 Dec 2009 12:50
Reporter:	Lenz Grimmer	Email Updates:
Status:	Can't repeat	Impact on me:	None
Category:	MySQL Server: GIS	Severity:	S3 (Non-critical)
Version:	5.1.35-snapshot20090630	OS:	Linux (openSUSE 11.0)
Assigned to:	Alexey Botchkov	CPU Architecture:	Any
Tags:	crash, gis, testfailure

Description:
I compiled a 32bit Linux binary using the 5.1.35-snapshot20090630 sources taken from pushbuild (based on the mysql-5.1-wl1326 source tree). Running the entire GIS tests results in a segfault when running the main.subselect_gis test - see the attached log file for a backtrace (BugDB would not let me post it in the description).

Interestingly, this only happens if there are other tests run before the failing test. Running subselect_gis alone passes, and I did not observe this failure in pushbuild as well:

[SNIP]
lenz@thebe:/mypool/space/tmp/mysql-5.1.35-snapshot20090630/mysql-test> ./mysql-test-run subselect_gis.test
Logging: ./mysql-test-run  subselect_gis.test
090723 11:55:06 [Warning] Forcing shutdown of 1 plugins
MySQL Version 5.1.35
Checking supported features...
 - skipping ndbcluster, mysqld not compiled with ndbcluster
 - SSL connections supported
Collecting tests...
vardir: /mypool/space/tmp/mysql-5.1.35-snapshot20090630/mysql-test/var
Checking leftover processes...
Removing old var directory...
Creating var directory '/mypool/space/tmp/mysql-5.1.35-snapshot20090630/mysql-test/var'...
Installing system database...
Using server port 15846

==============================================================================

TEST                                      RESULT   TIME (ms)
------------------------------------------------------------

worker[1] Using MTR_BUILD_THREAD 250, with reserved ports 12500..12509
main.subselect_gis                       [ pass ]   2006
------------------------------------------------------------
The servers were restarted 0 times
Spent 2.006 of 19 seconds executing testcases

All 1 tests were successful.
[SNIP]

How to repeat:
Take the mysql-5.1.35-snapshot20090630.tar.gz source tarball (from pushbuild) and compile it using the following options:

CFLAGS='-g -O2 -Wall -DUNIV_LINUX' CXX='g++' CXXFLAGS='-g -O2 -Wall -Wall   -fno-implicit-templates -fno-exceptions -fno-rtti'  LDFLAGS='-g -rdynamic '  ASFLAGS='-g' ./configure --enable-thread-safe-client --enable-local-infile --with-pic --with-client-ldflags=-static --with-mysqld-ldflags=-static --with-zlib-dir=bundled --without-ndb-debug --with-big-tables --with-ssl --with-readline --with-embedded-server --with-archive-storage-engine --with-blackhole-storage-engine --with-csv-storage-engine --with-example-storage-engine --with-federated-storage-engine --with-partition --with-extra-charsets=all --with-innodb

Now run the subselect_gis test after another test:

$ cd mysql-test
$ ./mysql-test-run gis.test subselect_gis.test

Observe the crash documented above.

Suggested fix:
No idea how to fix this - let me know if you need any additional information or a precompiled binary. I used gcc 4.3.1 for this build.

Console output incl. stack trace of the segfault

Attachment: bug46347.txt (text/plain), 12.41 KiB.

I just re-tested it with a more recent snapshot (mysql-5.1.43-snapshot20091222) and now it passes for me, too:

lenz@thebe:/tmp/mysql-5.1.43-snapshot20091222/mysql-test> ./mysql-test-run gis.test subselect_gis.test
Logging: ./mysql-test-run  gis.test subselect_gis.test
100205 21:15:58 [Note] Plugin 'FEDERATED' is disabled.
MySQL Version 5.1.43
Checking supported features...
 - skipping ndbcluster, mysqld not compiled with ndbcluster
 - SSL connections supported
Collecting tests...
vardir: /tmp/mysql-5.1.43-snapshot20091222/mysql-test/var
Removing old var directory...
Creating var directory '/tmp/mysql-5.1.43-snapshot20091222/mysql-test/var'...
Installing system database...
Using server port 48504

==============================================================================

TEST                                      RESULT   TIME (ms)
------------------------------------------------------------

worker[1] Using MTR_BUILD_THREAD 300, with reserved ports 13000..13009
main.gis                                 [ pass ]   1869
main.subselect_gis                       [ pass ]     53
------------------------------------------------------------
The servers were restarted 0 times
Spent 1.922 of 10 seconds executing testcases

All 2 tests were successful.

So I guess we can consider this bug as resolved. Thanks!