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 #ifdef CONFIG_IP_VS_NFCT
  29 #include <net/netfilter/nf_conntrack.h>
  30 #endif
  31
  32 /* Connections' size value needed by ip_vs_ctl.c */
  33 extern int ip_vs_conn_tab_size;
  34
  35
  36 struct ip_vs_iphdr {
  37         int len;
  38         __u8 protocol;
  39         union nf_inet_addr saddr;
  40         union nf_inet_addr daddr;
  41 };
  42
  43 static inline void
  44 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
  45 {
  46 #ifdef CONFIG_IP_VS_IPV6
  47         if (af == AF_INET6) {
  48                 const struct ipv6hdr *iph = nh;
  49                 iphdr->len = sizeof(struct ipv6hdr);
  50                 iphdr->protocol = iph->nexthdr;
  51                 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
  52                 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
  53         } else
  54 #endif
  55         {
  56                 const struct iphdr *iph = nh;
  57                 iphdr->len = iph->ihl * 4;
  58                 iphdr->protocol = iph->protocol;
  59                 iphdr->saddr.ip = iph->saddr;
  60                 iphdr->daddr.ip = iph->daddr;
  61         }
  62 }
  63
  64 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
  65                                    const union nf_inet_addr *src)
  66 {
  67 #ifdef CONFIG_IP_VS_IPV6
  68         if (af == AF_INET6)
  69                 ipv6_addr_copy(&dst->in6, &src->in6);
  70         else
  71 #endif
  72         dst->ip = src->ip;
  73 }
  74
  75 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
  76                                    const union nf_inet_addr *b)
  77 {
  78 #ifdef CONFIG_IP_VS_IPV6
  79         if (af == AF_INET6)
  80                 return ipv6_addr_equal(&a->in6, &b->in6);
  81 #endif
  82         return a->ip == b->ip;
  83 }
  84
  85 #ifdef CONFIG_IP_VS_DEBUG
  86 #include <linux/net.h>
  87
  88 extern int ip_vs_get_debug_level(void);
  89
  90 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
  91                                          const union nf_inet_addr *addr,
  92                                          int *idx)
  93 {
  94         int len;
  95 #ifdef CONFIG_IP_VS_IPV6
  96         if (af == AF_INET6)
  97                 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
  98                                &addr->in6) + 1;
  99         else
 100 #endif
 101                 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
 102                                &addr->ip) + 1;
 103
 104         *idx += len;
 105         BUG_ON(*idx > buf_len + 1);
 106         return &buf[*idx - len];
 107 }
 108
 109 #define IP_VS_DBG_BUF(level, msg, ...)                                  \
 110         do {                                                            \
 111                 char ip_vs_dbg_buf[160];                                \
 112                 int ip_vs_dbg_idx = 0;                                  \
 113                 if (level <= ip_vs_get_debug_level())                   \
 114                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 115         } while (0)
 116 #define IP_VS_ERR_BUF(msg...)                                           \
 117         do {                                                            \
 118                 char ip_vs_dbg_buf[160];                                \
 119                 int ip_vs_dbg_idx = 0;                                  \
 120                 pr_err(msg);                                            \
 121         } while (0)
 122
 123 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
 124 #define IP_VS_DBG_ADDR(af, addr)                                        \
 125         ip_vs_dbg_addr(af, ip_vs_dbg_buf,                               \
 126                        sizeof(ip_vs_dbg_buf), addr,                     \
 127                        &ip_vs_dbg_idx)
 128
 129 #define IP_VS_DBG(level, msg, ...)                                      \
 130         do {                                                            \
 131                 if (level <= ip_vs_get_debug_level())                   \
 132                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 133         } while (0)
 134 #define IP_VS_DBG_RL(msg, ...)                                          \
 135         do {                                                            \
 136                 if (net_ratelimit())                                    \
 137                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 138         } while (0)
