Bug 44001

Summary: [bisected PNV] oglc drawbuffer(basic.allCases) fails on pineview
Product: Mesa Reporter: fangxun <xunx.fang>
Component: Drivers/DRI/i915Assignee: Ian Romanick <idr>
Status: CLOSED WONTFIX QA Contact:
Severity: normal    
Priority: medium CC: michel
Version: git   
Hardware: All   
OS: Linux (All)   
i915 platform: i915 features:
Attachments: Skip invalidation unless the WindowPixmap changes
Flush pending writes to FakeFront

Description fangxun 2011-12-21 01:26:19 UTC
System Environment:
Arch:           i386
Platform:       pineview
Libdrm:         (master)2.4.29-3-gef20301a11afae50bfe127002913dbd0b81ddccc
Mesa:           (master)67e6a40605c482617afa9bd698003ea15e8f8f6e
Xserver:        (master)xorg-server-
Xf86_video_intel: (master)2.17.0-179-g4c2a97e9d2b1073f0e4b3f7b6670939e1b1c5121
Kernel:   (drm-intel-next)097354eb14fa94d31a09c64d640643f58e4a5a9a

Bug detailed description:
Bisect shows 6f916ffec7767eeab62132eb6575043969104c81 is the first bad commit.
commit 6f916ffec7767eeab62132eb6575043969104c81
Author:     Michel Daenzer <daenzer@vmware.com>
AuthorDate: Sun Dec 18 18:29:39 2011 +0200
Commit:     Keith Packard <keithp@keithp.com>
CommitDate: Mon Dec 19 22:31:01 2011 -0800

    dri2: Always re-generate front buffer information when asked for it.

    Otherwise we might keep stale cached information, e.g. after the driver
    performed page flipping.

    This is part of the fix for
    https://bugs.freedesktop.org/show_bug.cgi?id=35452 .

    Signed-off-by: Michel Dänzer <daenzer@vmware.com>
    Reviewed-by: Ville Syrjälä <ville.syrjala@nokia.com>
    Reviewed-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de>
    Signed-off-by: Keith Packard <keithp@keithp.com>

Reproduce steps:
1. start X
2. ./oglconform -z -s -suite all -v 2 -test drawbuffer basic.allCases
Comment 1 Michel Dänzer 2011-12-21 04:07:53 UTC
Not sure what's going on, and I probably won't be able to investigate anytime soon. Feel free to revert.
Comment 2 Gordon Jin 2012-01-12 01:01:21 UTC
Chris, any idea?
Comment 3 Chris Wilson 2012-01-12 08:14:45 UTC
On q35 with an up-to-date build:

$ ./oglconform -z -s -suite all -v 2 -test drawbuffer basic.allCases
Intel Conformance passed.

Total Passed : 32
Total Failed : 0
Total Not run: 0

Something amiss (or implicit?) in the reproduction steps?
Comment 4 Chris Wilson 2012-03-16 13:18:17 UTC
Created attachment 58577 [details] [review]
Skip invalidation unless the WindowPixmap changes
Comment 5 Chris Wilson 2012-03-16 13:19:14 UTC
Mesa should also be intelligent enough to squash this change as well, the logic is all there to not swap if the texture matches.
Comment 6 Gordon Jin 2012-03-31 18:08:40 UTC
any action plan, Chris, Ian?
Comment 7 Chris Wilson 2012-07-05 04:12:53 UTC
Created attachment 63858 [details] [review]
Flush pending writes to FakeFront
Comment 8 Kaveh 2014-08-05 01:08:12 UTC
Please confirm that this is seen on an SNB or newer platform.
Comment 9 Gordon Jin 2015-06-16 02:05:09 UTC
I don't think it impacts new platforms. Closing as won't fix.

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.