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 #include <asm/types.h>                  /* for __uXX types */
  12
  13 #include <linux/sysctl.h>               /* for ctl_path */
  14 #include <linux/list.h>                 /* for struct list_head */
  15 #include <linux/spinlock.h>             /* for struct rwlock_t */
  16 #include <linux/atomic.h>                 /* for struct atomic_t */
  17 #include <linux/compiler.h>
  18 #include <linux/timer.h>
  19 #include <linux/bug.h>
  20
  21 #include <net/checksum.h>
  22 #include <linux/netfilter.h>            /* for union nf_inet_addr */
  23 #include <linux/ip.h>
  24 #include <linux/ipv6.h>                 /* for struct ipv6hdr */
  25 #include <net/ipv6.h>
  26 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  27 #include <net/netfilter/nf_conntrack.h>
  28 #endif
  29 #include <net/net_namespace.h>          /* Netw namespace */
  30
  31 /*
  32  * Generic access of ipvs struct
  33  */
  34 static inline struct netns_ipvs *net_ipvs(struct net* net)
  35 {
  36         return net->ipvs;
  37 }
  38 /*
  39  * Get net ptr from skb in traffic cases
  40  * use skb_sknet when call is from userland (ioctl or netlink)
  41  */
  42 static inline struct net *skb_net(const struct sk_buff *skb)
  43 {
  44 #ifdef CONFIG_NET_NS
  45 #ifdef CONFIG_IP_VS_DEBUG
  46         /*
  47          * This is used for debug only.
  48          * Start with the most likely hit
  49          * End with BUG
  50          */
  51         if (likely(skb->dev && skb->dev->nd_net))
  52                 return dev_net(skb->dev);
  53         if (skb_dst(skb) && skb_dst(skb)->dev)
  54                 return dev_net(skb_dst(skb)->dev);
  55         WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
  56                       __func__, __LINE__);
  57         if (likely(skb->sk && skb->sk->sk_net))
  58                 return sock_net(skb->sk);
  59         pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
  60                 __func__, __LINE__);
  61         BUG();
  62 #else
  63         return dev_net(skb->dev ? : skb_dst(skb)->dev);
  64 #endif
  65 #else
  66         return &init_net;
  67 #endif
  68 }
  69
  70 static inline struct net *skb_sknet(const struct sk_buff *skb)
  71 {
  72 #ifdef CONFIG_NET_NS
  73 #ifdef CONFIG_IP_VS_DEBUG
  74         /* Start with the most likely hit */
  75         if (likely(skb->sk && skb->sk->sk_net))
  76                 return sock_net(skb->sk);
  77         WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
  78                        __func__, __LINE__);
  79         if (likely(skb->dev && skb->dev->nd_net))
  80                 return dev_net(skb->dev);
  81         pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
  82                 __func__, __LINE__);
  83         BUG();
  84 #else
  85         return sock_net(skb->sk);
  86 #endif
  87 #else
  88         return &init_net;
  89 #endif
  90 }
  91 /*
  92  * This one needed for single_open_net since net is stored directly in
  93  * private not as a struct i.e. seq_file_net can't be used.
  94  */
  95 static inline struct net *seq_file_single_net(struct seq_file *seq)
  96 {
  97 #ifdef CONFIG_NET_NS
  98         return (struct net *)seq->private;
  99 #else
 100         return &init_net;
 101 #endif
 102 }
 103
 104 /* Connections' size value needed by ip_vs_ctl.c */
 105 extern int ip_vs_conn_tab_size;
 106
 107
 108 struct ip_vs_iphdr {
 109         int len;
 110         __u8 protocol;
 111         union nf_inet_addr saddr;
 112         union nf_inet_addr daddr;
 113 };
 114
 115 static inline void
 116 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
 117 {
 118 #ifdef CONFIG_IP_VS_IPV6
 119         if (af == AF_INET6) {
 120                 const struct ipv6hdr *iph = nh;
 121                 iphdr->len = sizeof(struct ipv6hdr);
 122                 iphdr->protocol = iph->nexthdr;
 123                 iphdr->saddr.in6 = iph->saddr;
 124                 iphdr->daddr.in6 = iph->daddr;
 125         } else
 126 #endif
 127         {
 128                 const struct iphdr *iph = nh;
 129                 iphdr->len = iph->ihl * 4;
 130                 iphdr->protocol = iph->protocol;
 131                 iphdr->saddr.ip = iph->saddr;
 132                 iphdr->daddr.ip = iph->daddr;
 133         }
 134 }
 135
 136 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
 137                                    const union nf_inet_addr *src)
 138 {
 139 #ifdef CONFIG_IP_VS_IPV6
 140         if (af == AF_INET6)
 141                 dst->in6 = src->in6;
 142         else
 143 #endif
 144         dst->ip = src->ip;
 145 }
 146
 147 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
 148                                    const union nf_inet_addr *b)
 149 {
 150 #ifdef CONFIG_IP_VS_IPV6
 151         if (af == AF_INET6)
 152                 return ipv6_addr_equal(&a->in6, &b->in6);
 153 #endif
 154         return a->ip == b->ip;
 155 }
 156
 157 #ifdef CONFIG_IP_VS_DEBUG
 158 #include <linux/net.h>
 159
 160 extern int ip_vs_get_debug_level(void);
 161
 162 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
 163                                          const union nf_inet_addr *addr,
 164                                          int *idx)
 165 {
 166         int len;
 167 #ifdef CONFIG_IP_VS_IPV6
 168         if (af == AF_INET6)
 169                 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
 170                                &addr->in6) + 1;
 171         else
 172 #endif
 173                 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
 174                                &addr->ip) + 1;
 175
 176         *idx += len;
 177         BUG_ON(*idx > buf_len + 1);
 178         return &buf[*idx - len];
 179 }
 180
 181 #define IP_VS_DBG_BUF(level, msg, ...)                                  \
 182         do {                                                            \
 183                 char ip_vs_dbg_buf[160];                                \
 184                 int ip_vs_dbg_idx = 0;                                  \
