Mailing List Archive

[PATCH 04/10] viridian: use hypercall_vpmask in hvcall_ipi()
From: Paul Durrant <pdurrant@amazon.com>

A subsequent patch will need to IPI a mask of virtual processors potentially
wider than 64 bits. A previous patch introduced per-cpu hypercall_vpmask
to allow hvcall_flush() to deal with such wide masks. This patch modifies
the implementation of hvcall_ipi() to make use of the same mask structures,
introducing a for_each_vp() macro to facilitate traversing a mask.

Signed-off-by: Paul Durrant <pdurrant@amazon.com>
---
Cc: Wei Liu <wl@xen.org>
Cc: Jan Beulich <jbeulich@suse.com>
Cc: Andrew Cooper <andrew.cooper3@citrix.com>
Cc: "Roger Pau Monné" <roger.pau@citrix.com>
---
xen/arch/x86/hvm/viridian/viridian.c | 43 ++++++++++++++++++++++------
1 file changed, 35 insertions(+), 8 deletions(-)

diff --git a/xen/arch/x86/hvm/viridian/viridian.c b/xen/arch/x86/hvm/viridian/viridian.c
index 4ab1f14b2248..63f63093a513 100644
--- a/xen/arch/x86/hvm/viridian/viridian.c
+++ b/xen/arch/x86/hvm/viridian/viridian.c
@@ -533,6 +533,21 @@ static bool vpmask_test(struct hypercall_vpmask *vpmask, unsigned int vp)
return test_bit(vp, vpmask->mask);
}

+static unsigned int vpmask_first(struct hypercall_vpmask *vpmask)
+{
+ return find_first_bit(vpmask->mask, HVM_MAX_VCPUS);
+}
+
+static unsigned int vpmask_next(struct hypercall_vpmask *vpmask, unsigned int vp)
+{
+ return find_next_bit(vpmask->mask, HVM_MAX_VCPUS, vp + 1);
+}
+
+#define for_each_vp(vpmask, vp) \
+ for ((vp) = vpmask_first(vpmask); \
+ (vp) < HVM_MAX_VCPUS; \
+ (vp) = vpmask_next(vpmask, vp))
+
/*
* Windows should not issue the hypercalls requiring this callback in the
* case where vcpu_id would exceed the size of the mask.
@@ -624,15 +639,24 @@ static int hvcall_flush(union hypercall_input *input,
return 0;
}

+static void send_ipi(struct hypercall_vpmask *vpmask, uint8_t vector)
+{
+ struct domain *currd = current->domain;
+ unsigned int vp;
+
+ for_each_vp ( vpmask, vp )
+ vlapic_set_irq(vcpu_vlapic(currd->vcpu[vp]), vector, 0);
+}
+
static int hvcall_ipi(union hypercall_input *input,
union hypercall_output *output,
unsigned long input_params_gpa,
unsigned long output_params_gpa)
{
- struct domain *currd = current->domain;
- struct vcpu *v;
+ struct hypercall_vpmask *vpmask = &this_cpu(hypercall_vpmask);
uint32_t vector;
uint64_t vcpu_mask;
+ unsigned int vp;

/* Get input parameters. */
if ( input->fast )
@@ -669,17 +693,20 @@ static int hvcall_ipi(union hypercall_input *input,
if ( vector < 0x10 || vector > 0xff )
return -EINVAL;

- for_each_vcpu ( currd, v )
+ vpmask_empty(vpmask);
+ for (vp = 0; vp < 64; vp++)
{
- if ( v->vcpu_id >= (sizeof(vcpu_mask) * 8) )
- return -EINVAL;
+ if ( !vcpu_mask )
+ break;

- if ( !(vcpu_mask & (1ul << v->vcpu_id)) )
- continue;
+ if ( vcpu_mask & 1 )
+ vpmask_set(vpmask, vp);

