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) && 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 can't 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         struct ip_vs_stats_user ustats0;        /* reset values */
 378 };
 379
 380 struct dst_entry;
 381 struct iphdr;
 382 struct ip_vs_conn;
 383 struct ip_vs_app;
 384 struct sk_buff;
 385 struct ip_vs_proto_data;
 386
 387 struct ip_vs_protocol {
 388         struct ip_vs_protocol   *next;
 389         char                    *name;
 390         u16                     protocol;
 391         u16                     num_states;
 392         int                     dont_defrag;
 393
 394         void (*init)(struct ip_vs_protocol *pp);
 395
 396         void (*exit)(struct ip_vs_protocol *pp);
 397
 398         void (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
 399
 400         void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
 401
 402         int (*conn_schedule)(int af, struct sk_buff *skb,
 403                              struct ip_vs_proto_data *pd,
 404                              int *verdict, struct ip_vs_conn **cpp);
 405
 406         struct ip_vs_conn *
 407         (*conn_in_get)(int af,
 408                        const struct sk_buff *skb,
 409                        const struct ip_vs_iphdr *iph,
 410                        unsigned int proto_off,
 411                        int inverse);
 412
 413         struct ip_vs_conn *
 414         (*conn_out_get)(int af,
 415                         const struct sk_buff *skb,
 416                         const struct ip_vs_iphdr *iph,
 417                         unsigned int proto_off,
 418                         int inverse);
 419
 420         int (*snat_handler)(struct sk_buff *skb,
 421                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 422
 423         int (*dnat_handler)(struct sk_buff *skb,
 424                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 425
 426         int (*csum_check)(int af, struct sk_buff *skb,
 427                           struct ip_vs_protocol *pp);
 428
 429         const char *(*state_name)(int state);
 430
 431         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 432                                 const struct sk_buff *skb,
 433                                 struct ip_vs_proto_data *pd);
 434
 435         int (*register_app)(struct net *net, struct ip_vs_app *inc);
 436
 437         void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
 438
 439         int (*app_conn_bind)(struct ip_vs_conn *cp);
 440
 441         void (*debug_packet)(int af, struct ip_vs_protocol *pp,
 442                              const struct sk_buff *skb,
 443                              int offset,
 444                              const char *msg);
 445
 446         void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
 447 };
 448
 449 /*
 450  * protocol data per netns
 451  */
 452 struct ip_vs_proto_data {
 453         struct ip_vs_proto_data *next;
 454         struct ip_vs_protocol   *pp;
 455         int                     *timeout_table; /* protocol timeout table */
 456         atomic_t                appcnt;         /* counter of proto app incs. */
 457         struct tcp_states_t     *tcp_state_table;
 458 };
 459
 460 extern struct ip_vs_protocol   *ip_vs_proto_get(unsigned short proto);
 461 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
 462                                                      unsigned short proto);
 463
 464 struct ip_vs_conn_param {
 465         struct net                      *net;
 466         const union nf_inet_addr        *caddr;
 467         const union nf_inet_addr        *vaddr;
 468         __be16                          cport;
 469         __be16                          vport;
 470         __u16                           protocol;
 471         u16                             af;
 472
 473         const struct ip_vs_pe           *pe;
 474         char                            *pe_data;
 475         __u8                            pe_data_len;
 476 };
 477
 478 /*
 479  *      IP_VS structure allocated for each dynamically scheduled connection
 480  */
 481 struct ip_vs_conn {
 482         struct hlist_node       c_list;         /* hashed list heads */
 483 #ifdef CONFIG_NET_NS
 484         struct net              *net;           /* Name space */
 485 #endif
 486         /* Protocol, addresses and port numbers */
 487         u16                     af;             /* address family */
 488         __be16                  cport;
 489         __be16                  vport;
 490         __be16                  dport;
 491         __u32                   fwmark;         /* Fire wall mark from skb */
 492         union nf_inet_addr      caddr;          /* client address */
 493         union nf_inet_addr      vaddr;          /* virtual address */
 494         union nf_inet_addr      daddr;          /* destination address */
 495         volatile __u32          flags;          /* status flags */
 496         __u16                   protocol;       /* Which protocol (TCP/UDP) */
 497
 498         /* counter and timer */
 499         atomic_t                refcnt;         /* reference count */
 500         struct timer_list       timer;          /* Expiration timer */
 501         volatile unsigned long  timeout;        /* timeout */
 502
 503         /* Flags and state transition */
 504         spinlock_t              lock;           /* lock for state transition */
 505         volatile __u16          state;          /* state info */
 506         volatile __u16          old_state;      /* old state, to be used for
 507                                                  * state transition triggerd
 508                                                  * synchronization
 509                                                  */
 510
 511         /* Control members */
 512         struct ip_vs_conn       *control;       /* Master control connection */
 513         atomic_t                n_control;      /* Number of controlled ones */
 514         struct ip_vs_dest       *dest;          /* real server */
 515         atomic_t                in_pkts;        /* incoming packet counter */
 516
 517         /* packet transmitter for different forwarding methods.  If it
 518            mangles the packet, it must return NF_DROP or better NF_STOLEN,
 519            otherwise this must be changed to a sk_buff **.
