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 #include <net/net_namespace.h>          /* Netw namespace */
  32
  33 /*
  34  * Generic access of ipvs struct
  35  */
  36 static inline struct netns_ipvs *net_ipvs(struct net* net)
  37 {
  38         return net->ipvs;
  39 }
  40 /*
  41  * Get net ptr from skb in traffic cases
  42  * use skb_sknet when call is from userland (ioctl or netlink)
  43  */
  44 static inline struct net *skb_net(const struct sk_buff *skb)
  45 {
  46 #ifdef CONFIG_NET_NS
  47 #ifdef CONFIG_IP_VS_DEBUG
  48         /*
  49          * This is used for debug only.
  50          * Start with the most likely hit
  51          * End with BUG
  52          */
  53         if (likely(skb->dev && skb->dev->nd_net))
  54                 return dev_net(skb->dev);
  55         if (skb_dst(skb)->dev)
  56                 return dev_net(skb_dst(skb)->dev);
  57         WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
  58                       __func__, __LINE__);
  59         if (likely(skb->sk && skb->sk->sk_net))
  60                 return sock_net(skb->sk);
  61         pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
  62                 __func__, __LINE__);
  63         BUG();
  64 #else
  65         return dev_net(skb->dev ? : skb_dst(skb)->dev);
  66 #endif
  67 #else
  68         return &init_net;
  69 #endif
  70 }
  71
  72 static inline struct net *skb_sknet(const struct sk_buff *skb)
  73 {
  74 #ifdef CONFIG_NET_NS
  75 #ifdef CONFIG_IP_VS_DEBUG
  76         /* Start with the most likely hit */
  77         if (likely(skb->sk && skb->sk->sk_net))
  78                 return sock_net(skb->sk);
  79         WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
  80                        __func__, __LINE__);
  81         if (likely(skb->dev && skb->dev->nd_net))
  82                 return dev_net(skb->dev);
  83         pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
  84                 __func__, __LINE__);
  85         BUG();
  86 #else
  87         return sock_net(skb->sk);
  88 #endif
  89 #else
  90         return &init_net;
  91 #endif
  92 }
  93 /*
  94  * This one needed for single_open_net since net is stored directly in
  95  * private not as a struct i.e. seq_file_net cant be used.
  96  */
  97 static inline struct net *seq_file_single_net(struct seq_file *seq)
  98 {
  99 #ifdef CONFIG_NET_NS
 100         return (struct net *)seq->private;
 101 #else
 102         return &init_net;
 103 #endif
 104 }
 105
 106 /* Connections' size value needed by ip_vs_ctl.c */
 107 extern int ip_vs_conn_tab_size;
 108
 109
 110 struct ip_vs_iphdr {
 111         int len;
 112         __u8 protocol;
 113         union nf_inet_addr saddr;
 114         union nf_inet_addr daddr;
 115 };
 116
 117 static inline void
 118 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
 119 {
 120 #ifdef CONFIG_IP_VS_IPV6
 121         if (af == AF_INET6) {
 122                 const struct ipv6hdr *iph = nh;
 123                 iphdr->len = sizeof(struct ipv6hdr);
 124                 iphdr->protocol = iph->nexthdr;
 125                 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
 126                 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
 127         } else
 128 #endif
 129         {
 130                 const struct iphdr *iph = nh;
 131                 iphdr->len = iph->ihl * 4;
 132                 iphdr->protocol = iph->protocol;
 133                 iphdr->saddr.ip = iph->saddr;
 134                 iphdr->daddr.ip = iph->daddr;
 135         }
 136 }
 137
 138 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
 139                                    const union nf_inet_addr *src)
 140 {
 141 #ifdef CONFIG_IP_VS_IPV6
 142         if (af == AF_INET6)
 143                 ipv6_addr_copy(&dst->in6, &src->in6);
 144         else
 145 #endif
 146         dst->ip = src->ip;
 147 }
 148
 149 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
 150                                    const union nf_inet_addr *b)
 151 {
 152 #ifdef CONFIG_IP_VS_IPV6
 153         if (af == AF_INET6)
 154                 return ipv6_addr_equal(&a->in6, &b->in6);
 155 #endif
 156         return a->ip == b->ip;
 157 }
 158
 159 #ifdef CONFIG_IP_VS_DEBUG
 160 #include <linux/net.h>
 161
 162 extern int ip_vs_get_debug_level(void);
 163
 164 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
 165                                          const union nf_inet_addr *addr,
 166                                          int *idx)
 167 {
 168         int len;
 169 #ifdef CONFIG_IP_VS_IPV6
 170         if (af == AF_INET6)
 171                 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
 172                                &addr->in6) + 1;
 173         else
 174 #endif
 175                 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
 176                                &addr->ip) + 1;
 177
 178         *idx += len;
 179         BUG_ON(*idx > buf_len + 1);
 180         return &buf[*idx - len];
 181 }
 182
 183 #define IP_VS_DBG_BUF(level, msg, ...)                                  \
 184         do {                                                            \
 185                 char ip_vs_dbg_buf[160];                                \
 186                 int ip_vs_dbg_idx = 0;                                  \
 187                 if (level <= ip_vs_get_debug_level())                   \
 188                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 189         } while (0)
 190 #define IP_VS_ERR_BUF(msg...)                                           \
 191         do {                                                            \
 192                 char ip_vs_dbg_buf[160];                                \
 193                 int ip_vs_dbg_idx = 0;                                  \
 194                 pr_err(msg);                                            \
 195         } while (0)
 196
 197 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
 198 #define IP_VS_DBG_ADDR(af, addr)                                        \
 199         ip_vs_dbg_addr(af, ip_vs_dbg_buf,                               \
 200                        sizeof(ip_vs_dbg_buf), addr,                     \
 201                        &ip_vs_dbg_idx)
 202
 203 #define IP_VS_DBG(level, msg, ...)                                      \
 204         do {                                                            \
 205                 if (level <= ip_vs_get_debug_level())                   \
 206                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 207         } while (0)
 208 #define IP_VS_DBG_RL(msg, ...)                                          \
 209         do {                                                            \
 210                 if (net_ratelimit())                                    \
 211                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 212         } while (0)
 213 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)                     \
 214         do {                                                            \
 215                 if (level <= ip_vs_get_debug_level())                   \
 216                         pp->debug_packet(af, pp, skb, ofs, msg);        \
 217         } while (0)
 218 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)                  \
 219         do {                                                            \
 220                 if (level <= ip_vs_get_debug_level() &&                 \
 221                     net_ratelimit())                                    \
 222                         pp->debug_packet(af, pp, skb, ofs, msg);        \
 223         } while (0)
 224 #else   /* NO DEBUGGING at ALL */
 225 #define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
 226 #define IP_VS_ERR_BUF(msg...)  do {} while (0)
 227 #define IP_VS_DBG(level, msg...)  do {} while (0)
 228 #define IP_VS_DBG_RL(msg...)  do {} while (0)
 229 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)     do {} while (0)
 230 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)  do {} while (0)
 231 #endif
 232
 233 #define IP_VS_BUG() BUG()
 234 #define IP_VS_ERR_RL(msg, ...)                                          \
 235         do {                                                            \
 236                 if (net_ratelimit())                                    \
 237                         pr_err(msg, ##__VA_ARGS__);                     \
 238         } while (0)
 239
 240 #ifdef CONFIG_IP_VS_DEBUG
 241 #define EnterFunction(level)                                            \
 242         do {                                                            \
 243                 if (level <= ip_vs_get_debug_level())                   \
 244                         printk(KERN_DEBUG                               \
 245                                pr_fmt("Enter: %s, %s line %i\n"),       \
 246                                __func__, __FILE__, __LINE__);           \
 247         } while (0)
 248 #define LeaveFunction(level)                                            \
 249         do {                                                            \
 250                 if (level <= ip_vs_get_debug_level())                   \
 251                         printk(KERN_DEBUG                               \
 252                                pr_fmt("Leave: %s, %s line %i\n"),       \
 253                                __func__, __FILE__, __LINE__);           \
 254         } while (0)
 255 #else
 256 #define EnterFunction(level)   do {} while (0)
 257 #define LeaveFunction(level)   do {} while (0)
 258 #endif
 259
 260 #define IP_VS_WAIT_WHILE(expr)  while (expr) { cpu_relax(); }
 261
 262
 263 /*
 264  *      The port number of FTP service (in network order).
 265  */
 266 #define FTPPORT  cpu_to_be16(21)
 267 #define FTPDATA  cpu_to_be16(20)
 268
 269 /*
 270  *      TCP State Values
 271  */
 272 enum {
 273         IP_VS_TCP_S_NONE = 0,
 274         IP_VS_TCP_S_ESTABLISHED,
 275         IP_VS_TCP_S_SYN_SENT,
 276         IP_VS_TCP_S_SYN_RECV,
 277         IP_VS_TCP_S_FIN_WAIT,
 278         IP_VS_TCP_S_TIME_WAIT,
 279         IP_VS_TCP_S_CLOSE,
 280         IP_VS_TCP_S_CLOSE_WAIT,
 281         IP_VS_TCP_S_LAST_ACK,
 282         IP_VS_TCP_S_LISTEN,
 283         IP_VS_TCP_S_SYNACK,
 284         IP_VS_TCP_S_LAST
 285 };
 286
 287 /*
 288  *      UDP State Values
 289  */
 290 enum {
 291         IP_VS_UDP_S_NORMAL,
 292         IP_VS_UDP_S_LAST,
 293 };
 294
 295 /*
 296  *      ICMP State Values
 297  */
 298 enum {
 299         IP_VS_ICMP_S_NORMAL,
 300         IP_VS_ICMP_S_LAST,
 301 };
 302
 303 /*
 304  *      SCTP State Values
 305  */
 306 enum ip_vs_sctp_states {
 307         IP_VS_SCTP_S_NONE,
 308         IP_VS_SCTP_S_INIT_CLI,
 309         IP_VS_SCTP_S_INIT_SER,
 310         IP_VS_SCTP_S_INIT_ACK_CLI,
 311         IP_VS_SCTP_S_INIT_ACK_SER,
 312         IP_VS_SCTP_S_ECHO_CLI,
 313         IP_VS_SCTP_S_ECHO_SER,
 314         IP_VS_SCTP_S_ESTABLISHED,
 315         IP_VS_SCTP_S_SHUT_CLI,
 316         IP_VS_SCTP_S_SHUT_SER,
 317         IP_VS_SCTP_S_SHUT_ACK_CLI,
 318         IP_VS_SCTP_S_SHUT_ACK_SER,
 319         IP_VS_SCTP_S_CLOSED,
 320         IP_VS_SCTP_S_LAST
 321 };
 322
 323 /*
 324  *      Delta sequence info structure
 325  *      Each ip_vs_conn has 2 (output AND input seq. changes).
 326  *      Only used in the VS/NAT.
