io-uring: add napi busy poll support

Stefan Roesch · axboe · commit 8d0c12a80cde · 2024-02-09T11:54:19.000-07:00
This adds the napi busy polling support in io_uring.c. It adds a new napi_list to the io_ring_ctx structure. This list contains the list of napi_id's that are currently enabled for busy polling. The list is synchronized by the new napi_lock spin lock. The current default napi busy polling time is stored in napi_busy_poll_to. If napi busy polling is not enabled, the value is 0. In addition there is also a hash table. The hash table store the napi id and the pointer to the above list nodes. The hash table is used to speed up the lookup to the list elements. The hash table is synchronized with rcu. The NAPI_TIMEOUT is stored as a timeout to make sure that the time a napi entry is stored in the napi list is limited. The busy poll timeout is also stored as part of the io_wait_queue. This is necessary as for sq polling the poll interval needs to be adjusted and the napi callback allows only to pass in one value. This has been tested with two simple programs from the liburing library repository: the napi client and the napi server program. The client sends a request, which has a timestamp in its payload and the server replies with the same payload. The client calculates the roundtrip time and stores it to calculate the results. The client is running on host1 and the server is running on host 2 (in the same rack). The measured times below are roundtrip times. They are average times over 5 runs each. Each run measures 1 million roundtrips. no rx coal rx coal: frames=88,usecs=33 Default 57us 56us client_poll=100us 47us 46us server_poll=100us 51us 46us client_poll=100us+ 40us 40us server_poll=100us client_poll=100us+ 41us 39us server_poll=100us+ prefer napi busy poll on client client_poll=100us+ 41us 39us server_poll=100us+ prefer napi busy poll on server client_poll=100us+ 41us 39us server_poll=100us+ prefer napi busy poll on client + server Signed-off-by: Stefan Roesch <shr@devkernel.io> Suggested-by: Olivier Langlois <olivier@trillion01.com> Acked-by: Jakub Kicinski <kuba@kernel.org> Link: https://lore.kernel.org/r/20230608163839.2891748-5-shr@devkernel.io Signed-off-by: Jens Axboe <axboe@kernel.dk>
diff --git a/include/linux/io_uring_types.h b/include/linux/io_uring_types.h
@@ -2,6 +2,7 @@
 #define IO_URING_TYPES_H
 
 #include <linux/blkdev.h>
+#include <linux/hashtable.h>
 #include <linux/task_work.h>
 #include <linux/bitmap.h>
 #include <linux/llist.h>
@@ -247,6 +248,7 @@ struct io_ring_ctx {
 		struct percpu_ref	refs;
 
 		enum task_work_notify_mode	notify_method;
+		unsigned			sq_thread_idle;
 	} ____cacheline_aligned_in_smp;
 
 	/* submission data */
@@ -410,7 +412,18 @@ struct io_ring_ctx {
 
 	struct callback_head		poll_wq_task_work;
 	struct list_head		defer_list;
-	unsigned			sq_thread_idle;
+
+#ifdef CONFIG_NET_RX_BUSY_POLL
+	struct list_head	napi_list;	/* track busy poll napi_id */
+	spinlock_t		napi_lock;	/* napi_list lock */
+
+	/* napi busy poll default timeout */
+	unsigned int		napi_busy_poll_to;
+	bool			napi_prefer_busy_poll;
+
+	DECLARE_HASHTABLE(napi_ht, 4);
+#endif
+
 	/* protected by ->completion_lock */
 	unsigned			evfd_last_cq_tail;
 
diff --git a/io_uring/Makefile b/io_uring/Makefile
@@ -11,3 +11,4 @@ obj-$(CONFIG_IO_URING)		+= io_uring.o xattr.o nop.o fs.o splice.o \
 					notif.o waitid.o register.o truncate.o
 obj-$(CONFIG_IO_WQ)		+= io-wq.o
 obj-$(CONFIG_FUTEX)		+= futex.o
+obj-$(CONFIG_NET_RX_BUSY_POLL) += napi.o
diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
@@ -95,6 +95,7 @@
 #include "notif.h"
 #include "waitid.h"
 #include "futex.h"
+#include "napi.h"
 
 #include "timeout.h"
 #include "poll.h"
@@ -349,6 +350,8 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	INIT_DELAYED_WORK(&ctx->fallback_work, io_fallback_req_func);
 	INIT_WQ_LIST(&ctx->submit_state.compl_reqs);
 	INIT_HLIST_HEAD(&ctx->cancelable_uring_cmd);
+	io_napi_init(ctx);
+
 	return ctx;
 err:
 	kfree(ctx->cancel_table.hbs);
@@ -2602,9 +2605,13 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events,
 
 		if (get_timespec64(&ts, uts))
 			return -EFAULT;
+
 		iowq.timeout = ktime_add_ns(timespec64_to_ktime(ts), ktime_get_ns());
+		io_napi_adjust_timeout(ctx, &iowq, &ts);
 	}
 
