Re: [Xen-devel] [RFC PATCH] xen/arm: Deliver interrupts to vcpu specified in IROUTER

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Mon, 2014-09-08 at 19:55 +0530, Vijay Kilari wrote:
> On Mon, Sep 8, 2014 at 7:33 PM, Vijay Kilari <vijay.kilari@xxxxxxxxx> wrote:
> > On Mon, Sep 8, 2014 at 7:17 PM, Ian Campbell <Ian.Campbell@xxxxxxxxxx> 
> > wrote:
> >> On Mon, 2014-09-08 at 19:11 +0530, Vijay Kilari wrote:
> >>
> >> Why have we dropped the list?
> >
> > I have not dropped the list. I still see list in my email.
> >
> >>
> >>> Hi Stefano,
> >>>
> >>>  Apart from your comments below, I encounter below scenario.
> >>>
> >>> The below function vgic_irq_rank() in vgic.c is not generic. The rank is 
> >>> always
> >>> depends on register type it is going to access. So it cannot be
> >>> just hardcoded.
> >>>
> >>> struct vgic_irq_rank *(struct vcpu *v, unsigned int irq)
> >>> {
> >>>     return vgic_rank_offset(v, 8, irq, DABT_WORD);
> >>> }
> >>>
> >>> This function works ok for GIC v2. But this cannot be used for
> >>> GICv3 to access registers like IROUTER which are u64. The rank
> >>> calcuation goes wrong and there by takes wrong rank lock.
> >>
> >> This sounds to me like a bug which should be fixed. A rank is just a
> >> group of 32 consecutive IRQs and their associated register values, it
> >> shouldn't matter whether those registers are 8-, 16-, 32- or 64-bits.
> >
> >  Now I think of creating a callback to return the right rank for given irq
> > based on v2 or v3. This will fix and keep the old code intact.
> 
> This should fix the issue.

Looks plausible. Will you fold it into your series?

> 
>  struct vgic_irq_rank *vgic_rank_irq(struct vcpu *v, unsigned int irq)
>  {
> -    return vgic_rank_offset(v, 8, irq, DABT_WORD);
> +    int rank = irq/32;
> 
> +    if ( rank == 0 )
> +        return v->arch.vgic.private_irqs;
> +    else if ( rank <= DOMAIN_NR_RANKS(v->domain) )
> +        return &v->domain->arch.vgic.shared_irqs[rank - 1];
> +    else
> +        return NULL;
> }



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel