Closed Bug 992874 Opened 6 years ago Closed 6 years ago

Firefox 28 Crashes when creating a big WebGL Context


(Core :: Graphics: Layers, defect)

28 Branch
Windows 7
Not set



Tracking Status
firefox28 --- wontfix
firefox29 + wontfix
firefox30 + verified
firefox31 --- verified


(Reporter: nerom86, Assigned: jgilbert)


(Keywords: crash)

Crash Data


(2 files)

Attached file webglbasic.html
User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/33.0.1750.154 Safari/537.36

Steps to reproduce:

Create a WebGL Context (Simple example) and calling it

See attached file for a html file causing the issue

Actual results:

Browser crashes

Expected results:

It shouldnt crash ;) 
Chrome has no problems with it
Component: Untriaged → Canvas: WebGL
Product: Firefox → Core
Could you please take care of this until the next version? It is reproducible on multiple systems no matter what hardware. It always crashes firefox 28.

Just click on the attached html and FF28 crashes completely

here the one i am using:

Adapter Description	NVIDIA GeForce GT 630
Adapter Drivers	nvd3dumx,nvwgf2umx,nvwgf2umx nvd3dum,nvwgf2um,nvwgf2um
Adapter RAM	2048
Device ID	0x0fc2
Direct2D Enabled	true
DirectWrite Enabled	true (6.2.9200.16571)
Driver Date	3-4-2014
Driver Version
GPU #2 Active	false
GPU Accelerated Windows	1/1 Direct3D 10
Vendor ID	0x10de
WebGL Renderer	Google Inc. -- ANGLE (NVIDIA GeForce GT 630 Direct3D9Ex vs_3_0 ps_3_0)
windowLayerManagerRemote	false
AzureCanvasBackend	direct2d
AzureContentBackend	direct2d
AzureFallbackCanvasBackend	cairo
AzureSkiaAccelerated	0
I cannot reproduce this on my Win8.1 MBP, but I think I see what the problem is. It looks like mTexture creation fails in CanvasLayerD3D10, and later we dereference the null mTexture.
Component: Canvas: WebGL → Graphics: Layers
Attached patch handle-bad-texSplinter Review
Assignee: nobody → jgilbert
Ever confirmed: true
Attachment #8406407 - Flags: review?(bas)
Hi, thanks for looking into this. Maybe it works on Macbook Pro but i have tested several PCs (with Win7 and win8) , NVIDIA or AMD - they all crash. Also i dont know checking for mTexture == null does actually solve the issue (maybe just prevents the crash)

Both IE11 and Chrome have no issues at all with a big WebGL Context (actually chrome has too, because sometimes not rendered correctly). But IE working perfectly.
Right, my patch just fixes the crash. I'm not sure about the best way to handle these large layers. For non-basic ShSurfs (non-readback paths), we handle it fine.
Comment on attachment 8406407 [details] [diff] [review]

Review of attachment 8406407 [details] [diff] [review]:

::: gfx/layers/d3d10/CanvasLayerD3D10.cpp
@@ +117,5 @@
>    } else if (mIsD2DTexture) {
>      return;
>    }
> +  if (!mTexture)

nit: single line if-statements should have { }
Attachment #8406407 - Flags: review?(bas) → review+
Closed: 6 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla31
Comment on attachment 8406407 [details] [diff] [review]

[Approval Request Comment]
Bug caused by (feature/regressing bug #): unknown
User impact if declined: Some windows users could crash on large-webgl-canvas pages.
Testing completed (on m-c, etc.): on m-c
Risk to taking this patch (and alternatives if risky): very low
String or IDL/UUID changes made by this patch: none
Attachment #8406407 - Flags: approval-mozilla-beta?
Attachment #8406407 - Flags: approval-mozilla-aurora?
Comment on attachment 8406407 [details] [diff] [review]

Let's get this on Aurora and also see if QA can reproduce.  If we can reproduce the crash and know this low risk fix takes care of it, we can consider taking this in final 29 builds on Monday.
Attachment #8406407 - Flags: approval-mozilla-aurora? → approval-mozilla-aurora+
Flags: needinfo?(jbecerra)
Crash Signature: [@ mozilla::layers::CanvasLayerD3D10::UpdateSurface() ]
Keywords: crash
I haven't been able to reproduce this on the latest Fx30 or Fx29 on the hardware at hand, however if the reporter can with a variety of PCs with Win7 and Win 8, then we may want to consider this for Fx29. What do you think Jeff?
Flags: needinfo?(jbecerra) → needinfo?(jgilbert)
(In reply to juan becerra [:juanb] from comment #12)
> I haven't been able to reproduce this on the latest Fx30 or Fx29 on the
> hardware at hand, however if the reporter can with a variety of PCs with
> Win7 and Win 8, then we may want to consider this for Fx29. What do you
> think Jeff?

This is really low risk, so I think this is worth it.
Flags: needinfo?(jgilbert)
Comment on attachment 8406407 [details] [diff] [review]

We are not going to make an other build for 29. So, it is too late for this release...
Attachment #8406407 - Flags: approval-mozilla-beta? → approval-mozilla-beta-
Reproduced the crash using old Nightly (2014-04-07) on Windows 7 64bit, verified that the issue is fixed on Firefox 30 beta 3 and latest Aurora.
You need to log in before you can comment on or make changes to this bug.