 327  */
 328 struct ip_vs_seq {
 329         __u32                   init_seq;       /* Add delta from this seq */
 330         __u32                   delta;          /* Delta in sequence numbers */
 331         __u32                   previous_delta; /* Delta in sequence numbers
 332                                                    before last resized pkt */
 333 };
 334
 335 /*
 336  * counters per cpu
 337  */
 338 struct ip_vs_counters {
 339         __u32           conns;          /* connections scheduled */
 340         __u32           inpkts;         /* incoming packets */
 341         __u32           outpkts;        /* outgoing packets */
 342         __u64           inbytes;        /* incoming bytes */
 343         __u64           outbytes;       /* outgoing bytes */
 344 };
 345 /*
 346  * Stats per cpu
 347  */
 348 struct ip_vs_cpu_stats {
 349         struct ip_vs_counters   ustats;
 350         struct u64_stats_sync   syncp;
 351 };
 352
 353 /*
 354  *      IPVS statistics objects
 355  */
 356 struct ip_vs_estimator {
 357         struct list_head        list;
 358
 359         u64                     last_inbytes;
 360         u64                     last_outbytes;
 361         u32                     last_conns;
 362         u32                     last_inpkts;
 363         u32                     last_outpkts;
 364
 365         u32                     cps;
 366         u32                     inpps;
 367         u32                     outpps;
 368         u32                     inbps;
 369         u32                     outbps;
 370 };
 371
 372 struct ip_vs_stats {
 373         struct ip_vs_stats_user ustats;         /* statistics */
 374         struct ip_vs_estimator  est;            /* estimator */
 375         struct ip_vs_cpu_stats  *cpustats;      /* per cpu counters */
 376         spinlock_t              lock;           /* spin lock */
 377 };
 378
 379 /*
 380  * Helper Macros for per cpu
 381  * ipvs->tot_stats->ustats.count
 382  */
 383 #define IPVS_STAT_INC(ipvs, count)      \
 384         __this_cpu_inc((ipvs)->ustats->count)
 385
 386 #define IPVS_STAT_ADD(ipvs, count, value) \
 387         do {\
 388                 write_seqcount_begin(per_cpu_ptr((ipvs)->ustats_seq, \
 389                                      raw_smp_processor_id())); \
 390                 __this_cpu_add((ipvs)->ustats->count, value); \
 391                 write_seqcount_end(per_cpu_ptr((ipvs)->ustats_seq, \
 392                                    raw_smp_processor_id())); \
 393         } while (0)
 394
 395 struct dst_entry;
 396 struct iphdr;
 397 struct ip_vs_conn;
 398 struct ip_vs_app;
 399 struct sk_buff;
 400 struct ip_vs_proto_data;
 401
 402 struct ip_vs_protocol {
 403         struct ip_vs_protocol   *next;
 404         char                    *name;
 405         u16                     protocol;
 406         u16                     num_states;
 407         int                     dont_defrag;
 408
 409         void (*init)(struct ip_vs_protocol *pp);
 410
 411         void (*exit)(struct ip_vs_protocol *pp);
 412
 413         void (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
 414
 415         void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
 416
 417         int (*conn_schedule)(int af, struct sk_buff *skb,
 418                              struct ip_vs_proto_data *pd,
 419                              int *verdict, struct ip_vs_conn **cpp);
 420
 421         struct ip_vs_conn *
 422         (*conn_in_get)(int af,
 423                        const struct sk_buff *skb,
 424                        const struct ip_vs_iphdr *iph,
 425                        unsigned int proto_off,
 426                        int inverse);
 427
 428         struct ip_vs_conn *
 429         (*conn_out_get)(int af,
 430                         const struct sk_buff *skb,
 431                         const struct ip_vs_iphdr *iph,
 432                         unsigned int proto_off,
 433                         int inverse);
 434
 435         int (*snat_handler)(struct sk_buff *skb,
 436                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 437
 438         int (*dnat_handler)(struct sk_buff *skb,
 439                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 440
 441         int (*csum_check)(int af, struct sk_buff *skb,
 442                           struct ip_vs_protocol *pp);
 443
 444         const char *(*state_name)(int state);
 445
 446         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 447                                 const struct sk_buff *skb,
 448                                 struct ip_vs_proto_data *pd);
 449
 450         int (*register_app)(struct net *net, struct ip_vs_app *inc);
 451
 452         void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
 453
 454         int (*app_conn_bind)(struct ip_vs_conn *cp);
 455
 456         void (*debug_packet)(int af, struct ip_vs_protocol *pp,
 457                              const struct sk_buff *skb,
 458                              int offset,
 459                              const char *msg);
 460
 461         void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
 462 };
 463
 464 /*
 465  * protocol data per netns
 466  */
 467 struct ip_vs_proto_data {
 468         struct ip_vs_proto_data *next;
 469         struct ip_vs_protocol   *pp;
 470         int                     *timeout_table; /* protocol timeout table */
 471         atomic_t                appcnt;         /* counter of proto app incs. */
 472         struct tcp_states_t     *tcp_state_table;
 473 };
 474
 475 extern struct ip_vs_protocol   *ip_vs_proto_get(unsigned short proto);
 476 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
 477                                                      unsigned short proto);
 478
 479 struct ip_vs_conn_param {
 480         struct net                      *net;
 481         const union nf_inet_addr        *caddr;
 482         const union nf_inet_addr        *vaddr;
 483         __be16                          cport;
 484         __be16                          vport;
 485         __u16                           protocol;
 486         u16                             af;
 487
 488         const struct ip_vs_pe           *pe;
 489         char                            *pe_data;
 490         __u8                            pe_data_len;
 491 };
 492
 493 /*
 494  *      IP_VS structure allocated for each dynamically scheduled connection
 495  */
 496 struct ip_vs_conn {
 497         struct hlist_node       c_list;         /* hashed list heads */
 498 #ifdef CONFIG_NET_NS
 499         struct net              *net;           /* Name space */
 500 #endif
 501         /* Protocol, addresses and port numbers */
 502         u16                     af;             /* address family */
 503         __be16                  cport;
 504         __be16                  vport;
 505         __be16                  dport;
 506         __u32                   fwmark;         /* Fire wall mark from skb */
 507         union nf_inet_addr      caddr;          /* client address */
 508         union nf_inet_addr      vaddr;          /* virtual address */
 509         union nf_inet_addr      daddr;          /* destination address */
 510         volatile __u32          flags;          /* status flags */
 511         __u16                   protocol;       /* Which protocol (TCP/UDP) */
 512
 513         /* counter and timer */
 514         atomic_t                refcnt;         /* reference count */
 515         struct timer_list       timer;          /* Expiration timer */
 516         volatile unsigned long  timeout;        /* timeout */
 517
 518         /* Flags and state transition */
 519         spinlock_t              lock;           /* lock for state transition */
 520         volatile __u16          state;          /* state info */
 521         volatile __u16          old_state;      /* old state, to be used for
 522                                                  * state transition triggerd
 523                                                  * synchronization
 524                                                  */
 525
 526         /* Control members */
 527         struct ip_vs_conn       *control;       /* Master control connection */
 528         atomic_t                n_control;      /* Number of controlled ones */
 529         struct ip_vs_dest       *dest;          /* real server */
 530         atomic_t                in_pkts;        /* incoming packet counter */
 531
 532         /* packet transmitter for different forwarding methods.  If it
 533            mangles the packet, it must return NF_DROP or better NF_STOLEN,
 534            otherwise this must be changed to a sk_buff **.
 535            NF_ACCEPT can be returned when destination is local.
 536          */
 537         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
 538                            struct ip_vs_protocol *pp);
 539
 540         /* Note: we can group the following members into a structure,
 541            in order to save more space, and the following members are
 542            only used in VS/NAT anyway */
 543         struct ip_vs_app        *app;           /* bound ip_vs_app object */
 544         void                    *app_data;      /* Application private data */
 545         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
 546         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
 547
 548         const struct ip_vs_pe   *pe;
 549         char                    *pe_data;
 550         __u8                    pe_data_len;
 551 };
 552
 553 /*
 554  *  To save some memory in conn table when name space is disabled.
 555  */
 556 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
 557 {
 558 #ifdef CONFIG_NET_NS
 559         return cp->net;
 560 #else
 561         return &init_net;
 562 #endif
 563 }
 564 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
 565 {
 566 #ifdef CONFIG_NET_NS
 567         cp->net = net;
 568 #endif
 569 }
 570
 571 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
 572                                     struct net *net)
 573 {
 574 #ifdef CONFIG_NET_NS
 575         return cp->net == net;
 576 #else
 577         return 1;
 578 #endif
 579 }
 580
 581 /*
 582  *      Extended internal versions of struct ip_vs_service_user and
 583  *      ip_vs_dest_user for IPv6 support.
 584  *
 585  *      We need these to conveniently pass around service and destination
 586  *      options, but unfortunately, we also need to keep the old definitions to
 587  *      maintain userspace backwards compatibility for the setsockopt interface.
 588  */
 589 struct ip_vs_service_user_kern {
 590         /* virtual service addresses */
 591         u16                     af;
 592         u16                     protocol;
 593         union nf_inet_addr      addr;           /* virtual ip address */
 594         u16                     port;
 595         u32                     fwmark;         /* firwall mark of service */
 596
 597         /* virtual service options */
 598         char                    *sched_name;
 599         char                    *pe_name;
 600         unsigned                flags;          /* virtual service flags */
 601         unsigned                timeout;        /* persistent timeout in sec */
 602         u32                     netmask;        /* persistent netmask */
 603 };
 604
 605
 606 struct ip_vs_dest_user_kern {
 607         /* destination server address */
 608         union nf_inet_addr      addr;
 609         u16                     port;
 610
 611         /* real server options */
 612         unsigned                conn_flags;     /* connection flags */
 613         int                     weight;         /* destination weight */
 614
 615         /* thresholds for active connections */
 616         u32                     u_threshold;    /* upper threshold */
 617         u32                     l_threshold;    /* lower threshold */
 618 };
 619
 620
 621 /*
 622  *      The information about the virtual service offered to the net
 623  *      and the forwarding entries
 624  */
 625 struct ip_vs_service {
 626         struct list_head        s_list;   /* for normal service table */
 627         struct list_head        f_list;   /* for fwmark-based service table */
 628         atomic_t                refcnt;   /* reference counter */
 629         atomic_t                usecnt;   /* use counter */
 630
 631         u16                     af;       /* address family */
 632         __u16                   protocol; /* which protocol (TCP/UDP) */
 633         union nf_inet_addr      addr;     /* IP address for virtual service */
 634         __be16                  port;     /* port number for the service */
 635         __u32                   fwmark;   /* firewall mark of the service */
 636         unsigned                flags;    /* service status flags */
 637         unsigned                timeout;  /* persistent timeout in ticks */
 638         __be32                  netmask;  /* grouping granularity */
 639         struct net              *net;
 640
 641         struct list_head        destinations;  /* real server d-linked list */
 642         __u32                   num_dests;     /* number of servers */
 643         struct ip_vs_stats      stats;         /* statistics for the service */
 644         struct ip_vs_app        *inc;     /* bind conns to this app inc */
 645
 646         /* for scheduling */
 647         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
 648         rwlock_t                sched_lock;    /* lock sched_data */
 649         void                    *sched_data;   /* scheduler application data */
 650
 651         /* alternate persistence engine */
 652         struct ip_vs_pe         *pe;
 653 };
 654
 655
 656 /*
 657  *      The real server destination forwarding entry
 658  *      with ip address, port number, and so on.
