KVM: Unmap kernel-allocated memory on slot destruction

kvm_vm_ioctl_set_memory_region() is able to remove memory in addition to
adding it.  Therefore when using kernel swapping support for old userspaces,
we need to munmap the memory if the user request to remove it

Signed-off-by: Izik Eidus <izike@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>

drivers/kvm/kvm.h

@@ -353,6 +353,7 @@ struct kvm_memory_slot {
 	unsigned long *rmap;
 	unsigned long *dirty_bitmap;
 	unsigned long userspace_addr;
+	int user_alloc;
 };
 
 struct kvm {

drivers/kvm/kvm_main.c

@@ -713,6 +713,7 @@ static int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
 
 		memset(new.rmap, 0, npages * sizeof(*new.rmap));
 
+		new.user_alloc = user_alloc;
 		if (user_alloc)
 			new.userspace_addr = mem->userspace_addr;
 		else {
@@ -727,6 +728,19 @@ static int kvm_vm_ioctl_set_memory_region(struct kvm *kvm,
 			if (IS_ERR((void *)new.userspace_addr))
 				goto out_unlock;
 		}
+	} else {
+		if (!old.user_alloc && old.rmap) {
+			int ret;
+
+			down_write(&current->mm->mmap_sem);
+			ret = do_munmap(current->mm, old.userspace_addr,
+					old.npages * PAGE_SIZE);
+			up_write(&current->mm->mmap_sem);
+			if (ret < 0)
+				printk(KERN_WARNING
+				       "kvm_vm_ioctl_set_memory_region: "
+				       "failed to munmap memory\n");
+		}
 	}
 
 	/* Allocate page dirty bitmap if needed */