New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 873020 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Aug 29
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: ----

Blocked on:
issue 873022



Sign in to add a comment

rendering.mobile/capitolvolkswagen_mobile_sync_scroll_2018 failing on chromium.perf/Android Nexus6 WebView Perf

Project Member Reported by sheriff-...@appspot.gserviceaccount.com, Aug 10

Issue description

Filed by sheriff-o-matic@appspot.gserviceaccount.com on behalf of nednguyen@chromium.org

rendering.mobile/capitolvolkswagen_mobile_sync_scroll_2018 failing on chromium.perf/Android Nexus6 WebView Perf

Builders failed on: 
- Android Nexus6 WebView Perf: 
  https://ci.chromium.org/buildbot/chromium.perf/Android%20Nexus6%20WebView%20Perf


The failure just happens recently: 
https://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=rendering.mobile%2Fcapitolvolkswagen_mobile_sync_scroll_2018&testType=performance_webview_test_suite

 
Cc: sadrul@chromium.org vmi...@chromium.org
Blockedon: 873022
Cc: herb@google.com
Owner: herb@google.com
Status: Assigned (was: Available)
📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/166290b8640000

Move path fallback logic into drawer by herb@google.com
https://skia.googlesource.com/skia/+/9a52a3946844d50b0d8378a991659dd87728d3a3
0 → 1 (+1)

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions
This seems to be running muse/viz configuration. Is this in production?


signal 11 (SIGSEGV), code 1, fault addr 0x4 in tid 12283 (Chrome_InProcGp)
	pid: 12236, tid: 12283, name: Chrome_InProcGp  >>> org.chromium.webview_shell <<<
	signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x4
	     r0 00000000  r1 00000032  r2 9427e78c  r3 280da0eb
	     r4 9d9b74e0  r5 96488e08  r6 96488e00  r7 95c38580
	     r8 9f8be4cc  r9 00000000  sl 00000032  fp 91630b24
	     ip 00000020  sp 9f8be3b8  lr 9427e600  pc a2cf921a
	
	Stack Trace:
	  RELADDR   FUNCTION                                                                               FILE:LINE
	  0082521a  SkStrikeCache::desperationSearchForPath(SkDescriptor const&, unsigned short, SkPath*)  ../../third_party/skia/src/core/SkStrikeCache.cpp:337:38
This is a test failure on a perf bot--Sadrul, can you help figure out the misconfiguration?
mus is not used on android.

Feature Status:
  2d_canvas           : enabled
  flash_3d            : enabled
  flash_stage3d       : enabled
  flash_stage3d_baseline: enabled
  gpu_compositing     : enabled
  multiple_raster_threads: disabled_off
  native_gpu_memory_buffers: disabled_software
  oop_rasterization   : enabled
  protected_video_decode: unavailable_off
  rasterization       : enabled
  skia_deferred_display_list: disabled_off
  skia_renderer       : disabled_off
  surface_synchronization: disabled_off
  video_decode        : enabled
  viz_display_compositor: disabled_off
  webgl               : enabled
  webgl2              : enabled

oop-d/surface-sync are turned off, oop-r is turned on. Also: in_process_gpu is True.
Owner: sullivan@chromium.org
Owner: herb@google.com
herb, I'm not sure why you assigned this to me? My understanding is that:

1) The CL in #4 broke the test
2) You had some questions on the configuration, Sadrul answered them in #7.

Can you look into why your CL broke the test?
Cc: khushalsagar@google.com
I was confused by...
This is a test failure on a perf bot--Sadrul, can you help figure out the misconfiguration?

Is this misconfigured?

I don't think the desperationSearchForPath should ever be called in production. This call is specific for a muse/viz experiment.

Add Khushal who knows much more about the blink end of things.
The desperationSearchForPath is used by oop-raster which is turned on by default for the android bots. The experiment is also enabled for Android since M69 so it is used in the production path.

I'm not sure why this is crashing, but the fact that we're hitting this now indicates that this CL is likely causing us to hit these fallback cases as well.
Its likely crashing from de-referencing fPathData here: https://cs.chromium.org/chromium/src/third_party/skia/src/core/SkStrikeCache.cpp?q=SkStrikeCache::desperationSearchForPath&sq=package:chromium&g=0&l=337

from->fPathData->fPath != nullptr

There is no guarantee that fPathData can't be null at this point. 
Project Member

Comment 13 by bugdroid1@chromium.org, Aug 21

The following revision refers to this bug:
  https://skia.googlesource.com/skia/+/a5a26227620853d6b050d932465232636e540153

commit a5a26227620853d6b050d932465232636e540153
Author: Herb Derby <herb@google.com>
Date: Tue Aug 21 17:00:31 2018

Fix possible nullptr deref

Check that the fPathData object exists before
derefing to get the path.

