1 /* 2 * IP Virtual Server 3 * data structure and functionality definitions 4 */ 5 6 #ifndef _IP_VS_H 7 #define _IP_VS_H 8 9 #include <asm/types.h> /* For __uXX types */ 10 #include <linux/types.h> /* For __beXX types in userland */ 11 12 #include <linux/sysctl.h> /* For ctl_path */ 13 14 #define IP_VS_VERSION_CODE 0x010201 15 #define NVERSION(version) \ 16 (version >> 16) & 0xFF, \ 17 (version >> 8) & 0xFF, \ 18 version & 0xFF 19 20 /* 21 * Virtual Service Flags 22 */ 23 #define IP_VS_SVC_F_PERSISTENT 0x0001 /* persistent port */ 24 #define IP_VS_SVC_F_HASHED 0x0002 /* hashed entry */ 25 26 /* 27 * Destination Server Flags 28 */ 29 #define IP_VS_DEST_F_AVAILABLE 0x0001 /* server is available */ 30 #define IP_VS_DEST_F_OVERLOAD 0x0002 /* server is overloaded */ 31 32 /* 33 * IPVS sync daemon states 34 */ 35 #define IP_VS_STATE_NONE 0x0000 /* daemon is stopped */ 36 #define IP_VS_STATE_MASTER 0x0001 /* started as master */ 37 #define IP_VS_STATE_BACKUP 0x0002 /* started as backup */ 38 39 /* 40 * IPVS socket options 41 */ 42 #define IP_VS_BASE_CTL (64+1024+64) /* base */ 43 44 #define IP_VS_SO_SET_NONE IP_VS_BASE_CTL /* just peek */ 45 #define IP_VS_SO_SET_INSERT (IP_VS_BASE_CTL+1) 46 #define IP_VS_SO_SET_ADD (IP_VS_BASE_CTL+2) 47 #define IP_VS_SO_SET_EDIT (IP_VS_BASE_CTL+3) 48 #define IP_VS_SO_SET_DEL (IP_VS_BASE_CTL+4) 49 #define IP_VS_SO_SET_FLUSH (IP_VS_BASE_CTL+5) 50 #define IP_VS_SO_SET_LIST (IP_VS_BASE_CTL+6) 51 #define IP_VS_SO_SET_ADDDEST (IP_VS_BASE_CTL+7) 52 #define IP_VS_SO_SET_DELDEST (IP_VS_BASE_CTL+8) 53 #define IP_VS_SO_SET_EDITDEST (IP_VS_BASE_CTL+9) 54 #define IP_VS_SO_SET_TIMEOUT (IP_VS_BASE_CTL+10) 55 #define IP_VS_SO_SET_STARTDAEMON (IP_VS_BASE_CTL+11) 56 #define IP_VS_SO_SET_STOPDAEMON (IP_VS_BASE_CTL+12) 57 #define IP_VS_SO_SET_RESTORE (IP_VS_BASE_CTL+13) 58 #define IP_VS_SO_SET_SAVE (IP_VS_BASE_CTL+14) 59 #define IP_VS_SO_SET_ZERO (IP_VS_BASE_CTL+15) 60 #define IP_VS_SO_SET_MAX IP_VS_SO_SET_ZERO 61 62 #define IP_VS_SO_GET_VERSION IP_VS_BASE_CTL 63 #define IP_VS_SO_GET_INFO (IP_VS_BASE_CTL+1) 64 #define IP_VS_SO_GET_SERVICES (IP_VS_BASE_CTL+2) 65 #define IP_VS_SO_GET_SERVICE (IP_VS_BASE_CTL+3) 66 #define IP_VS_SO_GET_DESTS (IP_VS_BASE_CTL+4) 67 #define IP_VS_SO_GET_DEST (IP_VS_BASE_CTL+5) /* not used now */ 68 #define IP_VS_SO_GET_TIMEOUT (IP_VS_BASE_CTL+6) 69 #define IP_VS_SO_GET_DAEMON (IP_VS_BASE_CTL+7) 70 #define IP_VS_SO_GET_MAX IP_VS_SO_GET_DAEMON 71 72 73 /* 74 * IPVS Connection Flags 75 */ 76 #define IP_VS_CONN_F_FWD_MASK 0x0007 /* mask for the fwd methods */ 77 #define IP_VS_CONN_F_MASQ 0x0000 /* masquerading/NAT */ 78 #define IP_VS_CONN_F_LOCALNODE 0x0001 /* local node */ 79 #define IP_VS_CONN_F_TUNNEL 0x0002 /* tunneling */ 80 #define IP_VS_CONN_F_DROUTE 0x0003 /* direct routing */ 81 #define IP_VS_CONN_F_BYPASS 0x0004 /* cache bypass */ 82 #define IP_VS_CONN_F_SYNC 0x0020 /* entry created by sync */ 83 #define IP_VS_CONN_F_HASHED 0x0040 /* hashed entry */ 84 #define IP_VS_CONN_F_NOOUTPUT 0x0080 /* no output packets */ 85 #define IP_VS_CONN_F_INACTIVE 0x0100 /* not established */ 86 #define IP_VS_CONN_F_OUT_SEQ 0x0200 /* must do output seq adjust */ 87 #define IP_VS_CONN_F_IN_SEQ 0x0400 /* must do input seq adjust */ 88 #define IP_VS_CONN_F_SEQ_MASK 0x0600 /* in/out sequence mask */ 89 #define IP_VS_CONN_F_NO_CPORT 0x0800 /* no client port set yet */ 90 #define IP_VS_CONN_F_TEMPLATE 0x1000 /* template, not connection */ 91 92 /* Move it to better place one day, for now keep it unique */ 93 #define NFC_IPVS_PROPERTY 0x10000 94 95 #define IP_VS_SCHEDNAME_MAXLEN 16 96 #define IP_VS_IFNAME_MAXLEN 16 97 98 99 /* 100 * The struct ip_vs_service_user and struct ip_vs_dest_user are 101 * used to set IPVS rules through setsockopt. 