linux/include/net/ip_vs.h
<<
>>
Prefs
   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/list.h>                 /* for struct list_head */
  14#include <linux/spinlock.h>             /* for struct rwlock_t */
  15#include <linux/atomic.h>                 /* for struct atomic_t */
  16#include <linux/compiler.h>
  17#include <linux/timer.h>
  18#include <linux/bug.h>
  19
  20#include <net/checksum.h>
  21#include <linux/netfilter.h>            /* for union nf_inet_addr */
  22#include <linux/ip.h>
  23#include <linux/ipv6.h>                 /* for struct ipv6hdr */
  24#include <net/ipv6.h>
  25#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  26#include <net/netfilter/nf_conntrack.h>
  27#endif
  28#include <net/net_namespace.h>          /* Netw namespace */
  29
  30/*
  31 * Generic access of ipvs struct
  32 */
  33static inline struct netns_ipvs *net_ipvs(struct net* net)
  34{
  35        return net->ipvs;
  36}
  37/*
  38 * Get net ptr from skb in traffic cases
  39 * use skb_sknet when call is from userland (ioctl or netlink)
  40 */
  41static inline struct net *skb_net(const struct sk_buff *skb)
  42{
  43#ifdef CONFIG_NET_NS
  44#ifdef CONFIG_IP_VS_DEBUG
  45        /*
  46         * This is used for debug only.
  47         * Start with the most likely hit
  48         * End with BUG
  49         */
  50        if (likely(skb->dev && skb->dev->nd_net))
  51                return dev_net(skb->dev);
  52        if (skb_dst(skb) && skb_dst(skb)->dev)
  53                return dev_net(skb_dst(skb)->dev);
  54        WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
  55                      __func__, __LINE__);
  56        if (likely(skb->sk && skb->sk->sk_net))
  57                return sock_net(skb->sk);
  58        pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
  59                __func__, __LINE__);
  60        BUG();
  61#else
  62        return dev_net(skb->dev ? : skb_dst(skb)->dev);
  63#endif
  64#else
  65        return &init_net;
  66#endif
  67}
  68
  69static inline struct net *skb_sknet(const struct sk_buff *skb)
  70{
  71#ifdef CONFIG_NET_NS
  72#ifdef CONFIG_IP_VS_DEBUG
  73        /* Start with the most likely hit */
  74        if (likely(skb->sk && skb->sk->sk_net))
  75                return sock_net(skb->sk);
  76        WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
  77                       __func__, __LINE__);
  78        if (likely(skb->dev && skb->dev->nd_net))
  79                return dev_net(skb->dev);
  80        pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
  81                __func__, __LINE__);
  82        BUG();
  83#else
  84        return sock_net(skb->sk);
  85#endif
  86#else
  87        return &init_net;
  88#endif
  89}
  90/*
  91 * This one needed for single_open_net since net is stored directly in
  92 * private not as a struct i.e. seq_file_net can't be used.
  93 */
  94static inline struct net *seq_file_single_net(struct seq_file *seq)
  95{
  96#ifdef CONFIG_NET_NS
  97        return (struct net *)seq->private;
  98#else
  99        return &init_net;
 100#endif
 101}
 102
 103/* Connections' size value needed by ip_vs_ctl.c */
 104extern int ip_vs_conn_tab_size;
 105
 106
 107struct ip_vs_iphdr {
 108        int len;
 109        __u8 protocol;
 110        union nf_inet_addr saddr;
 111        union nf_inet_addr daddr;
 112};
 113
 114static inline void
 115ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
 116{
 117#ifdef CONFIG_IP_VS_IPV6
 118        if (af == AF_INET6) {
 119                const struct ipv6hdr *iph = nh;
 120                iphdr->len = sizeof(struct ipv6hdr);
 121                iphdr->protocol = iph->nexthdr;
 122                iphdr->saddr.in6 = iph->saddr;
 123                iphdr->daddr.in6 = iph->daddr;
 124        } else
 125#endif
 126        {
 127                const struct iphdr *iph = nh;
 128                iphdr->len = iph->ihl * 4;
 129                iphdr->protocol = iph->protocol;
 130                iphdr->saddr.ip = iph->saddr;
 131                iphdr->daddr.ip = iph->daddr;
 132        }
 133}
 134
 135static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
 136                                   const union nf_inet_addr *src)
 137{
 138#ifdef CONFIG_IP_VS_IPV6
 139        if (af == AF_INET6)
 140                dst->in6 = src->in6;
 141        else
 142#endif
 143        dst->ip = src->ip;
 144}
 145
 146static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
 147                                   const union nf_inet_addr *b)
 148{
 149#ifdef CONFIG_IP_VS_IPV6
 150        if (af == AF_INET6)
 151                return ipv6_addr_equal(&a->in6, &b->in6);
 152#endif
 153        return a->ip == b->ip;
 154}
 155
 156#ifdef CONFIG_IP_VS_DEBUG
 157#include <linux/net.h>
 158
 159extern int ip_vs_get_debug_level(void);
 160
 161static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
 162                                         const union nf_inet_addr *addr,
 163                                         int *idx)
 164{
 165        int len;
 166#ifdef CONFIG_IP_VS_IPV6
 167        if (af == AF_INET6)
