io_uring: fix poll file assign deadlock

commit cce64ef upstream.

We pass "unlocked" into io_assign_file() in io_poll_check_events(),
which can lead to double locking.

Fixes: 6bf9c47 ("io_uring: defer file assignment")
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/2476d4ae46554324b599ee4055447b105f20a75a.1649862516.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fs/io_uring.c

@@ -5510,8 +5510,9 @@ static int io_poll_check_events(struct io_kiocb *req, bool locked)
 
 		if (!req->result) {
 			struct poll_table_struct pt = { ._key = poll->events };
+			unsigned flags = locked ? 0 : IO_URING_F_UNLOCKED;
 
-			if (unlikely(!io_assign_file(req, IO_URING_F_UNLOCKED)))
+			if (unlikely(!io_assign_file(req, flags)))
 				req->result = -EBADF;
 			else
 				req->result = vfs_poll(req->file, &pt) & poll->events;