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 /*
 361  *      IP_VS structure allocated for each dynamically scheduled connection
 362  */
 363 struct ip_vs_conn {
 364         struct list_head        c_list;         /* hashed list heads */
 365
 366         /* Protocol, addresses and port numbers */
 367         u16                      af;            /* address family */
 368         union nf_inet_addr       caddr;          /* client address */
 369         union nf_inet_addr       vaddr;          /* virtual address */
 370         union nf_inet_addr       daddr;          /* destination address */
 371         volatile __u32           flags;          /* status flags */
 372         __be16                   cport;
 373         __be16                   vport;
 374         __be16                   dport;
 375         __u16                   protocol;       /* Which protocol (TCP/UDP) */
 376
 377         /* counter and timer */
 378         atomic_t                refcnt;         /* reference count */
 379         struct timer_list       timer;          /* Expiration timer */
 380         volatile unsigned long  timeout;        /* timeout */
 381
 382         /* Flags and state transition */
 383         spinlock_t              lock;           /* lock for state transition */
 384         volatile __u16          state;          /* state info */
 385         volatile __u16          old_state;      /* old state, to be used for
 386                                                  * state transition triggerd
 387                                                  * synchronization
 388                                                  */
 389
 390         /* Control members */
 391         struct ip_vs_conn       *control;       /* Master control connection */
 392         atomic_t                n_control;      /* Number of controlled ones */
 393         struct ip_vs_dest       *dest;          /* real server */
 394         atomic_t                in_pkts;        /* incoming packet counter */
 395
 396         /* packet transmitter for different forwarding methods.  If it
 397            mangles the packet, it must return NF_DROP or better NF_STOLEN,
 398            otherwise this must be changed to a sk_buff **.
 399          */
 400         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
 401                            struct ip_vs_protocol *pp);
 402
 403         /* Note: we can group the following members into a structure,
 404            in order to save more space, and the following members are
 405            only used in VS/NAT anyway */
 406         struct ip_vs_app        *app;           /* bound ip_vs_app object */
 407         void                    *app_data;      /* Application private data */
 408         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
 409         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
 410 };
 411
 412
 413 /*
 414  *      Extended internal versions of struct ip_vs_service_user and
 415  *      ip_vs_dest_user for IPv6 support.
 416  *
 417  *      We need these to conveniently pass around service and destination
 418  *      options, but unfortunately, we also need to keep the old definitions to
 419  *      maintain userspace backwards compatibility for the setsockopt interface.
 420  */
 421 struct ip_vs_service_user_kern {
 422         /* virtual service addresses */
 423         u16                     af;
 424         u16                     protocol;
 425         union nf_inet_addr      addr;           /* virtual ip address */
 426         u16                     port;
 427         u32                     fwmark;         /* firwall mark of service */
 428
 429         /* virtual service options */
 430         char                    *sched_name;
 431         unsigned                flags;          /* virtual service flags */
 432         unsigned                timeout;        /* persistent timeout in sec */
 433         u32                     netmask;        /* persistent netmask */
 434 };
 435
 436
 437 struct ip_vs_dest_user_kern {
 438         /* destination server address */
 439         union nf_inet_addr      addr;
 440         u16                     port;
 441
 442         /* real server options */
 443         unsigned                conn_flags;     /* connection flags */
 444         int                     weight;         /* destination weight */
 445
 446         /* thresholds for active connections */
 447         u32                     u_threshold;    /* upper threshold */
 448         u32                     l_threshold;    /* lower threshold */
 449 };
 450
 451
 452 /*
 453  *      The information about the virtual service offered to the net
 454  *      and the forwarding entries
 455  */
 456 struct ip_vs_service {
 457         struct list_head        s_list;   /* for normal service table */
 458         struct list_head        f_list;   /* for fwmark-based service table */
 459         atomic_t                refcnt;   /* reference counter */
 460         atomic_t                usecnt;   /* use counter */
 461
 462         u16                     af;       /* address family */
 463         __u16                   protocol; /* which protocol (TCP/UDP) */
 464         union nf_inet_addr      addr;     /* IP address for virtual service */
 465         __be16                  port;     /* port number for the service */
 466         __u32                   fwmark;   /* firewall mark of the service */
 467         unsigned                flags;    /* service status flags */
 468         unsigned                timeout;  /* persistent timeout in ticks */
 469         __be32                  netmask;  /* grouping granularity */
 470
 471         struct list_head        destinations;  /* real server d-linked list */
 472         __u32                   num_dests;     /* number of servers */
 473         struct ip_vs_stats      stats;         /* statistics for the service */
 474         struct ip_vs_app        *inc;     /* bind conns to this app inc */
 475
 476         /* for scheduling */
 477         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
 478         rwlock_t                sched_lock;    /* lock sched_data */
 479         void                    *sched_data;   /* scheduler application data */
 480 };
 481
 482
 483 /*
 484  *      The real server destination forwarding entry
 485  *      with ip address, port number, and so on.
