1348291 - Intermittent [@ mozilla::layers::RotatedContentBuffer::SetBufferProvider] | Assertion failure: !aClient || !mDTBuffer || !mDTBuffer->IsValid(), at /home/worker/workspace/build/src/gfx/layers/RotatedBuffer.h:342

Status: RESOLVED FIXED

(Core :: Graphics: Layers, defect)

Description

[Treeherder Bug Filer]

Filed by: cbook [at] mozilla.com

https://treeherder.mozilla.org/logviewer.html#?job_id=84427453&repo=mozilla-central

https://queue.taskcluster.net/v1/task/Mn8ooKgfSp6_pbCdbu5sKw/runs/0/artifacts/public/logs/live_backing.log

https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://queue.taskcluster.net/v1/task/Mn8ooKgfSp6_pbCdbu5sKw/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1

Comment 1

[Carsten Book [:Tomcat]]

Assertion failure: !aClient || !mDTBuffer || !mDTBuffer->IsValid(), at /home/worker/workspace/build/src/gfx/layers/RotatedBuffer.h:342
   NFO - #01: mozilla::layers::ContentClientRemoteBuffer::BeginPaint [mfbt/RefPtr.h:308]
[task 2017-03-16T21:32:50.982744Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.983839Z] 21:32:50     INFO - #02: mozilla::layers::ContentClientDoubleBuffered::BeginPaint [gfx/layers/client/ContentClient.cpp:538]
[task 2017-03-16T21:32:50.984591Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.984669Z] 21:32:50     INFO - #03: mozilla::layers::ClientPaintedLayer::PaintThebes [gfx/layers/client/ClientPaintedLayer.cpp:41]
[task 2017-03-16T21:32:50.985328Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.985404Z] 21:32:50     INFO - #04: mozilla::layers::ClientPaintedLayer::RenderLayerWithReadback [gfx/src/nsRegion.h:75]
[task 2017-03-16T21:32:50.986016Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.986675Z] 21:32:50     INFO - #05: mozilla::layers::ClientContainerLayer::RenderLayer [gfx/layers/client/ClientContainerLayer.h:57]
[task 2017-03-16T21:32:50.986722Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.987391Z] 21:32:50     INFO - #06: mozilla::layers::ClientLayerManager::EndTransactionInternal [gfx/layers/client/ClientLayerManager.cpp:358]
[task 2017-03-16T21:32:50.987440Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.988101Z] 21:32:50     INFO - #07: mozilla::layers::ClientLayerManager::EndTransaction [gfx/layers/client/ClientLayerManager.cpp:412]
[task 2017-03-16T21:32:50.988724Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.988793Z] 21:32:50     INFO - #08: nsDisplayList::PaintRoot [layout/painting/nsDisplayList.cpp:2256]
[task 2017-03-16T21:32:50.989404Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.989465Z] 21:32:50     INFO - #09: nsLayoutUtils::PaintFrame [mfbt/RefPtr.h:129]
[task 2017-03-16T21:32:50.990062Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.990128Z] 21:32:50     INFO - #10: mozilla::PresShell::Paint [layout/base/PresShell.cpp:6497]
[task 2017-03-16T21:32:50.990746Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.990813Z] 21:32:50     INFO - #11: nsViewManager::ProcessPendingUpdatesPaint [gfx/src/nsRegion.h:75]
[task 2017-03-16T21:32:50.991433Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.992671Z] 21:32:50     INFO - #12: nsViewManager::ProcessPendingUpdatesForView [view/nsViewManager.cpp:411]
[task 2017-03-16T21:32:50.993717Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.994931Z] 21:32:50     INFO - #13: nsViewManager::ProcessPendingUpdates [view/nsViewManager.cpp:1104]
[task 2017-03-16T21:32:50.995896Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.997019Z] 21:32:50     INFO - #14: nsRefreshDriver::Tick [layout/base/nsRefreshDriver.cpp:2031]
[task 2017-03-16T21:32:50.997987Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:50.999052Z] 21:32:50     INFO - #15: mozilla::RefreshDriverTimer::TickRefreshDrivers [layout/base/nsRefreshDriver.cpp:301]
[task 2017-03-16T21:32:50.999992Z] 21:32:50     INFO - 
[task 2017-03-16T21:32:51.001126Z] 21:32:51     INFO - #16: mozilla::RefreshDriverTimer::Tick [layout/base/nsRefreshDriver.cpp:323]
[task 2017-03-16T21:32:51.002061Z] 21:32:51     INFO - 
[task 2017-03-16T21:32:51.003168Z] 21:32:51     INFO - #17: mozilla::VsyncRefreshDriverTimer::RefreshDriverVsyncObserver::TickRefreshDriver [layout/base/nsRefreshDriver.cpp:627]
[task 2017-03-16T21:32:51.004106Z] 21:32:51     INFO - 
[task 2017-03-16T21:32:51.005206Z] 21:32:51     INFO - #18: mozilla::VsyncRefreshDriverTimer::RefreshDriverVsyncObserver::NotifyVsync [layout/base/nsRefreshDriver.cpp:529]
[task 2017-03-16T21:32:51.006161Z] 21:32:51     INFO - 
[task 2017-03-16T21:32:51.007289Z] 21:32:51     INFO - #19: mozilla::layout::VsyncChild::RecvNotify [layout/ipc/VsyncChild.cpp:67]
[task 2017-03-16T21:32:51.008233Z] 21:32:51     INFO - 
[task 2017-03-16T21:32:51.009319Z] 21:32:51     INFO - #20: mozilla::layout::PVsyncChild::OnMessageReceived [obj-firefox/ipc/ipdl/PVsyncChild.cpp:155]
[task 2017-03-16T21:32:51.010262Z] 21:32:51     INFO -

