Merge branch 'for-6.16/io_uring' into for-nextfor-next

* for-6.16/io_uring: (53 commits) 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 io_uring: count allocated requests io_uring: open code io_account_cq_overflow() io_uring: consolidate drain seq checking io_uring: remove drain prealloc checks io_uring: simplify drain ret passing io_uring: fix spurious drain flushing ...
author: Jens Axboe <axboe@kernel.dk> 2025-05-17 18:47:41 -0600
committer: Jens Axboe <axboe@kernel.dk> 2025-05-17 18:47:41 -0600
commit: f6873e3f089c6a386df3dac965abb5516c2eccda (patch)
tree: 53d49ad886d76202f08cd41d40f71bafa7064b54
parent: c2d8018325b6f0d0940c2544678822e88e41adbc (diff)
parent: f660fd2ca15a3743f65f6110ae60d5b80500d856 (diff)
download: linux-block-for-next.tar.gz
30 files changed, 866 insertions, 657 deletions
diff --git a/include/linux/io_uring_types.h b/include/linux/io_uring_types.h
index b44d201520d857..2922635986f521 100644
--- a/include/linux/io_uring_types.h
+++ b/include/linux/io_uring_types.h
@@ -40,8 +40,6 @@ enum io_uring_cmd_flags {
 	IO_URING_F_TASK_DEAD		= (1 << 13),
 };
 
-struct io_zcrx_ifq;
-
 struct io_wq_work_node {
 	struct io_wq_work_node *next;
 };
@@ -343,7 +341,6 @@ struct io_ring_ctx {
 		unsigned		cached_cq_tail;
 		unsigned		cq_entries;
 		struct io_ev_fd	__rcu	*io_ev_fd;
-		unsigned		cq_extra;
 
 		void			*cq_wait_arg;
 		size_t			cq_wait_size;
@@ -394,7 +391,8 @@ struct io_ring_ctx {
 	struct wait_queue_head		poll_wq;
 	struct io_restriction		restrictions;
 
-	struct io_zcrx_ifq		*ifq;
+	/* Stores zcrx object pointers of type struct io_zcrx_ifq */
+	struct xarray			zcrx_ctxs;
 
 	u32			pers_next;
 	struct xarray		personalities;
@@ -418,6 +416,7 @@ struct io_ring_ctx {
 
 	struct callback_head		poll_wq_task_work;
 	struct list_head		defer_list;
+	unsigned			nr_drained;
 
 	struct io_alloc_cache		msg_cache;
 	spinlock_t			msg_lock;
@@ -436,6 +435,7 @@ struct io_ring_ctx {
 
 	/* protected by ->completion_lock */
 	unsigned			evfd_last_cq_tail;
+	unsigned			nr_req_allocated;
 
 	/*
 	 * Protection for resize vs mmap races - both the mmap and resize
@@ -448,8 +448,6 @@ struct io_ring_ctx {
 	struct io_mapped_region		ring_region;
 	/* used for optimised request parameter and wait argument passing  */
 	struct io_mapped_region		param_region;
-	/* just one zcrx per ring for now, will move to io_zcrx_ifq eventually */
-	struct io_mapped_region		zcrx_region;
 };
 
 /*
@@ -653,8 +651,7 @@ struct io_kiocb {
 	u8				iopoll_completed;
 	/*
 	 * Can be either a fixed buffer index, or used with provided buffers.
-	 * For the latter, before issue it points to the buffer group ID,
-	 * and after selection it points to the buffer ID itself.
+	 * For the latter, it points to the selected buffer ID.
 	 */
 	u16				buf_index;
 
@@ -713,7 +710,7 @@ struct io_kiocb {
 	const struct cred		*creds;
 	struct io_wq_work		work;
 
-	struct {
+	struct io_big_cqe {
 		u64			extra1;
 		u64			extra2;
 	} big_cqe;
diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
index 50e372ea97c549..cfd17e382082fa 100644
--- a/include/uapi/linux/io_uring.h
+++ b/include/uapi/linux/io_uring.h
@@ -73,6 +73,7 @@ struct io_uring_sqe {
 		__u32		futex_flags;
 		__u32		install_fd_flags;
 		__u32		nop_flags;
+		__u32		pipe_flags;
 	};
 	__u64	user_data;	/* data to be passed back at completion time */
 	/* pack this to avoid bogus arm OABI complaints */
@@ -287,6 +288,7 @@ enum io_uring_op {
 	IORING_OP_EPOLL_WAIT,
 	IORING_OP_READV_FIXED,
 	IORING_OP_WRITEV_FIXED,
+	IORING_OP_PIPE,
 
 	/* this goes last, obviously */
 	IORING_OP_LAST,
@@ -992,12 +994,16 @@ struct io_uring_zcrx_offsets {
 	__u64	__resv[2];
 };
 
+enum io_uring_zcrx_area_flags {
+	IORING_ZCRX_AREA_DMABUF		= 1,
+};
+
 struct io_uring_zcrx_area_reg {
 	__u64	addr;
 	__u64	len;
 	__u64	rq_area_token;
 	__u32	flags;
-	__u32	__resv1;
+	__u32	dmabuf_fd;
 	__u64	__resv2[2];
 };
 
diff --git a/io_uring/Makefile b/io_uring/Makefile
index 3e28a741ca15c3..d97c6b51d58491 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/cmd_net.c b/io_uring/cmd_net.c
new file mode 100644
index 00000000000000..e99170c7d41add
--- /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/eventfd.c b/io_uring/eventfd.c
index 100d5da94cb95c..78f8ab7db104f0 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 d394f49c632105..e2f1985c2cf98e 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 e0d6a59a89fa1b..e9355276ab5d52 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/io-wq.c b/io_uring/io-wq.c
index 04a75d66619510..d52069b1177b0a 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;
@@ -612,10 +609,10 @@ 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 +931,8 @@ 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 +1192,6 @@ 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 +1201,6 @@ 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 d4fb2940e435f7..774abab54732ef 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 3f079713914154..d85657ecb8933e 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);
 	if (ctx->has_evfd)
-		io_eventfd_flush_signal(ctx);
+		io_eventfd_signal(ctx, true);
 }
 
 static inline void __io_cq_lock(struct io_ring_ctx *ctx)
@@ -700,27 +697,20 @@ static __cold void io_uring_drop_tctx_refs(struct task_struct *task)
 	}
 }
 
-static bool io_cqring_event_overflow(struct io_ring_ctx *ctx, u64 user_data,
-				     s32 res, u32 cflags, u64 extra1, u64 extra2)
+static __cold bool io_cqring_add_overflow(struct io_ring_ctx *ctx,
+					  struct io_overflow_cqe *ocqe)
 {
-	struct io_overflow_cqe *ocqe;
-	size_t ocq_size = sizeof(struct io_overflow_cqe);
-	bool is_cqe32 = (ctx->flags & IORING_SETUP_CQE32);
-
 	lockdep_assert_held(&ctx->completion_lock);
 
-	if (is_cqe32)
-		ocq_size += sizeof(struct io_uring_cqe);
-
-	ocqe = kmalloc(ocq_size, GFP_ATOMIC | __GFP_ACCOUNT);
-	trace_io_uring_cqe_overflow(ctx, user_data, res, cflags, ocqe);
 	if (!ocqe) {
+		struct io_rings *r = ctx->rings;
+
 		/*
 		 * If we're in ring overflow flush mode, or in task cancel mode,
 		 * or cannot allocate an overflow entry, then we need to drop it
 		 * on the floor.
 		 */
-		io_account_cq_overflow(ctx);
+		WRITE_ONCE(r->cq_overflow, READ_ONCE(r->cq_overflow) + 1);
 		set_bit(IO_CHECK_CQ_DROPPED_BIT, &ctx->check_cq);
 		return false;
 	}
@@ -729,23 +719,35 @@ static bool io_cqring_event_overflow(struct io_ring_ctx *ctx, u64 user_data,
 		atomic_or(IORING_SQ_CQ_OVERFLOW, &ctx->rings->sq_flags);
 
 	}
-	ocqe->cqe.user_data = user_data;
-	ocqe->cqe.res = res;
-	ocqe->cqe.flags = cflags;
-	if (is_cqe32) {
-		ocqe->cqe.big_cqe[0] = extra1;
-		ocqe->cqe.big_cqe[1] = extra2;
-	}
 	list_add_tail(&ocqe->list, &ctx->cq_overflow_list);
 	return true;
 }
 
-static void io_req_cqe_overflow(struct io_kiocb *req)
+static struct io_overflow_cqe *io_alloc_ocqe(struct io_ring_ctx *ctx,
+					     struct io_cqe *cqe,
+					     struct io_big_cqe *big_cqe, gfp_t gfp)
 {
-	io_cqring_event_overflow(req->ctx, req->cqe.user_data,
-				req->cqe.res, req->cqe.flags,
-				req->big_cqe.extra1, req->big_cqe.extra2);
-	memset(&req->big_cqe, 0, sizeof(req->big_cqe));
+	struct io_overflow_cqe *ocqe;
+	size_t ocq_size = sizeof(struct io_overflow_cqe);
+	bool is_cqe32 = (ctx->flags & IORING_SETUP_CQE32);
+
+	if (is_cqe32)
+		ocq_size += sizeof(struct io_uring_cqe);
+
+	ocqe = kzalloc(ocq_size, gfp | __GFP_ACCOUNT);
+	trace_io_uring_cqe_overflow(ctx, cqe->user_data, cqe->res, cqe->flags, ocqe);
+	if (ocqe) {
+		ocqe->cqe.user_data = cqe->user_data;
+		ocqe->cqe.res = cqe->res;
+		ocqe->cqe.flags = cqe->flags;
+		if (is_cqe32 && big_cqe) {
+			ocqe->cqe.big_cqe[0] = big_cqe->extra1;
+			ocqe->cqe.big_cqe[1] = big_cqe->extra2;
+		}
+	}
+	if (big_cqe)
+		big_cqe->extra1 = big_cqe->extra2 = 0;
+	return ocqe;
 }
 
 /*
@@ -790,13 +792,6 @@ static bool io_fill_cqe_aux(struct io_ring_ctx *ctx, u64 user_data, s32 res,
 {
 	struct io_uring_cqe *cqe;
 
-	ctx->cq_extra++;
-
-	/*
-	 * If we can't get a cq entry, userspace overflowed the
-	 * submission (by quite a lot). Increment the overflow count in
-	 * the ring.
-	 */
 	if (likely(io_get_cqe(ctx, &cqe))) {
 		WRITE_ONCE(cqe->user_data, user_data);
 		WRITE_ONCE(cqe->res, res);
@@ -813,14 +808,43 @@ static bool io_fill_cqe_aux(struct io_ring_ctx *ctx, u64 user_data, s32 res,
 	return false;
 }
 
+static inline struct io_cqe io_init_cqe(u64 user_data, s32 res, u32 cflags)
+{
+	return (struct io_cqe) { .user_data = user_data, .res = res, .flags = cflags };
+}
+
+static __cold void io_cqe_overflow(struct io_ring_ctx *ctx, struct io_cqe *cqe,
+			           struct io_big_cqe *big_cqe)
+{
+	struct io_overflow_cqe *ocqe;
+
+	ocqe = io_alloc_ocqe(ctx, cqe, big_cqe, GFP_KERNEL);
+	spin_lock(&ctx->completion_lock);
+	io_cqring_add_overflow(ctx, ocqe);
+	spin_unlock(&ctx->completion_lock);
+}
+
+static __cold bool io_cqe_overflow_locked(struct io_ring_ctx *ctx,
+					  struct io_cqe *cqe,
+					  struct io_big_cqe *big_cqe)
+{
+	struct io_overflow_cqe *ocqe;
+
+	ocqe = io_alloc_ocqe(ctx, cqe, big_cqe, GFP_ATOMIC);
+	return io_cqring_add_overflow(ctx, ocqe);
+}
+
 bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags)
 {
 	bool filled;
 
 	io_cq_lock(ctx);
 	filled = io_fill_cqe_aux(ctx, user_data, res, cflags);
-	if (!filled)
-		filled = io_cqring_event_overflow(ctx, user_data, res, cflags, 0, 0);
+	if (unlikely(!filled)) {
+		struct io_cqe cqe = io_init_cqe(user_data, res, cflags);
+
+		filled = io_cqe_overflow_locked(ctx, &cqe, NULL);
+	}
 	io_cq_unlock_post(ctx);
 	return filled;
 }
@@ -831,10 +855,13 @@ bool io_post_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags
  */
 void io_add_aux_cqe(struct io_ring_ctx *ctx, u64 user_data, s32 res, u32 cflags)
 {
+	lockdep_assert_held(&ctx->uring_lock);
+	lockdep_assert(ctx->lockless_cq);
+
 	if (!io_fill_cqe_aux(ctx, user_data, res, cflags)) {
-		spin_lock(&ctx->completion_lock);
-		io_cqring_event_overflow(ctx, user_data, res, cflags, 0, 0);
-		spin_unlock(&ctx->completion_lock);
+		struct io_cqe cqe = io_init_cqe(user_data, res, cflags);
+
+		io_cqe_overflow(ctx, &cqe, NULL);
 	}
 	ctx->submit_state.cq_flush = true;
 }
@@ -924,22 +951,6 @@ void io_req_defer_failed(struct io_kiocb *req, s32 res)
 }
 