 486  */
 487 struct ip_vs_dest {
 488         struct list_head        n_list;   /* for the dests in the service */
 489         struct list_head        d_list;   /* for table with all the dests */
 490
 491         u16                     af;             /* address family */
 492         union nf_inet_addr      addr;           /* IP address of the server */
 493         __be16                  port;           /* port number of the server */
 494         volatile unsigned       flags;          /* dest status flags */
 495         atomic_t                conn_flags;     /* flags to copy to conn */
 496         atomic_t                weight;         /* server weight */
 497
 498         atomic_t                refcnt;         /* reference counter */
 499         struct ip_vs_stats      stats;          /* statistics */
 500
 501         /* connection counters and thresholds */
 502         atomic_t                activeconns;    /* active connections */
 503         atomic_t                inactconns;     /* inactive connections */
 504         atomic_t                persistconns;   /* persistent connections */
 505         __u32                   u_threshold;    /* upper threshold */
 506         __u32                   l_threshold;    /* lower threshold */
 507
 508         /* for destination cache */
 509         spinlock_t              dst_lock;       /* lock of dst_cache */
 510         struct dst_entry        *dst_cache;     /* destination cache entry */
 511         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
 512
 513         /* for virtual service */
 514         struct ip_vs_service    *svc;           /* service it belongs to */
 515         __u16                   protocol;       /* which protocol (TCP/UDP) */
 516         union nf_inet_addr      vaddr;          /* virtual IP address */
 517         __be16                  vport;          /* virtual port number */
 518         __u32                   vfwmark;        /* firewall mark of service */
 519 };
 520
 521
 522 /*
 523  *      The scheduler object
 524  */
 525 struct ip_vs_scheduler {
 526         struct list_head        n_list;         /* d-linked list head */
 527         char                    *name;          /* scheduler name */
 528         atomic_t                refcnt;         /* reference counter */
 529         struct module           *module;        /* THIS_MODULE/NULL */
 530
 531         /* scheduler initializing service */
 532         int (*init_service)(struct ip_vs_service *svc);
 533         /* scheduling service finish */
 534         int (*done_service)(struct ip_vs_service *svc);
 535         /* scheduler updating service */
 536         int (*update_service)(struct ip_vs_service *svc);
 537
 538         /* selecting a server from the given service */
 539         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 540                                        const struct sk_buff *skb);
 541 };
 542
 543
 544 /*
 545  *      The application module object (a.k.a. app incarnation)
 546  */
 547 struct ip_vs_app {
 548         struct list_head        a_list;         /* member in app list */
 549         int                     type;           /* IP_VS_APP_TYPE_xxx */
 550         char                    *name;          /* application module name */
 551         __u16                   protocol;
 552         struct module           *module;        /* THIS_MODULE/NULL */
 553         struct list_head        incs_list;      /* list of incarnations */
 554
 555         /* members for application incarnations */
 556         struct list_head        p_list;         /* member in proto app list */
 557         struct ip_vs_app        *app;           /* its real application */
 558         __be16                  port;           /* port number in net order */
 559         atomic_t                usecnt;         /* usage counter */
 560
 561         /* output hook: return false if can't linearize. diff set for TCP.  */
 562         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 563                        struct sk_buff *, int *diff);
 564
 565         /* input hook: return false if can't linearize. diff set for TCP. */
 566         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 567                       struct sk_buff *, int *diff);
 568
 569         /* ip_vs_app initializer */
 570         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 571
 572         /* ip_vs_app finish */
 573         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 574
 575
 576         /* not used now */
 577         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 578                          struct ip_vs_protocol *);
 579
 580         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 581
 582         int *                   timeout_table;
 583         int *                   timeouts;
 584         int                     timeouts_size;
 585
 586         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 587                              int *verdict, struct ip_vs_conn **cpp);
 588
 589         struct ip_vs_conn *
 590         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 591                        const struct iphdr *iph, unsigned int proto_off,
