All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=850977

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=8c7607e25c8da54545da931e460a741dde27fddfb216ab05f2746f6c419179d8


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

android-nexus5

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/12c01909240000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/12c01909240000

Reland: Call blink::WebMemoryCoordinator::OnPurgeMemory in RenderThreadImpl::ReleaseFreeMemory by gyuyoung.kim@lge.com
https://chromium.googlesource.com/chromium/src/+/f3ab72bc34a678e7c8e164869a52b951c2cf051b
2172 → 2365 (+192.8)

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

Hmm, If I understand this perf regression correctly, my CL made Chromium 11% slower. It's slower than what I expected. Let me revert it soon.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/cdd97e247777254849ee41adb8efd25f52e48df3

commit cdd97e247777254849ee41adb8efd25f52e48df3
Author: Gyuyoung Kim <gyuyoung.kim@lge.com>
Date: Tue Jun 12 07:20:29 2018

Revert "Reland: Call blink::WebMemoryCoordinator::OnPurgeMemory in RenderThreadImpl::ReleaseFreeMemory"

This reverts commit f3ab72bc34a678e7c8e164869a52b951c2cf051b.

Because this CL caused about 11% performance regression on Android.

TBR=haraken@chromium.org, kbr@chromium.org

Bug:  850977 
Change-Id: Iaeb2697bbd4651bf40d6e437ff2a7efe19c647b0
Reviewed-on: https://chromium-review.googlesource.com/1096575
Reviewed-by: Gyuyoung Kim <gyuyoung.kim@chromium.org>
Reviewed-by: Kentaro Hara <haraken@chromium.org>
Commit-Queue: Gyuyoung Kim <gyuyoung.kim@lge.com>
Cr-Commit-Position: refs/heads/master@{#566349}
[modify] https://crrev.com/cdd97e247777254849ee41adb8efd25f52e48df3/content/renderer/render_thread_impl.cc

Once I reverted the CL. But I'm still thinking we can release more memory in RenderThreadImpl::ReleaseFreeMemory(). So I will prepare a new CL again. Thanks.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/67dc1b416c296c095015eb0413ba1c0361d115fe

commit 67dc1b416c296c095015eb0413ba1c0361d115fe
Author: Gyuyoung Kim <gyuyoung.kim@lge.com>
Date: Mon Jun 18 16:13:02 2018

Reland: Call blink::WebMemoryCoordinator::OnPurgeMemory in RenderThreadImpl::ReleaseFreeMemory - 2

This reverts commit cdd97e247777254849ee41adb8efd25f52e48df3.

After perf-bot reported a regression in timeToFirstMeaningfulPaint,
timeToFirstContentfulPaint, and timeToFirstPaint on android nexus5,
the original CL was reverted. However, the regression looks reasonable
because ReleaseMemory could be called before FCP and FMP came on low-end
devices if the website needs to use much memory. I guess that the regression
caused because the test website(edition.cnn.com) was consuming too much memory
on the mobile device.

Bug:  850977 ,  851868 

Cq-Include-Trybots: luci.chromium.try:linux_layout_tests_slimming_paint_v2;master.tryserver.blink:linux_trusty_blink_rel
Change-Id: Icc8e5f24c114947fe8d186f699d0457ea82b7416
Reviewed-on: https://chromium-review.googlesource.com/1096921
Reviewed-by: Kentaro Hara <haraken@chromium.org>
Commit-Queue: Gyuyoung Kim <gyuyoung.kim@lge.com>
Cr-Commit-Position: refs/heads/master@{#568023}
[modify] https://crrev.com/67dc1b416c296c095015eb0413ba1c0361d115fe/content/renderer/render_thread_impl.cc