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 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  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         char                    *pe_name;
 448         unsigned                flags;          /* virtual service flags */
 449         unsigned                timeout;        /* persistent timeout in sec */
 450         u32                     netmask;        /* persistent netmask */
 451 };
 452
 453
 454 struct ip_vs_dest_user_kern {
 455         /* destination server address */
 456         union nf_inet_addr      addr;
 457         u16                     port;
 458
 459         /* real server options */
 460         unsigned                conn_flags;     /* connection flags */
 461         int                     weight;         /* destination weight */
 462
 463         /* thresholds for active connections */
 464         u32                     u_threshold;    /* upper threshold */
 465         u32                     l_threshold;    /* lower threshold */
 466 };
 467
 468
 469 /*
 470  *      The information about the virtual service offered to the net
 471  *      and the forwarding entries
 472  */
 473 struct ip_vs_service {
 474         struct list_head        s_list;   /* for normal service table */
 475         struct list_head        f_list;   /* for fwmark-based service table */
 476         atomic_t                refcnt;   /* reference counter */
 477         atomic_t                usecnt;   /* use counter */
 478
 479         u16                     af;       /* address family */
 480         __u16                   protocol; /* which protocol (TCP/UDP) */
 481         union nf_inet_addr      addr;     /* IP address for virtual service */
 482         __be16                  port;     /* port number for the service */
 483         __u32                   fwmark;   /* firewall mark of the service */
 484         unsigned                flags;    /* service status flags */
 485         unsigned                timeout;  /* persistent timeout in ticks */
 486         __be32                  netmask;  /* grouping granularity */
 487
 488         struct list_head        destinations;  /* real server d-linked list */
 489         __u32                   num_dests;     /* number of servers */
 490         struct ip_vs_stats      stats;         /* statistics for the service */
 491         struct ip_vs_app        *inc;     /* bind conns to this app inc */
 492
 493         /* for scheduling */
 494         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
 495         rwlock_t                sched_lock;    /* lock sched_data */
 496         void                    *sched_data;   /* scheduler application data */
 497
 498         /* alternate persistence engine */
 499         struct ip_vs_pe         *pe;
 500 };
 501
 502
 503 /*
 504  *      The real server destination forwarding entry
 505  *      with ip address, port number, and so on.
 506  */
 507 struct ip_vs_dest {
 508         struct list_head        n_list;   /* for the dests in the service */
 509         struct list_head        d_list;   /* for table with all the dests */
 510
 511         u16                     af;             /* address family */
 512         union nf_inet_addr      addr;           /* IP address of the server */
 513         __be16                  port;           /* port number of the server */
 514         volatile unsigned       flags;          /* dest status flags */
 515         atomic_t                conn_flags;     /* flags to copy to conn */
 516         atomic_t                weight;         /* server weight */
 517
 518         atomic_t                refcnt;         /* reference counter */
 519         struct ip_vs_stats      stats;          /* statistics */
 520
 521         /* connection counters and thresholds */
 522         atomic_t                activeconns;    /* active connections */
 523         atomic_t                inactconns;     /* inactive connections */
 524         atomic_t                persistconns;   /* persistent connections */
 525         __u32                   u_threshold;    /* upper threshold */
 526         __u32                   l_threshold;    /* lower threshold */
 527
 528         /* for destination cache */
 529         spinlock_t              dst_lock;       /* lock of dst_cache */
 530         struct dst_entry        *dst_cache;     /* destination cache entry */
 531         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
 532         u32                     dst_cookie;
 533 #ifdef CONFIG_IP_VS_IPV6
 534         struct in6_addr         dst_saddr;
 535 #endif
 536
 537         /* for virtual service */
 538         struct ip_vs_service    *svc;           /* service it belongs to */
 539         __u16                   protocol;       /* which protocol (TCP/UDP) */
 540         union nf_inet_addr      vaddr;          /* virtual IP address */
 541         __be16                  vport;          /* virtual port number */
 542         __u32                   vfwmark;        /* firewall mark of service */
 543 };
 544
 545
 546 /*
 547  *      The scheduler object
 548  */
 549 struct ip_vs_scheduler {
 550         struct list_head        n_list;         /* d-linked list head */
