David Brownell wrote: > > > Kevin Brosius wrote: > > This is new in the last month or so, haven't tried this combination in a > > while... > > > > I have a usbaudio device which fails under xine at times. With latests > > kernel 2.6 bk tree, I notice that after xine locks up, and I kill -9 it, > > that powering down the usb audio device causes: > > See the comment immediately before that "badness" line: > > case ED_UNLINK: /* wait for hw to finish? */ > /* major IRQ delivery trouble loses INTR_SF too... */ > WARN_ON (limit-- == 0); > spin_unlock_irqrestore (&ohci->lock, flags); > set_current_state (TASK_UNINTERRUPTIBLE); > schedule_timeout (1); > goto rescan; > > So after waiting 1000 (the initial value of "limit") HZ it reports > this transition is way overdue ... suggesting IRQ delivery trouble. > > Is OHCI getting any interrupts at all during this period? Even > one IRQ should kick in finish_unlinks(), which should make that > ED transition from the UNLINK state (hardware may still see it) > to IDLE (it doesn't). What does sysfs say about ohci registers > and schedules when this is happening?
No, I'd bet it is not getting interrupts anymore. This is on my broken AMD OHCI hub, and one of the symptoms is loss of SF and completion interrupts. That explains it. Kevin > > > > > > Sep 28 11:41:09 sea kernel: ohci-hcd 0000:00:07.4: GetStatus > > roothub.portstatus > > [2] = 0x00030100 PESC CSC PPS > > Sep 28 11:41:09 sea kernel: hub 1-0:1.0: port 2, status 100, change 3, > > 12 Mb/s > > Sep 28 11:41:09 sea kernel: usb 1-2: USB disconnect, address 2 > > Sep 28 11:41:09 sea kernel: usb 1-2: usb_disable_device nuking all URBs > > Sep 28 11:41:09 sea kernel: ohci-hcd 0000:00:07.4: shutdown urb df1e6a80 > > pipe 00 > > 010200 ep2out-iso > > Sep 28 11:41:10 sea kernel: Badness in ohci_endpoint_disable at > > drivers/usb/host > > /ohci-hcd.c:342 > > Sep 28 11:41:10 sea kernel: Call Trace: > > Sep 28 11:41:10 sea kernel: [<c03123fd>] > > ohci_endpoint_disable+0x14d/0x250 > > Sep 28 11:41:10 sea kernel: [<c01267ed>] printk+0x1bd/0x2b0 > > Sep 28 11:41:10 sea kernel: [<c02fdb1f>] > > hcd_endpoint_disable+0x22f/0x3b0 > > Sep 28 11:41:10 sea kernel: [<c02ff4b8>] usb_disable_endpoint+0x58/0x80 > > Sep 28 11:41:10 sea kernel: [<c02ff63f>] usb_disable_device+0x10f/0x130 > > Sep 28 11:41:10 sea kernel: [<c02f7e34>] usb_disconnect+0xd4/0x160 > > Sep 28 11:41:10 sea kernel: [<c02faf11>] > > hub_port_connect_change+0x2a1/0x3b0 > > Sep 28 11:41:10 sea kernel: [<c02fb41b>] hub_events+0x3fb/0x560 > > Sep 28 11:41:10 sea kernel: [<c02fb5c5>] hub_thread+0x45/0x100 > > Sep 28 11:41:10 sea kernel: [<c010b736>] ret_from_fork+0x6/0x14 > > Sep 28 11:41:10 sea kernel: [<c0120410>] default_wake_function+0x0/0x30 > > Sep 28 11:41:10 sea kernel: [<c02fb580>] hub_thread+0x0/0x100 > > Sep 28 11:41:10 sea kernel: [<c010910d>] kernel_thread_helper+0x5/0x18 > > Sep 28 11:41:10 sea kernel: > > ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