 /*
- * Don't initialise the fields below on every allocation, but do that in
- * advance and keep them valid across allocations.
- */
-static void io_preinit_req(struct io_kiocb *req, struct io_ring_ctx *ctx)
-{
-	req->ctx = ctx;
-	req->buf_node = NULL;
-	req->file_node = NULL;
-	req->link = NULL;
-	req->async_data = NULL;
-	/* not necessary, but safer to zero */
-	memset(&req->cqe, 0, sizeof(req->cqe));
-	memset(&req->big_cqe, 0, sizeof(req->big_cqe));
-}
-
-/*
  * A request might get retired back into the request caches even before opcode
  * handlers and io_issue_sqe() are done with it, e.g. inline completion path.
  * Because of that, io_alloc_req() should be called only under ->uring_lock
@@ -948,7 +959,7 @@ static void io_preinit_req(struct io_kiocb *req, struct io_ring_ctx *ctx)
 __cold bool __io_alloc_req_refill(struct io_ring_ctx *ctx)
 	__must_hold(&ctx->uring_lock)
 {
-	gfp_t gfp = GFP_KERNEL | __GFP_NOWARN;
+	gfp_t gfp = GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO;
 	void *reqs[IO_REQ_ALLOC_BATCH];
 	int ret;
 
@@ -966,10 +977,11 @@ __cold bool __io_alloc_req_refill(struct io_ring_ctx *ctx)
 	}
 
 	percpu_ref_get_many(&ctx->refs, ret);
+	ctx->nr_req_allocated += ret;
+
 	while (ret--) {
 		struct io_kiocb *req = reqs[ret];
 
-		io_preinit_req(req, ctx);
 		io_req_add_to_cache(req, ctx);
 	}
 	return true;
@@ -1191,7 +1203,7 @@ static void io_req_local_work_add(struct io_kiocb *req, unsigned flags)
 		if (ctx->flags & IORING_SETUP_TASKRUN_FLAG)
 			atomic_or(IORING_SQ_TASKRUN, &ctx->rings->sq_flags);
 		if (ctx->has_evfd)
-			io_eventfd_signal(ctx);
+			io_eventfd_signal(ctx, false);
 	}
 
 	nr_wait = atomic_read(&ctx->cq_wait_nr);
@@ -1383,6 +1395,16 @@ void io_queue_next(struct io_kiocb *req)
 		io_req_task_queue(nxt);
 }
 
+static inline void io_req_put_rsrc_nodes(struct io_kiocb *req)
+{
+	if (req->file_node) {
+		io_put_rsrc_node(req->ctx, req->file_node);
+		req->file_node = NULL;
+	}
+	if (req->flags & REQ_F_BUF_NODE)
+		io_put_rsrc_node(req->ctx, req->buf_node);
+}
+
 static void io_free_batch_list(struct io_ring_ctx *ctx,
 			       struct io_wq_work_node *node)
 	__must_hold(&ctx->uring_lock)
@@ -1443,13 +1465,10 @@ void __io_submit_flush_completions(struct io_ring_ctx *ctx)
 		 */
 		if (!(req->flags & (REQ_F_CQE_SKIP | REQ_F_REISSUE)) &&
 		    unlikely(!io_fill_cqe_req(ctx, req))) {
-			if (ctx->lockless_cq) {
-				spin_lock(&ctx->completion_lock);
-				io_req_cqe_overflow(req);
-				spin_unlock(&ctx->completion_lock);
-			} else {
-				io_req_cqe_overflow(req);
-			}
+			if (ctx->lockless_cq)
+				io_cqe_overflow(ctx, &req->cqe, &req->big_cqe);
+			else
+				io_cqe_overflow_locked(ctx, &req->cqe, &req->big_cqe);
 		}
 	}
 	__io_cq_unlock_post(ctx);
@@ -1458,6 +1477,10 @@ void __io_submit_flush_completions(struct io_ring_ctx *ctx)
 		io_free_batch_list(ctx, state->compl_reqs.first);
 		INIT_WQ_LIST(&state->compl_reqs);
 	}
+
+	if (unlikely(ctx->drain_active))
+		io_queue_deferred(ctx);
+
 	ctx->submit_state.cq_flush = false;
 }
 
@@ -1645,56 +1668,28 @@ io_req_flags_t io_file_get_flags(struct file *file)
 	return res;
 }
 
-static u32 io_get_sequence(struct io_kiocb *req)
-{
-	u32 seq = req->ctx->cached_sq_head;
-	struct io_kiocb *cur;
-
-	/* need original cached_sq_head, but it was increased for each req */
-	io_for_each_link(cur, req)
-		seq--;
-	return seq;
-}
-
 static __cold void io_drain_req(struct io_kiocb *req)
 	__must_hold(&ctx->uring_lock)
 {
 	struct io_ring_ctx *ctx = req->ctx;
+	bool drain = req->flags & IOSQE_IO_DRAIN;
 	struct io_defer_entry *de;
-	int ret;
-	u32 seq = io_get_sequence(req);
-
-	/* Still need defer if there is pending req in defer list. */
-	spin_lock(&ctx->completion_lock);
-	if (!req_need_defer(req, seq) && list_empty_careful(&ctx->defer_list)) {
-		spin_unlock(&ctx->completion_lock);
-queue:
-		ctx->drain_active = false;
-		io_req_task_queue(req);
-		return;
-	}
-	spin_unlock(&ctx->completion_lock);
 
-	io_prep_async_link(req);
-	de = kmalloc(sizeof(*de), GFP_KERNEL);
+	de = kmalloc(sizeof(*de), GFP_KERNEL_ACCOUNT);
 	if (!de) {
-		ret = -ENOMEM;
-		io_req_defer_failed(req, ret);
+		io_req_defer_failed(req, -ENOMEM);
 		return;
 	}
 
-	spin_lock(&ctx->completion_lock);
-	if (!req_need_defer(req, seq) && list_empty(&ctx->defer_list)) {
-		spin_unlock(&ctx->completion_lock);
-		kfree(de);
-		goto queue;
-	}
-
+	io_prep_async_link(req);
 	trace_io_uring_defer(req);
 	de->req = req;
-	de->seq = seq;
+
+	ctx->nr_drained += io_linked_nr(req);
 	list_add_tail(&de->list, &ctx->defer_list);
-	spin_unlock(&ctx->completion_lock);
+	io_queue_deferred(ctx);
+	if (!drain && list_empty(&ctx->defer_list))
+		ctx->drain_active = false;
 }
 
 static bool io_assign_file(struct io_kiocb *req, const struct io_issue_def *def,
@@ -1815,7 +1810,7 @@ void io_wq_submit_work(struct io_wq_work *work)
 	bool needs_poll = false;
 	int ret = 0, err = -ECANCELED;
 
-	/* one will be dropped by ->io_wq_free_work() after returning to io-wq */
+	/* one will be dropped by io_wq_free_work() after returning to io-wq */
 	if (!(req->flags & REQ_F_REFCOUNT))
 		__io_req_set_refcount(req, 2);
 	else
@@ -1913,7 +1908,8 @@ inline struct file *io_file_get_fixed(struct io_kiocb *req, int fd,
 	io_ring_submit_lock(ctx, issue_flags);
 	node = io_rsrc_node_lookup(&ctx->file_table.data, fd);
 	if (node) {
-		io_req_assign_rsrc_node(&req->file_node, node);
+		node->refs++;
+		req->file_node = node;
 		req->flags |= io_slot_flags(node);
 		file = io_slot_file(node);
 	}
@@ -2046,7 +2042,7 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req,
 	int personality;
 	u8 opcode;
 
-	/* req is partially pre-initialised, see io_preinit_req() */
+	req->ctx = ctx;
 	req->opcode = opcode = READ_ONCE(sqe->opcode);
 	/* same numerical values with corresponding REQ_F_*, safe to copy */
 	sqe_flags = READ_ONCE(sqe->flags);
@@ -2277,10 +2273,6 @@ static bool io_get_sqe(struct io_ring_ctx *ctx, const struct io_uring_sqe **sqe)
 	    (!(ctx->flags & IORING_SETUP_NO_SQARRAY))) {
 		head = READ_ONCE(ctx->sq_array[head]);
 		if (unlikely(head >= ctx->sq_entries)) {
-			/* drop invalid entries */
-			spin_lock(&ctx->completion_lock);
-			ctx->cq_extra--;
-			spin_unlock(&ctx->completion_lock);
 			WRITE_ONCE(ctx->rings->sq_dropped,
 				   READ_ONCE(ctx->rings->sq_dropped) + 1);
 			return false;
@@ -2698,21 +2690,26 @@ unsigned long rings_size(unsigned int flags, unsigned int sq_entries,
 	return off;
 }
 
-static void io_req_caches_free(struct io_ring_ctx *ctx)
+static __cold void __io_req_caches_free(struct io_ring_ctx *ctx)
 {
 	struct io_kiocb *req;
 	int nr = 0;
 
-	mutex_lock(&ctx->uring_lock);
-
 	while (!io_req_cache_empty(ctx)) {
 		req = io_extract_req(ctx);
 		kmem_cache_free(req_cachep, req);
 		nr++;
 	}
-	if (nr)
+	if (nr) {
+		ctx->nr_req_allocated -= nr;
 		percpu_ref_put_many(&ctx->refs, nr);
-	mutex_unlock(&ctx->uring_lock);
+	}
+}
+
+static __cold void io_req_caches_free(struct io_ring_ctx *ctx)
+{
+	guard(mutex)(&ctx->uring_lock);
+	__io_req_caches_free(ctx);
 }
 
 static __cold void io_ring_ctx_free(struct io_ring_ctx *ctx)
@@ -2748,6 +2745,9 @@ static __cold void io_ring_ctx_free(struct io_ring_ctx *ctx)
 	percpu_ref_exit(&ctx->refs);
 	free_uid(ctx->user);
 	io_req_caches_free(ctx);
+
+	WARN_ON_ONCE(ctx->nr_req_allocated);
+
 	if (ctx->hash_map)
 		io_wq_put_hash(ctx->hash_map);
 	io_napi_free(ctx);
@@ -2882,7 +2882,7 @@ static __cold void io_ring_exit_work(struct work_struct *work)
 			io_cqring_overflow_kill(ctx);
 			mutex_unlock(&ctx->uring_lock);
 		}
-		if (ctx->ifq) {
+		if (!xa_empty(&ctx->zcrx_ctxs)) {
 			mutex_lock(&ctx->uring_lock);
 			io_shutdown_zcrx_ifqs(ctx);
 			mutex_unlock(&ctx->uring_lock);
@@ -3014,20 +3014,19 @@ static __cold bool io_cancel_defer_files(struct io_ring_ctx *ctx,
 	struct io_defer_entry *de;
 	LIST_HEAD(list);
 
