> Note that the example was a delay of a "normal" program by the console
> interrupt. Since presumably the console interrupt is "slow" and the RS232
> interrupt is "fast", there'd be no data loss.
Hmm. Unless the kernel changes this (I didn't look, but it isn't easy),
keyboard interrupts (which these should be) have a _higher_ priority than
everything else except for timer interrupts. Doing something slow in there
is a _very_ bad idea.
MfG Kai