102 */ 103 struct ip_vs_service_user { 104 /* virtual service addresses */ 105 u_int16_t protocol; 106 __be32 addr; /* virtual ip address */ 107 __be16 port; 108 u_int32_t fwmark; /* firwall mark of service */ 109 110 /* virtual service options */ 111 char sched_name[IP_VS_SCHEDNAME_MAXLEN]; 112 unsigned flags; /* virtual service flags */ 113 unsigned timeout; /* persistent timeout in sec */ 114 __be32 netmask; /* persistent netmask */ 115 }; 116 117 118 struct ip_vs_dest_user { 119 /* destination server address */ 120 __be32 addr; 121 __be16 port; 122 123 /* real server options */ 124 unsigned conn_flags; /* connection flags */ 125 int weight; /* destination weight */ 126 127 /* thresholds for active connections */ 128 u_int32_t u_threshold; /* upper threshold */ 129 u_int32_t l_threshold; /* lower threshold */ 130 }; 131 132 133 /* 134 * IPVS statistics object (for user space) 135 */ 136 struct ip_vs_stats_user 137 { 138 __u32 conns; /* connections scheduled */ 139 __u32 inpkts; /* incoming packets */ 140 __u32 outpkts; /* outgoing packets */ 141 __u64 inbytes; /* incoming bytes */ 142 __u64 outbytes; /* outgoing bytes */ 143 144 __u32 cps; /* current connection rate */ 145 __u32 inpps; /* current in packet rate */ 146 __u32 outpps; /* current out packet rate */ 147 __u32 inbps; /* current in byte rate */ 148 __u32 outbps; /* current out byte rate */ 149 }; 150 151 152 /* The argument to IP_VS_SO_GET_INFO */ 153 struct ip_vs_getinfo { 154 /* version number */ 155 unsigned int version; 156 157 /* size of connection hash table */ 158 unsigned int size; 159 160 /* number of virtual services */ 161 unsigned int num_services; 162 }; 163 164 165 /* The argument to IP_VS_SO_GET_SERVICE */ 166 struct ip_vs_service_entry { 167 /* which service: user fills in these */ 168 u_int16_t protocol; 169 __be32 addr; /* virtual address */ 170 __be16 port; 171 u_int32_t fwmark; /* firwall mark of service */ 172 173 /* service options */ 174 char sched_name[IP_VS_SCHEDNAME_MAXLEN]; 175 unsigned flags; /* virtual service flags */ 176 unsigned timeout; /* persistent timeout */ 177 __be32 netmask; /* persistent netmask */ 178 179 /* number of real servers */ 180 unsigned int num_dests; 181 182 /* statistics */ 183 struct ip_vs_stats_user stats; 184 }; 185 186 187 struct ip_vs_dest_entry { 188 __be32 addr; /* destination address */ 189 __be16 port; 190 unsigned conn_flags; /* connection flags */ 191 int weight; /* destination weight */ 192 193 u_int32_t u_threshold; /* upper threshold */ 194 u_int32_t l_threshold; /* lower threshold */ 195 196 u_int32_t activeconns; /* active connections */ 197 u_int32_t inactconns; /* inactive connections */ 198 u_int32_t persistconns; /* persistent connections */ 199 200 /* statistics */ 201 struct ip_vs_stats_user stats; 202 }; 203 204 205 /* The argument to IP_VS_SO_GET_DESTS */ 206 struct ip_vs_get_dests { 207 /* which service: user fills in these */ 208 u_int16_t protocol; 209 __be32 addr; /* virtual address */ 210 __be16 port; 211 u_int32_t fwmark; /* firwall mark of service */ 212 213 /* number of real servers */ 214 unsigned int num_dests; 215 216 /* the real servers */ 217 struct ip_vs_dest_entry entrytable[0]; 218 }; 219 220 221 /* The argument to IP_VS_SO_GET_SERVICES */ 222 struct ip_vs_get_services { 223 /* number of virtual services */ 224 unsigned int num_services; 225 226 /* service table */ 227 struct ip_vs_service_entry entrytable[0]; 228 }; 229 230 231 /* The argument to IP_VS_SO_GET_TIMEOUT */ 232 struct ip_vs_timeout_user { 233 int tcp_timeout; 234 int tcp_fin_timeout; 235 int udp_timeout; 236 }; 237 238 239 /* The argument to IP_VS_SO_GET_DAEMON */ 240 struct ip_vs_daemon_user { 241 /* sync daemon state (master/backup) */ 242 int state; 243 244 /* multicast interface name */ 245 char mcast_ifn[IP_VS_IFNAME_MAXLEN]; 246 247 /* SyncID we belong to */ 248 int syncid; 249 }; 250 251 252 #ifdef __KERNEL__ 253 254 #include <linux/list.h> /* for struct list_head */ 255 #include <linux/spinlock.h> /* for struct rwlock_t */ 256 #include <asm/atomic.h> /* for struct atomic_t */ 257 #include <linux/compiler.h> 258 #include <linux/timer.h> 259 260 #include <net/checksum.h> 261 262 #ifdef CONFIG_IP_VS_DEBUG 263 #include <linux/net.h> 264 265 extern int ip_vs_get_debug_level(void); 266 #define IP_VS_DBG(level, msg...) \ 267 do { \ 268 if (level <= ip_vs_get_debug_level()) \ 269 printk(KERN_DEBUG "IPVS: " msg); \ 270 } while (0) 271 #define IP_VS_DBG_RL(msg...) \ 272 do { \ 273 if (net_ratelimit()) \ 274 printk(KERN_DEBUG "IPVS: " msg); \ 275 } while (0) 276 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) \ 277 do { \ 278 if (level <= ip_vs_get_debug_level()) \ 279 pp->debug_packet(pp, skb, ofs, msg); \ 280 } while (0) 281 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) \ 282 do { \ 283 if (level <= ip_vs_get_debug_level() && \ 284 net_ratelimit()) \ 285 pp->debug_packet(pp, skb, ofs, msg); \ 286 } while (0) 287 #else /* NO DEBUGGING at ALL */ 288 #define IP_VS_DBG(level, msg...) do {} while (0) 289 #define IP_VS_DBG_RL(msg...) do {} while (0) 290 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) do {} while (0) 291 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) do {} while (0) 292 #endif 293 294 #define IP_VS_BUG() BUG() 295 #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg) 296 #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg) 297 #define IP_VS_WARNING(msg...) \ 298 printk(KERN_WARNING "IPVS: " msg) 299 #define IP_VS_ERR_RL(msg...) \ 300 do { \ 301 if (net_ratelimit()) \ 302 printk(KERN_ERR "IPVS: " msg); \ 303 } while (0) 304 305 #ifdef CONFIG_IP_VS_DEBUG 306 #define EnterFunction(level) \ 307 do { \ 308 if (level <= ip_vs_get_debug_level()) \ 309 printk(KERN_DEBUG "Enter: %s, %s line %i\n", \ 310 __FUNCTION__, __FILE__, __LINE__); \ 311 } while (0) 312 #define LeaveFunction(level) \ 313 do { \ 314 if (level <= ip_vs_get_debug_level()) \ 315 printk(KERN_DEBUG "Leave: %s, %s line %i\n", \ 316 __FUNCTION__, __FILE__, __LINE__); \ 317 } while (0) 318 #else 319 #define EnterFunction(level) do {} while (0) 320 #define LeaveFunction(level) do {} while (0) 321 #endif 322 323 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); } 324 325 326 /* 327 * The port number of FTP service (in network order). 328 */ 329 #define FTPPORT __constant_htons(21) 330 #define FTPDATA __constant_htons(20) 331 332 /* 333 * TCP State Values 334 */ 335 enum { 336 IP_VS_TCP_S_NONE = 0, 337 IP_VS_TCP_S_ESTABLISHED, 338 IP_VS_TCP_S_SYN_SENT, 339 IP_VS_TCP_S_SYN_RECV, 340 IP_VS_TCP_S_FIN_WAIT, 341 IP_VS_TCP_S_TIME_WAIT, 342 IP_VS_TCP_S_CLOSE, 343 IP_VS_TCP_S_CLOSE_WAIT, 344 IP_VS_TCP_S_LAST_ACK, 345 IP_VS_TCP_S_LISTEN, 346 IP_VS_TCP_S_SYNACK, 347 IP_VS_TCP_S_LAST 348 }; 349 350 /* 351 * UDP State Values 352 */ 353 enum { 354 IP_VS_UDP_S_NORMAL, 355 IP_VS_UDP_S_LAST, 356 }; 357 358 /* 359 * ICMP State Values 360 */ 361 enum { 362 IP_VS_ICMP_S_NORMAL, 363 IP_VS_ICMP_S_LAST, 364 }; 365 366 /* 367 * Delta sequence info structure 368 * Each ip_vs_conn has 2 (output AND input seq. changes). 369 * Only used in the VS/NAT. 370 */ 371 struct ip_vs_seq { 372 __u32 init_seq; /* Add delta from this seq */ 373 __u32 delta; /* Delta in sequence numbers */ 374 __u32 previous_delta; /* Delta in sequence numbers 375 before last resized pkt */ 376 }; 377 378 379 /* 380 * IPVS statistics object 381 */ 382 struct ip_vs_stats 383 { 384 __u32 conns; /* connections scheduled */ 385 __u32 inpkts; /* incoming packets */ 386 __u32 outpkts; /* outgoing packets */ 387 __u64 inbytes; /* incoming bytes */ 388 __u64 outbytes; /* outgoing bytes */ 389 390 __u32 cps; /* current connection rate */ 391 __u32 inpps; /* current in packet rate */ 392 __u32 outpps; /* current out packet rate */ 393 __u32 inbps; /* current in byte rate */ 394 __u32 outbps; /* current out byte rate */ 395 396 spinlock_t lock; /* spin lock */ 397 }; 398 399 struct dst_entry; 400 struct