 185                 if (level <= ip_vs_get_debug_level())                   \
 186                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 187         } while (0)
 188 #define IP_VS_ERR_BUF(msg...)                                           \
 189         do {                                                            \
 190                 char ip_vs_dbg_buf[160];                                \
 191                 int ip_vs_dbg_idx = 0;                                  \
 192                 pr_err(msg);                                            \
 193         } while (0)
 194
 195 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
 196 #define IP_VS_DBG_ADDR(af, addr)                                        \
 197         ip_vs_dbg_addr(af, ip_vs_dbg_buf,                               \
 198                        sizeof(ip_vs_dbg_buf), addr,                     \
 199                        &ip_vs_dbg_idx)
 200
 201 #define IP_VS_DBG(level, msg, ...)                                      \
 202         do {                                                            \
 203                 if (level <= ip_vs_get_debug_level())                   \
 204                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 205         } while (0)
 206 #define IP_VS_DBG_RL(msg, ...)                                          \
 207         do {                                                            \
 208                 if (net_ratelimit())                                    \
 209                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 210         } while (0)
 211 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)                     \
 212         do {                                                            \
 213                 if (level <= ip_vs_get_debug_level())                   \
 214                         pp->debug_packet(af, pp, skb, ofs, msg);        \
 215         } while (0)
 216 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)                  \
 217         do {                                                            \
 218                 if (level <= ip_vs_get_debug_level() &&                 \
 219                     net_ratelimit())                                    \
 220                         pp->debug_packet(af, pp, skb, ofs, msg);        \
 221         } while (0)
 222 #else   /* NO DEBUGGING at ALL */
 223 #define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
 224 #define IP_VS_ERR_BUF(msg...)  do {} while (0)
 225 #define IP_VS_DBG(level, msg...)  do {} while (0)
 226 #define IP_VS_DBG_RL(msg...)  do {} while (0)
 227 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)     do {} while (0)
 228 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)  do {} while (0)
 229 #endif
 230
 231 #define IP_VS_BUG() BUG()
 232 #define IP_VS_ERR_RL(msg, ...)                                          \
 233         do {                                                            \
 234                 if (net_ratelimit())                                    \
 235                         pr_err(msg, ##__VA_ARGS__);                     \
 236         } while (0)
 237
 238 #ifdef CONFIG_IP_VS_DEBUG
 239 #define EnterFunction(level)                                            \
 240         do {                                                            \
 241                 if (level <= ip_vs_get_debug_level())                   \
 242                         printk(KERN_DEBUG                               \
 243                                pr_fmt("Enter: %s, %s line %i\n"),       \
 244                                __func__, __FILE__, __LINE__);           \
 245         } while (0)
 246 #define LeaveFunction(level)                                            \
 247         do {                                                            \
 248                 if (level <= ip_vs_get_debug_level())                   \
 249                         printk(KERN_DEBUG                               \
 250                                pr_fmt("Leave: %s, %s line %i\n"),       \
 251                                __func__, __FILE__, __LINE__);           \
 252         } while (0)
 253 #else
 254 #define EnterFunction(level)   do {} while (0)
 255 #define LeaveFunction(level)   do {} while (0)
 256 #endif
 257
 258 #define IP_VS_WAIT_WHILE(expr)  while (expr) { cpu_relax(); }
 259
 260
 261 /*
 262  *      The port number of FTP service (in network order).
 263  */
 264 #define FTPPORT  cpu_to_be16(21)
 265 #define FTPDATA  cpu_to_be16(20)
 266
 267 /*
 268  *      TCP State Values
 269  */
 270 enum {
 271         IP_VS_TCP_S_NONE = 0,
 272         IP_VS_TCP_S_ESTABLISHED,
 273         IP_VS_TCP_S_SYN_SENT,
 274         IP_VS_TCP_S_SYN_RECV,
 275         IP_VS_TCP_S_FIN_WAIT,
 276         IP_VS_TCP_S_TIME_WAIT,
 277         IP_VS_TCP_S_CLOSE,
 278         IP_VS_TCP_S_CLOSE_WAIT,
 279         IP_VS_TCP_S_LAST_ACK,
 280         IP_VS_TCP_S_LISTEN,
 281         IP_VS_TCP_S_SYNACK,
 282         IP_VS_TCP_S_LAST
 283 };
 284
 285 /*
 286  *      UDP State Values
 287  */
 288 enum {
 289         IP_VS_UDP_S_NORMAL,
 290         IP_VS_UDP_S_LAST,
 291 };
 292
 293 /*
 294  *      ICMP State Values
 295  */
 296 enum {
 297         IP_VS_ICMP_S_NORMAL,
 298         IP_VS_ICMP_S_LAST,
 299 };
 300
 301 /*
 302  *      SCTP State Values
 303  */
 304 enum ip_vs_sctp_states {
 305         IP_VS_SCTP_S_NONE,
 306         IP_VS_SCTP_S_INIT_CLI,
 307         IP_VS_SCTP_S_INIT_SER,
 308         IP_VS_SCTP_S_INIT_ACK_CLI,
 309         IP_VS_SCTP_S_INIT_ACK_SER,
 310         IP_VS_SCTP_S_ECHO_CLI,
 311         IP_VS_SCTP_S_ECHO_SER,
 312         IP_VS_SCTP_S_ESTABLISHED,
 313         IP_VS_SCTP_S_SHUT_CLI,
 314         IP_VS_SCTP_S_SHUT_SER,
 315         IP_VS_SCTP_S_SHUT_ACK_CLI,
 316         IP_VS_SCTP_S_SHUT_ACK_SER,
 317         IP_VS_SCTP_S_CLOSED,
 318         IP_VS_SCTP_S_LAST
 319 };
 320
 321 /*
 322  *      Delta sequence info structure
 323  *      Each ip_vs_conn has 2 (output AND input seq. changes).
 324  *      Only used in the VS/NAT.
