David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 1 | /* SPDX-License-Identifier: GPL-2.0-or-later */ |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 2 | /* |
| 3 | * INET An implementation of the TCP/IP protocol suite for the LINUX |
| 4 | * operating system. INET is implemented using the BSD Socket |
| 5 | * interface as the means of communication with the user level. |
| 6 | * |
| 7 | * Definitions for the IP module. |
| 8 | * |
| 9 | * Version: @(#)ip.h 1.0.2 05/07/93 |
| 10 | * |
| 11 | * Authors: Ross Biro |
| 12 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
| 13 | * Alan Cox, <gw4pts@gw4pts.ampr.org> |
| 14 | * |
| 15 | * Changes: |
| 16 | * Mike McLagan : Routing by source |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 17 | */ |
| 18 | #ifndef _IP_H |
| 19 | #define _IP_H |
| 20 | |
| 21 | #include <linux/types.h> |
| 22 | #include <linux/ip.h> |
| 23 | #include <linux/in.h> |
| 24 | #include <linux/skbuff.h> |
| 25 | #include <linux/jhash.h> |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 26 | #include <linux/sockptr.h> |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 27 | |
| 28 | #include <net/inet_sock.h> |
| 29 | #include <net/route.h> |
| 30 | #include <net/snmp.h> |
| 31 | #include <net/flow.h> |
| 32 | #include <net/flow_dissector.h> |
| 33 | #include <net/netns/hash.h> |
Olivier Deprez | 0e64123 | 2021-09-23 10:07:05 +0200 | [diff] [blame] | 34 | #include <net/lwtunnel.h> |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 35 | |
| 36 | #define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */ |
| 37 | #define IPV4_MIN_MTU 68 /* RFC 791 */ |
| 38 | |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 39 | extern unsigned int sysctl_fib_sync_mem; |
| 40 | extern unsigned int sysctl_fib_sync_mem_min; |
| 41 | extern unsigned int sysctl_fib_sync_mem_max; |
| 42 | |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 43 | struct sock; |
| 44 | |
| 45 | struct inet_skb_parm { |
| 46 | int iif; |
| 47 | struct ip_options opt; /* Compiled IP options */ |
| 48 | u16 flags; |
| 49 | |
| 50 | #define IPSKB_FORWARDED BIT(0) |
| 51 | #define IPSKB_XFRM_TUNNEL_SIZE BIT(1) |
| 52 | #define IPSKB_XFRM_TRANSFORMED BIT(2) |
| 53 | #define IPSKB_FRAG_COMPLETE BIT(3) |
| 54 | #define IPSKB_REROUTED BIT(4) |
| 55 | #define IPSKB_DOREDIRECT BIT(5) |
| 56 | #define IPSKB_FRAG_PMTU BIT(6) |
| 57 | #define IPSKB_L3SLAVE BIT(7) |
Olivier Deprez | 92d4c21 | 2022-12-06 15:05:30 +0100 | [diff] [blame] | 58 | #define IPSKB_NOPOLICY BIT(8) |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 59 | |
| 60 | u16 frag_max_size; |
| 61 | }; |
| 62 | |
| 63 | static inline bool ipv4_l3mdev_skb(u16 flags) |
| 64 | { |
| 65 | return !!(flags & IPSKB_L3SLAVE); |
| 66 | } |
| 67 | |
| 68 | static inline unsigned int ip_hdrlen(const struct sk_buff *skb) |
| 69 | { |
| 70 | return ip_hdr(skb)->ihl * 4; |
| 71 | } |
| 72 | |
| 73 | struct ipcm_cookie { |
| 74 | struct sockcm_cookie sockc; |
| 75 | __be32 addr; |
| 76 | int oif; |
| 77 | struct ip_options_rcu *opt; |
| 78 | __u8 ttl; |
| 79 | __s16 tos; |
| 80 | char priority; |
| 81 | __u16 gso_size; |
| 82 | }; |
| 83 | |
| 84 | static inline void ipcm_init(struct ipcm_cookie *ipcm) |
| 85 | { |
| 86 | *ipcm = (struct ipcm_cookie) { .tos = -1 }; |
| 87 | } |
| 88 | |
| 89 | static inline void ipcm_init_sk(struct ipcm_cookie *ipcm, |
| 90 | const struct inet_sock *inet) |
| 91 | { |
| 92 | ipcm_init(ipcm); |
| 93 | |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 94 | ipcm->sockc.mark = inet->sk.sk_mark; |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 95 | ipcm->sockc.tsflags = inet->sk.sk_tsflags; |
| 96 | ipcm->oif = inet->sk.sk_bound_dev_if; |
| 97 | ipcm->addr = inet->inet_saddr; |
| 98 | } |
| 99 | |
| 100 | #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb)) |
| 101 | #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb)) |
