xref: /openbmc/linux/include/linux/net.h (revision f7eeb008)
1 /* SPDX-License-Identifier: GPL-2.0-or-later */
2 /*
3  * NET		An implementation of the SOCKET network access protocol.
4  *		This is the master header file for the Linux NET layer,
5  *		or, in plain English: the networking handling part of the
6  *		kernel.
7  *
8  * Version:	@(#)net.h	1.0.3	05/25/93
9  *
10  * Authors:	Orest Zborowski, <obz@Kodak.COM>
11  *		Ross Biro
12  *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13  */
14 #ifndef _LINUX_NET_H
15 #define _LINUX_NET_H
16 
17 #include <linux/stringify.h>
18 #include <linux/random.h>
19 #include <linux/wait.h>
20 #include <linux/fcntl.h>	/* For O_CLOEXEC and O_NONBLOCK */
21 #include <linux/rcupdate.h>
22 #include <linux/once.h>
23 #include <linux/fs.h>
24 #include <linux/mm.h>
25 #include <linux/sockptr.h>
26 
27 #include <uapi/linux/net.h>
28 
29 struct poll_table_struct;
30 struct pipe_inode_info;
31 struct inode;
32 struct file;
33 struct net;
34 
35 /* Historically, SOCKWQ_ASYNC_NOSPACE & SOCKWQ_ASYNC_WAITDATA were located
36  * in sock->flags, but moved into sk->sk_wq->flags to be RCU protected.
37  * Eventually all flags will be in sk->sk_wq->flags.
38  */
39 #define SOCKWQ_ASYNC_NOSPACE	0
40 #define SOCKWQ_ASYNC_WAITDATA	1
41 #define SOCK_NOSPACE		2
42 #define SOCK_PASSCRED		3
43 #define SOCK_PASSSEC		4
44 #define SOCK_SUPPORT_ZC		5
45 
46 #ifndef ARCH_HAS_SOCKET_TYPES
47 /**
48  * enum sock_type - Socket types
49  * @SOCK_STREAM: stream (connection) socket
50  * @SOCK_DGRAM: datagram (conn.less) socket
51  * @SOCK_RAW: raw socket
52  * @SOCK_RDM: reliably-delivered message
53  * @SOCK_SEQPACKET: sequential packet socket
54  * @SOCK_DCCP: Datagram Congestion Control Protocol socket
55  * @SOCK_PACKET: linux specific way of getting packets at the dev level.
56  *		  For writing rarp and other similar things on the user level.
57  *
58  * When adding some new socket type please
59  * grep ARCH_HAS_SOCKET_TYPE include/asm-* /socket.h, at least MIPS
60  * overrides this enum for binary compat reasons.
61  */
62 enum sock_type {
63 	SOCK_STREAM	= 1,
64 	SOCK_DGRAM	= 2,
65 	SOCK_RAW	= 3,
66 	SOCK_RDM	= 4,
67 	SOCK_SEQPACKET	= 5,
68 	SOCK_DCCP	= 6,
69 	SOCK_PACKET	= 10,
70 };
71 
72 #define SOCK_MAX (SOCK_PACKET + 1)
73 /* Mask which covers at least up to SOCK_MASK-1.  The
74  * remaining bits are used as flags. */
75 #define SOCK_TYPE_MASK 0xf
76 
77 /* Flags for socket, socketpair, accept4 */
78 #define SOCK_CLOEXEC	O_CLOEXEC
79 #ifndef SOCK_NONBLOCK
80 #define SOCK_NONBLOCK	O_NONBLOCK
81 #endif
82 
83 #endif /* ARCH_HAS_SOCKET_TYPES */
84 
85 /**
86  * enum sock_shutdown_cmd - Shutdown types
87  * @SHUT_RD: shutdown receptions
88  * @SHUT_WR: shutdown transmissions
89  * @SHUT_RDWR: shutdown receptions/transmissions
90  */
91 enum sock_shutdown_cmd {
92 	SHUT_RD,
93 	SHUT_WR,
94 	SHUT_RDWR,
95 };
96 
97 struct socket_wq {
98 	/* Note: wait MUST be first field of socket_wq */
99 	wait_queue_head_t	wait;
100 	struct fasync_struct	*fasync_list;
101 	unsigned long		flags; /* %SOCKWQ_ASYNC_NOSPACE, etc */
102 	struct rcu_head		rcu;
103 } ____cacheline_aligned_in_smp;
104 
105 /**
106  *  struct socket - general BSD socket
107  *  @state: socket state (%SS_CONNECTED, etc)
108  *  @type: socket type (%SOCK_STREAM, etc)
