All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=830969

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=aec77c2bb883c1fba5cc6cceaeeb59084666471a07743b4efc66583118948be4


Bot(s) for this bug's original alert(s):

cros-link

I'll bisect this one...

The toolchain is busted with simple chrome: 

https://bugs.chromium.org/p/chromium/issues/detail?id=831022

I'll bisect when that's back up or when I find a way around it...

Over to Kristian, since the framebuffers moved from tiled to linear.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromiumos/platform/minigbm/+/9c3fb32dd4a8b9a3cf870c533d1e88dd542d7c16

commit 9c3fb32dd4a8b9a3cf870c533d1e88dd542d7c16
Author: Kristian H. Kristensen <hoegsberg@chromium.org>
Date: Thu Apr 12 22:22:34 2018

i915: Allow allocating ARGB buffers for scanout

CL:991261 changed our primary framebuffer format from XRGB to ARGB. We
still only scanout as XRGB where the display controller doesn't
support ARGB, but minigbm doesn't know that. As a result, when we try
to allocate an ARGB buffer with the SCANOUT useflag, we get a linear
buffer.  This breaks PSR and regresses performance and, in general,
just isn't what we want.

This CL enables SCANOUT for all ARGB formats where the corresponding
XRGB formats supports scanout. In the end, it's up to the caller to
make sure a format works for scanout anyway.

BUG= 827188 , 830969 
TEST=test_that graphics_Idle on samus

Change-Id: Iab428e5b21abedcac7cee86fccc8df50dab3f471
Reviewed-on: https://chromium-review.googlesource.com/1008867
Commit-Ready: Kristian H. Kristensen <hoegsberg@chromium.org>
Tested-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>
Reviewed-by: Gurchetan Singh <gurchetansingh@chromium.org>

[modify] https://crrev.com/9c3fb32dd4a8b9a3cf870c533d1e88dd542d7c16/i915.c

Verified that the test is back at ~60fps.

+1