include/net/ip_vs.h

   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
  30 /* Connections' size value needed by ip_vs_ctl.c */
  31 extern int ip_vs_conn_tab_size;
  32
  33
  34 struct ip_vs_iphdr {
  35         int len;
  36         __u8 protocol;
  37         union nf_inet_addr saddr;
  38         union nf_inet_addr daddr;
  39 };
  40
  41 static inline void
  42 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
  43 {
  44 #ifdef CONFIG_IP_VS_IPV6
  45         if (af == AF_INET6) {
  46                 const struct ipv6hdr *iph = nh;
  47                 iphdr->len = sizeof(struct ipv6hdr);
  48                 iphdr->protocol = iph->nexthdr;
  49                 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
  50                 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
  51         } else
  52 #endif
  53         {
  54                 const struct iphdr *iph = nh;
  55                 iphdr->len = iph->ihl * 4;
  56                 iphdr->protocol = iph->protocol;
  57                 iphdr->saddr.ip = iph->saddr;
  58                 iphdr->daddr.ip = iph->daddr;
  59         }
  60 }
  61
  62 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
  63                                    const union nf_inet_addr *src)
  64 {
  65 #ifdef CONFIG_IP_VS_IPV6
  66         if (af == AF_INET6)
  67                 ipv6_addr_copy(&dst->in6, &src->in6);
  68         else
  69 #endif
  70         dst->ip = src->ip;
  71 }
  72
  73 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
  74                                    const union nf_inet_addr *b)
  75 {
  76 #ifdef CONFIG_IP_VS_IPV6
  77         if (af == AF_INET6)
  78                 return ipv6_addr_equal(&a->in6, &b->in6);
  79 #endif
  80         return a->ip == b->ip;
  81 }
  82
  83 #ifdef CONFIG_IP_VS_DEBUG
  84 #include <linux/net.h>
  85
  86 extern int ip_vs_get_debug_level(void);
  87
  88 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
  89                                          const union nf_inet_addr *addr,
  90                                          int *idx)
  91 {
  92         int len;
  93 #ifdef CONFIG_IP_VS_IPV6
  94         if (af == AF_INET6)
  95                 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
  96                                &addr->in6) + 1;
  97         else
  98 #endif
  99                 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
 100                                &addr->ip) + 1;
 101
 102         *idx += len;
 103         BUG_ON(*idx > buf_len + 1);
 104         return &buf[*idx - len];
 105 }
 106
 107 #define IP_VS_DBG_BUF(level, msg, ...)                                  \
 108         do {                                                            \
 109                 char ip_vs_dbg_buf[160];                                \
 110                 int ip_vs_dbg_idx = 0;                                  \
 111                 if (level <= ip_vs_get_debug_level())                   \
 112                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 113         } while (0)
 114 #define IP_VS_ERR_BUF(msg...)                                           \
 115         do {                                                            \
 116                 char ip_vs_dbg_buf[160];                                \
 117                 int ip_vs_dbg_idx = 0;                                  \
 118                 pr_err(msg);                                            \
 119         } while (0)
 120
 121 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
 122 #define IP_VS_DBG_ADDR(af, addr)                                        \
 123         ip_vs_dbg_addr(af, ip_vs_dbg_buf,                               \
 124                        sizeof(ip_vs_dbg_buf), addr,                     \
 125                        &ip_vs_dbg_idx)
 126
 127 #define IP_VS_DBG(level, msg, ...)                                      \
 128         do {                                                            \
 129                 if (level <= ip_vs_get_debug_level())                   \
 130                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 131         } while (0)
 132 #define IP_VS_DBG_RL(msg, ...)                                          \
 133         do {                                                            \
 134                 if (net_ratelimit())                                    \
 135                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 136         } while (0)
