[linux-2.6-block.git] / io_uring / msg_ring.c

// SPDX-License-Identifier: GPL-2.0
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/file.h>
#include <linux/slab.h>
#include <linux/nospec.h>
#include <linux/io_uring.h>

#include <uapi/linux/io_uring.h>

#include "io_uring.h"
#include "rsrc.h"
#include "filetable.h"
#include "alloc_cache.h"
#include "msg_ring.h"

/* All valid masks for MSG_RING */
#define IORING_MSG_RING_MASK		(IORING_MSG_RING_CQE_SKIP | \
					IORING_MSG_RING_FLAGS_PASS)

struct io_msg {
	struct file			*file;
	struct file			*src_file;
	struct callback_head		tw;
	u64 user_data;
	u32 len;
	u32 cmd;
	u32 src_fd;
	union {
		u32 dst_fd;
		u32 cqe_flags;
	};
	u32 flags;
};

static void io_double_unlock_ctx(struct io_ring_ctx *octx)
{
	mutex_unlock(&octx->uring_lock);
}

static int io_lock_external_ctx(struct io_ring_ctx *octx,
				unsigned int issue_flags)
{
	/*
	 * To ensure proper ordering between the two ctxs, we can only
	 * attempt a trylock on the target. If that fails and we already have
	 * the source ctx lock, punt to io-wq.
	 */
	if (!(issue_flags & IO_URING_F_UNLOCKED)) {
		if (!mutex_trylock(&octx->uring_lock))
			return -EAGAIN;
		return 0;
	}
	mutex_lock(&octx->uring_lock);
	return 0;
}

void io_msg_ring_cleanup(struct io_kiocb *req)
{
	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);

	if (WARN_ON_ONCE(!msg->src_file))
		return;

	fput(msg->src_file);
	msg->src_file = NULL;
}

static inline bool io_msg_need_remote(struct io_ring_ctx *target_ctx)
{
	return target_ctx->task_complete;
}

static void io_msg_tw_complete(struct io_kiocb *req, io_tw_token_t tw)
{
	struct io_ring_ctx *ctx = req->ctx;

	io_add_aux_cqe(ctx, req->cqe.user_data, req->cqe.res, req->cqe.flags);
	if (spin_trylock(&ctx->msg_lock)) {
		if (io_alloc_cache_put(&ctx->msg_cache, req))
			req = NULL;
		spin_unlock(&ctx->msg_lock);
	}
	if (req)
		kmem_cache_free(req_cachep, req);
	percpu_ref_put(&ctx->refs);
}

static int io_msg_remote_post(struct io_ring_ctx *ctx, struct io_kiocb *req,
			      int res, u32 cflags, u64 user_data)
{
	if (!READ_ONCE(ctx->submitter_task)) {
		kmem_cache_free(req_cachep, req);
		return -EOWNERDEAD;
	}
	req->opcode = IORING_OP_NOP;
	req->cqe.user_data = user_data;
	io_req_set_res(req, res, cflags);
	percpu_ref_get(&ctx->refs);
	req->ctx = ctx;
	req->tctx = NULL;
	req->io_task_work.func = io_msg_tw_complete;
	io_req_task_work_add_remote(req, IOU_F_TWQ_LAZY_WAKE);
	return 0;
}

static struct io_kiocb *io_msg_get_kiocb(struct io_ring_ctx *ctx)
{
	struct io_kiocb *req = NULL;

	if (spin_trylock(&ctx->msg_lock)) {
		req = io_alloc_cache_get(&ctx->msg_cache);
		spin_unlock(&ctx->msg_lock);
		if (req)
			return req;
	}
	return kmem_cache_alloc(req_cachep, GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO);
}

static int io_msg_data_remote(struct io_ring_ctx *target_ctx,
			      struct io_msg *msg)
{
	struct io_kiocb *target;
	u32 flags = 0;

	target = io_msg_get_kiocb(target_ctx);
	if (unlikely(!target))
		return -ENOMEM;

	if (msg->flags & IORING_MSG_RING_FLAGS_PASS)
		flags = msg->cqe_flags;

