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 29 struct ip_vs_iphdr { 30 int len; 31 __u8 protocol; 32 union nf_inet_addr saddr; 33 union nf_inet_addr daddr; 34 }; 35 36 static inline void 37 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr) 38 { 39 #ifdef CONFIG_IP_VS_IPV6 40 if (af == AF_INET6) { 41 const struct ipv6hdr *iph = nh; 42 iphdr->len = sizeof(struct ipv6hdr); 43 iphdr->protocol = iph->nexthdr; 44 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr); 45 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr); 46 } else 47 #endif 48 { 49 const struct iphdr *iph = nh; 50 iphdr->len = iph->ihl * 4; 51 iphdr->protocol = iph->protocol; 52 iphdr->saddr.ip = iph->saddr; 53 iphdr->daddr.ip = iph->daddr; 54 } 55 } 56 57 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst, 58 const union nf_inet_addr *src) 59 { 60 #ifdef CONFIG_IP_VS_IPV6 61 if (af == AF_INET6) 62 ipv6_addr_copy(&dst->in6, &src->in6); 63 else 64 #endif 65 dst->ip = src->ip; 66 } 67 68 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a, 69 const union nf_inet_addr *b) 70 { 71 #ifdef CONFIG_IP_VS_IPV6 72 if (af == AF_INET6) 73 return ipv6_addr_equal(&a->in6, &b->in6); 74 #endif 75 return a->ip == b->ip; 76 } 77 78 #ifdef CONFIG_IP_VS_DEBUG 79 #include <linux/net.h> 80 81 extern int ip_vs_get_debug_level(void); 82 83 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len, 84 const union nf_inet_addr *addr, 85 int *idx) 86 { 87 int len; 88 #ifdef CONFIG_IP_VS_IPV6 89 if (af == AF_INET6) 90 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]", 91 &addr->in6) + 1; 92 else 93 #endif 94 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4", 95 &addr->ip) + 1; 96 97 *idx += len; 98 BUG_ON(*idx > buf_len + 1); 99 return &buf[*idx - len]; 100 } 101 102 #define IP_VS_DBG_BUF(level, msg, ...) \ 103 do { \ 104 char ip_vs_dbg_buf[160]; \ 105 int ip_vs_dbg_idx = 0; \ 106 if (level <= ip_vs_get_debug_level()) \ 107 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \ 108 } while (0) 109 #define IP_VS_ERR_BUF(msg...) \ 110 do { \ 111 char ip_vs_dbg_buf[160]; \ 112 int ip_vs_dbg_idx = 0; \ 113 pr_err(msg); \ 114 } while (0) 115 116 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */ 117 #define IP_VS_DBG_ADDR(af, addr) \ 118 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \ 119 sizeof(ip_vs_dbg_buf), addr, \ 120 &ip_vs_dbg_idx) 121 122 #define IP_VS_DBG(level, msg, ...) \ 123 do { \ 124 if (level <= ip_vs_get_debug_level()) \ 125 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \ 126 } while (0) 127 #define IP_VS_DBG_RL(msg, ...) \ 128 do { \ 129 if (net_ratelimit()) \ 130 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \ 131 } while (0) 132 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) \ 133 do { \ 134 if (level <= ip_vs_get_debug_level()) \ 135 pp->debug_packet(pp, skb, ofs, msg); \ 136 } while (0) 137 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) \ 138 do { \ 139 if (level <= ip_vs_get_debug_level() && \ 140 net_ratelimit()) \ 141 pp->debug_packet(pp, skb, ofs, msg); \ 142 } while (0) 143 #else /* NO DEBUGGING at ALL */ 144 #define IP_VS_DBG_BUF(level, msg...) do {} while (0) 145 #define IP_VS_ERR_BUF(msg...) do {} while (0) 146 #define IP_VS_DBG(level, msg...) do {} while (0) 147 #define IP_VS_DBG_RL(msg...) do {} while (0) 148 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) do {} while (0) 149 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) do {} while (0) 150 #endif 151 152 #define IP_VS_BUG() BUG() 153 #define IP_VS_ERR_RL(msg, ...) \ 154 do { \ 155 if (net_ratelimit()) \ 156 pr_err(msg, ##__VA_ARGS__); \ 157 } while (0) 158 159 #ifdef CONFIG_IP_VS_DEBUG 160 #define EnterFunction(level) \ 161 do { \ 162 if (level <= ip_vs_get_debug_level()) \ 163 printk(KERN_DEBUG \ 164 pr_fmt("Enter: %s, %s line %i\n"), \ 165 __func__, __FILE__, __LINE__); \ 166 } while (0) 167 #define LeaveFunction(level) \ 168 do { \ 169 if (level <= ip_vs_get_debug_level()) \ 170 printk(KERN_DEBUG \ 171 pr_fmt("Leave: %s, %s line %i\n"), \ 172 __func__, __FILE__, __LINE__); \ 173 } while (0) 174 #else 175 #define EnterFunction(level) do {} while (0) 176 #define LeaveFunction(level) do {} while (0) 177 #endif 178 179 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); } 180 181 182 /* 183 * The port number of FTP service (in network order). 