| 102 | |
| 103 | /* return enslaved device index if relevant */ |
| 104 | static inline int inet_sdif(struct sk_buff *skb) |
| 105 | { |
| 106 | #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV) |
| 107 | if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags)) |
| 108 | return IPCB(skb)->iif; |
| 109 | #endif |
| 110 | return 0; |
| 111 | } |
| 112 | |
| 113 | /* Special input handler for packets caught by router alert option. |
| 114 | They are selected only by protocol field, and then processed likely |
| 115 | local ones; but only if someone wants them! Otherwise, router |
| 116 | not running rsvpd will kill RSVP. |
| 117 | |
| 118 | It is user level problem, what it will make with them. |
| 119 | I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), |
| 120 | but receiver should be enough clever f.e. to forward mtrace requests, |
| 121 | sent to multicast group to reach destination designated router. |
| 122 | */ |
| 123 | |
| 124 | struct ip_ra_chain { |
| 125 | struct ip_ra_chain __rcu *next; |
| 126 | struct sock *sk; |
| 127 | union { |
| 128 | void (*destructor)(struct sock *); |
| 129 | struct sock *saved_sk; |
| 130 | }; |
| 131 | struct rcu_head rcu; |
| 132 | }; |
| 133 | |
| 134 | /* IP flags. */ |
| 135 | #define IP_CE 0x8000 /* Flag: "Congestion" */ |
| 136 | #define IP_DF 0x4000 /* Flag: "Don't Fragment" */ |
| 137 | #define IP_MF 0x2000 /* Flag: "More Fragments" */ |
| 138 | #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */ |
| 139 | |
| 140 | #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */ |
| 141 | |
| 142 | struct msghdr; |
| 143 | struct net_device; |
| 144 | struct packet_type; |
| 145 | struct rtable; |
| 146 | struct sockaddr; |
| 147 | |
| 148 | int igmp_mc_init(void); |
| 149 | |
| 150 | /* |
| 151 | * Functions provided by ip.c |
| 152 | */ |
| 153 | |
| 154 | int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk, |
| 155 | __be32 saddr, __be32 daddr, |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 156 | struct ip_options_rcu *opt, u8 tos); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 157 | int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, |
| 158 | struct net_device *orig_dev); |
| 159 | void ip_list_rcv(struct list_head *head, struct packet_type *pt, |
| 160 | struct net_device *orig_dev); |
| 161 | int ip_local_deliver(struct sk_buff *skb); |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 162 | void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 163 | int ip_mr_input(struct sk_buff *skb); |
| 164 | int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb); |
| 165 | int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb); |
| 166 | int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, |
| 167 | int (*output)(struct net *, struct sock *, struct sk_buff *)); |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 168 | |
| 169 | struct ip_fraglist_iter { |
| 170 | struct sk_buff *frag; |
| 171 | struct iphdr *iph; |
| 172 | int offset; |
| 173 | unsigned int hlen; |
| 174 | }; |
| 175 | |
| 176 | void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph, |
| 177 | unsigned int hlen, struct ip_fraglist_iter *iter); |
| 178 | void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter); |
| 179 | |
| 180 | static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter) |
| 181 | { |
| 182 | struct sk_buff *skb = iter->frag; |
| 183 | |
| 184 | iter->frag = skb->next; |
| 185 | skb_mark_not_on_list(skb); |
| 186 | |
| 187 | return skb; |
| 188 | } |
| 189 | |
| 190 | struct ip_frag_state { |
| 191 | bool DF; |
| 192 | unsigned int hlen; |
| 193 | unsigned int ll_rs; |
| 194 | unsigned int mtu; |
| 195 | unsigned int left; |
| 196 | int offset; |
| 197 | int ptr; |
| 198 | __be16 not_last_frag; |
| 199 | }; |
| 200 | |
| 201 | void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs, |
| 202 | unsigned int mtu, bool DF, struct ip_frag_state *state); |