iphdr; 401 struct ip_vs_conn; 402 struct ip_vs_app; 403 struct sk_buff; 404 405 struct ip_vs_protocol { 406 struct ip_vs_protocol *next; 407 char *name; 408 u16 protocol; 409 u16 num_states; 410 int dont_defrag; 411 atomic_t appcnt; /* counter of proto app incs */ 412 int *timeout_table; /* protocol timeout table */ 413 414 void (*init)(struct ip_vs_protocol *pp); 415 416 void (*exit)(struct ip_vs_protocol *pp); 417 418 int (*conn_schedule)(struct sk_buff *skb, 419 struct ip_vs_protocol *pp, 420 int *verdict, struct ip_vs_conn **cpp); 421 422 struct ip_vs_conn * 423 (*conn_in_get)(const struct sk_buff *skb, 424 struct ip_vs_protocol *pp, 425 const struct iphdr *iph, 426 unsigned int proto_off, 427 int inverse); 428 429 struct ip_vs_conn * 430 (*conn_out_get)(const struct sk_buff *skb, 431 struct ip_vs_protocol *pp, 432 const struct iphdr *iph, 433 unsigned int proto_off, 434 int inverse); 435 436 int (*snat_handler)(struct sk_buff *skb, 437 struct ip_vs_protocol *pp, struct ip_vs_conn *cp); 438 439 int (*dnat_handler)(struct sk_buff *skb, 440 struct ip_vs_protocol *pp, struct ip_vs_conn *cp); 441 442 int (*csum_check)(struct sk_buff *skb, struct ip_vs_protocol *pp); 443 444 const char *(*state_name)(int state); 445 446 int (*state_transition)(struct ip_vs_conn *cp, int direction, 447 const struct sk_buff *skb, 448 struct ip_vs_protocol *pp); 449 450 int (*register_app)(struct ip_vs_app *inc); 451 452 void (*unregister_app)(struct ip_vs_app *inc); 453 454 int (*app_conn_bind)(struct ip_vs_conn *cp); 455 456 void (*debug_packet)(struct ip_vs_protocol *pp, 457 const struct sk_buff *skb, 458 int offset, 459 const char *msg); 460 461 void (*timeout_change)(struct ip_vs_protocol *pp, int flags); 462 463 int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to); 464 }; 465 466 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto); 467 468 /* 469 * IP_VS structure allocated for each dynamically scheduled connection 470 */ 471 struct ip_vs_conn { 472 struct list_head c_list; /* hashed list heads */ 473 474 /* Protocol, addresses and port numbers */ 475 __be32 caddr; /* client address */ 476 __be32 vaddr; /* virtual address */ 477 __be32 daddr; /* destination address */ 478 __be16 cport; 479 __be16 vport; 480 __be16 dport; 481 __u16 protocol; /* Which protocol (TCP/UDP) */ 482 483 /* counter and timer */ 484 atomic_t refcnt; /* reference count */ 485 struct timer_list timer; /* Expiration timer */ 486 volatile unsigned long timeout; /* timeout */ 487 488 /* Flags and state transition */ 489 spinlock_t lock; /* lock for state transition */ 490 volatile __u16 flags; /* status flags */ 491 volatile __u16 state; /* state info */ 492 volatile __u16 old_state; /* old state, to be used for 493 * state transition triggerd 494 * synchronization 495 */ 496 497 /* Control members */ 498 struct ip_vs_conn *control; /* Master control connection */ 499 atomic_t n_control; /* Number of controlled ones */ 500 struct ip_vs_dest *dest; /* real server */ 501 atomic_t in_pkts; /* incoming packet counter */ 502 503 /* packet transmitter for different forwarding methods. If it 504 mangles the packet, it must return NF_DROP or better NF_STOLEN, 505 otherwise this must be changed to a sk_buff **. 506 */ 507 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp, 508 struct ip_vs_protocol *pp); 509 510 /* Note: we can group the following members into a structure, 511 in order to save more space, and the following members are 512 only used in VS/NAT anyway */ 513 struct ip_vs_app *app; /* bound ip_vs_app object */ 514 void *app_data; /* Application private data */ 515 struct ip_vs_seq in_seq; /* incoming seq. struct */ 516 struct ip_vs_seq out_seq; /* outgoing seq. struct */ 517 }; 518 519 520 /* 521 * The information about the virtual service offered to the net 522 * and the forwarding entries 523 */ 524 struct ip_vs_service { 525 struct