Merge tag 'for-6.16/io_uring-20250523' of git://git.kernel.dk/linux

Pull io_uring updates from Jens Axboe:

 - Avoid indirect function calls in io-wq for executing and freeing
   work.

   The design of io-wq is such that it can be a generic mechanism, but
   as it's just used by io_uring now, may as well avoid these indirect
   calls

 - Clean up registered buffers for networking

 - Add support for IORING_OP_PIPE. Pretty straight forward, allows
   creating pipes with io_uring, particularly useful for having these be
   instantiated as direct descriptors

 - Clean up the coalescing support fore registered buffers

 - Add support for multiple interface queues for zero-copy rx
   networking. As this feature was merged for 6.15 it supported just a
   single ifq per ring

 - Clean up the eventfd support

 - Add dma-buf support to zero-copy rx

 - Clean up and improving the request draining support

 - Clean up provided buffer support, most notably with an eye toward
   making the legacy support less intrusive

 - Minor fdinfo cleanups, dropping support for dumping what credentials
   are registered

 - Improve support for overflow CQE handling, getting rid of GFP_ATOMIC
   for allocating overflow entries where possible

 - Improve detection of cases where io-wq doesn't need to spawn a new
   worker unnecessarily

 - Various little cleanups

* tag 'for-6.16/io_uring-20250523' of git://git.kernel.dk/linux: (59 commits)
  io_uring/cmd: warn on reg buf imports by ineligible cmds
  io_uring/io-wq: only create a new worker if it can make progress
  io_uring/io-wq: ignore non-busy worker going to sleep
  io_uring/io-wq: move hash helpers to the top
  trace/io_uring: fix io_uring_local_work_run ctx documentation
  io_uring: finish IOU_OK -> IOU_COMPLETE transition
  io_uring: add new helpers for posting overflows
  io_uring: pass in struct io_big_cqe to io_alloc_ocqe()
  io_uring: make io_alloc_ocqe() take a struct io_cqe pointer
  io_uring: split alloc and add of overflow
  io_uring: open code io_req_cqe_overflow()
  io_uring/fdinfo: get rid of dumping credentials
  io_uring/fdinfo: only compile if CONFIG_PROC_FS is set
  io_uring/kbuf: unify legacy buf provision and removal
  io_uring/kbuf: refactor __io_remove_buffers
  io_uring/kbuf: don't compute size twice on prep
  io_uring/kbuf: drop extra vars in io_register_pbuf_ring
  io_uring/kbuf: use mem_is_zero()
  io_uring/kbuf: account ring io_buffer_list memory
  io_uring: drain based on allocates reqs
  ...

42 files changed, 944 insertions, 703 deletions

diff --git a/io_uring/Makefile b/io_uring/Makefile
index 3e28a741ca15..d97c6b51d584 100644
--- a/io_uring/Makefile
+++ b/io_uring/Makefile
@@ -7,11 +7,11 @@ GCOV_PROFILE := y
 endif
 
 obj-$(CONFIG_IO_URING)		+= io_uring.o opdef.o kbuf.o rsrc.o notif.o \
-					tctx.o filetable.o rw.o net.o poll.o \
+					tctx.o filetable.o rw.o poll.o \
 					eventfd.o uring_cmd.o openclose.o \
 					sqpoll.o xattr.o nop.o fs.o splice.o \
 					sync.o msg_ring.o advise.o openclose.o \
-					statx.o timeout.o fdinfo.o cancel.o \
+					statx.o timeout.o cancel.o \
 					waitid.o register.o truncate.o \
 					memmap.o alloc_cache.o
 obj-$(CONFIG_IO_URING_ZCRX)	+= zcrx.o
@@ -19,3 +19,5 @@ obj-$(CONFIG_IO_WQ)		+= io-wq.o
 obj-$(CONFIG_FUTEX)		+= futex.o
 obj-$(CONFIG_EPOLL)		+= epoll.o
 obj-$(CONFIG_NET_RX_BUSY_POLL)	+= napi.o
+obj-$(CONFIG_NET) += net.o cmd_net.o
+obj-$(CONFIG_PROC_FS) += fdinfo.o
diff --git a/io_uring/advise.c b/io_uring/advise.c
index cb7b881665e5..0073f74e3658 100644
--- a/io_uring/advise.c
+++ b/io_uring/advise.c
@@ -58,7 +58,7 @@ int io_madvise(struct io_kiocb *req, unsigned int issue_flags)
 
