> > How about verifying that this problem is still present in the most > > recent kernel versions, and then sending a patch against those versions? > > Sure. I verified by eye that the problem still existed (by checking > that in usb-ohci.c, in dl_del_urb, there's a code path that doesn't > call urb->complete).
Which wouldn't be a bug if sohci_unlink_urb() or something else did it, as implied by the comment in dl_del_urb() ... :) Did you try the more obvious patch to usb-ohci ? Attached, but not tested. I'm puzzled why you'd want to patch the audio driver when you feel the root cause of the bug is elsewhere. Also, did you establish whether this shows up in any other host controller driver? A quick look at "ohci-hcd" suggests that it should not have this particular bug, for example, since it's got more consistent URB handling in return-to-driver code paths. I didn't look at the UHCI drivers. - Dave
ohci-0228.patch
Description: Binary data