| 203 | struct sk_buff *ip_frag_next(struct sk_buff *skb, |
| 204 | struct ip_frag_state *state); |
| 205 | |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 206 | void ip_send_check(struct iphdr *ip); |
| 207 | int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); |
| 208 | int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); |
| 209 | |
| 210 | int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl, |
| 211 | __u8 tos); |
| 212 | void ip_init(void); |
| 213 | int ip_append_data(struct sock *sk, struct flowi4 *fl4, |
| 214 | int getfrag(void *from, char *to, int offset, int len, |
| 215 | int odd, struct sk_buff *skb), |
| 216 | void *from, int len, int protolen, |
| 217 | struct ipcm_cookie *ipc, |
| 218 | struct rtable **rt, |
| 219 | unsigned int flags); |
| 220 | int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, |
| 221 | struct sk_buff *skb); |
| 222 | ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page, |
| 223 | int offset, size_t size, int flags); |
| 224 | struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, |
| 225 | struct sk_buff_head *queue, |
| 226 | struct inet_cork *cork); |
| 227 | int ip_send_skb(struct net *net, struct sk_buff *skb); |
| 228 | int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); |
| 229 | void ip_flush_pending_frames(struct sock *sk); |
| 230 | struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4, |
| 231 | int getfrag(void *from, char *to, int offset, |
| 232 | int len, int odd, struct sk_buff *skb), |
| 233 | void *from, int length, int transhdrlen, |
| 234 | struct ipcm_cookie *ipc, struct rtable **rtp, |
| 235 | struct inet_cork *cork, unsigned int flags); |
| 236 | |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 237 | int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 238 | |
| 239 | static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4) |
| 240 | { |
| 241 | return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base); |
| 242 | } |
| 243 | |
| 244 | static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet) |
| 245 | { |
| 246 | return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos); |
| 247 | } |
| 248 | |
| 249 | static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk) |
| 250 | { |
| 251 | return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk); |
| 252 | } |
| 253 | |
| 254 | /* datagram.c */ |
| 255 | int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); |
| 256 | int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); |
| 257 | |
| 258 | void ip4_datagram_release_cb(struct sock *sk); |
| 259 | |
| 260 | struct ip_reply_arg { |
| 261 | struct kvec iov[1]; |
| 262 | int flags; |
| 263 | __wsum csum; |
| 264 | int csumoffset; /* u16 offset of csum in iov[0].iov_base */ |
| 265 | /* -1 if not needed */ |
| 266 | int bound_dev_if; |
| 267 | u8 tos; |
| 268 | kuid_t uid; |
| 269 | }; |
| 270 | |
| 271 | #define IP_REPLY_ARG_NOSRCCHECK 1 |
| 272 | |
| 273 | static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg) |
| 274 | { |
| 275 | return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0; |
| 276 | } |
| 277 | |
| 278 | void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, |
| 279 | const struct ip_options *sopt, |
| 280 | __be32 daddr, __be32 saddr, |
| 281 | const struct ip_reply_arg *arg, |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 282 | unsigned int len, u64 transmit_time); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 283 | |
| 284 | #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field) |
| 285 | #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field) |
| 286 | #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) |
| 287 | #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) |
| 288 | #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) |
| 289 | #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) |
