Bug 26439

Summary:	[GMA3150] Errors while X startup
Product:	DRI	Reporter:	fangxun <xunx.fang>
Component:	DRM/Intel	Assignee:	Jesse Barnes <jbarnes>
Status:	CLOSED FIXED	QA Contact:
Severity:	major
Priority:	high	CC:	hodosiszabolcs
Version:	unspecified
Hardware:	All
OS:	Linux (All)
Whiteboard:
i915 platform:		i915 features:

Description fangxun 2010-02-05 00:36:25 UTC

System Environment:
--------------------------
Arch:           i386
Platform:       pinetrail
Libdrm:         (master)7b328122981a23f874105f333a5ec14ce887b0a8
Mesa:           (master)3c967a920718c2a74996ac23b8c1964915db88c2
Xserver:           (master)6f6a99abc12ddee82898fdabfb50c17e90e094b9
Xf86_video_intel:     (master)93cd943d41c646c794b8cb5a960d8f0805e15395
Kernel:  (master)7284ce6c9f6153d1777df5f310c959724d1bd446

Bug detailed description:
-------------------------
While X window start up, there are error info as following:
  (EE) intel(0): Failed to submit batch buffer, expect rendering corruption or even a frozen display: Input/output error.
(EE) intel(0): Failed to submit batch buffer, expect rendering corruption or even a frozen display: Input/output error.
(EE) intel(0): Failed to submit batch buffer, expect rendering corruption or even a frozen display: Input/output error.

And gnome desktop startup fails. With bisect and find a merge commit 635b3c9d5508d1dfe02ee5f882becea37e294111 is bad commit, but I am not sure if it is the first bad commit. By the way, commit 7959722b951cffcd61a0a35229d007deeed8c2dd is good commit.

Reproduce steps:
----------------
1.xinit&
2.gnome-session&

Comment 1 Gordon Jin 2010-02-06 03:00:33 UTC

moving to DRM component since it's a kernel regression.

Comment 2 Jesse Barnes 2010-02-06 10:26:21 UTC

Again, that's a merge commit, you have to bisect through that to find the real issue.  Also when reporting bisect results, please include the summary, e.g.

commit 2c1f1895ef2aa8f0e5497893eff71304aef332e1
Merge: 635b3c9 a81406b
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Wed Jan 6 20:26:42 2010 -0800

    Merge branch 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ai

for this one.

Comment 3 Jesse Barnes 2010-02-08 18:39:49 UTC

*** Bug 26066 has been marked as a duplicate of this bug. ***

Comment 4 Jesse Barnes 2010-02-08 18:40:36 UTC

I've heard reports that this issue may be libdrm related, specifically that a
checkout of libdrm from 11/25 works around this issue, can you confirm before
doing your kernel bisect?

Comment 5 Szabolcs Hodosi 2010-02-08 22:18:38 UTC

(In reply to comment #4)
> I've heard reports that this issue may be libdrm related, specifically that a
> checkout of libdrm from 11/25 works around this issue, can you confirm before
> doing your kernel bisect?
> 

I've used the latest intel driver with the latest libdrm (latest stable), which was recommended on the intel driver's web page. I mean the latest was around mid of january.

Comment 6 Gordon Jin 2010-02-09 00:59:49 UTC

Looking through the bisect range (already narrowed in the initial report), I bet it's caused by introducing the new intel_device_info which wrongly dropped .need_gfx_hws for pineview. And adding it back does fix this problem.

Later I found the same fix has already available in the latest drm-intel-next and 2.6.33-rc7:

commit 8a6c77d645d9111f7bc01f908464301f5c3e0a05
Author: Li Peng <peng.li@linux.intel.com>
Date:   Thu Jan 21 18:45:46 2010 +0800

    drm/i915: Fix the device info of Pineview
    
    Pineview doesn't has CXSR and need GTT-based hardware status page.
    It fixes a X boot hung issue on Pinview since commit cfdf1f

Comment 7 Gordon Jin 2010-02-09 01:00:08 UTC

verified with the latest drm-intel-next

Comment 8 Elizabeth 2017-10-06 14:54:20 UTC

Closing old verified.

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.