Merge tag 'kvm-arm-for-v4.9-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm into HEAD

KVM/ARM updates for v4.9-rc4

- Kick the vcpu when a pending interrupt becomes pending again
- Prevent access to invalid interrupt registers
- Invalid TLBs when two vcpus from the same VM share a CPU

Author: bonzini

arch/arm/include/asm/kvm_asm.h

@@ -66,6 +66,7 @@ extern char __kvm_hyp_vector[];
 extern void __kvm_flush_vm_context(void);
 extern void __kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa);
 extern void __kvm_tlb_flush_vmid(struct kvm *kvm);
+extern void __kvm_tlb_flush_local_vmid(struct kvm_vcpu *vcpu);
 
 extern int __kvm_vcpu_run(struct kvm_vcpu *vcpu);
 

arch/arm/include/asm/kvm_host.h

@@ -57,6 +57,9 @@ struct kvm_arch {
 	/* VTTBR value associated with below pgd and vmid */
 	u64    vttbr;
 
+	/* The last vcpu id that ran on each physical CPU */
+	int __percpu *last_vcpu_ran;
+
 	/* Timer */
 	struct arch_timer_kvm	timer;
 

arch/arm/include/asm/kvm_hyp.h

@@ -71,6 +71,7 @@
 #define ICIALLUIS	__ACCESS_CP15(c7, 0, c1, 0)
 #define ATS1CPR		__ACCESS_CP15(c7, 0, c8, 0)
 #define TLBIALLIS	__ACCESS_CP15(c8, 0, c3, 0)
+#define TLBIALL		__ACCESS_CP15(c8, 0, c7, 0)
 #define TLBIALLNSNHIS	__ACCESS_CP15(c8, 4, c3, 4)
 #define PRRR		__ACCESS_CP15(c10, 0, c2, 0)
 #define NMRR		__ACCESS_CP15(c10, 0, c2, 1)

arch/arm/kvm/arm.c

@@ -114,11 +114,18 @@ void kvm_arch_check_processor_compat(void *rtn)
  */
 int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 {
-	int ret = 0;
+	int ret, cpu;
 
 	if (type)
 		return -EINVAL;
 
+	kvm->arch.last_vcpu_ran = alloc_percpu(typeof(*kvm->arch.last_vcpu_ran));
+	if (!kvm->arch.last_vcpu_ran)
+		return -ENOMEM;
+
+	for_each_possible_cpu(cpu)
+		*per_cpu_ptr(kvm->arch.last_vcpu_ran, cpu) = -1;
+
 	ret = kvm_alloc_stage2_pgd(kvm);
 	if (ret)
 		goto out_fail_alloc;
@@ -141,6 +148,8 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
 out_free_stage2_pgd:
 	kvm_free_stage2_pgd(kvm);
 out_fail_alloc:
+	free_percpu(kvm->arch.last_vcpu_ran);
+	kvm->arch.last_vcpu_ran = NULL;
 	return ret;
 }
 
@@ -168,6 +177,9 @@ void kvm_arch_destroy_vm(struct kvm *kvm)
 {
 	int i;
 
+	free_percpu(kvm->arch.last_vcpu_ran);
+	kvm->arch.last_vcpu_ran = NULL;
+
 	for (i = 0; i < KVM_MAX_VCPUS; ++i) {
 		if (kvm->vcpus[i]) {
 			kvm_arch_vcpu_free(kvm->vcpus[i]);
@@ -312,6 +324,19 @@ int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu)
 
 void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 {
+	int *last_ran;
+
+	last_ran = this_cpu_ptr(vcpu->kvm->arch.last_vcpu_ran);
+
+	/*
+	 * We might get preempted before the vCPU actually runs, but
+	 * over-invalidation doesn't affect correctness.
+	 */
+	if (*last_ran != vcpu->vcpu_id) {
+		kvm_call_hyp(__kvm_tlb_flush_local_vmid, vcpu);
+		*last_ran = vcpu->vcpu_id;
+	}
+
 	vcpu->cpu = cpu;
 	vcpu->arch.host_cpu_context = this_cpu_ptr(kvm_host_cpu_state);
 

arch/arm/kvm/hyp/tlb.c

@@ -55,6 +55,21 @@ void __hyp_text __kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa)
 	__kvm_tlb_flush_vmid(kvm);
 }
 
