Hi, According to our results maximum throughput of our protocol is about 1.58 Mbps per device. Connecting more than one devices on the same or different buses(root hubs) doesnt change this maximum limit. 1.58Mbps makes almost 200bytes per frame, which makes 3 full bulk transactions (64 each at full speed) plus one partially filled. I wonder if UHCI controller pads the empty bytes of 4th transaction with dummy data or not. I am asking this question because 4 transactions make about 2Mbps. But when we try this throughput, although the device almost manages to process all submitted requests, we get a lot of outliers (which is what we called a urb transmission with latency more than 1ms). 200bytes per frame is ideal in terms of outliers, much lower than %1, and request processing rate(%100). So, do you have any ideas about this issue? thanks.. On Thu, Aug 13, 2009 at 5:27 PM, Alan Stern<stern@xxxxxxxxxxxxxxxxxxx> wrote: > On Thu, 13 Aug 2009, cihan öztürk wrote: > >> > It is possible for usbmon to lose events. You can tell if this happens >> > because the Bs file (where B is the bus number) contains a count of >> > lost events. >> > >> > Oddly, this fact doesn't seem to be mentioned explicitly in >> > Documentation/usb/usbmon.txt. >> > >> > Alan Stern >> > >> > >> I am not sure how to obtain Bs file. I tried >> #cat /sys/kernel/debug/usbmon/1u > logs1.txt >> in one shell to sniff 1st bus (which works perfectly) and >> # cat /sys/kernel/debug/usbmon/1s > events1.txt >> in another shell to capture events of 1st bus, but this command >> immediately returned, outputting a file with the text >> "nreaders 0 events 8626 text_lost" >> What am i doing wrong? > > You're not doing anything wrong. That's supposed to happen -- although > there is supposed to be a number following the "text_lost" string. > That number is a running counter of the number of events that were > dropped. > > Read the value text_lost value before starting your test and again > after ending the test. If the value hasn't changed then no events > were lost while the test was running. > > 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