 	ret = do_madvise(current->mm, ma->addr, ma->len, ma->advice);
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 #else
 	return -EOPNOTSUPP;
 #endif
@@ -104,5 +104,5 @@ int io_fadvise(struct io_kiocb *req, unsigned int issue_flags)
 	if (ret < 0)
 		req_set_fail(req);
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
diff --git a/io_uring/cancel.c b/io_uring/cancel.c
index 0870060bac7c..6d57602304df 100644
--- a/io_uring/cancel.c
+++ b/io_uring/cancel.c
@@ -229,7 +229,7 @@ done:
 	if (ret < 0)
 		req_set_fail(req);
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
 
 static int __io_sync_cancel(struct io_uring_task *tctx,
diff --git a/io_uring/cmd_net.c b/io_uring/cmd_net.c
new file mode 100644
index 000000000000..e99170c7d41a
--- /dev/null
+++ b/io_uring/cmd_net.c
@@ -0,0 +1,83 @@
+#include <asm/ioctls.h>
+#include <linux/io_uring/net.h>
+#include <net/sock.h>
+
+#include "uring_cmd.h"
+
+static inline int io_uring_cmd_getsockopt(struct socket *sock,
+					  struct io_uring_cmd *cmd,
+					  unsigned int issue_flags)
+{
+	const struct io_uring_sqe *sqe = cmd->sqe;
+	bool compat = !!(issue_flags & IO_URING_F_COMPAT);
+	int optlen, optname, level, err;
+	void __user *optval;
+
+	level = READ_ONCE(sqe->level);
+	if (level != SOL_SOCKET)
+		return -EOPNOTSUPP;
+
+	optval = u64_to_user_ptr(READ_ONCE(sqe->optval));
+	optname = READ_ONCE(sqe->optname);
+	optlen = READ_ONCE(sqe->optlen);
+
+	err = do_sock_getsockopt(sock, compat, level, optname,
+				 USER_SOCKPTR(optval),
+				 KERNEL_SOCKPTR(&optlen));
+	if (err)
+		return err;
+
+	/* On success, return optlen */
+	return optlen;
+}
+
+static inline int io_uring_cmd_setsockopt(struct socket *sock,
+					  struct io_uring_cmd *cmd,
+					  unsigned int issue_flags)
+{
+	const struct io_uring_sqe *sqe = cmd->sqe;
+	bool compat = !!(issue_flags & IO_URING_F_COMPAT);
+	int optname, optlen, level;
+	void __user *optval;
+	sockptr_t optval_s;
+
+	optval = u64_to_user_ptr(READ_ONCE(sqe->optval));
+	optname = READ_ONCE(sqe->optname);
+	optlen = READ_ONCE(sqe->optlen);
+	level = READ_ONCE(sqe->level);
+	optval_s = USER_SOCKPTR(optval);
+
+	return do_sock_setsockopt(sock, compat, level, optname, optval_s,
+				  optlen);
+}
+
+int io_uring_cmd_sock(struct io_uring_cmd *cmd, unsigned int issue_flags)
+{
+	struct socket *sock = cmd->file->private_data;
+	struct sock *sk = sock->sk;
+	struct proto *prot = READ_ONCE(sk->sk_prot);
+	int ret, arg = 0;
+
+	if (!prot || !prot->ioctl)
+		return -EOPNOTSUPP;
+
+	switch (cmd->cmd_op) {
+	case SOCKET_URING_OP_SIOCINQ:
+		ret = prot->ioctl(sk, SIOCINQ, &arg);
+		if (ret)
+			return ret;
+		return arg;
+	case SOCKET_URING_OP_SIOCOUTQ:
+		ret = prot->ioctl(sk, SIOCOUTQ, &arg);
+		if (ret)
+			return ret;
+		return arg;
+	case SOCKET_URING_OP_GETSOCKOPT:
+		return io_uring_cmd_getsockopt(sock, cmd, issue_flags);
+	case SOCKET_URING_OP_SETSOCKOPT:
+		return io_uring_cmd_setsockopt(sock, cmd, issue_flags);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+EXPORT_SYMBOL_GPL(io_uring_cmd_sock);
diff --git a/io_uring/epoll.c b/io_uring/epoll.c
index 6d2c48ba1923..8d4610246ba0 100644
--- a/io_uring/epoll.c
+++ b/io_uring/epoll.c
@@ -61,7 +61,7 @@ int io_epoll_ctl(struct io_kiocb *req, unsigned int issue_flags)
 	if (ret < 0)
 		req_set_fail(req);
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
 
