1 /* 2 * IP Virtual Server 3 * data structure and functionality definitions 4 */ 5 6 #ifndef _NET_IP_VS_H 7 #define _NET_IP_VS_H 8 9 #include <linux/ip_vs.h> /* definitions shared with userland */ 10 11 /* old ipvsadm versions still include this file directly */ 12 #ifdef __KERNEL__ 13 14 #include <asm/types.h> /* for __uXX types */ 15 16 #include <linux/sysctl.h> /* for ctl_path */ 17 #include <linux/list.h> /* for struct list_head */ 18 #include <linux/spinlock.h> /* for struct rwlock_t */ 19 #include <asm/atomic.h> /* for struct atomic_t */ 20 #include <linux/compiler.h> 21 #include <linux/timer.h> 22 23 #include <net/checksum.h> 24 #include <linux/netfilter.h> /* for union nf_inet_addr */ 25 #include <linux/ip.h> 26 #include <linux/ipv6.h> /* for struct ipv6hdr */ 27 #include <net/ipv6.h> /* for ipv6_addr_copy */ 28 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE) 29 #include <net/netfilter/nf_conntrack.h> 30 #endif 31 #include <net/net_namespace.h> /* Netw namespace */ 32 33 /* 34 * Generic access of ipvs struct 35 */ 36 static inline struct netns_ipvs *net_ipvs(struct net* net) 37 { 38 return net->ipvs; 39 } 40 /* 41 * Get net ptr from skb in traffic cases 42 * use skb_sknet when call is from userland (ioctl or netlink) 43 */ 44 static inline struct net *skb_net(const struct sk_buff *skb) 45 { 46 #ifdef CONFIG_NET_NS 47 #ifdef CONFIG_IP_VS_DEBUG 48 /* 49 * This is used for debug only. 50 * Start with the most likely hit 51 * End with BUG 52 */ 53 if (likely(skb->dev && skb->dev->nd_net)) 54 return dev_net(skb->dev); 55 if (skb_dst(skb)->dev) 56 return dev_net(skb_dst(skb)->dev); 57 WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n", 58 __func__, __LINE__); 59 if (likely(skb->sk && skb->sk->sk_net)) 60 return sock_net(skb->sk); 61 pr_err("There is no net ptr to find in the skb in %s() line:%d\n", 62 __func__, __LINE__); 63 BUG(); 64 #else 65 return dev_net(skb->dev ? : skb_dst(skb)->dev); 66 #endif 67 #else 68 return &init_net; 69 #endif 70 } 71 72 static inline struct net *skb_sknet(const struct sk_buff *skb) 73 { 74 #ifdef CONFIG_NET_NS 75 #ifdef CONFIG_IP_VS_DEBUG 76 /* Start with the most likely hit */ 77 if (likely(skb->sk && skb->sk->sk_net)) 78 return sock_net(skb->sk); 79 WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n", 80 __func__, __LINE__); 81 if (likely(skb->dev && skb->dev->nd_net)) 82 return dev_net(skb->dev); 83 pr_err("There is no net ptr to find in the skb in %s() line:%d\n", 84 __func__, __LINE__); 85 BUG(); 86 #else 87 return sock_net(skb->sk); 88 #endif 89 #else 90 return &init_net; 91 #endif 92 } 93 /* 94 * This one needed for single_open_net since net is stored directly in 95 * private not as a struct i.e. seq_file_net cant be used. 96 */ 97 static inline struct net *seq_file_single_net(struct seq_file *seq) 98 { 99 #ifdef CONFIG_NET_NS 100 return (struct net *)seq->private; 101 #else 102 return &init_net; 103 #endif 104 } 105 106 /* Connections' size value needed by ip_vs_ctl.c */ 107 extern int ip_vs_conn_tab_size; 108 109 110 struct ip_vs_iphdr { 111 int len; 112 __u8 protocol; 113 union nf_inet_addr saddr; 114 union nf_inet_addr daddr; 115 }; 116 117 static inline void 118 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr) 119 { 120 #ifdef CONFIG_IP_VS_IPV6 121 if (af == AF_INET6) { 122 const struct ipv6hdr *iph = nh; 123 iphdr->len = sizeof(struct ipv6hdr); 124 iphdr->protocol = iph->nexthdr; 125 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr); 126 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr); 127 } else 128 #endif 129 { 130 const struct iphdr *iph = nh; 131 iphdr->len = iph->ihl * 4; 132 iphdr->protocol = iph->protocol; 133 iphdr->saddr.ip = iph->saddr; 134 iphdr->daddr.ip = iph->daddr; 135 } 136 } 137 138 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst, 139 const union nf_inet_addr *src) 140 { 141 #ifdef CONFIG_IP_VS_IPV6 142 if (af == AF_INET6) 143 ipv6_addr_copy(&dst->in6, &src->in6); 144 else 145 #endif 146 dst->ip = src->ip; 147 } 148 149 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a, 150 const union nf_inet_addr *b) 151 { 152 #ifdef CONFIG_IP_VS_IPV6 153 if (af == AF_INET6) 154 return ipv6_addr_equal(&a->in6, &b->in6); 155 #endif 156 return a->ip == b->ip; 157 } 158 159 #ifdef CONFIG_IP_VS_DEBUG 160 #include <linux/net.h> 161 162 extern int ip_vs_get_debug_level(void); 163 164 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len, 165 const union nf_inet_addr *addr, 166 int *idx) 167 { 168 int len; 169 #ifdef CONFIG_IP_VS_IPV6 170 if (af == AF_INET6) 171 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]", 172 &addr->in6) + 1; 173 else 174 #endif 175 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4", 176 &addr->ip) + 1; 177 178 *idx += len; 179 BUG_ON(*idx > buf_len + 1); 180 return &buf[*idx - len]; 181 } 182 183 #define IP_VS_DBG_BUF(level, msg, ...) \ 184 do { \ 185 char ip_vs_dbg_buf[160]; \ 186 int ip_vs_dbg_idx = 0; \ 187 if (level <= ip_vs_get_debug_level()) \ 188 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \ 189 } while (0) 190 #define IP_VS_ERR_BUF(msg...) \ 191 do { \ 192 char ip_vs_dbg_buf[160]; \ 193 int ip_vs_dbg_idx = 0; \ 194 pr_err(msg); \ 195 } while (0) 196 197 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */ 198 #define IP_VS_DBG_ADDR(af, addr) \ 199 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \ 