Side remarks:

1) This was found by PoHsien's new failure alerts.

2) The regression in 1) is a completely broken Chrome and didn't make it through the PFQ. But we still got notified from the PFQ *before the uprev*, which is slightly unusual but actually also very cool (we are using the Chrome perf dashboard for notification, so that is going to have side effects when Chrome doesn't uprev for a while). 

@#1.2: i915_gem_active was removed from chromeos-4.4 by this CL:

https://chromium-review.googlesource.com/#/c/460579
UPSTREAM: drm/i915: Remove inactive/active list from debugfs

Its commit message says:
These two files (i915_gem_active, i915_gem_inactive) no longer give
pertinent information since active/inactive tracking is per-vm and so we
need the information per-vm. They are obsolete so remove them.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/third_party/autotest/+/de375a129945582548b40aefa22a6e1ded21d2ac

commit de375a129945582548b40aefa22a6e1ded21d2ac
Author: Ilja H. Friedel <ihf@chromium.org>
Date: Wed May 31 07:34:43 2017

graphics_Idle: disable gem_idle.

BUG= chromium:727983 
TEST=still runs verify_graphics_gem_idle on sentry (kernel 3.18).

Change-Id: I7ef82d8f4b7c204c7d7ea9317af75fe1ce76e553
Reviewed-on: https://chromium-review.googlesource.com/518922
Tested-by: Ilja H. Friedel <ihf@chromium.org>
Reviewed-by: Nicolas Boichat <drinkcat@chromium.org>

[modify] https://crrev.com/de375a129945582548b40aefa22a6e1ded21d2ac/client/site_tests/graphics_Idle/graphics_Idle.py

Are there any updates on this issue as to whether or not the graphics gem check will be re-enabled for 4.4 systems?

We would like to re-enable the test, but have not figured out how. Do you have suggestions?

I've discussed with the i915 team, and they have informed me that for these changes, the information for active/inactive gem objects can be found in each process listed in /sys/kernel/debug/dri/0/i915_gem_objects.

So, for 4.4 and up systems running the graphics_Idle test, we can iterate over each process in the i915_gem_objects file, parsing out the counts for active gem objects and summing them up.

Casey, you seem to have a good idea of what to do.  Can you submit a CL to fix this issue :-)?

I'll see if I can submit a CL either today or tomorrow to fix.

I've posted a fix at:
https://chromium-review.googlesource.com/#/c/567810/

With the check for whether the i915_gem_active file exists, the fix will retain support for kernel versions that use the i915_gem_active file, including older 4.4 kernels.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/third_party/autotest/+/b6915cac48d8df5d4304c83462e01182f6ce36b5

commit b6915cac48d8df5d4304c83462e01182f6ce36b5
Author: cgbowman <casey.g.bowman@intel.com>
Date: Tue Jul 18 07:32:41 2017

graphics_Idle: Add support for i915 per-process active gem object detection

The test will look at the i915_gem_objects file in each process for
active gem objects if the i915_gem_active file is missing.

The i915_gem_active debug file was removed in the 4.4 kernel as of
9198b9a4deecbf58654e0cdc4197b25c25f9612a

which corresponds to
https://chromium-review.googlesource.com/#/c/460579/

BUG= chromium:727983 
TEST=graphics_Idle on 4.4 and non-4.4 kernel images

Change-Id: I6225c0e60d0b7a1ed7ba8d93b4f0d7e35289f7fd
Reviewed-on: https://chromium-review.googlesource.com/567810
Commit-Ready: Gurchetan Singh <gurchetansingh@chromium.org>
Tested-by: Ilja H. Friedel <ihf@chromium.org>
Reviewed-by: Ilja H. Friedel <ihf@chromium.org>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>

[modify] https://crrev.com/b6915cac48d8df5d4304c83462e01182f6ce36b5/client/site_tests/graphics_Idle/graphics_Idle.py

Since the patch has been merged, is there any objection to closing this issue? :)

Thanks again Casey for fixing this (I would make you the owner -- but the bug trackers says "Issue owner must be a project member" [i.e, use Chromium account]))!

No worries, thanks for all the suggestions. :)