 int io_epoll_wait_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
@@ -88,5 +88,5 @@ int io_epoll_wait(struct io_kiocb *req, unsigned int issue_flags)
 		req_set_fail(req);
 
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
diff --git a/io_uring/eventfd.c b/io_uring/eventfd.c
index 100d5da94cb9..78f8ab7db104 100644
--- a/io_uring/eventfd.c
+++ b/io_uring/eventfd.c
@@ -47,13 +47,6 @@ static void io_eventfd_do_signal(struct rcu_head *rcu)
 	io_eventfd_put(ev_fd);
 }
 
-static void io_eventfd_release(struct io_ev_fd *ev_fd, bool put_ref)
-{
-	if (put_ref)
-		io_eventfd_put(ev_fd);
-	rcu_read_unlock();
-}
-
 /*
  * Returns true if the caller should put the ev_fd reference, false if not.
  */
@@ -72,63 +65,34 @@ static bool __io_eventfd_signal(struct io_ev_fd *ev_fd)
 
 /*
  * Trigger if eventfd_async isn't set, or if it's set and the caller is
- * an async worker. If ev_fd isn't valid, obviously return false.
+ * an async worker.
  */
 static bool io_eventfd_trigger(struct io_ev_fd *ev_fd)
 {
-	if (ev_fd)
-		return !ev_fd->eventfd_async || io_wq_current_is_worker();
-	return false;
+	return !ev_fd->eventfd_async || io_wq_current_is_worker();
 }
 
-/*
- * On success, returns with an ev_fd reference grabbed and the RCU read
- * lock held.
- */
-static struct io_ev_fd *io_eventfd_grab(struct io_ring_ctx *ctx)
+void io_eventfd_signal(struct io_ring_ctx *ctx, bool cqe_event)
 {
+	bool skip = false;
 	struct io_ev_fd *ev_fd;
 
 	if (READ_ONCE(ctx->rings->cq_flags) & IORING_CQ_EVENTFD_DISABLED)
-		return NULL;
-
-	rcu_read_lock();
+		return;
 
-	/*
-	 * rcu_dereference ctx->io_ev_fd once and use it for both for checking
-	 * and eventfd_signal
-	 */
+	guard(rcu)();
 	ev_fd = rcu_dereference(ctx->io_ev_fd);
-
 	/*
 	 * Check again if ev_fd exists in case 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 (io_eventfd_trigger(ev_fd) && refcount_inc_not_zero(&ev_fd->refs))
-		return ev_fd;
-
-	rcu_read_unlock();
-	return NULL;
-}
-
-void io_eventfd_signal(struct io_ring_ctx *ctx)
-{
-	struct io_ev_fd *ev_fd;
-
-	ev_fd = io_eventfd_grab(ctx);
-	if (ev_fd)
-		io_eventfd_release(ev_fd, __io_eventfd_signal(ev_fd));
-}
-
-void io_eventfd_flush_signal(struct io_ring_ctx *ctx)
-{
-	struct io_ev_fd *ev_fd;
-
-	ev_fd = io_eventfd_grab(ctx);
-	if (ev_fd) {
-		bool skip, put_ref = true;
+	if (!ev_fd)
+		return;
+	if (!io_eventfd_trigger(ev_fd) || !refcount_inc_not_zero(&ev_fd->refs))
+		return;
 
