Bug #53333 [Fbk]: Random crash
| From: | uw@php.net | Date: | Tue, 30 Nov 2010 11:24:46 +0000 |
| Subject: | Bug #53333 [Fbk]: Random crash | ||
| References: | 1 | Groups: | php.bugs |
| Request: | Send a blank email to php-bugs+get-156576@lists.php.net to get a copy of this message | ||
Edit report at http://bugs.php.net/bug.php?id=53333&edit=1
ID: 53333
Updated by: uw@php.net
Reported by: pajoye@php.net
Summary: Random crash
Status: Feedback
Type: Bug
Package: MySQLi related
Operating System: *
PHP Version: 5.3.3
Assigned To: mysql
Block user comment: N
Private report: N
New Comment:
Hi Pierre,
can you provide a mysqlnd debug log, do you have access to the php.ini?
In 5.3.4 mysqlnd learned to write proper debug logs on Windows. Try
running PHP with mysqlnd.debug="d:t:O,C:\mysqlnd_debug.log". You need
5.3.4. 5.3.3 won't do it.
Thanks,
Ulf
Previous Comments:
------------------------------------------------------------------------
[2010-11-29 14:12:14] pajoye@php.net
Yes, it happens as well with current SVN (5.3.4RC1 too).
------------------------------------------------------------------------
[2010-11-29 14:11:19] pajoye@php.net
As I said it is not possible yet to get an isolated case. I would have
done it already if yes.
However, anyone able to do a post mortem analysis could see where and
maybe why the crash occurs. memory dumb available if you have such
developer (or someone willing to try, not too hard :).
------------------------------------------------------------------------
[2010-11-29 14:09:13] uw@php.net
Pierre,
of course, we'd love to see an isolated phpt test case. As you state, it
is not necessarily a mysqli issue but may have some other cause.
I agree that Andrey should look at the back trace.
Can this reproduced with 5.3.4? If so, please check if you can give us a
mysqlnd debug log. It is going to be *very* long. As a first step,
please provide us with no more than the last 1000 lines from that log.
1000 lines is a a bit of a random pick to limit the expected long debug
log.
In 5.3.4 mysqlnd learned to write proper debug logs on Windows. Try
running PHP with mysqlnd.debug="d:t:O,C:\mysqlnd_debug.log". You need
5.3.4. 5.3.3 won't do it.
Andrey,
can you have a look at the backtrace?
Thanks,
Ulf
------------------------------------------------------------------------
[2010-11-20 13:45:02] pajoye@php.net
Can reproduce the crash with 5.3.4RC1 as well. It happens while
switching apps (new DB connection with different credential). Updated
bt:
php5ts!_zend_mm_free_int+0x57
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_alloc.c @ 2018]
php5ts!_efree+0x2e
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_alloc.c @ 2351]
php5ts!zend_hash_destroy+0x27
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_hash.c @ 529]
php5ts!zend_object_std_dtor+0x2b
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_objects.c @ 46]
php_mysqli!mysqli_result_free_storage+0x3f
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\ext\mysqli\mysqli.c @ 270]
php5ts!zend_objects_store_del_ref_by_handle_ex+0x1c1
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_objects_api.c @
220]
php5ts!zend_objects_store_del_ref+0x1a
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_objects_api.c @
173]
php5ts!_zval_dtor_func+0x7f
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_variables.c @ 60]
php5ts!_zval_ptr_dtor+0x56
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_execute_api.c @
443]
php5ts!zend_leave_helper_SPEC+0x6c
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_vm_execute.h @ 160]
php5ts!ZEND_RETURN_SPEC_TMP_HANDLER+0xa3
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_vm_execute.h @
4970]
php5ts!execute+0x2e8
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_vm_execute.h @ 107]
php5ts!zend_call_function+0x8c9
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_execute_api.c @
964]
php5ts!zif_call_user_func_array+0x63
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\ext\standard\basic_functions.c
@ 4796]
php5ts!zend_do_fcall_common_helper_SPEC+0x920
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_vm_execute.h @ 316]
php5ts!ZEND_DO_FCALL_SPEC_CONST_HANDLER+0x11a
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_vm_execute.h @
1606]
php5ts!execute+0x2e8
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend_vm_execute.h @ 107]
php5ts!zend_execute_scripts+0xfe
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\zend\zend.c @ 1195]
php5ts!php_execute_script+0x24c
[d:\php-sdk\snap_5_3\vc9\x86\php-5.3.4rc1\main\main.c @ 2265]
------------------------------------------------------------------------
[2010-11-18 01:26:55] pajoye@php.net
Description:
------------
See the backtrace here http://pastie.org/1307118
The crash is hard to reproduce but happens regularly in our tests (using
mediawiki, wordpress or other apps).
Does it ring a bell to one of you (mysql devs :)?
It could also be a GC bug. I can send the memory dump if one of you like
to do a post morterm debugging.
------------------------------------------------------------------------
--
Edit this bug report at http://bugs.php.net/bug.php?id=53333&edit=1