-	spin_lock(&ctx->completion_lock);
 	list_for_each_entry_reverse(de, &ctx->defer_list, list) {
 		if (io_match_task_safe(de->req, tctx, cancel_all)) {
 			list_cut_position(&list, &ctx->defer_list, &de->list);
 			break;
 		}
 	}
-	spin_unlock(&ctx->completion_lock);
 	if (list_empty(&list))
 		return false;
 
 	while (!list_empty(&list)) {
 		de = list_first_entry(&list, struct io_defer_entry, list);
 		list_del_init(&de->list);
+		ctx->nr_drained -= io_linked_nr(de->req);
 		io_req_task_queue_fail(de->req, -ECANCELED);
 		kfree(de);
 	}
@@ -3102,8 +3101,8 @@ static __cold bool io_uring_try_cancel_requests(struct io_ring_ctx *ctx,
 	if ((ctx->flags & IORING_SETUP_DEFER_TASKRUN) &&
 	    io_allowed_defer_tw_run(ctx))
 		ret |= io_run_local_work(ctx, INT_MAX, INT_MAX) > 0;
-	ret |= io_cancel_defer_files(ctx, tctx, cancel_all);
 	mutex_lock(&ctx->uring_lock);
+	ret |= io_cancel_defer_files(ctx, tctx, cancel_all);
 	ret |= io_poll_remove_all(ctx, tctx, cancel_all);
 	ret |= io_waitid_remove_all(ctx, tctx, cancel_all);
 	ret |= io_futex_remove_all(ctx, tctx, cancel_all);
diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h
index e4050b2d0821f5..81f22196a57de6 100644
--- a/io_uring/io_uring.h
+++ b/io_uring/io_uring.h
@@ -196,7 +196,6 @@ static inline bool io_defer_get_uncommited_cqe(struct io_ring_ctx *ctx,
 {
 	io_lockdep_assert_cq_locked(ctx);
 
-	ctx->cq_extra++;
 	ctx->submit_state.cq_flush = true;
 	return io_get_cqe(ctx, cqe_ret);
 }
@@ -414,7 +413,7 @@ static inline void io_req_complete_defer(struct io_kiocb *req)
 
 static inline void io_commit_cqring_flush(struct io_ring_ctx *ctx)
 {
-	if (unlikely(ctx->off_timeout_used || ctx->drain_active ||
+	if (unlikely(ctx->off_timeout_used ||
 		     ctx->has_evfd || ctx->poll_activated))
 		__io_commit_cqring_flush(ctx);
 }
diff --git a/io_uring/kbuf.c b/io_uring/kbuf.c
index 953d5e74256916..823e7eb15fb2a8 100644
--- a/io_uring/kbuf.c
+++ b/io_uring/kbuf.c
@@ -92,7 +92,6 @@ void io_kbuf_drop_legacy(struct io_kiocb *req)
 {
 	if (WARN_ON_ONCE(!(req->flags & REQ_F_BUFFER_SELECTED)))
 		return;
-	req->buf_index = req->kbuf->bgid;
 	req->flags &= ~REQ_F_BUFFER_SELECTED;
 	kfree(req->kbuf);
 	req->kbuf = NULL;
@@ -110,7 +109,6 @@ bool io_kbuf_recycle_legacy(struct io_kiocb *req, unsigned issue_flags)
 	bl = io_buffer_get_list(ctx, buf->bgid);
 	list_add(&buf->list, &bl->buf_list);
 	req->flags &= ~REQ_F_BUFFER_SELECTED;
-	req->buf_index = buf->bgid;
 
 	io_ring_submit_unlock(ctx, issue_flags);
 	return true;
@@ -193,7 +191,7 @@ static void __user *io_ring_buffer_select(struct io_kiocb *req, size_t *len,
 }
 
 void __user *io_buffer_select(struct io_kiocb *req, size_t *len,
-			      unsigned int issue_flags)
+			      unsigned buf_group, unsigned int issue_flags)
 {
 	struct io_ring_ctx *ctx = req->ctx;
 	struct io_buffer_list *bl;
@@ -201,7 +199,7 @@ void __user *io_buffer_select(struct io_kiocb *req, size_t *len,
 
 	io_ring_submit_lock(req->ctx, issue_flags);
 
-	bl = io_buffer_get_list(ctx, req->buf_index);
+	bl = io_buffer_get_list(ctx, buf_group);
 	if (likely(bl)) {
 		if (bl->flags & IOBL_BUF_RING)
 			ret = io_ring_buffer_select(req, len, bl, issue_flags);
@@ -302,7 +300,7 @@ int io_buffers_select(struct io_kiocb *req, struct buf_sel_arg *arg,
 	int ret = -ENOENT;
 
 	io_ring_submit_lock(ctx, issue_flags);
-	bl = io_buffer_get_list(ctx, req->buf_index);
+	bl = io_buffer_get_list(ctx, arg->buf_group);
 	if (unlikely(!bl))
 		goto out_unlock;
 
@@ -335,7 +333,7 @@ int io_buffers_peek(struct io_kiocb *req, struct buf_sel_arg *arg)
 
 	lockdep_assert_held(&ctx->uring_lock);
 
-	bl = io_buffer_get_list(ctx, req->buf_index);
+	bl = io_buffer_get_list(ctx, arg->buf_group);
 	if (unlikely(!bl))
 		return -ENOENT;
 
@@ -355,10 +353,9 @@ static inline bool __io_put_kbuf_ring(struct io_kiocb *req, int len, int nr)
 	struct io_buffer_list *bl = req->buf_list;
 	bool ret = true;
 
-	if (bl) {
+	if (bl)
 		ret = io_kbuf_commit(req, bl, len, nr);
-		req->buf_index = bl->bgid;
-	}
+
 	req->flags &= ~REQ_F_BUFFER_RING;
 	return ret;
 }
@@ -379,45 +376,33 @@ unsigned int __io_put_kbufs(struct io_kiocb *req, int len, int nbufs)
 	return ret;
 }
 
-static int __io_remove_buffers(struct io_ring_ctx *ctx,
-			       struct io_buffer_list *bl, unsigned nbufs)
+static int io_remove_buffers_legacy(struct io_ring_ctx *ctx,
+				    struct io_buffer_list *bl,
+				    unsigned long nbufs)
 {
-	unsigned i = 0;
-
-	/* shouldn't happen */
-	if (!nbufs)
-		return 0;
-
-	if (bl->flags & IOBL_BUF_RING) {
-		i = bl->buf_ring->tail - bl->head;
-		io_free_region(ctx, &bl->region);
-		/* make sure it's seen as empty */
-		INIT_LIST_HEAD(&bl->buf_list);
-		bl->flags &= ~IOBL_BUF_RING;
-		return i;
-	}
+	unsigned long i = 0;
+	struct io_buffer *nxt;
 
 	/* protects io_buffers_cache */
 	lockdep_assert_held(&ctx->uring_lock);
+	WARN_ON_ONCE(bl->flags & IOBL_BUF_RING);
 
-	while (!list_empty(&bl->buf_list)) {
-		struct io_buffer *nxt;
-
+	for (i = 0; i < nbufs && !list_empty(&bl->buf_list); i++) {
 		nxt = list_first_entry(&bl->buf_list, struct io_buffer, list);
 		list_del(&nxt->list);
 		kfree(nxt);
-
-		if (++i == nbufs)
-			return i;
 		cond_resched();
 	}
-
 	return i;
 }
 
 static void io_put_bl(struct io_ring_ctx *ctx, struct io_buffer_list *bl)
 {
-	__io_remove_buffers(ctx, bl, -1U);
+	if (bl->flags & IOBL_BUF_RING)
+		io_free_region(ctx, &bl->region);
+	else
+		io_remove_buffers_legacy(ctx, bl, -1U);
+
 	kfree(bl);
 }
 
@@ -465,30 +450,6 @@ int io_remove_buffers_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 	return 0;
 }
 
-int io_remove_buffers(struct io_kiocb *req, unsigned int issue_flags)
-{
-	struct io_provide_buf *p = io_kiocb_to_cmd(req, struct io_provide_buf);
-	struct io_ring_ctx *ctx = req->ctx;
-	struct io_buffer_list *bl;
-	int ret = 0;
-
-	io_ring_submit_lock(ctx, issue_flags);
-
-	ret = -ENOENT;
-	bl = io_buffer_get_list(ctx, p->bgid);
-	if (bl) {
-		ret = -EINVAL;
-		/* can't use provide/remove buffers command on mapped buffers */
-		if (!(bl->flags & IOBL_BUF_RING))
-			ret = __io_remove_buffers(ctx, bl, p->nbufs);
-	}
-	io_ring_submit_unlock(ctx, issue_flags);
-	if (ret < 0)
-		req_set_fail(req);
-	io_req_set_res(req, ret, 0);
-	return IOU_OK;
-}
-
 int io_provide_buffers_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
 	unsigned long size, tmp_check;
@@ -512,8 +473,6 @@ int io_provide_buffers_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe
 		return -EOVERFLOW;
 	if (check_add_overflow((unsigned long)p->addr, size, &tmp_check))
 		return -EOVERFLOW;
-
-	size = (unsigned long)p->len * p->nbufs;
 	if (!access_ok(u64_to_user_ptr(p->addr), size))
 		return -EFAULT;
 
@@ -552,37 +511,44 @@ static int io_add_buffers(struct io_ring_ctx *ctx, struct io_provide_buf *pbuf,
 	return i ? 0 : -ENOMEM;
 }
 
-int io_provide_buffers(struct io_kiocb *req, unsigned int issue_flags)
+static int __io_manage_buffers_legacy(struct io_kiocb *req,
+					struct io_buffer_list *bl)
 {
 	struct io_provide_buf *p = io_kiocb_to_cmd(req, struct io_provide_buf);
-	struct io_ring_ctx *ctx = req->ctx;
-	struct io_buffer_list *bl;
-	int ret = 0;
-
-	io_ring_submit_lock(ctx, issue_flags);
+	int ret;
 
-	bl = io_buffer_get_list(ctx, p->bgid);
-	if (unlikely(!bl)) {
+	if (!bl) {
+		if (req->opcode != IORING_OP_PROVIDE_BUFFERS)
+			return -ENOENT;
 		bl = kzalloc(sizeof(*bl), GFP_KERNEL_ACCOUNT);
-		if (!bl) {
-			ret = -ENOMEM;
-			goto err;
-		}
+		if (!bl)
+			return -ENOMEM;
+
 		INIT_LIST_HEAD(&bl->buf_list);
-		ret = io_buffer_add_list(ctx, bl, p->bgid);
+		ret = io_buffer_add_list(req->ctx, bl, p->bgid);
 		if (ret) {
 			kfree(bl);
-			goto err;
+			return ret;
 		}
 	}
-	/* can't add buffers via this command for a mapped buffer ring */
-	if (bl->flags & IOBL_BUF_RING) {
-		ret = -EINVAL;
-		goto err;
-	}
+	/* can't use provide/remove buffers command on mapped buffers */
+	if (bl->flags & IOBL_BUF_RING)
+		return -EINVAL;
+	if (req->opcode == IORING_OP_PROVIDE_BUFFERS)
+		return io_add_buffers(req->ctx, p, bl);
+	return io_remove_buffers_legacy(req->ctx, bl, p->nbufs);
+}
+
+int io_manage_buffers_legacy(struct io_kiocb *req, unsigned int issue_flags)
+{
+	struct io_provide_buf *p = io_kiocb_to_cmd(req, struct io_provide_buf);
+	struct io_ring_ctx *ctx = req->ctx;
+	struct io_buffer_list *bl;
+	int ret;
 
-	ret = io_add_buffers(ctx, p, bl);
-err:
+	io_ring_submit_lock(ctx, issue_flags);
+	bl = io_buffer_get_list(ctx, p->bgid);
+	ret = __io_manage_buffers_legacy(req, bl);
 	io_ring_submit_unlock(ctx, issue_flags);
 
 	if (ret < 0)
@@ -594,7 +560,7 @@ err:
 int io_register_pbuf_ring(struct io_ring_ctx *ctx, void __user *arg)
 {
 	struct io_uring_buf_reg reg;
-	struct io_buffer_list *bl, *free_bl = NULL;
+	struct io_buffer_list *bl;
 	struct io_uring_region_desc rd;
 	struct io_uring_buf_ring *br;
 	unsigned long mmap_offset;
@@ -605,8 +571,7 @@ int io_register_pbuf_ring(struct io_ring_ctx *ctx, void __user *arg)
 
 	if (copy_from_user(&reg, arg, sizeof(reg)))
 		return -EFAULT;
-
-	if (reg.resv[0] || reg.resv[1] || reg.resv[2])
+	if (!mem_is_zero(reg.resv, sizeof(reg.resv)))
 		return -EINVAL;
 	if (reg.flags & ~(IOU_PBUF_RING_MMAP | IOU_PBUF_RING_INC))
 		return -EINVAL;
@@ -624,7 +589,7 @@ int io_register_pbuf_ring(struct io_ring_ctx *ctx, void __user *arg)
 		io_destroy_bl(ctx, bl);
 	}
 
