Created attachment 125570 [details] CPU info Using Debian testing with kernel 4.6.4 and bzflag 2.4.6. $ uname -a Linux hp6200 4.6.0-1-amd64 #1 SMP Debian 4.6.4-1 (2016-07-18) x86_64 GNU/Linux CPU is Pentium G630 (cpuinfo attached) Hang occurs pretty much whenever I play bzflag, both with 32-bit and 64-bit kernels, though I haven't attempted with a 64-bit userspace under Debian (using 32-bit userspace only at this point.)
Created attachment 125571 [details] dmesg output
Created attachment 125572 [details] DRM error dump
What version of Mesa are you using?
Debian mesa package information, from squeeze (testing): $ dpkg -l \*-mesa\*| grep ^ii ii libgl1-mesa-dri:i386 11.2.2-1 i386 free implementation of the OpenGL API -- DRI modules ii libgl1-mesa-glx:i386 11.2.2-1 i386 free implementation of the OpenGL API -- GLX runtime ii libglapi-mesa:i386 11.2.2-1 i386 free implementation of the GL API -- shared library ii libgles1-mesa:i386 11.2.2-1 i386 free implementation of the OpenGL|ES 1.x API -- runtime ii libgles2-mesa:i386 11.2.2-1 i386 free implementation of the OpenGL|ES 2.x API -- runtime ii libglu1-mesa:i386 9.0.0-2.1 i386 Mesa OpenGL utility library (GLU) ii libwayland-egl1-mesa:i386 11.2.2-1 i386 implementation of the Wayland EGL platform -- runtime So I guess it's 11.2.2 plus whatever debian adds in their packaging. I don't have that info to hand, but I can download the source package to check if you like?
Please try a new version of Mesa (12.0.x or 13). Reopen if you can reproduce with a new version.
(In reply to Matt Turner from comment #5) > Please try a new version of Mesa (12.0.x or 13). Reopen if you can reproduce > with a new version. Tried Mesa 12.0.3 using Ubuntu 16.10, I get what looks like the same issue but with a slightly different error dump header: (Original dump header): GPU HANG: ecode 6:0:0x85fffffc, in bzflag [2693], reason: Ring hung, action: reset (12.0.3 dump header): GPU HANG: ecode 6:0:0x85fffffc, in bzflag [3867], reason: Hang on render ring, action: reset So the error text is slightly different, but it looks at first glance to be the same issue (same ecode for example.) I've attached the 12.0.3 error output.
Created attachment 127810 [details] Mesa 12.0.3 error dump for presumably same issue Error dump from Mesa 12.0.3 on Ubuntu 16.10 amd64. Feels like the same bug (bzflag hangs until GPU reset) and ecode is the same as 11.2 error dump.
PS. I'm happy to try this on a Fedora based installation to rule out any Debian based issue (original bug was Debian, 12.0.3 update was Ubuntu), if that helps?
I just joined a game of bzflag on kernel 4.9.5 and Mesa 13.0.5 and I don't get a hang. Is there something in particular you're doing when the hang occurs? About how long does it take to trigger? I'm on a 2540, which according to wikipedia is also a GT1.
(In reply to Matt Turner from comment #9) > I just joined a game of bzflag on kernel 4.9.5 and Mesa 13.0.5 and I don't > get a hang. Is there something in particular you're doing when the hang > occurs? About how long does it take to trigger? > > I'm on a 2540, which according to wikipedia is also a GT1. With Ubuntu 1610, using the Mesa 13.0.4 from: https://launchpad.net/~ubuntu-x-swat/+archive/ubuntu/updates I appeared to have no problems. I did some initial testing, and the graphics stayed smooth. I'd have reproduced the problem with the older Mesa versions within a minute, and I played at least 5 minutes with no issue. I don't recall off hand what kernel version was in use, it's whatever is latest with 16.10.
Cool. Since neither of us can reproduce it on newer Mesa, I'll mark as WORKSFORME.
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.