Also repro'd on Settings window. Attached screenshot

Deleted: Screenshot 2017-11-28 at 15.57.46 - Display 1.png 2.8 MB

	Screenshot 2017-11-28 at 15.57.46 - Display 1.png 2.8 MB View Download

wutao@ to triage

+oshima@

Can repro on EVE, but cannot repro this on Linux build.

Also looks like resolution depended. Some time need to change resolution to repro the bug while bisecting.

sdantuluri@, what resolution you used on EVE?

I am using default resolution 1200 * 800

Current bisect is between:

good: 10074.0.0	/ 64.0.3252.0
bad:  10077.0.0	/ 64.0.3253.0

There are not too much stand out between Chrome changes 64.0.3252.0 and 64.0.3253.0
https://chromium.googlesource.com/chromium/src/+log/64.0.3252.0..64.0.3253.0?n=10000

+marcheu@, could you please check the mesa update.


further bisect pointed good: 10076.0.0	/ 64.0.3252.0

At ChromeOS side, there is a mesa update from 10076.0.0 to 10077.0.0:
https://crosland.corp.google.com/log/10076.0.0..10077.0.0


https://chromium-review.googlesource.com/c/chromiumos/overlays/chromiumos-overlay/+/734620

These glitches are also observed while using Alt+Tab

Deleted: IMG-3743.JPG 2.8 MB

	IMG-3743.JPG 2.8 MB View Download

Kristian, do you have cycles to look at this one?

wutao@, have you seen this issue? do you think it might be related to the overview changes we've made recently?

reveman@, this bug happens before our changes to hide the header.
But it may be related to the tri-linear filtering used in overview and window switcher.

Do you think mesa change will affect the tri-linear filtering?

Yes, might be related to mipmaps. Next step is to bisect mesa to find the exact change that caused the issue. Assigning back to hoegsberg@.

 Issue 797196  has been merged into this issue.

 Issue 797148  has been merged into this issue.

#15, I removed the trilinear filtering in the window cycling and overview mode, and this bug cannot repro, which confirms it is mipmaps issue, triggered by mesa update.

Testing this on R65, the artifact is gone. I tried deploying master chrome on R64 and the issue goes away. There has been no mesa change between R64 and R65. Bisecting master now to see which commit made the artifact go away.

Bisected to:

commit 9cc56b283e51de94cc373a77214da2bb646efcbd
Author: Xu Xing <xing.xu@intel.com>
Date:   Fri Dec 22 08:40:31 2017 +0800

    [viz] Implement local resource in GLRenderer
    
    Local resource is resource used inside the same GL context and will not being sent
    into another process, no need to create fence and mailbox for these resources.
    For RenderPass in render_pass_textures_, it uses texture-backed resources, no need
    to support GpuMemoryBuffer.
    
    Currently there are three path may possible set is_overlay_ to true:
    ResourceProvider::CreateGpuMemoryBufferResource
    LayerTreeResourceProvider::CreateResourceFromTextureMailbox
    ResourceProvider::CreateGpuTextureResource (Used by GLRenderer::overlay_resource_pool_).
    Resource for RenderPass in render_pass_textures_ is not created by above three paths.
    Based on this, remove the support of overlay.
    
    BUG= 782044 
    
    Cq-Include-Trybots: master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel
    Change-Id: Ia334842b12b8297ae4a768979adb8b5a468b906a
    Reviewed-on: https://chromium-review.googlesource.com/771130
    Commit-Queue: Xing Xu <xing.xu@intel.com>
    Reviewed-by: danakj <danakj@chromium.org>
    Cr-Commit-Position: refs/heads/master@{#525903}

which is to say, that's the commit that fixes the issue on master.

On closer inspection and comparing scaling quality in overview mode, it looks like that commit breaks/disables mipmaps.

+danakj@

This is https://bugs.chromium.org/p/chromium/issues/detail?id=802924 ?

#23: Correct.

Bisected to:

commit ea0d2e98ecb369ab84e78c84709c0930ea8c293a (bad-commit)
Author: Kenneth Graunke <kenneth@whitecape.org>
Date:   Thu Oct 5 20:31:01 2017 -0700

    i965: Disable auxiliary buffers when there are self-dependencies.
    
