Description of problem: The wacom driver crashes X on logout. It also corrupts the X server making it impossible to VT switch. This prevents me from logging out or switching user. Version-Release number of selected component (if applicable): xorg-x11-server-Xorg-1.4.99.901-22.20080415.fc9.i386 linuxwacom-0.7.9.8-5.fc9.i386 xorg-x11-drv-i810-2.2.1-20.fc9.i386 kernel-2.6.25-1.fc9.i686 How reproducible: Always when using wacom and intel drivers together Steps to Reproduce: 1. Kill X. It may work the first time. 2. Kill X again and you are stuck with a corrupt screen and no input. (keyboard or mouse) This is also a problem when switching to a termnal. Actual results: On one tabletpc the system stop respond the everything. On the other it still accepts network connections. Expected results: Killing X should just start GDM again. Additional info: This is the x.org commit which causes the behavior: http://gitweb.freedesktop.org/?p=xorg/xserver.git;a=commitdiff;h=6d22a9615a0e6ab3d00b0bcb22ff001b6ece02ae Reverting that patch fixes the problem on both my tablet pcs. The tablet pcs use serial wacoms (builtin serial). This also happens if I attach a usb wacom and use that instead of the builtin wacom. This is the bug report which originated the above commit: http://bugs.freedesktop.org/show_bug.cgi?id=14418 I also put a backtrace of the wacom segfault in comment 8 of the x.org bug report. The problem doesn't exist on my laptops with nouveau drivers. I guess it could be because the intel driver is more advanced and it matters if isn't shut down in a proper fashion. I also tried compiling the hotplug branch of the linuxwacom driver (from git.debian.org). That doesn't work properly. It still causes corruption when killing X but it allows me to keep killing X. Most likely because the wacom driver segfaults before I kill X. I don't know how to make a backtrace of that so I'm stuck. There might be a more updated version of that branch soon (according to a developer from the linuxwacom mailling list).
nominating as F9Blocker.
Created attachment 303043 [details] Xorg.0.log from the upstream bug
This is the mentioned backtrace from the upstream bug: (gdb) bt full #0 xf86WcmDevProc (pWcm=0x9b0dad0, what=2) at ./xf86Wacom.c:682 local = (LocalDevicePtr) 0x9ba6568 priv = (WacomDevicePtr) 0x0 #1 0x0021cc8c in xf86WcmUninit (drv=0x99abd40, local=0x9afa350, flags=0) at ./wcmConfig.c:351 priv = (WacomDevicePtr) 0x9b1e968 #2 0x080d5292 in DeleteInputDeviceRequest (pDev=0x9b0dad0) at xf86Xinput.c:463 pInfo = (LocalDevicePtr) 0x9afa350 drv = (InputDriverPtr) 0x99abd40 idev = (IDevRec *) 0x998fdc8 #3 0x0807e6e3 in CloseDownDevices () at devices.c:621 dev = (DeviceIntPtr) 0x9b0dad0 next = (DeviceIntPtr) 0x9b0dc10 #4 0x0806b3f0 in main (argc=8, argv=0xbf96bd84, envp=0x0) at main.c:461 i = <value optimized out> error = 136273580 xauthfile = <value optimized out> alwaysCheckForInput = {0, 1}
Created attachment 303273 [details] Fix as described in the x.org bugzilla The upstream xorg bugzilla has comment which describes why this fixes the bug. This fix has been tested on both my tabletpcs. I have asked the linuxwacom developers to confirm if this is good enough for them.
Built for rawhide and F-9. Thanks!