[PATCH v3 04/13] viridian: introduce a per-cpu hypercall_vpmask and accessor functions...

[Paul Durrant <paul@xen.org>]

From: Paul Durrant <pdurrant@amazon.com>

... and make use of them in hvcall_flush()/need_flush().

Subsequent patches will need to deal with virtual processor masks potentially
wider than 64 bits. Thus, to avoid using too much stack, this patch
introduces global per-cpu virtual processor masks and converts the
implementation of hvcall_flush() to use them.

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>

v2:
 - Modified vpmask_set() to take a base 'vp' and a 64-bit 'mask', still
   looping over the mask as bitmap.h does not provide a primitive for copying
   one mask into another at an offset
 - Added ASSERTions to verify that we don't attempt to set or test bits
   beyond the limit of the map
---
 xen/arch/x86/hvm/viridian/viridian.c | 58 ++++++++++++++++++++++++++--
 1 file changed, 54 insertions(+), 4 deletions(-)

diff --git a/xen/arch/x86/hvm/viridian/viridian.c b/xen/arch/x86/hvm/viridian/viridian.c
index 8ac1b32565a8..d8f47b8528a2 100644
--- a/xen/arch/x86/hvm/viridian/viridian.c
+++ b/xen/arch/x86/hvm/viridian/viridian.c
@@ -507,15 +507,59 @@ void viridian_domain_deinit(struct domain *d)
     XFREE(d->arch.hvm.viridian);
 }
 
+struct hypercall_vpmask {
+    DECLARE_BITMAP(mask, HVM_MAX_VCPUS);
+};
+
+static DEFINE_PER_CPU(struct hypercall_vpmask, hypercall_vpmask);
+
+static void vpmask_empty(struct hypercall_vpmask *vpmask)
+{
+    bitmap_zero(vpmask->mask, HVM_MAX_VCPUS);
+}
+
+static void vpmask_set(struct hypercall_vpmask *vpmask, unsigned int vp,
+                       uint64_t mask)
+{
+    unsigned int count = sizeof(mask) * 8;
+
+    while ( count-- )
+    {
+        if ( !mask )
+            break;
+
+        if ( mask & 1 )
+        {
+            ASSERT(vp < HVM_MAX_VCPUS);
+            __set_bit(vp, vpmask->mask);
+        }
+
+        mask >>= 1;
+        vp++;
+    }
+}
+
+static void vpmask_fill(struct hypercall_vpmask *vpmask)
+{
+    bitmap_fill(vpmask->mask, HVM_MAX_VCPUS);
+}
+
+static bool vpmask_test(const struct hypercall_vpmask *vpmask,
+                        unsigned int vp)
+{
+    ASSERT(vp < HVM_MAX_VCPUS);
+    return test_bit(vp, vpmask->mask);
+}
+
 /*
  * Windows should not issue the hypercalls requiring this callback in the
  * case where vcpu_id would exceed the size of the mask.
  */
 static bool need_flush(void *ctxt, struct vcpu *v)
 {
-    uint64_t vcpu_mask = *(uint64_t *)ctxt;
+    struct hypercall_vpmask *vpmask = ctxt;
 
-    return vcpu_mask & (1ul << v->vcpu_id);
+    return vpmask_test(vpmask, v->vcpu_id);
 }
 
 union hypercall_input {
@@ -546,6 +590,7 @@ static int hvcall_flush(const union hypercall_input *input,
                         paddr_t input_params_gpa,
                         paddr_t output_params_gpa)
 {
+    struct hypercall_vpmask *vpmask = &this_cpu(hypercall_vpmask);
     struct {
         uint64_t address_space;
         uint64_t flags;
@@ -567,13 +612,18 @@ static int hvcall_flush(const union hypercall_input *input,
      * so err on the safe side.
      */
     if ( input_params.flags & HV_FLUSH_ALL_PROCESSORS )
-        input_params.vcpu_mask = ~0ul;
+        vpmask_fill(vpmask);
+    else
+    {
+        vpmask_empty(vpmask);
+        vpmask_set(vpmask, 0, input_params.vcpu_mask);
+    }
 
     /*
      * A false return means that another vcpu is currently trying
      * a similar operation, so back off.
      */
-    if ( !paging_flush_tlb(need_flush, &input_params.vcpu_mask) )
+    if ( !paging_flush_tlb(need_flush, vpmask) )
         return -ERESTART;
 
     output->rep_complete = input->rep_count;
-- 
2.20.1