list_head s_list; /* for normal service table */ 526 struct list_head f_list; /* for fwmark-based service table */ 527 atomic_t refcnt; /* reference counter */ 528 atomic_t usecnt; /* use counter */ 529 530 __u16 protocol; /* which protocol (TCP/UDP) */ 531 __be32 addr; /* IP address for virtual service */ 532 __be16 port; /* port number for the service */ 533 __u32 fwmark; /* firewall mark of the service */ 534 unsigned flags; /* service status flags */ 535 unsigned timeout; /* persistent timeout in ticks */ 536 __be32 netmask; /* grouping granularity */ 537 538 struct list_head destinations; /* real server d-linked list */ 539 __u32 num_dests; /* number of servers */ 540 struct ip_vs_stats stats; /* statistics for the service */ 541 struct ip_vs_app *inc; /* bind conns to this app inc */ 542 543 /* for scheduling */ 544 struct ip_vs_scheduler *scheduler; /* bound scheduler object */ 545 rwlock_t sched_lock; /* lock sched_data */ 546 void *sched_data; /* scheduler application data */ 547 }; 548 549 550 /* 551 * The real server destination forwarding entry 552 * with ip address, port number, and so on. 553 */ 554 struct ip_vs_dest { 555 struct list_head n_list; /* for the dests in the service */ 556 struct list_head d_list; /* for table with all the dests */ 557 558 __be32 addr; /* IP address of the server */ 559 __be16 port; /* port number of the server */ 560 volatile unsigned flags; /* dest status flags */ 561 atomic_t conn_flags; /* flags to copy to conn */ 562 atomic_t weight; /* server weight */ 563 564 atomic_t refcnt; /* reference counter */ 565 struct ip_vs_stats stats; /* statistics */ 566 567 /* connection counters and thresholds */ 568 atomic_t activeconns; /* active connections */ 569 atomic_t inactconns; /* inactive connections */ 570 atomic_t persistconns; /* persistent connections */ 571 __u32 u_threshold; /* upper threshold */ 572 __u32 l_threshold; /* lower threshold */ 573 574 /* for destination cache */ 575 spinlock_t dst_lock; /* lock of dst_cache */ 576 struct dst_entry *dst_cache; /* destination cache entry */ 577 u32 dst_rtos; /* RT_TOS(tos) for dst */ 578 579 /* for virtual service */ 580 struct ip_vs_service *svc; /* service it belongs to */ 581 __u16 protocol; /* which protocol (TCP/UDP) */ 582 __be32 vaddr; /* virtual IP address */ 583 __be16 vport; /* virtual port number */ 584 __u32 vfwmark; /* firewall mark of service */ 585 }; 586 587 588 /* 589 * The scheduler object 590 */ 591 struct ip_vs_scheduler { 592 struct list_head n_list; /* d-linked list head */ 593 char *name; /* scheduler name */ 594 atomic_t refcnt; /* reference counter */ 595 struct module *module; /* THIS_MODULE/NULL */ 596 597 /* scheduler initializing service */ 598 int (*init_service)(struct ip_vs_service *svc); 599 /* scheduling service finish */ 600 int (*done_service)(struct ip_vs_service *svc); 601 /* scheduler updating service */ 602 int (*update_service)(struct ip_vs_service *svc); 603 604 /* selecting a server from the given service */ 605 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc, 606 const struct sk_buff *skb); 607 }; 608 609 610 /* 611 * The application module object (a.k.a. app incarnation) 612 */ 613 struct ip_vs_app 614 { 615 struct list_head a_list; /* member in app list */ 616 int type; /* IP_VS_APP_TYPE_xxx */ 617 char *name; /* application module name */ 618 __u16 protocol; 619 struct module *module; /* THIS_MODULE/NULL */ 620 struct list_head incs_list; /* list of incarnations */ 621 622 /* members for application incarnations */ 623 struct list_head p_list; /* member in proto app list */ 624 struct ip_vs_app *app; /* its real application */ 625 __be16 port; /* port number in net order */ 626 atomic_t usecnt; /* usage counter */ 627 628 /* output hook: return false if can't linearize. diff set for TCP. */ 629 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *, 630 struct