Merge branch 'gpc-fixes' of git://git.infradead.org/users/dwmw2/linux into HEAD

Pull Xen-for-KVM changes from David Woodhouse:

* add support for 32-bit guests in SCHEDOP_poll

* the rest of the gfn-to-pfn cache API cleanup

"I still haven't reinstated the last of those patches to make gpc->len
immutable."

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

arch/x86/kvm/x86.c

@@ -2311,13 +2311,11 @@ static void kvm_write_system_time(struct kvm_vcpu *vcpu, gpa_t system_time,
 	kvm_make_request(KVM_REQ_GLOBAL_CLOCK_UPDATE, vcpu);
 
 	/* we verify if the enable bit is set... */
-	if (system_time & 1) {
+	if (system_time & 1)
-		kvm_gpc_activate(vcpu->kvm, &vcpu->arch.pv_time, vcpu,
+		kvm_gpc_activate(&vcpu->arch.pv_time, system_time & ~1ULL,
-				 KVM_HOST_USES_PFN, system_time & ~1ULL,
 				 sizeof(struct pvclock_vcpu_time_info));
-	} else {
+	else
-		kvm_gpc_deactivate(vcpu->kvm, &vcpu->arch.pv_time);
+		kvm_gpc_deactivate(&vcpu->arch.pv_time);
-	}
 
 	return;
 }
@@ -3047,12 +3045,10 @@ static void kvm_setup_guest_pvclock(struct kvm_vcpu *v,
 	unsigned long flags;
 
 	read_lock_irqsave(&gpc->lock, flags);
-	while (!kvm_gpc_check(v->kvm, gpc, gpc->gpa,
+	while (!kvm_gpc_check(gpc, offset + sizeof(*guest_hv_clock))) {
-			      offset + sizeof(*guest_hv_clock))) {
 		read_unlock_irqrestore(&gpc->lock, flags);
 
-		if (kvm_gpc_refresh(v->kvm, gpc, gpc->gpa,
+		if (kvm_gpc_refresh(gpc, offset + sizeof(*guest_hv_clock)))
-				    offset + sizeof(*guest_hv_clock)))
 			return;
 
 		read_lock_irqsave(&gpc->lock, flags);
@@ -3401,7 +3397,7 @@ static int kvm_pv_enable_async_pf_int(struct kvm_vcpu *vcpu, u64 data)
 
 static void kvmclock_reset(struct kvm_vcpu *vcpu)
 {
-	kvm_gpc_deactivate(vcpu->kvm, &vcpu->arch.pv_time);
+	kvm_gpc_deactivate(&vcpu->arch.pv_time);
 	vcpu->arch.time = 0;
 }
 
@@ -11559,7 +11555,7 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
 	vcpu->arch.regs_avail = ~0;
 	vcpu->arch.regs_dirty = ~0;
 
-	kvm_gpc_init(&vcpu->arch.pv_time);
+	kvm_gpc_init(&vcpu->arch.pv_time, vcpu->kvm, vcpu, KVM_HOST_USES_PFN);
 
 	if (!irqchip_in_kernel(vcpu->kvm) || kvm_vcpu_is_reset_bsp(vcpu))
 		vcpu->arch.mp_state = KVM_MP_STATE_RUNNABLE;

arch/x86/kvm/xen.c

@@ -42,13 +42,12 @@ static int kvm_xen_shared_info_init(struct kvm *kvm, gfn_t gfn)
 	int idx = srcu_read_lock(&kvm->srcu);
 
 	if (gfn == GPA_INVALID) {
-		kvm_gpc_deactivate(kvm, gpc);
+		kvm_gpc_deactivate(gpc);
 		goto out;
 	}
 
 	do {
-		ret = kvm_gpc_activate(kvm, gpc, NULL, KVM_HOST_USES_PFN, gpa,
+		ret = kvm_gpc_activate(gpc, gpa, PAGE_SIZE);
-				       PAGE_SIZE);
 		if (ret)
 			goto out;
 
@@ -273,14 +272,14 @@ static void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, bool atomic)
 	 * gfn_to_pfn caches that cover the region.
 	 */
 	read_lock_irqsave(&gpc1->lock, flags);
-	while (!kvm_gpc_check(v->kvm, gpc1, gpc1->gpa, user_len1)) {
+	while (!kvm_gpc_check(gpc1, user_len1)) {
 		read_unlock_irqrestore(&gpc1->lock, flags);
 
 		/* When invoked from kvm_sched_out() we cannot sleep */
 		if (atomic)
 			return;
 