200 sizeof(ip_vs_dbg_buf), addr, \ 201 &ip_vs_dbg_idx) 202 203 #define IP_VS_DBG(level, msg, ...) \ 204 do { \ 205 if (level <= ip_vs_get_debug_level()) \ 206 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \ 207 } while (0) 208 #define IP_VS_DBG_RL(msg, ...) \ 209 do { \ 210 if (net_ratelimit()) \ 211 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \ 212 } while (0) 213 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \ 214 do { \ 215 if (level <= ip_vs_get_debug_level()) \ 216 pp->debug_packet(af, pp, skb, ofs, msg); \ 217 } while (0) 218 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \ 219 do { \ 220 if (level <= ip_vs_get_debug_level() && \ 221 net_ratelimit()) \ 222 pp->debug_packet(af, pp, skb, ofs, msg); \ 223 } while (0) 224 #else /* NO DEBUGGING at ALL */ 225 #define IP_VS_DBG_BUF(level, msg...) do {} while (0) 226 #define IP_VS_ERR_BUF(msg...) do {} while (0) 227 #define IP_VS_DBG(level, msg...) do {} while (0) 228 #define IP_VS_DBG_RL(msg...) do {} while (0) 229 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0) 230 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0) 231 #endif 232 233 #define IP_VS_BUG() BUG() 234 #define IP_VS_ERR_RL(msg, ...) \ 235 do { \ 236 if (net_ratelimit()) \ 237 pr_err(msg, ##__VA_ARGS__); \ 238 } while (0) 239 240 #ifdef CONFIG_IP_VS_DEBUG 241 #define EnterFunction(level) \ 242 do { \ 243 if (level <= ip_vs_get_debug_level()) \ 244 printk(KERN_DEBUG \ 245 pr_fmt("Enter: %s, %s line %i\n"), \ 246 __func__, __FILE__, __LINE__); \ 247 } while (0) 248 #define LeaveFunction(level) \ 249 do { \ 250 if (level <= ip_vs_get_debug_level()) \ 251 printk(KERN_DEBUG \ 252 pr_fmt("Leave: %s, %s line %i\n"), \ 253 __func__, __FILE__, __LINE__); \ 254 } while (0) 255 #else 256 #define EnterFunction(level) do {} while (0) 257 #define LeaveFunction(level) do {} while (0) 258 #endif 259 260 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); } 261 262 263 /* 264 * The port number of FTP service (in network order). 265 */ 266 #define FTPPORT cpu_to_be16(21) 267 #define FTPDATA cpu_to_be16(20) 268 269 /* 270 * TCP State Values 271 */ 272 enum { 273 IP_VS_TCP_S_NONE = 0, 274 IP_VS_TCP_S_ESTABLISHED, 275 IP_VS_TCP_S_SYN_SENT, 276 IP_VS_TCP_S_SYN_RECV, 277 IP_VS_TCP_S_FIN_WAIT, 278 IP_VS_TCP_S_TIME_WAIT, 279 IP_VS_TCP_S_CLOSE, 280 IP_VS_TCP_S_CLOSE_WAIT, 281 IP_VS_TCP_S_LAST_ACK, 282 IP_VS_TCP_S_LISTEN, 283 IP_VS_TCP_S_SYNACK, 284 IP_VS_TCP_S_LAST 285 }; 286 287 /* 288 * UDP State Values 289 */ 290 enum { 291 IP_VS_UDP_S_NORMAL, 292 IP_VS_UDP_S_LAST, 293 }; 294 295 /* 296 * ICMP State Values 297 */ 298 enum { 299 IP_VS_ICMP_S_NORMAL, 300 IP_VS_ICMP_S_LAST, 301 }; 302 303 /* 304 * SCTP State Values 305 */ 306 enum ip_vs_sctp_states { 307 IP_VS_SCTP_S_NONE, 308 IP_VS_SCTP_S_INIT_CLI, 309 IP_VS_SCTP_S_INIT_SER, 310 IP_VS_SCTP_S_INIT_ACK_CLI, 311 IP_VS_SCTP_S_INIT_ACK_SER, 312 IP_VS_SCTP_S_ECHO_CLI, 313 IP_VS_SCTP_S_ECHO_SER, 314 IP_VS_SCTP_S_ESTABLISHED, 315 IP_VS_SCTP_S_SHUT_CLI, 316 IP_VS_SCTP_S_SHUT_SER, 317 IP_VS_SCTP_S_SHUT_ACK_CLI, 318 IP_VS_SCTP_S_SHUT_ACK_SER, 319 IP_VS_SCTP_S_CLOSED, 320 IP_VS_SCTP_S_LAST 321 }; 322 323 /* 324 * Delta sequence info structure 325 * Each ip_vs_conn has 2 (output AND input seq. changes). 326 * Only used in the VS/NAT. 327 */ 328 struct ip_vs_seq { 329 __u32 init_seq; /* Add delta from this seq */ 330 __u32 delta; /* Delta in sequence numbers */ 331 __u32 previous_delta; /* Delta in sequence numbers 332 before last resized pkt */ 333 }; 334 335 /* 336 * counters per cpu 337 */ 338 struct ip_vs_counters { 339 __u32 conns; /* connections scheduled */ 340 __u32 inpkts; /* incoming packets */ 341 __u32 outpkts; /* outgoing packets */ 342 __u64 inbytes; /* incoming bytes */ 343 __u64 outbytes; /* outgoing bytes */ 344 }; 345 /* 346 * Stats per cpu 347 */ 348 struct ip_vs_cpu_stats { 349 struct ip_vs_counters ustats; 350 struct u64_stats_sync syncp; 351 }; 352 353 /* 354 * IPVS statistics objects 355 */ 356 struct ip_vs_estimator { 357 struct list_head list; 358 359 u64 last_inbytes; 360 u64 last_outbytes; 361 u32 last_conns; 362 u32 last_inpkts; 363 u32 last_outpkts; 364 365 u32 cps; 366 u32 inpps; 367 u32 outpps; 368 u32 inbps; 369 u32 outbps; 370 }; 371 372 struct ip_vs_stats { 373 struct ip_vs_stats_user ustats; /* statistics */ 374 struct ip_vs_estimator est; /* estimator */ 375 struct ip_vs_cpu_stats *cpustats; /* per cpu counters */ 376 spinlock_t lock; /* spin lock */ 377 struct ip_vs_stats_user ustats0; /* reset values */ 378 }; 379 380 struct dst_entry; 381 struct iphdr; 382 struct ip_vs_conn; 383 struct ip_vs_app; 384 struct sk_buff; 385 struct ip_vs_proto_data; 386 387 struct ip_vs_protocol { 388 struct ip_vs_protocol *next; 389 char *name; 390 u16 protocol; 391 u16 num_states; 392 int dont_defrag; 393 394 void (*init)(struct ip_vs_protocol *pp); 395 396 void (*exit)(struct ip_vs_protocol *pp); 397 398 void (*init_netns)(struct net *net, struct ip_vs_proto_data *pd); 399 400 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd); 401 402 int (*conn_schedule)(int af, struct sk_buff *skb, 403 struct ip_vs_proto_data *pd, 404 int *verdict, struct ip_vs_conn **cpp); 405 406 struct ip_vs_conn * 407 (*conn_in_get)(int af, 408 const struct sk_buff *skb, 409 const struct ip_vs_iphdr *iph, 410 unsigned int proto_off, 411 int inverse); 412 