On Fri, Jun 09, 2006 at 08:01:44PM +0200, oliver munz @ s p e a g wrote: > The USB interrupt-mask-register-bits are not reliable to signal if an > transfer is running. This fix uses a variable for this purpose. > Furterhmore it fixes some the setting of the IMR in the bus_reset > function...
Hi Oliver I at long last got around to testing this patch. Unfortunately, it breaks the driver when tested with the eCos USB test program. The changes to usbs_at91_control_setup() stops the test program during its initial configuration phase where the host uses control transfers to talk to the target test client. I removed that part of the change and then the tests could exchange control messages. However bulk IN and OUT on the endpoints did not work. For IN transfers it does not receive even the first message. For OUT transfers it sends a few, but then the host reports the contents of the messages are wrong. I doubt i will have time to debug this further before i go on Holiday :-( Andrew