 168                len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
 169                               &addr->in6) + 1;
 170        else
 171#endif
 172                len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
 173                               &addr->ip) + 1;
 174
 175        *idx += len;
 176        BUG_ON(*idx > buf_len + 1);
 177        return &buf[*idx - len];
 178}
 179
 180#define IP_VS_DBG_BUF(level, msg, ...)                                  \
 181        do {                                                            \
 182                char ip_vs_dbg_buf[160];                                \
 183                int ip_vs_dbg_idx = 0;                                  \
 184                if (level <= ip_vs_get_debug_level())                   \
 185                        printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 186        } while (0)
 187#define IP_VS_ERR_BUF(msg...)                                           \
 188        do {                                                            \
 189                char ip_vs_dbg_buf[160];                                \
 190                int ip_vs_dbg_idx = 0;                                  \
 191                pr_err(msg);                                            \
 192        } while (0)
 193
 194/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
 195#define IP_VS_DBG_ADDR(af, addr)                                        \
 196        ip_vs_dbg_addr(af, ip_vs_dbg_buf,                               \
 197                       sizeof(ip_vs_dbg_buf), addr,                     \
 198                       &ip_vs_dbg_idx)
 199
 200#define IP_VS_DBG(level, msg, ...)                                      \
 201        do {                                                            \
 202                if (level <= ip_vs_get_debug_level())                   \
 203                        printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 204        } while (0)
 205#define IP_VS_DBG_RL(msg, ...)                                          \
 206        do {                                                            \
 207                if (net_ratelimit())                                    \
 208                        printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
 209        } while (0)
 210#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)                     \
 211        do {                                                            \
 212                if (level <= ip_vs_get_debug_level())                   \
 213                        pp->debug_packet(af, pp, skb, ofs, msg);        \
 214        } while (0)
 215#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)                  \
 216        do {                                                            \
 217                if (level <= ip_vs_get_debug_level() &&                 \
 218                    net_ratelimit())                                    \
 219                        pp->debug_packet(af, pp, skb, ofs, msg);        \
 220        } while (0)
 221#else   /* NO DEBUGGING at ALL */
 222#define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
 223#define IP_VS_ERR_BUF(msg...)  do {} while (0)
 224#define IP_VS_DBG(level, msg...)  do {} while (0)
 225#define IP_VS_DBG_RL(msg...)  do {} while (0)
 226#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)     do {} while (0)
 227#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)  do {} while (0)
 228#endif
 229
 230#define IP_VS_BUG() BUG()
 231#define IP_VS_ERR_RL(msg, ...)                                          \
 232        do {                                                            \
 233                if (net_ratelimit())                                    \
 234                        pr_err(msg, ##__VA_ARGS__);                     \
 235        } while (0)
 236
 237#ifdef CONFIG_IP_VS_DEBUG
 238#define EnterFunction(level)                                            \
 239        do {                                                            \
 240                if (level <= ip_vs_get_debug_level())                   \
 241                        printk(KERN_DEBUG                               \
 242                               pr_fmt("Enter: %s, %s line %i\n"),       \
 243                               __func__, __FILE__, __LINE__);           \
 244        } while (0)
 245#define LeaveFunction(level)                                            \
 246        do {                                                            \
 247                if (level <= ip_vs_get_debug_level())                   \
 248                        printk(KERN_DEBUG                               \
 249                               pr_fmt("Leave: %s, %s line %i\n"),       \
 250                               __func__, __FILE__, __LINE__);           \
 251        } while (0)
 252#else
 253#define EnterFunction(level)   do {} while (0)
 254#define LeaveFunction(level)   do {} while (0)
 255#endif
 256
 257#define IP_VS_WAIT_WHILE(expr)  while (expr) { cpu_relax(); }
 258
 259
 260/*
 261 *      The port number of FTP service (in network order).