109  *  @flags: socket flags (%SOCK_NOSPACE, etc)
110  *  @ops: protocol specific socket operations
111  *  @file: File back pointer for gc
112  *  @sk: internal networking protocol agnostic socket representation
113  *  @wq: wait queue for several uses
114  */
115 struct socket {
116 	socket_state		state;
117 
118 	short			type;
119 
120 	unsigned long		flags;
121 
122 	struct file		*file;
123 	struct sock		*sk;
124 	const struct proto_ops	*ops;
125 
126 	struct socket_wq	wq;
127 };
128 
129 /*
130  * "descriptor" for what we're up to with a read.
131  * This allows us to use the same read code yet
132  * have multiple different users of the data that
133  * we read from a file.
134  *
135  * The simplest case just copies the data to user
136  * mode.
137  */
138 typedef struct {
139 	size_t written;
140 	size_t count;
141 	union {
142 		char __user *buf;
143 		void *data;
144 	} arg;
145 	int error;
146 } read_descriptor_t;
147 
148 struct vm_area_struct;
149 struct page;
150 struct sockaddr;
151 struct msghdr;
152 struct module;
153 struct sk_buff;
154 typedef int (*sk_read_actor_t)(read_descriptor_t *, struct sk_buff *,
155 			       unsigned int, size_t);
156 typedef int (*skb_read_actor_t)(struct sock *, struct sk_buff *);
157 
158 
159 struct proto_ops {
160 	int		family;
161 	struct module	*owner;
162 	int		(*release)   (struct socket *sock);
163 	int		(*bind)	     (struct socket *sock,
164 				      struct sockaddr *myaddr,
165 				      int sockaddr_len);
166 	int		(*connect)   (struct socket *sock,
167 				      struct sockaddr *vaddr,
168 				      int sockaddr_len, int flags);
169 	int		(*socketpair)(struct socket *sock1,
170 				      struct socket *sock2);
171 	int		(*accept)    (struct socket *sock,
172 				      struct socket *newsock, int flags, bool kern);
173 	int		(*getname)   (struct socket *sock,
174 				      struct sockaddr *addr,
175 				      int peer);
176 	__poll_t	(*poll)	     (struct file *file, struct socket *sock,
177 				      struct poll_table_struct *wait);
178 	int		(*ioctl)     (struct socket *sock, unsigned int cmd,
179 				      unsigned long arg);
180 #ifdef CONFIG_COMPAT
181 	int	 	(*compat_ioctl) (struct socket *sock, unsigned int cmd,
182 				      unsigned long arg);
183 #endif
184 	int		(*gettstamp) (struct socket *sock, void __user *userstamp,
185 				      bool timeval, bool time32);
186 	int		(*listen)    (struct socket *sock, int len);
187 	int		(*shutdown)  (struct socket *sock, int flags);
188 	int		(*setsockopt)(struct socket *sock, int level,
189 				      int optname, sockptr_t optval,
190 				      unsigned int optlen);
191 	int		(*getsockopt)(struct socket *sock, int level,
192 				      int optname, char __user *optval, int __user *optlen);
193 	void		(*show_fdinfo)(struct seq_file *m, struct socket *sock);
194 	int		(*sendmsg)   (struct socket *sock, struct msghdr *m,
195 				      size_t total_len);
196 	/* Notes for implementing recvmsg:
197 	 * ===============================
198 	 * msg->msg_namelen should get updated by the recvmsg handlers
199 	 * iff msg_name != NULL. It is by default 0 to prevent
200 	 * returning uninitialized memory to user space.  The recvfrom
201 	 * handlers can assume that msg.msg_name is either NULL or has
202 	 * a minimum size of sizeof(struct sockaddr_storage).
