[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Xen-devel] [PATCH v4 05/17] xen/arm: ITS: implement hw_irq_controller for LPIs
- To: Marc Zyngier <marc.zyngier@xxxxxxx>
- From: Ian Campbell <ian.campbell@xxxxxxxxxx>
- Date: Thu, 16 Jul 2015 17:37:19 +0100
- Cc: Vijay Kilari <vijay.kilari@xxxxxxxxx>, Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>, "Prasun.Kapoor@xxxxxxxxxxxxxxxxxx" <Prasun.Kapoor@xxxxxxxxxxxxxxxxxx>, Vijaya Kumar K <vijaya.kumar@xxxxxxxxxxxxxxxxxx>, Tim Deegan <tim@xxxxxxx>, "xen-devel@xxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxx>, Julien Grall <julien.grall@xxxxxxxxxx>, "stefano.stabellini@xxxxxxxxxx" <stefano.stabellini@xxxxxxxxxx>, "manish.jaggi@xxxxxxxxxxxxxxxxxx" <manish.jaggi@xxxxxxxxxxxxxxxxxx>
- Delivery-date: Thu, 16 Jul 2015 16:37:37 +0000
- List-id: Xen developer discussion <xen-devel.lists.xen.org>
On Thu, 2015-07-16 at 17:27 +0100, Marc Zyngier wrote:
> On 16/07/15 17:18, Ian Campbell wrote:
> > On Thu, 2015-07-16 at 16:21 +0100, Marc Zyngier wrote:
> >> Hope this helps,
> >
> > It, plus the chat we had on IRC did, yes, thanks.
> >
> > In summary:
> >
> > I was very confusedly talking about INV when I meant SYNC.
> >
> > There is a real issue with the update of its_dev->collection in
> > its_set_affinity since it will result in the wrong collection being used
> > for subsequent affinity operations.
>
> Yeah, this is a very stupid bug on my side, and I realize that I never
> saw it because the whole ITS driver has been written on a model that can
> only generate a single interrupt per device. Lovely.
Oh dear!
> > Marc is intending to replace its_dev->collection with an array
> > (presumably dynamically allocated) in its_dev mapping eventid to a
> > collection.
>
> Exactly.
>
> > I think that is probably the right answer for Xen too.
>
> I'd think so.
>
> Thanks so much for putting me right!
No problem!
Ian.
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel
|