+	io_napi_busy_loop(ctx, &iowq);
+
 	trace_io_uring_cqring_wait(ctx, min_events);
 	do {
 		int nr_wait = (int) iowq.cq_tail - READ_ONCE(ctx->rings->cq.tail);
@@ -2897,6 +2904,7 @@ static __cold void io_ring_ctx_free(struct io_ring_ctx *ctx)
 	io_req_caches_free(ctx);
 	if (ctx->hash_map)
 		io_wq_put_hash(ctx->hash_map);
+	io_napi_free(ctx);
 	kfree(ctx->cancel_table.hbs);
 	kfree(ctx->cancel_table_locked.hbs);
 	kfree(ctx->io_bl);
diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h
@@ -42,6 +42,10 @@ struct io_wait_queue {
 	unsigned nr_timeouts;
 	ktime_t timeout;
 
+#ifdef CONFIG_NET_RX_BUSY_POLL
+	unsigned int napi_busy_poll_to;
+	bool napi_prefer_busy_poll;
+#endif
 };
 
 static inline bool io_should_wake(struct io_wait_queue *iowq)
diff --git a/io_uring/napi.c b/io_uring/napi.c
@@ -0,0 +1,255 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include "io_uring.h"
+#include "napi.h"
+
+#ifdef CONFIG_NET_RX_BUSY_POLL
+
+/* Timeout for cleanout of stale entries. */
+#define NAPI_TIMEOUT		(60 * SEC_CONVERSION)
+
+struct io_napi_entry {
+	unsigned int		napi_id;
+	struct list_head	list;
+
+	unsigned long		timeout;
+	struct hlist_node	node;
+
+	struct rcu_head		rcu;
+};
+
+static struct io_napi_entry *io_napi_hash_find(struct hlist_head *hash_list,
+					       unsigned int napi_id)
+{
+	struct io_napi_entry *e;
+
+	hlist_for_each_entry_rcu(e, hash_list, node) {
+		if (e->napi_id != napi_id)
+			continue;
+		e->timeout = jiffies + NAPI_TIMEOUT;
+		return e;
+	}
+
+	return NULL;
+}
+
+void __io_napi_add(struct io_ring_ctx *ctx, struct socket *sock)
+{
+	struct hlist_head *hash_list;
+	unsigned int napi_id;
+	struct sock *sk;
+	struct io_napi_entry *e;
+
+	sk = sock->sk;
+	if (!sk)
+		return;
+
+	napi_id = READ_ONCE(sk->sk_napi_id);
+
+	/* Non-NAPI IDs can be rejected. */
+	if (napi_id < MIN_NAPI_ID)
+		return;
+
+	hash_list = &ctx->napi_ht[hash_min(napi_id, HASH_BITS(ctx->napi_ht))];
+
+	rcu_read_lock();
+	e = io_napi_hash_find(hash_list, napi_id);
+	if (e) {
+		e->timeout = jiffies + NAPI_TIMEOUT;
+		rcu_read_unlock();
+		return;
+	}
+	rcu_read_unlock();
+
+	e = kmalloc(sizeof(*e), GFP_NOWAIT);
+	if (!e)
+		return;
+
+	e->napi_id = napi_id;
+	e->timeout = jiffies + NAPI_TIMEOUT;
+
+	spin_lock(&ctx->napi_lock);
+	if (unlikely(io_napi_hash_find(hash_list, napi_id))) {
+		spin_unlock(&ctx->napi_lock);
+		kfree(e);
+		return;
+	}
+
+	hlist_add_tail_rcu(&e->node, hash_list);
+	list_add_tail(&e->list, &ctx->napi_list);
+	spin_unlock(&ctx->napi_lock);
+}
+
+static void __io_napi_remove_stale(struct io_ring_ctx *ctx)
+{
+	struct io_napi_entry *e;
+	unsigned int i;
+
+	spin_lock(&ctx->napi_lock);
+	hash_for_each(ctx->napi_ht, i, e, node) {
+		if (time_after(jiffies, e->timeout)) {
+			list_del(&e->list);
+			hash_del_rcu(&e->node);
+			kfree_rcu(e, rcu);
+		}
+	}
+	spin_unlock(&ctx->napi_lock);
+}
+
+static inline void io_napi_remove_stale(struct io_ring_ctx *ctx, bool is_stale)
+{
+	if (is_stale)
+		__io_napi_remove_stale(ctx);
+}
+
+static inline bool io_napi_busy_loop_timeout(unsigned long start_time,
+					     unsigned long bp_usec)
+{
+	if (bp_usec) {
+		unsigned long end_time = start_time + bp_usec;
+		unsigned long now = busy_loop_current_time();
+
+		return time_after(now, end_time);
+	}
+
+	return true;
+}
+
+static bool io_napi_busy_loop_should_end(void *data,
+					 unsigned long start_time)
+{
+	struct io_wait_queue *iowq = data;
+
+	if (signal_pending(current))
+		return true;
+	if (io_should_wake(iowq))