 325  */
 326 struct ip_vs_seq {
 327         __u32                   init_seq;       /* Add delta from this seq */
 328         __u32                   delta;          /* Delta in sequence numbers */
 329         __u32                   previous_delta; /* Delta in sequence numbers
 330                                                    before last resized pkt */
 331 };
 332
 333 /*
 334  * counters per cpu
 335  */
 336 struct ip_vs_counters {
 337         __u32           conns;          /* connections scheduled */
 338         __u32           inpkts;         /* incoming packets */
 339         __u32           outpkts;        /* outgoing packets */
 340         __u64           inbytes;        /* incoming bytes */
 341         __u64           outbytes;       /* outgoing bytes */
 342 };
 343 /*
 344  * Stats per cpu
 345  */
 346 struct ip_vs_cpu_stats {
 347         struct ip_vs_counters   ustats;
 348         struct u64_stats_sync   syncp;
 349 };
 350
 351 /*
 352  *      IPVS statistics objects
 353  */
 354 struct ip_vs_estimator {
 355         struct list_head        list;
 356
 357         u64                     last_inbytes;
 358         u64                     last_outbytes;
 359         u32                     last_conns;
 360         u32                     last_inpkts;
 361         u32                     last_outpkts;
 362
 363         u32                     cps;
 364         u32                     inpps;
 365         u32                     outpps;
 366         u32                     inbps;
 367         u32                     outbps;
 368 };
 369
 370 struct ip_vs_stats {
 371         struct ip_vs_stats_user ustats;         /* statistics */
 372         struct ip_vs_estimator  est;            /* estimator */
 373         struct ip_vs_cpu_stats  *cpustats;      /* per cpu counters */
 374         spinlock_t              lock;           /* spin lock */
 375         struct ip_vs_stats_user ustats0;        /* reset values */
 376 };
 377
 378 struct dst_entry;
 379 struct iphdr;
 380 struct ip_vs_conn;
 381 struct ip_vs_app;
 382 struct sk_buff;
 383 struct ip_vs_proto_data;
 384
 385 struct ip_vs_protocol {
 386         struct ip_vs_protocol   *next;
 387         char                    *name;
 388         u16                     protocol;
 389         u16                     num_states;
 390         int                     dont_defrag;
 391
 392         void (*init)(struct ip_vs_protocol *pp);
 393
 394         void (*exit)(struct ip_vs_protocol *pp);
 395
 396         int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
 397
 398         void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
 399
 400         int (*conn_schedule)(int af, struct sk_buff *skb,
 401                              struct ip_vs_proto_data *pd,
 402                              int *verdict, struct ip_vs_conn **cpp);
 403
 404         struct ip_vs_conn *
 405         (*conn_in_get)(int af,
 406                        const struct sk_buff *skb,
 407                        const struct ip_vs_iphdr *iph,
 408                        unsigned int proto_off,
 409                        int inverse);
 410
 411         struct ip_vs_conn *
 412         (*conn_out_get)(int af,
 413                         const struct sk_buff *skb,
 414                         const struct ip_vs_iphdr *iph,
 415                         unsigned int proto_off,
 416                         int inverse);
 417
 418         int (*snat_handler)(struct sk_buff *skb,
 419                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 420
 421         int (*dnat_handler)(struct sk_buff *skb,
 422                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 423
 424         int (*csum_check)(int af, struct sk_buff *skb,
 425                           struct ip_vs_protocol *pp);
 426
 427         const char *(*state_name)(int state);
 428
 429         void (*state_transition)(struct ip_vs_conn *cp, int direction,
 430                                  const struct sk_buff *skb,
 431                                  struct ip_vs_proto_data *pd);
 432
 433         int (*register_app)(struct net *net, struct ip_vs_app *inc);
 434
 435         void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
 436
 437         int (*app_conn_bind)(struct ip_vs_conn *cp);
 438
 439         void (*debug_packet)(int af, struct ip_vs_protocol *pp,
 440                              const struct sk_buff *skb,
 441                              int offset,
 442                              const char *msg);
 443
 444         void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
 445 };
 446
 447 /*
 448  * protocol data per netns
 449  */
 450 struct ip_vs_proto_data {
 451         struct ip_vs_proto_data *next;
 452         struct ip_vs_protocol   *pp;
 453         int                     *timeout_table; /* protocol timeout table */
 454         atomic_t                appcnt;         /* counter of proto app incs. */
 455         struct tcp_states_t     *tcp_state_table;
 456 };
 457
 458 extern struct ip_vs_protocol   *ip_vs_proto_get(unsigned short proto);
 459 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
 460                                                      unsigned short proto);
 461
 462 struct ip_vs_conn_param {
 463         struct net                      *net;
 464         const union nf_inet_addr        *caddr;
 465         const union nf_inet_addr        *vaddr;
 466         __be16                          cport;
 467         __be16                          vport;
 468         __u16                           protocol;
 469         u16                             af;
 470
 471         const struct ip_vs_pe           *pe;
 472         char                            *pe_data;
 473         __u8                            pe_data_len;
 474 };
 475
 476 /*
 477  *      IP_VS structure allocated for each dynamically scheduled connection
 478  */
 479 struct ip_vs_conn {
 480         struct hlist_node       c_list;         /* hashed list heads */
 481 #ifdef CONFIG_NET_NS
 482         struct net              *net;           /* Name space */
 483 #endif
 484         /* Protocol, addresses and port numbers */
 485         u16                     af;             /* address family */
 486         __be16                  cport;
 487         __be16                  vport;
 488         __be16                  dport;
 489         __u32                   fwmark;         /* Fire wall mark from skb */
 490         union nf_inet_addr      caddr;          /* client address */
 491         union nf_inet_addr      vaddr;          /* virtual address */
 492         union nf_inet_addr      daddr;          /* destination address */
 493         volatile __u32          flags;          /* status flags */
 494         __u16                   protocol;       /* Which protocol (TCP/UDP) */
 495
 496         /* counter and timer */
 497         atomic_t                refcnt;         /* reference count */
 498         struct timer_list       timer;          /* Expiration timer */
 499         volatile unsigned long  timeout;        /* timeout */
 500
 501         /* Flags and state transition */
 502         spinlock_t              lock;           /* lock for state transition */
 503         volatile __u16          state;          /* state info */
 504         volatile __u16          old_state;      /* old state, to be used for
 505                                                  * state transition triggerd
 506                                                  * synchronization
 507                                                  */
 508
 509         /* Control members */
 510         struct ip_vs_conn       *control;       /* Master control connection */
 511         atomic_t                n_control;      /* Number of controlled ones */
 512         struct ip_vs_dest       *dest;          /* real server */
 513         atomic_t                in_pkts;        /* incoming packet counter */
 514
 515         /* packet transmitter for different forwarding methods.  If it
 516            mangles the packet, it must return NF_DROP or better NF_STOLEN,
 517            otherwise this must be changed to a sk_buff **.
