Bug #26774 | Crash on ARM platform even after REPAIR TABLE | ||
---|---|---|---|
Submitted: | 1 Mar 2007 22:16 | Modified: | 29 Oct 2007 16:41 |
Reporter: | Christian Hammers (Silver Quality Contributor) (OCA) | Email Updates: | |
Status: | No Feedback | Impact on me: | |
Category: | MySQL Server | Severity: | S3 (Non-critical) |
Version: | 5.0.32, 5.0.37 | OS: | Linux (Debian GNU/Linux etch) |
Assigned to: | CPU Architecture: | ARM | |
Tags: | qc |
[1 Mar 2007 22:16]
Christian Hammers
[1 Mar 2007 22:18]
Christian Hammers
dump of database wpg2
Attachment: wpg2.dmp.bz2 (application/x-bzip, text), 84.28 KiB.
[1 Mar 2007 22:43]
[ name withheld ]
for an eaysier test than the database sample that was send, the test describes in this bug works fine (in the way that I can reproduce the crash with it) : http://bugs.mysql.com/bug.php?id=25172 It could be architecture dependant as I am on a Linksys NSLU2 hardware, using an ARM arcitecture.
[2 Mar 2007 6:54]
Valeriy Kravchuk
Thank you for a problem report. As you identified bug #25172 as related, maybe, you should just wait for 5.0.36 sources/Debian build of 5.0.36 to be released officially, and check if that version will solve your problem also?
[3 Mar 2007 0:29]
[ name withheld ]
Here is the gdb debugging result : Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 180236 (LWP 5297)] 0x001db2bc in JOIN::exec () (gdb) backtrace #0 0x001db2bc in JOIN::exec () #1 0x001dc9a4 in mysql_select () #2 0x001dd1c8 in handle_select () #3 0x001941fc in mysql_execute_command () #4 0x001973e0 in mysql_parse () #5 0x001979e8 in dispatch_command () #6 0x00198ab8 in do_command () #7 0x00199570 in handle_one_connection () #8 0x40074fb0 in pthread_start_thread () from /lib/libpthread.so.0 Backtrace stopped: frame did not save the PC > Oh and try the precompiled binaries you get from dev.mysql.com. I really haven't seen the arm architecture (or it have an other name there). It is sometimes called ixp4, but I can't see it either.
[3 Mar 2007 0:32]
[ name withheld ]
The corrective patch mentionned in the (what I thought) related bug is supposed to be present in the 5.0.32 version on debian (dixit C. Hammers that is in charge of the mysql package in debian).
[3 Mar 2007 8:03]
Valeriy Kravchuk
Martin, Please, try to build 5.0.36 from sources and check with it. We do not have this platfrom here.
[7 Mar 2007 23:41]
[ name withheld ]
Hi, and where are the 5.0.36 sources? I only fond the 5.0.33 or the 5.1.16. Regards Martin
[8 Mar 2007 9:02]
Valeriy Kravchuk
Please, read http://dev.mysql.com/doc/refman/5.0/en/installing-source-tree.html on how to get 5.0.36 sources from development trees or wait for 5.0.35 sources released officially on the same page where we have 5.0.33 now.
[8 Mar 2007 21:29]
[ name withheld ]
Awaiting for your answer (very fast as usual), I downloaded and compiled the 5.0.33 version. As my hardware can be qualified of "tiny", it took 12h00 to obtain the binaries (compiled withe the "--debug" option). It works fine. So, regarding how long it takes to compile it, unless you have a strong argument to convince me to compile a newer version, I consider that it's ok for me Regards Martin
[9 Mar 2007 6:39]
Valeriy Kravchuk
5.0.37 will be officially released (instead of 5.0.35) really soon. But if you do not need any fixes it will contain (please, check the manual), 5.0.33 may just be an answer for you. I think, this report can be closed.
[25 Apr 2007 19:09]
[ name withheld ]
I kept on investigating. Thanks to my linux mailing list, I ended up by finding that the -02 CFLAGS compilation option causes this crash. But, I was told that, if there is a crash with the O2 option, it means that there is a piece of code that is poorly written. So, I am compiling mysql with all compilers warnings and mysql --with-debug="full" option. I hope it will help correcting the code. I'll send all logs tomorrow
[29 Apr 2007 20:30]
[ name withheld ]
last mysql debugs between the last successfull command and the server crash
Attachment: crash.log (text/x-log), 4.00 KiB.
[29 Apr 2007 20:33]
[ name withheld ]
full mysql debugging logs from the lient startup
Attachment: full_crash.log.tar.gz (application/x-gzip, text), 127.99 KiB.
[29 Apr 2007 20:34]
[ name withheld ]
Compilation log
Attachment: compile_debug.log.gz (application/x-gzip, text), 137.84 KiB.
[29 Apr 2007 20:43]
[ name withheld ]
If additionnal logs are required, please let me know (and tell me how to obtain them) Regards Martin Riondet
[29 Apr 2007 21:34]
[ name withheld ]
mysqld logs (probably more interesting). Complete log on request.
Attachment: crash_part_mysqld.log.gz (application/x-gzip, text), 17.16 KiB.
[2 May 2007 17:57]
Christian Hammers
Hi! I reopen this bug as the original submitter said that the crash does occur even on 5.0.38 when compiled with -O2 (his statement that 5.0.36 was fine most probably was due to a self compile without any -O options). bye, -christian-
[29 Sep 2007 16:41]
Valeriy Kravchuk
Please, try to repeat with a newer version, 5.0.45, and inform about the results.
[21 Oct 2007 12:47]
[ name withheld ]
Thank you. I works now, using directely the debian package, that is (I guess) compiled with the optimisation options. Regards Martin Riondet.
[30 Oct 2007 0:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".