+		return true;
+	if (io_napi_busy_loop_timeout(start_time, iowq->napi_busy_poll_to))
+		return true;
+
+	return false;
+}
+
+static bool __io_napi_do_busy_loop(struct io_ring_ctx *ctx,
+				   void *loop_end_arg)
+{
+	struct io_napi_entry *e;
+	bool (*loop_end)(void *, unsigned long) = NULL;
+	bool is_stale = false;
+
+	if (loop_end_arg)
+		loop_end = io_napi_busy_loop_should_end;
+
+	list_for_each_entry_rcu(e, &ctx->napi_list, list) {
+		napi_busy_loop_rcu(e->napi_id, loop_end, loop_end_arg,
+				   ctx->napi_prefer_busy_poll, BUSY_POLL_BUDGET);
+
+		if (time_after(jiffies, e->timeout))
+			is_stale = true;
+	}
+
+	return is_stale;
+}
+
+static void io_napi_blocking_busy_loop(struct io_ring_ctx *ctx,
+				       struct io_wait_queue *iowq)
+{
+	unsigned long start_time = busy_loop_current_time();
+	void *loop_end_arg = NULL;
+	bool is_stale = false;
+
+	/* Singular lists use a different napi loop end check function and are
+	 * only executed once.
+	 */
+	if (list_is_singular(&ctx->napi_list))
+		loop_end_arg = iowq;
+
+	rcu_read_lock();
+	do {
+		is_stale = __io_napi_do_busy_loop(ctx, loop_end_arg);
+	} while (!io_napi_busy_loop_should_end(iowq, start_time) && !loop_end_arg);
+	rcu_read_unlock();
+
+	io_napi_remove_stale(ctx, is_stale);
+}
+
+/*
+ * io_napi_init() - Init napi settings
+ * @ctx: pointer to io-uring context structure
+ *
+ * Init napi settings in the io-uring context.
+ */
+void io_napi_init(struct io_ring_ctx *ctx)
+{
+	INIT_LIST_HEAD(&ctx->napi_list);
+	spin_lock_init(&ctx->napi_lock);
+	ctx->napi_prefer_busy_poll = false;
+	ctx->napi_busy_poll_to = READ_ONCE(sysctl_net_busy_poll);
+}
+
+/*
+ * io_napi_free() - Deallocate napi
+ * @ctx: pointer to io-uring context structure
+ *
+ * Free the napi list and the hash table in the io-uring context.
+ */
+void io_napi_free(struct io_ring_ctx *ctx)
+{
+	struct io_napi_entry *e;
+	LIST_HEAD(napi_list);
+	unsigned int i;
+
+	spin_lock(&ctx->napi_lock);
+	hash_for_each(ctx->napi_ht, i, e, node) {
+		hash_del_rcu(&e->node);
+		kfree_rcu(e, rcu);
+	}
+	spin_unlock(&ctx->napi_lock);
+}
+
+/*
+ * __io_napi_adjust_timeout() - Add napi id to the busy poll list
+ * @ctx: pointer to io-uring context structure
+ * @iowq: pointer to io wait queue
+ * @ts: pointer to timespec or NULL
+ *
+ * Adjust the busy loop timeout according to timespec and busy poll timeout.
+ */
+void __io_napi_adjust_timeout(struct io_ring_ctx *ctx, struct io_wait_queue *iowq,
+			      struct timespec64 *ts)
+{
+	unsigned int poll_to = READ_ONCE(ctx->napi_busy_poll_to);
+
+	if (ts) {
+		struct timespec64 poll_to_ts = ns_to_timespec64(1000 * (s64)poll_to);
+
+		if (timespec64_compare(ts, &poll_to_ts) > 0) {
+			*ts = timespec64_sub(*ts, poll_to_ts);
+		} else {
+			u64 to = timespec64_to_ns(ts);
+
+			do_div(to, 1000);
+			ts->tv_sec = 0;
+			ts->tv_nsec = 0;
+		}
+	}
+
+	iowq->napi_busy_poll_to = poll_to;
+}
+
+/*
+ * __io_napi_busy_loop() - execute busy poll loop
+ * @ctx: pointer to io-uring context structure
+ * @iowq: pointer to io wait queue
+ *
+ * Execute the busy poll loop and merge the spliced off list.
+ */
+void __io_napi_busy_loop(struct io_ring_ctx *ctx, struct io_wait_queue *iowq)
+{
+	iowq->napi_prefer_busy_poll = READ_ONCE(ctx->napi_prefer_busy_poll);
+
+	if (!(ctx->flags & IORING_SETUP_SQPOLL) && iowq->napi_busy_poll_to)
+		io_napi_blocking_busy_loop(ctx, iowq);
+}
+
+#endif
diff --git a/io_uring/napi.h b/io_uring/napi.h
diff --git a/io_uring/poll.c b/io_uring/poll.c