 137 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)                         \
 138         do {                                                            \
 139                 if (level <= ip_vs_get_debug_level())                   \
 140                         pp->debug_packet(pp, skb, ofs, msg);            \
 141         } while (0)
 142 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)                      \
 143         do {                                                            \
 144                 if (level <= ip_vs_get_debug_level() &&                 \
 145                     net_ratelimit())                                    \
 146                         pp->debug_packet(pp, skb, ofs, msg);            \
 147         } while (0)
 148 #else   /* NO DEBUGGING at ALL */
 149 #define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
 150 #define IP_VS_ERR_BUF(msg...)  do {} while (0)
 151 #define IP_VS_DBG(level, msg...)  do {} while (0)
 152 #define IP_VS_DBG_RL(msg...)  do {} while (0)
 153 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)         do {} while (0)
 154 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)      do {} while (0)
 155 #endif
 156
 157 #define IP_VS_BUG() BUG()
 158 #define IP_VS_ERR_RL(msg, ...)                                          \
 159         do {                                                            \
 160                 if (net_ratelimit())                                    \
 161                         pr_err(msg, ##__VA_ARGS__);                     \
 162         } while (0)
 163
 164 #ifdef CONFIG_IP_VS_DEBUG
 165 #define EnterFunction(level)                                            \
 166         do {                                                            \
 167                 if (level <= ip_vs_get_debug_level())                   \
 168                         printk(KERN_DEBUG                               \
 169                                pr_fmt("Enter: %s, %s line %i\n"),       \
 170                                __func__, __FILE__, __LINE__);           \
 171         } while (0)
 172 #define LeaveFunction(level)                                            \
 173         do {                                                            \
 174                 if (level <= ip_vs_get_debug_level())                   \
 175                         printk(KERN_DEBUG                               \
 176                                pr_fmt("Leave: %s, %s line %i\n"),       \
 177                                __func__, __FILE__, __LINE__);           \
 178         } while (0)
 179 #else
 180 #define EnterFunction(level)   do {} while (0)
 181 #define LeaveFunction(level)   do {} while (0)
 182 #endif
 183
 184 #define IP_VS_WAIT_WHILE(expr)  while (expr) { cpu_relax(); }
 185
 186
 187 /*
 188  *      The port number of FTP service (in network order).
 189  */
 190 #define FTPPORT  cpu_to_be16(21)
 191 #define FTPDATA  cpu_to_be16(20)
 192
 193 /*
 194  *      TCP State Values
 195  */
 196 enum {
 197         IP_VS_TCP_S_NONE = 0,
 198         IP_VS_TCP_S_ESTABLISHED,
 199         IP_VS_TCP_S_SYN_SENT,
 200         IP_VS_TCP_S_SYN_RECV,
 201         IP_VS_TCP_S_FIN_WAIT,
 202         IP_VS_TCP_S_TIME_WAIT,
 203         IP_VS_TCP_S_CLOSE,
 204         IP_VS_TCP_S_CLOSE_WAIT,
 205         IP_VS_TCP_S_LAST_ACK,
 206         IP_VS_TCP_S_LISTEN,
 207         IP_VS_TCP_S_SYNACK,
 208         IP_VS_TCP_S_LAST
 209 };
 210
 211 /*
 212  *      UDP State Values
 213  */
 214 enum {
 215         IP_VS_UDP_S_NORMAL,
 216         IP_VS_UDP_S_LAST,
 217 };
 218
 219 /*
 220  *      ICMP State Values
 221  */
 222 enum {
 223         IP_VS_ICMP_S_NORMAL,
 224         IP_VS_ICMP_S_LAST,
 225 };
 226
 227 /*
 228  *      SCTP State Values
 229  */
 230 enum ip_vs_sctp_states {
 231         IP_VS_SCTP_S_NONE,
 232         IP_VS_SCTP_S_INIT_CLI,
 233         IP_VS_SCTP_S_INIT_SER,
 234         IP_VS_SCTP_S_INIT_ACK_CLI,
 235         IP_VS_SCTP_S_INIT_ACK_SER,
 236         IP_VS_SCTP_S_ECHO_CLI,
 237         IP_VS_SCTP_S_ECHO_SER,
 238         IP_VS_SCTP_S_ESTABLISHED,
 239         IP_VS_SCTP_S_SHUT_CLI,
 240         IP_VS_SCTP_S_SHUT_SER,
 241         IP_VS_SCTP_S_SHUT_ACK_CLI,
 242         IP_VS_SCTP_S_SHUT_ACK_SER,
 243         IP_VS_SCTP_S_CLOSED,
 244         IP_VS_SCTP_S_LAST
 245 };
 246
 247 /*
 248  *      Delta sequence info structure
 249  *      Each ip_vs_conn has 2 (output AND input seq. changes).