 520            NF_ACCEPT can be returned when destination is local.
 521          */
 522         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
 523                            struct ip_vs_protocol *pp);
 524
 525         /* Note: we can group the following members into a structure,
 526            in order to save more space, and the following members are
 527            only used in VS/NAT anyway */
 528         struct ip_vs_app        *app;           /* bound ip_vs_app object */
 529         void                    *app_data;      /* Application private data */
 530         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
 531         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
 532
 533         const struct ip_vs_pe   *pe;
 534         char                    *pe_data;
 535         __u8                    pe_data_len;
 536 };
 537
 538 /*
 539  *  To save some memory in conn table when name space is disabled.
 540  */
 541 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
 542 {
 543 #ifdef CONFIG_NET_NS
 544         return cp->net;
 545 #else
 546         return &init_net;
 547 #endif
 548 }
 549 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
 550 {
 551 #ifdef CONFIG_NET_NS
 552         cp->net = net;
 553 #endif
 554 }
 555
 556 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
 557                                     struct net *net)
 558 {
 559 #ifdef CONFIG_NET_NS
 560         return cp->net == net;
 561 #else
 562         return 1;
 563 #endif
 564 }
 565
 566 /*
 567  *      Extended internal versions of struct ip_vs_service_user and
 568  *      ip_vs_dest_user for IPv6 support.
 569  *
 570  *      We need these to conveniently pass around service and destination
 571  *      options, but unfortunately, we also need to keep the old definitions to
 572  *      maintain userspace backwards compatibility for the setsockopt interface.
 573  */
 574 struct ip_vs_service_user_kern {
 575         /* virtual service addresses */
 576         u16                     af;
 577         u16                     protocol;
 578         union nf_inet_addr      addr;           /* virtual ip address */
 579         u16                     port;
 580         u32                     fwmark;         /* firwall mark of service */
 581
 582         /* virtual service options */
 583         char                    *sched_name;
 584         char                    *pe_name;
 585         unsigned                flags;          /* virtual service flags */
 586         unsigned                timeout;        /* persistent timeout in sec */
 587         u32                     netmask;        /* persistent netmask */
 588 };
 589
 590
 591 struct ip_vs_dest_user_kern {
 592         /* destination server address */
 593         union nf_inet_addr      addr;
 594         u16                     port;
 595
 596         /* real server options */
 597         unsigned                conn_flags;     /* connection flags */
 598         int                     weight;         /* destination weight */
 599
 600         /* thresholds for active connections */
 601         u32                     u_threshold;    /* upper threshold */
 602         u32                     l_threshold;    /* lower threshold */
 603 };
 604
 605
 606 /*
 607  *      The information about the virtual service offered to the net
 608  *      and the forwarding entries
 609  */
 610 struct ip_vs_service {
 611         struct list_head        s_list;   /* for normal service table */
 612         struct list_head        f_list;   /* for fwmark-based service table */
 613         atomic_t                refcnt;   /* reference counter */
 614         atomic_t                usecnt;   /* use counter */
 615
 616         u16                     af;       /* address family */
 617         __u16                   protocol; /* which protocol (TCP/UDP) */
 618         union nf_inet_addr      addr;     /* IP address for virtual service */
 619         __be16                  port;     /* port number for the service */
 620         __u32                   fwmark;   /* firewall mark of the service */
 621         unsigned                flags;    /* service status flags */
 622         unsigned                timeout;  /* persistent timeout in ticks */
 623         __be32                  netmask;  /* grouping granularity */
 624         struct net              *net;
 625
 626         struct list_head        destinations;  /* real server d-linked list */
 627         __u32                   num_dests;     /* number of servers */
 628         struct ip_vs_stats      stats;         /* statistics for the service */
 629         struct ip_vs_app        *inc;     /* bind conns to this app inc */
 630
 631         /* for scheduling */
 632         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
 633         rwlock_t                sched_lock;    /* lock sched_data */
 634         void                    *sched_data;   /* scheduler application data */
 635
 636         /* alternate persistence engine */
 637         struct ip_vs_pe         *pe;
 638 };
 639
 640
 641 /*
 642  *      The real server destination forwarding entry
 643  *      with ip address, port number, and so on.