 659  */
 660 struct ip_vs_dest {
 661         struct list_head        n_list;   /* for the dests in the service */
 662         struct list_head        d_list;   /* for table with all the dests */
 663
 664         u16                     af;             /* address family */
 665         __be16                  port;           /* port number of the server */
 666         union nf_inet_addr      addr;           /* IP address of the server */
 667         volatile unsigned       flags;          /* dest status flags */
 668         atomic_t                conn_flags;     /* flags to copy to conn */
 669         atomic_t                weight;         /* server weight */
 670
 671         atomic_t                refcnt;         /* reference counter */
 672         struct ip_vs_stats      stats;          /* statistics */
 673
 674         /* connection counters and thresholds */
 675         atomic_t                activeconns;    /* active connections */
 676         atomic_t                inactconns;     /* inactive connections */
 677         atomic_t                persistconns;   /* persistent connections */
 678         __u32                   u_threshold;    /* upper threshold */
 679         __u32                   l_threshold;    /* lower threshold */
 680
 681         /* for destination cache */
 682         spinlock_t              dst_lock;       /* lock of dst_cache */
 683         struct dst_entry        *dst_cache;     /* destination cache entry */
 684         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
 685         u32                     dst_cookie;
 686 #ifdef CONFIG_IP_VS_IPV6
 687         struct in6_addr         dst_saddr;
 688 #endif
 689
 690         /* for virtual service */
 691         struct ip_vs_service    *svc;           /* service it belongs to */
 692         __u16                   protocol;       /* which protocol (TCP/UDP) */
 693         __be16                  vport;          /* virtual port number */
 694         union nf_inet_addr      vaddr;          /* virtual IP address */
 695         __u32                   vfwmark;        /* firewall mark of service */
 696 };
 697
 698
 699 /*
 700  *      The scheduler object
 701  */
 702 struct ip_vs_scheduler {
 703         struct list_head        n_list;         /* d-linked list head */
 704         char                    *name;          /* scheduler name */
 705         atomic_t                refcnt;         /* reference counter */
 706         struct module           *module;        /* THIS_MODULE/NULL */
 707
 708         /* scheduler initializing service */
 709         int (*init_service)(struct ip_vs_service *svc);
 710         /* scheduling service finish */
 711         int (*done_service)(struct ip_vs_service *svc);
 712         /* scheduler updating service */
 713         int (*update_service)(struct ip_vs_service *svc);
 714
 715         /* selecting a server from the given service */
 716         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 717                                        const struct sk_buff *skb);
 718 };
 719
 720 /* The persistence engine object */
 721 struct ip_vs_pe {
 722         struct list_head        n_list;         /* d-linked list head */
 723         char                    *name;          /* scheduler name */
 724         atomic_t                refcnt;         /* reference counter */
 725         struct module           *module;        /* THIS_MODULE/NULL */
 726
 727         /* get the connection template, if any */
 728         int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
 729         bool (*ct_match)(const struct ip_vs_conn_param *p,
 730                          struct ip_vs_conn *ct);
 731         u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
 732                            bool inverse);
 733         int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
 734 };
 735
 736 /*
 737  *      The application module object (a.k.a. app incarnation)
 738  */
 739 struct ip_vs_app {
 740         struct list_head        a_list;         /* member in app list */
 741         int                     type;           /* IP_VS_APP_TYPE_xxx */
 742         char                    *name;          /* application module name */
 743         __u16                   protocol;
 744         struct module           *module;        /* THIS_MODULE/NULL */
 745         struct list_head        incs_list;      /* list of incarnations */
 746
 747         /* members for application incarnations */
 748         struct list_head        p_list;         /* member in proto app list */
 749         struct ip_vs_app        *app;           /* its real application */
 750         __be16                  port;           /* port number in net order */
 751         atomic_t                usecnt;         /* usage counter */
 752
 753         /*
 754          * output hook: Process packet in inout direction, diff set for TCP.
 755          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 756          *         2=Mangled but checksum was not updated
 757          */
 758         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 759                        struct sk_buff *, int *diff);
 760
 761         /*
 762          * input hook: Process packet in outin direction, diff set for TCP.
 763          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 764          *         2=Mangled but checksum was not updated
 765          */
 766         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 767                       struct sk_buff *, int *diff);
 768
 769         /* ip_vs_app initializer */
 770         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 771
 772         /* ip_vs_app finish */
 773         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 774
 775
 776         /* not used now */
 777         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 778                          struct ip_vs_protocol *);
 779
 780         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 781
 782         int *                   timeout_table;
 783         int *                   timeouts;
 784         int                     timeouts_size;
 785
 786         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 787                              int *verdict, struct ip_vs_conn **cpp);
 788
 789         struct ip_vs_conn *
 790         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 791                        const struct iphdr *iph, unsigned int proto_off,
 792                        int inverse);
 793
 794         struct ip_vs_conn *
 795         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 796                         const struct iphdr *iph, unsigned int proto_off,
 797                         int inverse);
 798
 799         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 800                                 const struct sk_buff *skb,
 801                                 struct ip_vs_app *app);
 802
 803         void (*timeout_change)(struct ip_vs_app *app, int flags);
 804 };
 805
 806 /* IPVS in network namespace */
 807 struct netns_ipvs {
 808         int                     gen;            /* Generation */
 809         /*
 810          *      Hash table: for real service lookups
 811          */
 812         #define IP_VS_RTAB_BITS 4
 813         #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
 814         #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
 815
 816         struct list_head        rs_table[IP_VS_RTAB_SIZE];
 817         /* ip_vs_app */