+	if (cqe_event) {
 		/*
 		 * Eventfd should only get triggered when at least one event
 		 * has been posted. Some applications rely on the eventfd
@@ -142,12 +106,10 @@ void io_eventfd_flush_signal(struct io_ring_ctx *ctx)
 		skip = ctx->cached_cq_tail == ev_fd->last_cq_tail;
 		ev_fd->last_cq_tail = ctx->cached_cq_tail;
 		spin_unlock(&ctx->completion_lock);
-
-		if (!skip)
-			put_ref = __io_eventfd_signal(ev_fd);
-
-		io_eventfd_release(ev_fd, put_ref);
 	}
+
+	if (skip || __io_eventfd_signal(ev_fd))
+		io_eventfd_put(ev_fd);
 }
 
 int io_eventfd_register(struct io_ring_ctx *ctx, void __user *arg,
diff --git a/io_uring/eventfd.h b/io_uring/eventfd.h
index d394f49c6321..e2f1985c2cf9 100644
--- a/io_uring/eventfd.h
+++ b/io_uring/eventfd.h
@@ -4,5 +4,4 @@ int io_eventfd_register(struct io_ring_ctx *ctx, void __user *arg,
 			unsigned int eventfd_async);
 int io_eventfd_unregister(struct io_ring_ctx *ctx);
 
-void io_eventfd_flush_signal(struct io_ring_ctx *ctx);
-void io_eventfd_signal(struct io_ring_ctx *ctx);
+void io_eventfd_signal(struct io_ring_ctx *ctx, bool cqe_event);
diff --git a/io_uring/fdinfo.c b/io_uring/fdinfo.c
index e0d6a59a89fa..e9355276ab5d 100644
--- a/io_uring/fdinfo.c
+++ b/io_uring/fdinfo.c
@@ -15,37 +15,6 @@
 #include "cancel.h"
 #include "rsrc.h"
 
-#ifdef CONFIG_PROC_FS
-static __cold int io_uring_show_cred(struct seq_file *m, unsigned int id,
-		const struct cred *cred)
-{
-	struct user_namespace *uns = seq_user_ns(m);
-	struct group_info *gi;
-	kernel_cap_t cap;
-	int g;
-
-	seq_printf(m, "%5d\n", id);
-	seq_put_decimal_ull(m, "\tUid:\t", from_kuid_munged(uns, cred->uid));
-	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->euid));
-	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->suid));
-	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->fsuid));
-	seq_put_decimal_ull(m, "\n\tGid:\t", from_kgid_munged(uns, cred->gid));
-	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->egid));
-	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->sgid));
-	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->fsgid));
-	seq_puts(m, "\n\tGroups:\t");
-	gi = cred->group_info;
-	for (g = 0; g < gi->ngroups; g++) {
-		seq_put_decimal_ull(m, g ? " " : "",
-					from_kgid_munged(uns, gi->gid[g]));
-	}
-	seq_puts(m, "\n\tCapEff:\t");
-	cap = cred->cap_effective;
-	seq_put_hex_ll(m, NULL, cap.val, 16);
-	seq_putc(m, '\n');
-	return 0;
-}
-
 #ifdef CONFIG_NET_RX_BUSY_POLL
 static __cold void common_tracking_show_fdinfo(struct io_ring_ctx *ctx,
 					       struct seq_file *m,
@@ -214,14 +183,6 @@ static void __io_uring_show_fdinfo(struct io_ring_ctx *ctx, struct seq_file *m)
 		else
 			seq_printf(m, "%5u: <none>\n", i);
 	}
-	if (!xa_empty(&ctx->personalities)) {
-		unsigned long index;
-		const struct cred *cred;
-
-		seq_printf(m, "Personalities:\n");
-		xa_for_each(&ctx->personalities, index, cred)
-			io_uring_show_cred(m, index, cred);
-	}
 
 	seq_puts(m, "PollList:\n");
 	for (i = 0; i < (1U << ctx->cancel_table.hash_bits); i++) {
@@ -264,4 +225,3 @@ __cold void io_uring_show_fdinfo(struct seq_file *m, struct file *file)
 		mutex_unlock(&ctx->uring_lock);
 	}
 }
-#endif
diff --git a/io_uring/fs.c b/io_uring/fs.c
index eccea851dd5a..37079a414eab 100644
--- a/io_uring/fs.c
+++ b/io_uring/fs.c
@@ -90,7 +90,7 @@ int io_renameat(struct io_kiocb *req, unsigned int issue_flags)
 
