io_uring: Introduce IORING_OP_LISTEN

IORING_OP_LISTEN provides the semantic of listen(2) via io_uring.  While
this is an essentially synchronous system call, the main point is to
enable a network path to execute fully with io_uring registered and
descriptorless files.

Signed-off-by: Gabriel Krisman Bertazi <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jens Axboe <[email protected]>

Author: krisman

include/uapi/linux/io_uring.h

@@ -258,6 +258,7 @@ enum io_uring_op {
 	IORING_OP_FIXED_FD_INSTALL,
 	IORING_OP_FTRUNCATE,
 	IORING_OP_BIND,
+	IORING_OP_LISTEN,
 
 	/* this goes last, obviously */
 	IORING_OP_LAST,

io_uring/net.c

@@ -56,6 +56,11 @@ struct io_bind {
 	int				addr_len;
 };
 
+struct io_listen {
+	struct file			*file;
+	int				backlog;
+};
+
 struct io_sr_msg {
 	struct file			*file;
 	union {
@@ -1751,6 +1756,29 @@ int io_bind(struct io_kiocb *req, unsigned int issue_flags)
 	return 0;
 }
 
+int io_listen_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
+{
+	struct io_listen *listen = io_kiocb_to_cmd(req, struct io_listen);
+
+	if (sqe->addr || sqe->buf_index || sqe->rw_flags || sqe->splice_fd_in || sqe->addr2)
+		return -EINVAL;
+
+	listen->backlog = READ_ONCE(sqe->len);
+	return 0;
+}
+
+int io_listen(struct io_kiocb *req, unsigned int issue_flags)
+{
+	struct io_listen *listen = io_kiocb_to_cmd(req, struct io_listen);
+	int ret;
+
+	ret = __sys_listen_socket(sock_from_file(req->file), listen->backlog);
+	if (ret < 0)
+		req_set_fail(req);
+	io_req_set_res(req, ret, 0);
+	return 0;
+}
+
 void io_netmsg_cache_free(const void *entry)
 {
 	struct io_async_msghdr *kmsg = (struct io_async_msghdr *) entry;

io_uring/net.h

@@ -52,6 +52,9 @@ void io_send_zc_cleanup(struct io_kiocb *req);
 int io_bind_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
 int io_bind(struct io_kiocb *req, unsigned int issue_flags);
 
+int io_listen_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe);
+int io_listen(struct io_kiocb *req, unsigned int issue_flags);
+
 void io_netmsg_cache_free(const void *entry);
 #else
 static inline void io_netmsg_cache_free(const void *entry)

io_uring/opdef.c

@@ -503,6 +503,16 @@ const struct io_issue_def io_issue_defs[] = {
 		.async_size		= sizeof(struct io_async_msghdr),
 #else
 		.prep			= io_eopnotsupp_prep,
+#endif
+	},
+	[IORING_OP_LISTEN] = {
+#if defined(CONFIG_NET)
+		.needs_file		= 1,
+		.prep			= io_listen_prep,
+		.issue			= io_listen,
+		.async_size		= sizeof(struct io_async_msghdr),
+#else
+		.prep			= io_eopnotsupp_prep,
 #endif
 	},
 };
@@ -729,6 +739,9 @@ const struct io_cold_def io_cold_defs[] = {
 	[IORING_OP_BIND] = {
 		.name			= "BIND",
 	},
+	[IORING_OP_LISTEN] = {
+		.name			= "LISTEN",
+	},
 };
 
 const char *io_uring_get_opcode(u8 opcode)