-	free_bl = bl = kzalloc(sizeof(*bl), GFP_KERNEL);
+	bl = kzalloc(sizeof(*bl), GFP_KERNEL_ACCOUNT);
 	if (!bl)
 		return -ENOMEM;
 
@@ -669,7 +634,7 @@ int io_register_pbuf_ring(struct io_ring_ctx *ctx, void __user *arg)
 	return 0;
 fail:
 	io_free_region(ctx, &bl->region);
-	kfree(free_bl);
+	kfree(bl);
 	return ret;
 }
 
@@ -682,9 +647,7 @@ int io_unregister_pbuf_ring(struct io_ring_ctx *ctx, void __user *arg)
 
 	if (copy_from_user(&reg, arg, sizeof(reg)))
 		return -EFAULT;
-	if (reg.resv[0] || reg.resv[1] || reg.resv[2])
-		return -EINVAL;
-	if (reg.flags)
+	if (!mem_is_zero(reg.resv, sizeof(reg.resv)) || reg.flags)
 		return -EINVAL;
 
 	bl = io_buffer_get_list(ctx, reg.bgid);
@@ -704,14 +667,11 @@ int io_register_pbuf_status(struct io_ring_ctx *ctx, void __user *arg)
 {
 	struct io_uring_buf_status buf_status;
 	struct io_buffer_list *bl;
-	int i;
 
 	if (copy_from_user(&buf_status, arg, sizeof(buf_status)))
 		return -EFAULT;
-
-	for (i = 0; i < ARRAY_SIZE(buf_status.resv); i++)
-		if (buf_status.resv[i])
-			return -EINVAL;
+	if (!mem_is_zero(buf_status.resv, sizeof(buf_status.resv)))
+		return -EINVAL;
 
 	bl = io_buffer_get_list(ctx, buf_status.buf_group);
 	if (!bl)
diff --git a/io_uring/kbuf.h b/io_uring/kbuf.h
index 2ec0b983ce243c..4d2c209d1a4112 100644
--- a/io_uring/kbuf.h
+++ b/io_uring/kbuf.h
@@ -55,20 +55,19 @@ struct buf_sel_arg {
 	size_t max_len;
 	unsigned short nr_iovs;
 	unsigned short mode;
+	unsigned buf_group;
 };
 
 void __user *io_buffer_select(struct io_kiocb *req, size_t *len,
-			      unsigned int issue_flags);
+			      unsigned buf_group, unsigned int issue_flags);
 int io_buffers_select(struct io_kiocb *req, struct buf_sel_arg *arg,
 		      unsigned int issue_flags);
 int io_buffers_peek(struct io_kiocb *req, struct buf_sel_arg *arg);
 void io_destroy_buffers(struct io_ring_ctx *ctx);
 
 int io_remove_buffers_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
-int io_remove_buffers(struct io_kiocb *req, unsigned int issue_flags);
-
 int io_provide_buffers_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
-int io_provide_buffers(struct io_kiocb *req, unsigned int issue_flags);
+int io_manage_buffers_legacy(struct io_kiocb *req, unsigned int issue_flags);
 
 int io_register_pbuf_ring(struct io_ring_ctx *ctx, void __user *arg);
 int io_unregister_pbuf_ring(struct io_ring_ctx *ctx, void __user *arg);
@@ -94,7 +93,6 @@ static inline bool io_kbuf_recycle_ring(struct io_kiocb *req)
 	 * to monopolize the buffer.
 	 */
 	if (req->buf_list) {
-		req->buf_index = req->buf_list->bgid;
 		req->flags &= ~(REQ_F_BUFFER_RING|REQ_F_BUFFERS_COMMIT);
 		return true;
 	}
diff --git a/io_uring/memmap.c b/io_uring/memmap.c
index 07f8a5cbd37ec7..725dc0bec24c42 100644
--- a/io_uring/memmap.c
+++ b/io_uring/memmap.c
@@ -13,6 +13,7 @@
 #include "memmap.h"
 #include "kbuf.h"
 #include "rsrc.h"
+#include "zcrx.h"
 
 static void *io_mem_alloc_compound(struct page **pages, int nr_pages,
 				   size_t size, gfp_t gfp)
@@ -258,7 +259,8 @@ static struct io_mapped_region *io_mmap_get_region(struct io_ring_ctx *ctx,
 						   loff_t pgoff)
 {
 	loff_t offset = pgoff << PAGE_SHIFT;
-	unsigned int bgid;
+	unsigned int id;
+
 
 	switch (offset & IORING_OFF_MMAP_MASK) {
 	case IORING_OFF_SQ_RING:
@@ -267,12 +269,13 @@ static struct io_mapped_region *io_mmap_get_region(struct io_ring_ctx *ctx,
 	case IORING_OFF_SQES:
 		return &ctx->sq_region;
 	case IORING_OFF_PBUF_RING:
-		bgid = (offset & ~IORING_OFF_MMAP_MASK) >> IORING_OFF_PBUF_SHIFT;
-		return io_pbuf_get_region(ctx, bgid);
+		id = (offset & ~IORING_OFF_MMAP_MASK) >> IORING_OFF_PBUF_SHIFT;
+		return io_pbuf_get_region(ctx, id);
 	case IORING_MAP_OFF_PARAM_REGION:
 		return &ctx->param_region;
 	case IORING_MAP_OFF_ZCRX_REGION:
-		return &ctx->zcrx_region;
+		id = (offset & ~IORING_OFF_MMAP_MASK) >> IORING_OFF_ZCRX_SHIFT;
+		return io_zcrx_get_region(ctx, id);
 	}
 	return NULL;
 }
diff --git a/io_uring/memmap.h b/io_uring/memmap.h
index dad0aa5b1b4574..08419684e4bc91 100644
--- a/io_uring/memmap.h
+++ b/io_uring/memmap.h
@@ -4,7 +4,9 @@
 #define IORING_MAP_OFF_PARAM_REGION		0x20000000ULL
 #define IORING_MAP_OFF_ZCRX_REGION		0x30000000ULL
 
-struct page **io_pin_pages(unsigned long ubuf, unsigned long len, int *npages);
+#define IORING_OFF_ZCRX_SHIFT		16
+
+struct page **io_pin_pages(unsigned long uaddr, unsigned long len, int *npages);
 
 #ifndef CONFIG_MMU
 unsigned int io_uring_nommu_mmap_capabilities(struct file *file);
diff --git a/io_uring/net.c b/io_uring/net.c
index 24040bc3916a1b..1fbdb2bbb3f3fb 100644
--- a/io_uring/net.c
+++ b/io_uring/net.c
@@ -18,7 +18,6 @@
 #include "rsrc.h"
 #include "zcrx.h"
 
-#if defined(CONFIG_NET)
 struct io_shutdown {
 	struct file			*file;
 	int				how;
@@ -190,7 +189,6 @@ static inline void io_mshot_prep_retry(struct io_kiocb *req,
 	sr->done_io = 0;
 	sr->retry = false;
 	sr->len = 0; /* get from the provided buffer */
-	req->buf_index = sr->buf_group;
 }
 
 static int io_net_import_vec(struct io_kiocb *req, struct io_async_msghdr *iomsg,
@@ -359,15 +357,13 @@ static int io_send_setup(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 		kmsg->msg.msg_name = &kmsg->addr;
 		kmsg->msg.msg_namelen = addr_len;
 	}
-	if (sr->flags & IORING_RECVSEND_FIXED_BUF)
+	if (sr->flags & IORING_RECVSEND_FIXED_BUF) {
+		req->flags |= REQ_F_IMPORT_BUFFER;
 		return 0;
-	if (!io_do_buffer_select(req)) {
-		ret = import_ubuf(ITER_SOURCE, sr->buf, sr->len,
-				  &kmsg->msg.msg_iter);
-		if (unlikely(ret < 0))
-			return ret;
 	}
-	return 0;
+	if (req->flags & REQ_F_BUFFER_SELECT)
+		return 0;
+	return import_ubuf(ITER_SOURCE, sr->buf, sr->len, &kmsg->msg.msg_iter);
 }
 
 static int io_sendmsg_setup(struct io_kiocb *req, const struct io_uring_sqe *sqe)
@@ -409,13 +405,12 @@ int io_sendmsg_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 	sr->msg_flags = READ_ONCE(sqe->msg_flags) | MSG_NOSIGNAL;
 	if (sr->msg_flags & MSG_DONTWAIT)
 		req->flags |= REQ_F_NOWAIT;
+	if (req->flags & REQ_F_BUFFER_SELECT)
+		sr->buf_group = req->buf_index;
 	if (sr->flags & IORING_RECVSEND_BUNDLE) {
 		if (req->opcode == IORING_OP_SENDMSG)
 			return -EINVAL;
-		if (!(req->flags & REQ_F_BUFFER_SELECT))
-			return -EINVAL;
 		sr->msg_flags |= MSG_WAITALL;
-		sr->buf_group = req->buf_index;
 		req->buf_list = NULL;
 		req->flags |= REQ_F_MULTISHOT;
 	}
@@ -571,6 +566,7 @@ static int io_send_select_buffer(struct io_kiocb *req, unsigned int issue_flags,
 		.iovs = &kmsg->fast_iov,
 		.max_len = min_not_zero(sr->len, INT_MAX),
 		.nr_iovs = 1,
+		.buf_group = sr->buf_group,
 	};
 
 	if (kmsg->vec.iovec) {
@@ -723,7 +719,6 @@ static int io_recvmsg_prep_setup(struct io_kiocb *req)
 {
 	struct io_sr_msg *sr = io_kiocb_to_cmd(req, struct io_sr_msg);
 	struct io_async_msghdr *kmsg;
-	int ret;
 
 	kmsg = io_msg_alloc_async(req);
 	if (unlikely(!kmsg))
@@ -739,13 +734,10 @@ static int io_recvmsg_prep_setup(struct io_kiocb *req)
 		kmsg->msg.msg_iocb = NULL;
 		kmsg->msg.msg_ubuf = NULL;
 
-		if (!io_do_buffer_select(req)) {
-			ret = import_ubuf(ITER_DEST, sr->buf, sr->len,
-					  &kmsg->msg.msg_iter);
-			if (unlikely(ret))
-				return ret;
-		}
-		return 0;
+		if (req->flags & REQ_F_BUFFER_SELECT)
+			return 0;
+		return import_ubuf(ITER_DEST, sr->buf, sr->len,
+				   &kmsg->msg.msg_iter);
 	}
 
 	return io_recvmsg_copy_hdr(req, kmsg);
@@ -985,7 +977,7 @@ retry_multishot:
 		void __user *buf;
 		size_t len = sr->len;
 
-		buf = io_buffer_select(req, &len, issue_flags);
+		buf = io_buffer_select(req, &len, sr->buf_group, issue_flags);
 		if (!buf)
 			return -ENOBUFS;
 
@@ -1063,6 +1055,7 @@ static int io_recv_buf_select(struct io_kiocb *req, struct io_async_msghdr *kmsg
 			.iovs = &kmsg->fast_iov,
 			.nr_iovs = 1,
 			.mode = KBUF_MODE_EXPAND,
+			.buf_group = sr->buf_group,
 		};
 
 		if (kmsg->vec.iovec) {
@@ -1095,7 +1088,7 @@ static int io_recv_buf_select(struct io_kiocb *req, struct io_async_msghdr *kmsg
 		void __user *buf;
 
 		*len = sr->len;
-		buf = io_buffer_select(req, len, issue_flags);
+		buf = io_buffer_select(req, len, sr->buf_group, issue_flags);
 		if (!buf)
 			return -ENOBUFS;
 		sr->buf = buf;
@@ -1191,16 +1184,14 @@ int io_recvzc_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 	struct io_recvzc *zc = io_kiocb_to_cmd(req, struct io_recvzc);
 	unsigned ifq_idx;
 