| 290 | #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field) |
| 291 | #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field) |
| 292 | #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) |
| 293 | #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) |
| 294 | |
| 295 | u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct); |
| 296 | unsigned long snmp_fold_field(void __percpu *mib, int offt); |
| 297 | #if BITS_PER_LONG==32 |
| 298 | u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, |
| 299 | size_t syncp_offset); |
| 300 | u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off); |
| 301 | #else |
| 302 | static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, |
| 303 | size_t syncp_offset) |
| 304 | { |
| 305 | return snmp_get_cpu_field(mib, cpu, offct); |
| 306 | |
| 307 | } |
| 308 | |
| 309 | static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off) |
| 310 | { |
| 311 | return snmp_fold_field(mib, offt); |
| 312 | } |
| 313 | #endif |
| 314 | |
| 315 | #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \ |
| 316 | { \ |
| 317 | int i, c; \ |
| 318 | for_each_possible_cpu(c) { \ |
| 319 | for (i = 0; stats_list[i].name; i++) \ |
| 320 | buff64[i] += snmp_get_cpu_field64( \ |
| 321 | mib_statistic, \ |
| 322 | c, stats_list[i].entry, \ |
| 323 | offset); \ |
| 324 | } \ |
| 325 | } |
| 326 | |
| 327 | #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \ |
| 328 | { \ |
| 329 | int i, c; \ |
| 330 | for_each_possible_cpu(c) { \ |
| 331 | for (i = 0; stats_list[i].name; i++) \ |
| 332 | buff[i] += snmp_get_cpu_field( \ |
| 333 | mib_statistic, \ |
| 334 | c, stats_list[i].entry); \ |
| 335 | } \ |
| 336 | } |
| 337 | |
| 338 | void inet_get_local_port_range(struct net *net, int *low, int *high); |
| 339 | |
| 340 | #ifdef CONFIG_SYSCTL |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 341 | static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port) |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 342 | { |
| 343 | if (!net->ipv4.sysctl_local_reserved_ports) |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 344 | return false; |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 345 | return test_bit(port, net->ipv4.sysctl_local_reserved_ports); |
| 346 | } |
| 347 | |
| 348 | static inline bool sysctl_dev_name_is_allowed(const char *name) |
| 349 | { |
| 350 | return strcmp(name, "default") != 0 && strcmp(name, "all") != 0; |
| 351 | } |
| 352 | |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 353 | static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 354 | { |
Olivier Deprez | 92d4c21 | 2022-12-06 15:05:30 +0100 | [diff] [blame] | 355 | return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 356 | } |
| 357 | |
| 358 | #else |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 359 | static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port) |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 360 | { |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 361 | return false; |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 362 | } |
| 363 | |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 364 | static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 365 | { |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 366 | return port < PROT_SOCK; |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 367 | } |
| 368 | #endif |
| 369 | |
| 370 | __be32 inet_current_timestamp(void); |
| 371 | |
| 372 | /* From inetpeer.c */ |
| 373 | extern int inet_peer_threshold; |
| 374 | extern int inet_peer_minttl; |
| 375 | extern int inet_peer_maxttl; |
| 376 | |
| 377 | void ipfrag_init(void); |
| 378 | |
| 379 | void ip_static_sysctl_init(void); |
| 380 | |
| 381 | #define IP4_REPLY_MARK(net, mark) \ |
Olivier Deprez | 92d4c21 | 2022-12-06 15:05:30 +0100 | [diff] [blame] | 382 | (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0) |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 383 | |
