Description of problem: I am seeing the following issue: after I unmount and remove the USB key drive, the USB stack crashes shortly after that (usually coinciding with exiting and restarting X). The trace this generates follows: Mar 24 17:20:32 hagrid kernel: usb 1-1: USB disconnect, address 4 Mar 24 17:20:47 hagrid gconfd (icon-31893): Exiting Mar 24 17:20:47 hagrid gdm(pam_unix)[31812]: session closed for user icon Mar 24 17:20:49 hagrid kernel: irq 10: nobody cared! (screaming interrupt?) Mar 24 17:20:49 hagrid kernel: Call Trace: Mar 24 17:20:49 hagrid kernel: [<0210f1e9>] __report_bad_irq+0x2b/0x67 Mar 24 17:20:49 hagrid kernel: [<0210f281>] note_interrupt+0x43/0x66 Mar 24 17:20:49 hagrid kernel: [<0210f714>] do_IRQ+0x248/0x303 Mar 24 17:20:49 hagrid kernel: [<0212b3bc>] __do_softirq+0x2c/0x73 Mar 24 17:20:49 hagrid kernel: [<021103d0>] do_softirq+0x46/0x4d Mar 24 17:20:49 hagrid kernel: ======================= Mar 24 17:20:49 hagrid kernel: [<0210f7c3>] do_IRQ+0x2f7/0x303 Mar 24 17:20:49 hagrid kernel: Mar 24 17:20:49 hagrid kernel: handlers: Mar 24 17:20:49 hagrid kernel: [<32850017>] (usb_hcd_irq+0x0/0x4b [usbcore]) Mar 24 17:20:49 hagrid kernel: [<32850017>] (usb_hcd_irq+0x0/0x4b [usbcore]) Mar 24 17:20:49 hagrid kernel: Disabling IRQ #10 Nothing USB works after this (including my mouse), and X hangs. I have to reboot in order to be able to do something. Version-Release number of selected component (if applicable): kernel-0:2.6.3-2.1.253.2.1.i686 How reproducible: Always. Steps to Reproduce: 1. Unmount USB key drive 2. Remove USB key drive 3. Log out 4. Mouse no longer works, and nothing US either. 5. X is unkillable (-9 included) 6. Reboot.
Hmm... Correction. It occurred today again, but this time behavior was different. I unmounted the key-drive, then exited X, then disconnected the drive. As you can see from the following trace, the problem occurred right after gdm closed my session. Removal doesn't seem to matter. I'll do some more testing on Monday. Mar 26 17:03:21 hagrid gdm(pam_unix)[27958]: session closed for user icon Mar 26 17:03:24 hagrid kernel: irq 10: nobody cared! (screaming interrupt?) Mar 26 17:03:24 hagrid kernel: Call Trace: Mar 26 17:03:24 hagrid kernel: [<0210f1e9>] __report_bad_irq+0x2b/0x67 Mar 26 17:03:24 hagrid kernel: [<0210f281>] note_interrupt+0x43/0x66 Mar 26 17:03:24 hagrid kernel: [<0210f714>] do_IRQ+0x248/0x303 Mar 26 17:03:24 hagrid kernel: [<0212b3bc>] __do_softirq+0x2c/0x73 Mar 26 17:03:24 hagrid kernel: [<021103d0>] do_softirq+0x46/0x4d Mar 26 17:03:24 hagrid kernel: ======================= Mar 26 17:03:24 hagrid kernel: [<0210f7c3>] do_IRQ+0x2f7/0x303 Mar 26 17:03:24 hagrid kernel: Mar 26 17:03:24 hagrid kernel: handlers: Mar 26 17:03:24 hagrid kernel: [<32850017>] (usb_hcd_irq+0x0/0x4b [usbcore]) Mar 26 17:03:24 hagrid kernel: [<32850017>] (usb_hcd_irq+0x0/0x4b [usbcore]) Mar 26 17:03:24 hagrid kernel: Disabling IRQ #10 Mar 26 17:03:24 hagrid kernel: agpgart: Found an AGP 2.0 compliant device at 000Mar 26 17:03:24 hagrid kernel: agpgart: Putting AGP V2 device at 0000:00:00.0 inMar 26 17:03:24 hagrid kernel: agpgart: Putting AGP V2 device at 0000:01:00.0 inMar 26 17:03:35 hagrid kernel: usb 1-1: USB disconnect, address 4
Hmm... Further adjusting this, today this happened even without using the USB key drive. It seems that having a USB mouse is enough. Logging out from my gnome session seems to be sufficient to trigger the trace. It also occurs with 2.6.4-1.298.
I am seeing the same problem on a Sony Vaio. It occurs with kernel-2.6.3-2.1.253.2.1 and other 2.6.3-2.X and at least some 2.6.4-X versions -- I haven't tried the very newest. kernel-2.6.3-1.X seems fine. Booting seems fine and gdm runs. Login is fine, but any exiting of X11 causes a hard lock. E.g. any of the following will trigger it: logout, switch to console (ctl-alt-F2), ctl-alt-backspace, telinit 3, /usr/bin/poweroff. I am using the x11-org and gdm from the development tree.
I should have said, I am not sure this is related to USB, but more likely related to X.
I've not seen this happen in newer kernels, so I'm a happy camper. This bug can be closed, unless John is still seeing this.
This is a symptom caused by several root causes, essentially whenever a driver fails to consume an interrupt, so John needs to file his own. Closing...