1311952 - Flash video is black though sound is normal on certain site(NHK world TV Live)

Status: RESOLVED FIXED

(Core :: Graphics: Layers, defect)

Description

[Alice0775 White]

[Tracking Requested - why for this release]: NHK world TV Live Flash video would not playback properly

Build Identifier:
https://hg.mozilla.org/mozilla-central/rev/99a239e1866a57f987b08dad796528e4ea30e622
Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101 Firefox/52.0 ID:20161020030211

Reproducible: always

The problem is only in Nightly52.0a1.

Steps To Reproduce:
1. Open NHK world TV Live[1]
2. Wait to autostart the video

[1] http://www3.nhk.or.jp/nhkworld/en/live/

Actual Results:
The flash video is black though sound is normal.
And there are no longer available video controls when mouse hover over the video.

Expected Results:
The flash video should playback properly.
Video controls should appear when mouse hover over the video.


Regression window:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=61da7056917a6361f5499f5431b83421979c5233&tochange=1ecca76e7b94b1356a67cae946d7d4a2a117f114

Regressed by: 1ecca76e7b94	Sotaro Ikeda — Bug 1305490 - Use ALLOC_UPDATE_FROM_SURFACE flag r=mattwoodrow



Oddly enough, The NHK world live video[1] is continuously flickering if you play the "On-Demand" video[2] at least once.

[2] Open http://www3.nhk.or.jp/nhkworld/en/vod/ and click a video.

Comment 1

[Alice0775 White]

Attachment: aboutSupport.txt

Comment 2

[Sotaro Ikeda [:sotaro]]

Thanks for the reporting! I succeeded to reproduce the problem. I am going to look into it.

Comment 3

[Sotaro Ikeda [:sotaro]]

(In reply to Alice0775 White from comment #0)
> 
> Oddly enough, The NHK world live video[1] is continuously flickering if you
> play the "On-Demand" video[2] at least once.
> 
> [2] Open http://www3.nhk.or.jp/nhkworld/en/vod/ and click a video.

I failed to reproduce the flickering part.

Comment 4

[Sotaro Ikeda [:sotaro]]

By 1305490, TextureClient was fallback to BufferTexture. It seems related to the bug.

Comment 5

[Marcia Knous [:marcia]]

Tracking 52+ for this regression which covers the flash video content on some sites.

Comment 6

[Sotaro Ikeda [:sotaro]]

(In reply to Sotaro Ikeda [:sotaro] from comment #4)
> By 1305490, TextureClient was fallback to BufferTexture. It seems related to
> the bug.

I locally backed out bug 1305490 fix, but the problem did not addressed.

Comment 7

[Sotaro Ikeda [:sotaro]]

mozregression showed the following.

https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=da083c6367526ab8d30d1b07b94b2c56b77d7831&tochange=c941e8a139541661f1ca69c673959345e76f20bf

Comment 8

[Alice0775 White]

(In reply to Sotaro Ikeda [:sotaro] from comment #7)
> mozregression showed the following.
> 
> https://hg.mozilla.org/integration/mozilla-inbound/
> pushloghtml?fromchange=da083c6367526ab8d30d1b07b94b2c56b77d7831&tochange=c941
> e8a139541661f1ca69c673959345e76f20bf

NO.
Regression range in comment #0 is correct for me. I checked this with clean profile just now.

Comment 9

[Alice0775 White]

@Sotaro Ikeda
Just confirmation,
Hardware Acceleration of Flash Player should be Enabled(default).

Comment 10

[Sotaro Ikeda [:sotaro]]

(In reply to Alice0775 White from comment #9)
> @Sotaro Ikeda
> Just confirmation,
> Hardware Acceleration of Flash Player should be Enabled(default).

Thanks! When I enabled the Hardware Acceleration of Flash Player, I got the same regression.

Comment 11

[Sotaro Ikeda [:sotaro]]

(In reply to Sotaro Ikeda [:sotaro] from comment #7)
> mozregression showed the following.
> 
> https://hg.mozilla.org/integration/mozilla-inbound/
> pushloghtml?fromchange=da083c6367526ab8d30d1b07b94b2c56b77d7831&tochange=c941
> e8a139541661f1ca69c673959345e76f20bf

It was when the Hardware Acceleration of Flash Player was disabled.

Comment 12

[Sotaro Ikeda [:sotaro]]

When the Hardware Acceleration of Flash Player was enabled, PluginInstanceParent::RecvShowDirectDXGISurface() was called instead of PluginInstanceParent::RecvShowDirectBitmap().

Comment 13

[Sotaro Ikeda [:sotaro]]

The problem happens by failing at D3D11SurfaceHolder::CopyToTextureClient(). It expects only D3D11TextureData.
  https://dxr.mozilla.org/mozilla-central/source/dom/plugins/ipc/D3D11SurfaceHolder.cpp#43

Comment 14

[Sotaro Ikeda [:sotaro]]

Bug 1305490 added ALLOC_UPDATE_FROM_SURFACE flag also to RecvShowDirectDXGISurface(). But it is not necessary for RecvShowDirectDXGISurface() since it does not use TextureClient::UpdateFromSurface(), instead it uses D3D11SurfaceHolder::CopyToTextureClient().

Comment 15

[Sotaro Ikeda [:sotaro]]

Attachment: patch - Remove ALLOC_UPDATE_FROM_SURFACE flag from RecvShowDirectDXGISurface()

Comment 16

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/c72883b45f42
Remove ALLOC_UPDATE_FROM_SURFACE flag from RecvShowDirectDXGISurface() r=mattwoodrow

Comment 17

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/c72883b45f42

Comment 18

[Alice0775 White]

The problem is back, See Bug 1317131.