On Tue, Mar 25 2003, Nick Piggin wrote:
> Jens Axboe wrote:
>
> >On Tue, Mar 25 2003, Nick Piggin wrote:
> >
> >>
> >>Nice Jens. Very good in theory but I haven't looked at the
> >>code too much yet.
> >>
> >>Would it be possible to have all queues allocate out of
> >>the one global pool of free requests. This way you could
> >>have a big minimum (say 128) and a big maximum
> >>(say min(Mbytes, spindles).
> >>
> >
> >Well not really, as far as I can see we _need_ a pool per queue. Imagine
> >a bio handed to raid, needs to be split to 6 different queues. But our
> >minimum is 4, deadlock possibility. It could probably be made to work,
> >however I greatly prefer a per-queue reserve.
> >
> OK yeah you are right there. In light of your comment below
> I'm happy with that. I was mostly worried about queues being
> restricted to a small maximum.
Understandable, we'll make max tunable.
> >>This way memory usage is decoupled from the number of
> >>queues, and busy spindles could make use of more
> >>available free requests.
> >>
> >>Oh and the max value can easily be runtime tunable, right?
> >>
> >
> >Sure. However, they don't really mean _anything_. Max is just some
> >random number to prevent one queue going nuts, and could be completely
> >removed if the vm works perfectly. Beyond some limit there's little
> >benefit to doing that, though. But MAX could be runtime tunable. Min is
> >basically just to make sure we don't kill ourselves, I don't see any
> >point in making that runtime tunable. It's not really a tunable.
> >
> OK thats good then. I would like to see max removed, however perhaps
> the VM isn't up to that yet. I'll be testing this when your code
> solidifies!
Only testing will tell, so yes you are very welcome to give it a shot.
Let me release a known working version first :)
-- Jens Axboe- 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 Mar 31 2003 - 22:00:19 EST