 644  */
 645 struct ip_vs_dest {
 646         struct list_head        n_list;   /* for the dests in the service */
 647         struct list_head        d_list;   /* for table with all the dests */
 648
 649         u16                     af;             /* address family */
 650         __be16                  port;           /* port number of the server */
 651         union nf_inet_addr      addr;           /* IP address of the server */
 652         volatile unsigned       flags;          /* dest status flags */
 653         atomic_t                conn_flags;     /* flags to copy to conn */
 654         atomic_t                weight;         /* server weight */
 655
 656         atomic_t                refcnt;         /* reference counter */
 657         struct ip_vs_stats      stats;          /* statistics */
 658
 659         /* connection counters and thresholds */
 660         atomic_t                activeconns;    /* active connections */
 661         atomic_t                inactconns;     /* inactive connections */
 662         atomic_t                persistconns;   /* persistent connections */
 663         __u32                   u_threshold;    /* upper threshold */
 664         __u32                   l_threshold;    /* lower threshold */
 665
 666         /* for destination cache */
 667         spinlock_t              dst_lock;       /* lock of dst_cache */
 668         struct dst_entry        *dst_cache;     /* destination cache entry */
 669         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
 670         u32                     dst_cookie;
 671 #ifdef CONFIG_IP_VS_IPV6
 672         struct in6_addr         dst_saddr;
 673 #endif
 674
 675         /* for virtual service */
 676         struct ip_vs_service    *svc;           /* service it belongs to */
 677         __u16                   protocol;       /* which protocol (TCP/UDP) */
 678         __be16                  vport;          /* virtual port number */
 679         union nf_inet_addr      vaddr;          /* virtual IP address */
 680         __u32                   vfwmark;        /* firewall mark of service */
 681 };
 682
 683
 684 /*
 685  *      The scheduler object
 686  */
 687 struct ip_vs_scheduler {
 688         struct list_head        n_list;         /* d-linked list head */
 689         char                    *name;          /* scheduler name */
 690         atomic_t                refcnt;         /* reference counter */
 691         struct module           *module;        /* THIS_MODULE/NULL */
 692
 693         /* scheduler initializing service */
 694         int (*init_service)(struct ip_vs_service *svc);
 695         /* scheduling service finish */
 696         int (*done_service)(struct ip_vs_service *svc);
 697         /* scheduler updating service */
 698         int (*update_service)(struct ip_vs_service *svc);
 699
 700         /* selecting a server from the given service */
 701         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 702                                        const struct sk_buff *skb);
 703 };
 704
 705 /* The persistence engine object */
 706 struct ip_vs_pe {
 707         struct list_head        n_list;         /* d-linked list head */
 708         char                    *name;          /* scheduler name */
 709         atomic_t                refcnt;         /* reference counter */
 710         struct module           *module;        /* THIS_MODULE/NULL */
 711
 712         /* get the connection template, if any */
 713         int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
 714         bool (*ct_match)(const struct ip_vs_conn_param *p,
 715                          struct ip_vs_conn *ct);
 716         u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
 717                            bool inverse);
 718         int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
 719 };
 720
 721 /*
 722  *      The application module object (a.k.a. app incarnation)
 723  */
 724 struct ip_vs_app {
 725         struct list_head        a_list;         /* member in app list */
 726         int                     type;           /* IP_VS_APP_TYPE_xxx */
 727         char                    *name;          /* application module name */
 728         __u16                   protocol;
 729         struct module           *module;        /* THIS_MODULE/NULL */
 730         struct list_head        incs_list;      /* list of incarnations */
 731
 732         /* members for application incarnations */
 733         struct list_head        p_list;         /* member in proto app list */
 734         struct ip_vs_app        *app;           /* its real application */
 735         __be16                  port;           /* port number in net order */
 736         atomic_t                usecnt;         /* usage counter */
 737
 738         /*
 739          * output hook: Process packet in inout direction, diff set for TCP.