-		if (kvm_gpc_refresh(v->kvm, gpc1, gpc1->gpa, user_len1))
+		if (kvm_gpc_refresh(gpc1, user_len1))
 			return;
 
 		read_lock_irqsave(&gpc1->lock, flags);
@@ -309,7 +308,7 @@ static void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, bool atomic)
 		 */
 		read_lock(&gpc2->lock);
 
-		if (!kvm_gpc_check(v->kvm, gpc2, gpc2->gpa, user_len2)) {
+		if (!kvm_gpc_check(gpc2, user_len2)) {
 			read_unlock(&gpc2->lock);
 			read_unlock_irqrestore(&gpc1->lock, flags);
 
@@ -323,8 +322,8 @@ static void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, bool atomic)
 			 * to the second page now because the guest changed to
 			 * 64-bit mode, the second GPC won't have been set up.
 			 */
-			if (kvm_gpc_activate(v->kvm, gpc2, NULL, KVM_HOST_USES_PFN,
+			if (kvm_gpc_activate(gpc2, gpc1->gpa + user_len1,
-					     gpc1->gpa + user_len1, user_len2))
+					     user_len2))
 				return;
 
 			/*
@@ -489,12 +488,10 @@ void kvm_xen_inject_pending_events(struct kvm_vcpu *v)
 	 * little more honest about it.
 	 */
 	read_lock_irqsave(&gpc->lock, flags);
-	while (!kvm_gpc_check(v->kvm, gpc, gpc->gpa,
+	while (!kvm_gpc_check(gpc, sizeof(struct vcpu_info))) {
-			      sizeof(struct vcpu_info))) {
 		read_unlock_irqrestore(&gpc->lock, flags);
 
-		if (kvm_gpc_refresh(v->kvm, gpc, gpc->gpa,
+		if (kvm_gpc_refresh(gpc, sizeof(struct vcpu_info)))
-				    sizeof(struct vcpu_info)))
 			return;
 
 		read_lock_irqsave(&gpc->lock, flags);
@@ -554,8 +551,7 @@ int __kvm_xen_has_interrupt(struct kvm_vcpu *v)
 		     sizeof_field(struct compat_vcpu_info, evtchn_upcall_pending));
 
 	read_lock_irqsave(&gpc->lock, flags);
-	while (!kvm_gpc_check(v->kvm, gpc, gpc->gpa,
+	while (!kvm_gpc_check(gpc, sizeof(struct vcpu_info))) {
-			      sizeof(struct vcpu_info))) {
 		read_unlock_irqrestore(&gpc->lock, flags);
 
 		/*
@@ -569,8 +565,7 @@ int __kvm_xen_has_interrupt(struct kvm_vcpu *v)
 		if (in_atomic() || !task_is_running(current))
 			return 1;
 
-		if (kvm_gpc_refresh(v->kvm, gpc, gpc->gpa,
+		if (kvm_gpc_refresh(gpc, sizeof(struct vcpu_info))) {
-				    sizeof(struct vcpu_info))) {
 			/*
 			 * If this failed, userspace has screwed up the
 			 * vcpu_info mapping. No interrupts for you.
@@ -711,15 +706,13 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 			     offsetof(struct compat_vcpu_info, time));
 
 		if (data->u.gpa == GPA_INVALID) {
-			kvm_gpc_deactivate(vcpu->kvm, &vcpu->arch.xen.vcpu_info_cache);
+			kvm_gpc_deactivate(&vcpu->arch.xen.vcpu_info_cache);
 			r = 0;
 			break;
 		}
 
-		r = kvm_gpc_activate(vcpu->kvm,
+		r = kvm_gpc_activate(&vcpu->arch.xen.vcpu_info_cache,
-				     &vcpu->arch.xen.vcpu_info_cache, NULL,
+				     data->u.gpa, sizeof(struct vcpu_info));
-				     KVM_HOST_USES_PFN, data->u.gpa,
-				     sizeof(struct vcpu_info));
 		if (!r)
 			kvm_make_request(KVM_REQ_CLOCK_UPDATE, vcpu);
 
@@ -727,15 +720,13 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 
 	case KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO:
 		if (data->u.gpa == GPA_INVALID) {
-			kvm_gpc_deactivate(vcpu->kvm,
+			kvm_gpc_deactivate(&vcpu->arch.xen.vcpu_time_info_cache);
-					   &vcpu->arch.xen.vcpu_time_info_cache);
 			r = 0;
 			break;
 		}
 
