| 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 | enum socket_flags { |
| 40 | SOCKWQ_ASYNC_NOSPACE, |
| 41 | SOCKWQ_ASYNC_WAITDATA, |
| 42 | SOCK_NOSPACE, |
| 43 | SOCK_SUPPORT_ZC, |
| 44 | SOCK_CUSTOM_SOCKOPT, |
| 45 | }; |
| 46 | |
| 47 | #ifndef ARCH_HAS_SOCKET_TYPES |
| 48 | /** |
| 49 | * enum sock_type - Socket types |
| 50 | * @SOCK_STREAM: stream (connection) socket |
| 51 | * @SOCK_DGRAM: datagram (conn.less) socket |
| 52 | * @SOCK_RAW: raw socket |
| 53 | * @SOCK_RDM: reliably-delivered message |
| 54 | * @SOCK_SEQPACKET: sequential packet socket |
| 55 | * @SOCK_DCCP: Datagram Congestion Control Protocol socket |
| 56 | * @SOCK_PACKET: linux specific way of getting packets at the dev level. |
| 57 | * For writing rarp and other similar things on the user level. |
| 58 | * |
| 59 | * When adding some new socket type please |
| 60 | * grep ARCH_HAS_SOCKET_TYPE include/asm-* /socket.h, at least MIPS |
| 61 | * overrides this enum for binary compat reasons. |
| 62 | */ |
| 63 | enum sock_type { |
| 64 | SOCK_STREAM = 1, |
| 65 | SOCK_DGRAM = 2, |
| 66 | SOCK_RAW = 3, |
| 67 | SOCK_RDM = 4, |
| 68 | SOCK_SEQPACKET = 5, |
| 69 | SOCK_DCCP = 6, |
| 70 | SOCK_PACKET = 10, |
| 71 | }; |
| 72 | #endif /* ARCH_HAS_SOCKET_TYPES */ |
| 73 | |
| 74 | #define SOCK_MAX (SOCK_PACKET + 1) |
| 75 | /* Mask which covers at least up to SOCK_MASK-1. The |
| 76 | * remaining bits are used as flags. */ |
| 77 | #define SOCK_TYPE_MASK 0xf |
| 78 | |
| 79 | /* Flags for socket, socketpair, accept4 */ |
| 80 | #define SOCK_CLOEXEC O_CLOEXEC |
| 81 | #ifndef SOCK_NONBLOCK |
| 82 | #define SOCK_NONBLOCK O_NONBLOCK |
| 83 | #endif |
| 84 | #define SOCK_COREDUMP O_NOCTTY |
| 85 | |
| 86 | /** |
| 87 | * enum sock_shutdown_cmd - Shutdown types |
| 88 | * @SHUT_RD: shutdown receptions |
| 89 | * @SHUT_WR: shutdown transmissions |
| 90 | * @SHUT_RDWR: shutdown receptions/transmissions |
| 91 | */ |
| 92 | enum sock_shutdown_cmd { |
| 93 | SHUT_RD, |
| 94 | SHUT_WR, |
| 95 | SHUT_RDWR, |
| 96 | }; |
| 97 | |
| 98 | struct socket_wq { |
| 99 | /* Note: wait MUST be first field of socket_wq */ |
| 100 | wait_queue_head_t wait; |
| 101 | struct fasync_struct *fasync_list; |
| 102 | unsigned long flags; /* %SOCKWQ_ASYNC_NOSPACE, etc */ |
| 103 | struct rcu_head rcu; |
| 104 | } ____cacheline_aligned_in_smp; |
| 105 | |
| 106 | /** |
| 107 | * struct socket - general BSD socket |
| 108 | * @state: socket state (%SS_CONNECTED, etc) |
| 109 | * @type: socket type (%SOCK_STREAM, etc) |
| 110 | * @flags: socket flags (%SOCK_NOSPACE, etc) |
| 111 | * @ops: protocol specific socket operations |
| 112 | * @file: File back pointer for gc |
| 113 | * @sk: internal networking protocol agnostic socket representation |
| 114 | * @wq: wait queue for several uses |
| 115 | */ |
| 116 | struct socket { |
| 117 | socket_state state; |
| 118 | |
| 119 | short type; |
| 120 | |
| 121 | unsigned long flags; |
| 122 | |
| 123 | struct file *file; |
| 124 | struct sock *sk; |
| 125 | const struct proto_ops *ops; /* Might change with IPV6_ADDRFORM or MPTCP. */ |
| 126 | |
| 127 | struct socket_wq wq; |
| 128 | }; |
| 129 | |
| 130 | /* |
| 131 | * "descriptor" for what we're up to with a read. |
| 132 | * This allows us to use the same read code yet |
| 133 | * have multiple different users of the data that |
| 134 | * we read from a file. |
| 135 | * |
