Hi Paolo and Marc,
On Tue, Apr 07, 2015 at 06:20:15PM +0200, Paolo Bonzini wrote:
>
>
> On 18/03/2015 08:55, Christoffer Dall wrote:
> > Hi Stephen,
> >
> > On Wed, Mar 18, 2015 at 02:41:11PM +1100, Stephen Rothwell wrote:
> >> Hi all,
> >>
> >> Today's linux-next merge of the kvm-arm tree got a conflict in
> >> virt/kvm/arm/vgic.c between commit ae705930fca6 ("arm/arm64: KVM: Keep
> >> elrsr/aisr in sync with software model") from Linus' tree and commit
> >> 71760950bf3d ("arm/arm64: KVM: add a common vgic_queue_irq_to_lr fn")
> >> from the kvm-arm tree.
> >>
> >> I fixed it up (I think - see below) and can carry the fix as necessary
> >> (no action is required).
> >>
> >> --
> >> Cheers,
> >> Stephen Rothwell sfr@xxxxxxxxxxxxxxxx
> >>
> >> diff --cc virt/kvm/arm/vgic.c
> >> index c9f60f524588,ffd937ca5141..000000000000
> >> --- a/virt/kvm/arm/vgic.c
> >> +++ b/virt/kvm/arm/vgic.c
> >> @@@ -982,9 -1092,7 +1098,8 @@@ bool vgic_queue_irq(struct kvm_vcpu *vc
> >> if (vlr.source == sgi_source_id) {
> >> kvm_debug("LR%d piggyback for IRQ%d\n", lr, vlr.irq);
> >> BUG_ON(!test_bit(lr, vgic_cpu->lr_used));
> >> - vlr.state |= LR_STATE_PENDING;
> >> - vgic_set_lr(vcpu, lr, vlr);
> >> + vgic_queue_irq_to_lr(vcpu, irq, lr, vlr);
> >> + vgic_sync_lr_elrsr(vcpu, lr, vlr);
> >> return true;
> >> }
> >> }
> >> @@@ -1001,12 -1109,8 +1116,9 @@@
> >>
> >> vlr.irq = irq;
> >> vlr.source = sgi_source_id;
> >> - vlr.state = LR_STATE_PENDING;
> >> - if (!vgic_irq_is_edge(vcpu, irq))
> >> - vlr.state |= LR_EOI_INT;
> >> -
> >> - vgic_set_lr(vcpu, lr, vlr);
> >> + vlr.state = 0;
> >> + vgic_queue_irq_to_lr(vcpu, irq, lr, vlr);
> >> + vgic_sync_lr_elrsr(vcpu, lr, vlr);
> >>
> >> return true;
> >> }
> >
> > Looks great, thanks!
> > -Christoffer
>
> Got the same conflict when pulling from the kvm-arm tree, I used
> a different resolution though:
>
> diff --cc virt/kvm/arm/vgic.c
> index c9f60f524588,b70174e74868..8d550ff14700
> --- a/virt/kvm/arm/vgic.c
> +++ b/virt/kvm/arm/vgic.c
> @@@ -955,6 -1095,25 +1101,26 @@@ static void vgic_retire_disabled_irqs(s
> }
> }
>
> + static void vgic_queue_irq_to_lr(struct kvm_vcpu *vcpu, int irq,
> + int lr_nr, struct vgic_lr vlr)
> + {
> + if (vgic_irq_is_active(vcpu, irq)) {
> + vlr.state |= LR_STATE_ACTIVE;
> + kvm_debug("Set active, clear distributor: 0x%x\n", vlr.state);
> + vgic_irq_clear_active(vcpu, irq);
> + vgic_update_state(vcpu->kvm);
> + } else if (vgic_dist_irq_is_pending(vcpu, irq)) {
> + vlr.state |= LR_STATE_PENDING;
> + kvm_debug("Set pending: 0x%x\n", vlr.state);
> + }
> +
> + if (!vgic_irq_is_edge(vcpu, irq))
> + vlr.state |= LR_EOI_INT;
> +
> + vgic_set_lr(vcpu, lr_nr, vlr);
> ++ vgic_sync_lr_elrsr(vcpu, lr_nr, vlr);
> + }
> +
> /*
> * Queue an interrupt to a CPU virtual interface. Return true on success,
> * or false if it wasn't possible to queue it.
> @@@ -982,9 -1141,7 +1148,7 @@@ bool vgic_queue_irq(struct kvm_vcpu *vc
> if (vlr.source == sgi_source_id) {
> kvm_debug("LR%d piggyback for IRQ%d\n", lr, vlr.irq);
> BUG_ON(!test_bit(lr, vgic_cpu->lr_used));
> - vlr.state |= LR_STATE_PENDING;
> - vgic_set_lr(vcpu, lr, vlr);
> - vgic_sync_lr_elrsr(vcpu, lr, vlr);
> + vgic_queue_irq_to_lr(vcpu, irq, lr, vlr);
> return true;
> }
> }
> @@@ -1001,12 -1158,8 +1165,8 @@@
>
> vlr.irq = irq;
> vlr.source = sgi_source_id;
> - vlr.state = LR_STATE_PENDING;
> - if (!vgic_irq_is_edge(vcpu, irq))
> - vlr.state |= LR_EOI_INT;
> -
> - vgic_set_lr(vcpu, lr, vlr);
> - vgic_sync_lr_elrsr(vcpu, lr, vlr);
> + vlr.state = 0;
> + vgic_queue_irq_to_lr(vcpu, irq, lr, vlr);
>
> return true;
> }
>
>
> Christoffer, this is the same logic as Stephen's resolution, but
> can you confirm that it makes sense "semantically" as well?
>
> (Stephen, you'll still get the conflicts in linux-next for a
> couple of days as I finish local testing of KVM changes for 4.1).
>
As it turns out, it was not the same logic as Stephen's resolution.
Stephen's resolution is bussy, because vlr is passed by value to
vgic_queue_irq_to_lr() and therefore the call to sync the elrsr does not
have any effect.
Unfortunately, it seems Paolo's more correct fix did not end up in
Linus' tree, so I guess I should just send a patch?
-Christoffer
--
To unsubscribe from this list: send the line "unsubscribe linux-next" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