-	if (unlikely(sqe->file_index || sqe->addr2 || sqe->addr ||
-		     sqe->addr3))
+	if (unlikely(sqe->addr2 || sqe->addr || sqe->addr3))
 		return -EINVAL;
 
 	ifq_idx = READ_ONCE(sqe->zcrx_ifq_idx);
-	if (ifq_idx != 0)
-		return -EINVAL;
-	zc->ifq = req->ctx->ifq;
+	zc->ifq = xa_load(&req->ctx->zcrx_ctxs, ifq_idx);
 	if (!zc->ifq)
 		return -EINVAL;
+
 	zc->len = READ_ONCE(sqe->len);
 	zc->flags = READ_ONCE(sqe->ioprio);
 	zc->msg_flags = READ_ONCE(sqe->msg_flags);
@@ -1321,8 +1312,6 @@ int io_send_zc_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 		return -ENOMEM;
 
 	if (req->opcode == IORING_OP_SEND_ZC) {
-		if (zc->flags & IORING_RECVSEND_FIXED_BUF)
-			req->flags |= REQ_F_IMPORT_BUFFER;
 		ret = io_send_setup(req, sqe);
 	} else {
 		if (unlikely(sqe->addr2 || sqe->file_index))
@@ -1846,4 +1835,3 @@ void io_netmsg_cache_free(const void *entry)
 	io_vec_free(&kmsg->vec);
 	kfree(kmsg);
 }
-#endif
diff --git a/io_uring/notif.c b/io_uring/notif.c
index 7bd92538dccbc7..9a6f6e92d74242 100644
--- a/io_uring/notif.c
+++ b/io_uring/notif.c
@@ -112,6 +112,7 @@ struct io_kiocb *io_alloc_notif(struct io_ring_ctx *ctx)
 
 	if (unlikely(!io_alloc_req(ctx, &notif)))
 		return NULL;
+	notif->ctx = ctx;
 	notif->opcode = IORING_OP_NOP;
 	notif->flags = 0;
 	notif->file = NULL;
diff --git a/io_uring/opdef.c b/io_uring/opdef.c
index 489384c0438bd8..6e0882b051f93b 100644
--- a/io_uring/opdef.c
+++ b/io_uring/opdef.c
@@ -333,13 +333,13 @@ const struct io_issue_def io_issue_defs[] = {
 		.audit_skip		= 1,
 		.iopoll			= 1,
 		.prep			= io_provide_buffers_prep,
-		.issue			= io_provide_buffers,
+		.issue			= io_manage_buffers_legacy,
 	},
 	[IORING_OP_REMOVE_BUFFERS] = {
 		.audit_skip		= 1,
 		.iopoll			= 1,
 		.prep			= io_remove_buffers_prep,
-		.issue			= io_remove_buffers,
+		.issue			= io_manage_buffers_legacy,
 	},
 	[IORING_OP_TEE] = {
 		.needs_file		= 1,
@@ -569,6 +569,10 @@ const struct io_issue_def io_issue_defs[] = {
 		.prep			= io_prep_writev_fixed,
 		.issue			= io_write,
 	},
+	[IORING_OP_PIPE] = {
+		.prep			= io_pipe_prep,
+		.issue			= io_pipe,
+	},
 };
 
 const struct io_cold_def io_cold_defs[] = {
@@ -815,6 +819,9 @@ const struct io_cold_def io_cold_defs[] = {
 		.cleanup		= io_readv_writev_cleanup,
 		.fail			= io_rw_fail,
 	},
+	[IORING_OP_PIPE] = {
+		.name			= "PIPE",
+	},
 };
 
 const char *io_uring_get_opcode(u8 opcode)
diff --git a/io_uring/openclose.c b/io_uring/openclose.c
index e3357dfa14ca42..4dd46116345783 100644
--- a/io_uring/openclose.c
+++ b/io_uring/openclose.c
@@ -6,6 +6,8 @@
 #include <linux/fdtable.h>
 #include <linux/fsnotify.h>
 #include <linux/namei.h>
+#include <linux/pipe_fs_i.h>
+#include <linux/watch_queue.h>
 #include <linux/io_uring.h>
 
 #include <uapi/linux/io_uring.h>
@@ -302,3 +304,134 @@ int io_install_fixed_fd(struct io_kiocb *req, unsigned int issue_flags)
 	io_req_set_res(req, ret, 0);
 	return IOU_OK;
 }
+
+struct io_pipe {
+	struct file *file;
+	int __user *fds;
+	int flags;
+	int file_slot;
+	unsigned long nofile;
+};
+
+int io_pipe_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
+{
+	struct io_pipe *p = io_kiocb_to_cmd(req, struct io_pipe);
+
+	if (sqe->fd || sqe->off || sqe->addr3)
+		return -EINVAL;
+
+	p->fds = u64_to_user_ptr(READ_ONCE(sqe->addr));
+	p->flags = READ_ONCE(sqe->pipe_flags);
+	if (p->flags & ~(O_CLOEXEC | O_NONBLOCK | O_DIRECT | O_NOTIFICATION_PIPE))
+		return -EINVAL;
+
+	p->file_slot = READ_ONCE(sqe->file_index);
+	p->nofile = rlimit(RLIMIT_NOFILE);
+	return 0;
+}
+
+static int io_pipe_fixed(struct io_kiocb *req, struct file **files,
+			 unsigned int issue_flags)
+{
+	struct io_pipe *p = io_kiocb_to_cmd(req, struct io_pipe);
+	struct io_ring_ctx *ctx = req->ctx;
+	int ret, fds[2] = { -1, -1 };
+	int slot = p->file_slot;
+
+	if (p->flags & O_CLOEXEC)
+		return -EINVAL;
+
+	io_ring_submit_lock(ctx, issue_flags);
+
+	ret = __io_fixed_fd_install(ctx, files[0], slot);
+	if (ret < 0)
+		goto err;
+	fds[0] = ret;
+	files[0] = NULL;
+
+	/*
+	 * If a specific slot is given, next one will be used for
+	 * the write side.
+	 */
+	if (slot != IORING_FILE_INDEX_ALLOC)
+		slot++;
+
+	ret = __io_fixed_fd_install(ctx, files[1], slot);
+	if (ret < 0)
+		goto err;
+	fds[1] = ret;
+	files[1] = NULL;
+
+	io_ring_submit_unlock(ctx, issue_flags);
+
+	if (!copy_to_user(p->fds, fds, sizeof(fds)))
+		return 0;
+
+	ret = -EFAULT;
+	io_ring_submit_lock(ctx, issue_flags);
+err:
+	if (fds[0] != -1)
+		io_fixed_fd_remove(ctx, fds[0]);
+	if (fds[1] != -1)
+		io_fixed_fd_remove(ctx, fds[1]);
+	io_ring_submit_unlock(ctx, issue_flags);
+	return ret;
+}
+
+static int io_pipe_fd(struct io_kiocb *req, struct file **files)
+{
+	struct io_pipe *p = io_kiocb_to_cmd(req, struct io_pipe);
+	int ret, fds[2] = { -1, -1 };
+
+	ret = __get_unused_fd_flags(p->flags, p->nofile);
+	if (ret < 0)
+		goto err;
+	fds[0] = ret;
+
+	ret = __get_unused_fd_flags(p->flags, p->nofile);
+	if (ret < 0)
+		goto err;
+	fds[1] = ret;
+
+	if (!copy_to_user(p->fds, fds, sizeof(fds))) {
+		fd_install(fds[0], files[0]);
+		fd_install(fds[1], files[1]);
+		return 0;
+	}
+	ret = -EFAULT;
+err:
+	if (fds[0] != -1)
+		put_unused_fd(fds[0]);
+	if (fds[1] != -1)
+		put_unused_fd(fds[1]);
+	return ret;
+}
+
+int io_pipe(struct io_kiocb *req, unsigned int issue_flags)
+{
+	struct io_pipe *p = io_kiocb_to_cmd(req, struct io_pipe);
+	struct file *files[2];
+	int ret;
+
+	ret = create_pipe_files(files, p->flags);
+	if (ret)
+		return ret;
+	files[0]->f_mode |= FMODE_NOWAIT;
+	files[1]->f_mode |= FMODE_NOWAIT;
+
+	if (!!p->file_slot)
+		ret = io_pipe_fixed(req, files, issue_flags);
+	else
+		ret = io_pipe_fd(req, files);
+
+	io_req_set_res(req, ret, 0);
+	if (!ret)
+		return IOU_OK;
+
+	req_set_fail(req);
+	if (files[0])
+		fput(files[0]);
+	if (files[1])
+		fput(files[1]);
+	return ret;
+}
diff --git a/io_uring/openclose.h b/io_uring/openclose.h
index 8a93c98ad0adc6..4ca2a9935abc9c 100644
--- a/io_uring/openclose.h
+++ b/io_uring/openclose.h
@@ -13,5 +13,8 @@ int io_openat2(struct io_kiocb *req, unsigned int issue_flags);
 int io_close_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
 int io_close(struct io_kiocb *req, unsigned int issue_flags);
 
+int io_pipe_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
+int io_pipe(struct io_kiocb *req, unsigned int issue_flags);
+
 int io_install_fixed_fd_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
 int io_install_fixed_fd(struct io_kiocb *req, unsigned int issue_flags);
diff --git a/io_uring/rsrc.c b/io_uring/rsrc.c
index f80a77c4973f30..1657d775c8bab2 100644
--- a/io_uring/rsrc.c
+++ b/io_uring/rsrc.c
@@ -80,10 +80,21 @@ static int io_account_mem(struct io_ring_ctx *ctx, unsigned long nr_pages)
 	return 0;
 }
 
-int io_buffer_validate(struct iovec *iov)
+int io_validate_user_buf_range(u64 uaddr, u64 ulen)
 {
-	unsigned long tmp, acct_len = iov->iov_len + (PAGE_SIZE - 1);
+	unsigned long tmp, base = (unsigned long)uaddr;
+	unsigned long acct_len = (unsigned long)PAGE_ALIGN(ulen);
 
+	/* arbitrary limit, but we need something */
+	if (ulen > SZ_1G || !ulen)
+		return -EFAULT;
+	if (check_add_overflow(base, acct_len, &tmp))
+		return -EOVERFLOW;
+	return 0;
+}
+
+static int io_buffer_validate(struct iovec *iov)
+{
 	/*
 	 * Don't impose further limits on the size and buffer
 	 * constraints here, we'll -EINVAL later when IO is
@@ -91,17 +102,9 @@ int io_buffer_validate(struct iovec *iov)
 	 */
 	if (!iov->iov_base)
 		return iov->iov_len ? -EFAULT : 0;
-	if (!iov->iov_len)
-		return -EFAULT;
-
-	/* arbitrary limit, but we need something */
-	if (iov->iov_len > SZ_1G)
-		return -EFAULT;
 
-	if (check_add_overflow((unsigned long)iov->iov_base, acct_len, &tmp))
-		return -EOVERFLOW;
-
-	return 0;
+	return io_validate_user_buf_range((unsigned long)iov->iov_base,
+					  iov->iov_len);
 }
 
 static void io_release_ubuf(void *priv)
@@ -685,38 +688,34 @@ static bool io_coalesce_buffer(struct page ***pages, int *nr_pages,
 				struct io_imu_folio_data *data)
 {
 	struct page **page_array = *pages, **new_array = NULL;
-	int nr_pages_left = *nr_pages, i, j;
-	int nr_folios = data->nr_folios;
+	unsigned nr_pages_left = *nr_pages;
+	unsigned nr_folios = data->nr_folios;
+	unsigned i, j;
 
 	/* Store head pages only*/
-	new_array = kvmalloc_array(nr_folios, sizeof(struct page *),
-					GFP_KERNEL);
+	new_array = kvmalloc_array(nr_folios, sizeof(struct page *), GFP_KERNEL);
 	if (!new_array)
 		return false;
 
