Tim Waugh (tim@cyberelk.demon.co.uk)
Fri, 7 May 1999 19:15:47 +0100 (GMT)
On Fri, 7 May 1999, Thomas Sailer wrote:
> Basically because transactions are queued to the HC and then
> one sleeps until they complete. Basically the same is true
> for ECP ISA DMA, while I think there is some hackish special
> case in the tmw3 patch, it's quite ugly if you cannot sleep.
There's certainly special case code, yes. I haven't gotten around to
documenting parport_set_timeout yet, but a timeout of zero _is_ special --
it means that you should do all that you can without actually sleeping.
The worst case is in parport_wait_peripheral where we could end up
busy-waiting for 35ms.
It's needed for things like parallel printer console. I don't know if you
followed the 'non-blocking' phase of the tmw tree before it got ripped
out, or looked at Roger Schreiter's implementation, but the alternatives
start to get messy.
Tim.
*/
-- To unsubscribe, send mail to: linux-parport-request@torque.net --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2.0b3 on Fri 07 May 1999 - 14:24:27 EDT