* Alan Stern wrote: > On Fri, 8 Jul 2011, Thierry Reding wrote: > > > * Alan Stern wrote: > > > On Fri, 8 Jul 2011, Thierry Reding wrote: > > > > > > > Hi, > > > > > > > > I have a device that sometimes enters a strange state in which it keeps > > > > sending ISO transfers with apparently no content. What happens is that at > > > > some point the URB callback function keeps getting called, but summing up the > > > > actual_length field over all iso_frame_desc structures of the URB yields 0, > > > > which in turn causes the device to no longer respond to userspace. > > > > > > > > I'm wondering whether this might be an actual hardware bug. > > > > > > Or firmware bug. That seems quite likely. > > > > > > > Does anybody know > > > > whether such behaviour is to be expected, or what an appropriate response > > > > would be? > > > > > > Such behavior should not occur -- but in the real world it _does_ occur > > > often enough that you shouldn't be very surprised to encounter it. > > > > > > Update the firmware, if you can. Complain to the vendor. If nothing > > > else works, return the device for a refund. > > > > Okay, I'll try and contact the vendor then. If they're not responsive, is > > there anything else I could do? I was thinking something along the lines of > > marking the corresponding video buffer invalid, so that the error is in fact > > propagated instead of the driver just stopping to respond. I guess it > > wouldn't be very good to reset the device from within the driver, but perhaps > > some userspace application could react to the error code and initiate a reset > > instead. > > I don't know -- this is more the sort of thing you should ask the > linux-media people. > > It _is_ possible to reset a device from within the driver, if that's > what you decide to do. But the data stream would drop out during that > time, obviously. Okay, I'll have to see what works out best in the end. Thanks for you help! Cheers, Thierry
Attachment:
pgpAk36ul0rGJ.pgp
Description: PGP signature