| 136 | * The simplest case just copies the data to user |
| 137 | * mode. |
| 138 | */ |
| 139 | typedef struct { |
| 140 | size_t written; |
| 141 | size_t count; |
| 142 | union { |
| 143 | char __user *buf; |
| 144 | void *data; |
| 145 | } arg; |
| 146 | int error; |
| 147 | } read_descriptor_t; |
| 148 | |
| 149 | struct vm_area_struct; |
| 150 | struct page; |
| 151 | struct sockaddr; |
| 152 | struct msghdr; |
| 153 | struct module; |
| 154 | struct sk_buff; |
| 155 | struct proto_accept_arg; |
| 156 | typedef int (*sk_read_actor_t)(read_descriptor_t *, struct sk_buff *, |
| 157 | unsigned int, size_t); |
| 158 | typedef int (*skb_read_actor_t)(struct sock *, struct sk_buff *); |
| 159 | |
| 160 | |
| 161 | struct proto_ops { |
| 162 | int family; |
| 163 | struct module *owner; |
| 164 | int (*release) (struct socket *sock); |
| 165 | int (*bind) (struct socket *sock, |
| 166 | struct sockaddr *myaddr, |
| 167 | int sockaddr_len); |
| 168 | int (*connect) (struct socket *sock, |
| 169 | struct sockaddr *vaddr, |
| 170 | int sockaddr_len, int flags); |
| 171 | int (*socketpair)(struct socket *sock1, |
| 172 | struct socket *sock2); |
| 173 | int (*accept) (struct socket *sock, |
| 174 | struct socket *newsock, |
| 175 | struct proto_accept_arg *arg); |
| 176 | int (*getname) (struct socket *sock, |
| 177 | struct sockaddr *addr, |
| 178 | int peer); |
| 179 | __poll_t (*poll) (struct file *file, struct socket *sock, |
| 180 | struct poll_table_struct *wait); |
| 181 | int (*ioctl) (struct socket *sock, unsigned int cmd, |
| 182 | unsigned long arg); |
| 183 | #ifdef CONFIG_COMPAT |
| 184 | int (*compat_ioctl) (struct socket *sock, unsigned int cmd, |
| 185 | unsigned long arg); |
| 186 | #endif |
| 187 | int (*gettstamp) (struct socket *sock, void __user *userstamp, |
| 188 | bool timeval, bool time32); |
| 189 | int (*listen) (struct socket *sock, int len); |
| 190 | int (*shutdown) (struct socket *sock, int flags); |
| 191 | int (*setsockopt)(struct socket *sock, int level, |
| 192 | int optname, sockptr_t optval, |
| 193 | unsigned int optlen); |
| 194 | int (*getsockopt)(struct socket *sock, int level, |
| 195 | int optname, char __user *optval, int __user *optlen); |
| 196 | void (*show_fdinfo)(struct seq_file *m, struct socket *sock); |
| 197 | int (*sendmsg) (struct socket *sock, struct msghdr *m, |
| 198 | size_t total_len); |
| 199 | /* Notes for implementing recvmsg: |
| 200 | * =============================== |
| 201 | * msg->msg_namelen should get updated by the recvmsg handlers |
| 202 | * iff msg_name != NULL. It is by default 0 to prevent |
| 203 | * returning uninitialized memory to user space. The recvfrom |
| 204 | * handlers can assume that msg.msg_name is either NULL or has |
| 205 | * a minimum size of sizeof(struct sockaddr_storage). |
| 206 | */ |
| 207 | int (*recvmsg) (struct socket *sock, struct msghdr *m, |
| 208 | size_t total_len, int flags); |
| 209 | int (*mmap) (struct file *file, struct socket *sock, |
| 210 | struct vm_area_struct * vma); |
| 211 | ssize_t (*splice_read)(struct socket *sock, loff_t *ppos, |
| 212 | struct pipe_inode_info *pipe, size_t len, unsigned int flags); |
| 213 | void (*splice_eof)(struct socket *sock); |
| 214 | int (*set_peek_off)(struct sock *sk, int val); |
| 215 | int (*peek_len)(struct socket *sock); |
| 216 | |
