diff --git a/fs/io_uring.c b/fs/io_uring.c
index b40b640924db..e92029bd3936 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -169,7 +169,7 @@ struct io_rings {
 	 * The application needs a full memory barrier before checking
 	 * for IORING_SQ_NEED_WAKEUP after updating the sq tail.
 	 */
-	u32			sq_flags;
+	atomic_t		sq_flags;
 	/*
 	 * Runtime CQ flags
 	 *
@@ -2030,8 +2030,7 @@ static bool __io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force)
 	all_flushed = list_empty(&ctx->cq_overflow_list);
 	if (all_flushed) {
 		clear_bit(IO_CHECK_CQ_OVERFLOW_BIT, &ctx->check_cq);
-		WRITE_ONCE(ctx->rings->sq_flags,
-			   ctx->rings->sq_flags & ~IORING_SQ_CQ_OVERFLOW);
+		atomic_andnot(IORING_SQ_CQ_OVERFLOW, &ctx->rings->sq_flags);
 	}
 
 	io_commit_cqring(ctx);
@@ -2125,8 +2124,7 @@ static bool io_cqring_event_overflow(struct io_ring_ctx *ctx, u64 user_data,
 	}
 	if (list_empty(&ctx->cq_overflow_list)) {
 		set_bit(IO_CHECK_CQ_OVERFLOW_BIT, &ctx->check_cq);
-		WRITE_ONCE(ctx->rings->sq_flags,
-			   ctx->rings->sq_flags | IORING_SQ_CQ_OVERFLOW);
+		atomic_or(IORING_SQ_CQ_OVERFLOW, &ctx->rings->sq_flags);
 
 	}
 	ocqe->cqe.user_data = user_data;
@@ -8108,23 +8106,6 @@ static inline bool io_sqd_events_pending(struct io_sq_data *sqd)
 	return READ_ONCE(sqd->state);
 }
 
-static inline void io_ring_set_wakeup_flag(struct io_ring_ctx *ctx)
-{
-	/* Tell userspace we may need a wakeup call */
-	spin_lock(&ctx->completion_lock);
-	WRITE_ONCE(ctx->rings->sq_flags,
-		   ctx->rings->sq_flags | IORING_SQ_NEED_WAKEUP);
-	spin_unlock(&ctx->completion_lock);
-}
-
-static inline void io_ring_clear_wakeup_flag(struct io_ring_ctx *ctx)
-{
-	spin_lock(&ctx->completion_lock);
-	WRITE_ONCE(ctx->rings->sq_flags,
-		   ctx->rings->sq_flags & ~IORING_SQ_NEED_WAKEUP);
-	spin_unlock(&ctx->completion_lock);
-}
-
 static int __io_sq_thread(struct io_ring_ctx *ctx, bool cap_entries)
 {
 	unsigned int to_submit;
@@ -8240,8 +8221,8 @@ static int io_sq_thread(void *data)
 			bool needs_sched = true;
 
 			list_for_each_entry(ctx, &sqd->ctx_list, sqd_list) {
-				io_ring_set_wakeup_flag(ctx);
-
+				atomic_or(IORING_SQ_NEED_WAKEUP,
+						&ctx->rings->sq_flags);
 				if ((ctx->flags & IORING_SETUP_IOPOLL) &&
 				    !wq_list_empty(&ctx->iopoll_list)) {
 					needs_sched = false;
@@ -8266,7 +8247,8 @@ static int io_sq_thread(void *data)
 				mutex_lock(&sqd->lock);
 			}
 			list_for_each_entry(ctx, &sqd->ctx_list, sqd_list)
-				io_ring_clear_wakeup_flag(ctx);
+				atomic_andnot(IORING_SQ_NEED_WAKEUP,
+						&ctx->rings->sq_flags);
 		}
 
 		finish_wait(&sqd->wait, &wait);
@@ -8276,7 +8258,7 @@ static int io_sq_thread(void *data)
 	io_uring_cancel_generic(true, sqd);
 	sqd->thread = NULL;
 	list_for_each_entry(ctx, &sqd->ctx_list, sqd_list)
-		io_ring_set_wakeup_flag(ctx);
+		atomic_or(IORING_SQ_NEED_WAKEUP, &ctx->rings->sq_flags);
 	io_run_task_work();
 	mutex_unlock(&sqd->lock);
 
@@ -12029,6 +12011,8 @@ static int __init io_uring_init(void)
 	BUILD_BUG_ON(ARRAY_SIZE(io_op_defs) != IORING_OP_LAST);
 	BUILD_BUG_ON(__REQ_F_LAST_BIT > 8 * sizeof(int));
 
+	BUILD_BUG_ON(sizeof(atomic_t) != sizeof(u32));
+
 	req_cachep = KMEM_CACHE(io_kiocb, SLAB_HWCACHE_ALIGN | SLAB_PANIC |
 				SLAB_ACCOUNT);
 	return 0;