	return io_msg_remote_post(target_ctx, target, msg->len, flags,
					msg->user_data);
}

static int __io_msg_ring_data(struct io_ring_ctx *target_ctx,
			      struct io_msg *msg, unsigned int issue_flags)
{
	u32 flags = 0;
	int ret;

	if (msg->src_fd || msg->flags & ~IORING_MSG_RING_FLAGS_PASS)
		return -EINVAL;
	if (!(msg->flags & IORING_MSG_RING_FLAGS_PASS) && msg->dst_fd)
		return -EINVAL;
	if (target_ctx->flags & IORING_SETUP_R_DISABLED)
		return -EBADFD;

	if (io_msg_need_remote(target_ctx))
		return io_msg_data_remote(target_ctx, msg);

	if (msg->flags & IORING_MSG_RING_FLAGS_PASS)
		flags = msg->cqe_flags;

	ret = -EOVERFLOW;
	if (target_ctx->flags & IORING_SETUP_IOPOLL) {
		if (unlikely(io_lock_external_ctx(target_ctx, issue_flags)))
			return -EAGAIN;
	}
	if (io_post_aux_cqe(target_ctx, msg->user_data, msg->len, flags))
		ret = 0;
	if (target_ctx->flags & IORING_SETUP_IOPOLL)
		io_double_unlock_ctx(target_ctx);
	return ret;
}

static int io_msg_ring_data(struct io_kiocb *req, unsigned int issue_flags)
{
	struct io_ring_ctx *target_ctx = req->file->private_data;
	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);

	return __io_msg_ring_data(target_ctx, msg, issue_flags);
}

static int io_msg_grab_file(struct io_kiocb *req, unsigned int issue_flags)
{
	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	struct io_ring_ctx *ctx = req->ctx;
	struct io_rsrc_node *node;
	int ret = -EBADF;

	io_ring_submit_lock(ctx, issue_flags);
	node = io_rsrc_node_lookup(&ctx->file_table.data, msg->src_fd);
	if (node) {
		msg->src_file = io_slot_file(node);
		if (msg->src_file)
			get_file(msg->src_file);
		req->flags |= REQ_F_NEED_CLEANUP;
		ret = 0;
	}
	io_ring_submit_unlock(ctx, issue_flags);
	return ret;
}

static int io_msg_install_complete(struct io_kiocb *req, unsigned int issue_flags)
{
	struct io_ring_ctx *target_ctx = req->file->private_data;
	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	struct file *src_file = msg->src_file;
	int ret;

	if (unlikely(io_lock_external_ctx(target_ctx, issue_flags)))
		return -EAGAIN;

	ret = __io_fixed_fd_install(target_ctx, src_file, msg->dst_fd);
	if (ret < 0)
		goto out_unlock;

	msg->src_file = NULL;
	req->flags &= ~REQ_F_NEED_CLEANUP;

	if (msg->flags & IORING_MSG_RING_CQE_SKIP)
		goto out_unlock;
	/*
	 * If this fails, the target still received the file descriptor but
	 * wasn't notified of the fact. This means that if this request
	 * completes with -EOVERFLOW, then the sender must ensure that a
	 * later IORING_OP_MSG_RING delivers the message.
	 */
	if (!io_post_aux_cqe(target_ctx, msg->user_data, ret, 0))
		ret = -EOVERFLOW;
out_unlock:
	io_double_unlock_ctx(target_ctx);
	return ret;
}

static void io_msg_tw_fd_complete(struct callback_head *head)
{
	struct io_msg *msg = container_of(head, struct io_msg, tw);
	struct io_kiocb *req = cmd_to_io_kiocb(msg);
	int ret = -EOWNERDEAD;

	if (!(current->flags & PF_EXITING))
		ret = io_msg_install_complete(req, IO_URING_F_UNLOCKED);
	if (ret < 0)
		req_set_fail(req);
	io_req_queue_tw_complete(req, ret);
}

static int io_msg_fd_remote(struct io_kiocb *req)
{
	struct io_ring_ctx *ctx = req->file->private_data;
	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	struct task_struct *task = READ_ONCE(ctx->submitter_task);