413 struct ip_vs_conn * 414 (*conn_out_get)(int af, 415 const struct sk_buff *skb, 416 const struct ip_vs_iphdr *iph, 417 unsigned int proto_off, 418 int inverse); 419 420 int (*snat_handler)(struct sk_buff *skb, 421 struct ip_vs_protocol *pp, struct ip_vs_conn *cp); 422 423 int (*dnat_handler)(struct sk_buff *skb, 424 struct ip_vs_protocol *pp, struct ip_vs_conn *cp); 425 426 int (*csum_check)(int af, struct sk_buff *skb, 427 struct ip_vs_protocol *pp); 428 429 const char *(*state_name)(int state); 430 431 int (*state_transition)(struct ip_vs_conn *cp, int direction, 432 const struct sk_buff *skb, 433 struct ip_vs_proto_data *pd); 434 435 int (*register_app)(struct net *net, struct ip_vs_app *inc); 436 437 void (*unregister_app)(struct net *net, struct ip_vs_app *inc); 438 439 int (*app_conn_bind)(struct ip_vs_conn *cp); 440 441 void (*debug_packet)(int af, struct ip_vs_protocol *pp, 442 const struct sk_buff *skb, 443 int offset, 444 const char *msg); 445 446 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags); 447 }; 448 449 /* 450 * protocol data per netns 451 */ 452 struct ip_vs_proto_data { 453 struct ip_vs_proto_data *next; 454 struct ip_vs_protocol *pp; 455 int *timeout_table; /* protocol timeout table */ 456 atomic_t appcnt; /* counter of proto app incs. */ 457 struct tcp_states_t *tcp_state_table; 458 }; 459 460 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto); 461 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net, 462 unsigned short proto); 463 464 struct ip_vs_conn_param { 465 struct net *net; 466 const union nf_inet_addr *caddr; 467 const union nf_inet_addr *vaddr; 468 __be16 cport; 469 __be16 vport; 470 __u16 protocol; 471 u16 af; 472 473 const struct ip_vs_pe *pe; 474 char *pe_data; 475 __u8 pe_data_len; 476 }; 477 478 /* 479 * IP_VS structure allocated for each dynamically scheduled connection 480 */ 481 struct ip_vs_conn { 482 struct hlist_node c_list; /* hashed list heads */ 483 #ifdef CONFIG_NET_NS 484 struct net *net; /* Name space */ 485 #endif 486 /* Protocol, addresses and port numbers */ 487 u16 af; /* address family */ 488 __be16 cport; 489 __be16 vport; 490 __be16 dport; 491 __u32 fwmark; /* Fire wall mark from skb */ 492 union nf_inet_addr caddr; /* client address */ 493 union nf_inet_addr vaddr; /* virtual address */ 494 union nf_inet_addr daddr; /* destination address */ 495 volatile __u32 flags; /* status flags */ 496 __u16 protocol; /* Which protocol (TCP/UDP) */ 497 498 /* counter and timer */ 499 atomic_t refcnt; /* reference count */ 500 struct timer_list timer; /* Expiration timer */ 501 volatile unsigned long timeout; /* timeout */ 502 503 /* Flags and state transition */ 504 spinlock_t lock; /* lock for state transition */ 505 volatile __u16 state; /* state info */ 506 volatile __u16 old_state; /* old state, to be used for 507 * state transition triggerd 508 * synchronization 509 */ 510 511 /* Control members */ 512 struct ip_vs_conn *control; /* Master control connection */ 513 atomic_t n_control; /* Number of controlled ones */ 514 struct ip_vs_dest *dest; /* real server */ 515 atomic_t in_pkts; /* incoming packet counter */ 516 517 /* packet transmitter for different forwarding methods. If it 518 mangles the packet, it must return NF_DROP or better NF_STOLEN, 519 otherwise this must be changed to a sk_buff **. 520 NF_ACCEPT can be returned when destination is local. 521 */ 522 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp, 523 struct ip_vs_protocol *pp); 524 525 /* Note: we can group the following members into a structure, 526 in order to save more space, and the following members are 527 only used in VS/NAT anyway */ 528 struct ip_vs_app *app; /* bound ip_vs_app object */ 529 void *app_data; /* Application private data */ 530 struct ip_vs_seq in_seq; /* incoming seq. struct */ 531 struct ip_vs_seq out_seq; /* outgoing seq. struct */ 532 533 const struct ip_vs_pe *pe; 534 char *pe_data; 535 __u8 pe_data_len; 536 }; 537 538 /* 539 * To save some memory in conn table when name space is disabled. 540 */ 541 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp) 542 { 543 #ifdef CONFIG_NET_NS 544 return cp->net; 545 #else 546 return &init_net; 547 #endif 548 } 549 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net) 550 { 551 #ifdef CONFIG_NET_NS 552 cp->net = net; 553 #endif 554 } 555 556 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp, 557 struct net *net) 558 { 559 #ifdef CONFIG_NET_NS 560 return cp->net == net; 561 #else 562 return 1; 563 #endif 564 } 565 566 /* 567 * Extended internal versions of struct ip_vs_service_user and 568 * ip_vs_dest_user for IPv6 support. 569 * 570 * We need these to conveniently pass around service and destination 571 * options, but unfortunately, we also need to keep the old definitions to 572 * maintain userspace backwards compatibility for the setsockopt interface. 