This message is from the T13 list server.
Curtis S:
Some devices clear BSY, but leave DRQ set to allow them to do
whatever at the end of a data transfer. When the device has actually
completed the command internally, then they clear DRQ. ATAPI devices are
the biggest offenders here.
Did you say that exactly as you meant it?
Me, I had only seen ATAPI devices set BSY after cycling all the "word"s of an op xA1 "IDENTIFY" block. BSY, not DRQ. You mean to say you have also seen DRQ left set with BSY clear for significant time past the last fetch or store of a data word?
That device abuse would defeat any polling host disconnected from INTRQ: without INTRQ, the host couldn't reliably distinguish an old DRQ from a new DRQ. It would have to delay arbitrarily and poll again to confirm a seemingly leftover DRQ was really a new DRQ. Ick.
I've never myself tried to make an x86 host work without INTRQ - only embedded hosts, to save a pin.
From: Hale ... This would be a complete violation of the PATA PACKET command protocol.
Yep.
Curiously yours, Pat LaVarre