 818         struct list_head        app_list;
 819         struct mutex            app_mutex;
 820         struct lock_class_key   app_key;        /* mutex debuging */
 821
 822         /* ip_vs_proto */
 823         #define IP_VS_PROTO_TAB_SIZE    32      /* must be power of 2 */
 824         struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
 825         /* ip_vs_proto_tcp */
 826 #ifdef CONFIG_IP_VS_PROTO_TCP
 827         #define TCP_APP_TAB_BITS        4
 828         #define TCP_APP_TAB_SIZE        (1 << TCP_APP_TAB_BITS)
 829         #define TCP_APP_TAB_MASK        (TCP_APP_TAB_SIZE - 1)
 830         struct list_head        tcp_apps[TCP_APP_TAB_SIZE];
 831         spinlock_t              tcp_app_lock;
 832 #endif
 833         /* ip_vs_proto_udp */
 834 #ifdef CONFIG_IP_VS_PROTO_UDP
 835         #define UDP_APP_TAB_BITS        4
 836         #define UDP_APP_TAB_SIZE        (1 << UDP_APP_TAB_BITS)
 837         #define UDP_APP_TAB_MASK        (UDP_APP_TAB_SIZE - 1)
 838         struct list_head        udp_apps[UDP_APP_TAB_SIZE];
 839         spinlock_t              udp_app_lock;
 840 #endif
 841         /* ip_vs_proto_sctp */
 842 #ifdef CONFIG_IP_VS_PROTO_SCTP
 843         #define SCTP_APP_TAB_BITS       4
 844         #define SCTP_APP_TAB_SIZE       (1 << SCTP_APP_TAB_BITS)
 845         #define SCTP_APP_TAB_MASK       (SCTP_APP_TAB_SIZE - 1)
 846         /* Hash table for SCTP application incarnations  */
 847         struct list_head        sctp_apps[SCTP_APP_TAB_SIZE];
 848         spinlock_t              sctp_app_lock;
 849 #endif
 850         /* ip_vs_conn */
 851         atomic_t                conn_count;      /*  connection counter */
 852
 853         /* ip_vs_ctl */
 854         struct ip_vs_stats              tot_stats;  /* Statistics & est. */
 855         seqcount_t                      *ustats_seq; /* u64 read retry */
 856
 857         int                     num_services;    /* no of virtual services */
 858         /* 1/rate drop and drop-entry variables */
 859         struct delayed_work     defense_work;   /* Work handler */
 860         int                     drop_rate;
 861         int                     drop_counter;
 862         atomic_t                dropentry;
 863         /* locks in ctl.c */
 864         spinlock_t              dropentry_lock;  /* drop entry handling */
 865         spinlock_t              droppacket_lock; /* drop packet handling */
 866         spinlock_t              securetcp_lock;  /* state and timeout tables */
 867         rwlock_t                rs_lock;         /* real services table */
 868         /* semaphore for IPVS sockopts. And, [gs]etsockopt may sleep. */
 869         struct lock_class_key   ctl_key;        /* ctl_mutex debuging */
 870         /* Trash for destinations */
 871         struct list_head        dest_trash;
 872         /* Service counters */
 873         atomic_t                ftpsvc_counter;
 874         atomic_t                nullsvc_counter;
 875
 876         /* sys-ctl struct */
 877         struct ctl_table_header *sysctl_hdr;
 878         struct ctl_table        *sysctl_tbl;
 879         /* sysctl variables */
 880         int                     sysctl_amemthresh;
 881         int                     sysctl_am_droprate;
 882         int                     sysctl_drop_entry;
 883         int                     sysctl_drop_packet;
 884         int                     sysctl_secure_tcp;
 885 #ifdef CONFIG_IP_VS_NFCT
 886         int                     sysctl_conntrack;
 887 #endif
 888         int                     sysctl_snat_reroute;
 889         int                     sysctl_sync_ver;
 890         int                     sysctl_cache_bypass;
 891         int                     sysctl_expire_nodest_conn;
 892         int                     sysctl_expire_quiescent_template;
 893         int                     sysctl_sync_threshold[2];
 894         int                     sysctl_nat_icmp_send;
 895
 896         /* ip_vs_lblc */
 897         int                     sysctl_lblc_expiration;
 898         struct ctl_table_header *lblc_ctl_header;
 899         struct ctl_table        *lblc_ctl_table;
 900         /* ip_vs_lblcr */
 901         int                     sysctl_lblcr_expiration;
 902         struct ctl_table_header *lblcr_ctl_header;
 903         struct ctl_table        *lblcr_ctl_table;
 904         /* ip_vs_est */
 905         struct list_head        est_list;       /* estimator list */
 906         spinlock_t              est_lock;
 907         struct timer_list       est_timer;      /* Estimation timer */
 908         /* ip_vs_sync */
 909         struct list_head        sync_queue;
 910         spinlock_t              sync_lock;
 911         struct ip_vs_sync_buff  *sync_buff;
 912         spinlock_t              sync_buff_lock;
 913         struct sockaddr_in      sync_mcast_addr;
 914         struct task_struct      *master_thread;
 915         struct task_struct      *backup_thread;
 916         int                     send_mesg_maxlen;
 917         int                     recv_mesg_maxlen;
 918         volatile int            sync_state;
 919         volatile int            master_syncid;
 920         volatile int            backup_syncid;
 921         /* multicast interface name */
 922         char                    master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 923         char                    backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 924         /* net name space ptr */
 925         struct net              *net;            /* Needed by timer routines */
 926 };
 927
 928 /*
 929  *      IPVS core functions
 930  *      (from ip_vs_core.c)
 931  */
 932 extern const char *ip_vs_proto_name(unsigned proto);
 933 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
 934 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
 935
 936 #define IP_VS_APP_TYPE_FTP      1
 937
 938 /*
 939  *     ip_vs_conn handling functions
 940  *     (from ip_vs_conn.c)
 941  */
 942
 943 enum {
 944         IP_VS_DIR_INPUT = 0,
 945         IP_VS_DIR_OUTPUT,
 946         IP_VS_DIR_INPUT_ONLY,
 947         IP_VS_DIR_LAST,
 948 };
 949
 950 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
 951                                          const union nf_inet_addr *caddr,
 952                                          __be16 cport,
 953                                          const union nf_inet_addr *vaddr,
 954                                          __be16 vport,
 955                                          struct ip_vs_conn_param *p)
 956 {
 957         p->net = net;
 958         p->af = af;
 959         p->protocol = protocol;
 960         p->caddr = caddr;
 961         p->cport = cport;
 962         p->vaddr = vaddr;
 963         p->vport = vport;
 964         p->pe = NULL;
 965         p->pe_data = NULL;
 966 }
 967
 968 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
 969 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
 970
 971 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
 972                                             const struct ip_vs_iphdr *iph,
 973                                             unsigned int proto_off,
 974                                             int inverse);
 975
 976 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
 977
 978 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
 979                                              const struct ip_vs_iphdr *iph,
 980                                              unsigned int proto_off,
