Closed Bug 1060948 Opened 10 years ago Closed 10 years ago

Blank window when launching Firefox (turns black when resizing the window)

Categories

(Core :: Graphics, defect)

x86
macOS
defect
Not set
critical

Tracking

()

RESOLVED DUPLICATE of bug 1061209

People

(Reporter: stefanh, Unassigned)

Details

(Keywords: regression)

I get a blank window when launching Firefox on 10.6. Resizing the window turns it black. This is with an old Intel GMA 950 Mac mini on 10.6.8.

I get a warning in the console from gfx/gl/GLContextProviderCGL.mm, line 290 (”Couldn’t init gGlobalContext”) followed by a bunch of ”invalid pixel format” and ”invalid context” messages.

I could be wrong here, but my guess is that this is caused by some(?) of the patches in bug 996266. I don't know anything about that code, but I backed out to revision 201927 and that fixed the issue for me.
Keywords: regression
(In reply to Stefan [:stefanh] from comment #0)
> I get a blank window when launching Firefox on 10.6. Resizing the window
> turns it black. This is with an old Intel GMA 950 Mac mini on 10.6.8.
> 
> I get a warning in the console from gfx/gl/GLContextProviderCGL.mm, line 290
> (”Couldn’t init gGlobalContext”) followed by a bunch of ”invalid pixel
> format” and ”invalid context” messages.
> 
> I could be wrong here, but my guess is that this is caused by some(?) of the
> patches in bug 996266. I don't know anything about that code, but I backed
> out to revision 201927 and that fixed the issue for me.

These sequential revision numbers are different on m-c as compared to the integration branches. Can you post the hash part of the rev, and mark this bug as blocking the relevant bug that regressed this? Thanks!
Component: Canvas: WebGL → Graphics
Flags: needinfo?(stefanh)
Revision 201927:a2a10e38f42d is what I backed out to.

That revision is the revision just before 201928:fdd2bb040eb6 which is the first changset from bug 996266 (which this bug depends on). Note that I haven't isolated the relevant changeset, but my guess is that this regressed by bug 996266 (the console message is from GLContextProviderCGL.mm). I usually access the machine remotely and I can't build on that machine. Testing on the 10.6.8 machine involves building on another machine/packaging and then moving the .dmg to the 10.6.8 machine, but I'll see if I can isolate a specific changeset.
Flags: needinfo?(stefanh)
Turns out I was wrong about the cause of this. The last good nightly I've found was built from https://hg.mozilla.org/mozilla-central/rev/d697d649c765 which looks it was after the patches in bug 996266 landed.
No longer depends on: 996266
tinderbox builds:
https://hg.mozilla.org/mozilla-central/rev/d728b2f31c56 = OK
https://hg.mozilla.org/mozilla-central/rev/11e4f1678eab = Not OK

Which does indicates that this could be a dupe of bug 1061209...
(In reply to Stefan [:stefanh] from comment #4)
> Which does indicates that this could be a dupe of bug 1061209...

Yes it is - backing out the patch in bug 994541 made the problem go away.
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.