 139 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)                         \
 140         do {                                                            \
 141                 if (level <= ip_vs_get_debug_level())                   \
 142                         pp->debug_packet(pp, skb, ofs, msg);            \
 143         } while (0)
 144 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)                      \
 145         do {                                                            \
 146                 if (level <= ip_vs_get_debug_level() &&                 \
 147                     net_ratelimit())                                    \
 148                         pp->debug_packet(pp, skb, ofs, msg);            \
 149         } while (0)
 150 #else   /* NO DEBUGGING at ALL */
 151 #define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
 152 #define IP_VS_ERR_BUF(msg...)  do {} while (0)
 153 #define IP_VS_DBG(level, msg...)  do {} while (0)
 154 #define IP_VS_DBG_RL(msg...)  do {} while (0)
 155 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)         do {} while (0)
 156 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)      do {} while (0)
 157 #endif
 158
 159 #define IP_VS_BUG() BUG()
 160 #define IP_VS_ERR_RL(msg, ...)                                          \
 161         do {                                                            \
 162                 if (net_ratelimit())                                    \
 163                         pr_err(msg, ##__VA_ARGS__);                     \
 164         } while (0)
 165
 166 #ifdef CONFIG_IP_VS_DEBUG
 167 #define EnterFunction(level)                                            \
 168         do {                                                            \
 169                 if (level <= ip_vs_get_debug_level())                   \
 170                         printk(KERN_DEBUG                               \
 171                                pr_fmt("Enter: %s, %s line %i\n"),       \
 172                                __func__, __FILE__, __LINE__);           \
 173         } while (0)
 174 #define LeaveFunction(level)                                            \
 175         do {                                                            \
 176                 if (level <= ip_vs_get_debug_level())                   \
 177                         printk(KERN_DEBUG                               \
 178                                pr_fmt("Leave: %s, %s line %i\n"),       \
 179                                __func__, __FILE__, __LINE__);           \
 180         } while (0)
 181 #else
 182 #define EnterFunction(level)   do {} while (0)
 183 #define LeaveFunction(level)   do {} while (0)
 184 #endif
 185
 186 #define IP_VS_WAIT_WHILE(expr)  while (expr) { cpu_relax(); }
 187
 188
 189 /*
 190  *      The port number of FTP service (in network order).
 191  */
 192 #define FTPPORT  cpu_to_be16(21)
 193 #define FTPDATA  cpu_to_be16(20)
 194
 195 /*
 196  *      TCP State Values
 197  */
 198 enum {
 199         IP_VS_TCP_S_NONE = 0,
 200         IP_VS_TCP_S_ESTABLISHED,
 201         IP_VS_TCP_S_SYN_SENT,
 202         IP_VS_TCP_S_SYN_RECV,
 203         IP_VS_TCP_S_FIN_WAIT,
 204         IP_VS_TCP_S_TIME_WAIT,
 205         IP_VS_TCP_S_CLOSE,
 206         IP_VS_TCP_S_CLOSE_WAIT,
 207         IP_VS_TCP_S_LAST_ACK,
 208         IP_VS_TCP_S_LISTEN,
 209         IP_VS_TCP_S_SYNACK,
 210         IP_VS_TCP_S_LAST
 211 };
 212
 213 /*
 214  *      UDP State Values
 215  */
 216 enum {
 217         IP_VS_UDP_S_NORMAL,
 218         IP_VS_UDP_S_LAST,
 219 };
 220
 221 /*
 222  *      ICMP State Values
 223  */
 224 enum {
 225         IP_VS_ICMP_S_NORMAL,
 226         IP_VS_ICMP_S_LAST,
 227 };
 228
 229 /*
 230  *      SCTP State Values
 231  */
 232 enum ip_vs_sctp_states {
 233         IP_VS_SCTP_S_NONE,
 234         IP_VS_SCTP_S_INIT_CLI,
 235         IP_VS_SCTP_S_INIT_SER,
 236         IP_VS_SCTP_S_INIT_ACK_CLI,
 237         IP_VS_SCTP_S_INIT_ACK_SER,
 238         IP_VS_SCTP_S_ECHO_CLI,
 239         IP_VS_SCTP_S_ECHO_SER,
 240         IP_VS_SCTP_S_ESTABLISHED,
 241         IP_VS_SCTP_S_SHUT_CLI,
 242         IP_VS_SCTP_S_SHUT_SER,
 243         IP_VS_SCTP_S_SHUT_ACK_CLI,
 244         IP_VS_SCTP_S_SHUT_ACK_SER,
 245         IP_VS_SCTP_S_CLOSED,
 246         IP_VS_SCTP_S_LAST
 247 };
 248
 249 /*
 250  *      Delta sequence info structure
 251  *      Each ip_vs_conn has 2 (output AND input seq. changes).
 252  *      Only used in the VS/NAT.