 981                                              int inverse);
 982
 983 /* put back the conn without restarting its timer */
 984 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
 985 {
 986         atomic_dec(&cp->refcnt);
 987 }
 988 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
 989 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
 990
 991 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
 992                                   const union nf_inet_addr *daddr,
 993                                   __be16 dport, unsigned flags,
 994                                   struct ip_vs_dest *dest, __u32 fwmark);
 995 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
 996
 997 extern const char * ip_vs_state_name(__u16 proto, int state);
 998
 999 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1000 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1001 extern void ip_vs_random_dropentry(struct net *net);
1002 extern int ip_vs_conn_init(void);
1003 extern void ip_vs_conn_cleanup(void);
1004
1005 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1006 {
1007         struct ip_vs_conn *ctl_cp = cp->control;
1008         if (!ctl_cp) {
1009                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1010                               "%s:%d to %s:%d\n",
1011                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1012                               ntohs(cp->cport),
1013                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1014                               ntohs(cp->vport));
1015
1016                 return;
1017         }
1018
1019         IP_VS_DBG_BUF(7, "DELeting control for: "
1020                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1021                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1022                       ntohs(cp->cport),
1023                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1024                       ntohs(ctl_cp->cport));
1025
1026         cp->control = NULL;
1027         if (atomic_read(&ctl_cp->n_control) == 0) {
1028                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1029                               "%s:%d to %s:%d\n",
1030                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1031                               ntohs(cp->cport),
1032                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1033                               ntohs(cp->vport));
1034
1035                 return;
1036         }
1037         atomic_dec(&ctl_cp->n_control);
1038 }
1039
1040 static inline void
1041 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1042 {
1043         if (cp->control) {
1044                 IP_VS_ERR_BUF("request control ADD for already controlled: "
1045                               "%s:%d to %s:%d\n",
1046                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1047                               ntohs(cp->cport),
1048                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1049                               ntohs(cp->vport));
1050
1051                 ip_vs_control_del(cp);
1052         }
1053
1054         IP_VS_DBG_BUF(7, "ADDing control for: "
1055                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1056                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1057                       ntohs(cp->cport),
1058                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1059                       ntohs(ctl_cp->cport));
1060
1061         cp->control = ctl_cp;
1062         atomic_inc(&ctl_cp->n_control);
1063 }
1064
1065
1066 /*
1067  *      IPVS application functions
1068  *      (from ip_vs_app.c)
1069  */
1070 #define IP_VS_APP_MAX_PORTS  8
1071 extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1072 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1073 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1074 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1075 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1076                                   __u16 proto, __u16 port);
1077 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1078 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1079
1080 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1081 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1082 extern int ip_vs_app_init(void);
1083 extern void ip_vs_app_cleanup(void);
1084
1085 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1086 void ip_vs_unbind_pe(struct ip_vs_service *svc);
1087 int register_ip_vs_pe(struct ip_vs_pe *pe);
1088 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1089 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1090 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1091
1092 static inline void ip_vs_pe_get(const struct ip_vs_pe *pe)
1093 {
1094         if (pe && pe->module)
1095                 __module_get(pe->module);
1096 }
1097
1098 static inline void ip_vs_pe_put(const struct ip_vs_pe *pe)
1099 {
1100         if (pe && pe->module)
1101                 module_put(pe->module);
1102 }
1103
1104 /*
1105  *      IPVS protocol functions (from ip_vs_proto.c)
1106  */
1107 extern int ip_vs_protocol_init(void);
1108 extern void ip_vs_protocol_cleanup(void);
1109 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1110 extern int *ip_vs_create_timeout_table(int *table, int size);
1111 extern int
1112 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1113                         const char *name, int to);
1114 extern void
1115 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1116                           const struct sk_buff *skb,
1117                           int offset, const char *msg);
1118
1119 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1120 extern struct ip_vs_protocol ip_vs_protocol_udp;
1121 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1122 extern struct ip_vs_protocol ip_vs_protocol_esp;
1123 extern struct ip_vs_protocol ip_vs_protocol_ah;
1124 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1125
1126 /*
1127  *      Registering/unregistering scheduler functions
1128  *      (from ip_vs_sched.c)
1129  */
1130 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1131 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1132 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1133                                 struct ip_vs_scheduler *scheduler);
1134 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1135 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1136 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1137 extern struct ip_vs_conn *
1138 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1139                struct ip_vs_proto_data *pd, int *ignored);
1140 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1141                         struct ip_vs_proto_data *pd);
1142
1143 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1144
1145
1146 /*
1147  *      IPVS control data and functions (from ip_vs_ctl.c)
1148  */
1149 extern struct ip_vs_stats ip_vs_stats;
1150 extern const struct ctl_path net_vs_ctl_path[];
1151 extern int sysctl_ip_vs_sync_ver;
1152
1153 extern void ip_vs_sync_switch_mode(struct net *net, int mode);
1154 extern struct ip_vs_service *
1155 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1156                   const union nf_inet_addr *vaddr, __be16 vport);
1157
1158 static inline void ip_vs_service_put(struct ip_vs_service *svc)
1159 {
1160         atomic_dec(&svc->usecnt);
1161 }
1162
1163 extern struct ip_vs_dest *
1164 ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
