io_uring/io-wq: close io-wq full-stop gap

There is an old problem with io-wq cancellation where requests should be
killed and are in io-wq but are not discoverable, e.g. in @next_hashed
or @linked vars of io_worker_handle_work(). It adds some unreliability
to individual request canellation, but also may potentially get
__io_uring_cancel() stuck. For instance:

1) An __io_uring_cancel()'s cancellation round have not found any
   request but there are some as desribed.
2) __io_uring_cancel() goes to sleep
3) Then workers wake up and try to execute those hidden requests
   that happen to be unbound.

As we already cancel all requests of io-wq there, set IO_WQ_BIT_EXIT
in advance, so preventing 3) from executing unbound requests. The
workers will initially break looping because of getting a signal as they
are threads of the dying/exec()'ing user task.

Cc: [email protected]
Signed-off-by: Pavel Begunkov <[email protected]>
Link: https://lore.kernel.org/r/abfcf8c54cb9e8f7bfbad7e9a0cc5433cc70bdc2.1621781238.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <[email protected]>

Author: isilence

fs/io-wq.c

@@ -979,13 +979,16 @@ static bool io_task_work_match(struct callback_head *cb, void *data)
 	return cwd->wqe->wq == data;
 }
 
+void io_wq_exit_start(struct io_wq *wq)
+{
+	set_bit(IO_WQ_BIT_EXIT, &wq->state);
+}
+
 static void io_wq_exit_workers(struct io_wq *wq)
 {
 	struct callback_head *cb;
 	int node;
 
-	set_bit(IO_WQ_BIT_EXIT, &wq->state);
-
 	if (!wq->task)
 		return;
 
@@ -1020,8 +1023,6 @@ static void io_wq_destroy(struct io_wq *wq)
 
 	cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node);
 
-	io_wq_exit_workers(wq);
-
 	for_each_node(node) {
 		struct io_wqe *wqe = wq->wqes[node];
 		struct io_cb_cancel_data match = {
@@ -1036,16 +1037,13 @@ static void io_wq_destroy(struct io_wq *wq)
 	kfree(wq);
 }
 
-void io_wq_put(struct io_wq *wq)
-{
-	if (refcount_dec_and_test(&wq->refs))
-		io_wq_destroy(wq);
-}
-
 void io_wq_put_and_exit(struct io_wq *wq)
 {
+	WARN_ON_ONCE(!test_bit(IO_WQ_BIT_EXIT, &wq->state));
+
 	io_wq_exit_workers(wq);
-	io_wq_put(wq);
+	if (refcount_dec_and_test(&wq->refs))
+		io_wq_destroy(wq);
 }
 
 static bool io_wq_worker_affinity(struct io_worker *worker, void *data)

fs/io-wq.h

@@ -122,7 +122,7 @@ struct io_wq_data {
 };
 
 struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data);
-void io_wq_put(struct io_wq *wq);
+void io_wq_exit_start(struct io_wq *wq);
 void io_wq_put_and_exit(struct io_wq *wq);
 
 void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work);

fs/io_uring.c

@@ -9078,6 +9078,9 @@ static void io_uring_cancel_sqpoll(struct io_sq_data *sqd)
 
 	if (!current->io_uring)
 		return;
+	if (tctx->io_wq)
+		io_wq_exit_start(tctx->io_wq);
+
 	WARN_ON_ONCE(!sqd || sqd->thread != current);
 
 	atomic_inc(&tctx->in_idle);
@@ -9112,6 +9115,9 @@ void __io_uring_cancel(struct files_struct *files)
 	DEFINE_WAIT(wait);
 	s64 inflight;
 
+	if (tctx->io_wq)
+		io_wq_exit_start(tctx->io_wq);
+
 	/* make sure overflow events are dropped */
 	atomic_inc(&tctx->in_idle);
 	do {