userfaultfd: convert userfaultfd_ctx::refcount to refcount_t

Reference counters should use refcount_t rather than atomic_t, since the
refcount_t implementation can prevent overflows, reducing the
exploitability of reference leak bugs.  userfaultfd_ctx::refcount is a
reference counter with the usual semantics, so convert it to refcount_t.

Note: I replaced the BUG() on incrementing a 0 refcount with just
refcount_inc(), since part of the semantics of refcount_t is that that
incrementing a 0 refcount is not allowed; with CONFIG_REFCOUNT_FULL,
refcount_inc() already checks for it and warns.

Link: http://lkml.kernel.org/r/20181115003916.63381-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
diff --git a/fs/userfaultfd.c b/fs/userfaultfd.c
index 59dc280..e211b99 100644
--- a/fs/userfaultfd.c
+++ b/fs/userfaultfd.c
@@ -53,7 +53,7 @@ struct userfaultfd_ctx {
 	/* a refile sequence protected by fault_pending_wqh lock */
 	struct seqcount refile_seq;
 	/* pseudo fd refcounting */
-	atomic_t refcount;
+	refcount_t refcount;
 	/* userfaultfd syscall flags */
 	unsigned int flags;
 	/* features requested from the userspace */
@@ -140,8 +140,7 @@ static int userfaultfd_wake_function(wait_queue_entry_t *wq, unsigned mode,
  */
 static void userfaultfd_ctx_get(struct userfaultfd_ctx *ctx)
 {
-	if (!atomic_inc_not_zero(&ctx->refcount))
-		BUG();
+	refcount_inc(&ctx->refcount);
 }
 
 /**
@@ -154,7 +153,7 @@ static void userfaultfd_ctx_get(struct userfaultfd_ctx *ctx)
  */
 static void userfaultfd_ctx_put(struct userfaultfd_ctx *ctx)
 {
-	if (atomic_dec_and_test(&ctx->refcount)) {
+	if (refcount_dec_and_test(&ctx->refcount)) {
 		VM_BUG_ON(spin_is_locked(&ctx->fault_pending_wqh.lock));
 		VM_BUG_ON(waitqueue_active(&ctx->fault_pending_wqh));
 		VM_BUG_ON(spin_is_locked(&ctx->fault_wqh.lock));
@@ -686,7 +685,7 @@ int dup_userfaultfd(struct vm_area_struct *vma, struct list_head *fcs)
 			return -ENOMEM;
 		}
 
-		atomic_set(&ctx->refcount, 1);
+		refcount_set(&ctx->refcount, 1);
 		ctx->flags = octx->flags;
 		ctx->state = UFFD_STATE_RUNNING;
 		ctx->features = octx->features;
@@ -1927,7 +1926,7 @@ SYSCALL_DEFINE1(userfaultfd, int, flags)
 	if (!ctx)
 		return -ENOMEM;
 
-	atomic_set(&ctx->refcount, 1);
+	refcount_set(&ctx->refcount, 1);
 	ctx->flags = flags;
 	ctx->features = 0;
 	ctx->state = UFFD_STATE_WAIT_API;