- shouldn't part of what is done in the interrupt handler be done later in a
  tasklet?
- Sync warning: prolly no problem currently, but linked to our architecture +
  context (hardirq vs. softirq) (running etc.)
- Review again the FIFO rw code (WTF comments)
- (Low priority) check support for other XHFCs
- The init structure is not very clear (xhfc_init_and_configure)