sk_buff *, int *diff); 631 632 /* input hook: return false if can't linearize. diff set for TCP. */ 633 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *, 634 struct sk_buff *, int *diff); 635 636 /* ip_vs_app initializer */ 637 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *); 638 639 /* ip_vs_app finish */ 640 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *); 641 642 643 /* not used now */ 644 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *, 645 struct ip_vs_protocol *); 646 647 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *); 648 649 int * timeout_table; 650 int * timeouts; 651 int timeouts_size; 652 653 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app, 654 int *verdict, struct ip_vs_conn **cpp); 655 656 struct ip_vs_conn * 657 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app, 658 const struct iphdr *iph, unsigned int proto_off, 659 int inverse); 660 661 struct ip_vs_conn * 662 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app, 663 const struct iphdr *iph, unsigned int proto_off, 664 int inverse); 665 666 int (*state_transition)(struct ip_vs_conn *cp, int direction, 667 const struct sk_buff *skb, 668 struct ip_vs_app *app); 669 670 void (*timeout_change)(struct ip_vs_app *app, int flags); 671 }; 672 673 674 /* 675 * IPVS core functions 676 * (from ip_vs_core.c) 677 */ 678 extern const char *ip_vs_proto_name(unsigned proto); 679 extern void ip_vs_init_hash_table(struct list_head *table, int rows); 680 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table(t, sizeof(t)/sizeof(t[0])) 681 682 #define IP_VS_APP_TYPE_FTP 1 683 684 /* 685 * ip_vs_conn handling functions 686 * (from ip_vs_conn.c) 687 */ 688 689 /* 690 * IPVS connection entry hash table 691 */ 692 #ifndef CONFIG_IP_VS_TAB_BITS 693 #define CONFIG_IP_VS_TAB_BITS 12 694 #endif 695 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */ 696 #if CONFIG_IP_VS_TAB_BITS < 8 697 #define IP_VS_CONN_TAB_BITS 8 698 #endif 699 #if CONFIG_IP_VS_TAB_BITS > 20 700 #define IP_VS_CONN_TAB_BITS 20 701 #endif 702 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20 703 #define IP_VS_CONN_TAB_BITS CONFIG_IP_VS_TAB_BITS 704 #endif 705 #define IP_VS_CONN_TAB_SIZE (1 << IP_VS_CONN_TAB_BITS) 706 #define IP_VS_CONN_TAB_MASK (IP_VS_CONN_TAB_SIZE - 1) 707 708 enum { 709 IP_VS_DIR_INPUT = 0, 710 IP_VS_DIR_OUTPUT, 711 IP_VS_DIR_INPUT_ONLY, 712 IP_VS_DIR_LAST, 713 }; 714 715 extern struct ip_vs_conn *ip_vs_conn_in_get 716 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port); 717 extern struct ip_vs_conn *ip_vs_ct_in_get 718 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port); 719 extern struct ip_vs_conn *ip_vs_conn_out_get 720 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port); 721 722 /* put back the conn without restarting its timer */ 723 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp) 724 { 725 atomic_dec(&cp->refcnt); 726 } 727 extern void ip_vs_conn_put(struct ip_vs_conn *cp); 728 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport); 729 730 extern struct ip_vs_conn * 731 ip_vs_conn_new(int proto, __be32 caddr, __be16 cport, __be32 vaddr, __be16 vport, 732 __be32 daddr, __be16 dport, unsigned flags, 733 struct ip_vs_dest *dest); 734 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp); 735 736 extern const char * ip_vs_state_name(__u16 proto, int state); 737 738 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp); 739 extern int ip_vs_check_template(struct ip_vs_conn *ct); 740 extern void ip_vs_random_dropentry(void); 741 extern int ip_vs_conn_init(void); 742 extern void ip_vs_conn_cleanup(void); 743 744 static inline void ip_vs_control_del(struct ip_vs_conn *cp) 745 { 746 struct ip_vs_conn *ctl_cp = cp->control; 747 if (!ctl_cp) { 748 IP_VS_ERR("request control DEL for uncontrolled: " 749 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n", 750 