On Monday 24 January 2005 1:37 am, Vladimir Trukhin wrote: > >Hmm, does the following patch help at all? Presumably it compiles! > > > >It's kind of dirty; the real fix should probably be to not return > >the request from the queue until its ZLP gets ACKed. And it may > >well break with DMA (but nobody's using that lately). > > I'm afraid not :( Result is the same as before if I apply the patch as > it is.
Well that's too bad. I suspect the problem you're seeing is related to what Alexander Gordeyev was seeing: some confusion about the state of the IN fifo. You should probably change that code not to call done() for a request until its last packet is acked by the host; I suspect there's a window in there where the state's a bit confused if the queue is empty and the fifo isn't. The PXA hardware seemed to have curious expectations about what hardware does vs software; it wanted to enforce software policies in hardware, and have the software enforce policies more naturally handled in hardware. > Should I enable DMA in pxa2xx_udc? Not unless you want to debug that code and make it work reliably! - Dave ------------------------------------------------------- This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