573 */ 574 struct ip_vs_service_user_kern { 575 /* virtual service addresses */ 576 u16 af; 577 u16 protocol; 578 union nf_inet_addr addr; /* virtual ip address */ 579 u16 port; 580 u32 fwmark; /* firwall mark of service */ 581 582 /* virtual service options */ 583 char *sched_name; 584 char *pe_name; 585 unsigned flags; /* virtual service flags */ 586 unsigned timeout; /* persistent timeout in sec */ 587 u32 netmask; /* persistent netmask */ 588 }; 589 590 591 struct ip_vs_dest_user_kern { 592 /* destination server address */ 593 union nf_inet_addr addr; 594 u16 port; 595 596 /* real server options */ 597 unsigned conn_flags; /* connection flags */ 598 int weight; /* destination weight */ 599 600 /* thresholds for active connections */ 601 u32 u_threshold; /* upper threshold */ 602 u32 l_threshold; /* lower threshold */ 603 }; 604 605 606 /* 607 * The information about the virtual service offered to the net 608 * and the forwarding entries 609 */ 610 struct ip_vs_service { 611 struct list_head s_list; /* for normal service table */ 612 struct list_head f_list; /* for fwmark-based service table */ 613 atomic_t refcnt; /* reference counter */ 614 atomic_t usecnt; /* use counter */ 615 616 u16 af; /* address family */ 617 __u16 protocol; /* which protocol (TCP/UDP) */ 618 union nf_inet_addr addr; /* IP address for virtual service */ 619 __be16 port; /* port number for the service */ 620 __u32 fwmark; /* firewall mark of the service */ 621 unsigned flags; /* service status flags */ 622 unsigned timeout; /* persistent timeout in ticks */ 623 __be32 netmask; /* grouping granularity */ 624 struct net *net; 625 626 struct list_head destinations; /* real server d-linked list */ 627 __u32 num_dests; /* number of servers */ 628 struct ip_vs_stats stats; /* statistics for the service */ 629 struct ip_vs_app *inc; /* bind conns to this app inc */ 630 631 /* for scheduling */ 632 struct ip_vs_scheduler *scheduler; /* bound scheduler object */ 633 rwlock_t sched_lock; /* lock sched_data */ 634 void *sched_data; /* scheduler application data */ 635 636 /* alternate persistence engine */ 637 struct ip_vs_pe *pe; 638 }; 639 640 641 /* 642 * The real server destination forwarding entry 643 * with ip address, port number, and so on. 644 */ 645 struct ip_vs_dest { 646 struct list_head n_list; /* for the dests in the service */ 647 struct list_head d_list; /* for table with all the dests */ 648 649 u16 af; /* address family */ 650 __be16 port; /* port number of the server */ 651 union nf_inet_addr addr; /* IP address of the server */ 652 volatile unsigned flags; /* dest status flags */ 653 atomic_t conn_flags; /* flags to copy to conn */ 654 atomic_t weight; /* server weight */ 655 656 atomic_t refcnt; /* reference counter */ 657 struct ip_vs_stats stats; /* statistics */ 658 659 /* connection counters and thresholds */ 660 atomic_t activeconns; /* active connections */ 661 atomic_t inactconns; /* inactive connections */ 662 atomic_t persistconns; /* persistent connections */ 663 __u32 u_threshold; /* upper threshold */ 664 __u32 l_threshold; /* lower threshold */ 665 666 /* for destination cache */ 667 spinlock_t dst_lock; /* lock of dst_cache */ 668 struct dst_entry *dst_cache; /* destination cache entry */ 669 u32 dst_rtos; /* RT_TOS(tos) for dst */ 670 u32 dst_cookie; 671 #ifdef CONFIG_IP_VS_IPV6 672 struct in6_addr dst_saddr; 673 #endif 674 675 /* for virtual service */ 676 struct ip_vs_service *svc; /* service it belongs to */ 677 __u16 protocol; /* which protocol (TCP/UDP) */ 678 __be16 vport; /* virtual port number */ 679 union nf_inet_addr vaddr; /* virtual IP address */ 680 __u32 vfwmark; /* firewall mark of service */ 681 }; 682 683 684 /* 685 * The scheduler object 686 */ 687 struct ip_vs_scheduler { 688 struct list_head n_list; /* d-linked list head */ 689 char *name; /* scheduler name */ 690 atomic_t refcnt; /* reference counter */ 691 struct module *module; /* THIS_MODULE/NULL */ 692 693 /* scheduler initializing service */ 694 int (*init_service)(struct ip_vs_service *svc); 695 /* scheduling service finish */ 696 int (*done_service)(struct ip_vs_service *svc); 697 /* scheduler updating service */ 698 int (*update_service)(struct ip_vs_service *svc); 699 700 /* selecting a server from the given service */ 701 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc, 702 const struct sk_buff *skb); 703 }; 704 705 /* The persistence engine object */ 706 struct ip_vs_pe { 707 struct list_head n_list; /* d-linked list head */ 708 char *name; /* scheduler name */ 709 atomic_t refcnt; /* reference counter */ 710 struct module *module; /* THIS_MODULE/NULL */ 711 712 /* get the connection template, if any */ 713 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb); 714 bool (*ct_match)(const struct ip_vs_conn_param *p, 715 struct ip_vs_conn *ct); 716 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval, 717 bool inverse); 718 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf); 719 }; 720 721 /* 722 * The application module object (a.k.a. app incarnation) 723 */ 724 struct ip_vs_app { 725 struct list_head a_list; /* member in app list */ 726 int type; /* IP_VS_APP_TYPE_xxx */ 727 char *name; /* application module name */ 728 __u16 protocol; 729 struct module *module; /* THIS_MODULE/NULL */ 730 struct list_head incs_list; /* list of incarnations */ 731 732 /* members for application incarnations */ 733 struct list_head p_list; /* member in proto app list */ 734 struct ip_vs_app *app; /* its real application */ 735 __be16 port; /* port number in net order */ 736 atomic_t usecnt; /* usage counter */ 737 738 /* 739 * output hook: Process packet in inout direction, diff set for TCP. 740 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok, 741 * 2=Mangled but checksum was not updated 742 */ 743 