io-wq: invoke work cancelation with wqe->lock held

io_wqe_cancel_pending_work() grabs it internally, grab it upfront
instead. For the running work cancelation, grab the lock around it as
well.
Signed-off-by: Jens Axboe <axboe@kernel.dk>

fs/io-wq.c

@@ -1038,17 +1038,16 @@ static void io_wqe_cancel_pending_work(struct io_wqe *wqe,
 {
 	int i;
 retry:
-	raw_spin_lock(&wqe->lock);
 	for (i = 0; i < IO_WQ_ACCT_NR; i++) {
 		struct io_wqe_acct *acct = io_get_acct(wqe, i == 0);
 
 		if (io_acct_cancel_pending_work(wqe, acct, match)) {
+			raw_spin_lock(&wqe->lock);
 			if (match->cancel_all)
 				goto retry;
-			return;
+			break;
 		}
 	}
-	raw_spin_unlock(&wqe->lock);
 }
 
 static void io_wqe_cancel_running_work(struct io_wqe *wqe,
@@ -1077,7 +1076,9 @@ enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
 	for_each_node(node) {
 		struct io_wqe *wqe = wq->wqes[node];
 
+		raw_spin_lock(&wqe->lock);
 		io_wqe_cancel_pending_work(wqe, &match);
+		raw_spin_unlock(&wqe->lock);
 		if (match.nr_pending && !match.cancel_all)
 			return IO_WQ_CANCEL_OK;
 	}
@@ -1091,7 +1092,9 @@ enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
 	for_each_node(node) {
 		struct io_wqe *wqe = wq->wqes[node];
 
+		raw_spin_lock(&wqe->lock);
 		io_wqe_cancel_running_work(wqe, &match);
+		raw_spin_unlock(&wqe->lock);
 		if (match.nr_running && !match.cancel_all)
 			return IO_WQ_CANCEL_RUNNING;
 	}
@@ -1262,7 +1265,9 @@ static void io_wq_destroy(struct io_wq *wq)
 			.fn		= io_wq_work_match_all,
 			.cancel_all	= true,
 		};
+		raw_spin_lock(&wqe->lock);
 		io_wqe_cancel_pending_work(wqe, &match);
+		raw_spin_unlock(&wqe->lock);
 		free_cpumask_var(wqe->cpu_mask);
 		kfree(wqe);
 	}