	if (unlikely(!task))
		return -EOWNERDEAD;

	init_task_work(&msg->tw, io_msg_tw_fd_complete);
	if (task_work_add(task, &msg->tw, TWA_SIGNAL))
		return -EOWNERDEAD;

	return IOU_ISSUE_SKIP_COMPLETE;
}

static int io_msg_send_fd(struct io_kiocb *req, unsigned int issue_flags)
{
	struct io_ring_ctx *target_ctx = req->file->private_data;
	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	struct io_ring_ctx *ctx = req->ctx;

	if (msg->len)
		return -EINVAL;
	if (target_ctx == ctx)
		return -EINVAL;
	if (target_ctx->flags & IORING_SETUP_R_DISABLED)
		return -EBADFD;
	if (!msg->src_file) {
		int ret = io_msg_grab_file(req, issue_flags);
		if (unlikely(ret))
			return ret;
	}

	if (io_msg_need_remote(target_ctx))
		return io_msg_fd_remote(req);
	return io_msg_install_complete(req, issue_flags);
}

static int __io_msg_ring_prep(struct io_msg *msg, const struct io_uring_sqe *sqe)
{
	if (unlikely(sqe->buf_index || sqe->personality))
		return -EINVAL;

	msg->src_file = NULL;
	msg->user_data = READ_ONCE(sqe->off);
	msg->len = READ_ONCE(sqe->len);
	msg->cmd = READ_ONCE(sqe->addr);
	msg->src_fd = READ_ONCE(sqe->addr3);
	msg->dst_fd = READ_ONCE(sqe->file_index);
	msg->flags = READ_ONCE(sqe->msg_ring_flags);
	if (msg->flags & ~IORING_MSG_RING_MASK)
		return -EINVAL;

	return 0;
}

int io_msg_ring_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
	return __io_msg_ring_prep(io_kiocb_to_cmd(req, struct io_msg), sqe);
}

int io_msg_ring(struct io_kiocb *req, unsigned int issue_flags)
{
	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	int ret;

	ret = -EBADFD;
	if (!io_is_uring_fops(req->file))
		goto done;

	switch (msg->cmd) {
	case IORING_MSG_DATA:
		ret = io_msg_ring_data(req, issue_flags);
		break;
	case IORING_MSG_SEND_FD:
		ret = io_msg_send_fd(req, issue_flags);
		break;
	default:
		ret = -EINVAL;
		break;
	}

done:
	if (ret < 0) {
		if (ret == -EAGAIN || ret == IOU_ISSUE_SKIP_COMPLETE)
			return ret;
		req_set_fail(req);
	}
	io_req_set_res(req, ret, 0);
	return IOU_COMPLETE;
}

int io_uring_sync_msg_ring(struct io_uring_sqe *sqe)
{
	struct io_msg io_msg = { };
	int ret;

	ret = __io_msg_ring_prep(&io_msg, sqe);
	if (unlikely(ret))
		return ret;

	/*
	 * Only data sending supported, not IORING_MSG_SEND_FD as that one
	 * doesn't make sense without a source ring to send files from.
	 */
	if (io_msg.cmd != IORING_MSG_DATA)
		return -EINVAL;

