Agreed. I re-wrote the Rx PIO routine to do a proper interrupt at EOF, and added a DMA driven ring buffer, which eliminated a lot of the hackiness...