5.1%-22.1% regression in rendering.mobile at 580989:581036 |
|||||||||||||
Issue descriptionSee the link to graphs below.
,
Aug 8
📍 Pinpoint job started. https://pinpoint-dot-chromeperf.appspot.com/job/12e74778640000
,
Aug 9
📍 Found a significant difference after 1 commit. https://pinpoint-dot-chromeperf.appspot.com/job/12e74778640000 Only invalidate property trees and compositing for PaintLayers when needed, on style change. by chrishtr@chromium.org https://chromium.googlesource.com/chromium/src/+/ba8c87a1a32815e6061b7c38d31ddfbbaac7a124 15.25 → 18.49 (+3.244) Understanding performance regressions: http://g.co/ChromePerformanceRegressions
,
Aug 9
The regressions appear to all be on mobile WebViews.
,
Aug 9
My first theory after perusing the patch is that in cases where PaintLayer::AttemptDirectCompositingUpdate returned true, which previously avoided SetNeedsCompositingInputsUpdate() in PaintLayer::StyleDidChange. Now we may be able to have SetNeedsCompositingInputsUpdate happen in some of these cases. I think the code in LayoutObject::StyleDidChange in my patch might trigger in this case. If so, this is a real regression since the whole point of AttemptDirectCompositingUpdate is to provide a fast path for composited transform and opacity animations.
,
Aug 21
Issue 872596 has been merged into this issue.
,
Aug 21
Issue 873382 has been merged into this issue.
,
Aug 21
,
Aug 21
,
Aug 21
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/82efd6c5c86089eda2d54a0783e429ff96a36825 commit 82efd6c5c86089eda2d54a0783e429ff96a36825 Author: Chris Harrelson <chrishtr@chromium.org> Date: Tue Aug 21 20:45:52 2018 Reduce over-invalidation of compositing inputs. In particular, for cases when the PaintLayer direct compositing update fast-path applies. Bug: 872172 Cq-Include-Trybots: luci.chromium.try:linux_layout_tests_slimming_paint_v2;master.tryserver.blink:linux_trusty_blink_rel Change-Id: If059f2f76d9973c2a19d2f326f495e4891813d6e Reviewed-on: https://chromium-review.googlesource.com/1171797 Reviewed-by: vmpstr <vmpstr@chromium.org> Commit-Queue: Chris Harrelson <chrishtr@chromium.org> Cr-Commit-Position: refs/heads/master@{#584892} [modify] https://crrev.com/82efd6c5c86089eda2d54a0783e429ff96a36825/third_party/blink/renderer/core/layout/layout_object.cc [modify] https://crrev.com/82efd6c5c86089eda2d54a0783e429ff96a36825/third_party/blink/renderer/core/paint/paint_layer.cc
,
Aug 22
Two regressions are now resolved. Two remain: https://chromeperf.appspot.com/group_report?bug_id=874874 https://chromeperf.appspot.com/group_report?bug_id=873382
,
Aug 27
Issue 875981 has been merged into this issue.
,
Aug 27
,
Aug 27
,
Aug 31
I think this was caused by an over-invalidation in LayoutView when it recomputes overflow. If overflow did not actually change, compositing inputs don't need to be dirtied.
,
Aug 31
,
Sep 1
Fixed in: https://chromium.googlesource.com/chromium/src/+/d1042cc805130a5ed959f06593787dd0021106dc Will request merge after verifying bots recover.
,
Sep 3
This issue is marked as a release blocker with no OS labels associated. Please add an appropriate OS label. All release blocking issues should have OS labels associated to it, so that the issue can tracked and promptly verified, once it gets fixed. Thanks for your time! To disable nags, add the Disable-Nags label. For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Sep 4
,
Sep 4
Recovered.
,
Sep 5
Your change meets the bar and is auto-approved for M70. Please go ahead and merge the CL to branch 3538 manually. Please contact milestone owner if you have questions. Owners: benmason@(Android), kariahda@(iOS), geohsu@(ChromeOS), abdulsyed@(Desktop) For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Sep 5
https://chromium.googlesource.com/chromium/src/+/557dccd0a078ffd17d77f832b8affa23d965e596
,
Sep 10
This issue has been approved for a merge. Please merge the fix to any appropriate branches as soon as possible! If all merges have been completed, please remove any remaining Merge-Approved labels from this issue. Thanks for your time! To disable nags, add the Disable-Nags label. For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Sep 25
already merged. |
|||||||||||||
►
Sign in to add a comment |
|||||||||||||
Comment 1 by 42576172...@developer.gserviceaccount.com
, Aug 8