	CLASS(fd, f)(sqe->fd);
	if (fd_empty(f))
		return -EBADF;
	if (!io_is_uring_fops(fd_file(f)))
		return -EBADFD;
	return  __io_msg_ring_data(fd_file(f)->private_data,
				   &io_msg, IO_URING_F_UNLOCKED);
}
Commit	Line	Data
36404b09 JA	1	// SPDX-License-Identifier: GPL-2.0
	2	#include <linux/kernel.h>
	3	#include <linux/errno.h>
	4	#include <linux/file.h>
	5	#include <linux/slab.h>
e6130eba	6	#include <linux/nospec.h>
36404b09 JA	7	#include <linux/io_uring.h>
	8
	9	#include <uapi/linux/io_uring.h>
	10
36404b09	11	#include "io_uring.h"
e6130eba JA	12	#include "rsrc.h"
e6130eba JA	13	#include "filetable.h"
50cf5f38	14	#include "alloc_cache.h"
36404b09 JA	15	#include "msg_ring.h"
36404b09 JA	16
cbeb47a7 BL	17	/* All valid masks for MSG_RING */
	18	#define IORING_MSG_RING_MASK (IORING_MSG_RING_CQE_SKIP \| \
	19	IORING_MSG_RING_FLAGS_PASS)
	20
36404b09 JA	21	struct io_msg {
36404b09 JA	22	struct file *file;
11373026	23	struct file *src_file;
6d043ee1	24	struct callback_head tw;
36404b09 JA	25	u64 user_data;
36404b09 JA	26	u32 len;
e6130eba JA	27	u32 cmd;
e6130eba JA	28	u32 src_fd;
cbeb47a7 BL	29	union {
	30	u32 dst_fd;
	31	u32 cqe_flags;
	32	};
e6130eba	33	u32 flags;
36404b09 JA	34	};
36404b09 JA	35
423d5081 JA	36	static void io_double_unlock_ctx(struct io_ring_ctx *octx)
	37	{
	38	mutex_unlock(&octx->uring_lock);
	39	}
	40
b0e9570a PB	41	static int io_lock_external_ctx(struct io_ring_ctx *octx,
b0e9570a PB	42	unsigned int issue_flags)
423d5081 JA	43	{
	44	/*
	45	* To ensure proper ordering between the two ctxs, we can only
	46	* attempt a trylock on the target. If that fails and we already have
	47	* the source ctx lock, punt to io-wq.
	48	*/
	49	if (!(issue_flags & IO_URING_F_UNLOCKED)) {
	50	if (!mutex_trylock(&octx->uring_lock))
	51	return -EAGAIN;
	52	return 0;
	53	}
	54	mutex_lock(&octx->uring_lock);
	55	return 0;
	56	}
	57
11373026 PB	58	void io_msg_ring_cleanup(struct io_kiocb *req)
	59	{
	60	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	61
	62	if (WARN_ON_ONCE(!msg->src_file))
	63	return;
	64
	65	fput(msg->src_file);
	66	msg->src_file = NULL;
	67	}
	68
56d8e318 PB	69	static inline bool io_msg_need_remote(struct io_ring_ctx *target_ctx)
56d8e318 PB	70	{
d57afd8b	71	return target_ctx->task_complete;
56d8e318 PB	72	}
56d8e318 PB	73
bcf8a029	74	static void io_msg_tw_complete(struct io_kiocb *req, io_tw_token_t tw)
56d8e318	75	{
0617bb50	76	struct io_ring_ctx *ctx = req->ctx;
56d8e318	77
0617bb50	78	io_add_aux_cqe(ctx, req->cqe.user_data, req->cqe.res, req->cqe.flags);
50cf5f38 JA	79	if (spin_trylock(&ctx->msg_lock)) {
	80	if (io_alloc_cache_put(&ctx->msg_cache, req))
	81	req = NULL;
	82	spin_unlock(&ctx->msg_lock);
	83	}
	84	if (req)
be4f5d9c	85	kmem_cache_free(req_cachep, req);
0617bb50 JA	86	percpu_ref_put(&ctx->refs);
0617bb50 JA	87	}
56d8e318	88
b0727b12 JA	89	static int io_msg_remote_post(struct io_ring_ctx ctx, struct io_kiocb req,
b0727b12 JA	90	int res, u32 cflags, u64 user_data)
0617bb50	91	{
69a62e03	92	if (!READ_ONCE(ctx->submitter_task)) {
b0727b12 JA	93	kmem_cache_free(req_cachep, req);