 518            NF_ACCEPT can be returned when destination is local.
 519          */
 520         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
 521                            struct ip_vs_protocol *pp);
 522
 523         /* Note: we can group the following members into a structure,
 524            in order to save more space, and the following members are
 525            only used in VS/NAT anyway */
 526         struct ip_vs_app        *app;           /* bound ip_vs_app object */
 527         void                    *app_data;      /* Application private data */
 528         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
 529         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
 530
 531         const struct ip_vs_pe   *pe;
 532         char                    *pe_data;
 533         __u8                    pe_data_len;
 534 };
 535
 536 /*
 537  *  To save some memory in conn table when name space is disabled.
 538  */
 539 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
 540 {
 541 #ifdef CONFIG_NET_NS
 542         return cp->net;
 543 #else
 544         return &init_net;
 545 #endif
 546 }
 547 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
 548 {
 549 #ifdef CONFIG_NET_NS
 550         cp->net = net;
 551 #endif
 552 }
 553
 554 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
 555                                     struct net *net)
 556 {
 557 #ifdef CONFIG_NET_NS
 558         return cp->net == net;
 559 #else
 560         return 1;
 561 #endif
 562 }
 563
 564 /*
 565  *      Extended internal versions of struct ip_vs_service_user and
 566  *      ip_vs_dest_user for IPv6 support.
 567  *
 568  *      We need these to conveniently pass around service and destination
 569  *      options, but unfortunately, we also need to keep the old definitions to
 570  *      maintain userspace backwards compatibility for the setsockopt interface.
 571  */
 572 struct ip_vs_service_user_kern {
 573         /* virtual service addresses */
 574         u16                     af;
 575         u16                     protocol;
 576         union nf_inet_addr      addr;           /* virtual ip address */
 577         u16                     port;
 578         u32                     fwmark;         /* firwall mark of service */
 579
 580         /* virtual service options */
 581         char                    *sched_name;
 582         char                    *pe_name;
 583         unsigned                flags;          /* virtual service flags */
 584         unsigned                timeout;        /* persistent timeout in sec */
 585         u32                     netmask;        /* persistent netmask */
 586 };
 587
 588
 589 struct ip_vs_dest_user_kern {
 590         /* destination server address */
 591         union nf_inet_addr      addr;
 592         u16                     port;
 593
 594         /* real server options */
 595         unsigned                conn_flags;     /* connection flags */
 596         int                     weight;         /* destination weight */
 597
 598         /* thresholds for active connections */
 599         u32                     u_threshold;    /* upper threshold */
 600         u32                     l_threshold;    /* lower threshold */
 601 };
 602
 603
 604 /*
 605  *      The information about the virtual service offered to the net
 606  *      and the forwarding entries
 607  */
 608 struct ip_vs_service {
 609         struct list_head        s_list;   /* for normal service table */
 610         struct list_head        f_list;   /* for fwmark-based service table */
 611         atomic_t                refcnt;   /* reference counter */
 612         atomic_t                usecnt;   /* use counter */
 613
 614         u16                     af;       /* address family */
 615         __u16                   protocol; /* which protocol (TCP/UDP) */
 616         union nf_inet_addr      addr;     /* IP address for virtual service */
 617         __be16                  port;     /* port number for the service */
 618         __u32                   fwmark;   /* firewall mark of the service */
 619         unsigned                flags;    /* service status flags */
 620         unsigned                timeout;  /* persistent timeout in ticks */
 621         __be32                  netmask;  /* grouping granularity */
 622         struct net              *net;
 623
 624         struct list_head        destinations;  /* real server d-linked list */
 625         __u32                   num_dests;     /* number of servers */
 626         struct ip_vs_stats      stats;         /* statistics for the service */
 627         struct ip_vs_app        *inc;     /* bind conns to this app inc */
 628
 629         /* for scheduling */
 630         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
 631         rwlock_t                sched_lock;    /* lock sched_data */
 632         void                    *sched_data;   /* scheduler application data */
 633
 634         /* alternate persistence engine */
 635         struct ip_vs_pe         *pe;
 636 };
 637
 638
 639 /*
 640  *      The real server destination forwarding entry
 641  *      with ip address, port number, and so on.
