Re: O(1) Scheduler (tuning problem/live-lock)

From: Jim Houston (jim.houston@ccur.com)
Date: Tue Oct 01 2002 - 02:20:57 EST


Hi Andrea, Ingo,

Andrea I tried your patch and it does solve the live-lock
in the LTP waitpid06 test. The mouse movement gets a bit
jerky but atleast it doesn't lock up.

I guess the next question is how does it do on normal work loads?

I like the idea of making the child processes start with a smaller
sleep_avg value. Maybe it should just be a constant rather than a
fraction of the parents sleep_avg? Its really the child processes
inheriting the favorable sleep_avg that caused the problem with
waitpid06.

I liked the idea of giving interactive tasks special treatment.
Andrea please don't remove this. Always putting processes
(which have used up there time slice) into the rq->expired array
makes all processes round robin at the same priority. It makes
sense to do this to fail gracefully if the system is overloaded
but not all the time.

I hope this make sense. I'm falling asleep writing it:-)

Jim Houston - Concurrent Computer Corp.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Mon Oct 07 2002 - 22:00:25 EST