 551         char                    *name;          /* scheduler name */
 552         atomic_t                refcnt;         /* reference counter */
 553         struct module           *module;        /* THIS_MODULE/NULL */
 554
 555         /* scheduler initializing service */
 556         int (*init_service)(struct ip_vs_service *svc);
 557         /* scheduling service finish */
 558         int (*done_service)(struct ip_vs_service *svc);
 559         /* scheduler updating service */
 560         int (*update_service)(struct ip_vs_service *svc);
 561
 562         /* selecting a server from the given service */
 563         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 564                                        const struct sk_buff *skb);
 565 };
 566
 567 /* The persistence engine object */
 568 struct ip_vs_pe {
 569         struct list_head        n_list;         /* d-linked list head */
 570         char                    *name;          /* scheduler name */
 571         atomic_t                refcnt;         /* reference counter */
 572         struct module           *module;        /* THIS_MODULE/NULL */
 573
 574         /* get the connection template, if any */
 575         int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
 576         bool (*ct_match)(const struct ip_vs_conn_param *p,
 577                          struct ip_vs_conn *ct);
 578         u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
 579                            bool inverse);
 580         int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
 581 };
 582
 583 /*
 584  *      The application module object (a.k.a. app incarnation)
 585  */
 586 struct ip_vs_app {
 587         struct list_head        a_list;         /* member in app list */
 588         int                     type;           /* IP_VS_APP_TYPE_xxx */
 589         char                    *name;          /* application module name */
 590         __u16                   protocol;
 591         struct module           *module;        /* THIS_MODULE/NULL */
 592         struct list_head        incs_list;      /* list of incarnations */
 593
 594         /* members for application incarnations */
 595         struct list_head        p_list;         /* member in proto app list */
 596         struct ip_vs_app        *app;           /* its real application */
 597         __be16                  port;           /* port number in net order */
 598         atomic_t                usecnt;         /* usage counter */
 599
 600         /*
 601          * output hook: Process packet in inout direction, diff set for TCP.
 602          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 603          *         2=Mangled but checksum was not updated
 604          */
 605         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 606                        struct sk_buff *, int *diff);
 607
 608         /*
 609          * input hook: Process packet in outin direction, diff set for TCP.
 610          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 611          *         2=Mangled but checksum was not updated
 612          */
 613         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 614                       struct sk_buff *, int *diff);
 615
 616         /* ip_vs_app initializer */
 617         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 618
 619         /* ip_vs_app finish */
 620         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 621
 622
 623         /* not used now */
 624         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 625                          struct ip_vs_protocol *);
 626
 627         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 628
 629         int *                   timeout_table;
 630         int *                   timeouts;
 631         int                     timeouts_size;
 632
 633         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 634                              int *verdict, struct ip_vs_conn **cpp);
 635
 636         struct ip_vs_conn *
 637         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 638                        const struct iphdr *iph, unsigned int proto_off,
 639                        int inverse);
 640
 641         struct ip_vs_conn *
 642         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 643                         const struct iphdr *iph, unsigned int proto_off,
 644                         int inverse);
 645
 646         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 647                                 const struct sk_buff *skb,
 648                                 struct ip_vs_app *app);
 649
 650         void (*timeout_change)(struct ip_vs_app *app, int flags);
 651 };
 652
 653
 654 /*
 655  *      IPVS core functions
 656  *      (from ip_vs_core.c)
 657  */
 658 extern const char *ip_vs_proto_name(unsigned proto);
 659 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
 660 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
 661
 662 #define IP_VS_APP_TYPE_FTP      1
 663
 664 /*
 665  *     ip_vs_conn handling functions
 666  *     (from ip_vs_conn.c)
 667  */
 668
 669 enum {
 670         IP_VS_DIR_INPUT = 0,
 671         IP_VS_DIR_OUTPUT,
