1087932 - Some WebGL games/animations fail to work on some Radeon 3000 devices starting with Firefox 32.0.3

Status: RESOLVED WORKSFORME

(Core :: Graphics: CanvasWebGL, defect)

Description

[Camelia Badau [:cbadau], Desktop Test Engineering]

Reproducible on Firefox 33.0.1 (BuildID: 20141021130815)
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:33.0) Gecko/20100101 Firefox/33.0

Graphics: 
Adapter Description ATI Radeon 3000 Graphics
Adapter Drivers aticfx64 aticfx64 aticfx32 aticfx32 atiumd64 atidxx64 atiumdag atidxx32 atiumdva atiumd6a atitmm64
Adapter RAM 512
Device ID 0x9616
Direct2D Enabled true
DirectWrite Enabled true (6.2.9200.16492)
Driver Date 4-19-2011
Driver Version 8.850.0.0
GPU #2 Active false
GPU Accelerated Windows 1/1 Direct3D 11 (OMTC)
Vendor ID 0x1002
WebGL Renderer Google Inc. -- ANGLE (ATI Radeon 3000 Graphics Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote true
AzureCanvasBackend direct2d
AzureContentBackend direct2d
AzureFallbackCanvasBackend cairo
AzureSkiaAccelerated 0


Steps to reproduce: 
1. Launch Firefox.
2. Play some WebGL games/animations (http://www.3dsitelinks.com/webgl-games, http://www.webgl.com/category/webgl-games/, http://dl.dropboxusercontent.com/u/6983010/wserv/gexp_pulpo/index.html, http://madebyevan.com/webgl-water/).


Expected results: The games are opened and played without errors.

Actual results: Some games/animations fail to work. 
e.g: http://racer.nomo.hu/, https://triggerrally.com/, http://chandlerprall.github.io/Physijs/examples/jenga.html, http://radmars.com/tessitron/ludumDare/, https://dl.dropboxusercontent.com/u/6983010/wserv/gexp_pulpo/index.html (Controls are displayed, but the page is black), http://madebyevan.com/webgl-water/ ("Error: Rendering to this texture is not supported (incomplete framebuffer)")

Notes: 
1. The issue is reproducible also on Windows 8.1 32bit (with the same graphics settings).
2. All those games/animations DO NOT WORK in 33.0 and 33.0.1, but DO WORK in 32.0.3. 
3. The issue is only reproducible with this graphics settings.

Comment 1

[u279076]

[Tracking Requested - why for this release]: this is a regression from Firefox 32.0.3

Comment 2

[Sylvestre Ledru [:Sylvestre]]

Camelia, do you still see this issue with 33.1 candidate? Thanks

Comment 3

[Camelia Badau [:cbadau], Desktop Test Engineering]

Yes, the issue is reproducible with 33.1 candidate (on Windows 7 64bit, Windows 8.1 32bit - on AMD graphics).

Comment 4

[Sylvestre Ledru [:Sylvestre]]

Ok. Thanks.

Milan, anybody could help here? Thanks

Comment 5

[Milan Sreckovic [:milan] (needinfo for best results)]

Jeff, David, Benoit, any of you can reproduce this?

Comment 6

[Kelsey Gilbert [:jgilbert]]

http://racer.nomo.hu/ doesn't work in Nightly or Chrome, likely a problem with the site.

The others all look fine modulo a bug I think I'm seeing that's probably unrelated.

I'm on NVidia though. I don't have hardware for Win+AMD.

Comment 7

[Kelsey Gilbert [:jgilbert]]

(In reply to Jeff Gilbert [:jgilbert] from comment #6)
> http://racer.nomo.hu/ doesn't work in Nightly or Chrome, likely a problem
> with the site.
> 
> The others all look fine modulo a bug I think I'm seeing that's probably
> unrelated.
> 
> I'm on NVidia though. I don't have hardware for Win+AMD.

Actually, I do have hardware, I just don't have Window installed on it. I can take a look at doing this, but it would be better if someone else can give quicker turn-around.

Comment 8

[David Anderson [:dvander] - inactive, e-mail if emergency]

Looks like I don't have any ATI devices either.