 740          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 741          *         2=Mangled but checksum was not updated
 742          */
 743         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 744                        struct sk_buff *, int *diff);
 745
 746         /*
 747          * input hook: Process packet in outin direction, diff set for TCP.
 748          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 749          *         2=Mangled but checksum was not updated
 750          */
 751         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 752                       struct sk_buff *, int *diff);
 753
 754         /* ip_vs_app initializer */
 755         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 756
 757         /* ip_vs_app finish */
 758         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 759
 760
 761         /* not used now */
 762         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 763                          struct ip_vs_protocol *);
 764
 765         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 766
 767         int *                   timeout_table;
 768         int *                   timeouts;
 769         int                     timeouts_size;
 770
 771         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 772                              int *verdict, struct ip_vs_conn **cpp);
 773
 774         struct ip_vs_conn *
 775         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 776                        const struct iphdr *iph, unsigned int proto_off,
 777                        int inverse);
 778
 779         struct ip_vs_conn *
 780         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 781                         const struct iphdr *iph, unsigned int proto_off,
 782                         int inverse);
 783
 784         int (*state_transition)(struct ip_vs_conn *cp, int direction,
 785                                 const struct sk_buff *skb,
 786                                 struct ip_vs_app *app);
 787
 788         void (*timeout_change)(struct ip_vs_app *app, int flags);
 789 };
 790
 791 /* IPVS in network namespace */
 792 struct netns_ipvs {
 793         int                     gen;            /* Generation */
 794         int                     enable;         /* enable like nf_hooks do */
 795         /*
 796          *      Hash table: for real service lookups
 797          */
 798         #define IP_VS_RTAB_BITS 4
 799         #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
 800         #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
 801
 802         struct list_head        rs_table[IP_VS_RTAB_SIZE];
 803         /* ip_vs_app */
 804         struct list_head        app_list;
 805
 806         /* ip_vs_proto */
 807         #define IP_VS_PROTO_TAB_SIZE    32      /* must be power of 2 */
 808         struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
 809         /* ip_vs_proto_tcp */
 810 #ifdef CONFIG_IP_VS_PROTO_TCP
 811         #define TCP_APP_TAB_BITS        4
 812         #define TCP_APP_TAB_SIZE        (1 << TCP_APP_TAB_BITS)
 813         #define TCP_APP_TAB_MASK        (TCP_APP_TAB_SIZE - 1)
 814         struct list_head        tcp_apps[TCP_APP_TAB_SIZE];
 815         spinlock_t              tcp_app_lock;
 816 #endif
 817         /* ip_vs_proto_udp */
 818 #ifdef CONFIG_IP_VS_PROTO_UDP
 819         #define UDP_APP_TAB_BITS        4
 820         #define UDP_APP_TAB_SIZE        (1 << UDP_APP_TAB_BITS)
 821         #define UDP_APP_TAB_MASK        (UDP_APP_TAB_SIZE - 1)
 822         struct list_head        udp_apps[UDP_APP_TAB_SIZE];
 823         spinlock_t              udp_app_lock;
 824 #endif
 825         /* ip_vs_proto_sctp */
 826 #ifdef CONFIG_IP_VS_PROTO_SCTP
 827         #define SCTP_APP_TAB_BITS       4
 828         #define SCTP_APP_TAB_SIZE       (1 << SCTP_APP_TAB_BITS)
 829         #define SCTP_APP_TAB_MASK       (SCTP_APP_TAB_SIZE - 1)
 830         /* Hash table for SCTP application incarnations  */
 831         struct list_head        sctp_apps[SCTP_APP_TAB_SIZE];
 832         spinlock_t              sctp_app_lock;
 833 #endif
 834         /* ip_vs_conn */
 835         atomic_t                conn_count;      /*  connection counter */
 836
 837         /* ip_vs_ctl */
 838         struct ip_vs_stats              tot_stats;  /* Statistics & est. */
 839
 840         int                     num_services;    /* no of virtual services */
 841
 842         rwlock_t                rs_lock;         /* real services table */
 843         /* semaphore for IPVS sockopts. And, [gs]etsockopt may sleep. */
 844         struct lock_class_key   ctl_key;        /* ctl_mutex debuging */
 845         /* Trash for destinations */
 846         struct list_head        dest_trash;
 847         /* Service counters */
 848         atomic_t                ftpsvc_counter;
 849         atomic_t                nullsvc_counter;
 850
 851 #ifdef CONFIG_SYSCTL
 852         /* 1/rate drop and drop-entry variables */
 853         struct delayed_work     defense_work;   /* Work handler */
 854         int                     drop_rate;
 855         int                     drop_counter;
 856         atomic_t                dropentry;
 857         /* locks in ctl.c */
 858         spinlock_t              dropentry_lock;  /* drop entry handling */
 859         spinlock_t              droppacket_lock; /* drop packet handling */
