This regression happens on falco and wolf, whose processors are haswell as well as peppy.
https://chromeperf.appspot.com/report?sid=fd8179fd433e77d48e8715430e0d4cf9354d714a3da48fd984814f8c96cec2f5

I am concerned about the zig-zag change. I am going to wait and see in a day, and address tomorrow.

The same regression is caused on link and panther. link is IvyBridge though.
https://chromeperf.appspot.com/report?sid=1301706a92cde3f6844645dd7370d7505a097cee65858a20d6797babcd4f2c8a&rev=32530001007700000

I am investigating this regression using link.
ChromeOS range: 10076.0.0-10077.0.0
Chrome range: 64.0.3252.0-64.0.3253.0

The regression is caused by mesa version update. crosreview.com/734620.
It was resolved by merely reverting crosreview.com/734620 and all the media-libs/mesa patches merged after the version update (i.e. crosreview.com/196170, crosreview.com/742244 and crosreview.com/735325).

marcheu@, can you take a look or assign a better owner for this issue?

Thanks.

Hi, marcheu@. Is there any update?

Hmm, I look the graph again today.
The value is still higher.
On the other hand, the frame drop rate of video_PlaybackPerf is small.
So this regression will not affect to users directly.
However, may I please consider to solve this regression? 
https://chromeperf.appspot.com/report?
sid=598ce620285349ddc0f7cf090dbc0321af8b430fd74b398d06bbae8acefa4105&start_rev=32390271003202300&end_rev=32870001020800000&rev=32530001007700000

Thanks.

It seems related to kernel 3.8, which tells me it's probably the no_reloc change. pwang@ can tell us if it's the same issue or not.

In any case, it's only affecting 4K videos on intel-based kernel 3.8 devices, so it's not p1.

Test on my link.
I do not think this problem is the same as the no_reloc change as the value is high in mesa commit 12a77f391f, which is one commit before commit 3f353342a6 NO_RELOC introduced. 
It may be related to the previous graphics_VTSwitch flaky that we observed in  crbug.com/781884 .

BTW, one thing I noticed is that if we apply the tot mesa(34838c221260f961140040416b1a84b490448ac1) instead of our pinned version, the bug is fixed.

Ah ok, thanks for investigating. We can wait until the next mesa uprev then.

Thanks for investigating.
When is the next mesa version update going to happen?

May I ask for an update on the status of this issue please? From the graphs it appears that this is still happening? Thanks!

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/806b6d0fddc4ee12c3094e6e9d4691fa9c7d2b73

commit 806b6d0fddc4ee12c3094e6e9d4691fa9c7d2b73
Author: Po-Hsien Wang <pwang@chromium.org>
Date: Thu Jan 25 04:24:37 2018

mesa: flush after fallback texture

Fix issue that driver are failing to operates the same fallback texture.

BUG= chromium:780368 
TEST=test on my own link
     test_that link video_VDAPerf.h264.4k

Change-Id: Ie09de707c688fdd803987fe6fdb3e2654cab70df
Reviewed-on: https://chromium-review.googlesource.com/879963
Commit-Ready: Chad Versace <chadversary@chromium.org>
Tested-by: Po-Hsien Wang <pwang@chromium.org>
Reviewed-by: Chad Versace <chadversary@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

[modify] https://crrev.com/806b6d0fddc4ee12c3094e6e9d4691fa9c7d2b73/media-libs/mesa/mesa-17.2.3.ebuild
[delete] https://crrev.com/a7a547205057222f025d8d48180078b0df68c5a9/media-libs/mesa/mesa-17.2.3-r12.ebuild
[modify] https://crrev.com/806b6d0fddc4ee12c3094e6e9d4691fa9c7d2b73/media-libs/mesa/mesa-9999.ebuild
[add] https://crrev.com/806b6d0fddc4ee12c3094e6e9d4691fa9c7d2b73/media-libs/mesa/files/17.3-Flush-After-Fallback-Texture.patch
[add] https://crrev.com/806b6d0fddc4ee12c3094e6e9d4691fa9c7d2b73/media-libs/mesa/mesa-17.2.3-r13.ebuild

Since this was reported on 10077.0.0, both M64 and M65 are probably affected?
Should we consider merging? Thanks!

Would like to merge this to 64 65 after it is verified. Feel free to cherry-pick if you would like to make it pushed to the tree sooner.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/1437dc8bf1027271debf4edffce8511f4cec1104

commit 1437dc8bf1027271debf4edffce8511f4cec1104
Author: Po-Hsien Wang <pwang@chromium.org>
Date: Fri Feb 02 00:25:25 2018

mesa: flush after fallback texture

Fix issue that driver are failing to operates the same fallback texture.

BUG= chromium:780368 
TEST=test on my own link
     test_that link video_VDAPerf.h264.4k

Change-Id: Ie09de707c688fdd803987fe6fdb3e2654cab70df
Reviewed-on: https://chromium-review.googlesource.com/879963
Commit-Ready: Chad Versace <chadversary@chromium.org>
Tested-by: Po-Hsien Wang <pwang@chromium.org>
Reviewed-by: Chad Versace <chadversary@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>
(cherry picked from commit 806b6d0fddc4ee12c3094e6e9d4691fa9c7d2b73)
Reviewed-on: https://chromium-review.googlesource.com/887928
Reviewed-by: Po-Hsien Wang <pwang@chromium.org>
Commit-Queue: Po-Hsien Wang <pwang@chromium.org>

[modify] https://crrev.com/1437dc8bf1027271debf4edffce8511f4cec1104/media-libs/mesa/mesa-17.2.3.ebuild
[modify] https://crrev.com/1437dc8bf1027271debf4edffce8511f4cec1104/media-libs/mesa/mesa-9999.ebuild
[add] https://crrev.com/1437dc8bf1027271debf4edffce8511f4cec1104/media-libs/mesa/files/17.3-Flush-After-Fallback-Texture.patch