-	new_array[0] = compound_head(page_array[0]);
-	/*
-	 * The pages are bound to the folio, it doesn't
-	 * actually unpin them but drops all but one reference,
-	 * which is usually put down by io_buffer_unmap().
-	 * Note, needs a better helper.
-	 */
-	if (data->nr_pages_head > 1)
-		unpin_user_pages(&page_array[1], data->nr_pages_head - 1);
-
-	j = data->nr_pages_head;
-	nr_pages_left -= data->nr_pages_head;
-	for (i = 1; i < nr_folios; i++) {
-		unsigned int nr_unpin;
-
-		new_array[i] = page_array[j];
-		nr_unpin = min_t(unsigned int, nr_pages_left - 1,
-					data->nr_pages_mid - 1);
-		if (nr_unpin)
-			unpin_user_pages(&page_array[j+1], nr_unpin);
-		j += data->nr_pages_mid;
-		nr_pages_left -= data->nr_pages_mid;
+	for (i = 0, j = 0; i < nr_folios; i++) {
+		struct page *p = compound_head(page_array[j]);
+		struct folio *folio = page_folio(p);
+		unsigned int nr;
+
+		WARN_ON_ONCE(i > 0 && p != page_array[j]);
+
+		nr = i ? data->nr_pages_mid : data->nr_pages_head;
+		nr = min(nr, nr_pages_left);
+		/* Drop all but one ref, the entire folio will remain pinned. */
+		if (nr > 1)
+			unpin_user_folio(folio, nr - 1);
+		j += nr;
+		nr_pages_left -= nr;
+		new_array[i] = p;
 	}
+
+	WARN_ON_ONCE(j != *nr_pages);
+
 	kvfree(page_array);
 	*pages = new_array;
 	*nr_pages = nr_folios;
@@ -1062,8 +1061,6 @@ static int io_import_fixed(int ddir, struct iov_iter *iter,
 	size_t offset;
 	int ret;
 
-	if (WARN_ON_ONCE(!imu))
-		return -EFAULT;
 	ret = validate_fixed_range(buf_addr, len, imu);
 	if (unlikely(ret))
 		return ret;
@@ -1110,13 +1107,19 @@ inline struct io_rsrc_node *io_find_buf_node(struct io_kiocb *req,
 
 	if (req->flags & REQ_F_BUF_NODE)
 		return req->buf_node;
+	req->flags |= REQ_F_BUF_NODE;
 
 	io_ring_submit_lock(ctx, issue_flags);
 	node = io_rsrc_node_lookup(&ctx->buf_table, req->buf_index);
-	if (node)
-		io_req_assign_buf_node(req, node);
+	if (node) {
+		node->refs++;
+		req->buf_node = node;
+		io_ring_submit_unlock(ctx, issue_flags);
+		return node;
+	}
+	req->flags &= ~REQ_F_BUF_NODE;
 	io_ring_submit_unlock(ctx, issue_flags);
-	return node;
+	return NULL;
 }
 
 int io_import_reg_buf(struct io_kiocb *req, struct iov_iter *iter,
diff --git a/io_uring/rsrc.h b/io_uring/rsrc.h
index b52242852ff342..0d2138f16322b2 100644
--- a/io_uring/rsrc.h
+++ b/io_uring/rsrc.h
@@ -83,7 +83,7 @@ int io_register_rsrc_update(struct io_ring_ctx *ctx, void __user *arg,
 			    unsigned size, unsigned type);
 int io_register_rsrc(struct io_ring_ctx *ctx, void __user *arg,
 			unsigned int size, unsigned int type);
-int io_buffer_validate(struct iovec *iov);
+int io_validate_user_buf_range(u64 uaddr, u64 ulen);
 
 bool io_check_coalesce_buffer(struct page **page_array, int nr_pages,
 			      struct io_imu_folio_data *data);
@@ -115,32 +115,6 @@ static inline bool io_reset_rsrc_node(struct io_ring_ctx *ctx,
 	return true;
 }
 
-static inline void io_req_put_rsrc_nodes(struct io_kiocb *req)
-{
-	if (req->file_node) {
-		io_put_rsrc_node(req->ctx, req->file_node);
-		req->file_node = NULL;
-	}
-	if (req->flags & REQ_F_BUF_NODE) {
-		io_put_rsrc_node(req->ctx, req->buf_node);
-		req->buf_node = NULL;
-	}
-}
-
-static inline void io_req_assign_rsrc_node(struct io_rsrc_node **dst_node,
-					   struct io_rsrc_node *node)
-{
-	node->refs++;
-	*dst_node = node;
-}
-
-static inline void io_req_assign_buf_node(struct io_kiocb *req,
-					  struct io_rsrc_node *node)
-{
-	io_req_assign_rsrc_node(&req->buf_node, node);
-	req->flags |= REQ_F_BUF_NODE;
-}
-
 int io_files_update(struct io_kiocb *req, unsigned int issue_flags);
 int io_files_update_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
 
diff --git a/io_uring/rw.c b/io_uring/rw.c
index b8389674a747fc..303fdded375851 100644
--- a/io_uring/rw.c
+++ b/io_uring/rw.c
@@ -119,7 +119,7 @@ static int __io_import_rw_buffer(int ddir, struct io_kiocb *req,
 		return io_import_vec(ddir, req, io, buf, sqe_len);
 
 	if (io_do_buffer_select(req)) {
-		buf = io_buffer_select(req, &sqe_len, issue_flags);
+		buf = io_buffer_select(req, &sqe_len, io->buf_group, issue_flags);
 		if (!buf)
 			return -ENOBUFS;
 		rw->addr = (unsigned long) buf;
@@ -253,16 +253,19 @@ static int __io_prep_rw(struct io_kiocb *req, const struct io_uring_sqe *sqe,
 			int ddir)
 {
 	struct io_rw *rw = io_kiocb_to_cmd(req, struct io_rw);
+	struct io_async_rw *io;
 	unsigned ioprio;
 	u64 attr_type_mask;
 	int ret;
 
 	if (io_rw_alloc_async(req))
 		return -ENOMEM;
+	io = req->async_data;
 
 	rw->kiocb.ki_pos = READ_ONCE(sqe->off);
 	/* used for fixed read/write too - just read unconditionally */
 	req->buf_index = READ_ONCE(sqe->buf_index);
+	io->buf_group = req->buf_index;
 
 	ioprio = READ_ONCE(sqe->ioprio);
 	if (ioprio) {
diff --git a/io_uring/rw.h b/io_uring/rw.h
index 81d6d9a8cf6937..129a53fe54825d 100644
--- a/io_uring/rw.h
+++ b/io_uring/rw.h
@@ -16,6 +16,8 @@ struct io_async_rw {
 		struct iov_iter			iter;
 		struct iov_iter_state		iter_state;
 		struct iovec			fast_iov;
+		unsigned			buf_group;
+
 		/*
 		 * wpq is for buffered io, while meta fields are used with
 		 * direct io
diff --git a/io_uring/tctx.c b/io_uring/tctx.c
index adc6e42c14df6c..5b66755579c08f 100644
--- a/io_uring/tctx.c
+++ b/io_uring/tctx.c
@@ -35,8 +35,6 @@ static struct io_wq *io_init_wq_offload(struct io_ring_ctx *ctx,
 
 	data.hash = hash;
 	data.task = task;
-	data.free_work = io_wq_free_work;
-	data.do_work = io_wq_submit_work;
 
 	/* Do QD, or 4 * CPUS, whatever is smallest */
 	concurrency = min(ctx->sq_entries, 4 * num_online_cpus());
diff --git a/io_uring/timeout.c b/io_uring/timeout.c
index 2a107665230b07..a6ff8c026b1f22 100644
--- a/io_uring/timeout.c
+++ b/io_uring/timeout.c
@@ -35,6 +35,9 @@ struct io_timeout_rem {
 	bool				ltimeout;
 };
 
+static struct io_kiocb *__io_disarm_linked_timeout(struct io_kiocb *req,
+						   struct io_kiocb *link);
+
 static inline bool io_is_timeout_noseq(struct io_kiocb *req)
 {
 	struct io_timeout *timeout = io_kiocb_to_cmd(req, struct io_timeout);
@@ -218,7 +221,9 @@ void io_disarm_next(struct io_kiocb *req)
 		struct io_ring_ctx *ctx = req->ctx;
 
 		raw_spin_lock_irq(&ctx->timeout_lock);
-		link = io_disarm_linked_timeout(req);
+		if (req->link && req->link->opcode == IORING_OP_LINK_TIMEOUT)
+			link = __io_disarm_linked_timeout(req, req->link);
+
 		raw_spin_unlock_irq(&ctx->timeout_lock);
 		if (link)
 			io_req_queue_tw_complete(link, -ECANCELED);
@@ -228,8 +233,8 @@ void io_disarm_next(struct io_kiocb *req)
 		io_fail_links(req);
 }
 
-struct io_kiocb *__io_disarm_linked_timeout(struct io_kiocb *req,
-					    struct io_kiocb *link)
+static struct io_kiocb *__io_disarm_linked_timeout(struct io_kiocb *req,
+						   struct io_kiocb *link)
 	__must_hold(&req->ctx->completion_lock)
 	__must_hold(&req->ctx->timeout_lock)
 {
diff --git a/io_uring/timeout.h b/io_uring/timeout.h
index e91b32448dcf90..2b7c9ad729925c 100644
--- a/io_uring/timeout.h
+++ b/io_uring/timeout.h
@@ -8,19 +8,6 @@ struct io_timeout_data {
 	u32				flags;
 };
 
-struct io_kiocb *__io_disarm_linked_timeout(struct io_kiocb *req,
-					    struct io_kiocb *link);
-
-static inline struct io_kiocb *io_disarm_linked_timeout(struct io_kiocb *req)
-{
-	struct io_kiocb *link = req->link;
-
-	if (link && link->opcode == IORING_OP_LINK_TIMEOUT)
-		return __io_disarm_linked_timeout(req, link);
-
-	return NULL;
-}
-
 __cold void io_flush_timeouts(struct io_ring_ctx *ctx);
 struct io_cancel_data;
 int io_timeout_cancel(struct io_ring_ctx *ctx, struct io_cancel_data *cd);
diff --git a/io_uring/uring_cmd.c b/io_uring/uring_cmd.c
index 430ed620ddfe01..8a6b0ddef79644 100644
--- a/io_uring/uring_cmd.c
+++ b/io_uring/uring_cmd.c
@@ -3,13 +3,10 @@
 #include <linux/errno.h>
 #include <linux/file.h>
 #include <linux/io_uring/cmd.h>
-#include <linux/io_uring/net.h>
 #include <linux/security.h>
 #include <linux/nospec.h>
-#include <net/sock.h>
 
 #include <uapi/linux/io_uring.h>
-#include <asm/ioctls.h>
 
 #include "io_uring.h"
 #include "alloc_cache.h"
@@ -307,83 +304,3 @@ void io_uring_cmd_issue_blocking(struct io_uring_cmd *ioucmd)
 
 	io_req_queue_iowq(req);
 }
-
-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);
-}
-
-#if defined(CONFIG_NET)
-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);
-#endif
diff --git a/io_uring/zcrx.c b/io_uring/zcrx.c
index fe86606b9f304d..9a568d04920470 100644
--- a/io_uring/zcrx.c
+++ b/io_uring/zcrx.c
@@ -26,29 +26,207 @@
 #include "zcrx.h"
 #include "rsrc.h"
 
+#define IO_DMA_ATTR (DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_WEAK_ORDERING)
+
 static inline struct io_zcrx_ifq *io_pp_to_ifq(struct page_pool *pp)
 {
 	return pp->mp_priv;
 }
 
-#define IO_DMA_ATTR (DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_WEAK_ORDERING)
+static inline struct io_zcrx_area *io_zcrx_iov_to_area(const struct net_iov *niov)
+{
+	struct net_iov_area *owner = net_iov_owner(niov);
 
