With Wireshark capture, I found that the problem was that the ACK from Endpoint responding to Host's Interrupt OUT got stuck in Host's buffer inside AMD's USB controller. This ACK was never subsequently properly processed and fed back to the host program in the PC. For programs like DCS, it does not allow for any non response from Endpoint, therefore USB hangs causing DCS itself to freeze until USB gets reset. VPforce configurator, on the other hand, doesn't seem to care whether it gets ACK or not so it continues to chug along. The fact that the ACK finally showed up after either we unplugged the USB cable or Walmis' hidtest program cleared itself after HIDException means that the ACK was properly sent by Endpoint (VPforce controller) but, for some reason, was never processed by the USB Host controller inside AMD's chipset.