	94	return -EOWNERDEAD;
	95	}
9cc0bbda	96	req->opcode = IORING_OP_NOP;
0617bb50 JA	97	req->cqe.user_data = user_data;
	98	io_req_set_res(req, res, cflags);
	99	percpu_ref_get(&ctx->refs);
	100	req->ctx = ctx;
69a62e03	101	req->tctx = NULL;
0617bb50	102	req->io_task_work.func = io_msg_tw_complete;
ea910678	103	io_req_task_work_add_remote(req, IOU_F_TWQ_LAZY_WAKE);
b0727b12	104	return 0;
56d8e318 PB	105	}
56d8e318 PB	106
50cf5f38 JA	107	static struct io_kiocb io_msg_get_kiocb(struct io_ring_ctx ctx)
	108	{
	109	struct io_kiocb *req = NULL;
	110
	111	if (spin_trylock(&ctx->msg_lock)) {
	112	req = io_alloc_cache_get(&ctx->msg_cache);
	113	spin_unlock(&ctx->msg_lock);
0db4618e JA	114	if (req)
0db4618e JA	115	return req;
50cf5f38	116	}
0db4618e	117	return kmem_cache_alloc(req_cachep, GFP_KERNEL \| __GFP_NOWARN \| __GFP_ZERO);
50cf5f38 JA	118	}
50cf5f38 JA	119
95d6c922 JA	120	static int io_msg_data_remote(struct io_ring_ctx *target_ctx,
95d6c922 JA	121	struct io_msg *msg)
6d043ee1	122	{
0617bb50 JA	123	struct io_kiocb *target;
0617bb50 JA	124	u32 flags = 0;
6d043ee1	125
95d6c922	126	target = io_msg_get_kiocb(target_ctx);
0617bb50 JA	127	if (unlikely(!target))
	128	return -ENOMEM;
	129
	130	if (msg->flags & IORING_MSG_RING_FLAGS_PASS)
	131	flags = msg->cqe_flags;
	132
b0727b12 JA	133	return io_msg_remote_post(target_ctx, target, msg->len, flags,
b0727b12 JA	134	msg->user_data);
6d043ee1 PB	135	}
6d043ee1 PB	136
95d6c922 JA	137	static int __io_msg_ring_data(struct io_ring_ctx *target_ctx,
95d6c922 JA	138	struct io_msg *msg, unsigned int issue_flags)
e6130eba	139	{
cbeb47a7	140	u32 flags = 0;
e12d7a46	141	int ret;
e6130eba	142
cbeb47a7 BL	143	if (msg->src_fd \|\| msg->flags & ~IORING_MSG_RING_FLAGS_PASS)
	144	return -EINVAL;
	145	if (!(msg->flags & IORING_MSG_RING_FLAGS_PASS) && msg->dst_fd)
e6130eba	146	return -EINVAL;
8579538c PB	147	if (target_ctx->flags & IORING_SETUP_R_DISABLED)
8579538c PB	148	return -EBADFD;
e6130eba	149
56d8e318	150	if (io_msg_need_remote(target_ctx))
95d6c922	151	return io_msg_data_remote(target_ctx, msg);
6d043ee1	152
cbeb47a7 BL	153	if (msg->flags & IORING_MSG_RING_FLAGS_PASS)
	154	flags = msg->cqe_flags;
	155
e12d7a46 JA	156	ret = -EOVERFLOW;
e12d7a46 JA	157	if (target_ctx->flags & IORING_SETUP_IOPOLL) {
b0e9570a	158	if (unlikely(io_lock_external_ctx(target_ctx, issue_flags)))
e12d7a46	159	return -EAGAIN;
e12d7a46	160	}
59b28a6e JA	161	if (io_post_aux_cqe(target_ctx, msg->user_data, msg->len, flags))
	162	ret = 0;
	163	if (target_ctx->flags & IORING_SETUP_IOPOLL)
	164	io_double_unlock_ctx(target_ctx);
e12d7a46	165	return ret;
e6130eba JA	166	}
e6130eba JA	167
95d6c922 JA	168	static int io_msg_ring_data(struct io_kiocb *req, unsigned int issue_flags)
	169	{
	170	struct io_ring_ctx *target_ctx = req->file->private_data;
	171	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	172
	173	return __io_msg_ring_data(target_ctx, msg, issue_flags);
	174	}
	175
b54a1404	176	static int io_msg_grab_file(struct io_kiocb *req, unsigned int issue_flags)
11373026 PB	177	{