 672         IP_VS_DIR_INPUT_ONLY,
 673         IP_VS_DIR_LAST,
 674 };
 675
 676 static inline void ip_vs_conn_fill_param(int af, int protocol,
 677                                          const union nf_inet_addr *caddr,
 678                                          __be16 cport,
 679                                          const union nf_inet_addr *vaddr,
 680                                          __be16 vport,
 681                                          struct ip_vs_conn_param *p)
 682 {
 683         p->af = af;
 684         p->protocol = protocol;
 685         p->caddr = caddr;
 686         p->cport = cport;
 687         p->vaddr = vaddr;
 688         p->vport = vport;
 689         p->pe = NULL;
 690         p->pe_data = NULL;
 691 }
 692
 693 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
 694 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
 695
 696 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
 697                                             struct ip_vs_protocol *pp,
 698                                             const struct ip_vs_iphdr *iph,
 699                                             unsigned int proto_off,
 700                                             int inverse);
 701
 702 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
 703
 704 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
 705                                              struct ip_vs_protocol *pp,
 706                                              const struct ip_vs_iphdr *iph,
 707                                              unsigned int proto_off,
 708                                              int inverse);
 709
 710 /* put back the conn without restarting its timer */
 711 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
 712 {
 713         atomic_dec(&cp->refcnt);
 714 }
 715 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
 716 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
 717
 718 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
 719                                   const union nf_inet_addr *daddr,
 720                                   __be16 dport, unsigned flags,
 721                                   struct ip_vs_dest *dest);
 722 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
 723
 724 extern const char * ip_vs_state_name(__u16 proto, int state);
 725
 726 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
 727 extern int ip_vs_check_template(struct ip_vs_conn *ct);
 728 extern void ip_vs_random_dropentry(void);
 729 extern int ip_vs_conn_init(void);
 730 extern void ip_vs_conn_cleanup(void);
 731
 732 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
 733 {
 734         struct ip_vs_conn *ctl_cp = cp->control;
 735         if (!ctl_cp) {
 736                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
 737                               "%s:%d to %s:%d\n",
 738                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 739                               ntohs(cp->cport),
 740                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 741                               ntohs(cp->vport));
 742
 743                 return;
 744         }
 745
 746         IP_VS_DBG_BUF(7, "DELeting control for: "
 747                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
 748                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 749                       ntohs(cp->cport),
 750                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
 751                       ntohs(ctl_cp->cport));
 752
 753         cp->control = NULL;
 754         if (atomic_read(&ctl_cp->n_control) == 0) {
 755                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
 756                               "%s:%d to %s:%d\n",
 757                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 758                               ntohs(cp->cport),
 759                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 760                               ntohs(cp->vport));
 761
 762                 return;
 763         }
 764         atomic_dec(&ctl_cp->n_control);
 765 }
 766
 767 static inline void
 768 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
 769 {
 770         if (cp->control) {
 771                 IP_VS_ERR_BUF("request control ADD for already controlled: "
 772                               "%s:%d to %s:%d\n",
 773                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 774                               ntohs(cp->cport),
 775                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 776                               ntohs(cp->vport));
 777
 778                 ip_vs_control_del(cp);
 779         }
 780
 781         IP_VS_DBG_BUF(7, "ADDing control for: "
 782                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
 783                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 784                       ntohs(cp->cport),
 785                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
 786                       ntohs(ctl_cp->cport));
 787
 788         cp->control = ctl_cp;
 789         atomic_inc(&ctl_cp->n_control);
 790 }
 791
 792
 793 /*
 794  *      IPVS application functions
 795  *      (from ip_vs_app.c)
