include/linux/net.h

   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 #define SOCK_CUSTOM_SOCKOPT     6
  46 #define SOCK_PASSPIDFD          7
  47
  48 #ifndef ARCH_HAS_SOCKET_TYPES
  49 /**
  50  * enum sock_type - Socket types
  51  * @SOCK_STREAM: stream (connection) socket
  52  * @SOCK_DGRAM: datagram (conn.less) socket
  53  * @SOCK_RAW: raw socket
  54  * @SOCK_RDM: reliably-delivered message
  55  * @SOCK_SEQPACKET: sequential packet socket
  56  * @SOCK_DCCP: Datagram Congestion Control Protocol socket
  57  * @SOCK_PACKET: linux specific way of getting packets at the dev level.
  58  *                For writing rarp and other similar things on the user level.
  59  *
  60  * When adding some new socket type please
  61  * grep ARCH_HAS_SOCKET_TYPE include/asm-* /socket.h, at least MIPS
  62  * overrides this enum for binary compat reasons.
  63  */
  64 enum sock_type {
  65         SOCK_STREAM     = 1,
  66         SOCK_DGRAM      = 2,
  67         SOCK_RAW        = 3,
  68         SOCK_RDM        = 4,
  69         SOCK_SEQPACKET  = 5,
  70         SOCK_DCCP       = 6,
  71         SOCK_PACKET     = 10,
  72 };
  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
  85 #endif /* ARCH_HAS_SOCKET_TYPES */
  86
  87 /**
  88  * enum sock_shutdown_cmd - Shutdown types
  89  * @SHUT_RD: shutdown receptions
  90  * @SHUT_WR: shutdown transmissions
  91  * @SHUT_RDWR: shutdown receptions/transmissions
  92  */
  93 enum sock_shutdown_cmd {
  94         SHUT_RD,
  95         SHUT_WR,
  96         SHUT_RDWR,
  97 };
  98
  99 struct socket_wq {
 100         /* Note: wait MUST be first field of socket_wq */
 101         wait_queue_head_t       wait;
 102         struct fasync_struct    *fasync_list;
 103         unsigned long           flags; /* %SOCKWQ_ASYNC_NOSPACE, etc */
 104         struct rcu_head         rcu;
 105 } ____cacheline_aligned_in_smp;
 106
 107 /**
 108  *  struct socket - general BSD socket
 109  *  @state: socket state (%SS_CONNECTED, etc)
 110  *  @type: socket type (%SOCK_STREAM, etc)
 111  *  @flags: socket flags (%SOCK_NOSPACE, etc)
 112  *  @ops: protocol specific socket operations
 113  *  @file: File back pointer for gc
 114  *  @sk: internal networking protocol agnostic socket representation
 115  *  @wq: wait queue for several uses
 116  */
 117 struct socket {
 118         socket_state            state;
 119
 120         short                   type;
 121
 122         unsigned long           flags;
 123
 124         struct file             *file;
 125         struct sock             *sk;
 126         const struct proto_ops  *ops; /* Might change with IPV6_ADDRFORM or MPTCP. */
 127
 128         struct socket_wq        wq;
 129 };
 130
 131 /*
 132  * "descriptor" for what we're up to with a read.
 133  * This allows us to use the same read code yet
 134  * have multiple different users of the data that
 135  * we read from a file.
 136  *
 137  * The simplest case just copies the data to user
 138  * mode.
 139  */
 140 typedef struct {
 141         size_t written;
 142         size_t count;
 143         union {
 144                 char __user *buf;
 145                 void *data;
 146         } arg;
 147         int error;
 148 } read_descriptor_t;
 149
 150 struct vm_area_struct;
 151 struct page;
 152 struct sockaddr;
 153 struct msghdr;
 154 struct module;
 155 struct sk_buff;
 156 struct proto_accept_arg;
 157 typedef int (*sk_read_actor_t)(read_descriptor_t *, struct sk_buff *,
 158                                unsigned int, size_t);
 159 typedef int (*skb_read_actor_t)(struct sock *, struct sk_buff *);
 160
 161
 162 struct proto_ops {
 163         int             family;
 164         struct module   *owner;
 165         int             (*release)   (struct socket *sock);
 166         int             (*bind)      (struct socket *sock,
 167                                       struct sockaddr *myaddr,
 168                                       int sockaddr_len);
 169         int             (*connect)   (struct socket *sock,
 170                                       struct sockaddr *vaddr,
 171                                       int sockaddr_len, int flags);
 172         int             (*socketpair)(struct socket *sock1,
 173                                       struct socket *sock2);
 174         int             (*accept)    (struct socket *sock,
 175                                       struct socket *newsock,
 176                                       struct proto_accept_arg *arg);
 177         int             (*getname)   (struct socket *sock,
 178                                       struct sockaddr *addr,
 179                                       int peer);
 180         __poll_t        (*poll)      (struct file *file, struct socket *sock,
 181                                       struct poll_table_struct *wait);
 182         int             (*ioctl)     (struct socket *sock, unsigned int cmd,
 183                                       unsigned long arg);