NIPQUAD(cp->caddr),ntohs(cp->cport), 751 NIPQUAD(cp->vaddr),ntohs(cp->vport)); 752 return; 753 } 754 755 IP_VS_DBG(7, "DELeting control for: " 756 "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n", 757 NIPQUAD(cp->caddr),ntohs(cp->cport), 758 NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport)); 759 760 cp->control = NULL; 761 if (atomic_read(&ctl_cp->n_control) == 0) { 762 IP_VS_ERR("BUG control DEL with n=0 : " 763 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n", 764 NIPQUAD(cp->caddr),ntohs(cp->cport), 765 NIPQUAD(cp->vaddr),ntohs(cp->vport)); 766 return; 767 } 768 atomic_dec(&ctl_cp->n_control); 769 } 770 771 static inline void 772 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp) 773 { 774 if (cp->control) { 775 IP_VS_ERR("request control ADD for already controlled: " 776 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n", 777 NIPQUAD(cp->caddr),ntohs(cp->cport), 778 NIPQUAD(cp->vaddr),ntohs(cp->vport)); 779 ip_vs_control_del(cp); 780 } 781 782 IP_VS_DBG(7, "ADDing control for: " 783 "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n", 784 NIPQUAD(cp->caddr),ntohs(cp->cport), 785 NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport)); 786 787 cp->control = ctl_cp; 788 atomic_inc(&ctl_cp->n_control); 789 } 790 791 792 /* 793 * IPVS application functions 794 * (from ip_vs_app.c) 795 */ 796 #define IP_VS_APP_MAX_PORTS 8 797 extern int register_ip_vs_app(struct ip_vs_app *app); 798 extern void unregister_ip_vs_app(struct ip_vs_app *app); 799 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 800 extern void ip_vs_unbind_app(struct ip_vs_conn *cp); 801 extern int 802 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port); 803 extern int ip_vs_app_inc_get(struct ip_vs_app *inc); 804 extern void ip_vs_app_inc_put(struct ip_vs_app *inc); 805 806 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb); 807 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb); 808 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri, 809 char *o_buf, int o_len, char *n_buf, int n_len); 810 extern int ip_vs_app_init(void); 811 extern void ip_vs_app_cleanup(void); 812 813 814 /* 815 * IPVS protocol functions (from ip_vs_proto.c) 816 */ 817 extern int ip_vs_protocol_init(void); 818 extern void ip_vs_protocol_cleanup(void); 819 extern void ip_vs_protocol_timeout_change(int flags); 820 extern int *ip_vs_create_timeout_table(int *table, int size); 821 extern int 822 ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to); 823 extern void 824 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb, 825 int offset, const char *msg); 826 827 extern struct ip_vs_protocol ip_vs_protocol_tcp; 828 extern struct ip_vs_protocol ip_vs_protocol_udp; 829 extern struct ip_vs_protocol ip_vs_protocol_icmp; 830 extern struct ip_vs_protocol ip_vs_protocol_esp; 831 extern struct ip_vs_protocol ip_vs_protocol_ah; 832 833 834 /* 835 * Registering/unregistering scheduler functions 836 * (from ip_vs_sched.c) 837 */ 838 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler); 839 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler); 840 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc, 841 struct ip_vs_scheduler *scheduler); 842 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc); 843 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name); 844 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler); 845 extern struct ip_vs_conn * 846 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb); 847 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb, 848 struct ip_vs_protocol *pp); 849 850 851 /* 852 * IPVS control data and functions (from ip_vs_ctl.c) 853 */ 854 extern int sysctl_ip_vs_cache_bypass; 855 extern int sysctl_ip_vs_expire_nodest_conn; 856 extern int sysctl_ip_vs_expire_quiescent_template; 857 extern int sysctl_ip_vs_sync_threshold[2]; 858 extern int sysctl_ip_vs_nat_icmp_send; 