- vlapic_set_irq(vcpu_vlapic(v), vector, 0);
+ vcpu_mask >>= 1;
}

+ send_ipi(vpmask, vector);
+
return 0;
}

--
2.20.1
Re: [PATCH 04/10] viridian: use hypercall_vpmask in hvcall_ipi() [ In reply to ]
On 11.11.2020 21:07, Paul Durrant wrote:
> --- a/xen/arch/x86/hvm/viridian/viridian.c
> +++ b/xen/arch/x86/hvm/viridian/viridian.c
> @@ -533,6 +533,21 @@ static bool vpmask_test(struct hypercall_vpmask *vpmask, unsigned int vp)
> return test_bit(vp, vpmask->mask);
> }
>
> +static unsigned int vpmask_first(struct hypercall_vpmask *vpmask)
> +{
> + return find_first_bit(vpmask->mask, HVM_MAX_VCPUS);
> +}
> +
> +static unsigned int vpmask_next(struct hypercall_vpmask *vpmask, unsigned int vp)
> +{
> + return find_next_bit(vpmask->mask, HVM_MAX_VCPUS, vp + 1);

Perhaps again assert on vp's range?

> +}
> +
> +#define for_each_vp(vpmask, vp) \
> + for ((vp) = vpmask_first(vpmask); \
> + (vp) < HVM_MAX_VCPUS; \
> + (vp) = vpmask_next(vpmask, vp))

Nit again: Missing blanks here and ...

> @@ -669,17 +693,20 @@ static int hvcall_ipi(union hypercall_input *input,
> if ( vector < 0x10 || vector > 0xff )
> return -EINVAL;
>
> - for_each_vcpu ( currd, v )
> + vpmask_empty(vpmask);
> + for (vp = 0; vp < 64; vp++)

... here. I also wonder if the literal 64 couldn't be expressed in
some suitable way.

Jan
RE: [PATCH 04/10] viridian: use hypercall_vpmask in hvcall_ipi() [ In reply to ]
> -----Original Message-----
> From: Jan Beulich <jbeulich@suse.com>
> Sent: 12 November 2020 08:50
> To: Paul Durrant <paul@xen.org>
> Cc: Paul Durrant <pdurrant@amazon.com>; Wei Liu <wl@xen.org>; Andrew Cooper
> <andrew.cooper3@citrix.com>; Roger Pau Monné <roger.pau@citrix.com>; xen-devel@lists.xenproject.org
> Subject: Re: [PATCH 04/10] viridian: use hypercall_vpmask in hvcall_ipi()
>
> On 11.11.2020 21:07, Paul Durrant wrote:
> > --- a/xen/arch/x86/hvm/viridian/viridian.c
> > +++ b/xen/arch/x86/hvm/viridian/viridian.c
> > @@ -533,6 +533,21 @@ static bool vpmask_test(struct hypercall_vpmask *vpmask, unsigned int vp)
> > return test_bit(vp, vpmask->mask);
> > }
> >
> > +static unsigned int vpmask_first(struct hypercall_vpmask *vpmask)
> > +{
> > + return find_first_bit(vpmask->mask, HVM_MAX_VCPUS);
> > +}
> > +
> > +static unsigned int vpmask_next(struct hypercall_vpmask *vpmask, unsigned int vp)
> > +{
> > + return find_next_bit(vpmask->mask, HVM_MAX_VCPUS, vp + 1);
>
> Perhaps again assert on vp's range?
>

Ok.

> > +}
> > +
> > +#define for_each_vp(vpmask, vp) \
> > + for ((vp) = vpmask_first(vpmask); \
> > + (vp) < HVM_MAX_VCPUS; \
> > + (vp) = vpmask_next(vpmask, vp))
>
> Nit again: Missing blanks here and ...

Oh yes.

>
> > @@ -669,17 +693,20 @@ static int hvcall_ipi(union hypercall_input *input,
> > if ( vector < 0x10 || vector > 0xff )
> > return -EINVAL;
> >
> > - for_each_vcpu ( currd, v )
> > + vpmask_empty(vpmask);
> > + for (vp = 0; vp < 64; vp++)
>
> ... here. I also wonder if the literal 64 couldn't be expressed in
> some suitable way.
>

Ok. I'll see if I can suitably macrotize it.

Paul

> Jan