[PATCH v5 2/4] io_uring: avoid ring quiesce while registering/unregistering eventfd

[Usama Arif <usama.arif@bytedance.com>]

This is done by creating a new RCU data structure (io_ev_fd) as part of
io_ring_ctx that holds the eventfd_ctx.

The function io_eventfd_signal is executed under rcu_read_lock with a
single rcu_dereference to io_ev_fd so that if another thread unregisters
the eventfd while io_eventfd_signal is still being executed, the
eventfd_signal for which io_eventfd_signal was called completes
successfully.

The process of registering/unregistering eventfd is done under a lock
so multiple threads don't enter a race condition while
registering/unregistering eventfd.

With the above approach ring quiesce can be avoided which is much more
expensive then using RCU lock. On the system tested, io_uring_reigster with
IORING_REGISTER_EVENTFD takes less than 1ms with RCU lock, compared to 15ms
before with ring quiesce.

Signed-off-by: Usama Arif <usama.arif@bytedance.com>
---
 fs/io_uring.c | 116 ++++++++++++++++++++++++++++++++++++++++----------
 1 file changed, 93 insertions(+), 23 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 21531609a9c6..51602bddb9a8 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -326,6 +326,13 @@ struct io_submit_state {
 	struct blk_plug		plug;
 };
 
+struct io_ev_fd {
+	struct eventfd_ctx	*cq_ev_fd;
+	struct io_ring_ctx	*ctx;
+	struct rcu_head		rcu;
+	bool 			unregistering;
+};
+
 struct io_ring_ctx {
 	/* const or read-mostly hot data */
 	struct {
@@ -399,7 +406,8 @@ struct io_ring_ctx {
 	struct {
 		unsigned		cached_cq_tail;
 		unsigned		cq_entries;
-		struct eventfd_ctx	*cq_ev_fd;
+		struct io_ev_fd	__rcu	*io_ev_fd;
+		struct mutex		ev_fd_lock;
 		struct wait_queue_head	cq_wait;
 		unsigned		cq_extra;
 		atomic_t		cq_timeouts;
@@ -1448,6 +1456,7 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	xa_init_flags(&ctx->io_buffers, XA_FLAGS_ALLOC1);
 	xa_init_flags(&ctx->personalities, XA_FLAGS_ALLOC1);
 	mutex_init(&ctx->uring_lock);
+	mutex_init(&ctx->ev_fd_lock);
 	init_waitqueue_head(&ctx->cq_wait);
 	spin_lock_init(&ctx->completion_lock);
 	spin_lock_init(&ctx->timeout_lock);
@@ -1726,13 +1735,32 @@ static inline struct io_uring_cqe *io_get_cqe(struct io_ring_ctx *ctx)
 	return &rings->cqes[tail & mask];
 }
 
-static inline bool io_should_trigger_evfd(struct io_ring_ctx *ctx)
+static void io_eventfd_signal(struct io_ring_ctx *ctx)
 {
-	if (likely(!ctx->cq_ev_fd))
-		return false;
+	struct io_ev_fd *ev_fd;
+
+	/* Return quickly if ctx->io_ev_fd doesn't exist */
+	if (likely(!rcu_dereference_raw(ctx->io_ev_fd)))
+		return;
+
+	rcu_read_lock();
+	/* rcu_dereference ctx->io_ev_fd once and use it for both for checking and eventfd_signal */
+	ev_fd = rcu_dereference(ctx->io_ev_fd);
+
+	/*
+	 * Check again if ev_fd exists incase an io_eventfd_unregister call completed between
+	 * the NULL check of ctx->io_ev_fd at the start of the function and rcu_read_lock.
+	 */
+	if (unlikely(!ev_fd))
+		goto out;
 	if (READ_ONCE(ctx->rings->cq_flags) & IORING_CQ_EVENTFD_DISABLED)
-		return false;
-	return !ctx->eventfd_async || io_wq_current_is_worker();
+		goto out;
+
+	if (!ctx->eventfd_async || io_wq_current_is_worker())
+		eventfd_signal(ev_fd->cq_ev_fd, 1);
+
+out:
+	rcu_read_unlock();
 }
 
 /*
@@ -1751,8 +1779,7 @@ static void io_cqring_ev_posted(struct io_ring_ctx *ctx)
 	 */
 	if (wq_has_sleeper(&ctx->cq_wait))
 		wake_up_all(&ctx->cq_wait);
-	if (io_should_trigger_evfd(ctx))
-		eventfd_signal(ctx->cq_ev_fd, 1);
+	io_eventfd_signal(ctx);
 }
 
 static void io_cqring_ev_posted_iopoll(struct io_ring_ctx *ctx)
@@ -1764,8 +1791,7 @@ static void io_cqring_ev_posted_iopoll(struct io_ring_ctx *ctx)
 		if (waitqueue_active(&ctx->cq_wait))
 			wake_up_all(&ctx->cq_wait);
 	}
-	if (io_should_trigger_evfd(ctx))
-		eventfd_signal(ctx->cq_ev_fd, 1);
+	io_eventfd_signal(ctx);
 }
 
