The driver tests show that the reading always times out. I would suspect
the hardware, but under dos it works perfectly, every time.
I do register the io area (0x201) and I use do_gettimeofday() to time
my loop. I do not disable interrupts, which should only slightly distort
results rather than permanently disable any reading.
It seems as if the port is not being read again after the first time the
reading loop is entered. I use 'inb(io)' to read, and I hope gcc knows
not to optimize this out of the loop - true?
-- Eyal Lebedinsky (eyal@eyal.emu.id.au)