203 	 */
204 	int		(*recvmsg)   (struct socket *sock, struct msghdr *m,
205 				      size_t total_len, int flags);
206 	int		(*mmap)	     (struct file *file, struct socket *sock,
207 				      struct vm_area_struct * vma);
208 	ssize_t		(*sendpage)  (struct socket *sock, struct page *page,
209 				      int offset, size_t size, int flags);
210 	ssize_t 	(*splice_read)(struct socket *sock,  loff_t *ppos,
211 				       struct pipe_inode_info *pipe, size_t len, unsigned int flags);
212 	int		(*set_peek_off)(struct sock *sk, int val);
213 	int		(*peek_len)(struct socket *sock);
214 
215 	/* The following functions are called internally by kernel with
216 	 * sock lock already held.
217 	 */
218 	int		(*read_sock)(struct sock *sk, read_descriptor_t *desc,
219 				     sk_read_actor_t recv_actor);
220 	/* This is different from read_sock(), it reads an entire skb at a time. */
221 	int		(*read_skb)(struct sock *sk, skb_read_actor_t recv_actor);
222 	int		(*sendpage_locked)(struct sock *sk, struct page *page,
223 					   int offset, size_t size, int flags);
224 	int		(*sendmsg_locked)(struct sock *sk, struct msghdr *msg,
225 					  size_t size);
226 	int		(*set_rcvlowat)(struct sock *sk, int val);
227 };
228 
229 #define DECLARE_SOCKADDR(type, dst, src)	\
230 	type dst = ({ __sockaddr_check_size(sizeof(*dst)); (type) src; })
231 
232 struct net_proto_family {
233 	int		family;
234 	int		(*create)(struct net *net, struct socket *sock,
235 				  int protocol, int kern);
236 	struct module	*owner;
237 };
238 
239 struct iovec;
240 struct kvec;
241 
242 enum {
243 	SOCK_WAKE_IO,
244 	SOCK_WAKE_WAITD,
245 	SOCK_WAKE_SPACE,
246 	SOCK_WAKE_URG,
247 };
248 
249 int sock_wake_async(struct socket_wq *sk_wq, int how, int band);
250 int sock_register(const struct net_proto_family *fam);
251 void sock_unregister(int family);
252 bool sock_is_registered(int family);
253 int __sock_create(struct net *net, int family, int type, int proto,
254 		  struct socket **res, int kern);
255 int sock_create(int family, int type, int proto, struct socket **res);
256 int sock_create_kern(struct net *net, int family, int type, int proto, struct socket **res);
257 int sock_create_lite(int family, int type, int proto, struct socket **res);
258 struct socket *sock_alloc(void);
259 void sock_release(struct socket *sock);
260 int sock_sendmsg(struct socket *sock, struct msghdr *msg);
261 int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags);
262 struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname);
263 struct socket *sockfd_lookup(int fd, int *err);
264 struct socket *sock_from_file(struct file *file);
265 #define		     sockfd_put(sock) fput(sock->file)
266 int net_ratelimit(void);
267 
268 #define net_ratelimited_function(function, ...)			\
269 do {								\
270 	if (net_ratelimit())					\
271 		function(__VA_ARGS__);				\
272 } while (0)
273 
274 #define net_emerg_ratelimited(fmt, ...)				\
275 	net_ratelimited_function(pr_emerg, fmt, ##__VA_ARGS__)
276 #define net_alert_ratelimited(fmt, ...)				\
277 	net_ratelimited_function(pr_alert, fmt, ##__VA_ARGS__)
278 #define net_crit_ratelimited(fmt, ...)				\
279 	net_ratelimited_function(pr_crit, fmt, ##__VA_ARGS__)
280 #define net_err_ratelimited(fmt, ...)				\
281 	net_ratelimited_function(pr_err, fmt, ##__VA_ARGS__)
282 #define net_notice_ratelimited(fmt, ...)			\
283 	net_ratelimited_function(pr_notice, fmt, ##__VA_ARGS__)
284 #define net_warn_ratelimited(fmt, ...)				\