 642  */
 643 struct ip_vs_dest {
 644         struct list_head        n_list;   /* for the dests in the service */
 645         struct list_head        d_list;   /* for table with all the dests */
 646
 647         u16                     af;             /* address family */
 648         __be16                  port;           /* port number of the server */
 649         union nf_inet_addr      addr;           /* IP address of the server */
 650         volatile unsigned       flags;          /* dest status flags */
 651         atomic_t                conn_flags;     /* flags to copy to conn */
 652         atomic_t                weight;         /* server weight */
 653
 654         atomic_t                refcnt;         /* reference counter */
 655         struct ip_vs_stats      stats;          /* statistics */
 656
 657         /* connection counters and thresholds */
 658         atomic_t                activeconns;    /* active connections */
 659         atomic_t                inactconns;     /* inactive connections */
 660         atomic_t                persistconns;   /* persistent connections */
 661         __u32                   u_threshold;    /* upper threshold */
 662         __u32                   l_threshold;    /* lower threshold */
 663
 664         /* for destination cache */
 665         spinlock_t              dst_lock;       /* lock of dst_cache */
 666         struct dst_entry        *dst_cache;     /* destination cache entry */
 667         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
 668         u32                     dst_cookie;
 669         union nf_inet_addr      dst_saddr;
 670
 671         /* for virtual service */
 672         struct ip_vs_service    *svc;           /* service it belongs to */
 673         __u16                   protocol;       /* which protocol (TCP/UDP) */
 674         __be16                  vport;          /* virtual port number */
 675         union nf_inet_addr      vaddr;          /* virtual IP address */
 676         __u32                   vfwmark;        /* firewall mark of service */
 677 };
 678
 679
 680 /*
 681  *      The scheduler object
 682  */
 683 struct ip_vs_scheduler {
 684         struct list_head        n_list;         /* d-linked list head */
 685         char                    *name;          /* scheduler name */
 686         atomic_t                refcnt;         /* reference counter */
 687         struct module           *module;        /* THIS_MODULE/NULL */
 688
 689         /* scheduler initializing service */
 690         int (*init_service)(struct ip_vs_service *svc);
 691         /* scheduling service finish */
 692         int (*done_service)(struct ip_vs_service *svc);
 693         /* scheduler updating service */
 694         int (*update_service)(struct ip_vs_service *svc);
 695
 696         /* selecting a server from the given service */
 697         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 698                                        const struct sk_buff *skb);
 699 };
 700
 701 /* The persistence engine object */
 702 struct ip_vs_pe {
 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         /* get the connection template, if any */
 709         int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
 710         bool (*ct_match)(const struct ip_vs_conn_param *p,
 711                          struct ip_vs_conn *ct);
 712         u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
 713                            bool inverse);
 714         int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
 715 };
 716
 717 /*
 718  *      The application module object (a.k.a. app incarnation)
 719  */
 720 struct ip_vs_app {
 721         struct list_head        a_list;         /* member in app list */
 722         int                     type;           /* IP_VS_APP_TYPE_xxx */
 723         char                    *name;          /* application module name */
 724         __u16                   protocol;
 725         struct module           *module;        /* THIS_MODULE/NULL */
 726         struct list_head        incs_list;      /* list of incarnations */
 727
 728         /* members for application incarnations */
 729         struct list_head        p_list;         /* member in proto app list */
 730         struct ip_vs_app        *app;           /* its real application */
 731         __be16                  port;           /* port number in net order */
 732         atomic_t                usecnt;         /* usage counter */
 733
 734         /*
 735          * output hook: Process packet in inout direction, diff set for TCP.
 736          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 737          *         2=Mangled but checksum was not updated
 738          */
 739         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 740                        struct sk_buff *, int *diff);
 741
 742         /*
 743          * input hook: Process packet in outin direction, diff set for TCP.
 744          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 745          *         2=Mangled but checksum was not updated
 746          */
 747         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 748                       struct sk_buff *, int *diff);
 749
 750         /* ip_vs_app initializer */
 751         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 752
 753         /* ip_vs_app finish */
 754         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 755
 756
 757         /* not used now */
 758         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 759                          struct ip_vs_protocol *);
 760
 761         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 762
 763         int *                   timeout_table;
 764         int *                   timeouts;
 765         int                     timeouts_size;
 766
 767         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 768                              int *verdict, struct ip_vs_conn **cpp);
 769
 770         struct ip_vs_conn *
 771         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 772                        const struct iphdr *iph, unsigned int proto_off,
 773                        int inverse);
 774
 775         struct ip_vs_conn *
 776         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 777                         const struct iphdr *iph, unsigned int proto_off,
 778                         int inverse);
 779
 780         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 781                                 const struct sk_buff *skb,
 782                                 struct ip_vs_app *app);
 783
 784         void (*timeout_change)(struct ip_vs_app *app, int flags);
 785 };
 786
 787 /* IPVS in network namespace */
 788 struct netns_ipvs {
 789         int                     gen;            /* Generation */
 790         int                     enable;         /* enable like nf_hooks do */
 791         /*
 792          *      Hash table: for real service lookups
 793          */
 794         #define IP_VS_RTAB_BITS 4
 795         #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
 796         #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
 797
 798         struct list_head        rs_table[IP_VS_RTAB_SIZE];
 799         /* ip_vs_app */
 800         struct list_head        app_list;
 801         /* ip_vs_ftp */
 802         struct ip_vs_app        *ftp_app;
 803         /* ip_vs_proto */
 804         #define IP_VS_PROTO_TAB_SIZE    32      /* must be power of 2 */
 805         struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
 806         /* ip_vs_proto_tcp */
 807 #ifdef CONFIG_IP_VS_PROTO_TCP
 808         #define TCP_APP_TAB_BITS        4
 809         #define TCP_APP_TAB_SIZE        (1 << TCP_APP_TAB_BITS)
 810         #define TCP_APP_TAB_MASK        (TCP_APP_TAB_SIZE - 1)
 811         struct list_head        tcp_apps[TCP_APP_TAB_SIZE];