-static void __io_zcrx_unmap_area(struct io_zcrx_ifq *ifq,
-				 struct io_zcrx_area *area, int nr_mapped)
+	return container_of(owner, struct io_zcrx_area, nia);
+}
+
+static inline struct page *io_zcrx_iov_page(const struct net_iov *niov)
+{
+	struct io_zcrx_area *area = io_zcrx_iov_to_area(niov);
+
+	return area->mem.pages[net_iov_idx(niov)];
+}
+
+static void io_release_dmabuf(struct io_zcrx_mem *mem)
+{
+	if (!IS_ENABLED(CONFIG_DMA_SHARED_BUFFER))
+		return;
+
+	if (mem->sgt)
+		dma_buf_unmap_attachment_unlocked(mem->attach, mem->sgt,
+						  DMA_FROM_DEVICE);
+	if (mem->attach)
+		dma_buf_detach(mem->dmabuf, mem->attach);
+	if (mem->dmabuf)
+		dma_buf_put(mem->dmabuf);
+
+	mem->sgt = NULL;
+	mem->attach = NULL;
+	mem->dmabuf = NULL;
+}
+
+static int io_import_dmabuf(struct io_zcrx_ifq *ifq,
+			    struct io_zcrx_mem *mem,
+			    struct io_uring_zcrx_area_reg *area_reg)
+{
+	unsigned long off = (unsigned long)area_reg->addr;
+	unsigned long len = (unsigned long)area_reg->len;
+	unsigned long total_size = 0;
+	struct scatterlist *sg;
+	int dmabuf_fd = area_reg->dmabuf_fd;
+	int i, ret;
+
+	if (WARN_ON_ONCE(!ifq->dev))
+		return -EFAULT;
+	if (!IS_ENABLED(CONFIG_DMA_SHARED_BUFFER))
+		return -EINVAL;
+
+	mem->is_dmabuf = true;
+	mem->dmabuf = dma_buf_get(dmabuf_fd);
+	if (IS_ERR(mem->dmabuf)) {
+		ret = PTR_ERR(mem->dmabuf);
+		mem->dmabuf = NULL;
+		goto err;
+	}
+
+	mem->attach = dma_buf_attach(mem->dmabuf, ifq->dev);
+	if (IS_ERR(mem->attach)) {
+		ret = PTR_ERR(mem->attach);
+		mem->attach = NULL;
+		goto err;
+	}
+
+	mem->sgt = dma_buf_map_attachment_unlocked(mem->attach, DMA_FROM_DEVICE);
+	if (IS_ERR(mem->sgt)) {
+		ret = PTR_ERR(mem->sgt);
+		mem->sgt = NULL;
+		goto err;
+	}
+
+	for_each_sgtable_dma_sg(mem->sgt, sg, i)
+		total_size += sg_dma_len(sg);
+
+	if (total_size < off + len)
+		return -EINVAL;
+
+	mem->dmabuf_offset = off;
+	mem->size = len;
+	return 0;
+err:
+	io_release_dmabuf(mem);
+	return ret;
+}
+
+static int io_zcrx_map_area_dmabuf(struct io_zcrx_ifq *ifq, struct io_zcrx_area *area)
+{
+	unsigned long off = area->mem.dmabuf_offset;
+	struct scatterlist *sg;
+	unsigned i, niov_idx = 0;
+
+	if (!IS_ENABLED(CONFIG_DMA_SHARED_BUFFER))
+		return -EINVAL;
+
+	for_each_sgtable_dma_sg(area->mem.sgt, sg, i) {
+		dma_addr_t dma = sg_dma_address(sg);
+		unsigned long sg_len = sg_dma_len(sg);
+		unsigned long sg_off = min(sg_len, off);
+
+		off -= sg_off;
+		sg_len -= sg_off;
+		dma += sg_off;
+
+		while (sg_len && niov_idx < area->nia.num_niovs) {
+			struct net_iov *niov = &area->nia.niovs[niov_idx];
+
+			if (net_mp_niov_set_dma_addr(niov, dma))
+				return 0;
+			sg_len -= PAGE_SIZE;
+			dma += PAGE_SIZE;
+			niov_idx++;
+		}
+	}
+	return niov_idx;
+}
+
+static int io_import_umem(struct io_zcrx_ifq *ifq,
+			  struct io_zcrx_mem *mem,
+			  struct io_uring_zcrx_area_reg *area_reg)
+{
+	struct page **pages;
+	int nr_pages;
+
+	if (area_reg->dmabuf_fd)
+		return -EINVAL;
+	if (!area_reg->addr)
+		return -EFAULT;
+	pages = io_pin_pages((unsigned long)area_reg->addr, area_reg->len,
+				   &nr_pages);
+	if (IS_ERR(pages))
+		return PTR_ERR(pages);
+
+	mem->pages = pages;
+	mem->nr_folios = nr_pages;
+	mem->size = area_reg->len;
+	return 0;
+}
+
+static void io_release_area_mem(struct io_zcrx_mem *mem)
+{
+	if (mem->is_dmabuf) {
+		io_release_dmabuf(mem);
+		return;
+	}
+	if (mem->pages) {
+		unpin_user_pages(mem->pages, mem->nr_folios);
+		kvfree(mem->pages);
+	}
+}
+
+static int io_import_area(struct io_zcrx_ifq *ifq,
+			  struct io_zcrx_mem *mem,
+			  struct io_uring_zcrx_area_reg *area_reg)
+{
+	int ret;
+
+	ret = io_validate_user_buf_range(area_reg->addr, area_reg->len);
+	if (ret)
+		return ret;
+	if (area_reg->addr & ~PAGE_MASK || area_reg->len & ~PAGE_MASK)
+		return -EINVAL;
+
+	if (area_reg->flags & IORING_ZCRX_AREA_DMABUF)
+		return io_import_dmabuf(ifq, mem, area_reg);
+	return io_import_umem(ifq, mem, area_reg);
+}
+
+static void io_zcrx_unmap_umem(struct io_zcrx_ifq *ifq,
+				struct io_zcrx_area *area, int nr_mapped)
 {
 	int i;
 
 	for (i = 0; i < nr_mapped; i++) {
-		struct net_iov *niov = &area->nia.niovs[i];
-		dma_addr_t dma;
+		netmem_ref netmem = net_iov_to_netmem(&area->nia.niovs[i]);
+		dma_addr_t dma = page_pool_get_dma_addr_netmem(netmem);
 
-		dma = page_pool_get_dma_addr_netmem(net_iov_to_netmem(niov));
 		dma_unmap_page_attrs(ifq->dev, dma, PAGE_SIZE,
 				     DMA_FROM_DEVICE, IO_DMA_ATTR);
-		net_mp_niov_set_dma_addr(niov, 0);
 	}
 }
 
+static void __io_zcrx_unmap_area(struct io_zcrx_ifq *ifq,
+				 struct io_zcrx_area *area, int nr_mapped)
+{
+	int i;
+
+	if (area->mem.is_dmabuf)
+		io_release_dmabuf(&area->mem);
+	else
+		io_zcrx_unmap_umem(ifq, area, nr_mapped);
+
+	for (i = 0; i < area->nia.num_niovs; i++)
+		net_mp_niov_set_dma_addr(&area->nia.niovs[i], 0);
+}
+
 static void io_zcrx_unmap_area(struct io_zcrx_ifq *ifq, struct io_zcrx_area *area)
 {
 	guard(mutex)(&ifq->dma_lock);
@@ -58,20 +236,16 @@ static void io_zcrx_unmap_area(struct io_zcrx_ifq *ifq, struct io_zcrx_area *are
 	area->is_mapped = false;
 }
 
-static int io_zcrx_map_area(struct io_zcrx_ifq *ifq, struct io_zcrx_area *area)
+static int io_zcrx_map_area_umem(struct io_zcrx_ifq *ifq, struct io_zcrx_area *area)
 {
 	int i;
 
-	guard(mutex)(&ifq->dma_lock);
-	if (area->is_mapped)
-		return 0;
-
 	for (i = 0; i < area->nia.num_niovs; i++) {
 		struct net_iov *niov = &area->nia.niovs[i];
 		dma_addr_t dma;
 
-		dma = dma_map_page_attrs(ifq->dev, area->pages[i], 0, PAGE_SIZE,
-					 DMA_FROM_DEVICE, IO_DMA_ATTR);
+		dma = dma_map_page_attrs(ifq->dev, area->mem.pages[i], 0,
+					 PAGE_SIZE, DMA_FROM_DEVICE, IO_DMA_ATTR);
 		if (dma_mapping_error(ifq->dev, dma))
 			break;
 		if (net_mp_niov_set_dma_addr(niov, dma)) {
@@ -80,9 +254,24 @@ static int io_zcrx_map_area(struct io_zcrx_ifq *ifq, struct io_zcrx_area *area)
 			break;
 		}
 	}
+	return i;
+}
+
+static int io_zcrx_map_area(struct io_zcrx_ifq *ifq, struct io_zcrx_area *area)
+{
+	unsigned nr;
+
+	guard(mutex)(&ifq->dma_lock);
+	if (area->is_mapped)
+		return 0;
+
+	if (area->mem.is_dmabuf)
+		nr = io_zcrx_map_area_dmabuf(ifq, area);
+	else
+		nr = io_zcrx_map_area_umem(ifq, area);
 
-	if (i != area->nia.num_niovs) {
-		__io_zcrx_unmap_area(ifq, area, i);
+	if (nr != area->nia.num_niovs) {
+		__io_zcrx_unmap_area(ifq, area, nr);
 		return -EINVAL;
 	}
 
@@ -118,13 +307,6 @@ struct io_zcrx_args {
 
 static const struct memory_provider_ops io_uring_pp_zc_ops;
 
-static inline struct io_zcrx_area *io_zcrx_iov_to_area(const struct net_iov *niov)
-{
-	struct net_iov_area *owner = net_iov_owner(niov);
-
-	return container_of(owner, struct io_zcrx_area, nia);
-}
-
 static inline atomic_t *io_get_user_counter(struct net_iov *niov)
 {
 	struct io_zcrx_area *area = io_zcrx_iov_to_area(niov);
@@ -147,17 +329,12 @@ static void io_zcrx_get_niov_uref(struct net_iov *niov)
 	atomic_inc(io_get_user_counter(niov));
 }
 
-static inline struct page *io_zcrx_iov_page(const struct net_iov *niov)
-{
-	struct io_zcrx_area *area = io_zcrx_iov_to_area(niov);
-
-	return area->pages[net_iov_idx(niov)];
-}
-
 static int io_allocate_rbuf_ring(struct io_zcrx_ifq *ifq,
 				 struct io_uring_zcrx_ifq_reg *reg,
-				 struct io_uring_region_desc *rd)
+				 struct io_uring_region_desc *rd,
+				 u32 id)
 {
+	u64 mmap_offset;
 	size_t off, size;
 	void *ptr;
 	int ret;
@@ -167,12 +344,14 @@ static int io_allocate_rbuf_ring(struct io_zcrx_ifq *ifq,
 	if (size > rd->size)
 		return -EINVAL;
 
-	ret = io_create_region_mmap_safe(ifq->ctx, &ifq->ctx->zcrx_region, rd,
-					 IORING_MAP_OFF_ZCRX_REGION);
+	mmap_offset = IORING_MAP_OFF_ZCRX_REGION;
+	mmap_offset += id << IORING_OFF_PBUF_SHIFT;
+
+	ret = io_create_region(ifq->ctx, &ifq->region, rd, mmap_offset);
 	if (ret < 0)
 		return ret;
 
-	ptr = io_region_get_ptr(&ifq->ctx->zcrx_region);
+	ptr = io_region_get_ptr(&ifq->region);
 	ifq->rq_ring = (struct io_uring *)ptr;
 	ifq->rqes = (struct io_uring_zcrx_rqe *)(ptr + off);
 	return 0;
@@ -180,7 +359,7 @@ static int io_allocate_rbuf_ring(struct io_zcrx_ifq *ifq,
 
 static void io_free_rbuf_ring(struct io_zcrx_ifq *ifq)
 {
-	io_free_region(ifq->ctx, &ifq->ctx->zcrx_region);
+	io_free_region(ifq->ctx, &ifq->region);
 	ifq->rq_ring = NULL;
 	ifq->rqes = NULL;
 }
@@ -188,53 +367,44 @@ static void io_free_rbuf_ring(struct io_zcrx_ifq *ifq)
 static void io_zcrx_free_area(struct io_zcrx_area *area)
 {
 	io_zcrx_unmap_area(area->ifq, area);
+	io_release_area_mem(&area->mem);
 
 	kvfree(area->freelist);
 	kvfree(area->nia.niovs);
 	kvfree(area->user_refs);
-	if (area->pages) {
-		unpin_user_pages(area->pages, area->nr_folios);
-		kvfree(area->pages);
-	}
 	kfree(area);
 }
 
+#define IO_ZCRX_AREA_SUPPORTED_FLAGS	(IORING_ZCRX_AREA_DMABUF)
+
 static int io_zcrx_create_area(struct io_zcrx_ifq *ifq,
 			       struct io_zcrx_area **res,
 			       struct io_uring_zcrx_area_reg *area_reg)
 {
 	struct io_zcrx_area *area;
-	int i, ret, nr_pages, nr_iovs;
-	struct iovec iov;
+	unsigned nr_iovs;
+	int i, ret;
 