| 384 | static inline bool ip_is_fragment(const struct iphdr *iph) |
| 385 | { |
| 386 | return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0; |
| 387 | } |
| 388 | |
| 389 | #ifdef CONFIG_INET |
| 390 | #include <net/dst.h> |
| 391 | |
| 392 | /* The function in 2.2 was invalid, producing wrong result for |
| 393 | * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */ |
| 394 | static inline |
| 395 | int ip_decrease_ttl(struct iphdr *iph) |
| 396 | { |
| 397 | u32 check = (__force u32)iph->check; |
| 398 | check += (__force u32)htons(0x0100); |
| 399 | iph->check = (__force __sum16)(check + (check>=0xFFFF)); |
| 400 | return --iph->ttl; |
| 401 | } |
| 402 | |
| 403 | static inline int ip_mtu_locked(const struct dst_entry *dst) |
| 404 | { |
| 405 | const struct rtable *rt = (const struct rtable *)dst; |
| 406 | |
| 407 | return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU); |
| 408 | } |
| 409 | |
| 410 | static inline |
| 411 | int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst) |
| 412 | { |
| 413 | u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); |
| 414 | |
| 415 | return pmtudisc == IP_PMTUDISC_DO || |
| 416 | (pmtudisc == IP_PMTUDISC_WANT && |
| 417 | !ip_mtu_locked(dst)); |
| 418 | } |
| 419 | |
| 420 | static inline bool ip_sk_accept_pmtu(const struct sock *sk) |
| 421 | { |
| 422 | return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE && |
| 423 | inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT; |
| 424 | } |
| 425 | |
| 426 | static inline bool ip_sk_use_pmtu(const struct sock *sk) |
| 427 | { |
| 428 | return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE; |
| 429 | } |
| 430 | |
| 431 | static inline bool ip_sk_ignore_df(const struct sock *sk) |
| 432 | { |
| 433 | return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO || |
| 434 | inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT; |
| 435 | } |
| 436 | |
| 437 | static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst, |
| 438 | bool forwarding) |
| 439 | { |
| 440 | struct net *net = dev_net(dst->dev); |
Olivier Deprez | 0e64123 | 2021-09-23 10:07:05 +0200 | [diff] [blame] | 441 | unsigned int mtu; |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 442 | |
Olivier Deprez | 92d4c21 | 2022-12-06 15:05:30 +0100 | [diff] [blame] | 443 | if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) || |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 444 | ip_mtu_locked(dst) || |
| 445 | !forwarding) |
| 446 | return dst_mtu(dst); |
| 447 | |
Olivier Deprez | 0e64123 | 2021-09-23 10:07:05 +0200 | [diff] [blame] | 448 | /* 'forwarding = true' case should always honour route mtu */ |
| 449 | mtu = dst_metric_raw(dst, RTAX_MTU); |
| 450 | if (!mtu) |
| 451 | mtu = min(READ_ONCE(dst->dev->mtu), IP_MAX_MTU); |
| 452 | |
| 453 | return mtu - lwtunnel_headroom(dst->lwtstate, mtu); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 454 | } |
| 455 | |
| 456 | static inline unsigned int ip_skb_dst_mtu(struct sock *sk, |
| 457 | const struct sk_buff *skb) |
| 458 | { |
Olivier Deprez | 0e64123 | 2021-09-23 10:07:05 +0200 | [diff] [blame] | 459 | unsigned int mtu; |
| 460 | |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 461 | if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) { |
| 462 | bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; |
| 463 | |
| 464 | return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding); |
| 465 | } |
| 466 | |
Olivier Deprez | 0e64123 | 2021-09-23 10:07:05 +0200 | [diff] [blame] | 467 | mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU); |
| 468 | return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 469 | } |
| 470 | |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 471 | struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx, |
| 472 | int fc_mx_len, |
| 473 | struct netlink_ext_ack *extack); |