 812         spinlock_t              tcp_app_lock;
 813 #endif
 814         /* ip_vs_proto_udp */
 815 #ifdef CONFIG_IP_VS_PROTO_UDP
 816         #define UDP_APP_TAB_BITS        4
 817         #define UDP_APP_TAB_SIZE        (1 << UDP_APP_TAB_BITS)
 818         #define UDP_APP_TAB_MASK        (UDP_APP_TAB_SIZE - 1)
 819         struct list_head        udp_apps[UDP_APP_TAB_SIZE];
 820         spinlock_t              udp_app_lock;
 821 #endif
 822         /* ip_vs_proto_sctp */
 823 #ifdef CONFIG_IP_VS_PROTO_SCTP
 824         #define SCTP_APP_TAB_BITS       4
 825         #define SCTP_APP_TAB_SIZE       (1 << SCTP_APP_TAB_BITS)
 826         #define SCTP_APP_TAB_MASK       (SCTP_APP_TAB_SIZE - 1)
 827         /* Hash table for SCTP application incarnations  */
 828         struct list_head        sctp_apps[SCTP_APP_TAB_SIZE];
 829         spinlock_t              sctp_app_lock;
 830 #endif
 831         /* ip_vs_conn */
 832         atomic_t                conn_count;      /*  connection counter */
 833
 834         /* ip_vs_ctl */
 835         struct ip_vs_stats              tot_stats;  /* Statistics & est. */
 836
 837         int                     num_services;    /* no of virtual services */
 838
 839         rwlock_t                rs_lock;         /* real services table */
 840         /* Trash for destinations */
 841         struct list_head        dest_trash;
 842         /* Service counters */
 843         atomic_t                ftpsvc_counter;
 844         atomic_t                nullsvc_counter;
 845
 846 #ifdef CONFIG_SYSCTL
 847         /* 1/rate drop and drop-entry variables */
 848         struct delayed_work     defense_work;   /* Work handler */
 849         int                     drop_rate;
 850         int                     drop_counter;
 851         atomic_t                dropentry;
 852         /* locks in ctl.c */
 853         spinlock_t              dropentry_lock;  /* drop entry handling */
 854         spinlock_t              droppacket_lock; /* drop packet handling */
 855         spinlock_t              securetcp_lock;  /* state and timeout tables */
 856
 857         /* sys-ctl struct */
 858         struct ctl_table_header *sysctl_hdr;
 859         struct ctl_table        *sysctl_tbl;
 860 #endif
 861
 862         /* sysctl variables */
 863         int                     sysctl_amemthresh;
 864         int                     sysctl_am_droprate;
 865         int                     sysctl_drop_entry;
 866         int                     sysctl_drop_packet;
 867         int                     sysctl_secure_tcp;
 868 #ifdef CONFIG_IP_VS_NFCT
 869         int                     sysctl_conntrack;
 870 #endif
 871         int                     sysctl_snat_reroute;
 872         int                     sysctl_sync_ver;
 873         int                     sysctl_cache_bypass;
 874         int                     sysctl_expire_nodest_conn;
 875         int                     sysctl_expire_quiescent_template;
 876         int                     sysctl_sync_threshold[2];
 877         int                     sysctl_nat_icmp_send;
 878
 879         /* ip_vs_lblc */
 880         int                     sysctl_lblc_expiration;
 881         struct ctl_table_header *lblc_ctl_header;
 882         struct ctl_table        *lblc_ctl_table;
 883         /* ip_vs_lblcr */
 884         int                     sysctl_lblcr_expiration;
 885         struct ctl_table_header *lblcr_ctl_header;
 886         struct ctl_table        *lblcr_ctl_table;
 887         /* ip_vs_est */
 888         struct list_head        est_list;       /* estimator list */
 889         spinlock_t              est_lock;
 890         struct timer_list       est_timer;      /* Estimation timer */
 891         /* ip_vs_sync */
 892         struct list_head        sync_queue;
 893         spinlock_t              sync_lock;
 894         struct ip_vs_sync_buff  *sync_buff;
 895         spinlock_t              sync_buff_lock;
 896         struct sockaddr_in      sync_mcast_addr;
 897         struct task_struct      *master_thread;
 898         struct task_struct      *backup_thread;
 899         int                     send_mesg_maxlen;
 900         int                     recv_mesg_maxlen;
 901         volatile int            sync_state;
 902         volatile int            master_syncid;
 903         volatile int            backup_syncid;
 904         struct mutex            sync_mutex;
 905         /* multicast interface name */
 906         char                    master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 907         char                    backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 908         /* net name space ptr */
 909         struct net              *net;            /* Needed by timer routines */
 910 };
 911
 912 #define DEFAULT_SYNC_THRESHOLD  3
 913 #define DEFAULT_SYNC_PERIOD     50
 914 #define DEFAULT_SYNC_VER        1
 915
 916 #ifdef CONFIG_SYSCTL
 917
 918 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
 919 {
 920         return ipvs->sysctl_sync_threshold[0];
 921 }
 922
 923 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
 924 {
 925         return ipvs->sysctl_sync_threshold[1];
 926 }
 927
 928 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
 929 {
 930         return ipvs->sysctl_sync_ver;
 931 }
 932
 933 #else
 934
 935 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
 936 {
 937         return DEFAULT_SYNC_THRESHOLD;
 938 }
 939
 940 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
 941 {
 942         return DEFAULT_SYNC_PERIOD;
 943 }
 944
 945 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
 946 {
 947         return DEFAULT_SYNC_VER;
 948 }
 949
 950 #endif
 951
 952 /*
 953  *      IPVS core functions
 954  *      (from ip_vs_core.c)
 955  */
 956 extern const char *ip_vs_proto_name(unsigned proto);
 957 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
 958 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
 959
 960 #define IP_VS_APP_TYPE_FTP      1
 961
 962 /*
 963  *     ip_vs_conn handling functions
 964  *     (from ip_vs_conn.c)
 965  */
 966
 967 enum {
 968         IP_VS_DIR_INPUT = 0,
 969         IP_VS_DIR_OUTPUT,
 970         IP_VS_DIR_INPUT_ONLY,
 971         IP_VS_DIR_LAST,
 972 };
 973
 974 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
 975                                          const union nf_inet_addr *caddr,
 976                                          __be16 cport,
 977                                          const union nf_inet_addr *vaddr,
 978                                          __be16 vport,
 979                                          struct ip_vs_conn_param *p)
 980 {
 981         p->net = net;
 982         p->af = af;
 983         p->protocol = protocol;
 984         p->caddr = caddr;
 985         p->cport = cport;
 986         p->vaddr = vaddr;
 987         p->vport = vport;
 988         p->pe = NULL;
 989         p->pe_data = NULL;
 990 }
 991
 992 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
 993 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
 994
 995 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
 996                                             const struct ip_vs_iphdr *iph,
 997                                             unsigned int proto_off,
 998                                             int inverse);
 999
1000 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1001
1002 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1003                                              const struct ip_vs_iphdr *iph,
1004                                              unsigned int proto_off,
1005                                              int inverse);
1006
1007 /* put back the conn without restarting its timer */
1008 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1009 {
1010         atomic_dec(&cp->refcnt);
1011 }
