Check failed: context.NeedsVisualRectUpdate(object) browser tests flakes linux-chromeos-rel |
|||||||
Issue descriptionlinux-chromeos-rel is a CQ bot, and I'm seeing intermittent flakes in browser tests due to a check stop in paint invalidation. No record of it in crbug.com that I could find, so new bug, seems like a recent change in behavior then.
,
Apr 24 2018
,
Apr 24 2018
https://ci.chromium.org/buildbot/tryserver.chromium.chromiumos/linux-chromeos-rel/108405 build, for the record. Search the browser_tests (with patch) 'stdout' log to see the trace in #1.
,
Apr 29 2018
Another example of this CHECK stop: https://ci.chromium.org/buildbot/chromium.memory/Linux%20Chromium%20OS%20ASan%20LSan%20Tests%20(1)/27231 browser_tests
,
May 2 2018
,
May 2 2018
Flakiness dashboard: https://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=GalleryBrowserTest.EnableDisableOverwriteOriginalCheckboxOnDownloads. It's more flaky on linux-chromeos-dbg and Linux Chromium OS ASan LSan Tests (1), though all retries succeeded in the past week. I think the flakiness is not new. It had not been reported before this bug perhaps because the bots have tests that are much more flaky than this test.
,
May 2 2018
Yeap agree, retries are working so it's a TIMEOUT PASS. Low flake rate though, sometimes happens on the linux-chromeos-rel bot [1], reading browser_tests (with patch) in the flakiness dashboard. [1] Persistent browser_test flakiness cause on linux-chromeos-rel December 17 .. April 22 has been identified and fixed. Other flakes maybe went unreported during that time :/
,
May 3 2018
Didn't reproduce locally by repeating the test 100 times, with both release and debug builds. Will try to add some debugging information and see what the bot will log when the test fails. May need some time to fix. Keeping priority 3 because for now the flakiness doesn't fail the bots.
,
May 3 2018
For repro, with a DEBUG build browser_tests, run it in one window, and in another window do a clean compile of chrome: that might increase the chance of flake. P3 sounds good to me, btw.
,
May 3 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/71b78364f8a51dd48b4d21bf3c8aa8726d062197 commit 71b78364f8a51dd48b4d21bf3c8aa8726d062197 Author: Xianzhu Wang <wangxianzhu@chromium.org> Date: Thu May 03 22:01:04 2018 [SPv175] Disable more code when PartialRasterInvalidation is not enabled The code disabled in this CL had some side effects, e.g. left PartialInvalidationRect not cleared after PrePaintClean, and the next document lifecycle might do unexpected rect mapping while no NeedsVisualRect flag set. Bug: 836097 Cq-Include-Trybots: master.tryserver.blink:linux_trusty_blink_rel;master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2 Change-Id: Ib05ccfd439be78666771f8483a0b462128848c08 Reviewed-on: https://chromium-review.googlesource.com/1041570 Reviewed-by: Philip Rogers <pdr@chromium.org> Commit-Queue: Xianzhu Wang <wangxianzhu@chromium.org> Cr-Commit-Position: refs/heads/master@{#555889} [modify] https://crrev.com/71b78364f8a51dd48b4d21bf3c8aa8726d062197/third_party/blink/renderer/core/layout/layout_object.cc [modify] https://crrev.com/71b78364f8a51dd48b4d21bf3c8aa8726d062197/third_party/blink/renderer/core/paint/object_paint_invalidator.cc [modify] https://crrev.com/71b78364f8a51dd48b4d21bf3c8aa8726d062197/third_party/blink/renderer/core/paint/object_paint_invalidator_test.cc
,
May 4 2018
Seems no more flakiness after the CL.
,
May 4 2018
|
|||||||
►
Sign in to add a comment |
|||||||
Comment 1 by noel@chromium.org
, Apr 24 2018