+void __hyp_text __kvm_tlb_flush_local_vmid(struct kvm_vcpu *vcpu)
+{
+	struct kvm *kvm = kern_hyp_va(kern_hyp_va(vcpu)->kvm);
+
+	/* Switch to requested VMID */
+	write_sysreg(kvm->arch.vttbr, VTTBR);
+	isb();
+
+	write_sysreg(0, TLBIALL);
+	dsb(nsh);
+	isb();
+
+	write_sysreg(0, VTTBR);
+}
+
 void __hyp_text __kvm_flush_vm_context(void)
 {
 	write_sysreg(0, TLBIALLNSNHIS);

arch/arm64/include/asm/kvm_asm.h

@@ -54,6 +54,7 @@ extern char __kvm_hyp_vector[];
 extern void __kvm_flush_vm_context(void);
 extern void __kvm_tlb_flush_vmid_ipa(struct kvm *kvm, phys_addr_t ipa);
 extern void __kvm_tlb_flush_vmid(struct kvm *kvm);
+extern void __kvm_tlb_flush_local_vmid(struct kvm_vcpu *vcpu);
 
 extern int __kvm_vcpu_run(struct kvm_vcpu *vcpu);
 

arch/arm64/include/asm/kvm_host.h

@@ -62,6 +62,9 @@ struct kvm_arch {
 	/* VTTBR value associated with above pgd and vmid */
 	u64    vttbr;
 
+	/* The last vcpu id that ran on each physical CPU */
+	int __percpu *last_vcpu_ran;
+
 	/* The maximum number of vCPUs depends on the used GIC model */
 	int max_vcpus;
 

arch/arm64/include/asm/kvm_mmu.h

@@ -128,7 +128,7 @@ static inline unsigned long __kern_hyp_va(unsigned long v)
 	return v;
 }
 
-#define kern_hyp_va(v) 	(typeof(v))(__kern_hyp_va((unsigned long)(v)))
+#define kern_hyp_va(v) 	((typeof(v))(__kern_hyp_va((unsigned long)(v))))
 
 /*
  * We currently only support a 40bit IPA.

arch/arm64/kvm/hyp/tlb.c

@@ -64,6 +64,21 @@ void __hyp_text __kvm_tlb_flush_vmid(struct kvm *kvm)
 	write_sysreg(0, vttbr_el2);
 }
 
+void __hyp_text __kvm_tlb_flush_local_vmid(struct kvm_vcpu *vcpu)
+{
+	struct kvm *kvm = kern_hyp_va(kern_hyp_va(vcpu)->kvm);
+
+	/* Switch to requested VMID */
+	write_sysreg(kvm->arch.vttbr, vttbr_el2);
+	isb();
+
+	asm volatile("tlbi vmalle1" : : );
+	dsb(nsh);
+	isb();
+
+	write_sysreg(0, vttbr_el2);
+}
+
 void __hyp_text __kvm_flush_vm_context(void)
 {
 	dsb(ishst);

virt/kvm/arm/vgic/vgic-mmio.c

@@ -453,17 +453,33 @@ struct vgic_io_device *kvm_to_vgic_iodev(const struct kvm_io_device *dev)
 	return container_of(dev, struct vgic_io_device, dev);
 }
 