 184 #ifdef CONFIG_COMPAT
 185         int             (*compat_ioctl) (struct socket *sock, unsigned int cmd,
 186                                       unsigned long arg);
 187 #endif
 188         int             (*gettstamp) (struct socket *sock, void __user *userstamp,
 189                                       bool timeval, bool time32);
 190         int             (*listen)    (struct socket *sock, int len);
 191         int             (*shutdown)  (struct socket *sock, int flags);
 192         int             (*setsockopt)(struct socket *sock, int level,
 193                                       int optname, sockptr_t optval,
 194                                       unsigned int optlen);
 195         int             (*getsockopt)(struct socket *sock, int level,
 196                                       int optname, char __user *optval, int __user *optlen);
 197         void            (*show_fdinfo)(struct seq_file *m, struct socket *sock);
 198         int             (*sendmsg)   (struct socket *sock, struct msghdr *m,
 199                                       size_t total_len);
 200         /* Notes for implementing recvmsg:
 201          * ===============================
 202          * msg->msg_namelen should get updated by the recvmsg handlers
 203          * iff msg_name != NULL. It is by default 0 to prevent
 204          * returning uninitialized memory to user space.  The recvfrom
 205          * handlers can assume that msg.msg_name is either NULL or has
 206          * a minimum size of sizeof(struct sockaddr_storage).
 207          */
 208         int             (*recvmsg)   (struct socket *sock, struct msghdr *m,
 209                                       size_t total_len, int flags);
 210         int             (*mmap)      (struct file *file, struct socket *sock,
 211                                       struct vm_area_struct * vma);
 212         ssize_t         (*splice_read)(struct socket *sock,  loff_t *ppos,
 213                                        struct pipe_inode_info *pipe, size_t len, unsigned int flags);
 214         void            (*splice_eof)(struct socket *sock);
 215         int             (*set_peek_off)(struct sock *sk, int val);
 216         int             (*peek_len)(struct socket *sock);
 217
 218         /* The following functions are called internally by kernel with
 219          * sock lock already held.
 220          */
 221         int             (*read_sock)(struct sock *sk, read_descriptor_t *desc,
 222                                      sk_read_actor_t recv_actor);
 223         /* This is different from read_sock(), it reads an entire skb at a time. */
 224         int             (*read_skb)(struct sock *sk, skb_read_actor_t recv_actor);
 225         int             (*sendmsg_locked)(struct sock *sk, struct msghdr *msg,
 226                                           size_t size);
 227         int             (*set_rcvlowat)(struct sock *sk, int val);
 228 };
 229
 230 #define DECLARE_SOCKADDR(type, dst, src)        \
 231         type dst = ({ __sockaddr_check_size(sizeof(*dst)); (type) src; })
 232
 233 struct net_proto_family {
 234         int             family;
 235         int             (*create)(struct net *net, struct socket *sock,
 236                                   int protocol, int kern);
 237         struct module   *owner;
 238 };
 239
 240 struct iovec;
 241 struct kvec;
 242
 243 enum {
 244         SOCK_WAKE_IO,
 245         SOCK_WAKE_WAITD,
 246         SOCK_WAKE_SPACE,
 247         SOCK_WAKE_URG,
 248 };
 249
 250 int sock_wake_async(struct socket_wq *sk_wq, int how, int band);
 251 int sock_register(const struct net_proto_family *fam);
 252 void sock_unregister(int family);
 253 bool sock_is_registered(int family);
 254 int __sock_create(struct net *net, int family, int type, int proto,
 255                   struct socket **res, int kern);
 256 int sock_create(int family, int type, int proto, struct socket **res);
 257 int sock_create_kern(struct net *net, int family, int type, int proto, struct socket **res);
 258 int sock_create_lite(int family, int type, int proto, struct socket **res);
 259 struct socket *sock_alloc(void);
 260 void sock_release(struct socket *sock);
 261 int sock_sendmsg(struct socket *sock, struct msghdr *msg);
 262 int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags);
 263 struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname);
 264 struct socket *sockfd_lookup(int fd, int *err);
 265 struct socket *sock_from_file(struct file *file);
 266 #define              sockfd_put(sock) fput(sock->file)
 267 int net_ratelimit(void);
 268
 269 #define net_ratelimited_function(function, ...)                 \
 270 do {                                                            \
 271         if (net_ratelimit())                                    \
 272                 function(__VA_ARGS__);                          \
 273 } while (0)
 274
