907746 - Youtube/Flash/Java appears to only show the top left 1/4 of the content in HiDPI mode

Status: RESOLVED DUPLICATE of bug 907792

(Core Graveyard :: Plug-ins, defect)

Description

[David Bolter [:davidb] (NeedInfo me for attention)]

Graphics
Device ID	0x 166
GPU Accelerated Windows	2/2 OpenGL (OMTC)
Vendor ID	0x8086
WebGL Renderer	NVIDIA Corporation -- NVIDIA GeForce GT 650M OpenGL Engine
windowLayerManagerRemote	true
AzureCanvasBackend	quartz
AzureContentBackend	quartz
AzureFallbackCanvasBackend	none
AzureSkiaAccelerated	0

Macbook pro retina display.

Comment 1

[David Bolter [:davidb] (NeedInfo me for attention)]

Name 	Firefox
Version 	26.0a1
User Agent 	Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:26.0) Gecko/20100101 Firefox/26.0

Comment 2

[David Bolter [:davidb] (NeedInfo me for attention)]

This likely regressed very recently. I first noticed it with today's nightly.

Comment 3

[Steven Michaud [:smichaud] (Retired)]

I assume YouTube is running Flash :-)

Possibly a dup of bug 903296.

Please post a link to the YouTube video you tested with.

Comment 4

[Georg Fritzsche [:gfritzsche]]

Are you also seeing this with other Flash and Java content?

Comment 5

[David Bolter [:davidb] (NeedInfo me for attention)]

Yeah I see it on all Flash and Java.

Comment 6

[Georg Fritzsche [:gfritzsche]]

So, bug 903296 is also on Retina and Flash & Java, but isn't a recently introduced regression.
Maybe we are seeing some intermittent issues?

Comment 7

[Steven Michaud [:smichaud] (Retired)]

David, does this bug happen with a clean profile?

If it doesn't, look at about:support in your old profile for clues.

If it does, can you remember any OS-level changes you made recently, or any new software you installed?

What version of Flash do you have?

Comment 8

[David Bolter [:davidb] (NeedInfo me for attention)]

(In reply to Steven Michaud from comment #7)
> David, does this bug happen with a clean profile?

Yes - just checked.

> If it does, can you remember any OS-level changes you made recently, or any
> new software you installed?

Hmm... Nothing springs to mind. Note as for nightly regression window I'm not sure how recently I updated but I definitely updated this morning and the bug started happening.

I don't recall any other updates recently. Likely a red herring but yesterday I was running a few browsers I don't normally run (chrome, safari), just to test some interoperability during a w3c call.

I've been putting off an OSX update for printer and airport stuff.

> 
> What version of Flash do you have?

Version: 11.8.800.94

Comment 9

[Steven Michaud [:smichaud] (Retired)]

I *can* reproduce this bug in today's mozilla-central nightly, but only on a Retina MBP, and then only with HiDPI support on.

I couldn't reproduce bug 903296 a while ago.

I still think this bug and bug 903296 are most likely the same bug.  But something seems to have changed very recently on the trunk to make the situation worse.

I'll look for the regression range.

Comment 10

[Steven Michaud [:smichaud] (Retired)]

Here's the URL I used to reproduce this bug:
http://mirrors.creativecommons.org/getcreative/

Comment 11

[Steven Michaud [:smichaud] (Retired)]

For me these start with today's mozill-central nightly.  Which implies this regression range:

http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=bb025b6949e8&tochange=d136c8999d96

Comment 12

[Steven Michaud [:smichaud] (Retired)]

I can't begin to guess which patch in that range caused/triggered my ability to reproduce this bug.  I guess I'll have to hg bisect.

Comment 13

[Benoit Girard (:BenWa)]

You can use the tinderobx ftp mozilla-inbound builds to avoid building locally. It's annoying because you can't sort by landing order and have to check the .txt but it's faster.

Comment 14

[Steven Michaud [:smichaud] (Retired)]

I prefer to use hg bisect.  Yes it can take a long time.  But since it doesn't require much thought, I can do it "in the background", in between other tasks.  I can even leave it overnight and come back again the next day (as I did this time).

What hg bisect found puzzles me.  It tells me that the following patch triggered this bug.  (By the way I've confirmed this result "by hand".)

http://hg.mozilla.org/mozilla-central/rev/252dbb41d609

 Bug 900133 - Enable new textures on OSX. r=nical
author	Nicolas Silva <[e-mail]>
	Tue Jul 30 19:37:13 2013 +0200 (at Tue Jul 30 19:37:13 2013 +0200)

Comment 15

[Steven Michaud [:smichaud] (Retired)]

Since I'm the one who wrote the code that allows plugins to work in HiDPI mode, I'm probably best placed to figure this bug out.  But even for me it'll take a lot of debug logging and head scratching.

Do you have any ideas, Benoit?

Comment 16

[Cameron McCormack (:heycam)]

Seconding that this problem has returned just with the Nightly I updated today.  (My last update would have been one or two days ago.)

Comment 17

[Benoit Girard (:BenWa)]

(In reply to Steven Michaud from comment #15)
> Do you have any ideas, Benoit?

This was just discussed in a meeting I was in. Looks like this is a dupe of bug 907792. We're hoping to fix it ASAP to avoid turning off OMTC mac.