pixman glyph cache patch

Not sure if you are aware of this problem or if its a nvidia-driver problem but latest xorg-server package in archlinux (xorg-server 1.13.0-4) contain a pixman glyph cache patch which causes rendering issues with nvidia blob.
arch bug report: https://bugs.archlinux.org/task/32612?project=1&order=lastedit&sort=desc&order2=dateopened&sort2=desc

Short story is arch nvidia users downgraded xorg-server to xorg-server 1.13.0-3 to avoid this pixman patch.
The interesting thing is this patch was Acked by nvidia (Aaron Plattner) so you may want to take a look into this.

Interesting; I seem to remember we ironed out all the issues with Soeren after Aaron tried his patch locally. I’ll look into it, thanks for the report.

I am using Arch, and I have only noticed this rendering bug in one program - the gui for the Dolphin emulator. I thought it was just a dolphin bug, but after looking at the screenshots in the arch bugtracker, I am sure it is the same bug.

Søren’s patch causes a huge slowdown when the glyph cache is disabled (or can’t be used for a particular operation), but it shouldn’t cause crashes. I’ll see if I can reproduce the problem, but the fact that it’s crashing inside libpixman itself implies that it’s not an NVIDIA driver bug.

Aaron, I think there may be two different issues here. Some experience the crash, but some like myself just get the occasional font rendering wierdness. If you look at the Arch bug report link from the OP, and look at some of the attached screenshots, you’ll see the font corruption.

I can’t reproduce this problem. Could someone with the problem please follow the sticky posts’s instructions on attaching a bug report log file?


I’m not having a crash, but rather font rendering issues as reported here:

I can confirm this issue and that reverting the xorg version workarounds it.

I still haven’t reproduced the crash, but I’ve reproduced the corruption problem and have a fix that I’ll get into a future driver release.

Glad to hear it! Any chance for that fix to be backported to the 304.xx driver?

Thanks, Aaron!