	178	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	179	struct io_ring_ctx *ctx = req->ctx;
b54a1404 JA	180	struct io_rsrc_node *node;
b54a1404 JA	181	int ret = -EBADF;
11373026 PB	182
11373026 PB	183	io_ring_submit_lock(ctx, issue_flags);
b54a1404 JA	184	node = io_rsrc_node_lookup(&ctx->file_table.data, msg->src_fd);
	185	if (node) {
	186	msg->src_file = io_slot_file(node);
	187	if (msg->src_file)
	188	get_file(msg->src_file);
	189	req->flags \|= REQ_F_NEED_CLEANUP;
	190	ret = 0;
e6130eba	191	}
11373026	192	io_ring_submit_unlock(ctx, issue_flags);
b54a1404	193	return ret;
e6130eba JA	194	}
e6130eba JA	195
17211310	196	static int io_msg_install_complete(struct io_kiocb *req, unsigned int issue_flags)
e6130eba JA	197	{
e6130eba JA	198	struct io_ring_ctx *target_ctx = req->file->private_data;
f2ccb5ae	199	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
11373026	200	struct file *src_file = msg->src_file;
e6130eba JA	201	int ret;
e6130eba JA	202
b0e9570a	203	if (unlikely(io_lock_external_ctx(target_ctx, issue_flags)))
11373026	204	return -EAGAIN;
e6130eba JA	205
e6130eba JA	206	ret = __io_fixed_fd_install(target_ctx, src_file, msg->dst_fd);
11373026	207	if (ret < 0)
e6130eba	208	goto out_unlock;
17211310	209
11373026 PB	210	msg->src_file = NULL;
11373026 PB	211	req->flags &= ~REQ_F_NEED_CLEANUP;
e6130eba JA	212
	213	if (msg->flags & IORING_MSG_RING_CQE_SKIP)
	214	goto out_unlock;
e6130eba JA	215	/*
	216	* If this fails, the target still received the file descriptor but
	217	* wasn't notified of the fact. This means that if this request
	218	* completes with -EOVERFLOW, then the sender must ensure that a
	219	* later IORING_OP_MSG_RING delivers the message.
	220	*/
5da28edd	221	if (!io_post_aux_cqe(target_ctx, msg->user_data, ret, 0))
e6130eba JA	222	ret = -EOVERFLOW;
e6130eba JA	223	out_unlock:
423d5081	224	io_double_unlock_ctx(target_ctx);
e6130eba JA	225	return ret;
	226	}
	227
6d043ee1 PB	228	static void io_msg_tw_fd_complete(struct callback_head *head)
	229	{
	230	struct io_msg *msg = container_of(head, struct io_msg, tw);
	231	struct io_kiocb *req = cmd_to_io_kiocb(msg);
	232	int ret = -EOWNERDEAD;
	233
	234	if (!(current->flags & PF_EXITING))
	235	ret = io_msg_install_complete(req, IO_URING_F_UNLOCKED);
	236	if (ret < 0)
	237	req_set_fail(req);
	238	io_req_queue_tw_complete(req, ret);
	239	}
	240
0617bb50 JA	241	static int io_msg_fd_remote(struct io_kiocb *req)
	242	{
	243	struct io_ring_ctx *ctx = req->file->private_data;
	244	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	245	struct task_struct *task = READ_ONCE(ctx->submitter_task);
	246
	247	if (unlikely(!task))
	248	return -EOWNERDEAD;
	249
	250	init_task_work(&msg->tw, io_msg_tw_fd_complete);
	251	if (task_work_add(task, &msg->tw, TWA_SIGNAL))
	252	return -EOWNERDEAD;
	253
	254	return IOU_ISSUE_SKIP_COMPLETE;
	255	}
	256
17211310 PB	257	static int io_msg_send_fd(struct io_kiocb *req, unsigned int issue_flags)
	258	{
	259	struct io_ring_ctx *target_ctx = req->file->private_data;
	260	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