 262 */
 263#define FTPPORT  cpu_to_be16(21)
 264#define FTPDATA  cpu_to_be16(20)
 265
 266/*
 267 *      TCP State Values
 268 */
 269enum {
 270        IP_VS_TCP_S_NONE = 0,
 271        IP_VS_TCP_S_ESTABLISHED,
 272        IP_VS_TCP_S_SYN_SENT,
 273        IP_VS_TCP_S_SYN_RECV,
 274        IP_VS_TCP_S_FIN_WAIT,
 275        IP_VS_TCP_S_TIME_WAIT,
 276        IP_VS_TCP_S_CLOSE,
 277        IP_VS_TCP_S_CLOSE_WAIT,
 278        IP_VS_TCP_S_LAST_ACK,
 279        IP_VS_TCP_S_LISTEN,
 280        IP_VS_TCP_S_SYNACK,
 281        IP_VS_TCP_S_LAST
 282};
 283
 284/*
 285 *      UDP State Values
 286 */
 287enum {
 288        IP_VS_UDP_S_NORMAL,
 289        IP_VS_UDP_S_LAST,
 290};
 291
 292/*
 293 *      ICMP State Values
 294 */
 295enum {
 296        IP_VS_ICMP_S_NORMAL,
 297        IP_VS_ICMP_S_LAST,
 298};
 299
 300/*
 301 *      SCTP State Values
 302 */
 303enum ip_vs_sctp_states {
 304        IP_VS_SCTP_S_NONE,
 305        IP_VS_SCTP_S_INIT_CLI,
 306        IP_VS_SCTP_S_INIT_SER,
 307        IP_VS_SCTP_S_INIT_ACK_CLI,
 308        IP_VS_SCTP_S_INIT_ACK_SER,
 309        IP_VS_SCTP_S_ECHO_CLI,
 310        IP_VS_SCTP_S_ECHO_SER,
 311        IP_VS_SCTP_S_ESTABLISHED,
 312        IP_VS_SCTP_S_SHUT_CLI,
 313        IP_VS_SCTP_S_SHUT_SER,
 314        IP_VS_SCTP_S_SHUT_ACK_CLI,
 315        IP_VS_SCTP_S_SHUT_ACK_SER,
 316        IP_VS_SCTP_S_CLOSED,
 317        IP_VS_SCTP_S_LAST
 318};
 319
 320/*
 321 *      Delta sequence info structure
 322 *      Each ip_vs_conn has 2 (output AND input seq. changes).
 323 *      Only used in the VS/NAT.
 324 */
 325struct ip_vs_seq {
 326        __u32                   init_seq;       /* Add delta from this seq */
 327        __u32                   delta;          /* Delta in sequence numbers */
 328        __u32                   previous_delta; /* Delta in sequence numbers
 329                                                   before last resized pkt */
 330};
 331
 332/*
 333 * counters per cpu
 334 */
 335struct ip_vs_counters {
 336        __u32           conns;          /* connections scheduled */
 337        __u32           inpkts;         /* incoming packets */
 338        __u32           outpkts;        /* outgoing packets */
 339        __u64           inbytes;        /* incoming bytes */
 340        __u64           outbytes;       /* outgoing bytes */
 341};
 342/*
 343 * Stats per cpu
 344 */
 345struct ip_vs_cpu_stats {
 346        struct ip_vs_counters   ustats;
 347        struct u64_stats_sync   syncp;
 348};
 349
 350/*
 351 *      IPVS statistics objects
 352 */
 353struct ip_vs_estimator {
 354        struct list_head        list;
 355
 356        u64                     last_inbytes;
 357        u64                     last_outbytes;
 358        u32                     last_conns;
 359        u32                     last_inpkts;
 360        u32                     last_outpkts;
 361
 362        u32                     cps;
 363        u32                     inpps;
 364        u32                     outpps;
 365        u32                     inbps;
 366        u32                     outbps;
 367};
 368
 369struct ip_vs_stats {
 370        struct ip_vs_stats_user ustats;         /* statistics */
 371        struct ip_vs_estimator  est;            /* estimator */
 372        struct ip_vs_cpu_stats  *cpustats;      /* per cpu counters */
 373        spinlock_t              lock;           /* spin lock */
 374        struct ip_vs_stats_user ustats0;        /* reset values */
 375};
 376
 377struct dst_entry;
 378struct iphdr;
 379struct ip_vs_conn;
 380struct ip_vs_app;
 381struct sk_buff;
 382struct ip_vs_proto_data;
 383
 384struct ip_vs_protocol {
 385        struct ip_vs_protocol   *next;
 386        char                    *name;
 387        u16                     protocol;
 388        u16                     num_states;
 389        int                     dont_defrag;
 390
 391        void (*init)(struct ip_vs_protocol *pp);
 392
 393        void (*exit)(struct ip_vs_protocol *pp);
 394
 395        int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
 396
 397        void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
 398
 399        int (*conn_schedule)(int af, struct sk_buff *skb,
 400                             struct ip_vs_proto_data *pd,
 401                             int *verdict, struct ip_vs_conn **cpp);
 402
 403        struct ip_vs_conn *
 404        (*conn_in_get)(int af,
 405                       const struct sk_buff *skb,
 406                       const struct ip_vs_iphdr *iph,
 407                       unsigned int proto_off,
 408                       int inverse);
 409
 410        struct ip_vs_conn *
 411        (*conn_out_get)(int af,
 412                        const struct sk_buff *skb,
 413                        const struct ip_vs_iphdr *iph,
 414                        unsigned int proto_off,
 415                        int inverse);
 416
 417        int (*snat_handler)(struct sk_buff *skb,
 418                            struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 419
 420        int (*dnat_handler)(struct sk_buff *skb,
 421                            struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
 422
 423        int (*csum_check)(int af, struct sk_buff *skb,
 424                          struct ip_vs_protocol *pp);
 425
 426        const char *(*state_name)(int state);
 427
 428        void (*state_transition)(struct ip_vs_conn *cp, int direction,
 429                                 const struct sk_buff *skb,
 430                                 struct ip_vs_proto_data *pd);
 431
 432        int (*register_app)(struct net *net, struct ip_vs_app *inc);
 433
 434        void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
