1136507 - WebGL conformance test 1.0.2: conformance/textures/texture-attachment-formats.html

Status: RESOLVED FIXED

(Core :: Graphics: CanvasWebGL, defect)

Description

[Luke]

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:39.0) Gecko/20100101 Firefox/39.0
Build ID: 20150224030228

Steps to reproduce:

https://www.khronos.org/registry/webgl/sdk/tests/conformance/textures/texture-attachment-formats.html?webglVersion=1


Actual results:

conformance/textures/texture-attachment-formats.html (18 of 20 passed)

    failed: at (0, 0) expected: 63,63,63,255 was 64,0,0,255
    failed: at (0, 0) expected: 63,63,63,63 was 64,0,0,64


Expected results:

Chrome passes all the tests. Firefox fails the test with both the ANGLE and OpenGL backends.

Comment 1

[Kelsey Gilbert [:jgilbert]]

The driver is failing to expand L8 data to R8G8B8 as part of ReadPixels.

Comment 2

[Kelsey Gilbert [:jgilbert]]

Attachment: 0001-L8-LA8-and-A8-are-not-renderable.patch

I went through the work of getting L8->RGB8 expansion working with ReadPixels before realizing from a different outstanding conformance test failure that these formats should not be renderable. Oops.

Simple patch though!

Comment 3

[u480271]

Comment on attachment 8577556 [details] [diff] [review]
0001-L8-LA8-and-A8-are-not-renderable.patch

Review of attachment 8577556 [details] [diff] [review]:
-----------------------------------------------------------------

::: dom/canvas/WebGLFramebuffer.cpp
@@ -265,5 @@
>  {
>      switch (sizedFormat) {
> -    case LOCAL_GL_ALPHA8:
> -    case LOCAL_GL_LUMINANCE8:
> -    case LOCAL_GL_LUMINANCE8_ALPHA8:

I'm looking at the ES 2.0.25 spec and it only list:
  DEPTH_COMPONENT16
  RGBA4
  RGB5_A1
  RGB565
  STENCIL_INDEX8

Are RGB8, RGBA8, etc added via extensions?

Comment 4

[Kelsey Gilbert [:jgilbert]]

(In reply to Dan Glastonbury :djg :kamidphish from comment #3)
> Comment on attachment 8577556 [details] [diff] [review]
> 0001-L8-LA8-and-A8-are-not-renderable.patch
> 
> Review of attachment 8577556 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> ::: dom/canvas/WebGLFramebuffer.cpp
> @@ -265,5 @@
> >  {
> >      switch (sizedFormat) {
> > -    case LOCAL_GL_ALPHA8:
> > -    case LOCAL_GL_LUMINANCE8:
> > -    case LOCAL_GL_LUMINANCE8_ALPHA8:
> 
> I'm looking at the ES 2.0.25 spec and it only list:
>   DEPTH_COMPONENT16
>   RGBA4
>   RGB5_A1
>   RGB565
>   STENCIL_INDEX8
> 
> Are RGB8, RGBA8, etc added via extensions?

Textures that are RGB[A]/UNSIGNED_BYTE become RGB[A]8 by our format canonization.

I think that such textures are not guaranteed under GLES2, though I don't know whether they're forbidden by a close reading of the spec either.

The WebGL1 spec guarantees that a RGBA/UNSIGNED_BYTE texture is renderable.

Comment 5

[Kelsey Gilbert [:jgilbert]]

 https://hg.mozilla.org/integration/mozilla-inbound/rev/bd5ffbaf30a6

Comment 6

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/bd5ffbaf30a6