On Fri, Apr 12, 2013 at 09:30:18AM +0100, Marc Zyngier wrote:
> Hi Pekka,
>
> On 12/04/13 07:52, Pekka Enberg wrote:
> > On 04/11/2013 12:36 PM, Will Deacon wrote:
> >>> Without multi-queue, we can boot Debian Wheezy to a prompt in 38MB. With
> >>> the new changes, that increases to 170MB! Any chance we can try and tackle
> >>> this regression please? I keep getting bitten by the OOM killer :(
> >
> > On 04/11/2013 07:45 PM, Sasha Levin wrote:
> >> That's definitely unwanted.
> >>
> >> I'll look into it and try sending something out today/tomorrow.
> >
> > That's very unfortunate. Can you please confirm that reverting commit
> > e026314820acc3cc967308355f3746aca238fda4 ("kvm tools: virtio-net
> > multiqueue support") fixes it? If fixing the issue turns out to be
> > difficult, we should just take multi-queue out until the issues is resolved.
>
> It does, at least for an MMIO based setup.
>
> It would be a pity to completely loose the functionality though. Surely
> we can find a way to initialize the extra queues on demand. And if not,
> maybe introduce some kind of tunable parameter...
I suppose we *could* move the whole thing over to dynamic allocation and
let the maximum number of queues be an lkvm parameter, but it feels like a
lot of work without really solving the problem.
Anyway, thanks Sasha for starting a separate thread about this. Hopefully we
can resolve it in the spec and then update kvmtool/linux accordingly. As
long as the issue is being addressed, then I don't see that there's a
pressing need to revert the commit above.
Cheers,
Will
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