184 */ 185 #define FTPPORT cpu_to_be16(21) 186 #define FTPDATA cpu_to_be16(20) 187 188 /* 189 * TCP State Values 190 */ 191 enum { 192 IP_VS_TCP_S_NONE = 0, 193 IP_VS_TCP_S_ESTABLISHED, 194 IP_VS_TCP_S_SYN_SENT, 195 IP_VS_TCP_S_SYN_RECV, 196 IP_VS_TCP_S_FIN_WAIT, 197 IP_VS_TCP_S_TIME_WAIT, 198 IP_VS_TCP_S_CLOSE, 199 IP_VS_TCP_S_CLOSE_WAIT, 200 IP_VS_TCP_S_LAST_ACK, 201 IP_VS_TCP_S_LISTEN, 202 IP_VS_TCP_S_SYNACK, 203 IP_VS_TCP_S_LAST 204 }; 205 206 /* 207 * UDP State Values 208 */ 209 enum { 210 IP_VS_UDP_S_NORMAL, 211 IP_VS_UDP_S_LAST, 212 }; 213 214 /* 215 * ICMP State Values 216 */ 217 enum { 218 IP_VS_ICMP_S_NORMAL, 219 IP_VS_ICMP_S_LAST, 220 }; 221 222 /* 223 * Delta sequence info structure 224 * Each ip_vs_conn has 2 (output AND input seq. changes). 225 * Only used in the VS/NAT. 226 */ 227 struct ip_vs_seq { 228 __u32 init_seq; /* Add delta from this seq */ 229 __u32 delta; /* Delta in sequence numbers */ 230 __u32 previous_delta; /* Delta in sequence numbers 231 before last resized pkt */ 232 }; 233 234 235 /* 236 * IPVS statistics objects 237 */ 238 struct ip_vs_estimator { 239 struct list_head list; 240 241 u64 last_inbytes; 242 u64 last_outbytes; 243 u32 last_conns; 244 u32 last_inpkts; 245 u32 last_outpkts; 246 247 u32 cps; 248 u32 inpps; 249 u32 outpps; 250 u32 inbps; 251 u32 outbps; 252 }; 253 254 struct ip_vs_stats { 255 struct ip_vs_stats_user ustats; /* statistics */ 256 struct ip_vs_estimator est; /* estimator */ 257 258 spinlock_t lock; /* spin lock */ 259 }; 260 261 struct dst_entry; 262 struct iphdr; 263 struct ip_vs_conn; 264 struct ip_vs_app; 265 struct sk_buff; 266 267 struct ip_vs_protocol { 268 struct ip_vs_protocol *next; 269 char *name; 270 u16 protocol; 271 u16 num_states; 272 int dont_defrag; 273 atomic_t appcnt; /* counter of proto app incs */ 274 int *timeout_table; /* protocol timeout table */ 275 276 void (*init)(struct ip_vs_protocol *pp); 277 278 void (*exit)(struct ip_vs_protocol *pp); 279 280 int (*conn_schedule)(int af, struct sk_buff *skb, 281 struct ip_vs_protocol *pp, 282 int *verdict, struct ip_vs_conn **cpp); 283 284 struct ip_vs_conn * 285 (*conn_in_get)(int af, 286 const struct sk_buff *skb, 287 struct ip_vs_protocol *pp, 288 const struct ip_vs_iphdr *iph, 289 unsigned int proto_off, 290 int inverse); 291 292 struct ip_vs_conn * 293 (*conn_out_get)(int af, 294 const struct sk_buff *skb, 295 struct ip_vs_protocol *pp, 296 const struct ip_vs_iphdr *iph, 297 unsigned int proto_off, 298 int inverse); 299 300 int (*snat_handler)(struct sk_buff *skb, 301 struct ip_vs_protocol *pp, struct ip_vs_conn *cp); 302 303 int (*dnat_handler)(struct sk_buff *skb, 304 struct ip_vs_protocol *pp, struct ip_vs_conn *cp); 305 306 int (*csum_check)(int af, struct sk_buff *skb, 307 struct ip_vs_protocol *pp); 308 309 const char *(*state_name)(int state); 310 311 int (*state_transition)(struct ip_vs_conn *cp, int direction, 312 const struct sk_buff *skb, 313 struct ip_vs_protocol *pp); 314 315 int (*register_app)(struct ip_vs_app *inc); 316 317 void (*unregister_app)(struct ip_vs_app *inc); 318 319 int (*app_conn_bind)(struct ip_vs_conn *cp); 320 321 void (*debug_packet)(struct ip_vs_protocol *pp, 322 const struct sk_buff *skb, 323 int offset, 324 const char *msg); 325 326 void (*timeout_change)(struct ip_vs_protocol *pp, int flags); 327 328 int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to); 329 }; 330 331 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto); 332 333 /* 334 * IP_VS structure allocated for each dynamically scheduled connection 335 */ 336 struct ip_vs_conn { 337 struct list_head c_list; /* hashed list heads */ 338 339 /* Protocol, addresses