 435
 436        int (*app_conn_bind)(struct ip_vs_conn *cp);
 437
 438        void (*debug_packet)(int af, struct ip_vs_protocol *pp,
 439                             const struct sk_buff *skb,
 440                             int offset,
 441                             const char *msg);
 442
 443        void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
 444};
 445
 446/*
 447 * protocol data per netns
 448 */
 449struct ip_vs_proto_data {
 450        struct ip_vs_proto_data *next;
 451        struct ip_vs_protocol   *pp;
 452        int                     *timeout_table; /* protocol timeout table */
 453        atomic_t                appcnt;         /* counter of proto app incs. */
 454        struct tcp_states_t     *tcp_state_table;
 455};
 456
 457extern struct ip_vs_protocol   *ip_vs_proto_get(unsigned short proto);
 458extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
 459                                                     unsigned short proto);
 460
 461struct ip_vs_conn_param {
 462        struct net                      *net;
 463        const union nf_inet_addr        *caddr;
 464        const union nf_inet_addr        *vaddr;
 465        __be16                          cport;
 466        __be16                          vport;
 467        __u16                           protocol;
 468        u16                             af;
 469
 470        const struct ip_vs_pe           *pe;
 471        char                            *pe_data;
 472        __u8                            pe_data_len;
 473};
 474
 475/*
 476 *      IP_VS structure allocated for each dynamically scheduled connection
 477 */
 478struct ip_vs_conn {
 479        struct hlist_node       c_list;         /* hashed list heads */
 480#ifdef CONFIG_NET_NS
 481        struct net              *net;           /* Name space */
 482#endif
 483        /* Protocol, addresses and port numbers */
 484        u16                     af;             /* address family */
 485        __be16                  cport;
 486        __be16                  vport;
 487        __be16                  dport;
 488        __u32                   fwmark;         /* Fire wall mark from skb */
 489        union nf_inet_addr      caddr;          /* client address */
 490        union nf_inet_addr      vaddr;          /* virtual address */
 491        union nf_inet_addr      daddr;          /* destination address */
 492        volatile __u32          flags;          /* status flags */
 493        __u16                   protocol;       /* Which protocol (TCP/UDP) */
 494
 495        /* counter and timer */
 496        atomic_t                refcnt;         /* reference count */
 497        struct timer_list       timer;          /* Expiration timer */
 498        volatile unsigned long  timeout;        /* timeout */
 499
 500        /* Flags and state transition */
 501        spinlock_t              lock;           /* lock for state transition */
 502        volatile __u16          state;          /* state info */
 503        volatile __u16          old_state;      /* old state, to be used for
 504                                                 * state transition triggerd
 505                                                 * synchronization
 506                                                 */
 507        unsigned long           sync_endtime;   /* jiffies + sent_retries */
 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 */
 539static 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}
 547static 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
 554static 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 */
 572struct 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 int            flags;          /* virtual service flags */
 584        unsigned int            timeout;        /* persistent timeout in sec */
 585        u32                     netmask;        /* persistent netmask */
 586};
 587
 588
 589struct 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 int            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 */
 608struct 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 int            flags;    /* service status flags */
 620        unsigned int            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 */
 643struct 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 int   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 */
 683struct 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 */
 702struct 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 */
 720struct 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
 787struct ipvs_master_sync_state {
 788        struct list_head        sync_queue;
 789        struct ip_vs_sync_buff  *sync_buff;
 790        int                     sync_queue_len;
 791        unsigned int            sync_queue_delay;
 792        struct task_struct      *master_thread;
 793        struct delayed_work     master_wakeup_work;
 794        struct netns_ipvs       *ipvs;
 795};
 796
 797/* IPVS in network namespace */
 798struct netns_ipvs {
 799        int                     gen;            /* Generation */
 800        int                     enable;         /* enable like nf_hooks do */
 801        /*
 802         *      Hash table: for real service lookups
 803         */
 804        #define IP_VS_RTAB_BITS 4
 805        #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
 806        #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
 807
 808        struct list_head        rs_table[IP_VS_RTAB_SIZE];
 809        /* ip_vs_app */
 810        struct list_head        app_list;
 811        /* ip_vs_proto */