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *, 744 struct sk_buff *, int *diff); 745 746 /* 747 * input hook: Process packet in outin direction, diff set for TCP. 748 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok, 749 * 2=Mangled but checksum was not updated 750 */ 751 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *, 752 struct sk_buff *, int *diff); 753 754 /* ip_vs_app initializer */ 755 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *); 756 757 /* ip_vs_app finish */ 758 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *); 759 760 761 /* not used now */ 762 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *, 763 struct ip_vs_protocol *); 764 765 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *); 766 767 int * timeout_table; 768 int * timeouts; 769 int timeouts_size; 770 771 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app, 772 int *verdict, struct ip_vs_conn **cpp); 773 774 struct ip_vs_conn * 775 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app, 776 const struct iphdr *iph, unsigned int proto_off, 777 int inverse); 778 779 struct ip_vs_conn * 780 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app, 781 const struct iphdr *iph, unsigned int proto_off, 782 int inverse); 783 784 int (*state_transition)(struct ip_vs_conn *cp, int direction, 785 const struct sk_buff *skb, 786 struct ip_vs_app *app); 787 788 void (*timeout_change)(struct ip_vs_app *app, int flags); 789 }; 790 791 /* IPVS in network namespace */ 792 struct netns_ipvs { 793 int gen; /* Generation */ 794 /* 795 * Hash table: for real service lookups 796 */ 797 #define IP_VS_RTAB_BITS 4 798 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS) 799 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1) 800 801 struct list_head rs_table[IP_VS_RTAB_SIZE]; 802 /* ip_vs_app */ 803 struct list_head app_list; 804 struct mutex app_mutex; 805 struct lock_class_key app_key; /* mutex debuging */ 806 807 /* ip_vs_proto */ 808 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */ 809 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE]; 810 /* ip_vs_proto_tcp */ 811 #ifdef CONFIG_IP_VS_PROTO_TCP 812 #define TCP_APP_TAB_BITS 4 813 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS) 814 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1) 815 struct list_head tcp_apps[TCP_APP_TAB_SIZE]; 816 spinlock_t tcp_app_lock; 817 #endif 818 /* ip_vs_proto_udp */ 819 #ifdef CONFIG_IP_VS_PROTO_UDP 820 #define UDP_APP_TAB_BITS 4 821 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS) 822 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1) 823 struct list_head udp_apps[UDP_APP_TAB_SIZE]; 824 spinlock_t udp_app_lock; 825 #endif 826 /* ip_vs_proto_sctp */ 827 #ifdef CONFIG_IP_VS_PROTO_SCTP 828 #define SCTP_APP_TAB_BITS 4 829 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS) 830 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1) 831 /* Hash table for SCTP application incarnations */ 832 struct list_head sctp_apps[SCTP_APP_TAB_SIZE]; 833 spinlock_t sctp_app_lock; 834 #endif 835 /* ip_vs_conn */ 836 atomic_t conn_count; /* connection counter */ 837 838 /* ip_vs_ctl */ 839 struct ip_vs_stats tot_stats; /* Statistics & est. */ 840 841 int num_services; /* no of virtual services */ 842 843 rwlock_t rs_lock; /* real services table */ 844 /* semaphore for IPVS sockopts. And, [gs]etsockopt may sleep. */ 845 struct lock_class_key ctl_key; /* ctl_mutex debuging */ 846 /* Trash for destinations */ 847 struct list_head dest_trash; 848 /* Service counters */ 849 atomic_t ftpsvc_counter; 850 atomic_t nullsvc_counter; 851 852 #ifdef CONFIG_SYSCTL 853 /* 1/rate drop and drop-entry variables */ 854 struct delayed_work defense_work; /* Work handler */ 855 int drop_rate; 856 int drop_counter; 857 atomic_t dropentry; 858 /* locks in ctl.c */ 859 spinlock_t dropentry_lock; /* drop entry handling */ 860 spinlock_t droppacket_lock; /* drop packet handling */ 861 spinlock_t securetcp_lock; /* state and timeout tables */ 862 863 /* sys-ctl struct */ 864 struct ctl_table_header *sysctl_hdr; 865 struct ctl_table *sysctl_tbl; 866 #endif 867 868 /* sysctl variables */ 869 int sysctl_amemthresh; 870 int sysctl_am_droprate; 871 int sysctl_drop_entry; 872 int sysctl_drop_packet; 873 int sysctl_secure_tcp; 874 #ifdef CONFIG_IP_VS_NFCT 875 int sysctl_conntrack; 876 #endif 877 int sysctl_snat_reroute; 878 int sysctl_sync_ver; 879 int sysctl_cache_bypass; 880 int sysctl_expire_nodest_conn; 881 int sysctl_expire_quiescent_template; 882 int sysctl_sync_threshold[2]; 883 int sysctl_nat_icmp_send; 884 885 /* ip_vs_lblc */ 886 int sysctl_lblc_expiration; 887 struct ctl_table_header *lblc_ctl_header; 888 struct ctl_table *lblc_ctl_table; 889 /* ip_vs_lblcr */ 890 int sysctl_lblcr_expiration; 891 struct ctl_table_header *lblcr_ctl_header; 892 struct ctl_table *lblcr_ctl_table; 893 /* ip_vs_est */ 894 struct list_head est_list; /* estimator list */ 895 spinlock_t est_lock; 896 struct timer_list est_timer; /* Estimation timer */ 897 /* ip_vs_sync */ 898 struct list_head sync_queue; 899 spinlock_t sync_lock; 900 struct ip_vs_sync_buff *sync_buff; 901 spinlock_t sync_buff_lock; 902 struct sockaddr_in sync_mcast_addr; 903 struct task_struct *master_thread; 904 struct task_struct *backup_thread; 905 int send_mesg_maxlen; 906 int recv_mesg_maxlen; 907 volatile