On Sat, Jan 27, 2007 at 09:26:12PM +0100, Joris van Rantwijk wrote: > > I am experimenting with my own firmware on a USB development board. > Currently, it does full-speed bulk transfers with the Linux generic > usb-serial driver. > > It looks like I lose data when the device sends data faster than > the application handles it. In other words, there is no flow control. > I have some questions about this; hopefully someone on this list knows > the answers.
The linux-usb-devel list would probably be best, but I'll try to answer them here ;) > * The USB specs seem to suggest that there is a built-in notion of > flow control: A device can always throttle the host by sending NAK in > response to OUT transactions. Likewise, the host can always throttle > a device by not doing IN transactions. > Is this a correct interpretation of the standard? Yes. > * The Linux generic usb-serial driver never throttles the device > in this sense. It always sets up a new read URB immediately after > receiving a packet, even when the tty layer has no room to store > more data, resulting in possible loss of data. Yup. > Would it make sense to postpone setting up a new read URB until the > tty layer has room available? It seems that some of the other drivers > (digi_accelport for example) do this in response to a throttle message > from the tty. Maybe, that might be a good idea to add. Care to whip up a patch to do this? thanks, greg k-h ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Linux-usb-users@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-users