and port numbers */ 340 u16 af; /* address family */ 341 union nf_inet_addr caddr; /* client address */ 342 union nf_inet_addr vaddr; /* virtual address */ 343 union nf_inet_addr daddr; /* destination address */ 344 __be16 cport; 345 __be16 vport; 346 __be16 dport; 347 __u16 protocol; /* Which protocol (TCP/UDP) */ 348 349 /* counter and timer */ 350 atomic_t refcnt; /* reference count */ 351 struct timer_list timer; /* Expiration timer */ 352 volatile unsigned long timeout; /* timeout */ 353 354 /* Flags and state transition */ 355 spinlock_t lock; /* lock for state transition */ 356 volatile __u16 flags; /* status flags */ 357 volatile __u16 state; /* state info */ 358 volatile __u16 old_state; /* old state, to be used for 359 * state transition triggerd 360 * synchronization 361 */ 362 363 /* Control members */ 364 struct ip_vs_conn *control; /* Master control connection */ 365 atomic_t n_control; /* Number of controlled ones */ 366 struct ip_vs_dest *dest; /* real server */ 367 atomic_t in_pkts; /* incoming packet counter */ 368 369 /* packet transmitter for different forwarding methods. If it 370 mangles the packet, it must return NF_DROP or better NF_STOLEN, 371 otherwise this must be changed to a sk_buff **. 372 */ 373 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp, 374 struct ip_vs_protocol *pp); 375 376 /* Note: we can group the following members into a structure, 377 in order to save more space, and the following members are 378 only used in VS/NAT anyway */ 379 struct ip_vs_app *app; /* bound ip_vs_app object */ 380 void *app_data; /* Application private data */ 381 struct ip_vs_seq in_seq; /* incoming seq. struct */ 382 struct ip_vs_seq out_seq; /* outgoing seq. struct */ 383 }; 384 385 386 /* 387 * Extended internal versions of struct ip_vs_service_user and 388 * ip_vs_dest_user for IPv6 support. 389 * 390 * We need these to conveniently pass around service and destination 391 * options, but unfortunately, we also need to keep the old definitions to 392 * maintain userspace backwards compatibility for the setsockopt interface. 393 */ 394 struct ip_vs_service_user_kern { 395 /* virtual service addresses */ 396 u16 af; 397 u16 protocol; 398 union nf_inet_addr addr; /* virtual ip address */ 399 u16 port; 400 u32 fwmark; /* firwall mark of service */ 401 402 /* virtual service options */ 403 char *sched_name; 404 unsigned flags; /* virtual service flags */ 405 unsigned timeout; /* persistent timeout in sec */ 406 u32 netmask; /* persistent netmask */ 407 }; 408 409 410 struct ip_vs_dest_user_kern { 411 /* destination server address */ 412 union nf_inet_addr addr; 413 u16 port; 414 415 /* real server options */ 416 unsigned conn_flags; /* connection flags */ 417 int weight; /* destination weight */ 418 419 /* thresholds for active connections */ 420 u32 u_threshold; /* upper threshold */ 421 u32 l_threshold; /* lower threshold */ 422 }; 423 424 425 /* 426 * The information about the virtual service offered to the net 427 * and the forwarding entries 428 */ 429 struct ip_vs_service { 430 struct list_head s_list; /* for normal service table */ 431 struct list_head f_list; /* for fwmark-based service table */ 432 atomic_t refcnt; /* reference counter */ 433 atomic_t usecnt; /* use counter */ 434 435 u16 af; /* address family */ 436 __u16 protocol; /* which protocol (TCP/UDP) */ 437 union nf_inet_addr addr; /* IP address for virtual service */ 438 __be16 port; /* port number for the service */ 439 __u32 fwmark; /* firewall mark of the service */ 440 unsigned flags; /* service status flags */ 441 unsigned timeout; /* persistent timeout in ticks */ 442 __be32 netmask; /* grouping granularity */ 443 444 struct list_head destinations; /* real server d-linked list */ 445 __u32 num_dests; /* number of servers */ 446 struct ip_vs_stats stats; /* statistics for the service */ 447 struct ip_vs_app *inc; /* bind conns to this app inc */ 448 449 /* for scheduling */ 450 struct ip_vs_scheduler *scheduler; /* bound