On Wed, 13 Apr 2011, Dmitry Torokhov wrote: > On Wed, Apr 13, 2011 at 10:43:37AM -0400, Alan Stern wrote: > > On Tue, 12 Apr 2011, Dmitry Torokhov wrote: > > > > > Currently, when resetting a device, xHCI driver disables all but one > > > endpoints and frees their rings, but leaves alone any streams that > > > might have been allocated. Later, when users try to free allocated > > > streams, we oops in xhci_setup_no_streams_ep_input_ctx() because > > > ep->ring is NULL. > > > > > > Let's free not only rings but also stream data as well, so that > > > calling free_streams() on a device that was reset will be safe. > > > > What happens when the driver that allocated the streams initially wants > > to use them after the reset? Is this driver supposed to be responsible > > for re-allocating the streams in its post_reset method? > > Yes, I think so. It needs to re-initialize/set up the device anyway. > > FWIW the driver that I had issue with (the only one that uses streams at > the moment - UAS) did not want to use them after reset. It was just > trying to unbind from the device in otrderly fashion and was blowing up > spectacularly. > > > > > Wouldn't it make more sense to keep the streams around or > > That could be an option if we were keeping the original rings around as > well. My reasoning that if we blow away rings we should blow away > streams too so that we are consistent. > > > automatically re-allocate them? > > And if we fail? Or can't allocate as much as there was before? We'd have > to notify the driver somehow. Too messy IMO. Yes, that seems reasonable. Can you add a line to the kerneldoc for usb_alloc_streams() explaining that they will need to be reallocated following device resets? 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