 253  */
 254 struct ip_vs_seq {
 255         __u32                   init_seq;       /* Add delta from this seq */
 256         __u32                   delta;          /* Delta in sequence numbers */
 257         __u32                   previous_delta; /* Delta in sequence numbers
 258                                                    before last resized pkt */
 259 };
 260
 261
 262 /*
 263  *      IPVS statistics objects
 264  */
 265 struct ip_vs_estimator {
 266         struct list_head        list;
 267
 268         u64                     last_inbytes;
 269         u64                     last_outbytes;
 270         u32                     last_conns;
 271         u32                     last_inpkts;
 272         u32                     last_outpkts;
 273
 274         u32                     cps;
 275         u32                     inpps;
 276         u32                     outpps;
 277         u32                     inbps;
 278         u32                     outbps;
 279 };
 280
 281 struct ip_vs_stats {
 282         struct ip_vs_stats_user ustats;         /* statistics */
 283         struct ip_vs_estimator  est;            /* estimator */
 284
 285         spinlock_t              lock;           /* spin lock */
 286 };
 287
 288 struct dst_entry;
 289 struct iphdr;
 290 struct ip_vs_conn;
 291 struct ip_vs_app;
 292 struct sk_buff;
 293
 294 struct ip_vs_protocol {
 295         struct ip_vs_protocol   *next;
 296         char                    *name;
 297         u16                     protocol;
 298         u16                     num_states;
 299         int                     dont_defrag;
 300         atomic_t                appcnt;         /* counter of proto app incs */
 301         int                     *timeout_table; /* protocol timeout table */
 302
 303         void (*init)(struct ip_vs_protocol *pp);
 304
 305         void (*exit)(struct ip_vs_protocol *pp);
 306
 307         int (*conn_schedule)(int af, struct sk_buff *skb,
 308                              struct ip_vs_protocol *pp,
 309                              int *verdict, struct ip_vs_conn **cpp);
 310
 311         struct ip_vs_conn *
 312         (*conn_in_get)(int af,
 313                        const struct sk_buff *skb,
 314                        struct ip_vs_protocol *pp,
 315                        const struct ip_vs_iphdr *iph,
 316                        unsigned int proto_off,
 317                        int inverse);
 318
 319         struct ip_vs_conn *
 320         (*conn_out_get)(int af,
 321                         const struct sk_buff *skb,
 322                         struct ip_vs_protocol *pp,
 323                         const struct ip_vs_iphdr *iph,
 324                         unsigned int proto_off,
 325                         int inverse);
 326
 327         int (*snat_handler)(struct sk_buff *skb,
 328                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 329
 330         int (*dnat_handler)(struct sk_buff *skb,
 331                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 332
 333         int (*csum_check)(int af, struct sk_buff *skb,
 334                           struct ip_vs_protocol *pp);
 335
 336         const char *(*state_name)(int state);
 337
 338         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 339                                 const struct sk_buff *skb,
 340                                 struct ip_vs_protocol *pp);
 341
 342         int (*register_app)(struct ip_vs_app *inc);
 343
 344         void (*unregister_app)(struct ip_vs_app *inc);
 345
 346         int (*app_conn_bind)(struct ip_vs_conn *cp);
 347
 348         void (*debug_packet)(struct ip_vs_protocol *pp,
 349                              const struct sk_buff *skb,
 350                              int offset,
 351                              const char *msg);
 352
 353         void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
 354
 355         int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
 356 };
 357
 358 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
 359
 360 struct ip_vs_conn_param {
 361         const union nf_inet_addr        *caddr;
 362         const union nf_inet_addr        *vaddr;
 363         __be16                          cport;
 364         __be16                          vport;
 365         __u16                           protocol;
 366         u16                             af;
 367
 368         const struct ip_vs_pe           *pe;
 369         char                            *pe_data;
 370         __u8                            pe_data_len;
 371 };
 372
 373 /*
 374  *      IP_VS structure allocated for each dynamically scheduled connection
 375  */
 376 struct ip_vs_conn {
 377         struct list_head        c_list;         /* hashed list heads */
 378
 379         /* Protocol, addresses and port numbers */
 380         u16                      af;            /* address family */
 381         union nf_inet_addr       caddr;          /* client address */
 382         union nf_inet_addr       vaddr;          /* virtual address */
 383         union nf_inet_addr       daddr;          /* destination address */
 384         volatile __u32           flags;          /* status flags */
 385         __be16                   cport;
 386         __be16                   vport;
 387         __be16                   dport;
 388         __u16                   protocol;       /* Which protocol (TCP/UDP) */
 389
 390         /* counter and timer */
 391         atomic_t                refcnt;         /* reference count */
 392         struct timer_list       timer;          /* Expiration timer */
 393         volatile unsigned long  timeout;        /* timeout */
 394
 395         /* Flags and state transition */
 396         spinlock_t              lock;           /* lock for state transition */
 397         volatile __u16          state;          /* state info */
 398         volatile __u16          old_state;      /* old state, to be used for
 399                                                  * state transition triggerd
 400                                                  * synchronization
 401                                                  */
 402
 403         /* Control members */
 404         struct ip_vs_conn       *control;       /* Master control connection */
 405         atomic_t                n_control;      /* Number of controlled ones */
 406         struct ip_vs_dest       *dest;          /* real server */
 407         atomic_t                in_pkts;        /* incoming packet counter */
 408
 409         /* packet transmitter for different forwarding methods.  If it
 410            mangles the packet, it must return NF_DROP or better NF_STOLEN,
 411            otherwise this must be changed to a sk_buff **.