scheduler object */ 451 rwlock_t sched_lock; /* lock sched_data */ 452 void *sched_data; /* scheduler application data */ 453 }; 454 455 456 /* 457 * The real server destination forwarding entry 458 * with ip address, port number, and so on. 459 */ 460 struct ip_vs_dest { 461 struct list_head n_list; /* for the dests in the service */ 462 struct list_head d_list; /* for table with all the dests */ 463 464 u16 af; /* address family */ 465 union nf_inet_addr addr; /* IP address of the server */ 466 __be16 port; /* port number of the server */ 467 volatile unsigned flags; /* dest status flags */ 468 atomic_t conn_flags; /* flags to copy to conn */ 469 atomic_t weight; /* server weight */ 470 471 atomic_t refcnt; /* reference counter */ 472 struct ip_vs_stats stats; /* statistics */ 473 474 /* connection counters and thresholds */ 475 atomic_t activeconns; /* active connections */ 476 atomic_t inactconns; /* inactive connections */ 477 atomic_t persistconns; /* persistent connections */ 478 __u32 u_threshold; /* upper threshold */ 479 __u32 l_threshold; /* lower threshold */ 480 481 /* for destination cache */ 482 spinlock_t dst_lock; /* lock of dst_cache */ 483 struct dst_entry *dst_cache; /* destination cache entry */ 484 u32 dst_rtos; /* RT_TOS(tos) for dst */ 485 486 /* for virtual service */ 487 struct ip_vs_service *svc; /* service it belongs to */ 488 __u16 protocol; /* which protocol (TCP/UDP) */ 489 union nf_inet_addr vaddr; /* virtual IP address */ 490 __be16 vport; /* virtual port number */ 491 __u32 vfwmark; /* firewall mark of service */ 492 }; 493 494 495 /* 496 * The scheduler object 497 */ 498 struct ip_vs_scheduler { 499 struct list_head n_list; /* d-linked list head */ 500 char *name; /* scheduler name */ 501 atomic_t refcnt; /* reference counter */ 502 struct module *module; /* THIS_MODULE/NULL */ 503 504 /* scheduler initializing service */ 505 int (*init_service)(struct ip_vs_service *svc); 506 /* scheduling service finish */ 507 int (*done_service)(struct ip_vs_service *svc); 508 /* scheduler updating service */ 509 int (*update_service)(struct ip_vs_service *svc); 510 511 /* selecting a server from the given service */ 512 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc, 513 const struct sk_buff *skb); 514 }; 515 516 517 /* 518 * The application module object (a.k.a. app incarnation) 519 */ 520 struct ip_vs_app { 521 struct list_head a_list; /* member in app list */ 522 int type; /* IP_VS_APP_TYPE_xxx */ 523 char *name; /* application module name */ 524 __u16 protocol; 525 struct module *module; /* THIS_MODULE/NULL */ 526 struct list_head incs_list; /* list of incarnations */ 527 528 /* members for application incarnations */ 529 struct list_head p_list; /* member in proto app list */ 530 struct ip_vs_app *app; /* its real application */ 531 __be16 port; /* port number in net order */ 532 atomic_t usecnt; /* usage counter */ 533 534 /* output hook: return false if can't linearize. diff set for TCP. */ 535 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *, 536 struct sk_buff *, int *diff); 537 538 /* input hook: return false if can't linearize. diff set for TCP. */ 539 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *, 540 struct sk_buff *, int *diff); 541 542 /* ip_vs_app initializer */ 543 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *); 544 545 /* ip_vs_app finish */ 546 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *); 547 548 549 /* not used now */ 550 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *, 551 struct ip_vs_protocol *); 552 553 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *); 554 555 int * timeout_table; 556 int * timeouts; 557 int timeouts_size; 558 559 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app, 560 int *verdict, struct ip_vs_conn **cpp); 561 562 struct ip_vs_conn * 563 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app, 564 const struct iphdr *iph, unsigned int proto_off, 565 int inverse); 566 567 struct ip_vs_conn * 568 