1251382 - Animated gifs set as background blink

Status: VERIFIED FIXED

(Core :: Graphics, defect)

Description

[arni2033]

>>>   My Info:   Win7_64, Nightly 47, 32bit, ID 20160224030246
STR:
1. Open http://www.albinoblacksheep.com/games/missilegame3d
2. Scroll the page to the end
3. Hover mouse over the white area in #footer
4. Wait until "Mario" and "Luigi" gifs will appear at the bottom of that white area

AR:  Those 2 gifs do blink.
ER:  They shouldn't blink.

Please improve the Summary

This is regression between 2015-10-30 and 2015-10-31, Firefox 45. (regressionwindow-wanted)
> https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=540fe6d9b147b3e37713aeeae944fad7661a5685&tochange=58593839ebcedb976068fca5901af41cf48912d9

Comment 1

[Markus Stange [:mstange]]

Confirmed. I see this on OS X, too.

Comment 2

[Alice0775 White]

Attachment: reduced html

Regression window:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=540fe6d9b147b3e37713aeeae944fad7661a5685&tochange=58593839ebcedb976068fca5901af41cf48912d9

Comment 3

[Alice0775 White]

Via local build,
Last Good: d9da665e6f0d
First Bad: 7e3476b5b99d

Triggered by: 7e3476b5b99d	Markus Stange — Bug 947062 - Layerize background images with animated background-position if possible. r=mattwoodrow

Comment 4

[Markus Stange [:mstange]]

There seems to be an existing issue when this animated gif is used as an ImageLayer. Bug 947062 made us use an ImageLayer on this page, so it exposed the bug. But the underlying issue seems to have existed before.

Comment 5

[Markus Stange [:mstange]]

Attachment: different testcase without background-position (but still with ImageLayer)

Alice, can you find a new regression range with this testcase?

Comment 6

[Alice0775 White]

(In reply to Markus Stange [:mstange] from comment #5)
> Created attachment 8724027 [details]
> different testcase without background-position (but still with ImageLayer)
> 
> Alice, can you find a new regression range with this testcase?

Regression window:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=e0f5ea8e9082&tochange=bd4287c14070

Regressed by: bd4287c14070	Matt Woodrow — Bug 1039020 - Layerize background-attachment:fixed for all elements, not just the root one. r=roc

Comment 7

[Markus Stange [:mstange]]

Thanks! This is a surprising result - again we just found the cause of the ImageLayer, but not why the ImageLayer is buggy. So this goes even further back.

Seth, can you look into this? If it makes your job easier, I can try to create another testcase that uses still another method of forcing an ImageLayer, and then we can try to find a new regression range. Would you prefer that or do you just want to debug the bug?

Comment 8

[Seth Fowler [:seth] [:s2h]]

(In reply to Markus Stange [:mstange] from comment #7)
> Thanks! This is a surprising result - again we just found the cause of the
> ImageLayer, but not why the ImageLayer is buggy. So this goes even further
> back.
> 
> Seth, can you look into this? If it makes your job easier, I can try to
> create another testcase that uses still another method of forcing an
> ImageLayer, and then we can try to find a new regression range. Would you
> prefer that or do you just want to debug the bug?

Let me take a look. We've encountered some situations where we didn't invalidate the ImageContainer associated with the ImageLayer correctly during animation, but I don't think that should cause this bug. It looks a lot like there's some point in the animation where the ImageContainer just ends up containing garbage...

Comment 9

[Timothy Nikkel (:tnikkel)]

Looks like this got fixed by bug 1247152.

Comment 10

[Liz Henry (:lizzard) (relman/hg->git project)]

Regression from 45, fixed in 48. It seems too late in the 47 beta cycle to me to uplift the patches from bug 1247152, so, wontfix there as well.

Comment 11

[Md.Tarikul Islam Oashi]

I have reproduced this bug with Release 47.0 on Linux Mint, 64 Bit !

The Bug's fix is verified on Latest beta  48.0b1.

Build ID 	20160606200529
User Agent 	Mozilla/5.0 (X11; Linux x86_64; rv:48.0) Gecko/20100101 Firefox/48.0

Comment 12

[Mihai Boldan, Desktop QA [:mboldan]]

Thanks  Md.Tarikul Islam Oashi for helping us.
I also verified this issue on Firefox 48.0b6 and on Windows 10 x64, Mac OS X 10.10.5 and on Ubuntu 16.04 x64.
I am marking this issue Verified Fixed, since is no longer reproducible.