 412          */
 413         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
 414                            struct ip_vs_protocol *pp);
 415
 416         /* Note: we can group the following members into a structure,
 417            in order to save more space, and the following members are
 418            only used in VS/NAT anyway */
 419         struct ip_vs_app        *app;           /* bound ip_vs_app object */
 420         void                    *app_data;      /* Application private data */
 421         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
 422         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
 423
 424         char                    *pe_data;
 425         __u8                    pe_data_len;
 426 };
 427
 428
 429 /*
 430  *      Extended internal versions of struct ip_vs_service_user and
 431  *      ip_vs_dest_user for IPv6 support.
 432  *
 433  *      We need these to conveniently pass around service and destination
 434  *      options, but unfortunately, we also need to keep the old definitions to
 435  *      maintain userspace backwards compatibility for the setsockopt interface.
 436  */
 437 struct ip_vs_service_user_kern {
 438         /* virtual service addresses */
 439         u16                     af;
 440         u16                     protocol;
 441         union nf_inet_addr      addr;           /* virtual ip address */
 442         u16                     port;
 443         u32                     fwmark;         /* firwall mark of service */
 444
 445         /* virtual service options */
 446         char                    *sched_name;
 447         unsigned                flags;          /* virtual service flags */
 448         unsigned                timeout;        /* persistent timeout in sec */
 449         u32                     netmask;        /* persistent netmask */
 450 };
 451
 452
 453 struct ip_vs_dest_user_kern {
 454         /* destination server address */
 455         union nf_inet_addr      addr;
 456         u16                     port;
 457
 458         /* real server options */
 459         unsigned                conn_flags;     /* connection flags */
 460         int                     weight;         /* destination weight */
 461
 462         /* thresholds for active connections */
 463         u32                     u_threshold;    /* upper threshold */
 464         u32                     l_threshold;    /* lower threshold */
 465 };
 466
 467
 468 /*
 469  *      The information about the virtual service offered to the net
 470  *      and the forwarding entries
 471  */
 472 struct ip_vs_service {
 473         struct list_head        s_list;   /* for normal service table */
 474         struct list_head        f_list;   /* for fwmark-based service table */
 475         atomic_t                refcnt;   /* reference counter */
 476         atomic_t                usecnt;   /* use counter */
 477
 478         u16                     af;       /* address family */
 479         __u16                   protocol; /* which protocol (TCP/UDP) */
 480         union nf_inet_addr      addr;     /* IP address for virtual service */
 481         __be16                  port;     /* port number for the service */
 482         __u32                   fwmark;   /* firewall mark of the service */
 483         unsigned                flags;    /* service status flags */
 484         unsigned                timeout;  /* persistent timeout in ticks */
 485         __be32                  netmask;  /* grouping granularity */
 486
 487         struct list_head        destinations;  /* real server d-linked list */
 488         __u32                   num_dests;     /* number of servers */
 489         struct ip_vs_stats      stats;         /* statistics for the service */
 490         struct ip_vs_app        *inc;     /* bind conns to this app inc */
 491
 492         /* for scheduling */
 493         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
 494         rwlock_t                sched_lock;    /* lock sched_data */
 495         void                    *sched_data;   /* scheduler application data */
 496
 497         /* alternate persistence engine */
 498         struct ip_vs_pe         *pe;
 499 };
 500
 501
 502 /*
 503  *      The real server destination forwarding entry
 504  *      with ip address, port number, and so on.