-		r = kvm_gpc_activate(vcpu->kvm,
+		r = kvm_gpc_activate(&vcpu->arch.xen.vcpu_time_info_cache,
-				     &vcpu->arch.xen.vcpu_time_info_cache,
+				     data->u.gpa,
-				     NULL, KVM_HOST_USES_PFN, data->u.gpa,
 				     sizeof(struct pvclock_vcpu_time_info));
 		if (!r)
 			kvm_make_request(KVM_REQ_CLOCK_UPDATE, vcpu);
@@ -751,10 +742,8 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 		if (data->u.gpa == GPA_INVALID) {
 			r = 0;
 		deactivate_out:
-			kvm_gpc_deactivate(vcpu->kvm,
+			kvm_gpc_deactivate(&vcpu->arch.xen.runstate_cache);
-					   &vcpu->arch.xen.runstate_cache);
+			kvm_gpc_deactivate(&vcpu->arch.xen.runstate2_cache);
-			kvm_gpc_deactivate(vcpu->kvm,
-					   &vcpu->arch.xen.runstate2_cache);
 			break;
 		}
 
@@ -770,20 +759,18 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data)
 
 		/* How much fits in the (first) page? */
 		sz1 = PAGE_SIZE - (data->u.gpa & ~PAGE_MASK);
-		r = kvm_gpc_activate(vcpu->kvm, &vcpu->arch.xen.runstate_cache,
+		r = kvm_gpc_activate(&vcpu->arch.xen.runstate_cache,
-				     NULL, KVM_HOST_USES_PFN, data->u.gpa, sz1);
+				     data->u.gpa, sz1);
 		if (r)
 			goto deactivate_out;
 
 		/* Either map the second page, or deactivate the second GPC */
 		if (sz1 >= sz) {
-			kvm_gpc_deactivate(vcpu->kvm,
+			kvm_gpc_deactivate(&vcpu->arch.xen.runstate2_cache);
-					   &vcpu->arch.xen.runstate2_cache);
 		} else {
 			sz2 = sz - sz1;
 			BUG_ON((data->u.gpa + sz1) & ~PAGE_MASK);
-			r = kvm_gpc_activate(vcpu->kvm, &vcpu->arch.xen.runstate2_cache,
+			r = kvm_gpc_activate(&vcpu->arch.xen.runstate2_cache,
-					     NULL, KVM_HOST_USES_PFN,
 					     data->u.gpa + sz1, sz2);
 			if (r)
 				goto deactivate_out;
@@ -1167,7 +1154,7 @@ static bool wait_pending_event(struct kvm_vcpu *vcpu, int nr_ports,
 
 	idx = srcu_read_lock(&kvm->srcu);
 	read_lock_irqsave(&gpc->lock, flags);
-	if (!kvm_gpc_check(kvm, gpc, gpc->gpa, PAGE_SIZE))
+	if (!kvm_gpc_check(gpc, PAGE_SIZE))
 		goto out_rcu;
 
 	ret = false;
@@ -1201,20 +1188,45 @@ static bool kvm_xen_schedop_poll(struct kvm_vcpu *vcpu, bool longmode,
 	evtchn_port_t port, *ports;
 	gpa_t gpa;
 
-	if (!longmode || !lapic_in_kernel(vcpu) ||
+	if (!lapic_in_kernel(vcpu) ||
 	    !(vcpu->kvm->arch.xen_hvm_config.flags & KVM_XEN_HVM_CONFIG_EVTCHN_SEND))
 		return false;
 
 	idx = srcu_read_lock(&vcpu->kvm->srcu);
 	gpa = kvm_mmu_gva_to_gpa_system(vcpu, param, NULL);
 	srcu_read_unlock(&vcpu->kvm->srcu, idx);
-
+	if (!gpa) {
-	if (!gpa || kvm_vcpu_read_guest(vcpu, gpa, &sched_poll,
-					sizeof(sched_poll))) {
 		*r = -EFAULT;
 		return true;
 	}
 
+	if (IS_ENABLED(CONFIG_64BIT) && !longmode) {
+		struct compat_sched_poll sp32;
+
+		/* Sanity check that the compat struct definition is correct */
+		BUILD_BUG_ON(sizeof(sp32) != 16);
+
+		if (kvm_vcpu_read_guest(vcpu, gpa, &sp32, sizeof(sp32))) {
+			*r = -EFAULT;
+			return true;
+		}
+
+		/*
+		 * This is a 32-bit pointer to an array of evtchn_port_t which
+		 * are uint32_t, so once it's converted no further compat
+		 * handling is needed.
+		 */
+		sched_poll.ports = (void *)(unsigned long)(sp32.ports);
+		sched_poll.nr_ports = sp32.nr_ports;
+		sched_poll.timeout = sp32.timeout;
+	} else {
+		if (kvm_vcpu_read_guest(vcpu, gpa, &sched_poll,
+					sizeof(sched_poll))) {
+			*r = -EFAULT;
+			return true;
+		}
+	}
+
 	if (unlikely(sched_poll.nr_ports > 1)) {
 		/* Xen (unofficially) limits number of pollers to 128 */
 		if (sched_poll.nr_ports > 128) {
@@ -1564,7 +1576,7 @@ int kvm_xen_set_evtchn_fast(struct kvm_xen_evtchn *xe, struct kvm *kvm)
 	idx = srcu_read_lock(&kvm->srcu);
 
