Re: Priority Inheritance Test (Real-Time Preemption)

[john cooper]

Bill Huey (hui) wrote:
> On Mon, Nov 22, 2004 at 09:16:18AM -0500, john cooper wrote:
>
> > I'd hazard a guess the reason existing implementations do not
> > do this type of dependency-chain closure is the complexity of a
> > general approach.  Getting correct behavior and scaling on SMP
> > require some restrictions of how lock ownership is maintained,
> > otherwise fine grained locking is not possible.  Another likely
>
>
> What do you mean by that ? Are you talking about strict priority
> obedience by the system ?

Not quite if I understand your question.  I was referring to
avoiding having a global lock to synchronize the conglomerate
data structure when doing a PI dependency walk.  The problem
is the lock must be acquired not only in PI scenarios but in
any case which may possibly lead to one or affect a concurrent
PI in progress.

True this global lock is mostly an issue for large count
SMP systems.  But as witnessed by such voodoo[1] mechanisms
as rcu, scalability problems are real at that end of the
spectrum.

-john


[1] in a 'nice' way.


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