Bug 78411

Summary:	wine causes "Failed to submit rendering commands" since 5cc9ed4
Product:	Drivers	Reporter:	Christoph Haag (haagch.christoph)
Component:	Video(DRI - Intel)	Assignee:	Imre Deak (imre.deak)
Status:	RESOLVED CODE_FIX
Severity:	normal	CC:	chris, intel-gfx-bugs
Priority:	P2
Hardware:	All
OS:	Linux
Kernel Version:	3.16-rc1	Subsystem:
Regression:	Yes	Bisected commit-id:
Attachments:	xorg.0.log with --enable-debug=full not fixed --enable-debug=full with df89b4941130b5ff020c6ef3e244dcf5fec45f2a

Description Christoph Haag 2014-06-19 16:11:59 UTC

Created attachment 140431 [details]
xorg.0.log with --enable-debug=full

00:02.0 VGA compatible controller: Intel Corporation 3rd Gen Core processor Graphics Controller (rev 09)
01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Wimbledon XT [Radeon HD 7970M] (rev ff)

Since 5cc9ed4b9a7ac579362ccebac67f7a4cdb36de06 almost everything is fine, except wine with several programs like sometimes steam, skyrim, mass effect 2 causes this message in the xorg.0.log:
(EE) intel(0): Failed to submit rendering commands, disabling acceleration.

I have only been able to trigger the error with wine yet. Many complex native games and even all of piglit quick.py runs fine without triggering the error.

Here are two different complete xorg.0.log with xf86-video-intel compiled with --enable-debug:
http://pastebin.com/raw.php?i=cq21cs1W
http://pastebin.com/raw.php?i=cqnQAchm

wine is with d3dstream patches, but happens also with CSMT=disabled, a 1920x1080 virtual wine desktop is used
xf86-video-intel is git master
xorg server 1.15.1 with dri3 patches.
mesa is from git master + axeldavy's offloading patches. The games were also run with offloading to a radeon gpu, but the error also happens when everything is run on the intel gpu.

It happens always with tearfree enabled and disabled, with sna and with uxa and with dri3 and with --disable-dri3.


Here is dmesg with drm.debug=7
http://pastebin.com/raw.php?i=Dxwm0ytg
Timestamp reference: [    66.425] (EE) intel(0): Failed to submit rendering commands, disabling acceleration.


sorry for so much pastebin, I'm not too sure what is actually relevant to attach.

Attached is xorg.0.log with xf86-video-intel --enable-debug=full, compressed with xz because it's 33 megabyte uncompressed.

Comment 1 Chris Wilson 2014-06-20 05:55:52 UTC

commit df89b4941130b5ff020c6ef3e244dcf5fec45f2a
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Jun 19 20:39:59 2014 +0100

    sna: Mark upload from SHM segment as read-only
    
    As this may be mapped by the Xserver using a read-only SHM segment, we
    are forced to treat it always as read-only. And this being X, that it is
    using a SHM segment is opaque to the driver. Fantastic middlelayer.
    
    This was incorrectly removed in
    commit e680e54eab6ffa72e5e1eb6cc0e3fe4b235b06a1
    Author: Chris Wilson <chris@chris-wilson.co.uk>
    Date:   Wed Jun 11 10:48:26 2014 +0100
    
        sna: Ignore setting read-only for temporary userptr maps
    
    Reported-by: Christoph Haag <haagch.christoph@googlemail.com>
    Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=78411
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Comment 2 Christoph Haag 2014-06-20 08:37:50 UTC

Created attachment 140471 [details]
not fixed

Sorry, not fixed with this commit. I can still reproduce this immediately with wine. It also happens with uxa.

Shall I upload another log with --enable-debug=full?

Comment 3 Chris Wilson 2014-06-20 08:47:37 UTC

Strange. That should have been the efault. Please do upload a new trace.

Comment 4 Christoph Haag 2014-06-20 09:28:39 UTC

Created attachment 140481 [details]
--enable-debug=full with df89b4941130b5ff020c6ef3e244dcf5fec45f2a

Comment 5 Chris Wilson 2014-06-20 09:57:57 UTC

Gah.

commit 113a8b9be9dd7ffbc6f7c713bc40561bdf38e8b0
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Jun 19 20:39:59 2014 +0100

    sna: Mark upload from SHM segment as read-only
    
    As this may be mapped by the Xserver using a read-only SHM segment, we
    are forced to treat it always as read-only. And this being X, that it is
    using a SHM segment is opaque to the driver. Fantastic middlelayer.
    
    This was incorrectly removed in
    commit e680e54eab6ffa72e5e1eb6cc0e3fe4b235b06a1
    Author: Chris Wilson <chris@chris-wilson.co.uk>
    Date:   Wed Jun 11 10:48:26 2014 +0100
    
        sna: Ignore setting read-only for temporary userptr maps
    
    Also let's not forget the ShmPutImage -> CopyArea path.
    
    Reported-by: Christoph Haag <haagch.christoph@googlemail.com>
    Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=78411
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Comment 6 Christoph Haag 2014-06-20 10:07:13 UTC

After a quick test it seems like it is fixed.

Thanks!