Tim Waugh (tim@cyberelk.demon.co.uk)
Wed, 1 Apr 1998 19:15:21 +0100 (BST)
On Wed, 1 Apr 1998, Andrea Arcangeli wrote:
> If the parport lowlevel driver will autodetect the right irq I don' t see
> how can it cause an harm.
People who had their Ethernet cards set up to IRQ 5 in 2.0.x might find
that suddenly it doesn't work in newer kernels. That's because parport
has (correctly) autodetected 5 as belonging to the parallel port, and
grabbed it for some other device driver.
> I don' t see the problem how lp specific but instead as parport_lowlevel
> specific. If lp will cause an harm every other pardevice that will use
> interrupt on the same parport will do the same.
That's true, but lp is currently the only device driver (that I know of)
that _uses_ interrupts if they have been autodetected by parport.
A better solution then might be to have to explicitly ask parport
to autoprobe for IRQs if you want interrupt-driven operation.
I guess the problem is that some device drivers _need_ interrupts, while
others find them helpful but not essential, and there's currently no way
to sort them all out properly.
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 Wed 30 Dec 1998 - 10:17:34 EST