    Jason and I investigated several OpenGL CTS failures where the tests
    bind the same texture for rendering and texturing, at the same time.
    This has defined results as long as the reads happen before writes,
    or the regions are non-overlapping.  Normally, this just works out.
    
    However, CCS can cause problems.  If the shader is reading one set of
    pixels, and writing to different pixels that are adjacent, they may end
    up being covered by the same CCS block.  So rendering may be writing a
    CCS block, while the sampler is trying to read it.  Corruption ensues.
    
    Disabling CCS is unfortunate, but safe.
    
    Fixes several KHR-GL45.texture_barrier.* subtests.
    
    Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
    Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>

Mesa CL to backport upstream fix:

https://chromium-review.googlesource.com/c/chromiumos/overlays/chromiumos-overlay/+/871986

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/0f17d31e275064ee6d1b0835411e07dad4795991

commit 0f17d31e275064ee6d1b0835411e07dad4795991
Author: Kristian H. Kristensen <hoegsberg@chromium.org>
Date: Fri Jan 19 01:40:55 2018

media-libs/mesa: Fix mipmap generation for compressed textures

BUG= 789253 
TEST=Activate alt-tab or overlays mode, verify correct rendering.

Change-Id: Ie53f630b0edef8ea595c9225f28f750c77d4f978
Reviewed-on: https://chromium-review.googlesource.com/871986
Commit-Ready: ChromeOS CL Exonerator Bot <chromiumos-cl-exonerator@appspot.gserviceaccount.com>
Tested-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Reviewed-by: Chad Versace <chadversary@chromium.org>
Reviewed-by: David Reveman <reveman@chromium.org>

[add] https://crrev.com/0f17d31e275064ee6d1b0835411e07dad4795991/media-libs/mesa/files/17.2-0004-i965-Don-t-disable-aux-buffers-for-non-overlapping-m.patch
[add] https://crrev.com/0f17d31e275064ee6d1b0835411e07dad4795991/media-libs/mesa/files/17.2-0001-i965-Make-intel_miptree_prepare_texture-take-level-l.patch
[modify] https://crrev.com/0f17d31e275064ee6d1b0835411e07dad4795991/media-libs/mesa/mesa-17.2.3.ebuild
[modify] https://crrev.com/0f17d31e275064ee6d1b0835411e07dad4795991/media-libs/mesa/mesa-9999.ebuild
[add] https://crrev.com/0f17d31e275064ee6d1b0835411e07dad4795991/media-libs/mesa/files/17.2-0002-i965-Only-resolve-texture-levels-layers-that-are-acc.patch
[rename] https://crrev.com/0f17d31e275064ee6d1b0835411e07dad4795991/media-libs/mesa/mesa-17.2.3-r12.ebuild
[add] https://crrev.com/0f17d31e275064ee6d1b0835411e07dad4795991/media-libs/mesa/files/17.2-0003-i965-Remove-the-intel_miptree_prepare_fb_fetch-wrapp.patch

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/7124158d365824852b91bff6c074dd43d2b54269

commit 7124158d365824852b91bff6c074dd43d2b54269
Author: Kristian H. Kristensen <hoegsberg@chromium.org>
Date: Fri Jan 19 01:52:30 2018

media-libs/mesa: Fix mipmap generation for compressed textures

BUG= 789253 
TEST=Activate alt-tab or overlays mode, verify correct rendering.

