> Anyway, there's obscure test p->counter > current->counter + 3, which
> probably should be (3 * HZ) / 100. This could do some marginal
> difference in some cases.
nod, good spotting. (it's not obscure, but should be HZ-independent i
agree) decreasing +3 to something less just hides the real problem. (and
causes excessive reschedules between CPU-intensive processes)
> > but after a few seconds, your shell gathers enough 'interactiveness
> > priority' to not hang when you hit 'Enter'.
>
> That agrees with experiment. I just would like login to be fast even
> with CPU hogs running in background, automatically (renice them works).
note that this (fork-time priority sharing) mechanizm makes us 'survive'
fork bombs without problems. There should be no exploitable 'additional
timeslice source' in the system, other than 'time'.
-- mingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/