 505  */
 506 struct ip_vs_dest {
 507         struct list_head        n_list;   /* for the dests in the service */
 508         struct list_head        d_list;   /* for table with all the dests */
 509
 510         u16                     af;             /* address family */
 511         union nf_inet_addr      addr;           /* IP address of the server */
 512         __be16                  port;           /* port number of the server */
 513         volatile unsigned       flags;          /* dest status flags */
 514         atomic_t                conn_flags;     /* flags to copy to conn */
 515         atomic_t                weight;         /* server weight */
 516
 517         atomic_t                refcnt;         /* reference counter */
 518         struct ip_vs_stats      stats;          /* statistics */
 519
 520         /* connection counters and thresholds */
 521         atomic_t                activeconns;    /* active connections */
 522         atomic_t                inactconns;     /* inactive connections */
 523         atomic_t                persistconns;   /* persistent connections */
 524         __u32                   u_threshold;    /* upper threshold */
 525         __u32                   l_threshold;    /* lower threshold */
 526
 527         /* for destination cache */
 528         spinlock_t              dst_lock;       /* lock of dst_cache */
 529         struct dst_entry        *dst_cache;     /* destination cache entry */
 530         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
 531
 532         /* for virtual service */
 533         struct ip_vs_service    *svc;           /* service it belongs to */
 534         __u16                   protocol;       /* which protocol (TCP/UDP) */
 535         union nf_inet_addr      vaddr;          /* virtual IP address */
 536         __be16                  vport;          /* virtual port number */
 537         __u32                   vfwmark;        /* firewall mark of service */
 538 };
 539
 540
 541 /*
 542  *      The scheduler object
 543  */
 544 struct ip_vs_scheduler {
 545         struct list_head        n_list;         /* d-linked list head */
 546         char                    *name;          /* scheduler name */
 547         atomic_t                refcnt;         /* reference counter */
 548         struct module           *module;        /* THIS_MODULE/NULL */
 549
 550         /* scheduler initializing service */
 551         int (*init_service)(struct ip_vs_service *svc);
 552         /* scheduling service finish */
 553         int (*done_service)(struct ip_vs_service *svc);
 554         /* scheduler updating service */
 555         int (*update_service)(struct ip_vs_service *svc);
 556
 557         /* selecting a server from the given service */
 558         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 559                                        const struct sk_buff *skb);
 560 };
 561
 562 /* The persistence engine object */
 563 struct ip_vs_pe {
 564         struct list_head        n_list;         /* d-linked list head */
 565         char                    *name;          /* scheduler name */
 566         atomic_t                refcnt;         /* reference counter */
 567         struct module           *module;        /* THIS_MODULE/NULL */
 568
 569         /* get the connection template, if any */
 570         int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
 571         bool (*ct_match)(const struct ip_vs_conn_param *p,
 572                          struct ip_vs_conn *ct);
 573         u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
 574                            bool inverse);
 575         int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
 576 };
 577
 578 /*
 579  *      The application module object (a.k.a. app incarnation)
 580  */
 581 struct ip_vs_app {
 582         struct list_head        a_list;         /* member in app list */
 583         int                     type;           /* IP_VS_APP_TYPE_xxx */
 584         char                    *name;          /* application module name */
 585         __u16                   protocol;
 586         struct module           *module;        /* THIS_MODULE/NULL */
 587         struct list_head        incs_list;      /* list of incarnations */
 588
 589         /* members for application incarnations */
 590         struct list_head        p_list;         /* member in proto app list */
 591         struct ip_vs_app        *app;           /* its real application */
 592         __be16                  port;           /* port number in net order */
 593         atomic_t                usecnt;         /* usage counter */
 594
 595         /* output hook: return false if can't linearize. diff set for TCP.  */
 596         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 597                        struct sk_buff *, int *diff);
 598
 599         /* input hook: return false if can't linearize. diff set for TCP. */
 600         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 601                       struct sk_buff *, int *diff);
 602
 603         /* ip_vs_app initializer */
 604         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 605
 606         /* ip_vs_app finish */
 607         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 608
 609
 610         /* not used now */
 611         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 612                          struct ip_vs_protocol *);