 592                        int inverse);
 593
 594         struct ip_vs_conn *
 595         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 596                         const struct iphdr *iph, unsigned int proto_off,
 597                         int inverse);
 598
 599         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 600                                 const struct sk_buff *skb,
 601                                 struct ip_vs_app *app);
 602
 603         void (*timeout_change)(struct ip_vs_app *app, int flags);
 604 };
 605
 606
 607 /*
 608  *      IPVS core functions
 609  *      (from ip_vs_core.c)
 610  */
 611 extern const char *ip_vs_proto_name(unsigned proto);
 612 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
 613 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
 614
 615 #define IP_VS_APP_TYPE_FTP      1
 616
 617 /*
 618  *     ip_vs_conn handling functions
 619  *     (from ip_vs_conn.c)
 620  */
 621
 622 enum {
 623         IP_VS_DIR_INPUT = 0,
 624         IP_VS_DIR_OUTPUT,
 625         IP_VS_DIR_INPUT_ONLY,
 626         IP_VS_DIR_LAST,
 627 };
 628
 629 extern struct ip_vs_conn *ip_vs_conn_in_get
 630 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
 631  const union nf_inet_addr *d_addr, __be16 d_port);
 632
 633 extern struct ip_vs_conn *ip_vs_ct_in_get
 634 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
 635  const union nf_inet_addr *d_addr, __be16 d_port);
 636
 637 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
 638                                             struct ip_vs_protocol *pp,
 639                                             const struct ip_vs_iphdr *iph,
 640                                             unsigned int proto_off,
 641                                             int inverse);
 642
 643 extern struct ip_vs_conn *ip_vs_conn_out_get
 644 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
 645  const union nf_inet_addr *d_addr, __be16 d_port);
 646
 647 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
 648                                              struct ip_vs_protocol *pp,
 649                                              const struct ip_vs_iphdr *iph,
 650                                              unsigned int proto_off,
 651                                              int inverse);
 652
 653 /* put back the conn without restarting its timer */
 654 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
 655 {
 656         atomic_dec(&cp->refcnt);
 657 }
 658 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
 659 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
 660
 661 extern struct ip_vs_conn *
 662 ip_vs_conn_new(int af, int proto, const union nf_inet_addr *caddr, __be16 cport,
 663                const union nf_inet_addr *vaddr, __be16 vport,
 664                const union nf_inet_addr *daddr, __be16 dport, unsigned flags,
 665                struct ip_vs_dest *dest);
 666 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
 667
 668 extern const char * ip_vs_state_name(__u16 proto, int state);
 669
 670 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
 671 extern int ip_vs_check_template(struct ip_vs_conn *ct);
 672 extern void ip_vs_random_dropentry(void);
 673 extern int ip_vs_conn_init(void);
 674 extern void ip_vs_conn_cleanup(void);
 675
 676 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
 677 {
 678         struct ip_vs_conn *ctl_cp = cp->control;
 679         if (!ctl_cp) {
 680                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
 681                               "%s:%d to %s:%d\n",
 682                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 683                               ntohs(cp->cport),
 684                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 685                               ntohs(cp->vport));
 686
 687                 return;
 688         }
 689
 690         IP_VS_DBG_BUF(7, "DELeting control for: "
 691                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
 692                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 693                       ntohs(cp->cport),
 694                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
 695                       ntohs(ctl_cp->cport));
 696
 697         cp->control = NULL;
 698         if (atomic_read(&ctl_cp->n_control) == 0) {
 699                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
 700                               "%s:%d to %s:%d\n",
 701                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 702                               ntohs(cp->cport),
 703                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 704                               ntohs(cp->vport));
 705
 706                 return;
 707         }
 708         atomic_dec(&ctl_cp->n_control);
 709 }
 710
 711 static inline void
 712 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
 713 {
 714         if (cp->control) {
 715                 IP_VS_ERR_BUF("request control ADD for already controlled: "
 716                               "%s:%d to %s:%d\n",