int sync_state; 908 volatile int master_syncid; 909 volatile int backup_syncid; 910 /* multicast interface name */ 911 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN]; 912 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN]; 913 /* net name space ptr */ 914 struct net *net; /* Needed by timer routines */ 915 }; 916 917 #define DEFAULT_SYNC_THRESHOLD 3 918 #define DEFAULT_SYNC_PERIOD 50 919 #define DEFAULT_SYNC_VER 1 920 921 #ifdef CONFIG_SYSCTL 922 923 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs) 924 { 925 return ipvs->sysctl_sync_threshold[0]; 926 } 927 928 static inline int sysctl_sync_period(struct netns_ipvs *ipvs) 929 { 930 return ipvs->sysctl_sync_threshold[1]; 931 } 932 933 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs) 934 { 935 return ipvs->sysctl_sync_ver; 936 } 937 938 #else 939 940 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs) 941 { 942 return DEFAULT_SYNC_THRESHOLD; 943 } 944 945 static inline int sysctl_sync_period(struct netns_ipvs *ipvs) 946 { 947 return DEFAULT_SYNC_PERIOD; 948 } 949 950 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs) 951 { 952 return DEFAULT_SYNC_VER; 953 } 954 955 #endif 956 957 /* 958 * IPVS core functions 959 * (from ip_vs_core.c) 960 */ 961 extern const char *ip_vs_proto_name(unsigned proto); 962 extern void ip_vs_init_hash_table(struct list_head *table, int rows); 963 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t))) 964 965 #define IP_VS_APP_TYPE_FTP 1 966 967 /* 968 * ip_vs_conn handling functions 969 * (from ip_vs_conn.c) 970 */ 971 972 enum { 973 IP_VS_DIR_INPUT = 0, 974 IP_VS_DIR_OUTPUT, 975 IP_VS_DIR_INPUT_ONLY, 976 IP_VS_DIR_LAST, 977 }; 978 979 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol, 980 const union nf_inet_addr *caddr, 981 __be16 cport, 982 const union nf_inet_addr *vaddr, 983 __be16 vport, 984 struct ip_vs_conn_param *p) 985 { 986 p->net = net; 987 p->af = af; 988 p->protocol = protocol; 989 p->caddr = caddr; 990 p->cport = cport; 991 p->vaddr = vaddr; 992 p->vport = vport; 993 p->pe = NULL; 994 p->pe_data = NULL; 995 } 996 997 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p); 998 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p); 999 1000 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb, 1001 const struct ip_vs_iphdr *iph, 1002 unsigned int proto_off, 1003 int inverse); 1004 1005 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p); 1006 1007 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb, 1008 const struct ip_vs_iphdr *iph, 1009 unsigned int proto_off, 1010 int inverse); 1011 1012 /* put back the conn without restarting its timer */ 1013 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp) 1014 { 1015 atomic_dec(&cp->refcnt); 1016 } 1017 extern void ip_vs_conn_put(struct ip_vs_conn *cp); 1018 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport); 1019 1020 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, 1021 const union nf_inet_addr *daddr, 1022 __be16 dport, unsigned flags, 1023 struct ip_vs_dest *dest, __u32 fwmark); 1024 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp); 1025 1026 extern const char * ip_vs_state_name(__u16 proto, int state); 1027 1028 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp); 1029 extern int ip_vs_check_template(struct ip_vs_conn *ct); 1030 extern void ip_vs_random_dropentry(struct net *net); 1031 extern int ip_vs_conn_init(void); 1032 extern void ip_vs_conn_cleanup(void); 1033 1034 static inline void ip_vs_control_del(struct ip_vs_conn *cp) 1035 { 1036 struct ip_vs_conn *ctl_cp = cp->control; 1037 if (!ctl_cp) { 1038 IP_VS_ERR_BUF("request control DEL for uncontrolled: " 1039 "%s:%d to %s:%d\n", 1040 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 1041 ntohs(cp->cport), 1042 IP_VS_DBG_ADDR(cp->af, &cp->vaddr), 1043 ntohs(cp->vport)); 1044 1045 return; 1046 } 1047 1048 IP_VS_DBG_BUF(7, "DELeting control for: " 1049 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n", 1050 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 1051 ntohs(cp->cport), 1052 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr), 1053 ntohs(ctl_cp->cport)); 1054 1055 cp->control = NULL; 1056 if (atomic_read(&ctl_cp->n_control) == 0) { 1057 IP_VS_ERR_BUF("BUG control DEL with n=0 : " 1058 "%s:%d to %s:%d\n", 1059 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 1060 ntohs(cp->cport), 1061 IP_VS_DBG_ADDR(cp->af, &cp->vaddr), 1062 ntohs(cp->vport)); 1063 1064 return; 1065 } 1066 atomic_dec(&ctl_cp->n_control); 1067 } 1068 1069 static inline void 1070 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp) 1071 { 1072 if (cp->control) { 1073 IP_VS_ERR_BUF("request control ADD for already controlled: " 1074 "%s:%d to %s:%d\n", 1075 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 1076 ntohs(cp->cport), 1077 IP_VS_DBG_ADDR(cp->af, &cp->vaddr), 1078 ntohs(cp->vport)); 1079 1080 ip_vs_control_del(cp); 1081 } 1082 1083 IP_VS_DBG_BUF(7, "ADDing control for: " 1084 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n", 1085 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 1086 ntohs(cp->cport), 1087 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr), 1088 ntohs(ctl_cp->cport)); 1089 1090 cp->control = ctl_cp; 1091 atomic_inc(&ctl_cp->n_control); 1092 } 1093 1094 1095 /* 1096 * IPVS application functions 1097 * (from ip_vs_app.c) 