945243 - webgl-draw-buffers.html conformance test failures

Status: RESOLVED FIXED

(Core :: Graphics: CanvasWebGL, defect)

Description

[Jeff Muizelaar [:jrmuizel]]

I see a couple of failures on my OS X 10.7 machine:
https://www.khronos.org/registry/webgl/sdk/tests/conformance/extensions/webgl-draw-buffers.html

Comment 1

[Jeff Muizelaar [:jrmuizel]]

Guillaume, do you see these failures as well? If you do, do you have time to take a look at them?

Comment 2

[Guillaume Abadie]

Hi Jeff! =) I don't have any on Firefox 26, but I'm compiling the latest morzilla-central right now...

Comment 3

[Guillaume Abadie]

(In reply to Guillaume Abadie from comment #2)
> Hi Jeff! =) I don't have any on Firefox 26, but I'm compiling the latest
> morzilla-central right now...

Mozilla-Central* sorry :)

Comment 4

[Guillaume Abadie]

Oh... That is unfortunate... I'm having to many build error while trying to build on 10.9, and the campus' internet connection is to slow to download Firefox from the FTP right now (in fall 2013... -_-'). Will try tomorrow at the department!

Comment 5

[Benoit Girard (:BenWa)]

FAIL getError expected: INVALID_ENUM. Was NO_ERROR : should not be able to attach pass the max attachment point: gl.COLOR_ATTACHMENT0 + 8

Comment 6

[Jeff Muizelaar [:jrmuizel]]

(In reply to Benoit Girard (:BenWa) from comment #5)
> FAIL getError expected: INVALID_ENUM. Was NO_ERROR : should not be able to
> attach pass the max attachment point: gl.COLOR_ATTACHMENT0 + 8

BenWa gets this on 10.9

on 10.7 I also get:

--------- draw tests -----------
fragement shader
PASS gl.checkFramebufferStatus(gl.FRAMEBUFFER) is gl.FRAMEBUFFER_COMPLETE
PASS gl.checkFramebufferStatus(gl.FRAMEBUFFER) is gl.FRAMEBUFFER_COMPLETE
test that each texture got the correct color.
FAIL attachment 0 should be 255,0,0,0
at (24, 56) expected: 255,0,0,0 was 0,0,0,0
FAIL attachment 1 [details] [diff] [review] should be 0,255,0,0
at (2, 4) expected: 0,255,0,0 was 0,0,0,0
FAIL attachment 2 [details] [diff] [review] should be 255,255,0,0
at (4, 28) expected: 255,255,0,0 was 0,255,0,0
FAIL attachment 3 [details] [diff] [review] should be 0,0,255,0
at (12, 0) expected: 0,0,255,0 was 0,0,0,0
FAIL attachment 4 [details] [diff] [review] should be 255,0,255,0
at (28, 12) expected: 255,0,255,0 was 0,0,255,0
FAIL attachment 5 [details] should be 0,255,255,0
at (8, 20) expected: 0,255,255,0 was 0,0,0,0
FAIL attachment 6 [details] should be 255,255,255,0
at (10, 11) expected: 255,255,255,0 was 255,255,0,0
FAIL attachment 7 [details] should be 0,0,0,255
at (52, 2) expected: 0,0,0,255 was 0,0,0,0

Comment 7

[Kelsey Gilbert [:jgilbert]]

Gross. We'll have to dig a little into this, but the 10.7 failures sound like we should just blocklist it.
The error on 10.9 is easy enough to work around.

Comment 8

[Benoit Girard (:BenWa)]

(In reply to Jeff Muizelaar [:jrmuizel] from comment #6)
> (In reply to Benoit Girard (:BenWa) from comment #5)
> > FAIL getError expected: INVALID_ENUM. Was NO_ERROR : should not be able to
> > attach pass the max attachment point: gl.COLOR_ATTACHMENT0 + 8
> 
> BenWa gets this on 10.9
> 

... and on 10.8 with 'AMD Radeon HD 6750M 1024 MB' early 2011 MBP

Comment 9

[Guillaume Abadie]

Have found the bug :

http://dxr.mozilla.org/mozilla-central/source/content/canvas/src/WebGLFramebuffer.cpp#507

it is suppose to be a >= not >

Comment 10

[Guillaume Abadie]

Could I be assigned on this bug, and let me a little while for compiling firefox on 10.9 ? Thanks !

Comment 11

[Guillaume Abadie]

Attachment: patch revision 1

Here is the patch.

Push to try : https://tbpl.mozilla.org/?tree=Try&rev=7923b44458f4

Comment 12

[Kelsey Gilbert [:jgilbert]]

Comment on attachment 8342970 [details] [diff] [review]
patch revision 1

Review of attachment 8342970 [details] [diff] [review]:
-----------------------------------------------------------------

Awesome. I love when the issue is so simple.

Comment 13

[Kelsey Gilbert [:jgilbert]]

I believe this is what's affected.

Comment 14

[Guillaume Abadie]

(In reply to Jeff Gilbert [:jgilbert] from comment #12)
> Comment on attachment 8342970 [details] [diff] [review]
> patch revision 1
> 
> Review of attachment 8342970 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> Awesome. I love when the issue is so simple.

My Pleasure! Surly my smallest patch ever! =)

https://hg.mozilla.org/integration/mozilla-inbound/rev/701c8c3fb409

We keep in touch for any other small regression like that I can do aside of my university projects! =)

Comment 15

[Phil Ringnalda (:philor)]

https://hg.mozilla.org/mozilla-central/rev/701c8c3fb409