 	read_lock_irqsave(&gpc->lock, flags);
-	if (!kvm_gpc_check(kvm, gpc, gpc->gpa, PAGE_SIZE))
+	if (!kvm_gpc_check(gpc, PAGE_SIZE))
 		goto out_rcu;
 
 	if (IS_ENABLED(CONFIG_64BIT) && kvm->arch.xen.long_mode) {
@@ -1598,7 +1610,7 @@ int kvm_xen_set_evtchn_fast(struct kvm_xen_evtchn *xe, struct kvm *kvm)
 		gpc = &vcpu->arch.xen.vcpu_info_cache;
 
 		read_lock_irqsave(&gpc->lock, flags);
-		if (!kvm_gpc_check(kvm, gpc, gpc->gpa, sizeof(struct vcpu_info))) {
+		if (!kvm_gpc_check(gpc, sizeof(struct vcpu_info))) {
 			/*
 			 * Could not access the vcpu_info. Set the bit in-kernel
 			 * and prod the vCPU to deliver it for itself.
@@ -1696,7 +1708,7 @@ static int kvm_xen_set_evtchn(struct kvm_xen_evtchn *xe, struct kvm *kvm)
 			break;
 
 		idx = srcu_read_lock(&kvm->srcu);
-		rc = kvm_gpc_refresh(kvm, gpc, gpc->gpa, PAGE_SIZE);
+		rc = kvm_gpc_refresh(gpc, PAGE_SIZE);
 		srcu_read_unlock(&kvm->srcu, idx);
 	} while(!rc);
 
@@ -2026,10 +2038,14 @@ void kvm_xen_init_vcpu(struct kvm_vcpu *vcpu)
 
 	timer_setup(&vcpu->arch.xen.poll_timer, cancel_evtchn_poll, 0);
 
-	kvm_gpc_init(&vcpu->arch.xen.runstate_cache);
+	kvm_gpc_init(&vcpu->arch.xen.runstate_cache, vcpu->kvm, NULL,
-	kvm_gpc_init(&vcpu->arch.xen.runstate2_cache);
+		     KVM_HOST_USES_PFN);
-	kvm_gpc_init(&vcpu->arch.xen.vcpu_info_cache);
+	kvm_gpc_init(&vcpu->arch.xen.runstate2_cache, vcpu->kvm, NULL,
-	kvm_gpc_init(&vcpu->arch.xen.vcpu_time_info_cache);
+		     KVM_HOST_USES_PFN);
+	kvm_gpc_init(&vcpu->arch.xen.vcpu_info_cache, vcpu->kvm, NULL,
+		     KVM_HOST_USES_PFN);
+	kvm_gpc_init(&vcpu->arch.xen.vcpu_time_info_cache, vcpu->kvm, NULL,
+		     KVM_HOST_USES_PFN);
 }
 
 void kvm_xen_destroy_vcpu(struct kvm_vcpu *vcpu)
@@ -2037,10 +2053,10 @@ void kvm_xen_destroy_vcpu(struct kvm_vcpu *vcpu)
 	if (kvm_xen_timer_enabled(vcpu))
 		kvm_xen_stop_timer(vcpu);
 
-	kvm_gpc_deactivate(vcpu->kvm, &vcpu->arch.xen.runstate_cache);
+	kvm_gpc_deactivate(&vcpu->arch.xen.runstate_cache);
-	kvm_gpc_deactivate(vcpu->kvm, &vcpu->arch.xen.runstate2_cache);
+	kvm_gpc_deactivate(&vcpu->arch.xen.runstate2_cache);
-	kvm_gpc_deactivate(vcpu->kvm, &vcpu->arch.xen.vcpu_info_cache);
+	kvm_gpc_deactivate(&vcpu->arch.xen.vcpu_info_cache);
-	kvm_gpc_deactivate(vcpu->kvm, &vcpu->arch.xen.vcpu_time_info_cache);
+	kvm_gpc_deactivate(&vcpu->arch.xen.vcpu_time_info_cache);
 
 	del_timer_sync(&vcpu->arch.xen.poll_timer);
 }
@@ -2048,7 +2064,7 @@ void kvm_xen_destroy_vcpu(struct kvm_vcpu *vcpu)
 void kvm_xen_init_vm(struct kvm *kvm)
 {
 	idr_init(&kvm->arch.xen.evtchn_ports);
-	kvm_gpc_init(&kvm->arch.xen.shinfo_cache);
+	kvm_gpc_init(&kvm->arch.xen.shinfo_cache, kvm, NULL, KVM_HOST_USES_PFN);
 }
 
 void kvm_xen_destroy_vm(struct kvm *kvm)
@@ -2056,7 +2072,7 @@ void kvm_xen_destroy_vm(struct kvm *kvm)
 	struct evtchnfd *evtchnfd;
 	int i;
 
-	kvm_gpc_deactivate(kvm, &kvm->arch.xen.shinfo_cache);
+	kvm_gpc_deactivate(&kvm->arch.xen.shinfo_cache);
 
 	idr_for_each_entry(&kvm->arch.xen.evtchn_ports, evtchnfd, i) {
 		if (!evtchnfd->deliver.port.port)

arch/x86/kvm/xen.h

@@ -207,4 +207,11 @@ struct compat_vcpu_runstate_info {
     uint64_t time[4];
 } __attribute__((packed));
 
+struct compat_sched_poll {
+	/* This is actually a guest virtual address which points to ports. */
+	uint32_t ports;
+	unsigned int nr_ports;
+	uint64_t timeout;
+};
+
 #endif /* __ARCH_X86_KVM_XEN_H__ */

include/linux/kvm_host.h

@@ -1260,18 +1260,7 @@ void kvm_vcpu_mark_page_dirty(struct kvm_vcpu *vcpu, gfn_t gfn);
  * kvm_gpc_init - initialize gfn_to_pfn_cache.
  *
  * @gpc:	   struct gfn_to_pfn_cache object.
- *
- * This sets up a gfn_to_pfn_cache by initializing locks.  Note, the cache must
- * be zero-allocated (or zeroed by the caller before init).
- */
-void kvm_gpc_init(struct gfn_to_pfn_cache *gpc);
-
-/**
- * kvm_gpc_activate - prepare a cached kernel mapping and HPA for a given guest
- *                    physical address.
- *
  * @kvm:	   pointer to kvm instance.
- * @gpc:	   struct gfn_to_pfn_cache object.
  * @vcpu:	   vCPU to be used for marking pages dirty and to be woken on
  *		   invalidation.
  * @usage:	   indicates if the resulting host physical PFN is used while
@@ -1280,27 +1269,36 @@ void kvm_gpc_init(struct gfn_to_pfn_cache *gpc);
  *		   changes!---will also force @vcpu to exit the guest and
  *		   refresh the cache); and/or if the PFN used directly
  *		   by KVM (and thus needs a kernel virtual mapping).
+ *
+ * This sets up a gfn_to_pfn_cache by initializing locks and assigning the
+ * immutable attributes.  Note, the cache must be zero-allocated (or zeroed by
+ * the caller before init).
+ */
+void kvm_gpc_init(struct gfn_to_pfn_cache *gpc, struct kvm *kvm,
+		  struct kvm_vcpu *vcpu, enum pfn_cache_usage usage);
+
+/**
+ * kvm_gpc_activate - prepare a cached kernel mapping and HPA for a given guest
+ *                    physical address.
+ *
+ * @gpc:	   struct gfn_to_pfn_cache object.
  * @gpa:	   guest physical address to map.
  * @len:	   sanity check; the range being access must fit a single page.
  *
  * @return:	   0 for success.
  *		   -EINVAL for a mapping which would cross a page boundary.
- *                 -EFAULT for an untranslatable guest physical address.
+ *		   -EFAULT for an untranslatable guest physical address.
  *
- * This primes a gfn_to_pfn_cache and links it into the @kvm's list for
+ * This primes a gfn_to_pfn_cache and links it into the @gpc->kvm's list for
  * invalidations to be processed.  Callers are required to use kvm_gpc_check()
  * to ensure that the cache is valid before accessing the target page.
  */
-int kvm_gpc_activate(struct kvm *kvm, struct gfn_to_pfn_cache *gpc,
+int kvm_gpc_activate(struct gfn_to_pfn_cache *gpc, gpa_t gpa, unsigned long len);
-		     struct kvm_vcpu *vcpu, enum pfn_cache_usage usage,
-		     gpa_t gpa, unsigned long len);
 
 /**
  * kvm_gpc_check - check validity of a gfn_to_pfn_cache.
  *
- * @kvm:	   pointer to kvm instance.
  * @gpc:	   struct gfn_to_pfn_cache object.
- * @gpa:	   current guest physical address to map.
  * @len:	   sanity check; the range being access must fit a single page.
  *
  * @return:	   %true if the cache is still valid and the address matches.
@@ -1313,52 +1311,35 @@ int kvm_gpc_activate(struct kvm *kvm, struct gfn_to_pfn_cache *gpc,
  * Callers in IN_GUEST_MODE may do so without locking, although they should
  * still hold a read lock on kvm->scru for the memslot checks.
  */
-bool kvm_gpc_check(struct kvm *kvm, struct gfn_to_pfn_cache *gpc, gpa_t gpa,
+bool kvm_gpc_check(struct gfn_to_pfn_cache *gpc, unsigned long len);
-		   unsigned long len);
 
 /**
  * kvm_gpc_refresh - update a previously initialized cache.
  *
- * @kvm:	   pointer to kvm instance.
  * @gpc:	   struct gfn_to_pfn_cache object.
- * @gpa:	   updated guest physical address to map.
  * @len:	   sanity check; the range being access must fit a single page.
  *
  * @return:	   0 for success.
  *		   -EINVAL for a mapping which would cross a page boundary.
- *                 -EFAULT for an untranslatable guest physical address.
+ *		   -EFAULT for an untranslatable guest physical address.
  *
  * This will attempt to refresh a gfn_to_pfn_cache. Note that a successful
- * returm from this function does not mean the page can be immediately
+ * return from this function does not mean the page can be immediately
  * accessed because it may have raced with an invalidation. Callers must
  * still lock and check the cache status, as this function does not return
  * with the lock still held to permit access.
  */
-int kvm_gpc_refresh(struct kvm *kvm, struct gfn_to_pfn_cache *gpc, gpa_t gpa,
+int kvm_gpc_refresh(struct gfn_to_pfn_cache *gpc, unsigned long len);
-		    unsigned long len);
-
-/**
- * kvm_gpc_unmap - temporarily unmap a gfn_to_pfn_cache.
- *
- * @kvm:	   pointer to kvm instance.
- * @gpc:	   struct gfn_to_pfn_cache object.
- *
- * This unmaps the referenced page. The cache is left in the invalid state
- * but at least the mapping from GPA to userspace HVA will remain cached
- * and can be reused on a subsequent refresh.
- */
-void kvm_gpc_unmap(struct kvm *kvm, struct gfn_to_pfn_cache *gpc);
 
 /**
  * kvm_gpc_deactivate - deactivate and unlink a gfn_to_pfn_cache.
  *
- * @kvm:	   pointer to kvm instance.
  * @gpc:	   struct gfn_to_pfn_cache object.
  *
- * This removes a cache from the @kvm's list to be processed on MMU notifier
+ * This removes a cache from the VM's list to be processed on MMU notifier
  * invocation.
  */
-void kvm_gpc_deactivate(struct kvm *kvm, struct gfn_to_pfn_cache *gpc);
+void kvm_gpc_deactivate(struct gfn_to_pfn_cache *gpc);
 
 void kvm_sigset_activate(struct kvm_vcpu *vcpu);
 void kvm_sigset_deactivate(struct kvm_vcpu *vcpu);

include/linux/kvm_types.h

@@ -67,6 +67,7 @@ struct gfn_to_pfn_cache {
 	gpa_t gpa;
 	unsigned long uhva;
 	struct kvm_memory_slot *memslot;
+	struct kvm *kvm;
 	struct kvm_vcpu *vcpu;
 	struct list_head list;
 	rwlock_t lock;

virt/kvm/pfncache.c

@@ -76,19 +76,17 @@ void gfn_to_pfn_cache_invalidate_start(struct kvm *kvm, unsigned long start,
 	}
 }
 
-bool kvm_gpc_check(struct kvm *kvm, struct gfn_to_pfn_cache *gpc, gpa_t gpa,
+bool kvm_gpc_check(struct gfn_to_pfn_cache *gpc, unsigned long len)
-		   unsigned long len)
 {
-	struct kvm_memslots *slots = kvm_memslots(kvm);
+	struct kvm_memslots *slots = kvm_memslots(gpc->kvm);
 
 	if (!gpc->active)
 		return false;
 
-	if ((gpa & ~PAGE_MASK) + len > PAGE_SIZE)
+	if ((gpc->gpa & ~PAGE_MASK) + len > PAGE_SIZE)
 		return false;
 
-	if (gpc->gpa != gpa || gpc->generation != slots->generation ||
+	if (gpc->generation != slots->generation || kvm_is_error_hva(gpc->uhva))
-	    kvm_is_error_hva(gpc->uhva))
 		return false;
 
 	if (!gpc->valid)
@@ -139,7 +137,7 @@ static inline bool mmu_notifier_retry_cache(struct kvm *kvm, unsigned long mmu_s
 	return kvm->mmu_invalidate_seq != mmu_seq;
 }
 
-static kvm_pfn_t hva_to_pfn_retry(struct kvm *kvm, struct gfn_to_pfn_cache *gpc)
+static kvm_pfn_t hva_to_pfn_retry(struct gfn_to_pfn_cache *gpc)
 {
 	/* Note, the new page offset may be different than the old! */
 	void *old_khva = gpc->khva - offset_in_page(gpc->khva);
@@ -159,7 +157,7 @@ static kvm_pfn_t hva_to_pfn_retry(struct kvm *kvm, struct gfn_to_pfn_cache *gpc)
 	gpc->valid = false;
 
 	do {
-		mmu_seq = kvm->mmu_invalidate_seq;
+		mmu_seq = gpc->kvm->mmu_invalidate_seq;
 		smp_rmb();
 
 		write_unlock_irq(&gpc->lock);
@@ -217,7 +215,7 @@ static kvm_pfn_t hva_to_pfn_retry(struct kvm *kvm, struct gfn_to_pfn_cache *gpc)
 		 * attempting to refresh.
 		 */
 		WARN_ON_ONCE(gpc->valid);
-	} while (mmu_notifier_retry_cache(kvm, mmu_seq));
+	} while (mmu_notifier_retry_cache(gpc->kvm, mmu_seq));
 
 	gpc->valid = true;
 	gpc->pfn = new_pfn;
@@ -238,10 +236,10 @@ out_error:
 	return -EFAULT;
 }
 
-int kvm_gpc_refresh(struct kvm *kvm, struct gfn_to_pfn_cache *gpc, gpa_t gpa,
+static int __kvm_gpc_refresh(struct gfn_to_pfn_cache *gpc, gpa_t gpa,
-		    unsigned long len)
+			     unsigned long len)
 {
-	struct kvm_memslots *slots = kvm_memslots(kvm);
+	struct kvm_memslots *slots = kvm_memslots(gpc->kvm);
 	unsigned long page_offset = gpa & ~PAGE_MASK;
 	bool unmap_old = false;
 	unsigned long old_uhva;
@@ -295,7 +293,7 @@ int kvm_gpc_refresh(struct kvm *kvm, struct gfn_to_pfn_cache *gpc, gpa_t gpa,
 	 * drop the lock and do the HVA to PFN lookup again.
 	 */
 	if (!gpc->valid || old_uhva != gpc->uhva) {
-		ret = hva_to_pfn_retry(kvm, gpc);
+		ret = hva_to_pfn_retry(gpc);
 	} else {
 		/*
 		 * If the HVA→PFN mapping was already valid, don't unmap it.
@@ -303,9 +301,8 @@ int kvm_gpc_refresh(struct kvm *kvm, struct gfn_to_pfn_cache *gpc, gpa_t gpa,
 		 * may have changed.
 		 */
 		gpc->khva = old_khva + page_offset;
-		old_pfn = KVM_PFN_ERR_FAULT;
-		old_khva = NULL;
 		ret = 0;
+		goto out_unlock;
 	}
 
  out:
@@ -333,55 +330,37 @@ out_unlock:
 
 	return ret;
 }
+
+int kvm_gpc_refresh(struct gfn_to_pfn_cache *gpc, unsigned long len)
+{
+	return __kvm_gpc_refresh(gpc, gpc->gpa, len);
+}
 EXPORT_SYMBOL_GPL(kvm_gpc_refresh);
 
-void kvm_gpc_unmap(struct kvm *kvm, struct gfn_to_pfn_cache *gpc)
+void kvm_gpc_init(struct gfn_to_pfn_cache *gpc, struct kvm *kvm,
+		  struct kvm_vcpu *vcpu, enum pfn_cache_usage usage)
 {
-	void *old_khva;
+	WARN_ON_ONCE(!usage || (usage & KVM_GUEST_AND_HOST_USE_PFN) != usage);
-	kvm_pfn_t old_pfn;
+	WARN_ON_ONCE((usage & KVM_GUEST_USES_PFN) && !vcpu);
 