 /* Returns true if there are no backlogged entries after the flush */
@@ -9353,34 +9379,76 @@ static int __io_sqe_buffers_update(struct io_ring_ctx *ctx,
 
 static int io_eventfd_register(struct io_ring_ctx *ctx, void __user *arg)
 {
+	struct io_ev_fd *ev_fd;
 	__s32 __user *fds = arg;
-	int fd;
+	int fd, ret;
 
-	if (ctx->cq_ev_fd)
-		return -EBUSY;
+	mutex_lock(&ctx->ev_fd_lock);
+	ret = -EBUSY;
+	ev_fd = rcu_dereference_protected(ctx->io_ev_fd, lockdep_is_held(&ctx->ev_fd_lock));
+	if (ev_fd) {
+		/*
+		 * If ev_fd exists, there are 2 possibilities:
+		 * - The rcu_callback to io_eventfd_put hasn't finished while unregistering
+		 * (hence ev_fd->unregistering is true) and io_eventfd_register
+		 * can continue and overwrite ctx->io_ev_fd with the new eventfd.
+		 * - Or io_eventfd_register has been called on an io_uring that has
+		 * already registered a valid eventfd in which case return -EBUSY.
+		 */
+		if(!ev_fd->unregistering)
+			goto out;
+	}
 
+	ret = -EFAULT;
 	if (copy_from_user(&fd, fds, sizeof(*fds)))
-		return -EFAULT;
+		goto out;
 
-	ctx->cq_ev_fd = eventfd_ctx_fdget(fd);
-	if (IS_ERR(ctx->cq_ev_fd)) {
-		int ret = PTR_ERR(ctx->cq_ev_fd);
+	ret = -ENOMEM;
+	ev_fd = kmalloc(sizeof(*ev_fd), GFP_KERNEL);
+	if (!ev_fd)
+		goto out;
 
-		ctx->cq_ev_fd = NULL;
-		return ret;
+	ev_fd->cq_ev_fd = eventfd_ctx_fdget(fd);
+	if (IS_ERR(ev_fd->cq_ev_fd)) {
+		ret = PTR_ERR(ev_fd->cq_ev_fd);
+		kfree(ev_fd);
+		goto out;
 	}
+	ev_fd->ctx = ctx;
+	ev_fd->unregistering = false;
 
-	return 0;
+	rcu_assign_pointer(ctx->io_ev_fd, ev_fd);
+	ret = 0;
+
+out:
+	mutex_unlock(&ctx->ev_fd_lock);
+	return ret;
+}
+
+static void io_eventfd_put(struct rcu_head *rcu)
+{
+	struct io_ev_fd *ev_fd = container_of(rcu, struct io_ev_fd, rcu);
+	struct io_ring_ctx *ctx = ev_fd->ctx;
+
+	eventfd_ctx_put(ev_fd->cq_ev_fd);
+	kfree(ev_fd);
+	rcu_assign_pointer(ctx->io_ev_fd, NULL);
 }
 
 static int io_eventfd_unregister(struct io_ring_ctx *ctx)
 {
-	if (ctx->cq_ev_fd) {
-		eventfd_ctx_put(ctx->cq_ev_fd);
-		ctx->cq_ev_fd = NULL;
+	struct io_ev_fd *ev_fd;
+
+	mutex_lock(&ctx->ev_fd_lock);
+	ev_fd = rcu_dereference_protected(ctx->io_ev_fd, lockdep_is_held(&ctx->ev_fd_lock));
+	if (ev_fd) {
+		ev_fd->unregistering = true;
+		call_rcu(&ev_fd->rcu, io_eventfd_put);
+		mutex_unlock(&ctx->ev_fd_lock);
 		return 0;
 	}
 
+	mutex_unlock(&ctx->ev_fd_lock);
 	return -ENXIO;
 }
 
@@ -10960,6 +11028,8 @@ static bool io_register_op_must_quiesce(int op)
 	case IORING_REGISTER_FILES:
 	case IORING_UNREGISTER_FILES:
 	case IORING_REGISTER_FILES_UPDATE:
+	case IORING_REGISTER_EVENTFD:
+	case IORING_UNREGISTER_EVENTFD:
 	case IORING_REGISTER_PROBE:
 	case IORING_REGISTER_PERSONALITY:
 	case IORING_UNREGISTER_PERSONALITY:
-- 
2.25.1