 812        #define IP_VS_PROTO_TAB_SIZE    32      /* must be power of 2 */
 813        struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
 814        /* ip_vs_proto_tcp */
 815#ifdef CONFIG_IP_VS_PROTO_TCP
 816        #define TCP_APP_TAB_BITS        4
 817        #define TCP_APP_TAB_SIZE        (1 << TCP_APP_TAB_BITS)
 818        #define TCP_APP_TAB_MASK        (TCP_APP_TAB_SIZE - 1)
 819        struct list_head        tcp_apps[TCP_APP_TAB_SIZE];
 820        spinlock_t              tcp_app_lock;
 821#endif
 822        /* ip_vs_proto_udp */
 823#ifdef CONFIG_IP_VS_PROTO_UDP
 824        #define UDP_APP_TAB_BITS        4
 825        #define UDP_APP_TAB_SIZE        (1 << UDP_APP_TAB_BITS)
 826        #define UDP_APP_TAB_MASK        (UDP_APP_TAB_SIZE - 1)
 827        struct list_head        udp_apps[UDP_APP_TAB_SIZE];
 828        spinlock_t              udp_app_lock;
 829#endif
 830        /* ip_vs_proto_sctp */
 831#ifdef CONFIG_IP_VS_PROTO_SCTP
 832        #define SCTP_APP_TAB_BITS       4
 833        #define SCTP_APP_TAB_SIZE       (1 << SCTP_APP_TAB_BITS)
 834        #define SCTP_APP_TAB_MASK       (SCTP_APP_TAB_SIZE - 1)
 835        /* Hash table for SCTP application incarnations  */
 836        struct list_head        sctp_apps[SCTP_APP_TAB_SIZE];
 837        spinlock_t              sctp_app_lock;
 838#endif
 839        /* ip_vs_conn */
 840        atomic_t                conn_count;      /*  connection counter */
 841
 842        /* ip_vs_ctl */
 843        struct ip_vs_stats              tot_stats;  /* Statistics & est. */
 844
 845        int                     num_services;    /* no of virtual services */
 846
 847        rwlock_t                rs_lock;         /* real services table */
 848        /* Trash for destinations */
 849        struct list_head        dest_trash;
 850        /* Service counters */
 851        atomic_t                ftpsvc_counter;
 852        atomic_t                nullsvc_counter;
 853
 854#ifdef CONFIG_SYSCTL
 855        /* 1/rate drop and drop-entry variables */
 856        struct delayed_work     defense_work;   /* Work handler */
 857        int                     drop_rate;
 858        int                     drop_counter;
 859        atomic_t                dropentry;
 860        /* locks in ctl.c */
 861        spinlock_t              dropentry_lock;  /* drop entry handling */
 862        spinlock_t              droppacket_lock; /* drop packet handling */
 863        spinlock_t              securetcp_lock;  /* state and timeout tables */
 864
 865        /* sys-ctl struct */
 866        struct ctl_table_header *sysctl_hdr;
 867        struct ctl_table        *sysctl_tbl;
 868#endif
 869
 870        /* sysctl variables */
 871        int                     sysctl_amemthresh;
 872        int                     sysctl_am_droprate;
 873        int                     sysctl_drop_entry;
 874        int                     sysctl_drop_packet;
 875        int                     sysctl_secure_tcp;
 876#ifdef CONFIG_IP_VS_NFCT
 877        int                     sysctl_conntrack;
 878#endif
 879        int                     sysctl_snat_reroute;
 880        int                     sysctl_sync_ver;
 881        int                     sysctl_sync_ports;
 882        int                     sysctl_sync_qlen_max;
 883        int                     sysctl_sync_sock_size;
 884        int                     sysctl_cache_bypass;
 885        int                     sysctl_expire_nodest_conn;
 886        int                     sysctl_expire_quiescent_template;
 887        int                     sysctl_sync_threshold[2];
 888        unsigned int            sysctl_sync_refresh_period;
 889        int                     sysctl_sync_retries;
 890        int                     sysctl_nat_icmp_send;
 891        int                     sysctl_pmtu_disc;
 892
 893        /* ip_vs_lblc */
 894        int                     sysctl_lblc_expiration;
 895        struct ctl_table_header *lblc_ctl_header;
 896        struct ctl_table        *lblc_ctl_table;
 897        /* ip_vs_lblcr */
 898        int                     sysctl_lblcr_expiration;
 899        struct ctl_table_header *lblcr_ctl_header;
 900        struct ctl_table        *lblcr_ctl_table;
 901        /* ip_vs_est */
 902        struct list_head        est_list;       /* estimator list */
 903        spinlock_t              est_lock;
 904        struct timer_list       est_timer;      /* Estimation timer */
 905        /* ip_vs_sync */
 906        spinlock_t              sync_lock;
 907        struct ipvs_master_sync_state *ms;
 908        spinlock_t              sync_buff_lock;
 909        struct task_struct      **backup_threads;
 910        int                     threads_mask;
 911        int                     send_mesg_maxlen;
 912        int                     recv_mesg_maxlen;
 913        volatile int            sync_state;
 914        volatile int            master_syncid;
 915        volatile int            backup_syncid;
 916        struct mutex            sync_mutex;
 917        /* multicast interface name */
 918        char                    master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 919        char                    backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
 920        /* net name space ptr */
 921        struct net              *net;            /* Needed by timer routines */
 922};
 923
 924#define DEFAULT_SYNC_THRESHOLD  3
 925#define DEFAULT_SYNC_PERIOD     50
 926#define DEFAULT_SYNC_VER        1
 927#define DEFAULT_SYNC_REFRESH_PERIOD     (0U * HZ)
 928#define DEFAULT_SYNC_RETRIES            0
 929#define IPVS_SYNC_WAKEUP_RATE   8
 930#define IPVS_SYNC_QLEN_MAX      (IPVS_SYNC_WAKEUP_RATE * 4)