| 474 | static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics) |
| 475 | { |
| 476 | if (fib_metrics != &dst_default_metrics && |
| 477 | refcount_dec_and_test(&fib_metrics->refcnt)) |
| 478 | kfree(fib_metrics); |
| 479 | } |
| 480 | |
| 481 | /* ipv4 and ipv6 both use refcounted metrics if it is not the default */ |
| 482 | static inline |
| 483 | void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics) |
| 484 | { |
| 485 | dst_init_metrics(dst, fib_metrics->metrics, true); |
| 486 | |
| 487 | if (fib_metrics != &dst_default_metrics) { |
| 488 | dst->_metrics |= DST_METRICS_REFCOUNTED; |
| 489 | refcount_inc(&fib_metrics->refcnt); |
| 490 | } |
| 491 | } |
| 492 | |
| 493 | static inline |
| 494 | void ip_dst_metrics_put(struct dst_entry *dst) |
| 495 | { |
| 496 | struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst); |
| 497 | |
| 498 | if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt)) |
| 499 | kfree(p); |
| 500 | } |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 501 | |
| 502 | u32 ip_idents_reserve(u32 hash, int segs); |
| 503 | void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); |
| 504 | |
| 505 | static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, |
| 506 | struct sock *sk, int segs) |
| 507 | { |
| 508 | struct iphdr *iph = ip_hdr(skb); |
| 509 | |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 510 | /* We had many attacks based on IPID, use the private |
| 511 | * generator as much as we can. |
| 512 | */ |
| 513 | if (sk && inet_sk(sk)->inet_daddr) { |
| 514 | iph->id = htons(inet_sk(sk)->inet_id); |
| 515 | inet_sk(sk)->inet_id += segs; |
| 516 | return; |
| 517 | } |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 518 | if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 519 | iph->id = 0; |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 520 | } else { |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 521 | /* Unfortunately we need the big hammer to get a suitable IPID */ |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 522 | __ip_select_ident(net, iph, segs); |
| 523 | } |
| 524 | } |
| 525 | |
| 526 | static inline void ip_select_ident(struct net *net, struct sk_buff *skb, |
| 527 | struct sock *sk) |
| 528 | { |
| 529 | ip_select_ident_segs(net, skb, sk, 1); |
| 530 | } |
| 531 | |
| 532 | static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) |
| 533 | { |
| 534 | return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr, |
| 535 | skb->len, proto, 0); |
| 536 | } |
| 537 | |
| 538 | /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store |
| 539 | * Equivalent to : flow->v4addrs.src = iph->saddr; |
| 540 | * flow->v4addrs.dst = iph->daddr; |
| 541 | */ |
| 542 | static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow, |
| 543 | const struct iphdr *iph) |
| 544 | { |
| 545 | BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) != |
| 546 | offsetof(typeof(flow->addrs), v4addrs.src) + |
| 547 | sizeof(flow->addrs.v4addrs.src)); |
Olivier Deprez | 92d4c21 | 2022-12-06 15:05:30 +0100 | [diff] [blame] | 548 | memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs)); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 549 | flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; |
| 550 | } |
| 551 | |
| 552 | static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto) |
| 553 | { |
| 554 | const struct iphdr *iph = skb_gro_network_header(skb); |
| 555 | |
| 556 | return csum_tcpudp_nofold(iph->saddr, iph->daddr, |
| 557 | skb_gro_len(skb), proto, 0); |
| 558 | } |
| 559 | |
| 560 | /* |
| 561 | * Map a multicast IP onto multicast MAC for type ethernet. |
| 562 | */ |
| 563 | |
| 564 | static inline void ip_eth_mc_map(__be32 naddr, char *buf) |
| 565 | { |
| 566 | __u32 addr=ntohl(naddr); |
| 567 | buf[0]=0x01; |
| 568 | buf[1]=0x00; |
| 569 | buf[2]=0x5e; |
| 570 | buf[5]=addr&0xFF; |