 796  */
 797 #define IP_VS_APP_MAX_PORTS  8
 798 extern int register_ip_vs_app(struct ip_vs_app *app);
 799 extern void unregister_ip_vs_app(struct ip_vs_app *app);
 800 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 801 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
 802 extern int
 803 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
 804 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
 805 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
 806
 807 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
 808 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
 809 extern int ip_vs_app_init(void);
 810 extern void ip_vs_app_cleanup(void);
 811
 812 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
 813 void ip_vs_unbind_pe(struct ip_vs_service *svc);
 814 int register_ip_vs_pe(struct ip_vs_pe *pe);
 815 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
 816 extern struct ip_vs_pe *ip_vs_pe_get(const char *name);
 817 extern void ip_vs_pe_put(struct ip_vs_pe *pe);
 818
 819 /*
 820  *      IPVS protocol functions (from ip_vs_proto.c)
 821  */
 822 extern int ip_vs_protocol_init(void);
 823 extern void ip_vs_protocol_cleanup(void);
 824 extern void ip_vs_protocol_timeout_change(int flags);
 825 extern int *ip_vs_create_timeout_table(int *table, int size);
 826 extern int
 827 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
 828                         const char *name, int to);
 829 extern void
 830 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
 831                           int offset, const char *msg);
 832
 833 extern struct ip_vs_protocol ip_vs_protocol_tcp;
 834 extern struct ip_vs_protocol ip_vs_protocol_udp;
 835 extern struct ip_vs_protocol ip_vs_protocol_icmp;
 836 extern struct ip_vs_protocol ip_vs_protocol_esp;
 837 extern struct ip_vs_protocol ip_vs_protocol_ah;
 838 extern struct ip_vs_protocol ip_vs_protocol_sctp;
 839
 840 /*
 841  *      Registering/unregistering scheduler functions
 842  *      (from ip_vs_sched.c)
 843  */
 844 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
 845 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
 846 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
 847                                 struct ip_vs_scheduler *scheduler);
 848 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
 849 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
 850 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
 851 extern struct ip_vs_conn *
 852 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb);
 853 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
 854                         struct ip_vs_protocol *pp);
 855
 856
 857 /*
 858  *      IPVS control data and functions (from ip_vs_ctl.c)
 859  */
 860 extern int sysctl_ip_vs_cache_bypass;
 861 extern int sysctl_ip_vs_expire_nodest_conn;
 862 extern int sysctl_ip_vs_expire_quiescent_template;
 863 extern int sysctl_ip_vs_sync_threshold[2];
 864 extern int sysctl_ip_vs_nat_icmp_send;
 865 extern int sysctl_ip_vs_conntrack;
 866 extern int sysctl_ip_vs_snat_reroute;
 867 extern struct ip_vs_stats ip_vs_stats;
 868 extern const struct ctl_path net_vs_ctl_path[];
 869
 870 extern struct ip_vs_service *
 871 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
 872                   const union nf_inet_addr *vaddr, __be16 vport);
 873
 874 static inline void ip_vs_service_put(struct ip_vs_service *svc)
 875 {
 876         atomic_dec(&svc->usecnt);
 877 }
 878
 879 extern struct ip_vs_dest *
 880 ip_vs_lookup_real_service(int af, __u16 protocol,
 881                           const union nf_inet_addr *daddr, __be16 dport);
 882
 883 extern int ip_vs_use_count_inc(void);
 884 extern void ip_vs_use_count_dec(void);
 885 extern int ip_vs_control_init(void);
 886 extern void ip_vs_control_cleanup(void);
 887 extern struct ip_vs_dest *
 888 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
 889                 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol);
 890 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
 891
 892
 893 /*
 894  *      IPVS sync daemon data and function prototypes
 895  *      (from ip_vs_sync.c)
 896  */
 897 extern volatile int ip_vs_sync_state;
 898 extern volatile int ip_vs_master_syncid;
 899 extern volatile int ip_vs_backup_syncid;
 900 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 901 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 902 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
 903 extern int stop_sync_thread(int state);
 904 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
 905
 906
 907 /*
 908  *      IPVS rate estimator prototypes (from ip_vs_est.c)
 909  */
 910 extern int ip_vs_estimator_init(void);
 911 extern void ip_vs_estimator_cleanup(void);
 912 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
 913 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
 914 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
 915
 916 /*
 917  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
 918  */
 919 extern int ip_vs_null_xmit
 920 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 921 extern int ip_vs_bypass_xmit
 922 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 923 extern int ip_vs_nat_xmit
