io_uring: add multishot mode for IORING_OP_POLL_ADD

The default io_uring poll mode is one-shot, where once the event triggers,
the poll command is completed and won't trigger any further events. If
we're doing repeated polling on the same file or socket, then it can be
more efficient to do multishot, where we keep triggering whenever the
event becomes true.

This deviates from the usual norm of having one CQE per SQE submitted. Add
a CQE flag, IORING_CQE_F_MORE, which tells the application to expect
further completion events from the submitted SQE. Right now the only user
of this is POLL_ADD in multishot mode.

Since sqe->poll_events is using the space that we normally use for adding
flags to commands, use sqe->len for the flag space for POLL_ADD. Multishot
mode is selected by setting IORING_POLL_ADD_MULTI in sqe->len. An
application should expect more CQEs for the specificed SQE if the CQE is
flagged with IORING_CQE_F_MORE. In multishot mode, only cancelation or an
error will terminate the poll request, in which case the flag will be
cleared.
Signed-off-by: Jens Axboe <axboe@kernel.dk>

fs/io_uring.c

@@ -4898,17 +4898,25 @@ static void io_poll_remove_double(struct io_kiocb *req)
 	}
 }
 
-static void io_poll_complete(struct io_kiocb *req, __poll_t mask, int error)
+static bool io_poll_complete(struct io_kiocb *req, __poll_t mask, int error)
 {
 	struct io_ring_ctx *ctx = req->ctx;
+	unsigned flags = IORING_CQE_F_MORE;
 
-	if (!error && req->poll.canceled)
+	if (!error && req->poll.canceled) {
 		error = -ECANCELED;
-
+		req->poll.events |= EPOLLONESHOT;
+	}
+	if (error || (req->poll.events & EPOLLONESHOT)) {
 		io_poll_remove_double(req);
 		req->poll.done = true;
-	io_cqring_fill_event(req, error ? error : mangle_poll(mask));
+		flags = 0;
+	}
+	if (!error)
+		error = mangle_poll(mask);
+	__io_cqring_fill_event(req, error, flags);
 	io_commit_cqring(ctx);
+	return !(flags & IORING_CQE_F_MORE);
 }
 
 static void io_poll_task_func(struct callback_head *cb)
@@ -4920,15 +4928,26 @@ static void io_poll_task_func(struct callback_head *cb)
 	if (io_poll_rewait(req, &req->poll)) {
 		spin_unlock_irq(&ctx->completion_lock);
 	} else {
+		bool done, post_ev;
+
+		post_ev = done = io_poll_complete(req, req->result, 0);
+		if (done) {
 			hash_del(&req->hash_node);
-		io_poll_complete(req, req->result, 0);
+		} else if (!(req->poll.events & EPOLLONESHOT)) {
+			post_ev = true;
+			req->result = 0;
+			add_wait_queue(req->poll.head, &req->poll.wait);
+		}
 		spin_unlock_irq(&ctx->completion_lock);
 
-		nxt = io_put_req_find_next(req);
+		if (post_ev)
 			io_cqring_ev_posted(ctx);
+		if (done) {
+			nxt = io_put_req_find_next(req);
 			if (nxt)
 				__io_req_task_submit(nxt);
 		}
+	}
 }
 
 static int io_poll_double_wake(struct wait_queue_entry *wait, unsigned mode,
@@ -4941,6 +4960,8 @@ static int io_poll_double_wake(struct wait_queue_entry *wait, unsigned mode,
 	/* for instances that support it check for an event match first: */
 	if (mask && !(mask & poll->events))
 		return 0;
+	if (!(poll->events & EPOLLONESHOT))
+		return poll->wait.func(&poll->wait, mode, sync, key);
 
 	list_del_init(&wait->entry);
 
@@ -5106,7 +5127,7 @@ static __poll_t __io_arm_poll_handler(struct io_kiocb *req,
 			ipt->error = 0;
 			mask = 0;
 		}
-		if (mask || ipt->error)
+		if ((mask && (poll->events & EPOLLONESHOT)) || ipt->error)
 			list_del_init(&poll->wait.entry);
 		else if (cancel)
 			WRITE_ONCE(poll->canceled, true);
@@ -5149,7 +5170,7 @@ static bool io_arm_poll_handler(struct io_kiocb *req)
 	req->flags |= REQ_F_POLLED;
 	req->apoll = apoll;
 
-	mask = 0;
+	mask = EPOLLONESHOT;
 	if (def->pollin)
 		mask |= POLLIN | POLLRDNORM;
 	if (def->pollout)
@@ -5322,18 +5343,24 @@ static void io_poll_queue_proc(struct file *file, struct wait_queue_head *head,
 static int io_poll_add_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
 	struct io_poll_iocb *poll = &req->poll;
-	u32 events;
+	u32 events, flags;
 
 	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
 		return -EINVAL;
-	if (sqe->addr || sqe->ioprio || sqe->off || sqe->len || sqe->buf_index)
+	if (sqe->addr || sqe->ioprio || sqe->off || sqe->buf_index)
+		return -EINVAL;
+	flags = READ_ONCE(sqe->len);
+	if (flags & ~IORING_POLL_ADD_MULTI)
 		return -EINVAL;
 
 	events = READ_ONCE(sqe->poll32_events);
 #ifdef __BIG_ENDIAN
 	events = swahw32(events);
 #endif
-	poll->events = demangle_poll(events) | (events & EPOLLEXCLUSIVE);
+	if (!flags)
+		events |= EPOLLONESHOT;
+	poll->events = demangle_poll(events) |
+				(events & (EPOLLEXCLUSIVE|EPOLLONESHOT));
 	return 0;
 }
 
@@ -5357,6 +5384,7 @@ static int io_poll_add(struct io_kiocb *req, unsigned int issue_flags)
 
 	if (mask) {
 		io_cqring_ev_posted(ctx);
+		if (poll->events & EPOLLONESHOT)
 			io_put_req(req);
 	}
 	return ipt.error;

include/uapi/linux/io_uring.h

@@ -159,6 +159,16 @@ enum {
  */
 #define SPLICE_F_FD_IN_FIXED	(1U << 31) /* the last bit of __u32 */
 
+/*
+ * POLL_ADD flags. Note that since sqe->poll_events is the flag space, the
+ * command flags for POLL_ADD are stored in sqe->len.
+ *
+ * IORING_POLL_ADD_MULTI	Multishot poll. Sets IORING_CQE_F_MORE if
+ *				the poll handler will continue to report
+ *				CQEs on behalf of the same SQE.
+ */
+#define IORING_POLL_ADD_MULTI	(1U << 0)
+
 /*
  * IO completion data structure (Completion Queue Entry)
  */
@@ -172,8 +182,10 @@ struct io_uring_cqe {
  * cqe->flags
  *
  * IORING_CQE_F_BUFFER	If set, the upper 16 bits are the buffer ID
+ * IORING_CQE_F_MORE	If set, parent SQE will generate more CQE entries
  */
 #define IORING_CQE_F_BUFFER		(1U << 0)
+#define IORING_CQE_F_MORE		(1U << 1)
 
 enum {
 	IORING_CQE_BUFFER_SHIFT		= 16,