Xorg server using 100% CPU. Can reproduce. Need help to track the issue

[Chris Wilson]

On Wed, 12 May 2010 13:55:27 -0400, phil lemelin <phil.lemelin at gmail.com> wrote:
> I downloaded the 2.6.32.12 kernel and patched it since the patch would'nt
> work well with my 2.6.29.6 kernel and here is what I get after a couple of
> minutes playing with the magnifier. I seem to be able to play longer before
> experiencing the crash however.
> I can provide the corresponding Xorg.0.log if it is relevant.
> 
> [root at localhost ~]# /usr/local/bin/Xorg
> 
> X.Org X Server 1.7.6
> Release Date: 2010-03-17
> X Protocol Version 11, Revision 0
> Build Operating System: Linux 2.6.32.12 i686
> Current Operating System: Linux localhost 2.6.32.12 #4 SMP Wed May 12
> 10:12:55 EDT 2010 i686
> Kernel command line:
> Build Date: 12 May 2010  01:37:47PM
> 
> Current version of pixman: 0.15.20
>     Before reporting problems, check http://wiki.x.org
>     to make sure that you have the latest version.
> Markers: (--) probed, (**) from config file, (==) default setting,
>     (++) from command line, (!!) notice, (II) informational,
>     (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
> (==) Log file: "/usr/local/var/log/Xorg.0.log", Time: Wed May 12 13:44:57
> 2010
> (==) Using config file: "/etc/X11/xorg.conf"
> xf86TokenToOptinfo: table is NULL
> xf86TokenToOptinfo: table is NULL
> Setting master
> expected keysym, got XF86TouchpadToggle: line 122 of inet
> expected keysym, got XF86TouchpadToggle: line 122 of inet
> intel_bufmgr_gem.c:713: Error mapping buffer 65452 (pixmap): Cannot allocate
> memory .

So it seems to survive the original bug, only to trigger a later one by
hogging all the memory and causing ENOMEM. That segfault looks reminiscent
of:

[libdrm]
commit 08371bc29013370558728dcbeeed6a23ad2f5a70
Author: Chris Wilson <chris at chris-wilson.co.uk>
Date:   Tue Dec 8 22:35:24 2009 +0000

    intel: Clear virtual after failing to mmap_gtt.
    
    Don't store the error return in bo_gem->gtt_virtual or else we will
    attempt to use that as a valid pointer in future mappings.

I've pushed some improvements to xf86-video-intel that should prevent this
from occurring in the first place.

-- 
Chris Wilson, Intel Open Source Technology Centre