Bug 18880 - [945gm GEM]glean case/fbo cause oops due to OOM, with 512MB mem
Summary: [945gm GEM]glean case/fbo cause oops due to OOM, with 512MB mem
Status: VERIFIED FIXED
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i915 (show other bugs)
Version: unspecified
Hardware: Other All
: high critical
Assignee: Eric Anholt
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks: 18841 intel-3d-release
  Show dependency treegraph
 
Reported: 2008-12-03 19:20 UTC by liuhaien
Modified: 2009-03-08 19:23 UTC (History)
0 users

See Also:
i915 platform:
i915 features:


Attachments
dmesg (42.88 KB, text/plain)
2008-12-03 19:20 UTC, liuhaien
Details
xorg.0.log (65.38 KB, text/x-log)
2008-12-03 19:20 UTC, liuhaien
Details
xorg conf file (3.69 KB, text/plain)
2008-12-03 19:20 UTC, liuhaien
Details

Note You need to log in before you can comment on or make changes to this bug.
Description liuhaien 2008-12-03 19:20:03 UTC
Created attachment 20797 [details]
dmesg

System Environment:
--------------------------
Host:		x-945gm
Arch:		i386
OSD:		Fedora release 7 (Moonshine)
Libdrm:		(master)b0d93c74d884b40bd94469a5ef75fdb2fef17680
Mesa:		(master)154a9e5317f890618932cea0129ef887e16baf84
Xorg:		7.2
Xserver_stable:		(server-1.6-branch)b268458eab2f213ec14dfe8013aa714c187e3aab
Xf86_video_intel:		(master)07f5a8223187c1abc79c104d2fa5859a54cecd30
GEM_kernel:       (for-airlied)728ced8c47f99a2287cdd0d3e77f5ae1a3d410e6


Bug detailed description:
-------------------------
startx, then run glean/fbo ,we will get a kernel oops like below:


xterm:  fatal IO error 32 (Broken pipe) or KillClient on X server ":0.0"

Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel: Oops: 0000 [#1] SMP
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel: last sysfs file: /sys/class/drm/card0/dev
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel: Process glean (pid: 4330, ti=d78c4000 task=de96d4b0 task.ti=d78c4000)
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel: Stack:
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  de9a4924 e03005d2 e02fc52a 00000008 df3e8000 e0033130 00000008 de9a4924
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  d6c79940 de9a4924 c01644d6 d9d6f7c8 df020cc0 d6c79940 de5e71c0 00000000
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel: Call Trace:
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<e03004d1>] i915_gem_idle+0x1fe/0x2fa [i915]
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<e03005d2>] i915_gem_lastclose+0x5/0x1d [i915]
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<e02fc52a>] i915_driver_lastclose+0x13/0x33 [i915]
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<e0033130>] drm_lastclose+0x31/0x2e1 [drm]
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c01644d6>] __fput+0xb8/0x163
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c01622d9>] filp_close+0x4d/0x53
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c01232e6>] put_files_struct+0x60/0xa8
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c01242a4>] do_exit+0x179/0x65a
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c0129ae3>] __dequeue_signal+0xa9/0xd0
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c01247e0>] do_group_exit+0x5b/0x84
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c012bbea>] get_signal_to_deliver+0x307/0x31e
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c03fe849>] do_page_fault+0x0/0x574
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c0102638>] do_notify_resume+0x68/0x6fc
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c0150866>] __do_fault+0x2ea/0x32c
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c0151b0e>] handle_mm_fault+0x282/0x513
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c03feab2>] do_page_fault+0x269/0x574
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c03fe849>] do_page_fault+0x0/0x574
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c0102e8a>] work_notifysig+0x13/0x19
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel:  [<c03f0000>] tulip_read_eeprom+0x8a/0x12d
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel: Code: be 6c 0e 00 00 00 75 10 8d 86 38 0e 00 00 ba fa 00 00 00 e8 58 eb e2 df 5a 89 e8 5b 5e 5f 5d c3 57 56 89 c6 53 8b 5e 28 8b 40 08 <83> 7b 28 00 8b b8 48 02 00 00 8d 4b 08 8b 53 08 74 10 8b 41 04
Message from syslogd@ at Thu Dec  4 11:01:11 2008 ...
x-945gm kernel: EIP: [<e02ffc58>] i915_gem_object_move_to_inactive+0xb/0x69 [i915] SS:ESP 0068:d78c5de8
Comment 1 liuhaien 2008-12-03 19:20:32 UTC
Created attachment 20798 [details]
xorg.0.log
Comment 2 liuhaien 2008-12-03 19:20:54 UTC
Created attachment 20799 [details]
xorg conf file
Comment 3 Eric Anholt 2008-12-08 16:24:28 UTC
The original cause of your failure appears to be OOM condition on your system, which we recover from poorly.  Valid bug, and not one we'd see in development testing since we keep a lot of RAM in our systems.  Also one that'll go away with KMS.

However, for getting testing results without a bunch of false negatives, you'll want to populate the system with more than 512MB.
Comment 4 liuhaien 2008-12-09 23:16:09 UTC
(In reply to comment #3)
> The original cause of your failure appears to be OOM condition on your system,
> which we recover from poorly.  Valid bug, and not one we'd see in development
> testing since we keep a lot of RAM in our systems.  Also one that'll go away
> with KMS.
> 
> However, for getting testing results without a bunch of false negatives, you'll
> want to populate the system with more than 512MB.
> 

kernel oops will go away when increasing memory to 1536M. but the case glean/fbo still fails. 
Comment 5 Gordon Jin 2008-12-10 00:45:12 UTC
(In reply to comment #4)

> kernel oops will go away when increasing memory to 1536M. but the case
> glean/fbo still fails. 

This confirms Eric's explanation. Good.

case glean/fbo fail on this machine is not surprising, as it has been failed on all platforms, which needs a separate bug.
Comment 6 Gordon Jin 2009-03-04 19:37:58 UTC
Eric, I'm increasing priority since we've approaching Q1 release cycle and we need be strict for memory usage (considering there are still some old laptops and new netbooks using 512MB).
Comment 7 liuhaien 2009-03-05 00:13:12 UTC
hi, we retest it with 512M memory and the latest driver.kernel oops has gone now.
following is the commits:
Libdrm:		(master)a6dd0afa87558a670f970e61b023f45a396539eb
Mesa:		(mesa_7_4_branch)b65bfde84d2f0d83a432602cda425a63560e4034	
Xserver:	(server-1.6-branch)60c161545af80eb78eb790a05bde79409dfdf16e
Xf86_video_intel:		(master)14bb61e0c2e28725a2f6167d3263649bc845be18
GEM_kernel:       (for-airlied)edde72a59461d766997b469f6d20afdf5fe9b5b4
Comment 8 Gordon Jin 2009-03-05 00:38:45 UTC
Good, so I'm closing this.

Let's keep 512MB on this machine to see if it can find other issues caused by low memory.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.