BUG= chromium:873020 

Change-Id: Ie381178f49905a0dc3f5c3a3477453d4785507f4
Reviewed-on: https://skia-review.googlesource.com/148385
Auto-Submit: Herb Derby <herb@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
Reviewed-by: Mike Klein <mtklein@google.com>

[modify] https://crrev.com/a5a26227620853d6b050d932465232636e540153/src/core/SkStrikeCache.cpp

Project Member

Comment 14 by bugdroid1@chromium.org, Aug 21

The following revision refers to this bug:
  https://skia.googlesource.com/skia/+/20a6332c94aab2189f168ddc1fab02434fe61530

commit 20a6332c94aab2189f168ddc1fab02434fe61530
Author: Herb Derby <herb@google.com>
Date: Tue Aug 21 17:20:42 2018

Check for too small rect for path

This check was dropped in skia cl/145000. Add it back
in to bring it back into parity with the SkRemoteGlyphCache.

BUG= chromium:873020 

Change-Id: I04c12e54145e0ebd1bdec2f01f5f40f26648f78a
Reviewed-on: https://skia-review.googlesource.com/148384
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Herb Derby <herb@google.com>
Auto-Submit: Herb Derby <herb@google.com>

[modify] https://crrev.com/20a6332c94aab2189f168ddc1fab02434fe61530/src/core/SkGlyphRun.cpp

Project Member

Comment 15 by bugdroid1@chromium.org, Aug 22

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5c5f556a079770fe90ebfdd576cb2f1d3fbc3a29

commit 5c5f556a079770fe90ebfdd576cb2f1d3fbc3a29
Author: skia-chromium-autoroll <skia-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com>
Date: Wed Aug 22 16:51:17 2018

Roll src/third_party/skia ea60b951d735..e6ab998bc2e6 (21 commits)

https://skia.googlesource.com/skia.git/+log/ea60b951d735..e6ab998bc2e6


git log ea60b951d735..e6ab998bc2e6 --date=short --no-merges --format='%ad %ae %s'
2018-08-22 egdaniel@google.com Revert "added support for sk_Dimensions to SkSL"
2018-08-22 nigeltao@google.com Update SkCodec::rewindIfNeeded comment
2018-08-22 angle-skia-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com Roll third_party/externals/angle2 70aeda4942f5..44ae899ae9ce (3 commits)
2018-08-22 recipe-roller@chromium.org Roll recipe dependencies (trivial).
2018-08-22 angle-skia-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com Roll third_party/externals/angle2 d710fee3861a..70aeda4942f5 (5 commits)
2018-08-21 mtklein@google.com let's like, chill out about all these rules, man
2018-08-21 recipe-roller@chromium.org Roll recipe dependencies (trivial).
2018-08-21 herb@google.com Rename SkGlyphRunListDrawer to SkGlyphRunListPainter
2018-08-21 enne@chromium.org Add add_and_true_to_loop_condition workaround
2018-08-21 fmalita@chromium.org [skottie] Split-up Skottie.cpp
2018-08-21 herb@google.com Use SkGlyphRun directly in SkRemoteGlyphCache
2018-08-21 robertphillips@google.com Address fuzzer issue
2018-08-21 kjlubick@google.com Update Pixels to Android P
2018-08-21 egdaniel@google.com Disable FB Fetch in metal ios
2018-08-21 fmalita@chromium.org [skottie/cleanup] Refactor parser state handling
2018-08-21 angle-skia-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com Roll third_party/externals/angle2 7a4f6b8c7d5e..d710fee3861a (2 commits)
2018-08-21 herb@google.com Check for too small rect for path
2018-08-21 ethannicholas@google.com added support for sk_Dimensions to SkSL
2018-08-21 mtklein@google.com treat opaque dsts as src alpha type
2018-08-21 bsalomon@google.com Fix op chaining painter's order violation in GrRenderTargetOpList.
2018-08-21 herb@google.com Fix possible nullptr deref


Created with:
  gclient setdep -r src/third_party/skia@e6ab998bc2e6

The AutoRoll server is located here: https://autoroll.skia.org

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.

CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel;luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel

BUG=chromium: 644669,chromium:829614,chromium:873020,chromium:873020
TBR=fmalita@chromium.org

Change-Id: I4dc802075c67741f2c6db2225eb9e6d3772cd618
Reviewed-on: https://chromium-review.googlesource.com/1185061
Reviewed-by: skia-chromium-autoroll <skia-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com>
Commit-Queue: skia-chromium-autoroll <skia-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com>
Cr-Commit-Position: refs/heads/master@{#585088}
[modify] https://crrev.com/5c5f556a079770fe90ebfdd576cb2f1d3fbc3a29/DEPS

Status: Fixed (was: Assigned)

Sign in to add a comment