Comment 9

[Cristina Madaras, QA [:CristinaM]]

This WebGL games fail to work on Firefox 33.0.3 build 2 on Windows 7 x64bit:
http://www.3dtin.com/
http://www.holoduke.nl/3dbricks/

Comment 10

[[SV Manager] Florin Mezei, QA (:FlorinMezei)]

(In reply to Cristina Madaras, QA [:CristinaM] from comment #9)
> This WebGL games fail to work on Firefox 33.0.3 build 2 on Windows 7 x64bit:
> http://www.3dtin.com/
> http://www.holoduke.nl/3dbricks/

I can confirm that Cristina uses the same ATI Radeon 3000 Graphics card.

Comment 11

[Vasilica Tamas, QA [:vtamas]]

I was able to reproduce this issue on Windows 8.1 x32 (graphics card: ATI Radeon 3000 Graphics) using Firefox 33.0.3 build 2 for these games/animations: http://www.holoduke.nl/3dbricks/ ,
http://inear.se/fireshader/ .

Comment 12

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

The closest that we have to this hardware here in the Toronto office is a ATI Mobilidy Radeon HD 3650, with an older 8.641 ATI driver (that's from late 2009 or early 2010). It does not reproduce this issue, unfortunately.

It would be interesting to understand what is happening on this device / driver. However, given that this seems to only reproduce on some old ATI devices, the most natural way to think of this bug is as a likely driver bug. It's still unfortunate and strange that an update in Firefox or in ANGLE broke support for it, but there are lots of ways that this can be the case and has happened in the past. Typically, these bugs are not necessarily worth trying to work around.

Comment 13

[Sylvestre Ledru [:Sylvestre]]

We released 33.1. I don't think it is a driver for a 33.1.1.

Comment 14

[Camelia Badau [:cbadau], Desktop Test Engineering]

Here is the regression range: 

Last good revision: 2014-07-06 (buildID: 20140706030226)
First bad revision: 2014-07-07 (buildID: 20140707030202)

Pushlog: 
https://hg.mozilla.org/mozilla-central/pushloghtml?changeset=1dc6b294800d

Comment 15

[Arthur K. (he/him)]

(In reply to Camelia Badau, QA [:cbadau] from comment #3)
> Yes, the issue is reproducible with 33.1 candidate (on Windows 7 64bit,
> Windows 8.1 32bit - on AMD graphics).

Can you try AMD's Autodetect Tool (http://support.amd.com/en-us/download/auto-detect-tool) and see if it finds newer drivers for your Radeon 3000?

Comment 16

[Jeff Muizelaar [:jrmuizel]]

(In reply to Camelia Badau, QA [:cbadau] from comment #14)
> Here is the regression range: 
> 
> Last good revision: 2014-07-06 (buildID: 20140706030226)
> First bad revision: 2014-07-07 (buildID: 20140707030202)
> 
> Pushlog: 
> https://hg.mozilla.org/mozilla-central/pushloghtml?changeset=1dc6b294800d

Are you sure about that regression range? There's only one changeset listed and there's nothing remotely suspicious in that that range.

Comment 17

[Robert Kaiser]

(In reply to Camelia Badau, QA [:cbadau] from comment #14)
> Last good revision: 2014-07-06 (buildID: 20140706030226)
> First bad revision: 2014-07-07 (buildID: 20140707030202)

Putting those Build IDs into http://bsmedberg.github.io/firefox-regression-range-finder/ I get https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=81691a55e60f&tochange=1dc6b294800d

I guess Camelia mistakenly only put the changeset of the first bad build in here.

Comment 18

[Jeff Muizelaar [:jrmuizel]]

The ANGLE update is what I'd expect.

Comment 19

[Lawrence Mandel [:lmandel] (use needinfo)]

I reviewed this bug with Jeff. We're not going to be able to fix this in 34.

Comment 20

[Lukas Blakk [:lsblakk] use ?needinfo]

If we're not able to backout the ANGLE update, and users have lived with this fallout for three releases now, there's not much advantage to continuing to track this bug.

Comment 21

[u279076]

Jeff, is this something we still plan on addressing? If not, please close this bug.

Comment 22

[Milan Sreckovic [:milan] (needinfo for best results)]

Is it still a problem?  If it is, we should leave it open.

Comment 23

[Bogdan Maris, Desktop Test Engineering]

I reproduced the initial issue using Firefox 33.0.3 on Windows 7 64-bit with AMD Radeon HD 3000.
Using latest Nightly 45.0a1 I was unable to reproduce the issue on  http://www.holoduke.nl/3dbricks/ and http://inear.se/fireshader/.
I think it is safe to call this resolved worksforme.