-static bool check_region(const struct vgic_register_region *region,
+static bool check_region(const struct kvm *kvm,
+			 const struct vgic_register_region *region,
 			 gpa_t addr, int len)
 {
-	if ((region->access_flags & VGIC_ACCESS_8bit) && len == 1)
-		return true;
-	if ((region->access_flags & VGIC_ACCESS_32bit) &&
-	    len == sizeof(u32) && !(addr & 3))
-		return true;
-	if ((region->access_flags & VGIC_ACCESS_64bit) &&
-	    len == sizeof(u64) && !(addr & 7))
-		return true;
+	int flags, nr_irqs = kvm->arch.vgic.nr_spis + VGIC_NR_PRIVATE_IRQS;
+
+	switch (len) {
+	case sizeof(u8):
+		flags = VGIC_ACCESS_8bit;
+		break;
+	case sizeof(u32):
+		flags = VGIC_ACCESS_32bit;
+		break;
+	case sizeof(u64):
+		flags = VGIC_ACCESS_64bit;
+		break;
+	default:
+		return false;
+	}
+
+	if ((region->access_flags & flags) && IS_ALIGNED(addr, len)) {
+		if (!region->bits_per_irq)
+			return true;
+
+		/* Do we access a non-allocated IRQ? */
+		return VGIC_ADDR_TO_INTID(addr, region->bits_per_irq) < nr_irqs;
+	}
 
 	return false;
 }
@@ -477,7 +493,7 @@ static int dispatch_mmio_read(struct kvm_vcpu *vcpu, struct kvm_io_device *dev,
 
 	region = vgic_find_mmio_region(iodev->regions, iodev->nr_regions,
 				       addr - iodev->base_addr);
-	if (!region || !check_region(region, addr, len)) {
+	if (!region || !check_region(vcpu->kvm, region, addr, len)) {
 		memset(val, 0, len);
 		return 0;
 	}
@@ -510,10 +526,7 @@ static int dispatch_mmio_write(struct kvm_vcpu *vcpu, struct kvm_io_device *dev,
 
 	region = vgic_find_mmio_region(iodev->regions, iodev->nr_regions,
 				       addr - iodev->base_addr);
-	if (!region)
-		return 0;
-
-	if (!check_region(region, addr, len))
+	if (!region || !check_region(vcpu->kvm, region, addr, len))
 		return 0;
 
 	switch (iodev->iodev_type) {

virt/kvm/arm/vgic/vgic-mmio.h

@@ -50,15 +50,15 @@ extern struct kvm_io_device_ops kvm_io_gic_ops;
 #define VGIC_ADDR_IRQ_MASK(bits) (((bits) * 1024 / 8) - 1)
 
 /*
- * (addr & mask) gives us the byte offset for the INT ID, so we want to
- * divide this with 'bytes per irq' to get the INT ID, which is given
- * by '(bits) / 8'.  But we do this with fixed-point-arithmetic and
- * take advantage of the fact that division by a fraction equals
- * multiplication with the inverted fraction, and scale up both the
- * numerator and denominator with 8 to support at most 64 bits per IRQ:
+ * (addr & mask) gives us the _byte_ offset for the INT ID.
+ * We multiply this by 8 the get the _bit_ offset, then divide this by
+ * the number of bits to learn the actual INT ID.
+ * But instead of a division (which requires a "long long div" implementation),
+ * we shift by the binary logarithm of <bits>.
+ * This assumes that <bits> is a power of two.
  */
 #define VGIC_ADDR_TO_INTID(addr, bits)  (((addr) & VGIC_ADDR_IRQ_MASK(bits)) * \
-					64 / (bits) / 8)
+					8 >> ilog2(bits))
 
 /*
  * Some VGIC registers store per-IRQ information, with a different number

virt/kvm/arm/vgic/vgic.c

@@ -273,6 +273,18 @@ bool vgic_queue_irq_unlock(struct kvm *kvm, struct vgic_irq *irq)
 		 * no more work for us to do.
 		 */
 		spin_unlock(&irq->irq_lock);
+
+		/*
+		 * We have to kick the VCPU here, because we could be
+		 * queueing an edge-triggered interrupt for which we
+		 * get no EOI maintenance interrupt. In that case,
+		 * while the IRQ is already on the VCPU's AP list, the
+		 * VCPU could have EOI'ed the original interrupt and
+		 * won't see this one until it exits for some other
+		 * reason.
+		 */
+		if (vcpu)
+			kvm_vcpu_kick(vcpu);
 		return false;
 	}