 931#define IPVS_SYNC_SEND_DELAY    (HZ / 50)
 932#define IPVS_SYNC_CHECK_PERIOD  HZ
 933#define IPVS_SYNC_FLUSH_TIME    (HZ * 2)
 934#define IPVS_SYNC_PORTS_MAX     (1 << 6)
 935
 936#ifdef CONFIG_SYSCTL
 937
 938static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
 939{
 940        return ipvs->sysctl_sync_threshold[0];
 941}
 942
 943static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
 944{
 945        return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
 946}
 947
 948static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
 949{
 950        return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
 951}
 952
 953static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
 954{
 955        return ipvs->sysctl_sync_retries;
 956}
 957
 958static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
 959{
 960        return ipvs->sysctl_sync_ver;
 961}
 962
 963static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
 964{
 965        return ACCESS_ONCE(ipvs->sysctl_sync_ports);
 966}
 967
 968static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
 969{
 970        return ipvs->sysctl_sync_qlen_max;
 971}
 972
 973static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
 974{
 975        return ipvs->sysctl_sync_sock_size;
 976}
 977
 978static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
 979{
 980        return ipvs->sysctl_pmtu_disc;
 981}
 982
 983#else
 984
 985static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
 986{
 987        return DEFAULT_SYNC_THRESHOLD;
 988}
 989
 990static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
 991{
 992        return DEFAULT_SYNC_PERIOD;
 993}
 994
 995static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
 996{
 997        return DEFAULT_SYNC_REFRESH_PERIOD;
 998}
 999
1000static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1001{
1002        return DEFAULT_SYNC_RETRIES & 3;
1003}
1004
1005static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1006{
1007        return DEFAULT_SYNC_VER;
1008}
1009
1010static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1011{
1012        return 1;
1013}
1014
1015static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1016{
1017        return IPVS_SYNC_QLEN_MAX;
1018}
1019
1020static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1021{
1022        return 0;
1023}
1024
1025static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1026{
1027        return 1;
1028}
1029
1030#endif
1031
1032/*
1033 *      IPVS core functions
1034 *      (from ip_vs_core.c)
1035 */
1036extern const char *ip_vs_proto_name(unsigned int proto);
1037extern void ip_vs_init_hash_table(struct list_head *table, int rows);
1038#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1039
1040#define IP_VS_APP_TYPE_FTP      1
1041
1042/*
1043 *     ip_vs_conn handling functions
1044 *     (from ip_vs_conn.c)
1045 */
1046
1047enum {
1048        IP_VS_DIR_INPUT = 0,
1049        IP_VS_DIR_OUTPUT,
1050        IP_VS_DIR_INPUT_ONLY,
1051        IP_VS_DIR_LAST,
1052};
1053
1054static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1055                                         const union nf_inet_addr *caddr,
1056                                         __be16 cport,
1057                                         const union nf_inet_addr *vaddr,
1058                                         __be16 vport,
1059                                         struct ip_vs_conn_param *p)
1060{
1061        p->net = net;
1062        p->af = af;
1063        p->protocol = protocol;
1064        p->caddr = caddr;
1065        p->cport = cport;
1066        p->vaddr = vaddr;
1067        p->vport = vport;
1068        p->pe = NULL;
1069        p->pe_data = NULL;
1070}
1071
1072struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1073struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1074
1075struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1076                                            const struct ip_vs_iphdr *iph,
1077                                            unsigned int proto_off,
1078                                            int inverse);
1079
1080struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1081
1082struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1083                                             const struct ip_vs_iphdr *iph,
1084                                             unsigned int proto_off,
1085                                             int inverse);
1086
1087/* put back the conn without restarting its timer */
1088static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1089{
1090        atomic_dec(&cp->refcnt);
1091}
1092extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1093extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1094
1095struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1096                                  const union nf_inet_addr *daddr,
1097                                  __be16 dport, unsigned int flags,
1098                                  struct ip_vs_dest *dest, __u32 fwmark);
1099extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1100
1101extern const char * ip_vs_state_name(__u16 proto, int state);
1102
1103extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1104extern int ip_vs_check_template(struct ip_vs_conn *ct);
1105extern void ip_vs_random_dropentry(struct net *net);
1106extern int ip_vs_conn_init(void);
1107extern void ip_vs_conn_cleanup(void);
1108
1109static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1110{
1111        struct ip_vs_conn *ctl_cp = cp->control;
1112        if (!ctl_cp) {
1113                IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1114                              "%s:%d to %s:%d\n",
