You are viewing a plain text version of this content. The canonical link for it is here.
Posted to apreq-dev@httpd.apache.org by Karsten Kroesch <Ka...@trivadis.com> on 2009/02/26 17:20:40 UTC
Apache HTTPD segfaults w/ mod_perl 1.0
Hi!
I'm running an Apache 1.3.39 with mod_perl 1.0 freshly compiled by myself on Solaris 10 (The Perl application I'm dealing with, refused to work on Apache 2.x and mod_perl 2, though using compat. mode).
The application seems to work fine in the first place, but then showed some odd behaviour. The error_log says: "[notice] child pid 14240 exit signal Segmentation Fault (11), possible coredump in /etc/httpd" and when I enabled core dumping, I see the stack trace attached below. I cannot see whether a script is causing the failure.
Can anyone give me a hint where to search for the reason? Is it the fault of a script or the mod_perl lib itself?
Any help is appreciated; thanks in advance!
Karsten
-------8<--------------
bash-3.00# pstack core.httpd.14240
core 'core.httpd.14240' of 14240: /opt/apache-1.3.39/bin/httpd -DSSL
001322e4 Perl_av_undef (0, 1, a696c, 12ff24, 20bdf8, fef71414) + 4
0005a634 perl_shutdown (20e038, 23b240, 0, 0, 0, 0) + b0
0005a80c perl_child_exit (20e038, 23b240, ff222a00, 0, 0, 0) + 108
0005a824 perl_child_exit_cleanup (23b408, 1084, ffbfcb18, fffffffc, ff222a00, 1000) + 8
0007b118 run_cleanups (23b420, ffbfcb1c, 0, 10, fc00, fef73b00) + 24
00078994 ap_clear_pool (23b240, 1, 0, 0, 0, 0) + a8
00078a2c ap_destroy_pool (23b240, 20e038, 0, 0, 0, 0) + 14
0008f4a8 clean_child_exit (0, 0, 0, 0, 0, 0) + 70
00092728 just_die (10, 0, 0, 0, 0, 0) + 3c
0009275c usr1_handler (10, 0, ffbfce90, 1, 69626578, 65632f6d) + 24
fef44b28 __sighndlr (10, 0, ffbfce90, 92738, 0, 1) + c
fef39b00 call_user_handler (10, 0, 0, 0, ff222a00, ffbfce90) + 3b8
fef458e0 _read (4, 2b6088, 1000, 0, 0, 2dab89) + c
00040070 ssl_io_hook_read (2b6040, 2b6088, 1000, 6062feff, 80808080, 1010101) + 140
000bb1a4 ap_hook_call_func (ffbfd520, 216868, 218b80, 10, 10b4, fef73b00) + d30
000ba398 ap_hook_call (1cdec8, ffbfd53c, 2b6040, 2b6088, 1000, 7) + bc
0007ca14 ap_read (2b6040, 2b6088, 1000, 0, fef73700, ff222a00) + 2c
0007d488 buff_read (2b6040, 2b6088, 1000, 0, 0, 0) + 1c
0007d40c saferead_guts (2b6040, 2b6088, 1000, 0, 0, 0) + 40
0007d348 read_with_errors (2b6040, 2b6088, 1000, fffffff8, 0, 2db2c1) + 1c
0007d93c ap_bgets (ffbfd84c, 2000, 2b6040, fffffff8, 0, 2dab89) + dc
0009954c ap_getline (ffbfd84c, 2000, 2b6040, 0, 0, 2dad29) + 2c
00099af8 read_request_line (2dab80, 2dab80, 0, 10, 10b4, fef73b00) + 144
0009a76c ap_read_request (2fa1d0, 92738, 0, ffbff968, ffbff978, 7) + 230
00094fe0 child_main (7, 926ec, 0, 0, fef73700, ff222a00) + bec
00095330 make_child (20e038, 7, 49a6b574, 0, ffbffb34, ff222a00) + 1ac
00095934 perform_idle_server_maintenance (ffffffff, 0, 0, 20e038, 1d3f90, 1d2a60) + 4c0
00096334 standalone_main (2, ffbffd04, 2029e0, ffffffff, a4c50, 78ec0) + 674
00096d38 main (2, ffbffd04, ffbffd10, 20bdf8, ff220100, ff220140) + 5d0
00033b18 _start (0, 0, 0, 0, 0, 0) + 5c
RESOLVED: Apache HTTPD segfaults w/ mod_perl 1.0
Posted by Karsten Kroesch <Ka...@trivadis.com>.
The fix mentioned below works for me; additionally, nasty segfaults (plus core dumps) when restarting httpd have gone. The Perl version used by me is 5.10.0.
Have a good day!
Karsten
-----Original Message-----
From: Steve Hay [mailto:SteveHay@planit.com]
Sent: Do 2/26/2009 5:37
To: Karsten Kroesch; apreq-dev@httpd.apache.org
Cc: TVD_AD; dev@perl.apache.org
Subject: RE: Apache HTTPD segfaults w/ mod_perl 1.0
Karsten Kroesch wrote:
> Hi!
