Created attachment 81233 [details] screenshot of firefox I've just tried 2.21.10 on my SandyBridge-Notebook and there seem to be a few things off: - Many images in firefox are simply rendered black - Scrolling in some documents results in white area with AcroRead (that proprietary thing - as evince devs really tried hard to make evince as ugly as possible)
Can you please attach your Xorg.log so that I can narrow down the hardware and software likely required to reproduce? Firefox is still behaving on every machine I use, and I'm currently building an old chroot to run acroread.
I'm not seeing any failure in acroread on the first machine I installed it on either (a SNB).
Any chance you can also check the reproduction steps for anything peculiar I need to recreate?
Created attachment 81255 [details] Xorg log I can consistently reproduce this on Arch Linux x86_64 (firefox 21, linux 3.9.7, xorg 1.14.1, xf86-video intel 2.21.10) with SNA acceleration. For now I've just gone back to UXA acceleration.
By the way, here's the bug in our bug tracker, in case some useful information is posted there: https://bugs.archlinux.org/task/35896
Created attachment 81257 [details] another xorg.log
I see this with an official build of firefox-21 on fedora-19 (+ updates testing) and kernel-3.10.0.rc6.
Forgot to mention, I use XFCE. This bug is reproduceable with and without compositor. The acrobat-reader bug happens with acroread-9.5.3 and the attached document when scrolling too fast: http://youtu.be/OHMaEihUN1g
Created attachment 81258 [details] sample pdf document
Time to update my fedora laptop to f19 and see what breaks... Daniel, can you ask on the arch bug whether there are any machines affected that are not SNB/IVB?
Now running f19 (firefox 21, acroread 9.5.5, kernel 3.9.6, -intel.git) on a SNB i5-2520qm and not yet encountering any issues under gnome-shell or xfce4. Meh. Suggestions?
I can reproduce it on Arch Linux with kernel 3.9.7 and both stock intel 2.21.10 or intel-git with SNA acceleration. When reverting to UXA everything is properly rendered. The hardware used is an i5 2500K. It happens here in Mozilla Firefox and Mozilla Thunderbird while doesn't happen in Epiphany browser or any other application that I used today.
(In reply to comment #12) > I can reproduce it on Arch Linux with kernel 3.9.7 and both stock intel > 2.21.10 or intel-git with SNA acceleration. When reverting to UXA everything > is properly rendered. The hardware used is an i5 2500K. > > It happens here in Mozilla Firefox and Mozilla Thunderbird while doesn't > happen in Epiphany browser or any other application that I used today. You can always try to bisect to see what commit introduce the regression
Is anybody seeing this on a 32-bit build?
Yup. Booted a live f19 64-bit image, and the images disappeared. Sadly, so did the session, so time to try again.
The problem is gone with Firefox 23 (Aurora), but I think it's likely to still be a bug in the driver.
I'll try to bisect it tomorrow...
Ok, was able to bisect it sonner: 53c113c3cc2f8527debc185f0819139ca8637637 is the first bad commit commit 53c113c3cc2f8527debc185f0819139ca8637637 Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Fri Jun 21 19:27:24 2013 +0100 sna: Allow PutImage to write inplace using manual tiling Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> :040000 040000 a3ba69294e46d078d94dd5f6ad976ec6eaccbfb2 87e4890f2c7e21bf8fbf9c7237a91423b8bd1e76 M src
just verfied, without this commit both issues (FF and acrobat reader) are gone :)
commit 77ddadf0ec6b867a0638660bf06c82ad31577795 Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Sun Jun 23 22:01:12 2013 +0100 sna: Clear 'clear' hint upon uploading into tiled Fixes regression from commit 53c113c3cc2f8527debc185f0819139ca8637637 [2.21.10] Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Fri Jun 21 19:27:24 2013 +0100 sna: Allow PutImage to write inplace using manual tiling Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66059 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.