 613
 614         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 615
 616         int *                   timeout_table;
 617         int *                   timeouts;
 618         int                     timeouts_size;
 619
 620         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 621                              int *verdict, struct ip_vs_conn **cpp);
 622
 623         struct ip_vs_conn *
 624         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 625                        const struct iphdr *iph, unsigned int proto_off,
 626                        int inverse);
 627
 628         struct ip_vs_conn *
 629         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 630                         const struct iphdr *iph, unsigned int proto_off,
 631                         int inverse);
 632
 633         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 634                                 const struct sk_buff *skb,
 635                                 struct ip_vs_app *app);
 636
 637         void (*timeout_change)(struct ip_vs_app *app, int flags);
 638 };
 639
 640
 641 /*
 642  *      IPVS core functions
 643  *      (from ip_vs_core.c)
 644  */
 645 extern const char *ip_vs_proto_name(unsigned proto);
 646 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
 647 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
 648
 649 #define IP_VS_APP_TYPE_FTP      1
 650
 651 /*
 652  *     ip_vs_conn handling functions
 653  *     (from ip_vs_conn.c)
 654  */
 655
 656 enum {
 657         IP_VS_DIR_INPUT = 0,
 658         IP_VS_DIR_OUTPUT,
 659         IP_VS_DIR_INPUT_ONLY,
 660         IP_VS_DIR_LAST,
 661 };
 662
 663 static inline void ip_vs_conn_fill_param(int af, int protocol,
 664                                          const union nf_inet_addr *caddr,
 665                                          __be16 cport,
 666                                          const union nf_inet_addr *vaddr,
 667                                          __be16 vport,
 668                                          struct ip_vs_conn_param *p)
 669 {
 670         p->af = af;
 671         p->protocol = protocol;
 672         p->caddr = caddr;
 673         p->cport = cport;
 674         p->vaddr = vaddr;
 675         p->vport = vport;
 676         p->pe = NULL;
 677         p->pe_data = NULL;
 678 }
 679
 680 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
 681 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
 682
 683 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
 684                                             struct ip_vs_protocol *pp,
 685                                             const struct ip_vs_iphdr *iph,
 686                                             unsigned int proto_off,
 687                                             int inverse);
 688
 689 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
 690
 691 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
 692                                              struct ip_vs_protocol *pp,
 693                                              const struct ip_vs_iphdr *iph,
 694                                              unsigned int proto_off,
 695                                              int inverse);
 696
 697 /* put back the conn without restarting its timer */
 698 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
 699 {
 700         atomic_dec(&cp->refcnt);
 701 }
 702 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
 703 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
 704
 705 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
 706                                   const union nf_inet_addr *daddr,
 707                                   __be16 dport, unsigned flags,
 708                                   struct ip_vs_dest *dest);
 709 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
 710
 711 extern const char * ip_vs_state_name(__u16 proto, int state);
 712
 713 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
 714 extern int ip_vs_check_template(struct ip_vs_conn *ct);
 715 extern void ip_vs_random_dropentry(void);
 716 extern int ip_vs_conn_init(void);
 717 extern void ip_vs_conn_cleanup(void);
 718
 719 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
 720 {
 721         struct ip_vs_conn *ctl_cp = cp->control;
 722         if (!ctl_cp) {
 723                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
 724                               "%s:%d to %s:%d\n",
 725                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 726                               ntohs(cp->cport),
 727                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 728                               ntohs(cp->vport));
 729
 730                 return;
 731         }
 732
 733         IP_VS_DBG_BUF(7, "DELeting control for: "
 734                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
 735                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 736                       ntohs(cp->cport),
 737                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
 738                       ntohs(ctl_cp->cport));
 739
 740         cp->control = NULL;
 741         if (atomic_read(&ctl_cp->n_control) == 0) {
 742                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
 743                               "%s:%d to %s:%d\n",
 744                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 745                               ntohs(cp->cport),
