Ingo Molnar <mingo@chiara.csoma.elte.hu> writes:
> Here is the changelog of http://www.redhat.com/~mingo/lowprio-2.2.14-B1.
Speaking of "guaranteed low latency", keep in mind that with the current
scheduler on MP systems you cannot make any sort of guarantees. In fact,
scheduling latency is unbounded, even for RT processes, even when there are
several idle CPUs.
Consider the following scenario. Suppose you have several idle CPUs and two
RT processes A and B wake up and run reschedule_idle(). If their preferred
CPUs are busy, A and B will try to find an idle CPU. Both may find the same
CPU and send it one reschedule IPI each. Said CPU will schedule one of A and
B (maybe) and the other will be left on the run queue. It can stay there
indefinitely even though there are many other idle CPUs (which remain
blissfully unaware of the presence of runnable processes on the run queue).
Granted, the window is small, but if you want to *guarantee* a latency bound,
then not yet.
-- Dimitris Michailidis dimitris@engr.sgi.com- 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/
This archive was generated by hypermail 2b29 : Tue Feb 15 2000 - 21:00:18 EST