1115                              IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1116                              ntohs(cp->cport),
1117                              IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1118                              ntohs(cp->vport));
1119
1120                return;
1121        }
1122
1123        IP_VS_DBG_BUF(7, "DELeting control for: "
1124                      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1125                      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1126                      ntohs(cp->cport),
1127                      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1128                      ntohs(ctl_cp->cport));
1129
1130        cp->control = NULL;
1131        if (atomic_read(&ctl_cp->n_control) == 0) {
1132                IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1133                              "%s:%d to %s:%d\n",
1134                              IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1135                              ntohs(cp->cport),
1136                              IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1137                              ntohs(cp->vport));
1138
1139                return;
1140        }
1141        atomic_dec(&ctl_cp->n_control);
1142}
1143
1144static inline void
1145ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1146{
1147        if (cp->control) {
1148                IP_VS_ERR_BUF("request control ADD for already controlled: "
1149                              "%s:%d to %s:%d\n",
1150                              IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1151                              ntohs(cp->cport),
1152                              IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1153                              ntohs(cp->vport));
1154
1155                ip_vs_control_del(cp);
1156        }
1157
1158        IP_VS_DBG_BUF(7, "ADDing control for: "
1159                      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1160                      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1161                      ntohs(cp->cport),
1162                      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1163                      ntohs(ctl_cp->cport));
1164
1165        cp->control = ctl_cp;
1166        atomic_inc(&ctl_cp->n_control);
1167}
1168
1169/*
1170 * IPVS netns init & cleanup functions
1171 */
1172extern int ip_vs_estimator_net_init(struct net *net);
1173extern int ip_vs_control_net_init(struct net *net);
1174extern int ip_vs_protocol_net_init(struct net *net);
1175extern int ip_vs_app_net_init(struct net *net);
1176extern int ip_vs_conn_net_init(struct net *net);
1177extern int ip_vs_sync_net_init(struct net *net);
1178extern void ip_vs_conn_net_cleanup(struct net *net);
1179extern void ip_vs_app_net_cleanup(struct net *net);
1180extern void ip_vs_protocol_net_cleanup(struct net *net);
1181extern void ip_vs_control_net_cleanup(struct net *net);
1182extern void ip_vs_estimator_net_cleanup(struct net *net);
1183extern void ip_vs_sync_net_cleanup(struct net *net);
1184extern void ip_vs_service_net_cleanup(struct net *net);
1185
1186/*
1187 *      IPVS application functions
1188 *      (from ip_vs_app.c)
1189 */
1190#define IP_VS_APP_MAX_PORTS  8
1191extern struct ip_vs_app *register_ip_vs_app(struct net *net,
1192                                            struct ip_vs_app *app);
1193extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1194extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1195extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1196extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1197                                  __u16 proto, __u16 port);
1198extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1199extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1200
1201extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1202extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1203
1204void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1205void ip_vs_unbind_pe(struct ip_vs_service *svc);
1206int register_ip_vs_pe(struct ip_vs_pe *pe);
1207int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1208struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1209struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1210
1211/*
1212 * Use a #define to avoid all of module.h just for these trivial ops
1213 */
1214#define ip_vs_pe_get(pe)                        \
1215        if (pe && pe->module)                   \
1216                __module_get(pe->module);
1217
1218#define ip_vs_pe_put(pe)                        \
1219        if (pe && pe->module)                   \
1220                module_put(pe->module);
1221
1222/*
1223 *      IPVS protocol functions (from ip_vs_proto.c)
1224 */
1225extern int ip_vs_protocol_init(void);
1226extern void ip_vs_protocol_cleanup(void);
1227extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1228extern int *ip_vs_create_timeout_table(int *table, int size);
1229extern int
1230ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1231                        const char *name, int to);
1232extern void
1233ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1234                          const struct sk_buff *skb,
1235                          int offset, const char *msg);
1236
1237extern struct ip_vs_protocol ip_vs_protocol_tcp;
1238extern struct ip_vs_protocol ip_vs_protocol_udp;
1239extern struct ip_vs_protocol ip_vs_protocol_icmp;
1240extern struct ip_vs_protocol ip_vs_protocol_esp;
1241extern struct ip_vs_protocol ip_vs_protocol_ah;
1242extern struct ip_vs_protocol ip_vs_protocol_sctp;
