viridian: use softirq batching in hvcall_ipi()

vlapic_ipi() uses a softirq batching mechanism to improve the efficiency of
sending a IPIs to large number of processors. This patch modifies send_ipi()
(the worker function called by hvcall_ipi()) to also make use of the
mechanism when there multiple bits set the hypercall_vpmask.

Signed-off-by: Paul Durrant <pdurrant@amazon.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Wei Liu <wl@xen.org>

xen/arch/x86/hvm/viridian/viridian.c

@@ -11,6 +11,7 @@
 #include <xen/hypercall.h>
 #include <xen/domain_page.h>
 #include <xen/param.h>
+#include <xen/softirq.h>
 #include <asm/guest/hyperv-tlfs.h>
 #include <asm/paging.h>
 #include <asm/p2m.h>
@@ -571,6 +572,11 @@ static unsigned int vpmask_next(const struct hypercall_vpmask *vpmask,
 	      (vp) < HVM_MAX_VCPUS; \
 	      (vp) = vpmask_next(vpmask, vp) )
 
+static unsigned int vpmask_nr(const struct hypercall_vpmask *vpmask)
+{
+    return bitmap_weight(vpmask->mask, HVM_MAX_VCPUS);
+}
+
 /*
  * Windows should not issue the hypercalls requiring this callback in the
  * case where vcpu_id would exceed the size of the mask.
@@ -654,10 +660,17 @@ static int hvcall_flush(const union hypercall_input *input,
 static void send_ipi(struct hypercall_vpmask *vpmask, uint8_t vector)
 {
     struct domain *currd = current->domain;
+    unsigned int nr = vpmask_nr(vpmask);
     unsigned int vp;
 
+    if ( nr > 1 )
+        cpu_raise_softirq_batch_begin();
+
     for_each_vp ( vpmask, vp )
         vlapic_set_irq(vcpu_vlapic(currd->vcpu[vp]), vector, 0);
+
+    if ( nr > 1 )
+        cpu_raise_softirq_batch_finish();
 }
 
 static int hvcall_ipi(const union hypercall_input *input,