io_uring: enable option to only trigger eventfd for async completions

to #26323588

commit f2842ab5b72d7ee5f7f8385c2d4f32c133f5837b upstream.

If an application is using eventfd notifications with poll to know when
new SQEs can be issued, it's expecting the following read/writes to
complete inline. And with that, it knows that there are events available,
and don't want spurious wakeups on the eventfd for those requests.

This adds IORING_REGISTER_EVENTFD_ASYNC, which works just like
IORING_REGISTER_EVENTFD, except it only triggers notifications for events
that happen from async completions (IRQ, or io-wq worker completions).
Any completions inline from the submission itself will not trigger
notifications.
Suggested-by: NMark Papadakis <markuspapadakis@icloud.com>
Signed-off-by: NJens Axboe <axboe@kernel.dk>
Signed-off-by: NJoseph Qi <joseph.qi@linux.alibaba.com>
Acked-by: NXiaoguang Wang <xiaoguang.wang@linux.alibaba.com>

fs/io_uring.c

@@ -206,6 +206,7 @@ struct io_ring_ctx {
 		int			account_mem: 1;
 		int			cq_overflow_flushed: 1;
 		int			drain_next: 1;
+		int			eventfd_async: 1;
 
 		/*
 		 * Ring buffer of indices into array of io_uring_sqe, which is
@@ -963,13 +964,20 @@ static struct io_uring_cqe *io_get_cqring(struct io_ring_ctx *ctx)
 	return &rings->cqes[tail & ctx->cq_mask];
 }
 
+static inline bool io_should_trigger_evfd(struct io_ring_ctx *ctx)
+{
+	if (!ctx->eventfd_async)
+		return true;
+	return io_wq_current_is_worker() || in_interrupt();
+}
+
 static void io_cqring_ev_posted(struct io_ring_ctx *ctx)
 {
 	if (waitqueue_active(&ctx->wait))
 		wake_up(&ctx->wait);
 	if (waitqueue_active(&ctx->sqo_wait))
 		wake_up(&ctx->sqo_wait);
-	if (ctx->cq_ev_fd)
+	if (ctx->cq_ev_fd && io_should_trigger_evfd(ctx))
 		eventfd_signal(ctx->cq_ev_fd, 1);
 }
 
@@ -6541,10 +6549,17 @@ static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
 		ret = io_sqe_files_update(ctx, arg, nr_args);
 		break;
 	case IORING_REGISTER_EVENTFD:
+	case IORING_REGISTER_EVENTFD_ASYNC:
 		ret = -EINVAL;
 		if (nr_args != 1)
 			break;
 		ret = io_eventfd_register(ctx, arg);
+		if (ret)
+			break;
+		if (opcode == IORING_REGISTER_EVENTFD_ASYNC)
+			ctx->eventfd_async = 1;
+		else
+			ctx->eventfd_async = 0;
 		break;
 	case IORING_UNREGISTER_EVENTFD:
 		ret = -EINVAL;

include/uapi/linux/io_uring.h

@@ -192,6 +192,7 @@ struct io_uring_params {
 #define IORING_REGISTER_EVENTFD		4
 #define IORING_UNREGISTER_EVENTFD	5
 #define IORING_REGISTER_FILES_UPDATE	6
+#define IORING_REGISTER_EVENTFD_ASYNC	7
 
 struct io_uring_files_update {
 	__u32 offset;