We have been trying to track down a 'hang' bug on our platform and have
possibly narrowed it down to the UHCI driver. We have stress tests running
simultaneously on 3 serial ports, 2 network ports, the IDE/compact flash
interface and the USB driver.
The only time the test fails is if we assign the USB host and interrupt
(effectively 'enabling' the USB test, test does not start on USB if not
running). By 'fail' I mean complete hang of board/OS, no ping, no mouse,
etc. Naturally, when running the kernel debugger, the timing changes such
that the stress tests never fail :(
Is this a known issue ? Any one else have this problem ? Given other
problems we have had with the USB driver (arbitrary reset timeouts and such
in the driver that have to change based on using a hub or not) and the fact
the stress tests running continuously w/out USB this is the current area of
suspicion.
CE.NET 4.2
All QFEs applied
Intel PIII w/chipset (UCHI)
Thanks for any heads up,
Jason