KVM: switch vcpu context to use SRCU

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

arch/ia64/kvm/kvm-ia64.c

@@ -636,12 +636,9 @@ static void kvm_vcpu_post_transition(struct kvm_vcpu *vcpu)
 static int __vcpu_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 {
 	union context *host_ctx, *guest_ctx;
-	int r;
+	int r, idx;
 
-	/*
-	 * down_read() may sleep and return with interrupts enabled
-	 */
-	down_read(&vcpu->kvm->slots_lock);
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 
 again:
 	if (signal_pending(current)) {
@@ -663,7 +660,7 @@ static int __vcpu_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 	if (r < 0)
 		goto vcpu_run_fail;
 
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 	kvm_guest_enter();
 
 	/*
@@ -687,7 +684,7 @@ static int __vcpu_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 	kvm_guest_exit();
 	preempt_enable();
 
-	down_read(&vcpu->kvm->slots_lock);
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 
 	r = kvm_handle_exit(kvm_run, vcpu);
 
@@ -697,10 +694,10 @@ static int __vcpu_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 	}
 
 out:
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 	if (r > 0) {
 		kvm_resched(vcpu);
-		down_read(&vcpu->kvm->slots_lock);
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
 		goto again;
 	}
 

arch/s390/kvm/kvm-s390.h

@@ -67,10 +67,14 @@ static inline long kvm_s390_vcpu_get_memsize(struct kvm_vcpu *vcpu)
 
 static inline void kvm_s390_vcpu_set_mem(struct kvm_vcpu *vcpu)
 {
+	int idx;
 	struct kvm_memory_slot *mem;
+	struct kvm_memslots *memslots;
 
-	down_read(&vcpu->kvm->slots_lock);
-	mem = &vcpu->kvm->memslots[0];
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
+	memslots = rcu_dereference(vcpu->kvm->memslots);
+
+	mem = &memslots->memslots[0];
 
 	vcpu->arch.sie_block->gmsor = mem->userspace_addr;
 	vcpu->arch.sie_block->gmslm =
@@ -78,7 +82,7 @@ static inline void kvm_s390_vcpu_set_mem(struct kvm_vcpu *vcpu)
 		(mem->npages << PAGE_SHIFT) +
 		VIRTIODESCSPACE - 1ul;
 
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 }
 
 /* implemented in priv.c */

arch/x86/kvm/mmu.c

@@ -2933,10 +2933,9 @@ static int mmu_shrink(int nr_to_scan, gfp_t gfp_mask)
 	spin_lock(&kvm_lock);
 
 	list_for_each_entry(kvm, &vm_list, vm_list) {
-		int npages;
+		int npages, idx;
 
-		if (!down_read_trylock(&kvm->slots_lock))
-			continue;
+		idx = srcu_read_lock(&kvm->srcu);
 		spin_lock(&kvm->mmu_lock);
 		npages = kvm->arch.n_alloc_mmu_pages -
 			 kvm->arch.n_free_mmu_pages;
@@ -2949,7 +2948,7 @@ static int mmu_shrink(int nr_to_scan, gfp_t gfp_mask)
 		nr_to_scan--;
 
 		spin_unlock(&kvm->mmu_lock);
-		up_read(&kvm->slots_lock);
+		srcu_read_unlock(&kvm->srcu, idx);
 	}
 	if (kvm_freed)
 		list_move_tail(&kvm_freed->vm_list, &vm_list);

arch/x86/kvm/vmx.c

@@ -2478,10 +2478,10 @@ static int vmx_vcpu_reset(struct kvm_vcpu *vcpu)
 {
 	struct vcpu_vmx *vmx = to_vmx(vcpu);
 	u64 msr;
-	int ret;
+	int ret, idx;
 
 	vcpu->arch.regs_avail = ~((1 << VCPU_REGS_RIP) | (1 << VCPU_REGS_RSP));
-	down_read(&vcpu->kvm->slots_lock);
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 	if (!init_rmode(vmx->vcpu.kvm)) {
 		ret = -ENOMEM;
 		goto out;
@@ -2589,7 +2589,7 @@ static int vmx_vcpu_reset(struct kvm_vcpu *vcpu)
 	vmx->emulation_required = 0;
 
 out:
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 	return ret;
 }
 

arch/x86/kvm/x86.c

@@ -1306,15 +1306,15 @@ static int __msr_io(struct kvm_vcpu *vcpu, struct kvm_msrs *msrs,
 		    int (*do_msr)(struct kvm_vcpu *vcpu,
 				  unsigned index, u64 *data))
 {
-	int i;
+	int i, idx;
 
 	vcpu_load(vcpu);
 
-	down_read(&vcpu->kvm->slots_lock);
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 	for (i = 0; i < msrs->nmsrs; ++i)
 		if (do_msr(vcpu, entries[i].index, &entries[i].data))
 			break;
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 
 	vcpu_put(vcpu);
 
@@ -3900,14 +3900,15 @@ static void vapic_enter(struct kvm_vcpu *vcpu)
 static void vapic_exit(struct kvm_vcpu *vcpu)
 {
 	struct kvm_lapic *apic = vcpu->arch.apic;
+	int idx;
 
 	if (!apic || !apic->vapic_addr)
 		return;
 