	261	struct io_ring_ctx *ctx = req->ctx;
17211310	262
5da28edd PB	263	if (msg->len)
5da28edd PB	264	return -EINVAL;
17211310 PB	265	if (target_ctx == ctx)
17211310 PB	266	return -EINVAL;
8579538c PB	267	if (target_ctx->flags & IORING_SETUP_R_DISABLED)
8579538c PB	268	return -EBADFD;
b54a1404 JA	269	if (!msg->src_file) {
	270	int ret = io_msg_grab_file(req, issue_flags);
	271	if (unlikely(ret))
	272	return ret;
17211310	273	}
6d043ee1	274
56d8e318	275	if (io_msg_need_remote(target_ctx))
0617bb50	276	return io_msg_fd_remote(req);
17211310 PB	277	return io_msg_install_complete(req, issue_flags);
	278	}
	279
95d6c922	280	static int __io_msg_ring_prep(struct io_msg msg, const struct io_uring_sqe sqe)
36404b09	281	{
e6130eba	282	if (unlikely(sqe->buf_index \|\| sqe->personality))
36404b09 JA	283	return -EINVAL;
36404b09 JA	284
11373026	285	msg->src_file = NULL;
36404b09 JA	286	msg->user_data = READ_ONCE(sqe->off);
36404b09 JA	287	msg->len = READ_ONCE(sqe->len);
e6130eba JA	288	msg->cmd = READ_ONCE(sqe->addr);
	289	msg->src_fd = READ_ONCE(sqe->addr3);
	290	msg->dst_fd = READ_ONCE(sqe->file_index);
	291	msg->flags = READ_ONCE(sqe->msg_ring_flags);
cbeb47a7	292	if (msg->flags & ~IORING_MSG_RING_MASK)
e6130eba JA	293	return -EINVAL;
e6130eba JA	294
36404b09 JA	295	return 0;
	296	}
	297
95d6c922 JA	298	int io_msg_ring_prep(struct io_kiocb req, const struct io_uring_sqe sqe)
	299	{
	300	return __io_msg_ring_prep(io_kiocb_to_cmd(req, struct io_msg), sqe);
	301	}
	302
36404b09 JA	303	int io_msg_ring(struct io_kiocb *req, unsigned int issue_flags)
36404b09 JA	304	{
f2ccb5ae	305	struct io_msg *msg = io_kiocb_to_cmd(req, struct io_msg);
36404b09 JA	306	int ret;
	307
	308	ret = -EBADFD;
	309	if (!io_is_uring_fops(req->file))
	310	goto done;
	311
e6130eba JA	312	switch (msg->cmd) {
e6130eba JA	313	case IORING_MSG_DATA:
e12d7a46	314	ret = io_msg_ring_data(req, issue_flags);
e6130eba JA	315	break;
	316	case IORING_MSG_SEND_FD:
	317	ret = io_msg_send_fd(req, issue_flags);
	318	break;
	319	default:
	320	ret = -EINVAL;
	321	break;
	322	}
36404b09 JA	323
36404b09 JA	324	done:
6d043ee1 PB	325	if (ret < 0) {
	326	if (ret == -EAGAIN \|\| ret == IOU_ISSUE_SKIP_COMPLETE)
	327	return ret;
36404b09	328	req_set_fail(req);
6d043ee1	329	}
36404b09	330	io_req_set_res(req, ret, 0);
8bb9d6cc	331	return IOU_COMPLETE;
36404b09	332	}
50cf5f38	333
a3771321 JA	334	int io_uring_sync_msg_ring(struct io_uring_sqe *sqe)
	335	{
	336	struct io_msg io_msg = { };
a3771321 JA	337	int ret;
	338
	339	ret = __io_msg_ring_prep(&io_msg, sqe);
	340	if (unlikely(ret))
	341	return ret;
	342
	343	/*
	344	* Only data sending supported, not IORING_MSG_SEND_FD as that one
	345	* doesn't make sense without a source ring to send files from.
	346	*/
	347	if (io_msg.cmd != IORING_MSG_DATA)
	348	return -EINVAL;
	349
56cec28d AV	350	CLASS(fd, f)(sqe->fd);
	351	if (fd_empty(f))
	352	return -EBADF;
	353	if (!io_is_uring_fops(fd_file(f)))
	354	return -EBADFD;
	355	return __io_msg_ring_data(fd_file(f)->private_data,
	356	&io_msg, IO_URING_F_UNLOCKED);
a3771321	357	}