 250  *      Only used in the VS/NAT.
 251  */
 252 struct ip_vs_seq {
 253         __u32                   init_seq;       /* Add delta from this seq */
 254         __u32                   delta;          /* Delta in sequence numbers */
 255         __u32                   previous_delta; /* Delta in sequence numbers
 256                                                    before last resized pkt */
 257 };
 258
 259
 260 /*
 261  *      IPVS statistics objects
 262  */
 263 struct ip_vs_estimator {
 264         struct list_head        list;
 265
 266         u64                     last_inbytes;
 267         u64                     last_outbytes;
 268         u32                     last_conns;
 269         u32                     last_inpkts;
 270         u32                     last_outpkts;
 271
 272         u32                     cps;
 273         u32                     inpps;
 274         u32                     outpps;
 275         u32                     inbps;
 276         u32                     outbps;
 277 };
 278
 279 struct ip_vs_stats {
 280         struct ip_vs_stats_user ustats;         /* statistics */
 281         struct ip_vs_estimator  est;            /* estimator */
 282
 283         spinlock_t              lock;           /* spin lock */
 284 };
 285
 286 struct dst_entry;
 287 struct iphdr;
 288 struct ip_vs_conn;
 289 struct ip_vs_app;
 290 struct sk_buff;
 291
 292 struct ip_vs_protocol {
 293         struct ip_vs_protocol   *next;
 294         char                    *name;
 295         u16                     protocol;
 296         u16                     num_states;
 297         int                     dont_defrag;
 298         atomic_t                appcnt;         /* counter of proto app incs */
 299         int                     *timeout_table; /* protocol timeout table */
 300
 301         void (*init)(struct ip_vs_protocol *pp);
 302
 303         void (*exit)(struct ip_vs_protocol *pp);
 304
 305         int (*conn_schedule)(int af, struct sk_buff *skb,
 306                              struct ip_vs_protocol *pp,
 307                              int *verdict, struct ip_vs_conn **cpp);
 308
 309         struct ip_vs_conn *
 310         (*conn_in_get)(int af,
 311                        const struct sk_buff *skb,
 312                        struct ip_vs_protocol *pp,
 313                        const struct ip_vs_iphdr *iph,
 314                        unsigned int proto_off,
 315                        int inverse);
 316
 317         struct ip_vs_conn *
 318         (*conn_out_get)(int af,
 319                         const struct sk_buff *skb,
 320                         struct ip_vs_protocol *pp,
 321                         const struct ip_vs_iphdr *iph,
 322                         unsigned int proto_off,
 323                         int inverse);
 324
 325         int (*snat_handler)(struct sk_buff *skb,
 326                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 327
 328         int (*dnat_handler)(struct sk_buff *skb,
 329                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 330
 331         int (*csum_check)(int af, struct sk_buff *skb,
 332                           struct ip_vs_protocol *pp);
 333
 334         const char *(*state_name)(int state);
 335
 336         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 337                                 const struct sk_buff *skb,
 338                                 struct ip_vs_protocol *pp);
 339
 340         int (*register_app)(struct ip_vs_app *inc);
 341
 342         void (*unregister_app)(struct ip_vs_app *inc);
 343
 344         int (*app_conn_bind)(struct ip_vs_conn *cp);
 345
 346         void (*debug_packet)(struct ip_vs_protocol *pp,
 347                              const struct sk_buff *skb,
 348                              int offset,
 349                              const char *msg);
 350
 351         void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
 352
 353         int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
 354 };
 355
 356 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
 357
 358 /*
 359  *      IP_VS structure allocated for each dynamically scheduled connection
 360  */
 361 struct ip_vs_conn {
 362         struct list_head        c_list;         /* hashed list heads */
 363
 364         /* Protocol, addresses and port numbers */
 365         u16                      af;            /* address family */
 366         union nf_inet_addr       caddr;          /* client address */
 367         union nf_inet_addr       vaddr;          /* virtual address */
 368         union nf_inet_addr       daddr;          /* destination address */
 369         __be16                   cport;
 370         __be16                   vport;
 371         __be16                   dport;
 372         __u16                   protocol;       /* Which protocol (TCP/UDP) */
 373
 374         /* counter and timer */
 375         atomic_t                refcnt;         /* reference count */
 376         struct timer_list       timer;          /* Expiration timer */
 377         volatile unsigned long  timeout;        /* timeout */
 378
 379         /* Flags and state transition */