 746                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 747                               ntohs(cp->vport));
 748
 749                 return;
 750         }
 751         atomic_dec(&ctl_cp->n_control);
 752 }
 753
 754 static inline void
 755 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
 756 {
 757         if (cp->control) {
 758                 IP_VS_ERR_BUF("request control ADD for already controlled: "
 759                               "%s:%d to %s:%d\n",
 760                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 761                               ntohs(cp->cport),
 762                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 763                               ntohs(cp->vport));
 764
 765                 ip_vs_control_del(cp);
 766         }
 767
 768         IP_VS_DBG_BUF(7, "ADDing control for: "
 769                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
 770                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 771                       ntohs(cp->cport),
 772                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
 773                       ntohs(ctl_cp->cport));
 774
 775         cp->control = ctl_cp;
 776         atomic_inc(&ctl_cp->n_control);
 777 }
 778
 779
 780 /*
 781  *      IPVS application functions
 782  *      (from ip_vs_app.c)
 783  */
 784 #define IP_VS_APP_MAX_PORTS  8
 785 extern int register_ip_vs_app(struct ip_vs_app *app);
 786 extern void unregister_ip_vs_app(struct ip_vs_app *app);
 787 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 788 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
 789 extern int
 790 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
 791 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
 792 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
 793
 794 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
 795 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
 796 extern int ip_vs_app_init(void);
 797 extern void ip_vs_app_cleanup(void);
 798
 799
 800 /*
 801  *      IPVS protocol functions (from ip_vs_proto.c)
 802  */
 803 extern int ip_vs_protocol_init(void);
 804 extern void ip_vs_protocol_cleanup(void);
 805 extern void ip_vs_protocol_timeout_change(int flags);
 806 extern int *ip_vs_create_timeout_table(int *table, int size);
 807 extern int
 808 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
 809                         const char *name, int to);
 810 extern void
 811 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
 812                           int offset, const char *msg);
 813
 814 extern struct ip_vs_protocol ip_vs_protocol_tcp;
 815 extern struct ip_vs_protocol ip_vs_protocol_udp;
 816 extern struct ip_vs_protocol ip_vs_protocol_icmp;
 817 extern struct ip_vs_protocol ip_vs_protocol_esp;
 818 extern struct ip_vs_protocol ip_vs_protocol_ah;
 819 extern struct ip_vs_protocol ip_vs_protocol_sctp;
 820
 821 /*
 822  *      Registering/unregistering scheduler functions
 823  *      (from ip_vs_sched.c)
 824  */
 825 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
 826 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
 827 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
 828                                 struct ip_vs_scheduler *scheduler);
 829 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
 830 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
 831 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
 832 extern struct ip_vs_conn *
 833 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb);
 834 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
 835                         struct ip_vs_protocol *pp);
 836
 837
 838 /*
 839  *      IPVS control data and functions (from ip_vs_ctl.c)
 840  */
 841 extern int sysctl_ip_vs_cache_bypass;
 842 extern int sysctl_ip_vs_expire_nodest_conn;
 843 extern int sysctl_ip_vs_expire_quiescent_template;
 844 extern int sysctl_ip_vs_sync_threshold[2];
 845 extern int sysctl_ip_vs_nat_icmp_send;
 846 extern int sysctl_ip_vs_conntrack;
 847 extern int sysctl_ip_vs_snat_reroute;
 848 extern struct ip_vs_stats ip_vs_stats;
 849 extern const struct ctl_path net_vs_ctl_path[];
 850
 851 extern struct ip_vs_service *
 852 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
 853                   const union nf_inet_addr *vaddr, __be16 vport);
 854
 855 static inline void ip_vs_service_put(struct ip_vs_service *svc)
 856 {
 857         atomic_dec(&svc->usecnt);
 858 }
 859
 860 extern struct ip_vs_dest *
 861 ip_vs_lookup_real_service(int af, __u16 protocol,
 862                           const union nf_inet_addr *daddr, __be16 dport);
 863
 864 extern int ip_vs_use_count_inc(void);
 865 extern void ip_vs_use_count_dec(void);
 866 extern int ip_vs_control_init(void);
 867 extern void ip_vs_control_cleanup(void);
 868 extern struct ip_vs_dest *
 869 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
 870                 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol);
 871 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
 872
 873
 874 /*
 875  *      IPVS sync daemon data and function prototypes
 876  *      (from ip_vs_sync.c)
 877  */
 878 extern volatile int ip_vs_sync_state;
 879 extern volatile int ip_vs_master_syncid;
 880 extern volatile int ip_vs_backup_syncid;
 881 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 882 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 883 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