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app, 569 const struct iphdr *iph, unsigned int proto_off, 570 int inverse); 571 572 int (*state_transition)(struct ip_vs_conn *cp, int direction, 573 const struct sk_buff *skb, 574 struct ip_vs_app *app); 575 576 void (*timeout_change)(struct ip_vs_app *app, int flags); 577 }; 578 579 580 /* 581 * IPVS core functions 582 * (from ip_vs_core.c) 583 */ 584 extern const char *ip_vs_proto_name(unsigned proto); 585 extern void ip_vs_init_hash_table(struct list_head *table, int rows); 586 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t))) 587 588 #define IP_VS_APP_TYPE_FTP 1 589 590 /* 591 * ip_vs_conn handling functions 592 * (from ip_vs_conn.c) 593 */ 594 595 /* 596 * IPVS connection entry hash table 597 */ 598 #ifndef CONFIG_IP_VS_TAB_BITS 599 #define CONFIG_IP_VS_TAB_BITS 12 600 #endif 601 602 #define IP_VS_CONN_TAB_BITS CONFIG_IP_VS_TAB_BITS 603 #define IP_VS_CONN_TAB_SIZE (1 << IP_VS_CONN_TAB_BITS) 604 #define IP_VS_CONN_TAB_MASK (IP_VS_CONN_TAB_SIZE - 1) 605 606 enum { 607 IP_VS_DIR_INPUT = 0, 608 IP_VS_DIR_OUTPUT, 609 IP_VS_DIR_INPUT_ONLY, 610 IP_VS_DIR_LAST, 611 }; 612 613 extern struct ip_vs_conn *ip_vs_conn_in_get 614 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port, 615 const union nf_inet_addr *d_addr, __be16 d_port); 616 617 extern struct ip_vs_conn *ip_vs_ct_in_get 618 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port, 619 const union nf_inet_addr *d_addr, __be16 d_port); 620 621 extern struct ip_vs_conn *ip_vs_conn_out_get 622 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port, 623 const union nf_inet_addr *d_addr, __be16 d_port); 624 625 /* put back the conn without restarting its timer */ 626 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp) 627 { 628 atomic_dec(&cp->refcnt); 629 } 630 extern void ip_vs_conn_put(struct ip_vs_conn *cp); 631 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport); 632 633 extern struct ip_vs_conn * 634 ip_vs_conn_new(int af, int proto, const union nf_inet_addr *caddr, __be16 cport, 635 const union nf_inet_addr *vaddr, __be16 vport, 636 const union nf_inet_addr *daddr, __be16 dport, unsigned flags, 637 struct ip_vs_dest *dest); 638 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp); 639 640 extern const char * ip_vs_state_name(__u16 proto, int state); 641 642 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp); 643 extern int ip_vs_check_template(struct ip_vs_conn *ct); 644 extern void ip_vs_random_dropentry(void); 645 extern int ip_vs_conn_init(void); 646 extern void ip_vs_conn_cleanup(void); 647 648 static inline void ip_vs_control_del(struct ip_vs_conn *cp) 649 { 650 struct ip_vs_conn *ctl_cp = cp->control; 651 if (!ctl_cp) { 652 IP_VS_ERR_BUF("request control DEL for uncontrolled: " 653 "%s:%d to %s:%d\n", 654 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 655 ntohs(cp->cport), 656 IP_VS_DBG_ADDR(cp->af, &cp->vaddr), 657 ntohs(cp->vport)); 658 659 return; 660 } 661 662 IP_VS_DBG_BUF(7, "DELeting control for: " 663 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n", 664 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 665 ntohs(cp->cport), 666 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr), 667 ntohs(ctl_cp->cport)); 668 669 cp->control = NULL; 670 if (atomic_read(&ctl_cp->n_control) == 0) { 671 IP_VS_ERR_BUF("BUG control DEL with n=0 : " 672 "%s:%d to %s:%d\n", 673 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 674 ntohs(cp->cport), 675 IP_VS_DBG_ADDR(cp->af, &cp->vaddr), 676 ntohs(cp->vport)); 677 678 return; 679 } 680 atomic_dec(&ctl_cp->n_control); 681 } 682 683 static inline void 684 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp) 685 { 686 if (cp->control) { 687 IP_VS_ERR_BUF("request control ADD for already controlled: " 688 "%s:%d to %s:%d\n", 689 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 690 ntohs(cp->cport), 691 IP_VS_DBG_ADDR(cp->af, &cp->vaddr), 692 ntohs(cp->vport)); 693 694 ip_vs_control_del(cp); 695 } 696 697 