 380         spinlock_t              lock;           /* lock for state transition */
 381         volatile __u16          flags;          /* status flags */
 382         volatile __u16          state;          /* state info */
 383         volatile __u16          old_state;      /* old state, to be used for
 384                                                  * state transition triggerd
 385                                                  * synchronization
 386                                                  */
 387
 388         /* Control members */
 389         struct ip_vs_conn       *control;       /* Master control connection */
 390         atomic_t                n_control;      /* Number of controlled ones */
 391         struct ip_vs_dest       *dest;          /* real server */
 392         atomic_t                in_pkts;        /* incoming packet counter */
 393
 394         /* packet transmitter for different forwarding methods.  If it
 395            mangles the packet, it must return NF_DROP or better NF_STOLEN,
 396            otherwise this must be changed to a sk_buff **.
 397          */
 398         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
 399                            struct ip_vs_protocol *pp);
 400
 401         /* Note: we can group the following members into a structure,
 402            in order to save more space, and the following members are
 403            only used in VS/NAT anyway */
 404         struct ip_vs_app        *app;           /* bound ip_vs_app object */
 405         void                    *app_data;      /* Application private data */
 406         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
 407         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
 408 };
 409
 410
 411 /*
 412  *      Extended internal versions of struct ip_vs_service_user and
 413  *      ip_vs_dest_user for IPv6 support.
 414  *
 415  *      We need these to conveniently pass around service and destination
 416  *      options, but unfortunately, we also need to keep the old definitions to
 417  *      maintain userspace backwards compatibility for the setsockopt interface.
 418  */
 419 struct ip_vs_service_user_kern {
 420         /* virtual service addresses */
 421         u16                     af;
 422         u16                     protocol;
 423         union nf_inet_addr      addr;           /* virtual ip address */
 424         u16                     port;
 425         u32                     fwmark;         /* firwall mark of service */
 426
 427         /* virtual service options */
 428         char                    *sched_name;
 429         unsigned                flags;          /* virtual service flags */
 430         unsigned                timeout;        /* persistent timeout in sec */
 431         u32                     netmask;        /* persistent netmask */
 432 };
 433
 434
 435 struct ip_vs_dest_user_kern {
 436         /* destination server address */
 437         union nf_inet_addr      addr;
 438         u16                     port;
 439
 440         /* real server options */
 441         unsigned                conn_flags;     /* connection flags */
 442         int                     weight;         /* destination weight */
 443
 444         /* thresholds for active connections */
 445         u32                     u_threshold;    /* upper threshold */
 446         u32                     l_threshold;    /* lower threshold */
 447 };
 448
 449
 450 /*
 451  *      The information about the virtual service offered to the net
 452  *      and the forwarding entries
 453  */
 454 struct ip_vs_service {
 455         struct list_head        s_list;   /* for normal service table */
 456         struct list_head        f_list;   /* for fwmark-based service table */
 457         atomic_t                refcnt;   /* reference counter */
 458         atomic_t                usecnt;   /* use counter */
 459
 460         u16                     af;       /* address family */
 461         __u16                   protocol; /* which protocol (TCP/UDP) */
 462         union nf_inet_addr      addr;     /* IP address for virtual service */
 463         __be16                  port;     /* port number for the service */
 464         __u32                   fwmark;   /* firewall mark of the service */
 465         unsigned                flags;    /* service status flags */
 466         unsigned                timeout;  /* persistent timeout in ticks */
 467         __be32                  netmask;  /* grouping granularity */
 468
 469         struct list_head        destinations;  /* real server d-linked list */
 470         __u32                   num_dests;     /* number of servers */
 471         struct ip_vs_stats      stats;         /* statistics for the service */
 472         struct ip_vs_app        *inc;     /* bind conns to this app inc */
 473
 474         /* for scheduling */
 475         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
 476         rwlock_t                sched_lock;    /* lock sched_data */
 477         void                    *sched_data;   /* scheduler application data */
 478 };
 479
 480
 481 /*
 482  *      The real server destination forwarding entry
 483  *      with ip address, port number, and so on.