 860         spinlock_t              securetcp_lock;  /* state and timeout tables */
 861
 862         /* sys-ctl struct */
 863         struct ctl_table_header *sysctl_hdr;
 864         struct ctl_table        *sysctl_tbl;
 865 #endif
 866
 867         /* sysctl variables */
 868         int                     sysctl_amemthresh;
 869         int                     sysctl_am_droprate;
 870         int                     sysctl_drop_entry;
 871         int                     sysctl_drop_packet;
 872         int                     sysctl_secure_tcp;
 873 #ifdef CONFIG_IP_VS_NFCT
 874         int                     sysctl_conntrack;
 875 #endif
 876         int                     sysctl_snat_reroute;
 877         int                     sysctl_sync_ver;
 878         int                     sysctl_cache_bypass;
 879         int                     sysctl_expire_nodest_conn;
 880         int                     sysctl_expire_quiescent_template;
 881         int                     sysctl_sync_threshold[2];
 882         int                     sysctl_nat_icmp_send;
 883
 884         /* ip_vs_lblc */
 885         int                     sysctl_lblc_expiration;
 886         struct ctl_table_header *lblc_ctl_header;
 887         struct ctl_table        *lblc_ctl_table;
 888         /* ip_vs_lblcr */
 889         int                     sysctl_lblcr_expiration;
 890         struct ctl_table_header *lblcr_ctl_header;
 891         struct ctl_table        *lblcr_ctl_table;
 892         /* ip_vs_est */
 893         struct list_head        est_list;       /* estimator list */
 894         spinlock_t              est_lock;
 895         struct timer_list       est_timer;      /* Estimation timer */
 896         /* ip_vs_sync */
 897         struct list_head        sync_queue;
 898         spinlock_t              sync_lock;
 899         struct ip_vs_sync_buff  *sync_buff;
 900         spinlock_t              sync_buff_lock;
 901         struct sockaddr_in      sync_mcast_addr;
 902         struct task_struct      *master_thread;
 903         struct task_struct      *backup_thread;
 904         int                     send_mesg_maxlen;
 905         int                     recv_mesg_maxlen;
 906         volatile int            sync_state;
 907         volatile int            master_syncid;
 908         volatile int            backup_syncid;
 909         /* multicast interface name */
 910         char                    master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 911         char                    backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 912         /* net name space ptr */
 913         struct net              *net;            /* Needed by timer routines */
 914 };
 915
 916 #define DEFAULT_SYNC_THRESHOLD  3
 917 #define DEFAULT_SYNC_PERIOD     50
 918 #define DEFAULT_SYNC_VER        1
 919
 920 #ifdef CONFIG_SYSCTL
 921
 922 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
 923 {
 924         return ipvs->sysctl_sync_threshold[0];
 925 }
 926
 927 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
 928 {
 929         return ipvs->sysctl_sync_threshold[1];
 930 }
 931
 932 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
 933 {
 934         return ipvs->sysctl_sync_ver;
 935 }
 936
 937 #else
 938
 939 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
 940 {
 941         return DEFAULT_SYNC_THRESHOLD;
 942 }
 943
 944 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
 945 {
 946         return DEFAULT_SYNC_PERIOD;
 947 }
 948
 949 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
 950 {
 951         return DEFAULT_SYNC_VER;
 952 }
 953
 954 #endif
 955
 956 /*
 957  *      IPVS core functions
 958  *      (from ip_vs_core.c)
 959  */
 960 extern const char *ip_vs_proto_name(unsigned proto);
 961 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
 962 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
 963
 964 #define IP_VS_APP_TYPE_FTP      1
 965
 966 /*
 967  *     ip_vs_conn handling functions
 968  *     (from ip_vs_conn.c)
 969  */
 970
 971 enum {
 972         IP_VS_DIR_INPUT = 0,
 973         IP_VS_DIR_OUTPUT,
 974         IP_VS_DIR_INPUT_ONLY,
 975         IP_VS_DIR_LAST,
 976 };
 977
 978 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
 979                                          const union nf_inet_addr *caddr,
 980                                          __be16 cport,
 981                                          const union nf_inet_addr *vaddr,
 982                                          __be16 vport,
 983                                          struct ip_vs_conn_param *p)
 984 {
 985         p->net = net;
 986         p->af = af;
 987         p->protocol = protocol;
 988         p->caddr = caddr;
 989         p->cport = cport;
 990         p->vaddr = vaddr;
 991         p->vport = vport;
 992         p->pe = NULL;
 993         p->pe_data = NULL;
 994 }
 995
 996 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
 997 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
 998
 999 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1000                                             const struct ip_vs_iphdr *iph,
1001                                             unsigned int proto_off,
1002                                             int inverse);
1003
1004 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1005