 924 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 925 extern int ip_vs_tunnel_xmit
 926 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 927 extern int ip_vs_dr_xmit
 928 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 929 extern int ip_vs_icmp_xmit
 930 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
 931 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
 932
 933 #ifdef CONFIG_IP_VS_IPV6
 934 extern int ip_vs_bypass_xmit_v6
 935 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 936 extern int ip_vs_nat_xmit_v6
 937 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 938 extern int ip_vs_tunnel_xmit_v6
 939 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 940 extern int ip_vs_dr_xmit_v6
 941 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 942 extern int ip_vs_icmp_xmit_v6
 943 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
 944  int offset);
 945 #endif
 946
 947 /*
 948  *      This is a simple mechanism to ignore packets when
 949  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
 950  *      we start to drop 1/rate of the packets
 951  */
 952 extern int ip_vs_drop_rate;
 953 extern int ip_vs_drop_counter;
 954
 955 static __inline__ int ip_vs_todrop(void)
 956 {
 957         if (!ip_vs_drop_rate) return 0;
 958         if (--ip_vs_drop_counter > 0) return 0;
 959         ip_vs_drop_counter = ip_vs_drop_rate;
 960         return 1;
 961 }
 962
 963 /*
 964  *      ip_vs_fwd_tag returns the forwarding tag of the connection
 965  */
 966 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
 967
 968 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
 969 {
 970         char fwd;
 971
 972         switch (IP_VS_FWD_METHOD(cp)) {
 973         case IP_VS_CONN_F_MASQ:
 974                 fwd = 'M'; break;
 975         case IP_VS_CONN_F_LOCALNODE:
 976                 fwd = 'L'; break;
 977         case IP_VS_CONN_F_TUNNEL:
 978                 fwd = 'T'; break;
 979         case IP_VS_CONN_F_DROUTE:
 980                 fwd = 'R'; break;
 981         case IP_VS_CONN_F_BYPASS:
 982                 fwd = 'B'; break;
 983         default:
 984                 fwd = '?'; break;
 985         }
 986         return fwd;
 987 }
 988
 989 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
 990                            struct ip_vs_conn *cp, int dir);
 991
 992 #ifdef CONFIG_IP_VS_IPV6
 993 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
 994                               struct ip_vs_conn *cp, int dir);
 995 #endif
 996
 997 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
 998
 999 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1000 {
1001         __be32 diff[2] = { ~old, new };
1002
1003         return csum_partial(diff, sizeof(diff), oldsum);
1004 }
1005
1006 #ifdef CONFIG_IP_VS_IPV6
1007 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1008                                         __wsum oldsum)
1009 {
1010         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1011                             new[3],  new[2],  new[1],  new[0] };
1012
1013         return csum_partial(diff, sizeof(diff), oldsum);
1014 }
1015 #endif
1016
1017 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1018 {
1019         __be16 diff[2] = { ~old, new };
1020
1021         return csum_partial(diff, sizeof(diff), oldsum);
1022 }
1023
1024 /*
1025  * Forget current conntrack (unconfirmed) and attach notrack entry
1026  */
1027 static inline void ip_vs_notrack(struct sk_buff *skb)
1028 {
1029 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1030         enum ip_conntrack_info ctinfo;
1031         struct nf_conn *ct = ct = nf_ct_get(skb, &ctinfo);
1032
1033         if (!ct || !nf_ct_is_untracked(ct)) {
1034                 nf_reset(skb);
1035                 skb->nfct = &nf_ct_untracked_get()->ct_general;
1036                 skb->nfctinfo = IP_CT_NEW;
1037                 nf_conntrack_get(skb->nfct);
1038         }
1039 #endif
1040 }
1041
1042 #ifdef CONFIG_IP_VS_NFCT
1043 /*
1044  *      Netfilter connection tracking
1045  *      (from ip_vs_nfct.c)
1046  */
1047 static inline int ip_vs_conntrack_enabled(void)
1048 {
1049         return sysctl_ip_vs_conntrack;
1050 }
1051
1052 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1053                                    int outin);
1054 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1055 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1056                                       struct ip_vs_conn *cp, u_int8_t proto,
1057                                       const __be16 port, int from_rs);
1058 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1059
1060 #else
1061
1062 static inline int ip_vs_conntrack_enabled(void)
1063 {
1064         return 0;
1065 }
1066
1067 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1068                                           struct ip_vs_conn *cp, int outin)
1069 {
1070 }
1071
1072 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1073                                           struct ip_vs_conn *cp)
1074 {
1075         return NF_ACCEPT;
1076 }
1077
1078 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1079 {
1080 }
1081 /* CONFIG_IP_VS_NFCT */
1082 #endif
1083
1084 #endif /* __KERNEL__ */
1085
1086 #endif  /* _NET_IP_VS_H */