Re: Scheduler Cleanup

[Davide Libenzi]

On Wed, 5 Dec 2001, Mike Kravetz wrote:

> On Wed, Dec 05, 2001 at 03:44:42PM -0800, Davide Libenzi wrote:
> > Anyway me too have verified an increased latency with sched_yield() test
> > and next days I'm going to try the real one with the cycle counter
> > sampler.
> > I've a suspect, but i've to see the disassembly of schedule() before
> > talking :)
>
> One thing to note is that possible acquisition of the runqueue
> lock was reintroduced in sys_sched_yield().  From looking at
> the code, it seems the purpose was to ?add fairness? in the case
> of multiple yielders.  Is that correct Ingo?

Yep, suppose you've three running tasks A, B and C ( in that order ) and
suppose A and B are yield()ing.
You get:

A B A B A B A B A B A ...

until the priority of A or B drops, then C has a chance to execute.
Since with the new counter decay code priority remains up until a sudden
drop, why don't we decrease counter by K ( 1? ) for each sched_yield() call ?
In this way we can easily avoid the above pattern in case of multiple
yield()ers.

PS: sched_yield() code is very important because it is used inside the
pthread spinlock() wait path for a bunch of times before falling into
usleep().

- Davide