Following tests started crashing recently: inspector/canvas/recording-2d-frameCount.html inspector/canvas/recording-2d-saves.html inspector/canvas/recording-bitmaprenderer-frameCount.html inspector/canvas/recording-webgl-frameCount.html inspector/canvas/recording-webgl-snapshots.html inspector/canvas/setRecordingAutoCaptureFrameCount.html e.g.: https://ews-build.webkit.org/#/builders/17/builds/7186
Flakiness dashboard: https://results.webkit.org/?suite=layout-tests&suite=layout-tests&suite=layout-tests&suite=layout-tests&suite=layout-tests&suite=layout-tests&test=inspector%2Fcanvas%2Frecording-2d-frameCount.html&test=inspector%2Fcanvas%2Frecording-2d-saves.html&test=inspector%2Fcanvas%2Frecording-bitmaprenderer-frameCount.html&test=inspector%2Fcanvas%2Frecording-webgl-frameCount.html&test=inspector%2Fcanvas%2Frecording-webgl-snapshots.html&test=inspector%2Fcanvas%2FsetRecordingAutoCaptureFrameCount.html
Seems regression from either r252790 or r252792.
e.g.: https://build.webkit.org/results/Apple%20Mojave%20Debug%20WK1%20(Tests)/r252792%20(7061)/results.html https://ews-build.webkit.org/results/macOS-High-Sierra-Debug-WK1-Tests-EWS/r384184-7186-clean-tree/results.html
I'm pretty much positive it's r252792. Look at what files changed: https://results.webkit.org/commits?before_id=252792&after_id=252789
<rdar://problem/57446993>
(In reply to Jonathan Bedard from comment #4) > I'm pretty much positive it's r252792. > > Look at what files changed: > https://results.webkit.org/commits?before_id=252792&after_id=252789 Although I suppose that r252792 could have broken the canvas stuff and then r252790 independently broke the inspector stuff.
Oops, I just landed a patch which would make reverting this harder. Let me see if I can figure something out...
Created attachment 384220 [details] Fixes the bug
Created attachment 384222 [details] Fixes the bug
Comment on attachment 384220 [details] Fixes the bug r=me, this actually should be enough as we shouldn’t be adding anything to `m_recordingCanvasIdentifiers` in this loop. If anything, it should already be empty at the end of this loop since each item gets removed inside `InspectorCanvasAgent::didFinishRecordingCanvasFrame`. Thanks for the quick fix!
Comment on attachment 384220 [details] Fixes the bug Clearing flags on attachment: 384220 Committed r252823: <https://trac.webkit.org/changeset/252823>
All reviewed patches have been landed. Closing bug.