Change-Id: Ie53f630b0edef8ea595c9225f28f750c77d4f978
Reviewed-on: https://chromium-review.googlesource.com/872072
Reviewed-by: Bernie Thompson <bhthompson@chromium.org>
Commit-Queue: Kristian H. Kristensen <hoegsberg@chromium.org>
Tested-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Trybot-Ready: Kristian H. Kristensen <hoegsberg@chromium.org>

[add] https://crrev.com/7124158d365824852b91bff6c074dd43d2b54269/media-libs/mesa/files/17.2-0004-i965-Don-t-disable-aux-buffers-for-non-overlapping-m.patch
[add] https://crrev.com/7124158d365824852b91bff6c074dd43d2b54269/media-libs/mesa/files/17.2-0001-i965-Make-intel_miptree_prepare_texture-take-level-l.patch
[modify] https://crrev.com/7124158d365824852b91bff6c074dd43d2b54269/media-libs/mesa/mesa-17.2.3.ebuild
[modify] https://crrev.com/7124158d365824852b91bff6c074dd43d2b54269/media-libs/mesa/mesa-9999.ebuild
[add] https://crrev.com/7124158d365824852b91bff6c074dd43d2b54269/media-libs/mesa/files/17.2-0002-i965-Only-resolve-texture-levels-layers-that-are-acc.patch
[add] https://crrev.com/7124158d365824852b91bff6c074dd43d2b54269/media-libs/mesa/mesa-17.2.3-r12.ebuild
[add] https://crrev.com/7124158d365824852b91bff6c074dd43d2b54269/media-libs/mesa/files/17.2-0003-i965-Remove-the-intel_miptree_prepare_fb_fetch-wrapp.patch

That last CL could have broken the VM builds, mesa no longer builds and the start of the failures looks consistent with the CL hitting R64:
https://luci-milo.appspot.com/buildbot/chromeos_release/novato-release%20release-R64-10176.B/

The problem is that llvm isn't the right version or isn't installed, it has nothing to do with the above CL.

Or a bad ebuild breaks llvm detection...

Line 182 of https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/release-R64-10176.B/media-libs/mesa/mesa-17.2.3-r12.ebuild#182 is different from line 182 of https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/release-R64-10176.B/media-libs/mesa/mesa-17.2.3.ebuild#182

It does look like the CL is breaking the path to LLVM.

Hmm indeed, the ToT commit is good, but the backport ebuild is bad (it adds an extra file).

I cherry-picked this with the gerrit web tool.

It looks like mesa-17.2.3-r12.ebuild stopped being a symlink to mesa-17.2.3.ebuild.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/b6d96f41df705b603b9e882049c4c1483b13aaab

commit b6d96f41df705b603b9e882049c4c1483b13aaab
Author: Kristian H. Kristensen <hoegsberg@chromium.org>
Date: Mon Jan 22 23:36:30 2018

media-libs/mesa: Make mesa-17.2.3-r13.ebuild a symlink again

BUG= 789253 

Change-Id: Ia6af1f78aa35ba900a30b21053181edbd2ebbbbe
Reviewed-on: https://chromium-review.googlesource.com/879234
Reviewed-by: Bernie Thompson <bhthompson@chromium.org>
Reviewed-by: Nicolas Norvez <norvez@chromium.org>
Commit-Queue: Kristian H. Kristensen <hoegsberg@chromium.org>
Tested-by: Kristian H. Kristensen <hoegsberg@chromium.org>

[delete] https://crrev.com/da086e88f7d46f370200959a57a84a74609ccd94/media-libs/mesa/mesa-17.2.3-r12.ebuild
[rename] https://crrev.com/b6d96f41df705b603b9e882049c4c1483b13aaab/media-libs/mesa/mesa-17.2.3-r13.ebuild

Issue still exists on ToT but the glitches look different compared to those reported in the original bug description.

See settings window in the attached screenshot. Tested on build 10335.0.0, 66.0.3329.0 elm

Screenshot for c#37

Deleted: IMG-5457.JPG 2.6 MB

	IMG-5457.JPG 2.6 MB View Download

I tested on 10337, but could not reproduce the corruption in #38.

sdantuluri@, #38, which device it is? Similar bug in b/72298538

Re#40 It was on Elm. Yes, it is same as b/72298538

Ok, definitely a different issue then. Let's track that in b/72298538 and close this one.

 Issue 795203  has been merged into this issue.

This issue has been approved for a merge. Please merge the fix to any appropriate branches as soon as possible!

If all merges have been completed, please remove any remaining Merge-Approved labels from this issue.

Thanks for your time! To disable nags, add the Disable-Nags label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

10684.0.0, 68.0.3432.0