Vasilica Tamas, QA [:vtamas] Reporter
	Description • 8 years ago

[Affected versions]:
Firefox 47.0a1 (2016-03-06)
Firefox 46.0a2 (2016-03-06)
Firefox 45.0b10 with e10s and APZ enabled

[Affected platforms]:
Windows 10 64-bit

[Steps to reproduce]:
1.Launch Firefox with clean profile.
2.Navigate to https://www.linkedin.com/ and login.
3.Enter in Print Preview mode.
4.Scroll up and down using mouse wheel.

[Expected Results]:
The scrolling is smooth, without any jerkiness or rendering issues.

[Actual Results]:
Visual glitches appear while scrolling in Print Preview mode.
See screenshots: http://i.imgur.com/ThVil7a.jpg (Firefox 46.0a2)
                 http://i.imgur.com/tfXl0PP.jpg (Firefox 47.0a1) 

[Additional notes]:
- The issue doesn't reproduce with APZ pref disabled.
- I am marking Firefox 44 as unaffected since e10s cannot be enabled for this version.

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 1 • 8 years ago

This sounds like a dupe of bug 1254560. Can you look at that bug and let me know if you agree? If so please dupe it over. I'm marking 45 as unaffected because APZ is not on in 45 at all unless manually enabled which is not the default configuration. Even in 46 beta it's not on right now (but hopefully will be in a week or two) so leaving it as 'affected' there.

	Vasilica Tamas, QA [:vtamas] Reporter
	Comment 2 • 8 years ago

This issue is not a dupe of Bug 1254560 because:
- setting layers.acceleration.disabled to true does not fix the glitches. 
- I am not able to reproduce Bug 1254560 on Windows 10 64-bit, where I reproduced my issue.

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 3 • 8 years ago

Ok, thanks for checking.

I tried to reproduce this but was unable to on my windows 10 machine. I had to create a new linkedin test account since I don't have a LinkedIn account of my own, and there's very little content that shows up when I log in. Presumably you get a lot more content and that makes it easier to reproduce.

Are you able to save a standalone copy of the page that reproduces the problem? If so please email it to me or attach it to the bug, and I can try to reproduce with that.

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 4 • 8 years ago

For comment 3, if you can get a standalone page that reproduces the problem it would be great.

	Petruta Horea [:phorea], Desktop QA
	Comment 5 • 8 years ago

Reproduced with https://ro.yahoo.com/?p=us page after scrolling down before entering Preview mode in order to add more content (17 pages in print preview mode).

Results for Windows 10 64-bit: 
 48.0a1 and 47.0a2 2016-03-21 - http://i.imgur.com/sJHQ205.png
 46.0b2 - http://i.imgur.com/6XEfj3w.png

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 6 • 8 years ago

(In reply to Petruta Rasa [QA] [:petruta] from comment #5)
> Results for Windows 10 64-bit: 
>  48.0a1 and 47.0a2 2016-03-21 - http://i.imgur.com/sJHQ205.png

I can reproduce this issue on 48 if I bump up the mousewheel acceleration. It's standard checkerboarding, as far as I can tell. Can you confirm that the glitches you're seeing on 47 and 48 are temporary and go away by themselves?

>  46.0b2 - http://i.imgur.com/6XEfj3w.png

I don't see this issue on Nightly, it might be a different root cause that's already been fixed on 47 and 48.

	Petruta Horea [:phorea], Desktop QA
	Comment 7 • 8 years ago

(In reply to Kartikaya Gupta (email:kats@mozilla.com) from comment #6)
> (In reply to Petruta Rasa [QA] [:petruta] from comment #5)
> > Results for Windows 10 64-bit: 
> >  48.0a1 and 47.0a2 2016-03-21 - http://i.imgur.com/sJHQ205.png
> 
> I can reproduce this issue on 48 if I bump up the mousewheel acceleration.
> It's standard checkerboarding, as far as I can tell. Can you confirm that
> the glitches you're seeing on 47 and 48 are temporary and go away by
> themselves?

I haven't modified the mousewheel acceleration in order to see those glitches.
They are only temporary and disappear fast.

Are the checkerboard issues tracked separately? 

> >  46.0b2 - http://i.imgur.com/6XEfj3w.png
> 
> I don't see this issue on Nightly, it might be a different root cause that's
> already been fixed on 47 and 48.

This was specific to 46 beta. Vasilica had a different problem for 46 too, as she mentioned in comment 0.

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 8 • 8 years ago

(In reply to Petruta Rasa [QA] [:petruta] from comment #7)
> I haven't modified the mousewheel acceleration in order to see those
> glitches.
> They are only temporary and disappear fast.

Thanks. The mousewheel acceleration is probably not needed for you if your mouse is fast enough. My mouse is old and slow so I needed the acceleration to reproduce the problem.

> Are the checkerboard issues tracked separately? 

We are not going to track checkerboard issues individually any more, but bug 1256677 is tracking checkerboard issues overall. See bug 1256677 comment 0 for details.

> > >  46.0b2 - http://i.imgur.com/6XEfj3w.png
> > 
> > I don't see this issue on Nightly, it might be a different root cause that's
> > already been fixed on 47 and 48.
> 
> This was specific to 46 beta. Vasilica had a different problem for 46 too,
> as she mentioned in comment 0.

Is it possible to get a fix window for this issue, since presumably it is fixed in newer versions?

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 9 • 8 years ago

Marking 47 and 48 unaffected since the problems seen on those branches are regular checkerboarding. 46 is the one with the weird-looking behaviour, as demonstrated in the screenshot at http://imgur.com/6XEfj3w

	Petruta Horea [:phorea], Desktop QA
	Comment 10 • 8 years ago

(In reply to Kartikaya Gupta (email:kats@mozilla.com) from comment #8)
> > > >  46.0b2 - http://i.imgur.com/6XEfj3w.png
> > > 
> > > I don't see this issue on Nightly, it might be a different root cause that's
> > > already been fixed on 47 and 48.
> > 
> > This was specific to 46 beta. Vasilica had a different problem for 46 too,
> > as she mentioned in comment 0.
> 
> Is it possible to get a fix window for this issue, since presumably it is
> fixed in newer versions?

https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=ae1d72ffd535a25dd29454d4503ccfef80cb78fb&tochange=ff6fc2d41f4517df623a1e0912cc5b5f5e2c7027

Mozregression suggests bug 1240073 from this pushlog fixed this on 47: "Found commit message:
Bug 1240073 - Use the transformed frame as the AGR for active transform so that FrameLayerBuilder knows that they can move independently. r=tnikkel"

	Kartikaya Gupta (email:kats@mozilla.staktrace.com)
	Comment 11 • 8 years ago

Seems to me from that pushlog that it could be either bug 1240073 or bug 1230774 that fixed it. It might also be one of the other ones in the range although those are less likely. However bug 1230774 is already on 46 so yeah, bug 1240073 is the most likely one.

I'm going to close this as fixed by bug 1240073. Given that APZ isn't going to the release population on 46, I'm not sure there's any value in uplifting that fix to 46 as it will only add risk.