On Friday 18 June 2004 08:17, Martin Husemann wrote:
> On Thu, Jun 17, 2004 at 05:02:15PM -0400, Greg A. Woods wrote:
> > Which seems to show that there's a design flaw or bug of some sort in
> > the lpt driver.
>
> More in the hardware: it does not have (IIRC) an interrupt status register
> that could indicate "yes, I requested an interrupt".
>
> The lpt driver will not claim an interrupt if the hardware does not
> indicate the printer being ready or having an error condition. This could
> be considered bogus - maybe we should just declare "lpt can not share
> interrupts" and have the driver always return 1 from the interrupt handler.

That would be equally bogus.  There are plenty of existing systems that do, in 
fact, share this interrupt.

Occasional "stray interrupts" are a fact of life with edge-triggered 
interrupts.  With most hardware, there are race conditions that are 
*impossible* to avoid.