KVM: x86: Shove vp_bitmap handling down into sparse_set_to_vcpu_mask()

Move the vp_bitmap "allocation" that's needed to handle mismatched vp_index
values down into sparse_set_to_vcpu_mask() and drop __always_inline from
said helper.  The need for an intermediate vp_bitmap is a detail that's
specific to the sparse translation with mismatched VP<=>vCPU indexes and
does not need to be exposed to the caller.

Regarding the __always_inline, prior to commit f21dd494 ("KVM: x86:
hyperv: optimize sparse VP set processing") the helper, then named
hv_vcpu_in_sparse_set(), was a tiny bit of code that effectively boiled
down to a handful of bit ops.  The __always_inline was understandable, if
not justifiable.  Since the aforementioned change, sparse_set_to_vcpu_mask()
is a chunky 350-450+ bytes of code without KASAN=y, and balloons to 1100+
with KASAN=y.  In other words, it has no business being forcefully inlined.
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Message-Id: <20211207220926.718794-7-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

arch/x86/kvm/hyperv.c

@@ -1713,32 +1713,47 @@ int kvm_hv_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata, bool host)
 		return kvm_hv_get_msr(vcpu, msr, pdata, host);
 }
 
-static __always_inline unsigned long *sparse_set_to_vcpu_mask(
-	struct kvm *kvm, u64 *sparse_banks, u64 valid_bank_mask,
-	u64 *vp_bitmap, unsigned long *vcpu_bitmap)
+static void sparse_set_to_vcpu_mask(struct kvm *kvm, u64 *sparse_banks,
+				    u64 valid_bank_mask, unsigned long *vcpu_mask)
 {
 	struct kvm_hv *hv = to_kvm_hv(kvm);
+	bool has_mismatch = atomic_read(&hv->num_mismatched_vp_indexes);
+	u64 vp_bitmap[KVM_HV_MAX_SPARSE_VCPU_SET_BITS];
 	struct kvm_vcpu *vcpu;
 	int bank, sbank = 0;
 	unsigned long i;
+	u64 *bitmap;
+
+	BUILD_BUG_ON(sizeof(vp_bitmap) >
+		     sizeof(*vcpu_mask) * BITS_TO_LONGS(KVM_MAX_VCPUS));
+
+	/*
+	 * If vp_index == vcpu_idx for all vCPUs, fill vcpu_mask directly, else
+	 * fill a temporary buffer and manually test each vCPU's VP index.
+	 */
+	if (likely(!has_mismatch))
+		bitmap = (u64 *)vcpu_mask;
+	else
+		bitmap = vp_bitmap;
 
-	memset(vp_bitmap, 0,
-	       KVM_HV_MAX_SPARSE_VCPU_SET_BITS * sizeof(*vp_bitmap));
+	/*
+	 * Each set of 64 VPs is packed into sparse_banks, with valid_bank_mask
+	 * having a '1' for each bank that exists in sparse_banks.  Sets must
+	 * be in ascending order, i.e. bank0..bankN.
+	 */
+	memset(bitmap, 0, sizeof(vp_bitmap));
 	for_each_set_bit(bank, (unsigned long *)&valid_bank_mask,
 			 KVM_HV_MAX_SPARSE_VCPU_SET_BITS)
-		vp_bitmap[bank] = sparse_banks[sbank++];
+		bitmap[bank] = sparse_banks[sbank++];
 
-	if (likely(!atomic_read(&hv->num_mismatched_vp_indexes))) {
-		/* for all vcpus vp_index == vcpu_idx */
-		return (unsigned long *)vp_bitmap;
-	}
+	if (likely(!has_mismatch))
+		return;
 
-	bitmap_zero(vcpu_bitmap, KVM_MAX_VCPUS);
+	bitmap_zero(vcpu_mask, KVM_MAX_VCPUS);
 	kvm_for_each_vcpu(i, vcpu, kvm) {
 		if (test_bit(kvm_hv_get_vpindex(vcpu), (unsigned long *)vp_bitmap))
-			__set_bit(i, vcpu_bitmap);
+			__set_bit(i, vcpu_mask);
 	}
-	return vcpu_bitmap;
 }
 
 struct kvm_hv_hcall {
@@ -1775,9 +1790,7 @@ static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc, bool
 	struct kvm *kvm = vcpu->kvm;
 	struct hv_tlb_flush_ex flush_ex;
 	struct hv_tlb_flush flush;
-	u64 vp_bitmap[KVM_HV_MAX_SPARSE_VCPU_SET_BITS];
-	DECLARE_BITMAP(vcpu_bitmap, KVM_MAX_VCPUS);
-	unsigned long *vcpu_mask;
+	DECLARE_BITMAP(vcpu_mask, KVM_MAX_VCPUS);
 	u64 valid_bank_mask;
 	u64 sparse_banks[KVM_HV_MAX_SPARSE_VCPU_SET_BITS];
 	bool all_cpus;
@@ -1870,11 +1883,9 @@ static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc, bool
 	if (all_cpus) {
 		kvm_make_all_cpus_request(kvm, KVM_REQ_TLB_FLUSH_GUEST);
 	} else {
-		vcpu_mask = sparse_set_to_vcpu_mask(kvm, sparse_banks, valid_bank_mask,
-						    vp_bitmap, vcpu_bitmap);
+		sparse_set_to_vcpu_mask(kvm, sparse_banks, valid_bank_mask, vcpu_mask);
 
-		kvm_make_vcpus_request_mask(kvm, KVM_REQ_TLB_FLUSH_GUEST,
-					    vcpu_mask);
+		kvm_make_vcpus_request_mask(kvm, KVM_REQ_TLB_FLUSH_GUEST, vcpu_mask);
 	}
 
 ret_success:
@@ -1907,9 +1918,7 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc, bool
 	struct kvm *kvm = vcpu->kvm;
 	struct hv_send_ipi_ex send_ipi_ex;
 	struct hv_send_ipi send_ipi;
-	u64 vp_bitmap[KVM_HV_MAX_SPARSE_VCPU_SET_BITS];
-	DECLARE_BITMAP(vcpu_bitmap, KVM_MAX_VCPUS);
-	unsigned long *vcpu_mask;
+	DECLARE_BITMAP(vcpu_mask, KVM_MAX_VCPUS);
 	unsigned long valid_bank_mask;
 	u64 sparse_banks[KVM_HV_MAX_SPARSE_VCPU_SET_BITS];
 	u32 vector;
@@ -1965,11 +1974,13 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc, bool
 	if ((vector < HV_IPI_LOW_VECTOR) || (vector > HV_IPI_HIGH_VECTOR))
 		return HV_STATUS_INVALID_HYPERCALL_INPUT;
 
-	vcpu_mask = all_cpus ? NULL :
-		sparse_set_to_vcpu_mask(kvm, sparse_banks, valid_bank_mask,
-					vp_bitmap, vcpu_bitmap);
+	if (all_cpus) {
+		kvm_send_ipi_to_many(kvm, vector, NULL);
+	} else {
+		sparse_set_to_vcpu_mask(kvm, sparse_banks, valid_bank_mask, vcpu_mask);
 
-	kvm_send_ipi_to_many(kvm, vector, vcpu_mask);
+		kvm_send_ipi_to_many(kvm, vector, vcpu_mask);
+	}
 
 ret_success:
 	return HV_STATUS_SUCCESS;