1387467 - Crash in mozilla::layers::FrameBuilder::ProcessContainerLayer

Status: RESOLVED WORKSFORME

(Core :: Graphics: Layers, defect, P3)

Description

[Julian Seward [:jseward]]

This bug was filed from the Socorro interface and is 
report bp-5696fe0d-3ece-4c47-8020-38c730170804.
=============================================================

This is Windows topcrash #17 for the nightly of 20170803134456.

Comment 1

[David Anderson [:dvander] - inactive, e-mail if emergency]

Attachment: diagnostics

I have no earthly idea how these two lines could crash. The first implies we've attached a LayerComposite to a LayerManagerMLGPU - or possibly some unsupported layer type is being used.

The second crash implies that a RefLayerMLGPU requires an intermediate surface. I don't know how that's possible, and it would be a minor pain to support.

Anyway, this patch will make the crashes prerelease only, and log some extra information.

Comment 2

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/c044d8842861
Add diagnostics for the crash in bug 1387467. r=milan

Comment 3

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

Backed out for build bustage at FrameBuilder.cpp:154 (no member GetName):

https://hg.mozilla.org/integration/mozilla-inbound/rev/34567477dd06c7287fc8b742d26220da855e5358

Push with bustage: https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&revision=c044d884286153f6f3912e394d0c90469161503b&filter-resultStatus=testfailed&filter-resultStatus=busted&filter-resultStatus=exception&filter-resultStatus=retry&filter-resultStatus=usercancel
Failure log: https://treeherder.mozilla.org/logviewer.html#?job_id=121166358&repo=mozilla-inbound

[task 2017-08-05T08:44:12.155529Z] 08:44:12     INFO -  In file included from /home/worker/workspace/build/src/obj-firefox/gfx/layers/Unified_cpp_gfx_layers9.cpp:29:0:
[task 2017-08-05T08:44:12.155660Z] 08:44:12     INFO -  /home/worker/workspace/build/src/gfx/layers/mlgpu/FrameBuilder.cpp: In member function 'bool mozilla::layers::FrameBuilder::ProcessContainerLayer(mozilla::layers::ContainerLayer*, mozilla::layers::RenderViewMLGPU*, const RenderTargetIntRect&, mozilla::Maybe<mozilla::gfx::PolygonTyped<mozilla::gfx::UnknownUnits> >&)':
[task 2017-08-05T08:44:12.155731Z] 08:44:12     INFO -  /home/worker/workspace/build/src/gfx/layers/mlgpu/FrameBuilder.cpp:154:48: error: 'class mozilla::layers::ContainerLayer' has no member named 'GetName'
[task 2017-08-05T08:44:12.155774Z] 08:44:12     INFO -         "Layer type is invalid: " << aContainer->GetName();
[task 2017-08-05T08:44:12.155815Z] 08:44:12     INFO -                                                  ^
[task 2017-08-05T08:44:12.155885Z] 08:44:12     INFO -  /home/worker/workspace/build/src/gfx/layers/mlgpu/FrameBuilder.cpp:183:53: error: 'class mozilla::layers::LayerMLGPU' has no member named 'GetName'
[task 2017-08-05T08:44:12.156075Z] 08:44:12     INFO -         "Container layer type is invalid: " << layer->GetName();
[task 2017-08-05T08:44:12.156492Z] 08:44:12     INFO -                                                       ^
[task 2017-08-05T08:44:12.156801Z] 08:44:12     INFO -  /home/worker/workspace/build/src/config/rules.mk:1050: recipe for target 'Unified_cpp_gfx_layers9.o' failed
[task 2017-08-05T08:44:12.156988Z] 08:44:12     INFO -  gmake[5]: *** [Unified_cpp_gfx_layers9.o] Error 1

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/360fa5a78fc9
Add diagnostics for the crash in bug 1387467. r=milan

Comment 5

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

https://hg.mozilla.org/mozilla-central/rev/360fa5a78fc9

Comment 6

[Kan-Ru Chen [:kanru] (UTC+9)]

Still happening and the GraphicsCriticalError annotation looks like: Container layer type is invalid: RefLayerMLGPU

Comment 7

[Milan Sreckovic [:milan] (needinfo for best results)]

Given comment 6, this is new, so let's see if it's nasty.

Comment 8

[David Anderson [:dvander] - inactive, e-mail if emergency]

It looks like a specific theme might be causing this, perhaps by styling browser tabs such that the RefLayer needs an intermediate surface. Advanced Layers can't handle this and it's theoretically not supposed to happen. I'll see if I can find the theme and reproduce the bug.

Comment 9

[David Anderson [:dvander] - inactive, e-mail if emergency]

Every crash has "Simple White X" as the theme, and it's a legacy addon. But even when enabling legacy addons in 57, and modifying the addon manifest to support 57, I can't reproduce this. So maybe it's a false correlation, or maybe someone who works on the addon is testing out 57 support.

Another oddity is that the diagnostic metadata doesn't match the stack trace. The stack is at line 153 but the log message is printing from line 183. Line 153 should be unreachable, so that's confusing.

Anyway, this is a pre-release-only crash so it should be okay to continue monitoring it.

Comment 10

[Firefox Bug Husbandry Bot]

https://wiki.mozilla.org/Bug_Triage/Projects/Bug_Handling/Bug_Husbandry#Release_Cycle_Queries_and_Activities

Move optional bugs to next release

Comment 11

[Sylvestre Ledru [:Sylvestre]]

Moving to p3 because no activity for at least 24 weeks.

Comment 12

[BugBot [:suhaib / :marco/ :calixte]]

Closing because no crashes reported for 12 weeks.