 484  */
 485 struct ip_vs_dest {
 486         struct list_head        n_list;   /* for the dests in the service */
 487         struct list_head        d_list;   /* for table with all the dests */
 488
 489         u16                     af;             /* address family */
 490         union nf_inet_addr      addr;           /* IP address of the server */
 491         __be16                  port;           /* port number of the server */
 492         volatile unsigned       flags;          /* dest status flags */
 493         atomic_t                conn_flags;     /* flags to copy to conn */
 494         atomic_t                weight;         /* server weight */
 495
 496         atomic_t                refcnt;         /* reference counter */
 497         struct ip_vs_stats      stats;          /* statistics */
 498
 499         /* connection counters and thresholds */
 500         atomic_t                activeconns;    /* active connections */
 501         atomic_t                inactconns;     /* inactive connections */
 502         atomic_t                persistconns;   /* persistent connections */
 503         __u32                   u_threshold;    /* upper threshold */
 504         __u32                   l_threshold;    /* lower threshold */
 505
 506         /* for destination cache */
 507         spinlock_t              dst_lock;       /* lock of dst_cache */
 508         struct dst_entry        *dst_cache;     /* destination cache entry */
 509         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
 510
 511         /* for virtual service */
 512         struct ip_vs_service    *svc;           /* service it belongs to */
 513         __u16                   protocol;       /* which protocol (TCP/UDP) */
 514         union nf_inet_addr      vaddr;          /* virtual IP address */
 515         __be16                  vport;          /* virtual port number */
 516         __u32                   vfwmark;        /* firewall mark of service */
 517 };
 518
 519
 520 /*
 521  *      The scheduler object
 522  */
 523 struct ip_vs_scheduler {
 524         struct list_head        n_list;         /* d-linked list head */
 525         char                    *name;          /* scheduler name */
 526         atomic_t                refcnt;         /* reference counter */
 527         struct module           *module;        /* THIS_MODULE/NULL */
 528
 529         /* scheduler initializing service */
 530         int (*init_service)(struct ip_vs_service *svc);
 531         /* scheduling service finish */
 532         int (*done_service)(struct ip_vs_service *svc);
 533         /* scheduler updating service */
 534         int (*update_service)(struct ip_vs_service *svc);
 535
 536         /* selecting a server from the given service */
 537         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 538                                        const struct sk_buff *skb);
 539 };
 540
 541
 542 /*
 543  *      The application module object (a.k.a. app incarnation)
 544  */
 545 struct ip_vs_app {
 546         struct list_head        a_list;         /* member in app list */
 547         int                     type;           /* IP_VS_APP_TYPE_xxx */
 548         char                    *name;          /* application module name */
 549         __u16                   protocol;
 550         struct module           *module;        /* THIS_MODULE/NULL */
 551         struct list_head        incs_list;      /* list of incarnations */
 552
 553         /* members for application incarnations */
 554         struct list_head        p_list;         /* member in proto app list */
 555         struct ip_vs_app        *app;           /* its real application */
 556         __be16                  port;           /* port number in net order */
 557         atomic_t                usecnt;         /* usage counter */
 558
 559         /* output hook: return false if can't linearize. diff set for TCP.  */
 560         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 561                        struct sk_buff *, int *diff);
 562
 563         /* input hook: return false if can't linearize. diff set for TCP. */
 564         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 565                       struct sk_buff *, int *diff);
 566
 567         /* ip_vs_app initializer */
 568         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 569
 570         /* ip_vs_app finish */
 571         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 572
 573
 574         /* not used now */
 575         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 576                          struct ip_vs_protocol *);
 577
 578         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 579
 580         int *                   timeout_table;
 581         int *                   timeouts;
