Bug 578577

Summary: Race condition in libXext
Product: [openSUSE] openSUSE 11.2 Reporter: Lei Zhang <thestig>
Component: X.OrgAssignee: Stefan Dirsch <sndirsch>
Status: RESOLVED FIXED QA Contact: E-mail List <xorg-maintainer-bugs>
Severity: Normal    
Priority: P2 - High CC: antoine.mechelynck, ast, dmueller, meissner
Version: Final   
Target Milestone: ---   
Hardware: PC   
OS: openSUSE 11.2   
Whiteboard: maint:released:11.2:33026
Found By: --- Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---

Description Lei Zhang 2010-02-10 00:50:13 UTC
User-Agent:       Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.0.17) Gecko/2010010604 Ubuntu/8.04 (hardy) Firefox/3.0.17

There's a race condition in libXext that causes apps that use the X shared memory extensions to occasionally crash. [1][2] This has been fixed upstream with a small patch. [3] Can we cherrypick this for all affected releases? (All releases up to 11.2.)

[1] http://lists.freedesktop.org/archives/xcb/2009-October/005102.html
[2] http://crbug.com/25324
[3] http://cgit.freedesktop.org/xorg/lib/libXext/commit/?id=956fd30e1046e5779ac0b6c07ec4f0e87250869a

Reproducible: Didn't try

Steps to Reproduce:
1.
2.
3.
Comment 1 Stefan Dirsch 2010-02-17 13:36:34 UTC
I'm going to fix this for openSUSE:Factory, i.e. upcoming openSUSE 11.3.
Comment 2 Stefan Dirsch 2010-02-17 21:39:58 UTC
The fix is already included in libxcb 1.1.1, which we use for openSUSE:Factory, i.e. upcoming openSUSE 11.3.
Comment 3 Lei Zhang 2010-02-17 21:57:13 UTC
Indeed, this is in openSUSE 11.3. Which is why the original bug report asks to cherrypick the fix into the affected releases - OpenSUSE release up to 11.2.
Comment 4 Stefan Dirsch 2010-02-17 23:31:17 UTC
We only do updates for released products for security or critical issues. Why do you think this is one of these?
Comment 5 Lei Zhang 2010-02-17 23:58:56 UTC
I am not aware of this policy. Good to know.

I believe this fixes the assertion failure for xcb_io.c line 378. If you search it, you'll see there's tons of bug reports for that bug. Example [4]. It affects lots of X programs, including Evolution, Firefox, various parts of Gnome, etc. I wouldn't say it's critical, but it seems pretty serious for affected users.

[4] https://bugs.launchpad.net/ubuntu/+source/libx11/+bug/458989
Comment 6 Stefan Dirsch 2010-02-18 10:19:57 UTC
I've now submitted it also for upcoming SLE11-SP1.
Comment 7 Stefan Dirsch 2010-02-18 10:26:00 UTC
@openSUSE maintenace team
Should we do an online update to address this one for openSUSE 11.2?

@Lei
If you feel as part of the openSUSE community, you can help with the maintenance update for openSUSE 11.2 (in case an update gets accepted)

You can find more information on these wiki pages:

* http://en.opensuse.org/Maintenance
* http://en.opensuse.org/Maintenance/Policy
* http://en.opensuse.org/Build_Service/Concepts/Maintenance
Comment 8 Stefan Dirsch 2010-02-18 10:26:58 UTC
A possible update for openSUSE 11.0/11.1 needs to be decided by Anja/Dirk. Adding to Cc.
Comment 9 Christian Dengler 2010-02-18 10:55:45 UTC
I think we can arrange it with the new maintenance policy of 11.2 to do an update 
here, if Lei will submit an update.

So for an update: +1



For the older releases and their stricter policy, I will avoid to trigger an update now. Here we will pick the fix up with the next package update (planned for the next update).
Comment 10 Marcus Meissner 2010-02-18 10:59:25 UTC
would be fine for me for 11.2 too.
Comment 11 Christian Dengler 2010-02-18 11:06:24 UTC
So Lei, would you do the update? :-)
Comment 12 Stefan Dirsch 2010-02-18 11:16:45 UTC
(In reply to comment #9)
> For the older releases and their stricter policy, I will avoid to trigger an
> update now. Here we will pick the fix up with the next package update (planned
> for the next update).