1006 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1007                                              const struct ip_vs_iphdr *iph,
1008                                              unsigned int proto_off,
1009                                              int inverse);
1010
1011 /* put back the conn without restarting its timer */
1012 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1013 {
1014         atomic_dec(&cp->refcnt);
1015 }
1016 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1017 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1018
1019 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1020                                   const union nf_inet_addr *daddr,
1021                                   __be16 dport, unsigned flags,
1022                                   struct ip_vs_dest *dest, __u32 fwmark);
1023 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1024
1025 extern const char * ip_vs_state_name(__u16 proto, int state);
1026
1027 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1028 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1029 extern void ip_vs_random_dropentry(struct net *net);
1030 extern int ip_vs_conn_init(void);
1031 extern void ip_vs_conn_cleanup(void);
1032
1033 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1034 {
1035         struct ip_vs_conn *ctl_cp = cp->control;
1036         if (!ctl_cp) {
1037                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1038                               "%s:%d to %s:%d\n",
1039                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1040                               ntohs(cp->cport),
1041                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1042                               ntohs(cp->vport));
1043
1044                 return;
1045         }
1046
1047         IP_VS_DBG_BUF(7, "DELeting control for: "
1048                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1049                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1050                       ntohs(cp->cport),
1051                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1052                       ntohs(ctl_cp->cport));
1053
1054         cp->control = NULL;
1055         if (atomic_read(&ctl_cp->n_control) == 0) {
1056                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1057                               "%s:%d to %s:%d\n",
1058                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1059                               ntohs(cp->cport),
1060                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1061                               ntohs(cp->vport));
1062
1063                 return;
1064         }
1065         atomic_dec(&ctl_cp->n_control);
1066 }
1067
1068 static inline void
1069 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1070 {
1071         if (cp->control) {
1072                 IP_VS_ERR_BUF("request control ADD for already controlled: "
1073                               "%s:%d to %s:%d\n",
1074                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1075                               ntohs(cp->cport),
1076                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1077                               ntohs(cp->vport));
1078
1079                 ip_vs_control_del(cp);
1080         }
1081
1082         IP_VS_DBG_BUF(7, "ADDing control for: "
1083                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1084                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1085                       ntohs(cp->cport),
1086                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1087                       ntohs(ctl_cp->cport));
1088
1089         cp->control = ctl_cp;
1090         atomic_inc(&ctl_cp->n_control);
1091 }
1092
1093 /*
1094  * IPVS netns init & cleanup functions
1095  */
1096 extern int __ip_vs_estimator_init(struct net *net);
1097 extern int __ip_vs_control_init(struct net *net);
1098 extern int __ip_vs_protocol_init(struct net *net);
1099 extern int __ip_vs_app_init(struct net *net);
1100 extern int __ip_vs_conn_init(struct net *net);
1101 extern int __ip_vs_sync_init(struct net *net);
1102 extern void __ip_vs_conn_cleanup(struct net *net);
1103 extern void __ip_vs_app_cleanup(struct net *net);
1104 extern void __ip_vs_protocol_cleanup(struct net *net);
1105 extern void __ip_vs_control_cleanup(struct net *net);
1106 extern void __ip_vs_estimator_cleanup(struct net *net);
1107 extern void __ip_vs_sync_cleanup(struct net *net);
1108 extern void __ip_vs_service_cleanup(struct net *net);
1109
1110 /*
1111  *      IPVS application functions
1112  *      (from ip_vs_app.c)
1113  */
1114 #define IP_VS_APP_MAX_PORTS  8
1115 extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1116 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1117 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1118 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1119 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1120                                   __u16 proto, __u16 port);
1121 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1122 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1123
1124 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1125 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1126 extern int ip_vs_app_init(void);
1127 extern void ip_vs_app_cleanup(void);
1128
1129 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1130 void ip_vs_unbind_pe(struct ip_vs_service *svc);
