Segfault on POST

Bug #697105 reported by Antono Vasiljev
52
This bug affects 9 people
Affects Status Importance Assigned to Milestone
Apache2 Web Server
Fix Released
High
apache2 (Mandriva)
Unknown
Critical
apache2 (Ubuntu)
Fix Released
High
Chuck Short
Natty
Fix Released
High
Chuck Short

Bug Description

Binary package hint: apache2

It seems that problem is well known:

https://qa.mandriva.com/show_bug.cgi?id=61384
https://issues.apache.org/bugzilla/show_bug.cgi?id=50190

(gdb) bt
#0 0x00000058 in ?? ()
#1 0x0018f024 in ap_discard_request_body (r=0x218eee38) at /build/buildd/apache2-2.2.16/modules/http/http_filters.c:1394
#2 0x0016f96f in default_handler (r=0x218eee38) at /build/buildd/apache2-2.2.16/server/core.c:3642
#3 0x0017b181 in ap_run_handler (r=0x218eee38) at /build/buildd/apache2-2.2.16/server/config.c:159
#4 0x0017b589 in ap_invoke_handler (r=0x218eee38) at /build/buildd/apache2-2.2.16/server/config.c:377
#5 0x0018ce58 in ap_process_request (r=0x218eee38) at /build/buildd/apache2-2.2.16/modules/http/http_request.c:282
#6 0x00189ba8 in ap_process_http_connection (c=0x218deb20) at /build/buildd/apache2-2.2.16/modules/http/http_core.c:190
#7 0x00182761 in ap_run_process_connection (c=0x218deb20) at /build/buildd/apache2-2.2.16/server/connection.c:43
#8 0x00193343 in process_socket (thd=0x218a54d0, dummy=0x218dbbe0) at /build/buildd/apache2-2.2.16/server/mpm/worker/worker.c:544
#9 worker_thread (thd=0x218a54d0, dummy=0x218dbbe0) at /build/buildd/apache2-2.2.16/server/mpm/worker/worker.c:894
#10 0x00ae8a35 in ?? () from /usr/lib/libapr-1.so.0
#11 0x00e3fe79 in start_thread (arg=0xb0748b70) at pthread_create.c:304
#12 0x006bd6ae in clone () at ../sysdeps/unix/sysv/linux/i386/clone.S:130
(gdb) ^CQuit
(gdb)

Tags: natty
Antono Vasiljev (antono)
Changed in apache2 (Ubuntu):
status: New → Confirmed
Revision history for this message
shankao (shankao) wrote :

In upstreams bug comments, is suggested that the bug can be a regression in gcc 4.5 optimizations [0]

This produces lots of problems with apache in natty (gcc 4.5). Should we mark it as medium importance?

[0] http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46488

shankao (shankao)
Changed in apache2 (Ubuntu):
assignee: nobody → shankao (shankao)
status: Confirmed → In Progress
shankao (shankao)
tags: added: natty
Revision history for this message
shankao (shankao) wrote :

Here a patch that fixes the problem, taken from mandriva (thanks!)

Changed in apache2 (Ubuntu):
status: In Progress → New
Revision history for this message
shankao (shankao) wrote :

I'm sending the patch upstream

Revision history for this message
Stefan Fritsch (sf-sfritsch) wrote :

I haven't really looked into this issue in detail and don't know yet if the proposed fix is ok.

But in case you want to ahead with it: Your patch is against the wrong package. srclib/ in apache2 is never compiled. You would have to patch the apr package and rebuild apache2 with the fixed libapr1-dev.

Revision history for this message
shankao (shankao) wrote :

That's weird, the proposed patch is also applied in my personal PPA [0] (where I did the testing) and apache is now working perfectly in my system.

[0] - https://launchpad.net/~shankao/+archive/shankao-test

Revision history for this message
Stefan Fritsch (sf-sfritsch) wrote :

That's weird indeed. Maybe the package in your PPA and the package in natty were compiled with slightly different versions of gcc? Are the buildlogs available somewhere?

Revision history for this message
shankao (shankao) wrote :

Sure, every build in launchpad has the logs avaliable in the corresponding "builds" section [0], and it sould have being with the last natty versions as the build process takes them at the beginning (see the logs). I have made those builds also in my own computer and with the same build tools it crashes without the patch and works with it.

[0] https://launchpad.net/~shankao/+archive/shankao-test/+builds?build_state=built
The working patch is in the "shankao2" logs, I tested another proposed solution in the shankao1 version :)

Revision history for this message
o00o (lbr-linux) wrote :

Shankao thank your packages fix my problem

Revision history for this message
TorrentialStorm (x-admin-torrentialstorm-net-deactivatedaccount) wrote :

Making shankao's changes to libapr and rebuilding apache2 using the new libapr1-dev fixes the segfault for me.

Dave Walker (davewalker)
Changed in apache2 (Ubuntu):
status: New → Confirmed
importance: Undecided → High
Chuck Short (zulcss)
Changed in apache2 (Ubuntu Natty):
assignee: shankao (shankao) → Chuck Short (zulcss)
Dave Walker (davewalker)
Changed in apache2 (Ubuntu Natty):
milestone: none → natty-alpha-2
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package apache2 - 2.2.16-6ubuntu2

---------------
apache2 (2.2.16-6ubuntu2) natty; urgency=low

  * debian/rules: Use "-fno-strict-aliasing" to work around a gcc bug.
   (LP: #697105)
 -- Chuck Short <email address hidden> Tue, 25 Jan 2011 11:14:58 -0500

Changed in apache2 (Ubuntu Natty):
status: Confirmed → Fix Released
Revision history for this message
Antono Vasiljev (antono) wrote :

Thank You guys!

Changed in apache2:
importance: Unknown → High
status: Unknown → Confirmed
Changed in apache2 (Mandriva):
importance: Unknown → Critical
Changed in apache2:
status: Confirmed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.