 	req->flags &= ~REQ_F_NEED_CLEANUP;
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
 
 void io_renameat_cleanup(struct io_kiocb *req)
@@ -141,7 +141,7 @@ int io_unlinkat(struct io_kiocb *req, unsigned int issue_flags)
 
 	req->flags &= ~REQ_F_NEED_CLEANUP;
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
 
 void io_unlinkat_cleanup(struct io_kiocb *req)
@@ -185,7 +185,7 @@ int io_mkdirat(struct io_kiocb *req, unsigned int issue_flags)
 
 	req->flags &= ~REQ_F_NEED_CLEANUP;
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
 
 void io_mkdirat_cleanup(struct io_kiocb *req)
@@ -235,7 +235,7 @@ int io_symlinkat(struct io_kiocb *req, unsigned int issue_flags)
 
 	req->flags &= ~REQ_F_NEED_CLEANUP;
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
 
 int io_linkat_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
@@ -281,7 +281,7 @@ int io_linkat(struct io_kiocb *req, unsigned int issue_flags)
 
 	req->flags &= ~REQ_F_NEED_CLEANUP;
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
 
 void io_link_cleanup(struct io_kiocb *req)
diff --git a/io_uring/futex.c b/io_uring/futex.c
index 0ea4820cd8ff..b34695022baa 100644
--- a/io_uring/futex.c
+++ b/io_uring/futex.c
@@ -234,7 +234,7 @@ int io_futexv_wait(struct io_kiocb *req, unsigned int issue_flags)
 		kfree(futexv);
 		req->async_data = NULL;
 		req->flags &= ~REQ_F_ASYNC_DATA;
-		return IOU_OK;
+		return IOU_COMPLETE;
 	}
 
 	/*
@@ -311,7 +311,7 @@ done:
 		req_set_fail(req);
 	io_req_set_res(req, ret, 0);
 	kfree(ifd);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
 
 int io_futex_wake(struct io_kiocb *req, unsigned int issue_flags)
@@ -328,5 +328,5 @@ int io_futex_wake(struct io_kiocb *req, unsigned int issue_flags)
 	if (ret < 0)
 		req_set_fail(req);
 	io_req_set_res(req, ret, 0);
-	return IOU_OK;
+	return IOU_COMPLETE;
 }
diff --git a/io_uring/io-wq.c b/io_uring/io-wq.c
index 04a75d666195..cd1fcb115739 100644
--- a/io_uring/io-wq.c
+++ b/io_uring/io-wq.c
@@ -114,9 +114,6 @@ enum {
 struct io_wq {
 	unsigned long state;
 
-	free_work_fn *free_work;
-	io_wq_work_fn *do_work;
-
 	struct io_wq_hash *hash;
 
 	atomic_t worker_refs;
@@ -153,6 +150,16 @@ static bool io_acct_cancel_pending_work(struct io_wq *wq,
 static void create_worker_cb(struct callback_head *cb);
 static void io_wq_cancel_tw_create(struct io_wq *wq);
 
+static inline unsigned int __io_get_work_hash(unsigned int work_flags)
+{
+	return work_flags >> IO_WQ_HASH_SHIFT;
+}
+
+static inline unsigned int io_get_work_hash(struct io_wq_work *work)
+{
+	return __io_get_work_hash(atomic_read(&work->flags));
+}
+
 static bool io_worker_get(struct io_worker *worker)
 {
 	return refcount_inc_not_zero(&worker->ref);
@@ -412,6 +419,30 @@ fail:
 	return false;
 }
 