1012 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1013 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1014
1015 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1016                                   const union nf_inet_addr *daddr,
1017                                   __be16 dport, unsigned flags,
1018                                   struct ip_vs_dest *dest, __u32 fwmark);
1019 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1020
1021 extern const char * ip_vs_state_name(__u16 proto, int state);
1022
1023 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1024 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1025 extern void ip_vs_random_dropentry(struct net *net);
1026 extern int ip_vs_conn_init(void);
1027 extern void ip_vs_conn_cleanup(void);
1028
1029 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1030 {
1031         struct ip_vs_conn *ctl_cp = cp->control;
1032         if (!ctl_cp) {
1033                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1034                               "%s:%d to %s:%d\n",
1035                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1036                               ntohs(cp->cport),
1037                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1038                               ntohs(cp->vport));
1039
1040                 return;
1041         }
1042
1043         IP_VS_DBG_BUF(7, "DELeting control for: "
1044                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1045                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1046                       ntohs(cp->cport),
1047                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1048                       ntohs(ctl_cp->cport));
1049
1050         cp->control = NULL;
1051         if (atomic_read(&ctl_cp->n_control) == 0) {
1052                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1053                               "%s:%d to %s:%d\n",
1054                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1055                               ntohs(cp->cport),
1056                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1057                               ntohs(cp->vport));
1058
1059                 return;
1060         }
1061         atomic_dec(&ctl_cp->n_control);
1062 }
1063
1064 static inline void
1065 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1066 {
1067         if (cp->control) {
1068                 IP_VS_ERR_BUF("request control ADD for already controlled: "
1069                               "%s:%d to %s:%d\n",
1070                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1071                               ntohs(cp->cport),
1072                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1073                               ntohs(cp->vport));
1074
1075                 ip_vs_control_del(cp);
1076         }
1077
1078         IP_VS_DBG_BUF(7, "ADDing control for: "
1079                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1080                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1081                       ntohs(cp->cport),
1082                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1083                       ntohs(ctl_cp->cport));
1084
1085         cp->control = ctl_cp;
1086         atomic_inc(&ctl_cp->n_control);
1087 }
1088
1089 /*
1090  * IPVS netns init & cleanup functions
1091  */
1092 extern int ip_vs_estimator_net_init(struct net *net);
1093 extern int ip_vs_control_net_init(struct net *net);
1094 extern int ip_vs_protocol_net_init(struct net *net);
1095 extern int ip_vs_app_net_init(struct net *net);
1096 extern int ip_vs_conn_net_init(struct net *net);
1097 extern int ip_vs_sync_net_init(struct net *net);
1098 extern void ip_vs_conn_net_cleanup(struct net *net);
1099 extern void ip_vs_app_net_cleanup(struct net *net);
1100 extern void ip_vs_protocol_net_cleanup(struct net *net);
1101 extern void ip_vs_control_net_cleanup(struct net *net);
1102 extern void ip_vs_estimator_net_cleanup(struct net *net);
1103 extern void ip_vs_sync_net_cleanup(struct net *net);
1104 extern void ip_vs_service_net_cleanup(struct net *net);
1105
1106 /*
1107  *      IPVS application functions
1108  *      (from ip_vs_app.c)
1109  */
1110 #define IP_VS_APP_MAX_PORTS  8
1111 extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1112 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1113 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1114 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1115 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1116                                   __u16 proto, __u16 port);
1117 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1118 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1119
1120 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1121 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1122
1123 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1124 void ip_vs_unbind_pe(struct ip_vs_service *svc);
1125 int register_ip_vs_pe(struct ip_vs_pe *pe);
1126 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1127 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1128 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1129
1130 /*
1131  * Use a #define to avoid all of module.h just for these trivial ops
1132  */
1133 #define ip_vs_pe_get(pe)                        \
1134         if (pe && pe->module)                   \
1135                 __module_get(pe->module);
1136
1137 #define ip_vs_pe_put(pe)                        \
1138         if (pe && pe->module)                   \
1139                 module_put(pe->module);
1140
1141 /*
1142  *      IPVS protocol functions (from ip_vs_proto.c)
1143  */
1144 extern int ip_vs_protocol_init(void);
1145 extern void ip_vs_protocol_cleanup(void);
1146 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1147 extern int *ip_vs_create_timeout_table(int *table, int size);
1148 extern int
1149 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1150                         const char *name, int to);
1151 extern void
1152 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1153                           const struct sk_buff *skb,
1154                           int offset, const char *msg);
1155
1156 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1157 extern struct ip_vs_protocol ip_vs_protocol_udp;
1158 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1159 extern struct ip_vs_protocol ip_vs_protocol_esp;
1160 extern struct ip_vs_protocol ip_vs_protocol_ah;
1161 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1162
1163 /*
1164  *      Registering/unregistering scheduler functions
1165  *      (from ip_vs_sched.c)
1166  */
1167 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1168 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1169 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1170                                 struct ip_vs_scheduler *scheduler);
1171 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1172 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1173 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1174 extern struct ip_vs_conn *
1175 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1176                struct ip_vs_proto_data *pd, int *ignored);
1177 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1178                         struct ip_vs_proto_data *pd);
1179
1180 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1181
1182
1183 /*
1184  *      IPVS control data and functions (from ip_vs_ctl.c)
1185  */
1186 extern struct ip_vs_stats ip_vs_stats;
1187 extern const struct ctl_path net_vs_ctl_path[];
1188 extern int sysctl_ip_vs_sync_ver;
1189
1190 extern void ip_vs_sync_switch_mode(struct net *net, int mode);
1191 extern struct ip_vs_service *
1192 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1193                   const union nf_inet_addr *vaddr, __be16 vport);
