On Wed, 25 Mar 2009, Dave Mielke wrote: > [quoted lines by Alan Stern on 2009/03/25 at 10:35 -0400] > > >> Does doing a read into a zero-length buffer make any > >> sense? > > > >No. It's more likely either to time out or cause an overflow error > >when the device tries to send you more than 0 bytes of data. It's a > >little safer to do a read into a buffer whose length is the endpoint's > >maxpacket size; then you can't get an overflow. But you still can time > >out if the device doesn't have any data to send. > > What should happen if a read into a packet-sized buffer is done and the toggle > is wrong? Should it yield zero bytes and no erorr, a specific type of error, > or what? And, of coruse, will the toggle state then be correct? What should happen is this: The host will ask the device for data. If no data is available, the read will eventually time out and the toggle will still be wrong. Otherwise the device will send a data packet with the wrong toggle value. The host will accept the packet and ignore it. All the data in the packet will be lost. But now the toggle will be correct. The host will continue asking the device for data, since the data received earlier was ignored. Either the device will send more data or the read will time out. If the device does send data, that will be the result of the read. If the read times out, your program has no way of knowing whether or not the toggle has been corrected. 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