Not sure what this means.
Comment 13 Christian Dengler 2010-02-18 11:23:34 UTC
The maintenance policy and the requirements for updates were loosened for 11.2.
Comment 19 Lei Zhang 2010-02-18 19:41:37 UTC
I'd be happy to submit an update, assuming it means using the osc tool to pull the package source, modify and commit. This is my first time so it'll be a learning experience. I'm going to the So Cal Linux Expo for the next 3 days, so I won't have time to actually do this until next week.

If that's ok, then I'll do the update next week. If someone else can do it, that's fine with me too.
Comment 20 Stefan Dirsch 2010-02-18 23:22:10 UTC
Wonderful. Thanks for taking over this task. :-)
Comment 21 Stefan Dirsch 2010-04-28 03:13:30 UTC
(In reply to comment #20)
> Wonderful. Thanks for taking over this task. :-)

Looks like I've been too optimistic here. I just submitted the fix for openSUSE:11.2:Update:Test (SR #38965).
Comment 22 Lei Zhang 2010-04-28 03:19:20 UTC
Sorry, life got busy, and the one time I played with osc, I didn't get far enough. Thanks for submitting the fix.
Comment 23 Stefan Dirsch 2010-04-28 03:19:46 UTC
@maintenance team: Could you create a SWAMPID for that issue on 11.2? Thanks.
Comment 24 Marcus Meissner 2010-04-28 06:48:32 UTC
lets do it.
Comment 25 Swamp Workflow Management 2010-04-28 11:51:02 UTC
The SWAMPID for this issue is 33022.
This issue was rated as ${rating}.
Please submit fixed packages as soon as possible.
Also create a patchinfo file using this link:
https://swamp.suse.de/webswamp/wf/33022
Comment 26 Christian Dengler 2010-04-28 11:53:30 UTC
SwampID provided. Be so kind and also submit a patchinfo. TIA
Comment 27 Stefan Dirsch 2010-04-28 12:59:45 UTC
Packages and patchinfo are now submitted.

openSUSE:11.2:Update:Test (SR #38965)
https://swamp.suse.de/webswamp/swamp/template/DisplayWorkflow.vm/workflowid/33026
Comment 30 Tony Mechelynck 2010-04-29 16:05:29 UTC
When I saw this patch on the 11.2 Update-Test repo, I thought it might perhaps cure the sporadic bug I was having, https://bugzilla.mozilla.org/show_bug.cgi?id=550810 -- i.e., sometimes, even when I'm away, SeaMonkey gets (got?) killed with no stack trace and only the line "seamonkey-bin: Fatal IO error 10 (No child processes) on X server :0.0." at the end of an otherwise variable sysout/syserr log (often, but not always, after up to thousands of "XID collision, trouble ahead" warnings). However, the symptoms seem different from those mentioned in the previous comments, and I have neither a stack trace nor a reproducible testcase.

Comments welcome, and if you don't want to spam either bug, my email addy is authentic (and the same @gmail.com works too).

In reply to comment #5: For a user like me, any bug that causes crashes, hangs, or dataloss, even sporadically, is critical stuff. Now maybe SuSE has a different definition. ;-)
Comment 31 Marcus Meissner 2010-04-29 16:12:16 UTC
tony, please open new bugs for new issues then.
Comment 32 Tony Mechelynck 2010-04-29 18:58:01 UTC
(In reply to comment #31)
> tony, please open new bugs for new issues then.

OK, if the problem appears again I will.
Comment 33 Swamp Workflow Management 2010-05-04 11:52:19 UTC
Update released for: xorg-x11-libXext, xorg-x11-libXext-debuginfo, xorg-x11-libXext-debuginfo-32bit, xorg-x11-libXext-debuginfo-x86, xorg-x11-libXext-debugsource, xorg-x11-libXext-devel
Products:
openSUSE 11.2 (debug, i586, x86_64)