1736793 - Add SurfaceTexture support to GLBlitHelper::BlitSdToFramebuffer()

Status: RESOLVED FIXED

(Core :: Graphics: CanvasWebGL, enhancement)

Description

[Sotaro Ikeda [:sotaro]]

Blit support of SurfaceTexture was removed by Bug 1640607.

Comment 1

[Sotaro Ikeda [:sotaro]]

Attachment: Bug 1736793 - Add SurfaceTexture support to GLBlitHelper::BlitSdToFramebuffer()

Comment 2

[Sotaro Ikeda [:sotaro]]

I confirmed that that the patch work with pref "media.geckoview.autoplay.request" = false on geckoview example app at the following.
https://developer.mozilla.org/en-US/docs/Web/API/WebGL_API/Tutorial/Animating_textures_in_WebGL

Comment 3

[Sotaro Ikeda [:sotaro]]

The patch seems to cause some test failures on Pixel2.
https://treeherder.mozilla.org/jobs?repo=try&revision=3faac087a5da40dbdfa297aab91c3b9923754d2f&selectedTaskRun=d3U4sCwZQDS-D7DjXfErpA.0

Comment 4

[Sotaro Ikeda [:sotaro]]

The test is disabled on emulator.
https://searchfox.org/mozilla-central/rev/011ed92913b38e950977ab3fc56ae68a8f3bca12/dom/canvas/test/webgl-conf/generated-mochitest.ini#6

Comment 5

[Sotaro Ikeda [:sotaro]]

Attachment: npot-video-1920x1080.mpeg

Comment 6

[Sotaro Ikeda [:sotaro]]

Attachment: npot-video-1920x1080.mp4

Comment 7

[Sotaro Ikeda [:sotaro]]

(In reply to Sotaro Ikeda [:sotaro] from comment #4)

The test is disabled on emulator.
https://searchfox.org/mozilla-central/rev/011ed92913b38e950977ab3fc56ae68a8f3bca12/dom/canvas/test/webgl-conf/generated-mochitest.ini#6

It seems that "textureSize(tex, 0).y" == 0 when TexUnpackImage::TexOrSubImage() was used.
"textureSize(tex, 0).y" was valid when TexUnpackSurface::TexOrSubImage() (slow path)was used.

Comment 8

[Sotaro Ikeda [:sotaro]]

:jgilbert, do you have any ideas about why textureSize(tex, 0).y" was 0?

Comment 9

[Imanol Fernandez]

I have tested this patch on Oculus Quest 2 using this sample: https://immersive-web.github.io/webxr-samples/stereo-video.html

I found two problems:

• The video is upside down
• There is some flickering only when the video plays (are we missing any makeCurrent() calls?)

Comment 10

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

There's a r+ patch which didn't land and no activity in this bug for 2 weeks.
:sotaro, could you have a look please?
For more information, please visit auto_nag documentation.

Comment 11

[Kelsey Gilbert [:jgilbert]]

(In reply to Sotaro Ikeda [:sotaro] from comment #7)

(In reply to Sotaro Ikeda [:sotaro] from comment #4)

The test is disabled on emulator.
https://searchfox.org/mozilla-central/rev/011ed92913b38e950977ab3fc56ae68a8f3bca12/dom/canvas/test/webgl-conf/generated-mochitest.ini#6

It seems that "textureSize(tex, 0).y" == 0 when TexUnpackImage::TexOrSubImage() was used.
"textureSize(tex, 0).y" was valid when TexUnpackSurface::TexOrSubImage() (slow path)was used.

textureSize == 0 is generally from incomplete textures, or maybe we just never initialized it.
It's possible for e.g. TEXTURE_MIN_FILTER to default to trying to mipmap, which will cause incomplete textures like this.
My guess is either the upload is failing or the filter is asking for mipmaps and causing an "incomplete" texture.

Comment 12

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/c333871e1c59
Add SurfaceTexture support to GLBlitHelper::BlitSdToFramebuffer() r=jgilbert,gfx-reviewers

Comment 13

[Cristina Cozmuta (:CrissCozmuta)]

https://hg.mozilla.org/mozilla-central/rev/c333871e1c59