1098 */ 1099 #define IP_VS_APP_MAX_PORTS 8 1100 extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app); 1101 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app); 1102 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1103 extern void ip_vs_unbind_app(struct ip_vs_conn *cp); 1104 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app, 1105 __u16 proto, __u16 port); 1106 extern int ip_vs_app_inc_get(struct ip_vs_app *inc); 1107 extern void ip_vs_app_inc_put(struct ip_vs_app *inc); 1108 1109 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb); 1110 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb); 1111 extern int ip_vs_app_init(void); 1112 extern void ip_vs_app_cleanup(void); 1113 1114 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe); 1115 void ip_vs_unbind_pe(struct ip_vs_service *svc); 1116 int register_ip_vs_pe(struct ip_vs_pe *pe); 1117 int unregister_ip_vs_pe(struct ip_vs_pe *pe); 1118 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name); 1119 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name); 1120 1121 static inline void ip_vs_pe_get(const struct ip_vs_pe *pe) 1122 { 1123 if (pe && pe->module) 1124 __module_get(pe->module); 1125 } 1126 1127 static inline void ip_vs_pe_put(const struct ip_vs_pe *pe) 1128 { 1129 if (pe && pe->module) 1130 module_put(pe->module); 1131 } 1132 1133 /* 1134 * IPVS protocol functions (from ip_vs_proto.c) 1135 */ 1136 extern int ip_vs_protocol_init(void); 1137 extern void ip_vs_protocol_cleanup(void); 1138 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags); 1139 extern int *ip_vs_create_timeout_table(int *table, int size); 1140 extern int 1141 ip_vs_set_state_timeout(int *table, int num, const char *const *names, 1142 const char *name, int to); 1143 extern void 1144 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp, 1145 const struct sk_buff *skb, 1146 int offset, const char *msg); 1147 1148 extern struct ip_vs_protocol ip_vs_protocol_tcp; 1149 extern struct ip_vs_protocol ip_vs_protocol_udp; 1150 extern struct ip_vs_protocol ip_vs_protocol_icmp; 1151 extern struct ip_vs_protocol ip_vs_protocol_esp; 1152 extern struct ip_vs_protocol ip_vs_protocol_ah; 1153 extern struct ip_vs_protocol ip_vs_protocol_sctp; 1154 1155 /* 1156 * Registering/unregistering scheduler functions 1157 * (from ip_vs_sched.c) 1158 */ 1159 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler); 1160 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler); 1161 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc, 1162 struct ip_vs_scheduler *scheduler); 1163 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc); 1164 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name); 1165 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler); 1166 extern struct ip_vs_conn * 1167 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb, 1168 struct ip_vs_proto_data *pd, int *ignored); 1169 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb, 1170 struct ip_vs_proto_data *pd); 1171 1172 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg); 1173 1174 1175 /* 1176 * IPVS control data and functions (from ip_vs_ctl.c) 1177 */ 1178 extern struct ip_vs_stats ip_vs_stats; 1179 extern const struct ctl_path net_vs_ctl_path[]; 1180 extern int sysctl_ip_vs_sync_ver; 1181 1182 extern void ip_vs_sync_switch_mode(struct net *net, int mode); 1183 extern struct ip_vs_service * 1184 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol, 1185 const union nf_inet_addr *vaddr, __be16 vport); 1186 1187 static inline void ip_vs_service_put(struct ip_vs_service *svc) 1188 { 1189 atomic_dec(&svc->usecnt); 1190 } 1191 1192 extern struct ip_vs_dest * 1193 ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol, 1194 const union nf_inet_addr *daddr, __be16 dport); 1195 1196 extern int ip_vs_use_count_inc(void); 1197 extern void ip_vs_use_count_dec(void); 1198 extern int ip_vs_control_init(void); 1199 extern void ip_vs_control_cleanup(void); 1200 extern struct ip_vs_dest * 1201 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr, 1202 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport, 1203 __u16 protocol, __u32 fwmark); 1204 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp); 1205 1206 1207 /* 1208 * IPVS sync daemon data and function prototypes 1209 * (from ip_vs_sync.c) 1210 */ 1211 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn, 1212 __u8 syncid); 1213 extern int stop_sync_thread(struct net *net, int state); 1214 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp); 1215 extern int ip_vs_sync_init(void); 1216 extern void ip_vs_sync_cleanup(void); 1217 1218 1219 /* 1220 * IPVS rate estimator prototypes (from ip_vs_est.c) 1221 */ 1222 extern int ip_vs_estimator_init(void); 1223 extern void ip_vs_estimator_cleanup(void); 1224 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats); 1225 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats); 1226 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats); 1227 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst, 1228 struct ip_vs_stats *stats); 1229 1230 /* 1231 * Various IPVS packet transmitters (from ip_vs_xmit.c) 1232 */ 1233 extern int ip_vs_null_xmit 1234 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1235 extern int ip_vs_bypass_xmit 1236 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1237 