Re: RFC for a new Scheduling policy/class in the Linux-kernel

["Chris Friesen" <cfriesen@nortel.com>]

Raistlin wrote:

> Very basically: from the analysis point of view one easy and effective
> solution would be to have the blocked-running tasks --i.e., the tasks
> blocked on some lock that have been left on the rq to proxy-execute the
> lock owner-- busy waiting while the lock owner is running. This allows
> for retaining a lot of nice properties BWI already has, as far as
> analyzability is concerned.
> 
> On the other hand, from the practical end efficiency point of view, it
> would be not that difficult to block these otherwise-spinning tasks, in
> order to avoid wasting CPU time too much... The only important thing is
> to properly account the budget of the correct server/group (which
> probably must be the otherwise-spinning task's one), or the analysis is
> gone again! :-O

Could you elaborate on this "proper accounting"?

If task A is blocked waiting for a lock (held by a task B on another
cpu) and we run task C instead, how would you propose that the
accounting be handled?

Chris