 717                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 718                               ntohs(cp->cport),
 719                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
 720                               ntohs(cp->vport));
 721
 722                 ip_vs_control_del(cp);
 723         }
 724
 725         IP_VS_DBG_BUF(7, "ADDing control for: "
 726                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
 727                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
 728                       ntohs(cp->cport),
 729                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
 730                       ntohs(ctl_cp->cport));
 731
 732         cp->control = ctl_cp;
 733         atomic_inc(&ctl_cp->n_control);
 734 }
 735
 736
 737 /*
 738  *      IPVS application functions
 739  *      (from ip_vs_app.c)
 740  */
 741 #define IP_VS_APP_MAX_PORTS  8
 742 extern int register_ip_vs_app(struct ip_vs_app *app);
 743 extern void unregister_ip_vs_app(struct ip_vs_app *app);
 744 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 745 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
 746 extern int
 747 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
 748 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
 749 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
 750
 751 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
 752 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
 753 extern int ip_vs_app_init(void);
 754 extern void ip_vs_app_cleanup(void);
 755
 756
 757 /*
 758  *      IPVS protocol functions (from ip_vs_proto.c)
 759  */
 760 extern int ip_vs_protocol_init(void);
 761 extern void ip_vs_protocol_cleanup(void);
 762 extern void ip_vs_protocol_timeout_change(int flags);
 763 extern int *ip_vs_create_timeout_table(int *table, int size);
 764 extern int
 765 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
 766                         const char *name, int to);
 767 extern void
 768 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
 769                           int offset, const char *msg);
 770
 771 extern struct ip_vs_protocol ip_vs_protocol_tcp;
 772 extern struct ip_vs_protocol ip_vs_protocol_udp;
 773 extern struct ip_vs_protocol ip_vs_protocol_icmp;
 774 extern struct ip_vs_protocol ip_vs_protocol_esp;
 775 extern struct ip_vs_protocol ip_vs_protocol_ah;
 776 extern struct ip_vs_protocol ip_vs_protocol_sctp;
 777
 778 /*
 779  *      Registering/unregistering scheduler functions
 780  *      (from ip_vs_sched.c)
 781  */
 782 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
 783 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
 784 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
 785                                 struct ip_vs_scheduler *scheduler);
 786 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
 787 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
 788 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
 789 extern struct ip_vs_conn *
 790 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
 791 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
 792                         struct ip_vs_protocol *pp);
 793
 794
 795 /*
 796  *      IPVS control data and functions (from ip_vs_ctl.c)
 797  */
 798 extern int sysctl_ip_vs_cache_bypass;
 799 extern int sysctl_ip_vs_expire_nodest_conn;
 800 extern int sysctl_ip_vs_expire_quiescent_template;
 801 extern int sysctl_ip_vs_sync_threshold[2];
 802 extern int sysctl_ip_vs_nat_icmp_send;
 803 extern int sysctl_ip_vs_conntrack;
 804 extern struct ip_vs_stats ip_vs_stats;
 805 extern const struct ctl_path net_vs_ctl_path[];
 806
 807 extern struct ip_vs_service *
 808 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
 809                   const union nf_inet_addr *vaddr, __be16 vport);
 810
 811 static inline void ip_vs_service_put(struct ip_vs_service *svc)
 812 {
 813         atomic_dec(&svc->usecnt);
 814 }
 815
 816 extern struct ip_vs_dest *
 817 ip_vs_lookup_real_service(int af, __u16 protocol,
 818                           const union nf_inet_addr *daddr, __be16 dport);
 819
 820 extern int ip_vs_use_count_inc(void);
 821 extern void ip_vs_use_count_dec(void);
 822 extern int ip_vs_control_init(void);
 823 extern void ip_vs_control_cleanup(void);
 824 extern struct ip_vs_dest *
 825 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
 826                 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol);
 827 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
 828
 829
 830 /*
 831  *      IPVS sync daemon data and function prototypes
 832  *      (from ip_vs_sync.c)
 833  */
 834 extern volatile int ip_vs_sync_state;
 835 extern volatile int ip_vs_master_syncid;
 836 extern volatile int ip_vs_backup_syncid;
 837 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 838 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 839 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
 840 extern int stop_sync_thread(int state);
 841 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
 842
 843
 844 /*
 845  *      IPVS rate estimator prototypes (from ip_vs_est.c)
 846  */
 847 extern int ip_vs_estimator_init(void);