 275 #define net_emerg_ratelimited(fmt, ...)                         \
 276         net_ratelimited_function(pr_emerg, fmt, ##__VA_ARGS__)
 277 #define net_alert_ratelimited(fmt, ...)                         \
 278         net_ratelimited_function(pr_alert, fmt, ##__VA_ARGS__)
 279 #define net_crit_ratelimited(fmt, ...)                          \
 280         net_ratelimited_function(pr_crit, fmt, ##__VA_ARGS__)
 281 #define net_err_ratelimited(fmt, ...)                           \
 282         net_ratelimited_function(pr_err, fmt, ##__VA_ARGS__)
 283 #define net_notice_ratelimited(fmt, ...)                        \
 284         net_ratelimited_function(pr_notice, fmt, ##__VA_ARGS__)
 285 #define net_warn_ratelimited(fmt, ...)                          \
 286         net_ratelimited_function(pr_warn, fmt, ##__VA_ARGS__)
 287 #define net_info_ratelimited(fmt, ...)                          \
 288         net_ratelimited_function(pr_info, fmt, ##__VA_ARGS__)
 289 #if defined(CONFIG_DYNAMIC_DEBUG) || \
 290         (defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
 291 #define net_dbg_ratelimited(fmt, ...)                                   \
 292 do {                                                                    \
 293         DEFINE_DYNAMIC_DEBUG_METADATA(descriptor, fmt);                 \
 294         if (DYNAMIC_DEBUG_BRANCH(descriptor) &&                         \
 295             net_ratelimit())                                            \
 296                 __dynamic_pr_debug(&descriptor, pr_fmt(fmt),            \
 297                                    ##__VA_ARGS__);                      \
 298 } while (0)
 299 #elif defined(DEBUG)
 300 #define net_dbg_ratelimited(fmt, ...)                           \
 301         net_ratelimited_function(pr_debug, fmt, ##__VA_ARGS__)
 302 #else
 303 #define net_dbg_ratelimited(fmt, ...)                           \
 304         no_printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__)
 305 #endif
 306
 307 #define net_get_random_once(buf, nbytes)                        \
 308         get_random_once((buf), (nbytes))
 309
 310 /*
 311  * E.g. XFS meta- & log-data is in slab pages, or bcache meta
 312  * data pages, or other high order pages allocated by
 313  * __get_free_pages() without __GFP_COMP, which have a page_count
 314  * of 0 and/or have PageSlab() set. We cannot use send_page for
 315  * those, as that does get_page(); put_page(); and would cause
 316  * either a VM_BUG directly, or __page_cache_release a page that
 317  * would actually still be referenced by someone, leading to some
 318  * obscure delayed Oops somewhere else.
 319  */
 320 static inline bool sendpage_ok(struct page *page)
 321 {
 322         return !PageSlab(page) && page_count(page) >= 1;
 323 }
 324
 325 /*
 326  * Check sendpage_ok on contiguous pages.
 327  */
 328 static inline bool sendpages_ok(struct page *page, size_t len, size_t offset)
 329 {
 330         struct page *p = page + (offset >> PAGE_SHIFT);
 331         size_t count = 0;
 332
 333         while (count < len) {
 334                 if (!sendpage_ok(p))
 335                         return false;
 336
 337                 p++;
 338                 count += PAGE_SIZE;
 339         }
 340
 341         return true;
 342 }
 343
 344 int kernel_sendmsg(struct socket *sock, struct msghdr *msg, struct kvec *vec,
 345                    size_t num, size_t len);
 346 int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg,
 347                           struct kvec *vec, size_t num, size_t len);
 348 int kernel_recvmsg(struct socket *sock, struct msghdr *msg, struct kvec *vec,
 349                    size_t num, size_t len, int flags);
 350
 351 int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen);
 352 int kernel_listen(struct socket *sock, int backlog);
 353 int kernel_accept(struct socket *sock, struct socket **newsock, int flags);
 354 int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen,
 355                    int flags);
 356 int kernel_getsockname(struct socket *sock, struct sockaddr *addr);
 357 int kernel_getpeername(struct socket *sock, struct sockaddr *addr);
 358 int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how);
 359
 360 /* Routine returns the IP overhead imposed by a (caller-protected) socket. */
 361 u32 kernel_sock_ip_overhead(struct sock *sk);
 362
 363 #define MODULE_ALIAS_NETPROTO(proto) \
 364         MODULE_ALIAS("net-pf-" __stringify(proto))
 365
 366 #define MODULE_ALIAS_NET_PF_PROTO(pf, proto) \
 367         MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto))
 368
 369 #define MODULE_ALIAS_NET_PF_PROTO_TYPE(pf, proto, type) \
 370         MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto) \
 371                      "-type-" __stringify(type))
 372
 373 #define MODULE_ALIAS_NET_PF_PROTO_NAME(pf, proto, name) \
 374         MODULE_ALIAS("net-pf-" __stringify(pf) "-proto-" __stringify(proto) \
 375                      name)
 376 #endif  /* _LINUX_NET_H */