 884 extern int stop_sync_thread(int state);
 885 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
 886
 887
 888 /*
 889  *      IPVS rate estimator prototypes (from ip_vs_est.c)
 890  */
 891 extern int ip_vs_estimator_init(void);
 892 extern void ip_vs_estimator_cleanup(void);
 893 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
 894 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
 895 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
 896
 897 /*
 898  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
 899  */
 900 extern int ip_vs_null_xmit
 901 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 902 extern int ip_vs_bypass_xmit
 903 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 904 extern int ip_vs_nat_xmit
 905 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 906 extern int ip_vs_tunnel_xmit
 907 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 908 extern int ip_vs_dr_xmit
 909 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 910 extern int ip_vs_icmp_xmit
 911 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
 912 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
 913
 914 #ifdef CONFIG_IP_VS_IPV6
 915 extern int ip_vs_bypass_xmit_v6
 916 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 917 extern int ip_vs_nat_xmit_v6
 918 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 919 extern int ip_vs_tunnel_xmit_v6
 920 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 921 extern int ip_vs_dr_xmit_v6
 922 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 923 extern int ip_vs_icmp_xmit_v6
 924 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
 925  int offset);
 926 #endif
 927
 928 /*
 929  *      This is a simple mechanism to ignore packets when
 930  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
 931  *      we start to drop 1/rate of the packets
 932  */
 933 extern int ip_vs_drop_rate;
 934 extern int ip_vs_drop_counter;
 935
 936 static __inline__ int ip_vs_todrop(void)
 937 {
 938         if (!ip_vs_drop_rate) return 0;
 939         if (--ip_vs_drop_counter > 0) return 0;
 940         ip_vs_drop_counter = ip_vs_drop_rate;
 941         return 1;
 942 }
 943
 944 /*
 945  *      ip_vs_fwd_tag returns the forwarding tag of the connection
 946  */
 947 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
 948
 949 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
 950 {
 951         char fwd;
 952
 953         switch (IP_VS_FWD_METHOD(cp)) {
 954         case IP_VS_CONN_F_MASQ:
 955                 fwd = 'M'; break;
 956         case IP_VS_CONN_F_LOCALNODE:
 957                 fwd = 'L'; break;
 958         case IP_VS_CONN_F_TUNNEL:
 959                 fwd = 'T'; break;
 960         case IP_VS_CONN_F_DROUTE:
 961                 fwd = 'R'; break;
 962         case IP_VS_CONN_F_BYPASS:
 963                 fwd = 'B'; break;
 964         default:
 965                 fwd = '?'; break;
 966         }
 967         return fwd;
 968 }
 969
 970 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
 971                            struct ip_vs_conn *cp, int dir);
 972
 973 #ifdef CONFIG_IP_VS_IPV6
 974 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
 975                               struct ip_vs_conn *cp, int dir);
 976 #endif
 977
 978 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
 979
 980 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
 981 {
 982         __be32 diff[2] = { ~old, new };
 983
 984         return csum_partial(diff, sizeof(diff), oldsum);
 985 }
 986
 987 #ifdef CONFIG_IP_VS_IPV6
 988 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
 989                                         __wsum oldsum)
 990 {
 991         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
 992                             new[3],  new[2],  new[1],  new[0] };
 993
 994         return csum_partial(diff, sizeof(diff), oldsum);
 995 }
 996 #endif
 997
 998 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
 999 {
1000         __be16 diff[2] = { ~old, new };
1001
1002         return csum_partial(diff, sizeof(diff), oldsum);
1003 }
1004
1005 #ifdef CONFIG_IP_VS_NFCT
1006 /*
1007  *      Netfilter connection tracking
1008  *      (from ip_vs_nfct.c)
1009  */
1010 static inline int ip_vs_conntrack_enabled(void)
1011 {
1012         return sysctl_ip_vs_conntrack;
1013 }
1014
1015 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1016                                    int outin);
1017 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1018 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1019                                       struct ip_vs_conn *cp, u_int8_t proto,
1020                                       const __be16 port, int from_rs);
1021 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1022
1023 #else
1024
1025 static inline int ip_vs_conntrack_enabled(void)
1026 {
1027         return 0;
1028 }
1029
1030 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1031                                           struct ip_vs_conn *cp, int outin)
1032 {
1033 }
1034
1035 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1036                                           struct ip_vs_conn *cp)
1037 {
1038         return NF_ACCEPT;
1039 }
1040
1041 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1042 {
1043 }
1044 /* CONFIG_IP_VS_NFCT */
1045 #endif
1046
1047 #endif /* __KERNEL__ */
1048
1049 #endif  /* _NET_IP_VS_H */