fuse: fix possibly missed wake-up after abort

In current fuse_drop_waiting() implementation it's possible that
fuse_wait_aborted() will not be woken up in the unlikely case that
fuse_abort_conn() + fuse_wait_aborted() runs in between checking
fc->connected and calling atomic_dec(&fc->num_waiting).

Do the atomic_dec_and_test() unconditionally, which also provides the
necessary barrier against reordering with the fc->connected check.

The explicit smp_mb() in fuse_wait_aborted() is not actually needed, since
the spin_unlock() in fuse_abort_conn() provides the necessary RELEASE
barrier after resetting fc->connected.  However, this is not a performance
sensitive path, and adding the explicit barrier makes it easier to
document.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Fixes: b8f95e5d ("fuse: umount should wait for all requests")
Cc: <stable@vger.kernel.org> #v4.19

fs/fuse/dev.c

@@ -165,9 +165,13 @@ static bool fuse_block_alloc(struct fuse_conn *fc, bool for_background)
 
 static void fuse_drop_waiting(struct fuse_conn *fc)
 {
-	if (fc->connected) {
-		atomic_dec(&fc->num_waiting);
-	} else if (atomic_dec_and_test(&fc->num_waiting)) {
+	/*
+	 * lockess check of fc->connected is okay, because atomic_dec_and_test()
+	 * provides a memory barrier mached with the one in fuse_wait_aborted()
+	 * to ensure no wake-up is missed.
+	 */
+	if (atomic_dec_and_test(&fc->num_waiting) &&
+	    !READ_ONCE(fc->connected)) {
 		/* wake up aborters */
 		wake_up_all(&fc->blocked_waitq);
 	}
@@ -2221,6 +2225,8 @@ EXPORT_SYMBOL_GPL(fuse_abort_conn);
 
 void fuse_wait_aborted(struct fuse_conn *fc)
 {
+	/* matches implicit memory barrier in fuse_drop_waiting() */
+	smp_mb();
 	wait_event(fc->blocked_waitq, atomic_read(&fc->num_waiting) == 0);
 }