 848 extern void ip_vs_estimator_cleanup(void);
 849 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
 850 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
 851 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
 852
 853 /*
 854  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
 855  */
 856 extern int ip_vs_null_xmit
 857 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 858 extern int ip_vs_bypass_xmit
 859 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 860 extern int ip_vs_nat_xmit
 861 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 862 extern int ip_vs_tunnel_xmit
 863 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 864 extern int ip_vs_dr_xmit
 865 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 866 extern int ip_vs_icmp_xmit
 867 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
 868 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
 869
 870 #ifdef CONFIG_IP_VS_IPV6
 871 extern int ip_vs_bypass_xmit_v6
 872 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 873 extern int ip_vs_nat_xmit_v6
 874 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 875 extern int ip_vs_tunnel_xmit_v6
 876 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 877 extern int ip_vs_dr_xmit_v6
 878 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
 879 extern int ip_vs_icmp_xmit_v6
 880 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
 881  int offset);
 882 #endif
 883
 884 /*
 885  *      This is a simple mechanism to ignore packets when
 886  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
 887  *      we start to drop 1/rate of the packets
 888  */
 889 extern int ip_vs_drop_rate;
 890 extern int ip_vs_drop_counter;
 891
 892 static __inline__ int ip_vs_todrop(void)
 893 {
 894         if (!ip_vs_drop_rate) return 0;
 895         if (--ip_vs_drop_counter > 0) return 0;
 896         ip_vs_drop_counter = ip_vs_drop_rate;
 897         return 1;
 898 }
 899
 900 /*
 901  *      ip_vs_fwd_tag returns the forwarding tag of the connection
 902  */
 903 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
 904
 905 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
 906 {
 907         char fwd;
 908
 909         switch (IP_VS_FWD_METHOD(cp)) {
 910         case IP_VS_CONN_F_MASQ:
 911                 fwd = 'M'; break;
 912         case IP_VS_CONN_F_LOCALNODE:
 913                 fwd = 'L'; break;
 914         case IP_VS_CONN_F_TUNNEL:
 915                 fwd = 'T'; break;
 916         case IP_VS_CONN_F_DROUTE:
 917                 fwd = 'R'; break;
 918         case IP_VS_CONN_F_BYPASS:
 919                 fwd = 'B'; break;
 920         default:
 921                 fwd = '?'; break;
 922         }
 923         return fwd;
 924 }
 925
 926 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
 927                            struct ip_vs_conn *cp, int dir);
 928
 929 #ifdef CONFIG_IP_VS_IPV6
 930 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
 931                               struct ip_vs_conn *cp, int dir);
 932 #endif
 933
 934 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
 935
 936 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
 937 {
 938         __be32 diff[2] = { ~old, new };
 939
 940         return csum_partial(diff, sizeof(diff), oldsum);
 941 }
 942
 943 #ifdef CONFIG_IP_VS_IPV6
 944 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
 945                                         __wsum oldsum)
 946 {
 947         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
 948                             new[3],  new[2],  new[1],  new[0] };
 949
 950         return csum_partial(diff, sizeof(diff), oldsum);
 951 }
 952 #endif
 953
 954 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
 955 {
 956         __be16 diff[2] = { ~old, new };
 957
 958         return csum_partial(diff, sizeof(diff), oldsum);
 959 }
 960
 961 #ifdef CONFIG_IP_VS_NFCT
 962 /*
 963  *      Netfilter connection tracking
 964  *      (from ip_vs_nfct.c)
 965  */
 966 static inline int ip_vs_conntrack_enabled(void)
 967 {
 968         return sysctl_ip_vs_conntrack;
 969 }
 970
 971 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
 972                                    int outin);
 973 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
 974 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
 975                                       struct ip_vs_conn *cp, u_int8_t proto,
 976                                       const __be16 port, int from_rs);
 977 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
 978
 979 #else
 980
 981 static inline int ip_vs_conntrack_enabled(void)
 982 {
 983         return 0;
 984 }
 985
 986 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
 987                                           struct ip_vs_conn *cp, int outin)
 988 {
 989 }
 990
 991 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
 992                                           struct ip_vs_conn *cp)
 993 {
 994         return NF_ACCEPT;
 995 }
 996
 997 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
 998 {
 999 }
1000 /* CONFIG_IP_VS_NFCT */
1001 #endif
1002
1003 #endif /* __KERNEL__ */
1004
1005 #endif  /* _NET_IP_VS_H */