+/* Defer if current and next work are both hashed to the same chain */
+static bool io_wq_hash_defer(struct io_wq_work *work, struct io_wq_acct *acct)
+{
+	unsigned int hash, work_flags;
+	struct io_wq_work *next;
+
+	lockdep_assert_held(&acct->lock);
+
+	work_flags = atomic_read(&work->flags);
+	if (!__io_wq_is_hashed(work_flags))
+		return false;
+
+	/* should not happen, io_acct_run_queue() said we had work */
+	if (wq_list_empty(&acct->work_list))
+		return true;
+
+	hash = __io_get_work_hash(work_flags);
+	next = container_of(acct->work_list.first, struct io_wq_work, list);
+	work_flags = atomic_read(&next->flags);
+	if (!__io_wq_is_hashed(work_flags))
+		return false;
+	return hash == __io_get_work_hash(work_flags);
+}
+
 static void io_wq_dec_running(struct io_worker *worker)
 {
 	struct io_wq_acct *acct = io_wq_get_acct(worker);
@@ -422,8 +453,14 @@ static void io_wq_dec_running(struct io_worker *worker)
 
 	if (!atomic_dec_and_test(&acct->nr_running))
 		return;
+	if (!worker->cur_work)
+		return;
 	if (!io_acct_run_queue(acct))
 		return;
+	if (io_wq_hash_defer(worker->cur_work, acct)) {
+		raw_spin_unlock(&acct->lock);
+		return;
+	}
 
 	raw_spin_unlock(&acct->lock);
 	atomic_inc(&acct->nr_running);
@@ -457,16 +494,6 @@ static void __io_worker_idle(struct io_wq_acct *acct, struct io_worker *worker)
 	}
 }
 
-static inline unsigned int __io_get_work_hash(unsigned int work_flags)
-{
-	return work_flags >> IO_WQ_HASH_SHIFT;
-}
-
-static inline unsigned int io_get_work_hash(struct io_wq_work *work)
-{
-	return __io_get_work_hash(atomic_read(&work->flags));
-}
-
 static bool io_wait_on_hash(struct io_wq *wq, unsigned int hash)
 {
 	bool ret = false;
@@ -612,10 +639,10 @@ static void io_worker_handle_work(struct io_wq_acct *acct,
 			if (do_kill &&
 			    (work_flags & IO_WQ_WORK_UNBOUND))
 				atomic_or(IO_WQ_WORK_CANCEL, &work->flags);
-			wq->do_work(work);
+			io_wq_submit_work(work);
 			io_assign_current_work(worker, NULL);
 
-			linked = wq->free_work(work);
+			linked = io_wq_free_work(work);
 			work = next_hashed;
 			if (!work && linked && !io_wq_is_hashed(linked)) {
 				work = linked;
@@ -934,8 +961,8 @@ static void io_run_cancel(struct io_wq_work *work, struct io_wq *wq)
 {
 	do {
 		atomic_or(IO_WQ_WORK_CANCEL, &work->flags);
-		wq->do_work(work);
-		work = wq->free_work(work);
+		io_wq_submit_work(work);
+		work = io_wq_free_work(work);
 	} while (work);
 }
 
@@ -1195,8 +1222,6 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
 	int ret, i;
 	struct io_wq *wq;
 
-	if (WARN_ON_ONCE(!data->free_work || !data->do_work))
-		return ERR_PTR(-EINVAL);
 	if (WARN_ON_ONCE(!bounded))
 		return ERR_PTR(-EINVAL);
 
@@ -1206,8 +1231,6 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
 
 	refcount_inc(&data->hash->refs);
 	wq->hash = data->hash;
-	wq->free_work = data->free_work;
-	wq->do_work = data->do_work;
 
 	ret = -ENOMEM;
 
diff --git a/io_uring/io-wq.h b/io_uring/io-wq.h
index d4fb2940e435..774abab54732 100644
--- a/io_uring/io-wq.h
+++ b/io_uring/io-wq.h
@@ -21,9 +21,6 @@ enum io_wq_cancel {
 	IO_WQ_CANCEL_NOTFOUND,	/* work not found */
 };
 