 582         int                     timeouts_size;
 583
 584         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 585                              int *verdict, struct ip_vs_conn **cpp);
 586
 587         struct ip_vs_conn *
 588         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 589                        const struct iphdr *iph, unsigned int proto_off,
 590                        int inverse);
 591
 592         struct ip_vs_conn *
 593         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 594                         const struct iphdr *iph, unsigned int proto_off,
 595                         int inverse);
 596
 597         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 598                                 const struct sk_buff *skb,
 599                                 struct ip_vs_app *app);
 600
 601         void (*timeout_change)(struct ip_vs_app *app, int flags);
 602 };
 603
 604
 605 /*
 606  *      IPVS core functions
 607  *      (from ip_vs_core.c)
 608  */
 609 extern const char *ip_vs_proto_name(unsigned proto);
 610 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
 611 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
 612
 613 #define IP_VS_APP_TYPE_FTP      1
 614
 615 /*
 616  *     ip_vs_conn handling functions
 617  *     (from ip_vs_conn.c)
 618  */
 619
 620 enum {
 621         IP_VS_DIR_INPUT = 0,
 622         IP_VS_DIR_OUTPUT,
 623         IP_VS_DIR_INPUT_ONLY,
 624         IP_VS_DIR_LAST,
 625 };
 626
 627 extern struct ip_vs_conn *ip_vs_conn_in_get
 628 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
 629  const union nf_inet_addr *d_addr, __be16 d_port);
 630
 631 extern struct ip_vs_conn *ip_vs_ct_in_get
 632 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
 633  const union nf_inet_addr *d_addr, __be16 d_port);
 634
 635 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
 636                                             struct ip_vs_protocol *pp,
 637                                             const struct ip_vs_iphdr *iph,
 638                                             unsigned int proto_off,
 639                                             int inverse);
 640
 641 extern struct ip_vs_conn *ip_vs_conn_out_get
 642 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
 643  const union nf_inet_addr *d_addr, __be16 d_port);
 644
 645 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
 646                                              struct ip_vs_protocol *pp,
 647                                              const struct ip_vs_iphdr *iph,
 648                                              unsigned int proto_off,
 649                                              int inverse);
 650
 651 /* put back the conn without restarting its timer */
 652 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
 653 {
 654         atomic_dec(&cp->refcnt);
 655 }
 656 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
 657 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
 658
 659 extern struct ip_vs_conn *
 660 ip_vs_conn_new(int af, int proto, const union nf_inet_addr *caddr, __be16 cport,
 661                const union nf_inet_addr *vaddr, __be16 vport,
 662                const union nf_inet_addr *daddr, __be16 dport, unsigned flags,
 663                struct ip_vs_dest *dest);
 664 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
 665
 666 extern const char * ip_vs_state_name(__u16 proto, int state);
 667
 668 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
 669 extern int ip_vs_check_template(struct ip_vs_conn *ct);
 670 extern void ip_vs_random_dropentry(void);
 671 extern int ip_vs_conn_init(void);
 672 extern void ip_vs_conn_cleanup(void);
 673
 674 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
 675 {
 676         struct ip_vs_conn *ctl_cp = cp->control;
 677         if (!ctl_cp) {
 678                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
 679                               "%s:%d to %s:%d\n",
 680                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 681                               ntohs(cp->cport),
 682                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 683                               ntohs(cp->vport));
 684
 685                 return;
 686         }
 687
 688         IP_VS_DBG_BUF(7, "DELeting control for: "
 689                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
 690                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 691                       ntohs(cp->cport),
 692                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
 693                       ntohs(ctl_cp->cport));
 694
 695         cp->control = NULL;
 696         if (atomic_read(&ctl_cp->n_control) == 0) {
 697                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
 698                               "%s:%d to %s:%d\n",