IP_VS_DBG_BUF(7, "ADDing control for: " 698 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n", 699 IP_VS_DBG_ADDR(cp->af, &cp->caddr), 700 ntohs(cp->cport), 701 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr), 702 ntohs(ctl_cp->cport)); 703 704 cp->control = ctl_cp; 705 atomic_inc(&ctl_cp->n_control); 706 } 707 708 709 /* 710 * IPVS application functions 711 * (from ip_vs_app.c) 712 */ 713 #define IP_VS_APP_MAX_PORTS 8 714 extern int register_ip_vs_app(struct ip_vs_app *app); 715 extern void unregister_ip_vs_app(struct ip_vs_app *app); 716 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 717 extern void ip_vs_unbind_app(struct ip_vs_conn *cp); 718 extern int 719 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port); 720 extern int ip_vs_app_inc_get(struct ip_vs_app *inc); 721 extern void ip_vs_app_inc_put(struct ip_vs_app *inc); 722 723 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb); 724 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb); 725 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri, 726 char *o_buf, int o_len, char *n_buf, int n_len); 727 extern int ip_vs_app_init(void); 728 extern void ip_vs_app_cleanup(void); 729 730 731 /* 732 * IPVS protocol functions (from ip_vs_proto.c) 733 */ 734 extern int ip_vs_protocol_init(void); 735 extern void ip_vs_protocol_cleanup(void); 736 extern void ip_vs_protocol_timeout_change(int flags); 737 extern int *ip_vs_create_timeout_table(int *table, int size); 738 extern int 739 ip_vs_set_state_timeout(int *table, int num, const char *const *names, 740 const char *name, int to); 741 extern void 742 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb, 743 int offset, const char *msg); 744 745 extern struct ip_vs_protocol ip_vs_protocol_tcp; 746 extern struct ip_vs_protocol ip_vs_protocol_udp; 747 extern struct ip_vs_protocol ip_vs_protocol_icmp; 748 extern struct ip_vs_protocol ip_vs_protocol_esp; 749 extern struct ip_vs_protocol ip_vs_protocol_ah; 750 751 752 /* 753 * Registering/unregistering scheduler functions 754 * (from ip_vs_sched.c) 755 */ 756 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler); 757 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler); 758 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc, 759 struct ip_vs_scheduler *scheduler); 760 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc); 761 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name); 762 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler); 763 extern struct ip_vs_conn * 764 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb); 765 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb, 766 struct ip_vs_protocol *pp); 767 768 769 /* 770 * IPVS control data and functions (from ip_vs_ctl.c) 771 */ 772 extern int sysctl_ip_vs_cache_bypass; 773 extern int sysctl_ip_vs_expire_nodest_conn; 774 extern int sysctl_ip_vs_expire_quiescent_template; 775 extern int sysctl_ip_vs_sync_threshold[2]; 776 extern int sysctl_ip_vs_nat_icmp_send; 777 extern struct ip_vs_stats ip_vs_stats; 778 extern const struct ctl_path net_vs_ctl_path[]; 779 780 extern struct ip_vs_service * 781 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol, 782 const union nf_inet_addr *vaddr, __be16 vport); 783 784 static inline void ip_vs_service_put(struct ip_vs_service *svc) 785 { 786 atomic_dec(&svc->usecnt); 787 } 788 789 extern struct ip_vs_dest * 790 ip_vs_lookup_real_service(int af, __u16 protocol, 791 const union nf_inet_addr *daddr, __be16 dport); 792 793 extern int ip_vs_use_count_inc(void); 794 extern void ip_vs_use_count_dec(void); 795 extern int ip_vs_control_init(void); 796 extern void ip_vs_control_cleanup(void); 797 extern struct ip_vs_dest * 798 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport, 799 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol); 800 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp); 801 802 803 /* 804 * IPVS sync daemon data and function prototypes 805 * (from ip_vs_sync.c) 806 */ 807 extern volatile