| 217 | /* The following functions are called internally by kernel with |
| 218 | * sock lock already held. |
| 219 | */ |
| 220 | int (*read_sock)(struct sock *sk, read_descriptor_t *desc, |
| 221 | sk_read_actor_t recv_actor); |
| 222 | /* This is different from read_sock(), it reads an entire skb at a time. */ |
| 223 | int (*read_skb)(struct sock *sk, skb_read_actor_t recv_actor); |
| 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 | no_printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__) |
| 304 | #endif |
| 305 | |
| 306 | #define net_get_random_once(buf, nbytes) \ |
| 307 | get_random_once((buf), (nbytes)) |
| 308 | |
| 309 | /* |
| 310 | * E.g. XFS meta- & log-data is in slab pages, or bcache meta |
| 311 | * data pages, or other high order pages allocated by |
| 312 | * __get_free_pages() without __GFP_COMP, which have a page_count |
| 313 | * of 0 and/or have PageSlab() set. We cannot use send_page for |
| 314 | * those, as that does get_page(); put_page(); and would cause |
| 315 | * either a VM_BUG directly, or __page_cache_release a page that |
| 316 | * would actually still be referenced by someone, leading to some |
| 317 | * obscure delayed Oops somewhere else. |
| 318 | */ |
| 319 | static inline bool sendpage_ok(struct page *page) |
| 320 | { |
| 321 | return !PageSlab(page) && page_count(page) >= 1; |
| 322 | } |
| 323 | |
| 324 | /* |
| 325 | * Check sendpage_ok on contiguous pages. |
| 326 | */ |
| 327 | static inline bool sendpages_ok(struct page *page, size_t len, size_t offset) |
| 328 | { |
| 329 | struct page *p = page + (offset >> PAGE_SHIFT); |
| 330 | size_t count = 0; |
| 331 | |
| 332 | while (count < len) { |
| 333 | if (!sendpage_ok(page: p)) |
| 334 | return false; |
| 335 | |
| 336 | p++; |
| 337 | count += PAGE_SIZE; |
| 338 | } |
| 339 | |
| 340 | return true; |
| 341 | } |
| 342 | |
| 343 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, struct kvec *vec, |
| 344 | size_t num, size_t len); |
| 345 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, struct kvec *vec, |
| 346 | size_t num, size_t len, int flags); |
| 347 | |
| 348 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen); |
| 349 | int kernel_listen(struct socket *sock, int backlog); |
| 350 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags); |
| 351 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
| 352 | int flags); |
| 353 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr); |
| 354 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr); |
| 355 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how); |
| 356 | |
| 357 | /* Routine returns the IP overhead imposed by a (caller-protected) socket. */ |
| 358 | u32 kernel_sock_ip_overhead(struct sock *sk); |
| 359 | |
| 360 | #define MODULE_ALIAS_NETPROTO(proto) \ |
| 361 | MODULE_ALIAS("net-pf-" __stringify(proto)) |
| 362 | |
| 363 | #define MODULE_ALIAS_NET_PF_PROTO(pf, proto) \ |
| 364 | MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto)) |
| 365 | |
| 366 | #define MODULE_ALIAS_NET_PF_PROTO_TYPE(pf, proto, type) \ |
| 367 | MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto) \ |
| 368 | "-type-" __stringify(type)) |
| 369 | |
| 370 | #define MODULE_ALIAS_NET_PF_PROTO_NAME(pf, proto, name) \ |
| 371 | MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto) \ |
| 372 | name) |
| 373 | #endif /* _LINUX_NET_H */ |
| 374 | |