859 extern struct ip_vs_stats ip_vs_stats; 860 extern struct ctl_path net_vs_ctl_path[]; 861 862 extern struct ip_vs_service * 863 ip_vs_service_get(__u32 fwmark, __u16 protocol, __be32 vaddr, __be16 vport); 864 865 static inline void ip_vs_service_put(struct ip_vs_service *svc) 866 { 867 atomic_dec(&svc->usecnt); 868 } 869 870 extern struct ip_vs_dest * 871 ip_vs_lookup_real_service(__u16 protocol, __be32 daddr, __be16 dport); 872 extern int ip_vs_use_count_inc(void); 873 extern void ip_vs_use_count_dec(void); 874 extern int ip_vs_control_init(void); 875 extern void ip_vs_control_cleanup(void); 876 extern struct ip_vs_dest * 877 ip_vs_find_dest(__be32 daddr, __be16 dport, 878 __be32 vaddr, __be16 vport, __u16 protocol); 879 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp); 880 881 882 /* 883 * IPVS sync daemon data and function prototypes 884 * (from ip_vs_sync.c) 885 */ 886 extern volatile int ip_vs_sync_state; 887 extern volatile int ip_vs_master_syncid; 888 extern volatile int ip_vs_backup_syncid; 889 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN]; 890 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN]; 891 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid); 892 extern int stop_sync_thread(int state); 893 extern void ip_vs_sync_conn(struct ip_vs_conn *cp); 894 895 896 /* 897 * IPVS rate estimator prototypes (from ip_vs_est.c) 898 */ 899 extern int ip_vs_new_estimator(struct ip_vs_stats *stats); 900 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats); 901 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats); 902 903 /* 904 * Various IPVS packet transmitters (from ip_vs_xmit.c) 905 */ 906 extern int ip_vs_null_xmit 907 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 908 extern int ip_vs_bypass_xmit 909 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 910 extern int ip_vs_nat_xmit 911 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 912 extern int ip_vs_tunnel_xmit 913 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 914 extern int ip_vs_dr_xmit 915 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); 916 extern int ip_vs_icmp_xmit 917 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset); 918 extern void ip_vs_dst_reset(struct ip_vs_dest *dest); 919 920 921 /* 922 * This is a simple mechanism to ignore packets when 923 * we are loaded. Just set ip_vs_drop_rate to 'n' and 924 * we start to drop 1/rate of the packets 925 */ 926 extern int ip_vs_drop_rate; 927 extern int ip_vs_drop_counter; 928 929 static __inline__ int ip_vs_todrop(void) 930 { 931 if (!ip_vs_drop_rate) return 0; 932 if (--ip_vs_drop_counter > 0) return 0; 933 ip_vs_drop_counter = ip_vs_drop_rate; 934 return 1; 935 } 936 937 /* 938 * ip_vs_fwd_tag returns the forwarding tag of the connection 939 */ 940 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK) 941 942 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp) 943 { 944 char fwd; 945 946 switch (IP_VS_FWD_METHOD(cp)) { 947 case IP_VS_CONN_F_MASQ: 948 fwd = 'M'; break; 949 case IP_VS_CONN_F_LOCALNODE: 950 fwd = 'L'; break; 951 case IP_VS_CONN_F_TUNNEL: 952 fwd = 'T'; break; 953 case IP_VS_CONN_F_DROUTE: 954 fwd = 'R'; break; 955 case IP_VS_CONN_F_BYPASS: 956 fwd = 'B'; break; 957 default: 958 fwd = '?'; break; 959 } 960 return fwd; 961 } 962 963 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp, 964 struct ip_vs_conn *cp, int dir); 965 966 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset); 967 968 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum) 969 { 970 __be32 diff[2] = { ~old, new }; 971 972 return csum_partial((char *) diff, sizeof(diff), oldsum); 973 } 974 975 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum) 976 { 977 __be16 diff[2] = { ~old, new }; 978 979 return csum_partial((char *) diff, sizeof(diff), oldsum); 980 } 981 982 #endif /* __KERNEL__ */ 983 984 #endif /* _IP_VS_H */ 985