1523522 - Skip test_videocontrols_onclickplay.html test

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, enhancement)

Description

[Sotaro Ikeda [:sotaro]]

During testing test_videocontrols_onclickplay.html, it hit ASSERT ain WebRenderBridgeParent.cpp

https://treeherder.mozilla.org/#/jobs?repo=try&revision=4a901d5835974f5f679ac06a11a964dc82be046c&selectedJob=224412557

https://searchfox.org/mozilla-central/rev/bce40231eacba55a6d0dbdf222ff482b5b1fbc71/gfx/layers/wr/WebRenderBridgeParent.cpp#805

Comment 1

[Sotaro Ikeda [:sotaro]]

test_videocontrols_onclickplay.html belongs to mochitest-chrome and it runs test with e10s disabled. It caused hitting the assert.

Comment 2

[Sotaro Ikeda [:sotaro]]

(In reply to Sotaro Ikeda [:sotaro] from comment #1)

test_videocontrols_onclickplay.html belongs to mochitest-chrome and it runs test with e10s disabled. It caused hitting the assert.

In this case ImageBridgeChild::UpdateTextureFactoryIdentifier() is called at first for LayersBackend::LAYERS_WR and then for LayersBackend::LAYERS_BASIC.

Comment 3

[Sotaro Ikeda [:sotaro]]

(In reply to Sotaro Ikeda [:sotaro] from comment #2)

In this case ImageBridgeChild::UpdateTextureFactoryIdentifier() is called at first for LayersBackend::LAYERS_WR and then for LayersBackend::LAYERS_BASIC.

ImageBridgeChild::UpdateTextureFactoryIdentifier() for LayersBackend::LAYERS_BASIC was always called for eWindowType_invisible in parent process. The testing test_videocontrols_onclickplay.html run with e10s disabled, then it caused hitting the assert.

Comment 4

[Sotaro Ikeda [:sotaro]]

Attachment: patch - Do not call ImageBridgeChild::IdentifyCompositorTextureHost() for eWindowType_invisible

Comment 5

[Sotaro Ikeda [:sotaro]]

attachment 9040006 [details] [diff] [review] was not enough to address the test failure, since pop-up windows was also created.

https://treeherder.mozilla.org/#/jobs?repo=try&author=sikeda%40mozilla.com&selectedJob=224898163

Comment 6

[Sotaro Ikeda [:sotaro]]

Attachment: patch - Do not call ImageBridgeChild::IdentifyCompositorTextureHost() for non-webrender widget

Comment 7

[Sotaro Ikeda [:sotaro]]

attachment 9040030 [details] [diff] [review] addressed the test failure.
https://treeherder.mozilla.org/#/jobs?repo=try&author=sikeda%40mozilla.com&selectedJob=224905561

Comment 8

[Sotaro Ikeda [:sotaro]]

Comment on attachment 9040030 [details] [diff] [review]
patch - Do not call ImageBridgeChild::IdentifyCompositorTextureHost() for non-webrender widget

This patch is not a good way to address the test failure.

Comment 9

[Sotaro Ikeda [:sotaro]]

I re-think about test_videocontrols_onclickplay.html. The test run with e10s disabled, and the test failure happens by disabling e10s. But WebRender is not used on windows with disabling e10s. Then we could skip the test for MVP.

Comment 10

[Sotaro Ikeda [:sotaro]]

Attachment: Bug 1523522 - Skip test_videocontrols_onclickplay.html

Comment 11

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/164bfd05f3f7
Skip test_videocontrols_onclickplay.html r=jrmuizel

Comment 12

[Raul Gurzau (:RaulG)]

https://hg.mozilla.org/mozilla-central/rev/164bfd05f3f7