extern int ip_vs_nat_xmit 1238 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1239 extern int ip_vs_tunnel_xmit 1240 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1241 extern int ip_vs_dr_xmit 1242 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1243 extern int ip_vs_icmp_xmit 1244 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset); 1245 extern void ip_vs_dst_reset(struct ip_vs_dest *dest); 1246 1247 #ifdef CONFIG_IP_VS_IPV6 1248 extern int ip_vs_bypass_xmit_v6 1249 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1250 extern int ip_vs_nat_xmit_v6 1251 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1252 extern int ip_vs_tunnel_xmit_v6 1253 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1254 extern int ip_vs_dr_xmit_v6 1255 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 1256 extern int ip_vs_icmp_xmit_v6 1257 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, 1258 int offset); 1259 #endif 1260 1261 #ifdef CONFIG_SYSCTL 1262 /* 1263 * This is a simple mechanism to ignore packets when 1264 * we are loaded. Just set ip_vs_drop_rate to 'n' and 1265 * we start to drop 1/rate of the packets 1266 */ 1267 1268 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) 1269 { 1270 if (!ipvs->drop_rate) 1271 return 0; 1272 if (--ipvs->drop_counter > 0) 1273 return 0; 1274 ipvs->drop_counter = ipvs->drop_rate; 1275 return 1; 1276 } 1277 #else 1278 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; } 1279 #endif 1280 1281 /* 1282 * ip_vs_fwd_tag returns the forwarding tag of the connection 1283 */ 1284 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK) 1285 1286 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp) 1287 { 1288 char fwd; 1289 1290 switch (IP_VS_FWD_METHOD(cp)) { 1291 case IP_VS_CONN_F_MASQ: 1292 fwd = 'M'; break; 1293 case IP_VS_CONN_F_LOCALNODE: 1294 fwd = 'L'; break; 1295 case IP_VS_CONN_F_TUNNEL: 1296 fwd = 'T'; break; 1297 case IP_VS_CONN_F_DROUTE: 1298 fwd = 'R'; break; 1299 case IP_VS_CONN_F_BYPASS: 1300 fwd = 'B'; break; 1301 default: 1302 fwd = '?'; break; 1303 } 1304 return fwd; 1305 } 1306 1307 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp, 1308 struct ip_vs_conn *cp, int dir); 1309 1310 #ifdef CONFIG_IP_VS_IPV6 1311 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp, 1312 struct ip_vs_conn *cp, int dir); 1313 #endif 1314 1315 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset); 1316 1317 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum) 1318 { 1319 __be32 diff[2] = { ~old, new }; 1320 1321 return csum_partial(diff, sizeof(diff), oldsum); 1322 } 1323 1324 #ifdef CONFIG_IP_VS_IPV6 1325 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new, 1326 __wsum oldsum) 1327 { 1328 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0], 1329 new[3], new[2], new[1], new[0] }; 1330 1331 return csum_partial(diff, sizeof(diff), oldsum); 1332 } 1333 #endif 1334 1335 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum) 1336 { 1337 __be16 diff[2] = { ~old, new }; 1338 1339 return csum_partial(diff, sizeof(diff), oldsum); 1340 } 1341 1342 /* 1343 * Forget current conntrack (unconfirmed) and attach notrack entry 1344 */ 1345 static inline void ip_vs_notrack(struct sk_buff *skb) 1346 { 1347 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE) 1348 enum ip_conntrack_info ctinfo; 1349 struct nf_conn *ct = nf_ct_get(skb, &ctinfo); 1350 1351 if (!ct || !nf_ct_is_untracked(ct)) { 1352 nf_reset(skb); 1353 skb->nfct = &nf_ct_untracked_get()->ct_general; 1354 skb->nfctinfo = IP_CT_NEW; 1355 nf_conntrack_get(skb->nfct); 1356 } 1357 #endif 1358 } 1359 1360 #ifdef CONFIG_IP_VS_NFCT 1361 /* 1362 * Netfilter connection tracking 1363 * (from ip_vs_nfct.c) 1364 */ 1365 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs) 1366 { 1367 #ifdef CONFIG_SYSCTL 1368 return ipvs->sysctl_conntrack; 1369 #else 1370 return 0; 1371 #endif 1372 } 1373 1374 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp, 1375 int outin); 1376 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp); 1377 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct, 1378 struct ip_vs_conn *cp, u_int8_t proto, 1379 const __be16 port, int from_rs); 1380 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp); 1381 1382 #else 1383 1384 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs) 1385 { 1386 return 0; 1387 } 1388 1389 static inline void ip_vs_update_conntrack(struct sk_buff *skb, 1390 struct ip_vs_conn *cp, int outin) 1391 { 1392 } 1393 1394 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb, 1395 struct ip_vs_conn *cp) 1396 { 1397 return NF_ACCEPT; 1398 } 1399 1400 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp) 1401 { 1402 } 1403 /* CONFIG_IP_VS_NFCT */ 1404 #endif 1405 1406 static inline unsigned int 1407 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest) 1408 { 1409 /* 1410 * We think the overhead of processing active connections is 256 1411 * times higher than that of inactive connections in average. (This 1412 * 256 times might not be accurate, we will change it later) We 1413 * use the following formula to estimate the overhead now: 1414 * dest->activeconns*256 + dest->inactconns 1415 */ 1416 return (atomic_read(&dest->activeconns) << 8) + 1417 atomic_read(&dest->inactconns); 1418 } 1419 1420 #endif /* __KERNEL__ */ 1421 1422 #endif /* _NET_IP_VS_H */ 1423