Hi guys,

I've done some more debugging -- first, here's an extract from the trace 
log:

   [enumeration trimmed]
   > SET_CONFIGURATION(1)
   > BULK_WRITE(09 11)
   > BULK_WRITE(0B 80 64)
   < BULK_READ(64 bytes) = 0xFE
   > BULK_WRITE(09 00)
(application ends here)
   > SET_CONFIGURATION(1)
   > BULK_WRITE(09 11)
   > BULK_WRITE(0B 80 64)
   < BULK_READ(64 bytes) = [timed out]

Now, on Windows, that second SET_CONFIGURATION request just doesn't 
happen. On Linux, libusb *always* sends a SET_CONFIGURATION req when you 
call usb_set_configuration(), even if the device is already set to the 
specified configuration.

This isn't normally an issue, and AIUI there's nothing in the USB spec 
that says this is incorrect. What actually happens on the Microchip 
stack is that the endpoints end up disabled -- the OUT endpoint will 
accept data from the host (and ACK it), but the IN endpoint NAKs everything.

For extra bonus points, you can't even send a SET_CONFIGURATION(0) 
followed by a SET_CONFIGURATION(n) -- that is, put the device into the 
Addressed state and wake it up again. That locks the stack up as well.


Trapping the code with the ICD2 after the problem occurs leads me to 
suspect that the stack isn't re-enabling the endpoints correctly. For 
instance, this expression in ProcessIO() evaluates to TRUE when data is 
available, but after the second SET_CONFIGURATION, it evaluates to FALSE 
and ProcessIO() assumes there's no data available:
     //Check if the endpoint has received any data from the host.
     if(!USBHandleBusy(USBGenericOutHandle)) {
       // ... USB OUT packet handling code goes here ...
     }

Has anyone seen anything even remotely like this going on with the 
Microchip stack? I'm probably going to pull the thing apart some time 
next week and debug it in greater detail, but for now I'd love to know 
if there are any known issues with Framework 2.5b that aren't in the 
README...

Thanks,
-- 
Phil.
piclist@philpem.me.uk
http://www.philpem.me.uk/
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist