KVM: unmap pages from the iommu when slots are removed

commit 32f6daad4651a748a58a3ab6da0611862175722f upstream.

We've been adding new mappings, but not destroying old mappings.
This can lead to a page leak as pages are pinned using
get_user_pages, but only unpinned with put_page if they still
exist in the memslots list on vm shutdown.  A memslot that is
destroyed while an iommu domain is enabled for the guest will
therefore result in an elevated page reference count that is
never cleared.

Additionally, without this fix, the iommu is only programmed
with the first translation for a gpa.  This can result in
peer-to-peer errors if a mapping is destroyed and replaced by a
new mapping at the same gpa as the iommu will still be pointing
to the original, pinned memory address.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

3 files changed, 17 insertions, 6 deletions

diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index d52623199978..35410ef9cf69 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -562,6 +562,7 @@ void kvm_free_irq_source_id(struct kvm *kvm, int irq_source_id);
 
 #ifdef CONFIG_IOMMU_API
 int kvm_iommu_map_pages(struct kvm *kvm, struct kvm_memory_slot *slot);
+void kvm_iommu_unmap_pages(struct kvm *kvm, struct kvm_memory_slot *slot);
 int kvm_iommu_map_guest(struct kvm *kvm);
 int kvm_iommu_unmap_guest(struct kvm *kvm);
 int kvm_assign_device(struct kvm *kvm,
@@ -575,6 +576,11 @@ static inline int kvm_iommu_map_pages(struct kvm *kvm,
 	return 0;
 }
 
+static inline void kvm_iommu_unmap_pages(struct kvm *kvm,
+					 struct kvm_memory_slot *slot)
+{
+}
+
 static inline int kvm_iommu_map_guest(struct kvm *kvm)
 {
 	return -ENODEV;
diff --git a/virt/kvm/iommu.c b/virt/kvm/iommu.c
index a195c07fa829..fd817a2cb240 100644
--- a/virt/kvm/iommu.c
+++ b/virt/kvm/iommu.c
@@ -309,6 +309,11 @@ static void kvm_iommu_put_pages(struct kvm *kvm,
 	}
 }
 
+void kvm_iommu_unmap_pages(struct kvm *kvm, struct kvm_memory_slot *slot)
+{
+	kvm_iommu_put_pages(kvm, slot->base_gfn, slot->npages);
+}
+
 static int kvm_iommu_unmap_memslots(struct kvm *kvm)
 {
 	int i, idx;
@@ -317,10 +322,9 @@ static int kvm_iommu_unmap_memslots(struct kvm *kvm)
 	idx = srcu_read_lock(&kvm->srcu);
 	slots = kvm_memslots(kvm);
 
-	for (i = 0; i < slots->nmemslots; i++) {
-		kvm_iommu_put_pages(kvm, slots->memslots[i].base_gfn,
-				    slots->memslots[i].npages);
-	}
+	for (i = 0; i < slots->nmemslots; i++)
+		kvm_iommu_unmap_pages(kvm, &slots->memslots[i]);
+
 	srcu_read_unlock(&kvm->srcu, idx);
 
 	return 0;
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index d9cfb782cb81..e401c1b118e9 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -802,12 +802,13 @@ skip_lpage:
 	if (r)
 		goto out_free;
 
-	/* map the pages in iommu page table */
+	/* map/unmap the pages in iommu page table */
 	if (npages) {
 		r = kvm_iommu_map_pages(kvm, &new);
 		if (r)
 			goto out_free;
-	}
+	} else
+		kvm_iommu_unmap_pages(kvm, &old);
 
 	r = -ENOMEM;
 	slots = kzalloc(sizeof(struct kvm_memslots), GFP_KERNEL);