On Wed, 24 Aug 2011, Al Ivanov wrote: > Hello Alan, > > We are using sierra cellular modules, which use sierra.c module, and are > connected to OHCI USB controller. In the course of testing we have > noticed that when module gets unplugged while process is writing to it > usb_kill_urb() called from usb_hcd_flush_endpoint() hangs indefinitely > and the user-space process goes to uninterruptible sleep. In case device > gets plugged back in, after the failure, then kernel completely fails to > enumerate it. Once this condition happens the only way to recover from > it is to power cycle all the equipment. While doing research on the > possible causes and fixes for this problem I came quite a few posts > where people have been experiencing it with different USB devices. Few > of the examples, which also shows the history of system calls are > https://bugzilla.novell.com/show_bug.cgi?id=370872#c8, > http://ubuntuforums.org/showthread.php?t=1474287 (I have also posted > message at the bottom of this email for convenience). This problem has > been reproduced with 2.6.27.59 and 2.6.32.45 kernels. Would you happen > to know if this known issue has been fixed in any mainstream release of > the Linux kernel and what the patch is. Thank you very much for your > time and help. Bug reports like this should always be sent to the linux-usb mailing list. No, I don't know of any reports like this or whether it has been fixed. If you want to pursue the matter further, you should use a more recent kernel, such as 3.0.2 or 3.1-rc3, and you should build the kernel with CONFIG_USB_DEBUG enabled. After the problem occurs, get a copy of the dmesg log and post it. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