Comment 2

[Carsten Book [:Tomcat]]

matt, nical, could you take a look at this assertion failure. Thanks!

Comment 3

[Nicolas Silva [:nical]]

This seems to indicate ContentClient::BeginPaint() was called during the previous frame but not ContentClient::EndPaint(). My guess is that we early-returned here http://searchfox.org/mozilla-central/rev/c48398abd9f0f074c69f2223260939e30e8f99a8/gfx/layers/client/ClientPaintedLayer.cpp#60 without calling EndPaint.
Let's make sure EndPaint is called before returning.

Comment 4

[Nicolas Silva [:nical]]

Attachment: Call end paint if painting is skipped after calling BeginPaint.

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/acf5abb425bc
Call EndPaint when painting is skipped if BeginPaint was called. r=mattwoodrow

Comment 6

[Wes Kocher (:KWierso)]

https://hg.mozilla.org/mozilla-central/rev/acf5abb425bc

Comment 7

[Ryan VanderMeulen [:RyanVM]]

How far back does this bug go?

Comment 8

[Nicolas Silva [:nical]]

The assertion was added in April 2016, while the code that can cause EndPaint to not be called looks like it's been written in April 2014. back then it might not have been an issue (to not call EndPaint), not sure, but in 2016 it certainly was, at least in the context of taking a snapshot of the page, since the assertion in bug 1262681.

Comment 9

[Ryan VanderMeulen [:RyanVM]]

This grafts cleanly to Aurora/Beta. Per IRC discussion with nical, it's not entirely clear what user impact this bug has, but it's also a low-risk fix.

Comment 10

[Nicolas Silva [:nical]]

Comment on attachment 8850929 [details] [diff] [review]
Call end paint if painting is skipped after calling BeginPaint.

Approval Request Comment
[Feature/Bug causing the regression]:
[User impact if declined]: Not sure, most likely bugs related to snapshots, see bug 1262681. This also causes intermittent assertion failures in debug builds.
[Is this code covered by automated tests?]: Not specifically, although automated tests do trip on the bug this is trying to fix intermittently.
[Has the fix been verified in Nightly?]: Landed without making waves, would require baking for a long while before we are 100% sure the intermittent issue is fixed.
[Needs manual test from QE? If yes, steps to reproduce]: No.
[List of other uplifts needed for the feature/fix]: None.
[Is the change risky?]: Not risky.
[Why is the change risky/not risky?]: Very simple patch that makes sure we leave a function in a more consistent state.
[String changes made/needed]: None.

Comment 11

[Liz Henry (:lizzard)]

Comment on attachment 8850929 [details] [diff] [review]
Call end paint if painting is skipped after calling BeginPaint.

Fixes a failing test, seems low risk, let's uplift to beta.

Comment 12

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/cb1ffcc22ab1

Comment 13

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/releases/mozilla-beta/rev/939c485875a5a118ff7b6c79c79986d818b70a38