1131 int register_ip_vs_pe(struct ip_vs_pe *pe);
1132 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1133 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1134 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1135
1136 static inline void ip_vs_pe_get(const struct ip_vs_pe *pe)
1137 {
1138         if (pe && pe->module)
1139                 __module_get(pe->module);
1140 }
1141
1142 static inline void ip_vs_pe_put(const struct ip_vs_pe *pe)
1143 {
1144         if (pe && pe->module)
1145                 module_put(pe->module);
1146 }
1147
1148 /*
1149  *      IPVS protocol functions (from ip_vs_proto.c)
1150  */
1151 extern int ip_vs_protocol_init(void);
1152 extern void ip_vs_protocol_cleanup(void);
1153 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1154 extern int *ip_vs_create_timeout_table(int *table, int size);
1155 extern int
1156 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1157                         const char *name, int to);
1158 extern void
1159 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1160                           const struct sk_buff *skb,
1161                           int offset, const char *msg);
1162
1163 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1164 extern struct ip_vs_protocol ip_vs_protocol_udp;
1165 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1166 extern struct ip_vs_protocol ip_vs_protocol_esp;
1167 extern struct ip_vs_protocol ip_vs_protocol_ah;
1168 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1169
1170 /*
1171  *      Registering/unregistering scheduler functions
1172  *      (from ip_vs_sched.c)
1173  */
1174 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1175 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1176 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1177                                 struct ip_vs_scheduler *scheduler);
1178 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1179 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1180 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1181 extern struct ip_vs_conn *
1182 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1183                struct ip_vs_proto_data *pd, int *ignored);
1184 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1185                         struct ip_vs_proto_data *pd);
1186
1187 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1188
1189
1190 /*
1191  *      IPVS control data and functions (from ip_vs_ctl.c)
1192  */
1193 extern struct ip_vs_stats ip_vs_stats;
1194 extern const struct ctl_path net_vs_ctl_path[];
1195 extern int sysctl_ip_vs_sync_ver;
1196
1197 extern void ip_vs_sync_switch_mode(struct net *net, int mode);
1198 extern struct ip_vs_service *
1199 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1200                   const union nf_inet_addr *vaddr, __be16 vport);
1201
1202 static inline void ip_vs_service_put(struct ip_vs_service *svc)
1203 {
1204         atomic_dec(&svc->usecnt);
1205 }
1206
1207 extern struct ip_vs_dest *
1208 ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
1209                           const union nf_inet_addr *daddr, __be16 dport);
1210
1211 extern int ip_vs_use_count_inc(void);
1212 extern void ip_vs_use_count_dec(void);
1213 extern int ip_vs_control_init(void);
1214 extern void ip_vs_control_cleanup(void);
1215 extern struct ip_vs_dest *
1216 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1217                 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1218                 __u16 protocol, __u32 fwmark);
1219 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1220
1221
1222 /*
1223  *      IPVS sync daemon data and function prototypes
1224  *      (from ip_vs_sync.c)
1225  */
1226 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1227                              __u8 syncid);
1228 extern int stop_sync_thread(struct net *net, int state);
1229 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
1230 extern int ip_vs_sync_init(void);
1231 extern void ip_vs_sync_cleanup(void);
1232
1233
1234 /*
1235  *      IPVS rate estimator prototypes (from ip_vs_est.c)
1236  */
1237 extern int ip_vs_estimator_init(void);
1238 extern void ip_vs_estimator_cleanup(void);
1239 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1240 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1241 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1242 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1243                                  struct ip_vs_stats *stats);
1244
1245 /*
1246  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
1247  */
1248 extern int ip_vs_null_xmit
1249 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1250 extern int ip_vs_bypass_xmit
1251 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1252 extern int ip_vs_nat_xmit
1253 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1254 extern int ip_vs_tunnel_xmit
1255 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1256 extern int ip_vs_dr_xmit
1257 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1258 extern int ip_vs_icmp_xmit
1259 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
1260 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1261
1262 #ifdef CONFIG_IP_VS_IPV6
1263 extern int ip_vs_bypass_xmit_v6
1264 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1265 extern int ip_vs_nat_xmit_v6
1266 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1267 extern int ip_vs_tunnel_xmit_v6