-	if (area_reg->flags || area_reg->rq_area_token)
+	if (area_reg->flags & ~IO_ZCRX_AREA_SUPPORTED_FLAGS)
 		return -EINVAL;
-	if (area_reg->__resv1 || area_reg->__resv2[0] || area_reg->__resv2[1])
+	if (area_reg->rq_area_token)
 		return -EINVAL;
-	if (area_reg->addr & ~PAGE_MASK || area_reg->len & ~PAGE_MASK)
+	if (area_reg->__resv2[0] || area_reg->__resv2[1])
 		return -EINVAL;
 
-	iov.iov_base = u64_to_user_ptr(area_reg->addr);
-	iov.iov_len = area_reg->len;
-	ret = io_buffer_validate(&iov);
-	if (ret)
-		return ret;
-
 	ret = -ENOMEM;
 	area = kzalloc(sizeof(*area), GFP_KERNEL);
 	if (!area)
 		goto err;
 
-	area->pages = io_pin_pages((unsigned long)area_reg->addr, area_reg->len,
-				   &nr_pages);
-	if (IS_ERR(area->pages)) {
-		ret = PTR_ERR(area->pages);
-		area->pages = NULL;
+	ret = io_import_area(ifq, &area->mem, area_reg);
+	if (ret)
 		goto err;
-	}
-	area->nr_folios = nr_iovs = nr_pages;
+
+	nr_iovs = area->mem.size >> PAGE_SHIFT;
 	area->nia.num_niovs = nr_iovs;
 
+	ret = -ENOMEM;
 	area->nia.niovs = kvmalloc_array(nr_iovs, sizeof(area->nia.niovs[0]),
 					 GFP_KERNEL | __GFP_ZERO);
 	if (!area->nia.niovs)
@@ -245,9 +415,6 @@ static int io_zcrx_create_area(struct io_zcrx_ifq *ifq,
 	if (!area->freelist)
 		goto err;
 
-	for (i = 0; i < nr_iovs; i++)
-		area->freelist[i] = i;
-
 	area->user_refs = kvmalloc_array(nr_iovs, sizeof(area->user_refs[0]),
 					GFP_KERNEL | __GFP_ZERO);
 	if (!area->user_refs)
@@ -341,6 +508,16 @@ static void io_zcrx_ifq_free(struct io_zcrx_ifq *ifq)
 	kfree(ifq);
 }
 
+struct io_mapped_region *io_zcrx_get_region(struct io_ring_ctx *ctx,
+					    unsigned int id)
+{
+	struct io_zcrx_ifq *ifq = xa_load(&ctx->zcrx_ctxs, id);
+
+	lockdep_assert_held(&ctx->mmap_lock);
+
+	return ifq ? &ifq->region : NULL;
+}
+
 int io_register_zcrx_ifq(struct io_ring_ctx *ctx,
 			  struct io_uring_zcrx_ifq_reg __user *arg)
 {
@@ -350,6 +527,7 @@ int io_register_zcrx_ifq(struct io_ring_ctx *ctx,
 	struct io_uring_region_desc rd;
 	struct io_zcrx_ifq *ifq;
 	int ret;
+	u32 id;
 
 	/*
 	 * 1. Interface queue allocation.
@@ -362,8 +540,6 @@ int io_register_zcrx_ifq(struct io_ring_ctx *ctx,
 	if (!(ctx->flags & IORING_SETUP_DEFER_TASKRUN &&
 	      ctx->flags & IORING_SETUP_CQE32))
 		return -EINVAL;
-	if (ctx->ifq)
-		return -EBUSY;
 	if (copy_from_user(&reg, arg, sizeof(reg)))
 		return -EFAULT;
 	if (copy_from_user(&rd, u64_to_user_ptr(reg.region_ptr), sizeof(rd)))
@@ -386,29 +562,37 @@ int io_register_zcrx_ifq(struct io_ring_ctx *ctx,
 	ifq = io_zcrx_ifq_alloc(ctx);
 	if (!ifq)
 		return -ENOMEM;
+	ifq->rq_entries = reg.rq_entries;
 
-	ret = io_allocate_rbuf_ring(ifq, &reg, &rd);
-	if (ret)
-		goto err;
+	scoped_guard(mutex, &ctx->mmap_lock) {
+		/* preallocate id */
+		ret = xa_alloc(&ctx->zcrx_ctxs, &id, NULL, xa_limit_31b, GFP_KERNEL);
+		if (ret)
+			goto ifq_free;
+	}
 
-	ret = io_zcrx_create_area(ifq, &ifq->area, &area);
+	ret = io_allocate_rbuf_ring(ifq, &reg, &rd, id);
 	if (ret)
 		goto err;
 
-	ifq->rq_entries = reg.rq_entries;
-
-	ret = -ENODEV;
 	ifq->netdev = netdev_get_by_index(current->nsproxy->net_ns, reg.if_idx,
 					  &ifq->netdev_tracker, GFP_KERNEL);
-	if (!ifq->netdev)
+	if (!ifq->netdev) {
+		ret = -ENODEV;
 		goto err;
+	}
 
 	ifq->dev = ifq->netdev->dev.parent;
-	ret = -EOPNOTSUPP;
-	if (!ifq->dev)
+	if (!ifq->dev) {
+		ret = -EOPNOTSUPP;
 		goto err;
+	}
 	get_device(ifq->dev);
 
+	ret = io_zcrx_create_area(ifq, &ifq->area, &area);
+	if (ret)
+		goto err;
+
 	mp_param.mp_ops = &io_uring_pp_zc_ops;
 	mp_param.mp_priv = ifq;
 	ret = net_mp_open_rxq(ifq->netdev, reg.if_rxq, &mp_param);
@@ -419,6 +603,14 @@ int io_register_zcrx_ifq(struct io_ring_ctx *ctx,
 	reg.offsets.rqes = sizeof(struct io_uring);
 	reg.offsets.head = offsetof(struct io_uring, head);
 	reg.offsets.tail = offsetof(struct io_uring, tail);
+	reg.zcrx_id = id;
+
+	scoped_guard(mutex, &ctx->mmap_lock) {
+		/* publish ifq */
+		ret = -ENOMEM;
+		if (xa_store(&ctx->zcrx_ctxs, id, ifq, GFP_KERNEL))
+			goto err;
+	}
 
 	if (copy_to_user(arg, &reg, sizeof(reg)) ||
 	    copy_to_user(u64_to_user_ptr(reg.region_ptr), &rd, sizeof(rd)) ||
@@ -426,24 +618,34 @@ int io_register_zcrx_ifq(struct io_ring_ctx *ctx,
 		ret = -EFAULT;
 		goto err;
 	}
-	ctx->ifq = ifq;
 	return 0;
 err:
+	scoped_guard(mutex, &ctx->mmap_lock)
+		xa_erase(&ctx->zcrx_ctxs, id);
+ifq_free:
 	io_zcrx_ifq_free(ifq);
 	return ret;
 }
 
 void io_unregister_zcrx_ifqs(struct io_ring_ctx *ctx)
 {
-	struct io_zcrx_ifq *ifq = ctx->ifq;
+	struct io_zcrx_ifq *ifq;
+	unsigned long id;
 
 	lockdep_assert_held(&ctx->uring_lock);
 
-	if (!ifq)
-		return;
+	while (1) {
+		scoped_guard(mutex, &ctx->mmap_lock) {
+			ifq = xa_find(&ctx->zcrx_ctxs, &id, ULONG_MAX, XA_PRESENT);
+			if (ifq)
+				xa_erase(&ctx->zcrx_ctxs, id);
+		}
+		if (!ifq)
+			break;
+		io_zcrx_ifq_free(ifq);
+	}
 
-	ctx->ifq = NULL;
-	io_zcrx_ifq_free(ifq);
+	xa_destroy(&ctx->zcrx_ctxs);
 }
 
 static struct net_iov *__io_zcrx_get_free_niov(struct io_zcrx_area *area)
@@ -500,12 +702,15 @@ static void io_zcrx_scrub(struct io_zcrx_ifq *ifq)
 
 void io_shutdown_zcrx_ifqs(struct io_ring_ctx *ctx)
 {
+	struct io_zcrx_ifq *ifq;
+	unsigned long index;
+
 	lockdep_assert_held(&ctx->uring_lock);
 
-	if (!ctx->ifq)
-		return;
-	io_zcrx_scrub(ctx->ifq);
-	io_close_queue(ctx->ifq);
+	xa_for_each(&ctx->zcrx_ctxs, index, ifq) {
+		io_zcrx_scrub(ifq);
+		io_close_queue(ifq);
+	}
 }
 
 static inline u32 io_zcrx_rqring_entries(struct io_zcrx_ifq *ifq)
@@ -742,6 +947,9 @@ static ssize_t io_zcrx_copy_chunk(struct io_kiocb *req, struct io_zcrx_ifq *ifq,
 	size_t copied = 0;
 	int ret = 0;
 
+	if (area->mem.is_dmabuf)
+		return -EFAULT;
+
 	while (len) {
 		size_t copy_size = min_t(size_t, PAGE_SIZE, len);
 		const int dst_off = 0;
diff --git a/io_uring/zcrx.h b/io_uring/zcrx.h
index f2bc811f022c67..2f5e26389f2218 100644
--- a/io_uring/zcrx.h
+++ b/io_uring/zcrx.h
@@ -3,10 +3,24 @@
 #define IOU_ZC_RX_H
 
 #include <linux/io_uring_types.h>
+#include <linux/dma-buf.h>
 #include <linux/socket.h>
 #include <net/page_pool/types.h>
 #include <net/net_trackers.h>
 
+struct io_zcrx_mem {
+	unsigned long			size;
+	bool				is_dmabuf;
+
+	struct page			**pages;
+	unsigned long			nr_folios;
+
+	struct dma_buf_attachment	*attach;
+	struct dma_buf			*dmabuf;
+	struct sg_table			*sgt;
+	unsigned long			dmabuf_offset;
+};
+
 struct io_zcrx_area {
 	struct net_iov_area	nia;
 	struct io_zcrx_ifq	*ifq;
@@ -14,13 +28,13 @@ struct io_zcrx_area {
 
 	bool			is_mapped;
 	u16			area_id;
-	struct page		**pages;
-	unsigned long		nr_folios;
 
 	/* freelist */
 	spinlock_t		freelist_lock ____cacheline_aligned_in_smp;
 	u32			free_count;
 	u32			*freelist;
+
+	struct io_zcrx_mem	mem;
 };
 
 struct io_zcrx_ifq {
@@ -39,6 +53,7 @@ struct io_zcrx_ifq {
 	netdevice_tracker		netdev_tracker;
 	spinlock_t			lock;
 	struct mutex			dma_lock;
+	struct io_mapped_region		region;
 };
 
 #if defined(CONFIG_IO_URING_ZCRX)
@@ -49,6 +64,8 @@ void io_shutdown_zcrx_ifqs(struct io_ring_ctx *ctx);
 int io_zcrx_recv(struct io_kiocb *req, struct io_zcrx_ifq *ifq,
 		 struct socket *sock, unsigned int flags,
 		 unsigned issue_flags, unsigned int *len);
+struct io_mapped_region *io_zcrx_get_region(struct io_ring_ctx *ctx,
+					    unsigned int id);
 #else
 static inline int io_register_zcrx_ifq(struct io_ring_ctx *ctx,
 					struct io_uring_zcrx_ifq_reg __user *arg)
@@ -67,6 +84,11 @@ static inline int io_zcrx_recv(struct io_kiocb *req, struct io_zcrx_ifq *ifq,
 {
 	return -EOPNOTSUPP;
 }
+static inline struct io_mapped_region *io_zcrx_get_region(struct io_ring_ctx *ctx,
+							  unsigned int id)
+{
+	return NULL;
+}
 #endif
 
 int io_recvzc(struct io_kiocb *req, unsigned int issue_flags);
author	Jens Axboe <axboe@kernel.dk>	2025-05-17 18:47:41 -0600
committer	Jens Axboe <axboe@kernel.dk>	2025-05-17 18:47:41 -0600
commit	f6873e3f089c6a386df3dac965abb5516c2eccda (patch)
tree	53d49ad886d76202f08cd41d40f71bafa7064b54
parent	c2d8018325b6f0d0940c2544678822e88e41adbc (diff)
parent	f660fd2ca15a3743f65f6110ae60d5b80500d856 (diff)
download	linux-block-for-next.tar.gz