| 571 | addr>>=8; |
| 572 | buf[4]=addr&0xFF; |
| 573 | addr>>=8; |
| 574 | buf[3]=addr&0x7F; |
| 575 | } |
| 576 | |
| 577 | /* |
| 578 | * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. |
| 579 | * Leave P_Key as 0 to be filled in by driver. |
| 580 | */ |
| 581 | |
| 582 | static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) |
| 583 | { |
| 584 | __u32 addr; |
| 585 | unsigned char scope = broadcast[5] & 0xF; |
| 586 | |
| 587 | buf[0] = 0; /* Reserved */ |
| 588 | buf[1] = 0xff; /* Multicast QPN */ |
| 589 | buf[2] = 0xff; |
| 590 | buf[3] = 0xff; |
| 591 | addr = ntohl(naddr); |
| 592 | buf[4] = 0xff; |
| 593 | buf[5] = 0x10 | scope; /* scope from broadcast address */ |
| 594 | buf[6] = 0x40; /* IPv4 signature */ |
| 595 | buf[7] = 0x1b; |
| 596 | buf[8] = broadcast[8]; /* P_Key */ |
| 597 | buf[9] = broadcast[9]; |
| 598 | buf[10] = 0; |
| 599 | buf[11] = 0; |
| 600 | buf[12] = 0; |
| 601 | buf[13] = 0; |
| 602 | buf[14] = 0; |
| 603 | buf[15] = 0; |
| 604 | buf[19] = addr & 0xff; |
| 605 | addr >>= 8; |
| 606 | buf[18] = addr & 0xff; |
| 607 | addr >>= 8; |
| 608 | buf[17] = addr & 0xff; |
| 609 | addr >>= 8; |
| 610 | buf[16] = addr & 0x0f; |
| 611 | } |
| 612 | |
| 613 | static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) |
| 614 | { |
| 615 | if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) |
| 616 | memcpy(buf, broadcast, 4); |
| 617 | else |
| 618 | memcpy(buf, &naddr, sizeof(naddr)); |
| 619 | } |
| 620 | |
| 621 | #if IS_ENABLED(CONFIG_IPV6) |
| 622 | #include <linux/ipv6.h> |
| 623 | #endif |
| 624 | |
| 625 | static __inline__ void inet_reset_saddr(struct sock *sk) |
| 626 | { |
| 627 | inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; |
| 628 | #if IS_ENABLED(CONFIG_IPV6) |
| 629 | if (sk->sk_family == PF_INET6) { |
| 630 | struct ipv6_pinfo *np = inet6_sk(sk); |
| 631 | |
| 632 | memset(&np->saddr, 0, sizeof(np->saddr)); |
| 633 | memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); |
| 634 | } |
| 635 | #endif |
| 636 | } |
| 637 | |
| 638 | #endif |
| 639 | |
| 640 | static inline unsigned int ipv4_addr_hash(__be32 ip) |
| 641 | { |
| 642 | return (__force unsigned int) ip; |
| 643 | } |
| 644 | |
| 645 | static inline u32 ipv4_portaddr_hash(const struct net *net, |
| 646 | __be32 saddr, |
| 647 | unsigned int port) |
| 648 | { |
| 649 | return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port; |
| 650 | } |
| 651 | |
| 652 | bool ip_call_ra_chain(struct sk_buff *skb); |
| 653 | |
| 654 | /* |
| 655 | * Functions provided by ip_fragment.c |
| 656 | */ |
| 657 | |
| 658 | enum ip_defrag_users { |
| 659 | IP_DEFRAG_LOCAL_DELIVER, |
| 660 | IP_DEFRAG_CALL_RA_CHAIN, |
| 661 | IP_DEFRAG_CONNTRACK_IN, |
| 662 | __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, |
| 663 | IP_DEFRAG_CONNTRACK_OUT, |
| 664 | __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, |
| 665 | IP_DEFRAG_CONNTRACK_BRIDGE_IN, |
| 666 | __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, |
| 667 | IP_DEFRAG_VS_IN, |
| 668 | IP_DEFRAG_VS_OUT, |
| 669 | IP_DEFRAG_VS_FWD, |
| 670 | IP_DEFRAG_AF_PACKET, |
| 671 | IP_DEFRAG_MACVLAN, |
| 672 | }; |
| 673 | |
| 674 | /* Return true if the value of 'user' is between 'lower_bond' |
| 675 | * and 'upper_bond' inclusively. |
| 676 | */ |
| 677 | static inline bool ip_defrag_user_in_between(u32 user, |
| 678 | enum ip_defrag_users lower_bond, |
| 679 | enum ip_defrag_users upper_bond) |
| 680 | { |
| 681 | return user >= lower_bond && user <= upper_bond; |
| 682 | } |
| 683 | |
| 684 | int ip_defrag(struct net *net, struct sk_buff *skb, u32 user); |
| 685 | #ifdef CONFIG_INET |
| 686 | struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user); |
| 687 | #else |
| 688 | static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user) |
| 689 | { |
| 690 | return skb; |
| 691 | } |
| 692 | #endif |
| 693 | |
| 694 | /* |
| 695 | * Functions provided by ip_forward.c |
