892769 - WebGL fake-black-texture bug in DrawArrays and DrawElements

Status: VERIFIED FIXED

(Core :: Graphics: CanvasWebGL, defect)

Description

[Guillaume Abadie]

Forget to undo fake black texture if fake vertex attribute 0 fail in WebGLContext::DrawArrays and WebGLContext::DrawElements.

Comment 1

[Guillaume Abadie]

Attachment: patch for landing

Fixing the problem

Comment 2

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/315e1eb41e82

Comment 3

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/315e1eb41e82

Comment 4

[Kelsey Gilbert [:jgilbert]]

Comment on attachment 774335 [details] [diff] [review]
patch for landing

[Approval Request Comment]
Bug caused by (feature/regressing bug #): WebGL
User impact if declined: Some likely-erroneous draw calls can cause subsequent draw calls to render incorrectly.
Testing completed (on m-c, etc.): m-c
Risk to taking this patch (and alternatives if risky): very low
String or IDL/UUID changes made by this patch: none

I think the reward is small enough that we shouldn't try to take this in Beta, though.

Comment 5

[bhavana bajaj [:bajaj]]

Comment on attachment 774335 [details] [diff] [review]
patch for landing

low risk enough for aurora uplift and looks like a Fx22 WebGl regression.

Comment 6

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/d3a462c5d035

Comment 7

[u279076]

Guillaume, is there a testcase QA can use to verify this is fixed?

Comment 8

[Guillaume Abadie]

(In reply to Anthony Hughes, Mozilla QA (:ashughes) from comment #7)
> Guillaume, is there a testcase QA can use to verify this is fixed?

Yes, it has been committed in the official WebGL conformance test suite 1.0.3. But we don't run it on tbpl yet because we have the 1.0.1 and upgrading is very very painful right now.

Comment 9

[u279076]

Are there instructions or documentation on how to run the WebGL CTS 1.0.3?

Comment 10

[Benoit Jacob [:bjacob] (mostly away)]

(In reply to Anthony Hughes, Mozilla QA (:ashughes) from comment #9)
> Are there instructions or documentation on how to run the WebGL CTS 1.0.3?

https://www.khronos.org/registry/webgl/sdk/tests/webgl-conformance-tests.html

Comment 11

[u279076]

Thank you Benoit.

Comment 12

[Ioana (away)]

I tried running all the conformance tests on Firefox 25 beta 1, on Mac OS X 10.8.4 32bit and 64bit, Windows 7 64bit, and Ubuntu 13.04 32bit. On Mac and Ubuntu, the build freezes until it gets restarted manually. On Windows it crashed: https://crash-stats.mozilla.com/report/index/98e11a76-a56e-48c2-97a6-d29082130920.

Which set(s) of tests should I run for this bug in particular?

Comment 13

[Benoit Jacob [:bjacob] (mostly away)]

(In reply to Ioana Budnar, QA [:ioana] from comment #12)
> I tried running all the conformance tests on Firefox 25 beta 1, on Mac OS X
> 10.8.4 32bit and 64bit, Windows 7 64bit, and Ubuntu 13.04 32bit. On Mac and
> Ubuntu, the build freezes until it gets restarted manually. On Windows it
> crashed:
> https://crash-stats.mozilla.com/report/index/98e11a76-a56e-48c2-97a6-
> d29082130920.

Ouch. Yes, that 1.0.3 WebGL test suite is very heavy-hitting. Didn't realize that it brought both Mac and Windows on their knees. It appears that we have some more working-around-system-bugs to do... sorry.

> 
> Which set(s) of tests should I run for this bug in particular?

You only need to run this single test:
https://www.khronos.org/registry/webgl/sdk/tests/conformance/textures/texture-fakeblack.html

Comment 14

[Ioana (away)]

Verified as fixed on Firefox 25 beta 3 (20130926170421), on Windows 7 64bit, Mac OSX 10.8.4 64bit and Ubuntu 13.04 32bit.

The texture-fakeblack test passes every time.

Comment 15

[u279076]

(In reply to Benoit Jacob [:bjacob] from comment #13)
> (In reply to Ioana Budnar, QA [:ioana] from comment #12)
> > I tried running all the conformance tests on Firefox 25 beta 1, on Mac OS X
> > 10.8.4 32bit and 64bit, Windows 7 64bit, and Ubuntu 13.04 32bit. On Mac and
> > Ubuntu, the build freezes until it gets restarted manually. On Windows it
> > crashed:
> > https://crash-stats.mozilla.com/report/index/98e11a76-a56e-48c2-97a6-
> > d29082130920.
> 
> Ouch. Yes, that 1.0.3 WebGL test suite is very heavy-hitting. Didn't realize
> that it brought both Mac and Windows on their knees. It appears that we have
> some more working-around-system-bugs to do... sorry.

Benoit, do you need a follow-up bug report?

Comment 16

[Benoit Jacob [:bjacob] (mostly away)]

No, that would be too much of a grab-bag. There are probably upwards of 10 different issues to solve here. If you find a particular WebGL conformance test page (out of the hundreds there) that triggers some crash or other bad behavior on a particular OS/driver, then you are very welcome to file a bug on it, that is very useful, but be warned that doing it consistently for all the WebGL conformance tests is going to be a fairly large task (though a useful one).