As written in the Gentoo bug report... ---------------- Most full screen apps (i.e games) are crashing on exit in modular X. Neverball and Trackballs both exit with "*** glibc detected *** free(): invalid pointer" and fail to set the resolution back. Unreal Tournament and Return To Castle Wolfenstein appear to crash the X server. xscreensaver is okay though. I tried setting the resolution of Trackballs to be the same as my desktop but the crash still occurs. I also tried recompiling Neverball. I have an ATI Radeon 345M IGP. I've had random crashing problems with the DRI module for months now (only just found out roughly what it is) but I think this is a separate issue. ---------------- One of the Gentoo guys reproduced this on his 855GM. Disabling RENDER acceleration does nothing. These programs worked fine on 6.8.2.
Created attachment 3925 [details] xorg.conf
Created attachment 3926 [details] Xorg.0.log
We need backtraces for the crashes. As for the DRI instabilities, see if not changing the default AGP settings makes a difference.
I recompiled Neverball and Trackballs with -g3 and ran them through gdb but all I got was this, which doesn't tell you very much, does it? 0xffffe410 in __kernel_vsyscall Do I need to recompile Xorg or SDL or something else with debug symbols? Sorry, I can code but debugging is something I was never really taught properly. Thanks for the DRI tip, will try that soon.
Ah sorry, ignore all that, I was being stupid. Here's the backtrace from Neverball. #0 0xffffe410 in __kernel_vsyscall () #1 0xb7af0d1d in raise () from /lib/tls/libc.so.6 #2 0xb7af2333 in abort () from /lib/tls/libc.so.6 #3 0xb7b23b52 in __fsetlocking () from /lib/tls/libc.so.6 #4 0xb7b29497 in malloc_usable_size () from /lib/tls/libc.so.6 #5 0xb7b29e9b in free () from /lib/tls/libc.so.6 #6 0xb74b65c1 in _tnl_free_vertices () from /usr/lib/xorg/modules/dri/radeon_dri.so #7 0xb73fd9a1 in radeonDestroyContext () from /usr/lib/xorg/modules/dri/radeon_dri.so #8 0xb73fa2a6 in __driUtilUpdateDrawableInfo () from /usr/lib/xorg/modules/dri/radeon_dri.so #9 0xb7d7c1bb in __glXFreeContext () from //usr/lib/opengl/xorg-x11/lib/libGL.so.1 #10 0xb7cef833 in X11_GL_Shutdown (this=0x0) at SDL_x11gl.c:242 #11 0xb7ceda97 in X11_DestroyWindow (this=0x810dcf0, screen=0x8112900) at SDL_x11video.c:476 #12 0xb7cef43f in X11_VideoQuit (this=0x8108f00) at SDL_x11video.c:1233 #13 0xb7ce5f03 in SDL_VideoQuit () at SDL_video.c:1330 #14 0xb7cbda38 in SDL_QuitSubSystem (flags=32) at SDL.c:206 #15 0xb7cbdadf in SDL_Quit () at SDL.c:223 #16 0xb7af35a7 in exit () from /lib/tls/libc.so.6 #17 0x0806eb9c in MenuMode::doSelection (this=0x8500160) at menuMode.cc:127
I just recompiled Mesa with debug symbols as well. This might tell you a bit more. #0 0xffffe410 in __kernel_vsyscall () #1 0xb7c0ed1d in raise () from /lib/tls/libc.so.6 #2 0xb7c10333 in abort () from /lib/tls/libc.so.6 #3 0xb7c41b52 in __fsetlocking () from /lib/tls/libc.so.6 #4 0xb7c47497 in malloc_usable_size () from /lib/tls/libc.so.6 #5 0xb7c47e9b in free () from /lib/tls/libc.so.6 #6 0xb6e0a5c1 in _tnl_free_vertices (ctx=0x0) at t_vertex.c:497 #7 0xb6d519a1 in radeonDestroyContext (driContextPriv=0x8133301) at radeon_context.c:499 #8 0xb6d4e2a6 in driDestroyContext (dpy=0x8089070, scrn=0, contextPrivate=0x8133330) at dri_util.c:703 #9 0xb7d461bb in DestroyContext (dpy=0x8089070, gc=0x812e478) at glxcmds.c:472 #10 0xb7eaf833 in X11_GL_Shutdown (this=0x8084008) at SDL_x11gl.c:242 #11 0xb7eada97 in X11_DestroyWindow (this=0x8084008, screen=0x808df88) at SDL_x11video.c:476 #12 0xb7eaf43f in X11_VideoQuit (this=0x8084008) at SDL_x11video.c:1233 #13 0xb7ea5f03 in SDL_VideoQuit () at SDL_video.c:1330 #14 0xb7e7da38 in SDL_QuitSubSystem (flags=32) at SDL.c:206 #15 0xb7e7dadf in SDL_Quit () at SDL.c:223 #16 0x08065743 in main (argc=1, argv=0x27cc) at main.c:289
Changing product as this seems to be a Mesa issue. If you have backtraces for the X server crashes as well, please attach them to another bug.
Actually, this looks like a duplicate of bug 5185, which has been fixed in Mesa CVS. Please attach backtraces for the X server crashes (and change the product/component fields back) or resolve as duplicate.
Mesa CVS fixed everything, thanks a lot. But I don't know which bug this is a duplicate of because you gave the ID for this bug.
(In reply to comment #9) > Mesa CVS fixed everything, thanks a lot. But I don't know which bug this is a > duplicate of because you gave the ID for this bug. Whoops, thanks for verifying this! *** This bug has been marked as a duplicate of 5131 ***
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.