The UI doesn't even start on the peach_pit device. I will need to bisect....

Bisect ended ... This CL is the culprit https://codereview.chromium.org/1914343003.
Assigning to the owner. This failure is likely to make it to the PFQ soon unfortunately.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7

commit 5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7
Author: lionel.g.landwerlin <lionel.g.landwerlin@intel.com>
Date: Thu May 05 09:23:39 2016

Revert of ash: reset color management when new screens are hotplugged (patchset #3 id:30009 of https://codereview.chromium.org/1914343003/ )

Reason for revert:
This is causing issues on ARM/exynos5420-peach-pit Chromebook.
Let's revert this for now as I don't have a device to investigate.

BUG= 609256 

Original issue's description:
> ash: reset color management when new screens are plugged
>
> With the DRM drivers on ChromeOS, the color management tables and matrices
> are stored at the pipe level (part of the display hardware that is
> configurable regardless of the actual connector it is attached to). This
> allows display configuration to remain active while different processes are
> using the driver (for example switching VT).
>
> As a result, when an external screen is connected to a Chromebook, a given
> color configuration might be applied to it and remain stored in the driver
> after the screen is disconnected. If another external screen is now
> connected the previously applied color management will remain if there is
> not a profile for that display.
>
> This change takes care of resetting the color management matrices/tables on
> newly connected screen before we apply a new configuration.
>
> BUG= 495196 
> TEST=Manual testing with hotplugging of a display and identify correct
> behavior.
>
> Committed: https://crrev.com/5932a0131851f35b74be5d2fc86a3834defd4d8c
> Cr-Commit-Position: refs/heads/master@{#391491}

TBR=robert.bradford@intel.com,oshima@chromium.org,spang@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG= 495196 

Review-Url: https://codereview.chromium.org/1945353004
Cr-Commit-Position: refs/heads/master@{#391781}

[modify] https://crrev.com/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7/ash/display/display_color_manager_chromeos.cc
[modify] https://crrev.com/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7/ash/display/display_color_manager_chromeos_unittest.cc
[modify] https://crrev.com/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7/ui/ozone/platform/drm/gpu/drm_device.cc
[modify] https://crrev.com/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7/ui/ozone/platform/drm/gpu/drm_device.h
[modify] https://crrev.com/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7/ui/ozone/platform/drm/gpu/drm_display.cc
[modify] https://crrev.com/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7/ui/ozone/platform/drm/gpu/drm_gpu_display_manager.h
[modify] https://crrev.com/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7/ui/ozone/platform/drm/gpu/mock_drm_device.cc
[modify] https://crrev.com/5f1b82e8460a3d4d0a45346cdcdf5862be27c2c7/ui/ozone/platform/drm/gpu/mock_drm_device.h

Let's revert https://codereview.chromium.org/1914343003 for now.
I've tested this on several Intel Chromebook (running different kernel versions),  I'm wondering if this could be an issue in the exynos drm driver.

Could you paste more of the traces on this bug?
Thanks!

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/7af25197088e372b01dbb1ea7b7793e5c1e404d2

commit 7af25197088e372b01dbb1ea7b7793e5c1e404d2
Author: lionel.g.landwerlin <lionel.g.landwerlin@intel.com>
Date: Thu May 05 15:44:33 2016

Reland: ash: reset color management when new screens are plugged

Reland with fix for platforms that don't support drmModeCrtcSetGamma.

With the DRM drivers on ChromeOS, the color management tables and matrices
are stored at the pipe level (part of the display hardware that is
configurable regardless of the actual connector it is attached to). This
allows display configuration to remain active while different processes are
using the driver (for example switching VT).

As a result, when an external screen is connected to a Chromebook, a given
color configuration might be applied to it and remain stored in the driver
after the screen is disconnected. If another external screen is now
connected the previously applied color management will remain if there is
not a profile for that display.

This change takes care of resetting the color management matrices/tables on
newly connected screen before we apply a new configuration.

BUG= 495196 , 609256 
TEST=Manual testing with hotplugging of a display and identify correct
behavior.

Committed: https://crrev.com/5932a0131851f35b74be5d2fc86a3834defd4d8c
Cr-Commit-Position: refs/heads/master@{#391491}

Review-Url: https://codereview.chromium.org/1914343003
Cr-Commit-Position: refs/heads/master@{#391810}

[modify] https://crrev.com/7af25197088e372b01dbb1ea7b7793e5c1e404d2/ash/display/display_color_manager_chromeos.cc
[modify] https://crrev.com/7af25197088e372b01dbb1ea7b7793e5c1e404d2/ash/display/display_color_manager_chromeos_unittest.cc
[modify] https://crrev.com/7af25197088e372b01dbb1ea7b7793e5c1e404d2/ui/ozone/platform/drm/gpu/drm_device.cc
[modify] https://crrev.com/7af25197088e372b01dbb1ea7b7793e5c1e404d2/ui/ozone/platform/drm/gpu/drm_device.h
[modify] https://crrev.com/7af25197088e372b01dbb1ea7b7793e5c1e404d2/ui/ozone/platform/drm/gpu/drm_display.cc
[modify] https://crrev.com/7af25197088e372b01dbb1ea7b7793e5c1e404d2/ui/ozone/platform/drm/gpu/drm_gpu_display_manager.h
[modify] https://crrev.com/7af25197088e372b01dbb1ea7b7793e5c1e404d2/ui/ozone/platform/drm/gpu/mock_drm_device.cc
[modify] https://crrev.com/7af25197088e372b01dbb1ea7b7793e5c1e404d2/ui/ozone/platform/drm/gpu/mock_drm_device.h

Thanks for the fix. The peach_pit-tot-chrome-pfq-informational builder is green again. However the PFQ is not; the fix hasn't made it there yet.

PFQ turned green with the fix in.

Bulk verified

bulk verified

Issue 599172 has been merged into this issue.