KVM: x86/mmu: Trigger APIC-access page reload iff vendor code cares

Request an APIC-access page reload when the backing page is migrated (or
unmapped) if and only if vendor code actually plugs the backing pfn into
structures that reside outside of KVM's MMU.  This avoids kicking all
vCPUs in the (hopefully infrequent) scenario where the backing page is
migrated/invalidated.

Unlike VMX's APICv, SVM's AVIC doesn't plug the backing pfn directly into
the VMCB and so doesn't need a hook to invalidate an out-of-MMU "mapping".

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Link: https://lore.kernel.org/r/20230602011518.787006-4-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>

arch/x86/kvm/mmu/mmu.c

@@ -1600,7 +1600,8 @@ bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
 	if (tdp_mmu_enabled)
 		flush = kvm_tdp_mmu_unmap_gfn_range(kvm, range, flush);
 
-	if (range->slot->id == APIC_ACCESS_PAGE_PRIVATE_MEMSLOT)
+	if (kvm_x86_ops.set_apic_access_page_addr &&
+	    range->slot->id == APIC_ACCESS_PAGE_PRIVATE_MEMSLOT)
 		kvm_make_all_cpus_request(kvm, KVM_REQ_APIC_PAGE_RELOAD);
 
 	return flush;