 699                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 700                               ntohs(cp->cport),
 701                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 702                               ntohs(cp->vport));
 703
 704                 return;
 705         }
 706         atomic_dec(&ctl_cp->n_control);
 707 }
 708
 709 static inline void
 710 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
 711 {
 712         if (cp->control) {
 713                 IP_VS_ERR_BUF("request control ADD for already controlled: "
 714                               "%s:%d to %s:%d\n",
 715                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 716                               ntohs(cp->cport),
 717                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 718                               ntohs(cp->vport));
 719
 720                 ip_vs_control_del(cp);
 721         }
 722
 723         IP_VS_DBG_BUF(7, "ADDing control for: "
 724                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
 725                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 726                       ntohs(cp->cport),
 727                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
 728                       ntohs(ctl_cp->cport));
 729
 730         cp->control = ctl_cp;
 731         atomic_inc(&ctl_cp->n_control);
 732 }
 733
 734
 735 /*
 736  *      IPVS application functions
 737  *      (from ip_vs_app.c)
 738  */
 739 #define IP_VS_APP_MAX_PORTS  8
 740 extern int register_ip_vs_app(struct ip_vs_app *app);
 741 extern void unregister_ip_vs_app(struct ip_vs_app *app);
 742 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 743 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
 744 extern int
 745 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
 746 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
 747 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
 748
 749 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
 750 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
 751 extern int ip_vs_app_init(void);
 752 extern void ip_vs_app_cleanup(void);
 753
 754
 755 /*
 756  *      IPVS protocol functions (from ip_vs_proto.c)
 757  */
 758 extern int ip_vs_protocol_init(void);
 759 extern void ip_vs_protocol_cleanup(void);
 760 extern void ip_vs_protocol_timeout_change(int flags);
 761 extern int *ip_vs_create_timeout_table(int *table, int size);
 762 extern int
 763 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
 764                         const char *name, int to);
 765 extern void
 766 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
 767                           int offset, const char *msg);
 768
 769 extern struct ip_vs_protocol ip_vs_protocol_tcp;
 770 extern struct ip_vs_protocol ip_vs_protocol_udp;
 771 extern struct ip_vs_protocol ip_vs_protocol_icmp;
 772 extern struct ip_vs_protocol ip_vs_protocol_esp;
 773 extern struct ip_vs_protocol ip_vs_protocol_ah;
 774 extern struct ip_vs_protocol ip_vs_protocol_sctp;
 775
 776 /*
 777  *      Registering/unregistering scheduler functions
 778  *      (from ip_vs_sched.c)
 779  */
 780 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
 781 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
 782 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
 783                                 struct ip_vs_scheduler *scheduler);
 784 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
 785 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
 786 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
 787 extern struct ip_vs_conn *
 788 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
 789 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
 790                         struct ip_vs_protocol *pp);
 791
 792
 793 /*
 794  *      IPVS control data and functions (from ip_vs_ctl.c)
 795  */
 796 extern int sysctl_ip_vs_cache_bypass;
 797 extern int sysctl_ip_vs_expire_nodest_conn;
 798 extern int sysctl_ip_vs_expire_quiescent_template;
 799 extern int sysctl_ip_vs_sync_threshold[2];
 800 extern int sysctl_ip_vs_nat_icmp_send;
 801 extern struct ip_vs_stats ip_vs_stats;
 802 extern const struct ctl_path net_vs_ctl_path[];
 803
 804 extern struct ip_vs_service *
 805 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
 806                   const union nf_inet_addr *vaddr, __be16 vport);
 807
 808 static inline void ip_vs_service_put(struct ip_vs_service *svc)
 809 {
 810         atomic_dec(&svc->usecnt);
 811 }
 812
 813 extern struct ip_vs_dest *
 814 ip_vs_lookup_real_service(int af, __u16 protocol,
 815                           const union nf_inet_addr *daddr, __be16 dport);
 816
 817 extern int ip_vs_use_count_inc(void);
 818 extern void ip_vs_use_count_dec(void);
 819 extern int ip_vs_control_init(void);