-typedef struct io_wq_work *(free_work_fn)(struct io_wq_work *);
-typedef void (io_wq_work_fn)(struct io_wq_work *);
-
 struct io_wq_hash {
 	refcount_t refs;
 	unsigned long map;
@@ -39,8 +36,6 @@ static inline void io_wq_put_hash(struct io_wq_hash *hash)
 struct io_wq_data {
 	struct io_wq_hash *hash;
 	struct task_struct *task;
-	io_wq_work_fn *do_work;
-	free_work_fn *free_work;
 };
 
 struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data);
diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index 46373549a733..c7a9cecf528e 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -129,7 +129,6 @@
 struct io_defer_entry {
 	struct list_head	list;
 	struct io_kiocb		*req;
-	u32			seq;
 };
 
 /* requests with any of those set should undergo io_disarm_next() */
@@ -149,6 +148,7 @@ static bool io_uring_try_cancel_requests(struct io_ring_ctx *ctx,
 					 bool is_sqpoll_thread);
 
 static void io_queue_sqe(struct io_kiocb *req);
+static void __io_req_caches_free(struct io_ring_ctx *ctx);
 
 static __read_mostly DEFINE_STATIC_KEY_FALSE(io_key_has_sqarray);
 
@@ -359,6 +359,7 @@ static __cold struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	INIT_LIST_HEAD(&ctx->tctx_list);
 	ctx->submit_state.free_list.next = NULL;
 	INIT_HLIST_HEAD(&ctx->waitid_list);
+	xa_init_flags(&ctx->zcrx_ctxs, XA_FLAGS_ALLOC);
 #ifdef CONFIG_FUTEX
 	INIT_HLIST_HEAD(&ctx->futex_list);
 #endif
@@ -380,25 +381,6 @@ err:
 	return NULL;
 }
 
-static void io_account_cq_overflow(struct io_ring_ctx *ctx)
-{
-	struct io_rings *r = ctx->rings;
-
-	WRITE_ONCE(r->cq_overflow, READ_ONCE(r->cq_overflow) + 1);
-	ctx->cq_extra--;
-}
-
-static bool req_need_defer(struct io_kiocb *req, u32 seq)
-{
-	if (unlikely(req->flags & REQ_F_IO_DRAIN)) {
-		struct io_ring_ctx *ctx = req->ctx;
-
-		return seq + READ_ONCE(ctx->cq_extra) != ctx->cached_cq_tail;
-	}
-
-	return false;
-}
-
 static void io_clean_op(struct io_kiocb *req)
 {
 	if (unlikely(req->flags & REQ_F_BUFFER_SELECTED))
@@ -537,20 +519,37 @@ void io_req_queue_iowq(struct io_kiocb *req)
 	io_req_task_work_add(req);
 }
 
+static unsigned io_linked_nr(struct io_kiocb *req)
+{
+	struct io_kiocb *tmp;
+	unsigned nr = 0;
+
+	io_for_each_link(tmp, req)
+		nr++;
+	return nr;
+}
+
 static __cold noinline void io_queue_deferred(struct io_ring_ctx *ctx)
 {
-	spin_lock(&ctx->completion_lock);
+	bool drain_seen = false, first = true;
+
+	lockdep_assert_held(&ctx->uring_lock);
+	__io_req_caches_free(ctx);
+
 	while (!list_empty(&ctx->defer_list)) {
 		struct io_defer_entry *de = list_first_entry(&ctx->defer_list,
 						struct io_defer_entry, list);
 
-		if (req_need_defer(de->req, de->seq))
-			break;
+		drain_seen |= de->req->flags & REQ_F_IO_DRAIN;
+		if ((drain_seen || first) && ctx->nr_req_allocated != ctx->nr_drained)
+			return;
+
 		list_del_init(&de->list);
+		ctx->nr_drained -= io_linked_nr(de->req);
 		io_req_task_queue(de->req);
 		kfree(de);
+		first = false;
 	}
-	spin_unlock(&ctx->completion_lock);
 }
 
 void __io_commit_cqring_flush(struct io_ring_ctx *ctx)
@@ -559,10 +558,8 @@ void __io_commit_cqring_flush(struct io_ring_ctx *ctx)
 		io_poll_wq_wake(ctx);
 	if (ctx->off_timeout_used)
 		io_flush_timeouts(ctx);
-	if (ctx->drain_active)
-		io_queue_deferred(ctx);<