-	mutex_lock(&gpc->refresh_lock);
-	write_lock_irq(&gpc->lock);
-
-	gpc->valid = false;
-
-	old_khva = gpc->khva - offset_in_page(gpc->khva);
-	old_pfn = gpc->pfn;
-
-	/*
-	 * We can leave the GPA → uHVA map cache intact but the PFN
-	 * lookup will need to be redone even for the same page.
-	 */
-	gpc->khva = NULL;
-	gpc->pfn = KVM_PFN_ERR_FAULT;
-
-	write_unlock_irq(&gpc->lock);
-	mutex_unlock(&gpc->refresh_lock);
-
-	gpc_unmap_khva(old_pfn, old_khva);
-}
-EXPORT_SYMBOL_GPL(kvm_gpc_unmap);
-
-void kvm_gpc_init(struct gfn_to_pfn_cache *gpc)
-{
 	rwlock_init(&gpc->lock);
 	mutex_init(&gpc->refresh_lock);
+
+	gpc->kvm = kvm;
+	gpc->vcpu = vcpu;
+	gpc->usage = usage;
+	gpc->pfn = KVM_PFN_ERR_FAULT;
+	gpc->uhva = KVM_HVA_ERR_BAD;
 }
 EXPORT_SYMBOL_GPL(kvm_gpc_init);
 
-int kvm_gpc_activate(struct kvm *kvm, struct gfn_to_pfn_cache *gpc,
+int kvm_gpc_activate(struct gfn_to_pfn_cache *gpc, gpa_t gpa, unsigned long len)
-		     struct kvm_vcpu *vcpu, enum pfn_cache_usage usage,
-		     gpa_t gpa, unsigned long len)
 {
-	WARN_ON_ONCE(!usage || (usage & KVM_GUEST_AND_HOST_USE_PFN) != usage);
+	struct kvm *kvm = gpc->kvm;
 
 	if (!gpc->active) {
-		gpc->khva = NULL;
+		if (KVM_BUG_ON(gpc->valid, kvm))
-		gpc->pfn = KVM_PFN_ERR_FAULT;
+			return -EIO;
-		gpc->uhva = KVM_HVA_ERR_BAD;
-		gpc->vcpu = vcpu;
-		gpc->usage = usage;
-		gpc->valid = false;
 
 		spin_lock(&kvm->gpc_lock);
 		list_add(&gpc->list, &kvm->gpc_list);
@@ -396,12 +375,16 @@ int kvm_gpc_activate(struct kvm *kvm, struct gfn_to_pfn_cache *gpc,
 		gpc->active = true;
 		write_unlock_irq(&gpc->lock);
 	}
-	return kvm_gpc_refresh(kvm, gpc, gpa, len);
+	return __kvm_gpc_refresh(gpc, gpa, len);
 }
 EXPORT_SYMBOL_GPL(kvm_gpc_activate);
 
-void kvm_gpc_deactivate(struct kvm *kvm, struct gfn_to_pfn_cache *gpc)
+void kvm_gpc_deactivate(struct gfn_to_pfn_cache *gpc)
 {
+	struct kvm *kvm = gpc->kvm;
+	kvm_pfn_t old_pfn;
+	void *old_khva;
+
 	if (gpc->active) {
 		/*
 		 * Deactivate the cache before removing it from the list, KVM
@@ -410,13 +393,26 @@ void kvm_gpc_deactivate(struct kvm *kvm, struct gfn_to_pfn_cache *gpc)
 		 */
 		write_lock_irq(&gpc->lock);
 		gpc->active = false;
+		gpc->valid = false;
+
+		/*
+		 * Leave the GPA => uHVA cache intact, it's protected by the
+		 * memslot generation.  The PFN lookup needs to be redone every
+		 * time as mmu_notifier protection is lost when the cache is
+		 * removed from the VM's gpc_list.
+		 */
+		old_khva = gpc->khva - offset_in_page(gpc->khva);
+		gpc->khva = NULL;
+
+		old_pfn = gpc->pfn;
+		gpc->pfn = KVM_PFN_ERR_FAULT;
 		write_unlock_irq(&gpc->lock);
 
 		spin_lock(&kvm->gpc_lock);
 		list_del(&gpc->list);
 		spin_unlock(&kvm->gpc_lock);
 
-		kvm_gpc_unmap(kvm, gpc);
+		gpc_unmap_khva(old_pfn, old_khva);
 	}
 }
 EXPORT_SYMBOL_GPL(kvm_gpc_deactivate);