>
> I'm running an Apache 1.3.39 with mod_perl 1.0 freshly compiled by
> myself on Solaris 10 (The Perl application I'm dealing with, refused
> to work on Apache 2.x and mod_perl 2, though using compat. mode).
>
> The application seems to work fine in the first place, but then
> showed some odd behaviour. The error_log says: "[notice] child pid
> 14240 exit signal Segmentation Fault (11), possible coredump in
> /etc/httpd" and when I enabled core dumping, I see the stack trace
> attached below. I cannot see whether a script is causing the failure.
>
> Can anyone give me a hint where to search for the reason? Is it the
> fault of a script or the mod_perl lib itself?
>
> Any help is appreciated; thanks in advance!
>
> Karsten
>
> -------8<--------------
>
> bash-3.00# pstack core.httpd.14240
> core 'core.httpd.14240' of 14240: /opt/apache-1.3.39/bin/httpd
> -DSSL 001322e4 Perl_av_undef (0, 1, a696c, 12ff24, 20bdf8, fef71414)
> + 4 0005a634
What version of perl are you using? If its 5.10.x then from your stack
trace I would guess that you've been bitten by the bug that was fixed
here:
http://svn.apache.org/viewvc?view=rev&revision=555908
Unfortunately, that was committed just after mod_perl-1.30 was released
and 1.31 still hasn't been released yet. Try manually applying that
change to your mod_perl source and rebuilding it (or else get the
current dev version from SVN).
RE: Apache HTTPD segfaults w/ mod_perl 1.0
Posted by Steve Hay <St...@planit.com>.
Karsten Kroesch wrote:
> Hi!
>
> I'm running an Apache 1.3.39 with mod_perl 1.0 freshly compiled by
> myself on Solaris 10 (The Perl application I'm dealing with, refused
> to work on Apache 2.x and mod_perl 2, though using compat. mode).
>
> The application seems to work fine in the first place, but then
> showed some odd behaviour. The error_log says: "[notice] child pid
> 14240 exit signal Segmentation Fault (11), possible coredump in
> /etc/httpd" and when I enabled core dumping, I see the stack trace
> attached below. I cannot see whether a script is causing the failure.
>
> Can anyone give me a hint where to search for the reason? Is it the
> fault of a script or the mod_perl lib itself?
>
> Any help is appreciated; thanks in advance!
>
> Karsten
>
> -------8<--------------
>
> bash-3.00# pstack core.httpd.14240
> core 'core.httpd.14240' of 14240: /opt/apache-1.3.39/bin/httpd
> -DSSL 001322e4 Perl_av_undef (0, 1, a696c, 12ff24, 20bdf8, fef71414)
> + 4 0005a634
What version of perl are you using? If its 5.10.x then from your stack
trace I would guess that you've been bitten by the bug that was fixed
here:
http://svn.apache.org/viewvc?view=rev&revision=555908
Unfortunately, that was committed just after mod_perl-1.30 was released
and 1.31 still hasn't been released yet. Try manually applying that
change to your mod_perl source and rebuilding it (or else get the
current dev version from SVN).
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@perl.apache.org
For additional commands, e-mail: dev-help@perl.apache.org
RE: Apache HTTPD segfaults w/ mod_perl 1.0
Posted by Steve Hay <St...@planit.com>.
Karsten Kroesch wrote:
> Hi!
>
> I'm running an Apache 1.3.39 with mod_perl 1.0 freshly compiled by
> myself on Solaris 10 (The Perl application I'm dealing with, refused
> to work on Apache 2.x and mod_perl 2, though using compat. mode).
>
> The application seems to work fine in the first place, but then
> showed some odd behaviour. The error_log says: "[notice] child pid
> 14240 exit signal Segmentation Fault (11), possible coredump in
> /etc/httpd" and when I enabled core dumping, I see the stack trace
> attached below. I cannot see whether a script is causing the failure.
>
> Can anyone give me a hint where to search for the reason? Is it the
> fault of a script or the mod_perl lib itself?
>
> Any help is appreciated; thanks in advance!
>
> Karsten
>
> -------8<--------------
>
> bash-3.00# pstack core.httpd.14240
> core 'core.httpd.14240' of 14240: /opt/apache-1.3.39/bin/httpd
> -DSSL 001322e4 Perl_av_undef (0, 1, a696c, 12ff24, 20bdf8, fef71414)
> + 4 0005a634
What version of perl are you using? If its 5.10.x then from your stack
trace I would guess that you've been bitten by the bug that was fixed
here:
http://svn.apache.org/viewvc?view=rev&revision=555908
Unfortunately, that was committed just after mod_perl-1.30 was released
and 1.31 still hasn't been released yet. Try manually applying that
change to your mod_perl source and rebuilding it (or else get the
current dev version from SVN).