https://bugs.freedesktop.org/show_bug.cgi?id=66515
--- Comment #22 from Mauro Carvalho Chehab <[email protected]> --- (In reply to comment #21) > I'm a little confused about the symptoms of the problem: > > The original report and the attached gdb output show what looks like a > fairly standard X server crash. When this happens, you should get kicked > back to console or the display manager login screen. > > But the bug title and later comments instead make me think of something like > a GPU lockup / reset cycle, which sometimes ends with a completely hung > system. > > What is the connection between these two rather different kinds of symptoms? > Do both of them occur every time? Together, or one after another? ... When I opened the BZ, the original symptom were a Xorg crash. On a few cases the machine just hangs (something like 10% of the crashes). In order to track the root cause and see if this issue was not solved on a latter fix, I upgraded from kernel 3.9 to kernel 3.10.3 (and later to 3.10.4) and updated all packages at the machine. With the updated packages, the symptom changed: when the GPU stops to answer to mesa/Glamor, instead of causing Xorg to crash (and when the machine doesn't hang) it now reports: [15225.379837] radeon 0000:01:00.0: GPU lockup CP stall for more than 10000msec As shown on this attachment: https://bugs.freedesktop.org/attachment.cgi?id=83300 The workaround to avoid the buggy code was to disable glx-tls at Mesa. -- You are receiving this mail because: You are on the CC list for the bug.
_______________________________________________ xorg-driver-ati mailing list [email protected] http://lists.x.org/mailman/listinfo/xorg-driver-ati
