On Mon, 12 Jun 2006, Greg KH wrote: > On Mon, Jun 12, 2006 at 03:16:31PM -0400, Alan Stern wrote: > > Greg: > > > > This patch (as712) fixes the usb-skeleton example driver so that it won't > > try to submit URBs after skel_disconnect() has returned. This could cause > > errors, if the driver was unbound and then a different driver was bound to > > the device. > > Huh? How could that happen? Ah, the struct usb_device would be now > owned by another driver... > > Ick. > > Yeah, I guess this patch is needed, oh well, maybe I'll think of some > other way to do this without a lock being needed... I just want to keep > it simple wherever possible...
I sympathize with the desire to keep things simple. In this case the extra mutex wasn't really needed; the regular usb_device lock could have been used instead. But that lock controls other things; it wouldn't be friendly to hog it in a multi-interface device, for example. And even then, you would still need to explicitly acquire and release the lock. I don't think there _is_ any way to fix this without some sort of synchronization mechanism. The race between read/write and disconnect makes it unavoidable. A mutex seemed like the simplest solution. Alan Stern _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