-	down_read(&vcpu->kvm->slots_lock);
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 	kvm_release_page_dirty(apic->vapic_page);
 	mark_page_dirty(vcpu->kvm, apic->vapic_addr >> PAGE_SHIFT);
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 }
 
 static void update_cr8_intercept(struct kvm_vcpu *vcpu)
@@ -4036,7 +4037,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
 		kvm_lapic_sync_to_vapic(vcpu);
 	}
 
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
 
 	kvm_guest_enter();
 
@@ -4078,7 +4079,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
 
 	preempt_enable();
 
-	down_read(&vcpu->kvm->slots_lock);
+	vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
 
 	/*
 	 * Profile KVM exit RIPs:
@@ -4100,6 +4101,7 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
 static int __vcpu_run(struct kvm_vcpu *vcpu)
 {
 	int r;
+	struct kvm *kvm = vcpu->kvm;
 
 	if (unlikely(vcpu->arch.mp_state == KVM_MP_STATE_SIPI_RECEIVED)) {
 		pr_debug("vcpu %d received sipi with vector # %x\n",
@@ -4111,7 +4113,7 @@ static int __vcpu_run(struct kvm_vcpu *vcpu)
 		vcpu->arch.mp_state = KVM_MP_STATE_RUNNABLE;
 	}
 
-	down_read(&vcpu->kvm->slots_lock);
+	vcpu->srcu_idx = srcu_read_lock(&kvm->srcu);
 	vapic_enter(vcpu);
 
 	r = 1;
@@ -4119,9 +4121,9 @@ static int __vcpu_run(struct kvm_vcpu *vcpu)
 		if (vcpu->arch.mp_state == KVM_MP_STATE_RUNNABLE)
 			r = vcpu_enter_guest(vcpu);
 		else {
-			up_read(&vcpu->kvm->slots_lock);
+			srcu_read_unlock(&kvm->srcu, vcpu->srcu_idx);
 			kvm_vcpu_block(vcpu);
-			down_read(&vcpu->kvm->slots_lock);
+			vcpu->srcu_idx = srcu_read_lock(&kvm->srcu);
 			if (test_and_clear_bit(KVM_REQ_UNHALT, &vcpu->requests))
 			{
 				switch(vcpu->arch.mp_state) {
@@ -4156,13 +4158,13 @@ static int __vcpu_run(struct kvm_vcpu *vcpu)
 			++vcpu->stat.signal_exits;
 		}
 		if (need_resched()) {
-			up_read(&vcpu->kvm->slots_lock);
+			srcu_read_unlock(&kvm->srcu, vcpu->srcu_idx);
 			kvm_resched(vcpu);
-			down_read(&vcpu->kvm->slots_lock);
+			vcpu->srcu_idx = srcu_read_lock(&kvm->srcu);
 		}
 	}
 
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&kvm->srcu, vcpu->srcu_idx);
 	post_kvm_run_save(vcpu);
 
 	vapic_exit(vcpu);
@@ -4201,10 +4203,10 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 		vcpu->mmio_read_completed = 1;
 		vcpu->mmio_needed = 0;
 
-		down_read(&vcpu->kvm->slots_lock);
+		vcpu->srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
 		r = emulate_instruction(vcpu, vcpu->arch.mmio_fault_cr2, 0,
 					EMULTYPE_NO_DECODE);
-		up_read(&vcpu->kvm->slots_lock);
+		srcu_read_unlock(&vcpu->kvm->srcu, vcpu->srcu_idx);
 		if (r == EMULATE_DO_MMIO) {
 			/*
 			 * Read-modify-write.  Back to userspace.
@@ -4967,11 +4969,12 @@ int kvm_arch_vcpu_ioctl_translate(struct kvm_vcpu *vcpu,
 {
 	unsigned long vaddr = tr->linear_address;
 	gpa_t gpa;
+	int idx;
 
 	vcpu_load(vcpu);
-	down_read(&vcpu->kvm->slots_lock);
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 	gpa = vcpu->arch.mmu.gva_to_gpa(vcpu, vaddr);
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 	tr->physical_address = gpa;
 	tr->valid = gpa != UNMAPPED_GVA;
 	tr->writeable = 1;
@@ -5223,11 +5226,13 @@ int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu)
 
 void kvm_arch_vcpu_uninit(struct kvm_vcpu *vcpu)
 {
+	int idx;
+
 	kfree(vcpu->arch.mce_banks);
 	kvm_free_lapic(vcpu);
-	down_read(&vcpu->kvm->slots_lock);
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 	kvm_mmu_destroy(vcpu);
-	up_read(&vcpu->kvm->slots_lock);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 	free_page((unsigned long)vcpu->arch.pio_data);
 }
 

include/linux/kvm_host.h

@@ -83,6 +83,8 @@ struct kvm_vcpu {
 	struct kvm_run *run;
 	unsigned long requests;
 	unsigned long guest_debug;
+	int srcu_idx;
+
 	int fpu_active;
 	int guest_fpu_loaded;
 	wait_queue_head_t wq;