On Monday 30 May 2005 9:26 am, Alan Stern wrote: > On Sun, 29 May 2005, Patrick Boettcher wrote: > > > Is there method or rule how to calcuate the number of frames/urb needed? > > Do I have to calculate it at all? > > There is no hard-and-fast rule. However some host controller drivers are > limited as to how far in the future they can schedule an ISO transfer. > You should probably have not much more than 1/2 second worth of frames > queued at any time. You can divide that up into URBs almost any way you > like, subject to your latency requirements.
And there can be hardware limits too. 1/2 second can exceed them. This comment in ehci-sched gives more portable guidelines: /* * This scheduler plans almost as far into the future as it has actual * periodic schedule slots. (Affected by TUNE_FLS, which defaults to * "as small as possible" to be cache-friendlier.) That limits the size * transfers you can stream reliably; avoid more than 64 msec per urb. * Also avoid queue depths of less than ehci's worst irq latency (affected * by the per-urb URB_NO_INTERRUPT hint, the log2_irq_thresh module parameter, * and other factors); or more than about 230 msec total (for portability, * given EHCI_TUNE_FLS and the slop). Or, write a smarter scheduler! */ Frames-per-URB isn't the most useful measure; that'd be msec-per-URB, or in some cases KBytes-per-URB. Requiring more than a page of buffer per URB can be troublesome due to kernel memory allocation issues; you can't reliably allocate buffers that take more than one page. > > I always get the following errors: > > ehci_hcd 0000:00:1d.7: devpath 2.1 ep2in 3strikes > > > > What does the error mean exactly? I assume that I did something wrong... > > but what? > > The "ehci_hcd 0000:00:1d.7" part tells you which host controller driver > and which controller encountered the error. The "devpath 2.1 ep2in" part > tells you the error occurred when communicating with device usb-2.1, > endpoint 2, while performing an IN transfer. The "3strikes" part means > three errors occurred so the controller is giving up. > > From that message alone it's impossible to tell exactly what the error > was. Basically all it means is that the computer was expecting a response > from the USB device and didn't get one. It does however strongly suggest you weren't using isochronous endpoints, since "3strikes" is not an ISO error. :) - Dave ------------------------------------------------------- This SF.Net email is sponsored by Yahoo. Introducing Yahoo! Search Developer Network - Create apps using Yahoo! Search APIs Find out how you can build Yahoo! directly into your own Applications - visit http://developer.yahoo.net/?fr=offad-ysdn-ostg-q22005 _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
