kvm: x86: hyperv: add KVM_CAP_HYPERV_SYNIC2

There is a flaw in the Hyper-V SynIC implementation in KVM: when message
page or event flags page is enabled by setting the corresponding msr,
KVM zeroes it out.  This is problematic because on migration the
corresponding MSRs are loaded on the destination, so the content of
those pages is lost.

This went unnoticed so far because the only user of those pages was
in-KVM hyperv synic timers, which could continue working despite that
zeroing.

Newer QEMU uses those pages for Hyper-V VMBus implementation, and
zeroing them breaks the migration.

Besides, in newer QEMU the content of those pages is fully managed by
QEMU, so zeroing them is undesirable even when writing the MSRs from the
guest side.

To support this new scheme, introduce a new capability,
KVM_CAP_HYPERV_SYNIC2, which, when enabled, makes sure that the synic
pages aren't zeroed out in KVM.
Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

Documentation/virtual/kvm/api.txt

@@ -4329,3 +4329,12 @@ Querying this capability returns a bitmap indicating the possible
 virtual SMT modes that can be set using KVM_CAP_PPC_SMT.  If bit N
 (counting from the right) is set, then a virtual SMT mode of 2^N is
 available.
+
+8.11 KVM_CAP_HYPERV_SYNIC2
+
+Architectures: x86
+
+This capability enables a newer version of Hyper-V Synthetic interrupt
+controller (SynIC).  The only difference with KVM_CAP_HYPERV_SYNIC is that KVM
+doesn't clear SynIC message and event flags pages when they are enabled by
+writing to the respective MSRs.

arch/x86/include/asm/kvm_host.h

@@ -462,6 +462,7 @@ struct kvm_vcpu_hv_synic {
 	DECLARE_BITMAP(auto_eoi_bitmap, 256);
 	DECLARE_BITMAP(vec_bitmap, 256);
 	bool active;
+	bool dont_zero_synic_pages;
 };
 
 /* Hyper-V per vcpu emulation context */

arch/x86/kvm/hyperv.c

@@ -221,7 +221,8 @@ static int synic_set_msr(struct kvm_vcpu_hv_synic *synic,
 		synic->version = data;
 		break;
 	case HV_X64_MSR_SIEFP:
-		if (data & HV_SYNIC_SIEFP_ENABLE)
+		if ((data & HV_SYNIC_SIEFP_ENABLE) && !host &&
+		    !synic->dont_zero_synic_pages)
 			if (kvm_clear_guest(vcpu->kvm,
 					    data & PAGE_MASK, PAGE_SIZE)) {
 				ret = 1;
@@ -232,7 +233,8 @@ static int synic_set_msr(struct kvm_vcpu_hv_synic *synic,
 			synic_exit(synic, msr);
 		break;
 	case HV_X64_MSR_SIMP:
-		if (data & HV_SYNIC_SIMP_ENABLE)
+		if ((data & HV_SYNIC_SIMP_ENABLE) && !host &&
+		    !synic->dont_zero_synic_pages)
 			if (kvm_clear_guest(vcpu->kvm,
 					    data & PAGE_MASK, PAGE_SIZE)) {
 				ret = 1;
@@ -687,14 +689,17 @@ void kvm_hv_vcpu_init(struct kvm_vcpu *vcpu)
 		stimer_init(&hv_vcpu->stimer[i], i);
 }
 
-int kvm_hv_activate_synic(struct kvm_vcpu *vcpu)
+int kvm_hv_activate_synic(struct kvm_vcpu *vcpu, bool dont_zero_synic_pages)
 {
+	struct kvm_vcpu_hv_synic *synic = vcpu_to_synic(vcpu);
+
 	/*
 	 * Hyper-V SynIC auto EOI SINT's are
 	 * not compatible with APICV, so deactivate APICV
 	 */
 	kvm_vcpu_deactivate_apicv(vcpu);
-	vcpu_to_synic(vcpu)->active = true;
+	synic->active = true;
+	synic->dont_zero_synic_pages = dont_zero_synic_pages;
 	return 0;
 }
 

arch/x86/kvm/hyperv.h

@@ -56,7 +56,7 @@ int kvm_hv_hypercall(struct kvm_vcpu *vcpu);
 void kvm_hv_irq_routing_update(struct kvm *kvm);
 int kvm_hv_synic_set_irq(struct kvm *kvm, u32 vcpu_id, u32 sint);
 void kvm_hv_synic_send_eoi(struct kvm_vcpu *vcpu, int vector);
-int kvm_hv_activate_synic(struct kvm_vcpu *vcpu);
+int kvm_hv_activate_synic(struct kvm_vcpu *vcpu, bool dont_zero_synic_pages);
 
 void kvm_hv_vcpu_init(struct kvm_vcpu *vcpu);
 void kvm_hv_vcpu_uninit(struct kvm_vcpu *vcpu);

arch/x86/kvm/x86.c

@@ -2659,6 +2659,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_HYPERV_VAPIC:
 	case KVM_CAP_HYPERV_SPIN:
 	case KVM_CAP_HYPERV_SYNIC:
+	case KVM_CAP_HYPERV_SYNIC2:
 	case KVM_CAP_PCI_SEGMENT:
 	case KVM_CAP_DEBUGREGS:
 	case KVM_CAP_X86_ROBUST_SINGLESTEP:
@@ -3382,10 +3383,14 @@ static int kvm_vcpu_ioctl_enable_cap(struct kvm_vcpu *vcpu,
 		return -EINVAL;
 
 	switch (cap->cap) {
+	case KVM_CAP_HYPERV_SYNIC2:
+		if (cap->args[0])
+			return -EINVAL;
 	case KVM_CAP_HYPERV_SYNIC:
 		if (!irqchip_in_kernel(vcpu->kvm))
 			return -EINVAL;
-		return kvm_hv_activate_synic(vcpu);
+		return kvm_hv_activate_synic(vcpu, cap->cap ==
+					     KVM_CAP_HYPERV_SYNIC2);
 	default:
 		return -EINVAL;
 	}

include/uapi/linux/kvm.h

@@ -927,6 +927,7 @@ struct kvm_ppc_resize_hpt {
 #define KVM_CAP_S390_CMMA_MIGRATION 145
 #define KVM_CAP_PPC_FWNMI 146
 #define KVM_CAP_PPC_SMT_POSSIBLE 147
+#define KVM_CAP_HYPERV_SYNIC2 148
 
 #ifdef KVM_CAP_IRQ_ROUTING