285 	net_ratelimited_function(pr_warn, fmt, ##__VA_ARGS__)
286 #define net_info_ratelimited(fmt, ...)				\
287 	net_ratelimited_function(pr_info, fmt, ##__VA_ARGS__)
288 #if defined(CONFIG_DYNAMIC_DEBUG) || \
289 	(defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
290 #define net_dbg_ratelimited(fmt, ...)					\
291 do {									\
292 	DEFINE_DYNAMIC_DEBUG_METADATA(descriptor, fmt);			\
293 	if (DYNAMIC_DEBUG_BRANCH(descriptor) &&				\
294 	    net_ratelimit())						\
295 		__dynamic_pr_debug(&descriptor, pr_fmt(fmt),		\
296 		                   ##__VA_ARGS__);			\
297 } while (0)
298 #elif defined(DEBUG)
299 #define net_dbg_ratelimited(fmt, ...)				\
300 	net_ratelimited_function(pr_debug, fmt, ##__VA_ARGS__)
301 #else
302 #define net_dbg_ratelimited(fmt, ...)				\
303 	do {							\
304 		if (0)						\
305 			no_printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__); \
306 	} while (0)
307 #endif
308 
309 #define net_get_random_once(buf, nbytes)			\
310 	get_random_once((buf), (nbytes))
311 
312 /*
313  * E.g. XFS meta- & log-data is in slab pages, or bcache meta
314  * data pages, or other high order pages allocated by
315  * __get_free_pages() without __GFP_COMP, which have a page_count
316  * of 0 and/or have PageSlab() set. We cannot use send_page for
317  * those, as that does get_page(); put_page(); and would cause
318  * either a VM_BUG directly, or __page_cache_release a page that
319  * would actually still be referenced by someone, leading to some
320  * obscure delayed Oops somewhere else.
321  */
322 static inline bool sendpage_ok(struct page *page)
323 {
324 	return !PageSlab(page) && page_count(page) >= 1;
325 }
326 
327 int kernel_sendmsg(struct socket *sock, struct msghdr *msg, struct kvec *vec,
328 		   size_t num, size_t len);
329 int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg,
330 			  struct kvec *vec, size_t num, size_t len);
331 int kernel_recvmsg(struct socket *sock, struct msghdr *msg, struct kvec *vec,
332 		   size_t num, size_t len, int flags);
333 
334 int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen);
335 int kernel_listen(struct socket *sock, int backlog);
336 int kernel_accept(struct socket *sock, struct socket **newsock, int flags);
337 int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen,
338 		   int flags);
339 int kernel_getsockname(struct socket *sock, struct sockaddr *addr);
340 int kernel_getpeername(struct socket *sock, struct sockaddr *addr);
341 int kernel_sendpage(struct socket *sock, struct page *page, int offset,
342 		    size_t size, int flags);
343 int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset,
344 			   size_t size, int flags);
345 int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how);
346 
347 /* Routine returns the IP overhead imposed by a (caller-protected) socket. */
348 u32 kernel_sock_ip_overhead(struct sock *sk);
349 
350 #define MODULE_ALIAS_NETPROTO(proto) \
351 	MODULE_ALIAS("net-pf-" __stringify(proto))
352 
353 #define MODULE_ALIAS_NET_PF_PROTO(pf, proto) \
354 	MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto))
355 
356 #define MODULE_ALIAS_NET_PF_PROTO_TYPE(pf, proto, type) \
357 	MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto) \
358 		     "-type-" __stringify(type))
359 
360 #define MODULE_ALIAS_NET_PF_PROTO_NAME(pf, proto, name) \
361 	MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto) \
362 		     name)
363 #endif	/* _LINUX_NET_H */
364