| 696 | */ |
| 697 | |
| 698 | int ip_forward(struct sk_buff *skb); |
| 699 | |
| 700 | /* |
| 701 | * Functions provided by ip_options.c |
| 702 | */ |
| 703 | |
| 704 | void ip_options_build(struct sk_buff *skb, struct ip_options *opt, |
| 705 | __be32 daddr, struct rtable *rt, int is_frag); |
| 706 | |
| 707 | int __ip_options_echo(struct net *net, struct ip_options *dopt, |
| 708 | struct sk_buff *skb, const struct ip_options *sopt); |
| 709 | static inline int ip_options_echo(struct net *net, struct ip_options *dopt, |
| 710 | struct sk_buff *skb) |
| 711 | { |
| 712 | return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt); |
| 713 | } |
| 714 | |
| 715 | void ip_options_fragment(struct sk_buff *skb); |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 716 | int __ip_options_compile(struct net *net, struct ip_options *opt, |
| 717 | struct sk_buff *skb, __be32 *info); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 718 | int ip_options_compile(struct net *net, struct ip_options *opt, |
| 719 | struct sk_buff *skb); |
| 720 | int ip_options_get(struct net *net, struct ip_options_rcu **optp, |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 721 | sockptr_t data, int optlen); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 722 | void ip_options_undo(struct ip_options *opt); |
| 723 | void ip_forward_options(struct sk_buff *skb); |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 724 | int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 725 | |
| 726 | /* |
| 727 | * Functions provided by ip_sockglue.c |
| 728 | */ |
| 729 | |
| 730 | void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb); |
| 731 | void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk, |
| 732 | struct sk_buff *skb, int tlen, int offset); |
| 733 | int ip_cmsg_send(struct sock *sk, struct msghdr *msg, |
| 734 | struct ipcm_cookie *ipc, bool allow_ipv6); |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 735 | int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 736 | unsigned int optlen); |
| 737 | int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, |
| 738 | int __user *optlen); |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 739 | int ip_ra_control(struct sock *sk, unsigned char on, |
| 740 | void (*destructor)(struct sock *)); |
| 741 | |
| 742 | int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); |
| 743 | void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, |
| 744 | u32 info, u8 *payload); |
| 745 | void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, |
| 746 | u32 info); |
| 747 | |
| 748 | static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) |
| 749 | { |
| 750 | ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0); |
| 751 | } |
| 752 | |
| 753 | bool icmp_global_allow(void); |
| 754 | extern int sysctl_icmp_msgs_per_sec; |
| 755 | extern int sysctl_icmp_msgs_burst; |
| 756 | |
| 757 | #ifdef CONFIG_PROC_FS |
| 758 | int ip_misc_proc_init(void); |
| 759 | #endif |
| 760 | |
David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 761 | int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family, |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 762 | struct netlink_ext_ack *extack); |
| 763 | |
Olivier Deprez | 0e64123 | 2021-09-23 10:07:05 +0200 | [diff] [blame] | 764 | static inline bool inetdev_valid_mtu(unsigned int mtu) |
| 765 | { |
| 766 | return likely(mtu >= IPV4_MIN_MTU); |
| 767 | } |
| 768 | |
Olivier Deprez | 157378f | 2022-04-04 15:47:50 +0200 | [diff] [blame] | 769 | void ip_sock_set_freebind(struct sock *sk); |
| 770 | int ip_sock_set_mtu_discover(struct sock *sk, int val); |
| 771 | void ip_sock_set_pktinfo(struct sock *sk); |
| 772 | void ip_sock_set_recverr(struct sock *sk); |
| 773 | void ip_sock_set_tos(struct sock *sk, int val); |
| 774 | |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 775 | #endif /* _IP_H */ |