On Thu, 2 Mar 2000, der Mouse wrote:

> I had such a problem on my i386 box.  I determined that the problem
> was that the printer would accept a byte and be ready for the next in
> less time than the CPU could get out of the interrupt routine.  Then
> when it allowed interrupts just prior to returning, it would be
> interrupted immediately -> instant stack overflow, because the buffer
> was too big.
> 
> Cutting down the buffer size stopped the crashes, but performance was
> pitiful.  Switching to the polled driver was the real cure; you might
> try that.  (This is done by setting the 0x80 bit in the device minor
> number; for example, my /dev/lpt0 is device 16,128.)

	Is this noted in a PR anywhere?

		David/absolute