On Fri, Dec 13, 2013 at 06:25:20PM +0100, Paolo Bonzini wrote:
> Il 13/12/2013 17:07, Radim Krčmář ha scritto:
> > This bug can only be hit when the destination cpu is > 256, so the
> > request itself is buggy -- we don't support that many in kvm and it
> > would crash when initializing the vcpus if we did.
> > => It looks like we should just ignore the ipi, because we have no
> > vcpus in that cluster.
>
> That's what should happen in physical mode. Something like this patch:
>
> diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
> index 5439117d5c4c..1f8e9e1abd3b 100644
> --- a/arch/x86/kvm/lapic.c
> +++ b/arch/x86/kvm/lapic.c
> @@ -613,9 +615,10 @@ bool kvm_irq_delivery_to_apic_fast(struct kvm *kvm, struct kvm_lapic *src,
>
> if (irq->dest_mode == 0) { /* physical mode */
> if (irq->delivery_mode == APIC_DM_LOWEST ||
> - irq->dest_id == 0xff)
> + irq->dest_id == 0xff ||
> + (apic_x2apic_mode(src) && irq->dest_id > 0xff))
Here you fall back to slow path wich still uses irq->dest_id == 0xff as broadcast test.
> goto out;
> - dst = &map->phys_map[irq->dest_id & 0xff];
> + dst = &map->phys_map[irq->dest_id];
> } else {
> u32 mda = irq->dest_id << (32 - map->ldr_bits);
>
>
> On top of this, the x2apic spec documents a "broadcast" destination ID that
> could be implemented as follows. But I have no idea if this is correct and
> how it differs from the usual broadcast shorthand:
>
> @@ -815,9 +818,11 @@ static void apic_send_ipi(struct kvm_lapic *apic)
I'd rather add kvm_apic_is_broadcast() function like that:
bool kvm_apic_is_broadcast(struct kvm_lapic *l, u32 dest)
{
return dest == (apic_x2apic_mode(l) ? 0xffffffff : 0xff);
}
and use everywhere instead of irq->dest_id == 0xff test.
> irq.level = icr_low & APIC_INT_ASSERT;
> irq.trig_mode = icr_low & APIC_INT_LEVELTRIG;
> irq.shorthand = icr_low & APIC_SHORT_MASK;
> - if (apic_x2apic_mode(apic))
> + if (apic_x2apic_mode(apic)) {
> irq.dest_id = icr_high;
> - else
> + if (icr_high == 0xFFFFFFFF)
> + irq.shorthand = APIC_DEST_ALLINC;
> + } else
> irq.dest_id = GET_APIC_DEST_FIELD(icr_high);
>
> trace_kvm_apic_ipi(icr_low, irq.dest_id);
>
>
> > - Where does the 'only one supported cluster' come from?
> >
> > I only see we use 'struct kvm_lapic *logical_map[16][16];', which
> > supports 16 clusters of 16 apics = first 256 vcpus, so if we map
> > everything to logical_map[0][0:15], we would not work correctly in
> > the cluster x2apic, with > 16 vcpus.
>
> I think you're right. Something like this would be a better fix:
>
> diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
See my answer to Radim. There is not point in maintaining cid mapping that cannot be
addressed.
> index dec48bfaddb8..58745cbbb7e6 100644
> --- a/arch/x86/kvm/lapic.c
> +++ b/arch/x86/kvm/lapic.c
> @@ -198,7 +198,7 @@ static void recalculate_apic_map(struct kvm *kvm)
> cid = apic_cluster_id(new, ldr);
> lid = apic_logical_id(new, ldr);
>
> - if (lid)
> + if (cid < ARRAY_SIZE(new->logical_map) && lid)
> new->logical_map[cid][ffs(lid) - 1] = apic;
> }
> out:
> @@ -621,9 +621,12 @@ bool kvm_irq_delivery_to_apic_fast(struct kvm *kvm, struct kvm_lapic *src,
> dst = &map->phys_map[irq->dest_id & 0xff];
> } else {
> u32 mda = irq->dest_id << (32 - map->ldr_bits);
> + u32 cid = apic_cluster_id(map, mda);
>
> - dst = map->logical_map[apic_cluster_id(map, mda)];
> + if (cid >= ARRAY_SIZE(map->logical_map))
> + goto out;
>
> + dst = map->logical_map[cid];
> bitmap = apic_logical_id(map, mda);
>
> if (irq->delivery_mode == APIC_DM_LOWEST) {
> diff --git a/arch/x86/kvm/lapic.h b/arch/x86/kvm/lapic.h
> index c8b0d0d2da5c..5ccb71658894 100644
> --- a/arch/x86/kvm/lapic.h
> +++ b/arch/x86/kvm/lapic.h
> @@ -152,8 +152,6 @@ static inline u16 apic_cluster_id(struct kvm_apic_map *map, u32 ldr)
> ldr >>= 32 - map->ldr_bits;
> cid = (ldr >> map->cid_shift) & map->cid_mask;
>
> - BUG_ON(cid >= ARRAY_SIZE(map->logical_map));
> -
> return cid;
> }
>
> playground:~/work/upstream/linux-2.6/arch/x86 pbonzini$ vi kvm/lapic.c
> playground:~/work/upstream/linux-2.6/arch/x86 pbonzini$ git diff !$
> git diff kvm/lapic.c
> diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
> index dec48bfaddb8..1005185972a9 100644
> --- a/arch/x86/kvm/lapic.c
> +++ b/arch/x86/kvm/lapic.c
> @@ -143,8 +143,6 @@ static inline int kvm_apic_id(struct kvm_lapic *apic)
> return (kvm_apic_get_reg(apic, APIC_ID) >> 24) & 0xff;
> }
>
> -#define KVM_X2APIC_CID_BITS 0
> -
> static void recalculate_apic_map(struct kvm *kvm)
> {
> struct kvm_apic_map *new, *old = NULL;
> @@ -182,8 +180,7 @@ static void recalculate_apic_map(struct kvm *kvm)
> if (apic_x2apic_mode(apic)) {
> new->ldr_bits = 32;
> new->cid_shift = 16;
> - new->cid_mask = (1 << KVM_X2APIC_CID_BITS) - 1;
> - new->lid_mask = 0xffff;
> + new->cid_mask = new->lid_mask = 0xffff;
> } else if (kvm_apic_sw_enabled(apic) &&
> !new->cid_mask /* flat mode */ &&
> kvm_apic_get_reg(apic, APIC_DFR) == APIC_DFR_CLUSTER) {
> @@ -198,7 +195,7 @@ static void recalculate_apic_map(struct kvm *kvm)
> cid = apic_cluster_id(new, ldr);
> lid = apic_logical_id(new, ldr);
>
> - if (lid)
> + if (cid < ARRAY_SIZE(new->logical_map) && lid)
> new->logical_map[cid][ffs(lid) - 1] = apic;
> }
> out:
> @@ -621,9 +618,12 @@ bool kvm_irq_delivery_to_apic_fast(struct kvm *kvm, struct kvm_lapic *src,
> dst = &map->phys_map[irq->dest_id & 0xff];
> } else {
> u32 mda = irq->dest_id << (32 - map->ldr_bits);
> + u32 cid = apic_cluster_id(map, mda);
>
> - dst = map->logical_map[apic_cluster_id(map, mda)];
> + if (cid >= ARRAY_SIZE(map->logical_map))
> + goto out;
>
> + dst = map->logical_map[cid];
> bitmap = apic_logical_id(map, mda);
>
> if (irq->delivery_mode == APIC_DM_LOWEST) {
>
--
Gleb.
--
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