1268 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1269 extern int ip_vs_dr_xmit_v6
1270 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1271 extern int ip_vs_icmp_xmit_v6
1272 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1273  int offset);
1274 #endif
1275
1276 #ifdef CONFIG_SYSCTL
1277 /*
1278  *      This is a simple mechanism to ignore packets when
1279  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
1280  *      we start to drop 1/rate of the packets
1281  */
1282
1283 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1284 {
1285         if (!ipvs->drop_rate)
1286                 return 0;
1287         if (--ipvs->drop_counter > 0)
1288                 return 0;
1289         ipvs->drop_counter = ipvs->drop_rate;
1290         return 1;
1291 }
1292 #else
1293 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1294 #endif
1295
1296 /*
1297  *      ip_vs_fwd_tag returns the forwarding tag of the connection
1298  */
1299 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
1300
1301 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1302 {
1303         char fwd;
1304
1305         switch (IP_VS_FWD_METHOD(cp)) {
1306         case IP_VS_CONN_F_MASQ:
1307                 fwd = 'M'; break;
1308         case IP_VS_CONN_F_LOCALNODE:
1309                 fwd = 'L'; break;
1310         case IP_VS_CONN_F_TUNNEL:
1311                 fwd = 'T'; break;
1312         case IP_VS_CONN_F_DROUTE:
1313                 fwd = 'R'; break;
1314         case IP_VS_CONN_F_BYPASS:
1315                 fwd = 'B'; break;
1316         default:
1317                 fwd = '?'; break;
1318         }
1319         return fwd;
1320 }
1321
1322 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1323                            struct ip_vs_conn *cp, int dir);
1324
1325 #ifdef CONFIG_IP_VS_IPV6
1326 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1327                               struct ip_vs_conn *cp, int dir);
1328 #endif
1329
1330 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1331
1332 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1333 {
1334         __be32 diff[2] = { ~old, new };
1335
1336         return csum_partial(diff, sizeof(diff), oldsum);
1337 }
1338
1339 #ifdef CONFIG_IP_VS_IPV6
1340 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1341                                         __wsum oldsum)
1342 {
1343         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1344                             new[3],  new[2],  new[1],  new[0] };
1345
1346         return csum_partial(diff, sizeof(diff), oldsum);
1347 }
1348 #endif
1349
1350 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1351 {
1352         __be16 diff[2] = { ~old, new };
1353
1354         return csum_partial(diff, sizeof(diff), oldsum);
1355 }
1356
1357 /*
1358  * Forget current conntrack (unconfirmed) and attach notrack entry
1359  */
1360 static inline void ip_vs_notrack(struct sk_buff *skb)
1361 {
1362 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1363         enum ip_conntrack_info ctinfo;
1364         struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1365
1366         if (!ct || !nf_ct_is_untracked(ct)) {
1367                 nf_reset(skb);
1368                 skb->nfct = &nf_ct_untracked_get()->ct_general;
1369                 skb->nfctinfo = IP_CT_NEW;
1370                 nf_conntrack_get(skb->nfct);
1371         }
1372 #endif
1373 }
1374
1375 #ifdef CONFIG_IP_VS_NFCT
1376 /*
1377  *      Netfilter connection tracking
1378  *      (from ip_vs_nfct.c)
1379  */
1380 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1381 {
1382 #ifdef CONFIG_SYSCTL
1383         return ipvs->sysctl_conntrack;
1384 #else
1385         return 0;
1386 #endif
1387 }
1388
1389 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1390                                    int outin);
1391 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1392 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1393                                       struct ip_vs_conn *cp, u_int8_t proto,
1394                                       const __be16 port, int from_rs);
1395 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1396
1397 #else
1398
1399 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1400 {
1401         return 0;
1402 }
1403
1404 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1405                                           struct ip_vs_conn *cp, int outin)
1406 {
1407 }
1408
1409 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1410                                           struct ip_vs_conn *cp)
1411 {
1412         return NF_ACCEPT;
1413 }
1414
1415 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1416 {
1417 }
1418 /* CONFIG_IP_VS_NFCT */
1419 #endif
1420
1421 static inline unsigned int
1422 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1423 {
1424         /*
1425          * We think the overhead of processing active connections is 256
1426          * times higher than that of inactive connections in average. (This
1427          * 256 times might not be accurate, we will change it later) We
1428          * use the following formula to estimate the overhead now:
1429          *                dest->activeconns*256 + dest->inactconns
1430          */
1431         return (atomic_read(&dest->activeconns) << 8) +
1432                 atomic_read(&dest->inactconns);
1433 }
1434
1435 #endif /* __KERNEL__ */
1436
1437 #endif  /* _NET_IP_VS_H */