int ip_vs_sync_state; 808 extern volatile int ip_vs_master_syncid; 809 extern volatile int ip_vs_backup_syncid; 810 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN]; 811 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN]; 812 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid); 813 extern int stop_sync_thread(int state); 814 extern void ip_vs_sync_conn(struct ip_vs_conn *cp); 815 816 817 /* 818 * IPVS rate estimator prototypes (from ip_vs_est.c) 819 */ 820 extern int ip_vs_estimator_init(void); 821 extern void ip_vs_estimator_cleanup(void); 822 extern void ip_vs_new_estimator(struct ip_vs_stats *stats); 823 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats); 824 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats); 825 826 /* 827 * Various IPVS packet transmitters (from ip_vs_xmit.c) 828 */ 829 extern int ip_vs_null_xmit 830 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 831 extern int ip_vs_bypass_xmit 832 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 833 extern int ip_vs_nat_xmit 834 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 835 extern int ip_vs_tunnel_xmit 836 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 837 extern int ip_vs_dr_xmit 838 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 839 extern int ip_vs_icmp_xmit 840 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset); 841 extern void ip_vs_dst_reset(struct ip_vs_dest *dest); 842 843 #ifdef CONFIG_IP_VS_IPV6 844 extern int ip_vs_bypass_xmit_v6 845 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 846 extern int ip_vs_nat_xmit_v6 847 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 848 extern int ip_vs_tunnel_xmit_v6 849 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 850 extern int ip_vs_dr_xmit_v6 851 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 852 extern int ip_vs_icmp_xmit_v6 853 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, 854 int offset); 855 #endif 856 857 /* 858 * This is a simple mechanism to ignore packets when 859 * we are loaded. Just set ip_vs_drop_rate to 'n' and 860 * we start to drop 1/rate of the packets 861 */ 862 extern int ip_vs_drop_rate; 863 extern int ip_vs_drop_counter; 864 865 static __inline__ int ip_vs_todrop(void) 866 { 867 if (!ip_vs_drop_rate) return 0; 868 if (--ip_vs_drop_counter > 0) return 0; 869 ip_vs_drop_counter = ip_vs_drop_rate; 870 return 1; 871 } 872 873 /* 874 * ip_vs_fwd_tag returns the forwarding tag of the connection 875 */ 876 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK) 877 878 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp) 879 { 880 char fwd; 881 882 switch (IP_VS_FWD_METHOD(cp)) { 883 case IP_VS_CONN_F_MASQ: 884 fwd = 'M'; break; 885 case IP_VS_CONN_F_LOCALNODE: 886 fwd = 'L'; break; 887 case IP_VS_CONN_F_TUNNEL: 888 fwd = 'T'; break; 889 case IP_VS_CONN_F_DROUTE: 890 fwd = 'R'; break; 891 case IP_VS_CONN_F_BYPASS: 892 fwd = 'B'; break; 893 default: 894 fwd = '?'; break; 895 } 896 return fwd; 897 } 898 899 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp, 900 struct ip_vs_conn *cp, int dir); 901 902 #ifdef CONFIG_IP_VS_IPV6 903 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp, 904 struct ip_vs_conn *cp, int dir); 905 #endif 906 907 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset); 908 909 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum) 910 { 911 __be32 diff[2] = { ~old, new }; 912 913 return csum_partial(diff, sizeof(diff), oldsum); 914 } 915 916 #ifdef CONFIG_IP_VS_IPV6 917 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new, 918 __wsum oldsum) 919 { 920 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0], 921 new[3], new[2], new[1], new[0] }; 922 923 return csum_partial(diff, sizeof(diff), oldsum); 924 } 925 #endif 926 927 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum) 928 { 929 __be16 diff[2] = { ~old, new }; 930 931 return csum_partial(diff, sizeof(diff), oldsum); 932 } 933 934 #endif /* __KERNEL__ */ 935 936 #endif /* _NET_IP_VS_H */ 937