1194
1195 static inline void ip_vs_service_put(struct ip_vs_service *svc)
1196 {
1197         atomic_dec(&svc->usecnt);
1198 }
1199
1200 extern struct ip_vs_dest *
1201 ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
1202                           const union nf_inet_addr *daddr, __be16 dport);
1203
1204 extern int ip_vs_use_count_inc(void);
1205 extern void ip_vs_use_count_dec(void);
1206 extern int ip_vs_register_nl_ioctl(void);
1207 extern void ip_vs_unregister_nl_ioctl(void);
1208 extern int ip_vs_control_init(void);
1209 extern void ip_vs_control_cleanup(void);
1210 extern struct ip_vs_dest *
1211 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1212                 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1213                 __u16 protocol, __u32 fwmark, __u32 flags);
1214 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1215
1216
1217 /*
1218  *      IPVS sync daemon data and function prototypes
1219  *      (from ip_vs_sync.c)
1220  */
1221 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1222                              __u8 syncid);
1223 extern int stop_sync_thread(struct net *net, int state);
1224 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
1225
1226
1227 /*
1228  *      IPVS rate estimator prototypes (from ip_vs_est.c)
1229  */
1230 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1231 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1232 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1233 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1234                                  struct ip_vs_stats *stats);
1235
1236 /*
1237  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
1238  */
1239 extern int ip_vs_null_xmit
1240 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1241 extern int ip_vs_bypass_xmit
1242 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1243 extern int ip_vs_nat_xmit
1244 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1245 extern int ip_vs_tunnel_xmit
1246 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1247 extern int ip_vs_dr_xmit
1248 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1249 extern int ip_vs_icmp_xmit
1250 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1251  int offset, unsigned int hooknum);
1252 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1253
1254 #ifdef CONFIG_IP_VS_IPV6
1255 extern int ip_vs_bypass_xmit_v6
1256 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1257 extern int ip_vs_nat_xmit_v6
1258 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1259 extern int ip_vs_tunnel_xmit_v6
1260 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1261 extern int ip_vs_dr_xmit_v6
1262 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1263 extern int ip_vs_icmp_xmit_v6
1264 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1265  int offset, unsigned int hooknum);
1266 #endif
1267
1268 #ifdef CONFIG_SYSCTL
1269 /*
1270  *      This is a simple mechanism to ignore packets when
1271  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
1272  *      we start to drop 1/rate of the packets
1273  */
1274
1275 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1276 {
1277         if (!ipvs->drop_rate)
1278                 return 0;
1279         if (--ipvs->drop_counter > 0)
1280                 return 0;
1281         ipvs->drop_counter = ipvs->drop_rate;
1282         return 1;
1283 }
1284 #else
1285 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1286 #endif
1287
1288 /*
1289  *      ip_vs_fwd_tag returns the forwarding tag of the connection
1290  */
1291 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
1292
1293 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1294 {
1295         char fwd;
1296
1297         switch (IP_VS_FWD_METHOD(cp)) {
1298         case IP_VS_CONN_F_MASQ:
1299                 fwd = 'M'; break;
1300         case IP_VS_CONN_F_LOCALNODE:
1301                 fwd = 'L'; break;
1302         case IP_VS_CONN_F_TUNNEL:
1303                 fwd = 'T'; break;
1304         case IP_VS_CONN_F_DROUTE:
1305                 fwd = 'R'; break;
1306         case IP_VS_CONN_F_BYPASS:
1307                 fwd = 'B'; break;
1308         default:
1309                 fwd = '?'; break;
1310         }
1311         return fwd;
1312 }
1313
1314 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1315                            struct ip_vs_conn *cp, int dir);
1316
1317 #ifdef CONFIG_IP_VS_IPV6
1318 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1319                               struct ip_vs_conn *cp, int dir);
1320 #endif
1321
1322 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1323
1324 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1325 {
1326         __be32 diff[2] = { ~old, new };
1327
1328         return csum_partial(diff, sizeof(diff), oldsum);
1329 }
1330
1331 #ifdef CONFIG_IP_VS_IPV6
1332 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1333                                         __wsum oldsum)
1334 {
1335         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1336                             new[3],  new[2],  new[1],  new[0] };
1337
1338         return csum_partial(diff, sizeof(diff), oldsum);
1339 }
1340 #endif
1341
1342 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1343 {
1344         __be16 diff[2] = { ~old, new };
1345
1346         return csum_partial(diff, sizeof(diff), oldsum);
1347 }
1348
1349 /*
1350  * Forget current conntrack (unconfirmed) and attach notrack entry
1351  */
1352 static inline void ip_vs_notrack(struct sk_buff *skb)
1353 {
1354 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1355         enum ip_conntrack_info ctinfo;
1356         struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1357
1358         if (!ct || !nf_ct_is_untracked(ct)) {
1359                 nf_reset(skb);
1360                 skb->nfct = &nf_ct_untracked_get()->ct_general;
1361                 skb->nfctinfo = IP_CT_NEW;
1362                 nf_conntrack_get(skb->nfct);
1363         }
1364 #endif
1365 }
1366
1367 #ifdef CONFIG_IP_VS_NFCT
1368 /*
1369  *      Netfilter connection tracking
1370  *      (from ip_vs_nfct.c)
1371  */
1372 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1373 {
1374 #ifdef CONFIG_SYSCTL
1375         return ipvs->sysctl_conntrack;
1376 #else
1377         return 0;
1378 #endif
1379 }
1380
1381 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1382                                    int outin);
1383 extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1384 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1385                                       struct ip_vs_conn *cp, u_int8_t proto,
1386                                       const __be16 port, int from_rs);
1387 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1388
1389 #else
1390
1391 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1392 {
1393         return 0;
1394 }
1395
1396 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1397                                           struct ip_vs_conn *cp, int outin)
1398 {
1399 }
1400
1401 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1402 {
1403         return NF_ACCEPT;
1404 }
1405
1406 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1407 {
1408 }
1409 /* CONFIG_IP_VS_NFCT */
1410 #endif
1411
1412 static inline unsigned int
1413 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1414 {
1415         /*
1416          * We think the overhead of processing active connections is 256
1417          * times higher than that of inactive connections in average. (This
1418          * 256 times might not be accurate, we will change it later) We
1419          * use the following formula to estimate the overhead now:
1420          *                dest->activeconns*256 + dest->inactconns
1421          */
1422         return (atomic_read(&dest->activeconns) << 8) +
1423                 atomic_read(&dest->inactconns);
1424 }
1425
1426 #endif  /* _NET_IP_VS_H */