On Mon, 9 Aug 2010, Chris Beauchamp wrote: > On 6 August 2010 19:51, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > > On Fri, 6 Aug 2010, Chris Beauchamp wrote: > > > > > That's a little unclear. Under 2.6.27 or later, suppose you plug it > > in, unplug it, wait 10 seconds, then plug it in again. Does it work? > > What if you wait 60 seconds? 60 minutes? 3 days? A year? At what > > point does the device stop working? > > Thanks for the reply - Did some experimenting this morning: > > Under 10 seconds: Still doesn't work properly - tried this 2-3 times > > 2-3 minutes: works fine > > 30 minutes: works fine > > I can do some more experimenting to narrow down when it starts > working, if you think its relevant > > > > >> Doesn't seem to be dependant on whether its on UHCI (Intel) or OHCI (nVidia). > >> > >> I can do usbmon traces, but what the best way of putting them up? The > >> files are quite large... > > > > You can post the files on a web site or an FTP server. Or you can > > compress them and attach them to an email message. > > Three traces: > > Freshly power cycled server, freshly powered Edgeport, first plugin: > > http://www.chillibean.net/usbmon/firstplug.usbmon > > Now unplug (for completeness) > http://www.chillibean.net/usbmon/firstunplug.usbmon > > Then plugged back in, after 2-3 minutes > http://www.chillibean.net/usbmon/secondplug.usbmon > > > Corresponding kern.log extracts with debug=1 for io_ti & usbserial modules: > > http://www.chillibean.net/usbmon/firstplug.kern.log > http://www.chillibean.net/usbmon/firstunplug.kern.log > http://www.chillibean.net/usbmon/secondplug.kern.log > > > > > Besides, the files shouldn't be all that big if all you collect is the > > data sent when the device is plugged in (it helps if you have no other > > USB devices plugged into the same bus at the same time). > > Made sure there wasn't anything else plugged in at the same time, and > only grabbed traces from the relevant bus (3) > > They're only a MB in total, but I'm new here and not sure what people > think is reasonable to attach these days :) > > > It beats me why the error value is printed in hex. A quick calculation > > shows that this is equal to -71, which indicates a problem in the > > device. > > OK - I get the same behaviour with two of these, so either they're > both broken, or its something systematic with them I'm busy now with LinuxCon; I'll try to look at this next week. 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