 820 extern void ip_vs_control_cleanup(void);
 821 extern struct ip_vs_dest *
 822 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
 823                 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol);
 824 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
 825
 826
 827 /*
 828  *      IPVS sync daemon data and function prototypes
 829  *      (from ip_vs_sync.c)
 830  */
 831 extern volatile int ip_vs_sync_state;
 832 extern volatile int ip_vs_master_syncid;
 833 extern volatile int ip_vs_backup_syncid;
 834 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 835 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 836 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
 837 extern int stop_sync_thread(int state);
 838 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
 839
 840
 841 /*
 842  *      IPVS rate estimator prototypes (from ip_vs_est.c)
 843  */
 844 extern int ip_vs_estimator_init(void);
 845 extern void ip_vs_estimator_cleanup(void);
 846 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
 847 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
 848 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
 849
 850 /*
 851  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
 852  */
 853 extern int ip_vs_null_xmit
 854 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 855 extern int ip_vs_bypass_xmit
 856 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 857 extern int ip_vs_nat_xmit
 858 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 859 extern int ip_vs_tunnel_xmit
 860 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 861 extern int ip_vs_dr_xmit
 862 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 863 extern int ip_vs_icmp_xmit
 864 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
 865 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
 866
 867 #ifdef CONFIG_IP_VS_IPV6
 868 extern int ip_vs_bypass_xmit_v6
 869 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 870 extern int ip_vs_nat_xmit_v6
 871 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 872 extern int ip_vs_tunnel_xmit_v6
 873 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 874 extern int ip_vs_dr_xmit_v6
 875 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 876 extern int ip_vs_icmp_xmit_v6
 877 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
 878  int offset);
 879 #endif
 880
 881 /*
 882  *      This is a simple mechanism to ignore packets when
 883  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
 884  *      we start to drop 1/rate of the packets
 885  */
 886 extern int ip_vs_drop_rate;
 887 extern int ip_vs_drop_counter;
 888
 889 static __inline__ int ip_vs_todrop(void)
 890 {
 891         if (!ip_vs_drop_rate) return 0;
 892         if (--ip_vs_drop_counter > 0) return 0;
 893         ip_vs_drop_counter = ip_vs_drop_rate;
 894         return 1;
 895 }
 896
 897 /*
 898  *      ip_vs_fwd_tag returns the forwarding tag of the connection
 899  */
 900 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
 901
 902 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
 903 {
 904         char fwd;
 905
 906         switch (IP_VS_FWD_METHOD(cp)) {
 907         case IP_VS_CONN_F_MASQ:
 908                 fwd = 'M'; break;
 909         case IP_VS_CONN_F_LOCALNODE:
 910                 fwd = 'L'; break;
 911         case IP_VS_CONN_F_TUNNEL:
 912                 fwd = 'T'; break;
 913         case IP_VS_CONN_F_DROUTE:
 914                 fwd = 'R'; break;
 915         case IP_VS_CONN_F_BYPASS:
 916                 fwd = 'B'; break;
 917         default:
 918                 fwd = '?'; break;
 919         }
 920         return fwd;
 921 }
 922
 923 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
 924                            struct ip_vs_conn *cp, int dir);
 925
 926 #ifdef CONFIG_IP_VS_IPV6
 927 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
 928                               struct ip_vs_conn *cp, int dir);
 929 #endif
 930
 931 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
 932
 933 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
 934 {
 935         __be32 diff[2] = { ~old, new };
 936
 937         return csum_partial(diff, sizeof(diff), oldsum);
 938 }
 939
 940 #ifdef CONFIG_IP_VS_IPV6
 941 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
 942                                         __wsum oldsum)
 943 {
 944         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
 945                             new[3],  new[2],  new[1],  new[0] };
 946
 947         return csum_partial(diff, sizeof(diff), oldsum);
 948 }
 949 #endif
 950
 951 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
 952 {
 953         __be16 diff[2] = { ~old, new };
 954
 955         return csum_partial(diff, sizeof(diff), oldsum);
 956 }
 957
 958 #endif /* __KERNEL__ */
 959
 960 #endif  /* _NET_IP_VS_H */