From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040510 Epiphany/1.2.4 Description of problem: On a Dell GX240, with a ATI Rage Pro 128 video card (r128 driver), the *whole* computer crashes at GDM logoff. Version-Release number of selected component (if applicable): kernel-2.6.5-1.358 & xorg-x11-6.7.0-2 How reproducible: Always Steps to Reproduce: 1. Login using GDM 2. Do what you want :) 3. Logoff Actual Results: Computer is dead crashed ! Expected Results: X server should restart normally :) Additional info: Solution 1: disable DRI with '#Load "dri"' in /etc/X11/xorg.conf. Solution 2: disable GDM and run X server using 'startx' instead.
I analyzed the problem lately and the computer doesn't actually crash. The system stays into some kind of koma without I/O available. The reason is a bug in shared interrupt handling. As long as the r128 driver shares an interrupt with any other device, the 'crash' occurs. If you manage to keep keyboard drivers alive Ctrl+Alt+F1, you will manage to see the following: 00:34:42 xdm(pam_unix)[1781]: session closed for user achim 00:34:44 kernel: irq 11: nobody cared! 00:34:44 kernel: Call Trace: 00:34:44 kernel: [<c0107423>] __report_bad_irq+0x23/0x80 00:34:44 kernel: [<c0107510>] note_interrupt+0x50/0x80 00:34:44 kernel: [<c0107762>] do_IRQ+0x122/0x140 00:34:44 kernel: [<c0105de8>] common_interrupt+0x18/0x20 00:34:44 kernel: [<c0118d30>] __do_softirq+0x30/0xa0 00:34:44 kernel: [<c0118dcd>] do_softirq+0x2d/0x40 00:34:44 kernel: [<c0107745>] do_IRQ+0x105/0x140 00:34:44 kernel: [<c0105de8>] common_interrupt+0x18/0x20 00:34:44 kernel: 00:34:44 kernel: handlers: 00:34:44 kernel: [<d08829a0>] (usb_hcd_irq+0x0/0x60 [usbcore]) 00:34:44 kernel: [<d0974a40>] (tulip_interrupt+0x0/0x720 [tulip]) 00:34:44 kernel: Disabling IRQ #11 00:34:44 kernel: agpgart: Found an AGP 1.0 compliant device at 0000:00:00.0. or something like that. If you manage to put r128 on it's own interrupt the problem disappears. Which is almost impossible though, since there's no way to force device drivers to use specific interrupts. Stephan
FWiW, I have the same computer, and I'm experiencing the same gdm issues. I turned on Debug for gdm, and here's what /var/log/messages says after I log off, but before my computer totally crashes: Jun 10 16:33:20 dhcp-10-22-101-108 gconfd (mlinford-2553): Exiting Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: slave_waitpid: done_waiting Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: Session: start_time: 1086910184 end_time: 1086910400 Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: Sending SESSPID == 0 for slave 2407 Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: Sending SESSPID 2407 0 Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_child_handler Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_child_handler: 2504 died Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_child_handler: 2504 returned 0 Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2235]: Handling message: 'SESSPID 2407 0' Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2235]: Got SESSPID == 0 Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: gdm_slave_session_stop: mlinford on :0 Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_usr2_handler: :0 got USR2 signal Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: gdm_slave_session_stop: Running post session script Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: gdm_auth_user_remove: Removing cookie from /home/mlinford/.Xauthority (0) Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: gdm_auth_purge: :0 Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: Running gdm_verify_cleanup and pamh != NULL Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2407]: Running pam_close_session Jun 10 16:33:20 dhcp-10-22-101-108 gdm(pam_unix)[2407]: session closed for user mlinford Jun 10 16:33:20 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_child_handler Jun 10 16:33:21 dhcp-10-22-101-108 gdm-binary[2407]: Running pam_setcred with PAM_DELETE_CRED Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_child_handler Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_child_handler: 2741 died Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_child_handler: 2741 returned 0 Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: gdm_slave_session_start: Session ended OK (now all finished) Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: Sending LOGGED_IN == 0 for slave 2407 Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: Sending LOGGED_IN 2407 0 Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: Handling message: 'LOGGED_IN 2407 0' Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: Got logged in == FALSE Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: Sending LOGIN == for slave 2407 Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: Sending LOGIN 2407 Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_usr2_handler: :0 got USR2 signal Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: Handling message: 'LOGIN 2407 ' Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: Got LOGIN == Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: gdm_slave_start: Reinitializing things Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_usr2_handler: :0 got USR2 signal Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: gdm_auth_secure_display: Setting up access for :0 Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: gdm_auth_secure_display: Setting up access Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: gdm_auth_secure_display: Setting up access for :0 - 1 entries Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: Sending COOKIE == <secret> for slave 2407 Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: Sending COOKIE 2407 a1... Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: Handling message: 'COOKIE 2407 a1...' Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: Got COOKIE == <secret> Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: gdm_server_reinit: Server for :0 is about to be reinitialized! Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2235]: (child 2407) gdm_slave_usr2_handler: :0 got USR2 signal Jun 10 16:33:21 dhcp-10-22-101-108 gdm[2407]: do_server_wait: Before mainloop waiting for server In the meantime, I've commented out "Load dri" in my xorg.conf, as suggested above, which solves my "crashing" problem.
*** This bug has been marked as a duplicate of 123424 ***
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.