1004417 - Delayed redrawing of Apps window on Google's homepage

Status: RESOLVED WORKSFORME

(Core :: Graphics: Layers, defect)

Description

[Evangelos Foutras]

Attachment: Video screen capture of the issue

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:29.0) Gecko/20100101 Firefox/29.0 (Beta/Release)
Build ID: 20140430131441

Steps to reproduce:

The following is tested on a Linux system; Windows doesn't seem to be affected.

1) Go to Google.com
2) Click on the "Apps" button on the top right corner of the page
3) Use the mouse wheel on the Apps window to scroll down


Actual results:

Part (or all) of the Apps window becomes blank. With no further user input, the window is redrawn and displayed properly after approximately four seconds.


Expected results:

The contents of the Apps window should remain visible at all times.

The issue doesn't seem to exist in Firefox 28; it started with Firefox 29. I was also able to reproduce it using a x86_64 nightly build (32.0a1).

The issue was first reported on our Arch Linux bug tracker:
https://bugs.archlinux.org/task/40160

Comment 1

[:Gijs (he/him)]

Could you provide the contents of the video section of about:support, and, assuming you're comfortable on a commandline, try to get a regression range for this issue? You can use the "mozregression" tool for this purpose, which should make it relatively short work, assuming you have a fast internet connection.

Comment 2

[:Gijs (he/him)]

(In reply to :Gijs Kruitbosch from comment #1)
> Could you provide the contents of the video section

Eh, I meant the *graphics* section, sorry.

Comment 3

[Evangelos Foutras]

Thanks for mentioning the mozregression tool; it quickly pointed out the following commit:

http://hg.mozilla.org/integration/mozilla-inbound/rev/d8fb025ca7d2

After building Firefox 29 with the above changeset reverted, the issue is no longer observable.

Also, below is the graphics section of about:support from a x86_64 desktop and a i686 netbook (issue is reproducible on both):

Adapter Description: ATI Technologies Inc. -- AMD Radeon HD 6800 Series
Device ID: AMD Radeon HD 6800 Series
Driver Version: 4.4.12874 Compatibility Profile Context 8.97.100.7
GPU Accelerated Windows: 0/1 Basic
Vendor ID: ATI Technologies Inc.
WebGL Renderer: ATI Technologies Inc. -- AMD Radeon HD 6800 Series
windowLayerManagerRemote: false
AzureCanvasBackend: cairo
AzureContentBackend: cairo
AzureFallbackCanvasBackend: none
AzureSkiaAccelerated: 0

Adapter Description: Intel Open Source Technology Center -- Mesa DRI Intel(R) 945GME x86/MMX/SSE2
Device ID: Mesa DRI Intel(R) 945GME x86/MMX/SSE2
Driver Version: 2.1 Mesa 10.1.1
GPU Accelerated Windows: 0/1 Basic
Vendor ID: Intel Open Source Technology Center
WebGL Renderer: Intel Open Source Technology Center -- Mesa DRI Intel(R) 945GME x86/MMX/SSE2
windowLayerManagerRemote: false
AzureCanvasBackend: cairo
AzureContentBackend: cairo
AzureFallbackCanvasBackend: none
AzureSkiaAccelerated: 0

Comment 4

[:Gijs (he/him)]

(In reply to Evangelos Foutras from comment #3)
> Thanks for mentioning the mozregression tool; it quickly pointed out the
> following commit:
> 
> http://hg.mozilla.org/integration/mozilla-inbound/rev/d8fb025ca7d2

Thanks for quickly getting a regression window down to one commit! Nice!

> After building Firefox 29 with the above changeset reverted, the issue is no
> longer observable.

Great. Let's check with Benoît what's going on here...

Comment 5

[Benoit Girard (:BenWa)]

This is a dupe of bug 984939 I think. But this is very valuable information because it reproduces on Linux and on a non Intel GPU. It's very unlikely that it's a driver bug.

Comment 6

[Evangelos Foutras]

Since bug 984939 was marked as a duplicate of bug 987497, and the latter has a Mac OS X specific fix in Firefox 31, can we reopen this report? The issue still exists in Firefox 31 on Linux.

Comment 7

[Evangelos Foutras]

Appears to be fixed in Firefox 32 but I wasn't able to track down (using mozregression) the specific commit that fixed it.