1165                           const union nf_inet_addr *daddr, __be16 dport);
1166
1167 extern int ip_vs_use_count_inc(void);
1168 extern void ip_vs_use_count_dec(void);
1169 extern int ip_vs_control_init(void);
1170 extern void ip_vs_control_cleanup(void);
1171 extern struct ip_vs_dest *
1172 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1173                 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1174                 __u16 protocol, __u32 fwmark);
1175 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1176
1177
1178 /*
1179  *      IPVS sync daemon data and function prototypes
1180  *      (from ip_vs_sync.c)
1181  */
1182 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1183                              __u8 syncid);
1184 extern int stop_sync_thread(struct net *net, int state);
1185 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
1186 extern int ip_vs_sync_init(void);
1187 extern void ip_vs_sync_cleanup(void);
1188
1189
1190 /*
1191  *      IPVS rate estimator prototypes (from ip_vs_est.c)
1192  */
1193 extern int ip_vs_estimator_init(void);
1194 extern void ip_vs_estimator_cleanup(void);
1195 extern void ip_vs_new_estimator(struct net *net, struct ip_vs_stats *stats);
1196 extern void ip_vs_kill_estimator(struct net *net, struct ip_vs_stats *stats);
1197 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1198
1199 /*
1200  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
1201  */
1202 extern int ip_vs_null_xmit
1203 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1204 extern int ip_vs_bypass_xmit
1205 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1206 extern int ip_vs_nat_xmit
1207 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1208 extern int ip_vs_tunnel_xmit
1209 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1210 extern int ip_vs_dr_xmit
1211 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1212 extern int ip_vs_icmp_xmit
1213 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
1214 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1215
1216 #ifdef CONFIG_IP_VS_IPV6
1217 extern int ip_vs_bypass_xmit_v6
1218 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1219 extern int ip_vs_nat_xmit_v6
1220 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1221 extern int ip_vs_tunnel_xmit_v6
1222 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1223 extern int ip_vs_dr_xmit_v6
1224 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1225 extern int ip_vs_icmp_xmit_v6
1226 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1227  int offset);
1228 #endif
1229
1230 /*
1231  *      This is a simple mechanism to ignore packets when
1232  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
1233  *      we start to drop 1/rate of the packets
1234  */
1235
1236 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1237 {
1238         if (!ipvs->drop_rate)
1239                 return 0;
1240         if (--ipvs->drop_counter > 0)
1241                 return 0;
1242         ipvs->drop_counter = ipvs->drop_rate;
1243         return 1;
1244 }
1245
1246 /*
1247  *      ip_vs_fwd_tag returns the forwarding tag of the connection
1248  */
1249 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
1250
1251 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1252 {
1253         char fwd;
1254
1255         switch (IP_VS_FWD_METHOD(cp)) {
1256         case IP_VS_CONN_F_MASQ:
1257                 fwd = 'M'; break;
1258         case IP_VS_CONN_F_LOCALNODE:
1259                 fwd = 'L'; break;
1260         case IP_VS_CONN_F_TUNNEL:
1261                 fwd = 'T'; break;
1262         case IP_VS_CONN_F_DROUTE:
1263                 fwd = 'R'; break;
1264         case IP_VS_CONN_F_BYPASS:
1265                 fwd = 'B'; break;
1266         default:
1267                 fwd = '?'; break;
1268         }
1269         return fwd;
1270 }
1271
1272 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1273                            struct ip_vs_conn *cp, int dir);
1274
1275 #ifdef CONFIG_IP_VS_IPV6
1276 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1277                               struct ip_vs_conn *cp, int dir);
1278 #endif
1279
1280 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1281
1282 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1283 {
1284         __be32 diff[2] = { ~old, new };
1285
1286         return csum_partial(diff, sizeof(diff), oldsum);
1287 }
1288
1289 #ifdef CONFIG_IP_VS_IPV6
1290 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1291                                         __wsum oldsum)
1292 {
1293         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1294                             new[3],  new[2],  new[1],  new[0] };
1295
1296         return csum_partial(diff, sizeof(diff), oldsum);
1297 }
1298 #endif
1299
1300 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1301 {
1302         __be16 diff[2] = { ~old, new };
1303
1304         return csum_partial(diff, sizeof(diff), oldsum);
1305 }
1306
1307 /*
1308  * Forget current conntrack (unconfirmed) and attach notrack entry
1309  */
1310 static inline void ip_vs_notrack(struct sk_buff *skb)
1311 {
1312 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1313         enum ip_conntrack_info ctinfo;
1314         struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1315
1316         if (!ct || !nf_ct_is_untracked(ct)) {
1317                 nf_reset(skb);
1318                 skb->nfct = &nf_ct_untracked_get()->ct_general;
1319                 skb->nfctinfo = IP_CT_NEW;
1320                 nf_conntrack_get(skb->nfct);
1321         }
1322 #endif
1323 }
1324
1325 #ifdef CONFIG_IP_VS_NFCT
1326 /*
1327  *      Netfilter connection tracking
1328  *      (from ip_vs_nfct.c)
1329  */
1330 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1331 {
1332         return ipvs->sysctl_conntrack;
1333 }
1334
1335 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1336                                    int outin);
1337 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1338 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1339                                       struct ip_vs_conn *cp, u_int8_t proto,
1340                                       const __be16 port, int from_rs);
1341 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1342
1343 #else
1344
1345 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1346 {
1347         return 0;
1348 }
1349
1350 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1351                                           struct ip_vs_conn *cp, int outin)
1352 {
1353 }
1354
1355 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1356                                           struct ip_vs_conn *cp)
1357 {
1358         return NF_ACCEPT;
1359 }
1360
1361 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1362 {
1363 }
1364 /* CONFIG_IP_VS_NFCT */
1365 #endif
1366
1367 static inline unsigned int
1368 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1369 {
1370         /*
1371          * We think the overhead of processing active connections is 256
1372          * times higher than that of inactive connections in average. (This
1373          * 256 times might not be accurate, we will change it later) We
1374          * use the following formula to estimate the overhead now:
1375          *                dest->activeconns*256 + dest->inactconns
1376          */
1377         return (atomic_read(&dest->activeconns) << 8) +
1378                 atomic_read(&dest->inactconns);
1379 }
1380
1381 #endif /* __KERNEL__ */
1382
1383 #endif  /* _NET_IP_VS_H */