1243
1244/*
1245 *      Registering/unregistering scheduler functions
1246 *      (from ip_vs_sched.c)
1247 */
1248extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1249extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1250extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1251                                struct ip_vs_scheduler *scheduler);
1252extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1253extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1254extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1255extern struct ip_vs_conn *
1256ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1257               struct ip_vs_proto_data *pd, int *ignored);
1258extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1259                        struct ip_vs_proto_data *pd);
1260
1261extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1262
1263
1264/*
1265 *      IPVS control data and functions (from ip_vs_ctl.c)
1266 */
1267extern struct ip_vs_stats ip_vs_stats;
1268extern int sysctl_ip_vs_sync_ver;
1269
1270extern struct ip_vs_service *
1271ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1272                  const union nf_inet_addr *vaddr, __be16 vport);
1273
1274static inline void ip_vs_service_put(struct ip_vs_service *svc)
1275{
1276        atomic_dec(&svc->usecnt);
1277}
1278
1279extern struct ip_vs_dest *
1280ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
1281                          const union nf_inet_addr *daddr, __be16 dport);
1282
1283extern int ip_vs_use_count_inc(void);
1284extern void ip_vs_use_count_dec(void);
1285extern int ip_vs_register_nl_ioctl(void);
1286extern void ip_vs_unregister_nl_ioctl(void);
1287extern int ip_vs_control_init(void);
1288extern void ip_vs_control_cleanup(void);
1289extern struct ip_vs_dest *
1290ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1291                __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1292                __u16 protocol, __u32 fwmark, __u32 flags);
1293extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1294
1295
1296/*
1297 *      IPVS sync daemon data and function prototypes
1298 *      (from ip_vs_sync.c)
1299 */
1300extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1301                             __u8 syncid);
1302extern int stop_sync_thread(struct net *net, int state);
1303extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1304
1305
1306/*
1307 *      IPVS rate estimator prototypes (from ip_vs_est.c)
1308 */
1309extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1310extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1311extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1312extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1313                                 struct ip_vs_stats *stats);
1314
1315/*
1316 *      Various IPVS packet transmitters (from ip_vs_xmit.c)
1317 */
1318extern int ip_vs_null_xmit
1319(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1320extern int ip_vs_bypass_xmit
1321(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1322extern int ip_vs_nat_xmit
1323(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1324extern int ip_vs_tunnel_xmit
1325(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1326extern int ip_vs_dr_xmit
1327(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1328extern int ip_vs_icmp_xmit
1329(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1330 int offset, unsigned int hooknum);
1331extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1332
1333#ifdef CONFIG_IP_VS_IPV6
1334extern int ip_vs_bypass_xmit_v6
1335(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1336extern int ip_vs_nat_xmit_v6
1337(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1338extern int ip_vs_tunnel_xmit_v6
1339(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1340extern int ip_vs_dr_xmit_v6
1341(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1342extern int ip_vs_icmp_xmit_v6
1343(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1344 int offset, unsigned int hooknum);
1345#endif
1346
1347#ifdef CONFIG_SYSCTL
1348/*
1349 *      This is a simple mechanism to ignore packets when
1350 *      we are loaded. Just set ip_vs_drop_rate to 'n' and
1351 *      we start to drop 1/rate of the packets
1352 */
1353
1354static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1355{
1356        if (!ipvs->drop_rate)
1357                return 0;
1358        if (--ipvs->drop_counter > 0)
1359                return 0;
1360        ipvs->drop_counter = ipvs->drop_rate;
1361        return 1;
1362}
1363#else
1364static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1365#endif
1366
1367/*
1368 *      ip_vs_fwd_tag returns the forwarding tag of the connection
1369 */
1370#define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
1371
1372static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1373{
1374        char fwd;
1375
1376        switch (IP_VS_FWD_METHOD(cp)) {
1377        case IP_VS_CONN_F_MASQ:
1378                fwd = 'M'; break;
1379        case IP_VS_CONN_F_LOCALNODE:
1380                fwd = 'L'; break;
1381        case IP_VS_CONN_F_TUNNEL:
1382                fwd = 'T'; break;
1383        case IP_VS_CONN_F_DROUTE:
1384                fwd = 'R'; break;
1385        case IP_VS_CONN_F_BYPASS:
1386                fwd = 'B'; break;
1387        default:
1388                fwd = '?'; break;
1389        }
1390        return fwd;
1391}
1392
1393extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1394                           struct ip_vs_conn *cp, int dir);
1395
1396#ifdef CONFIG_IP_VS_IPV6
1397extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1398                              struct ip_vs_conn *cp, int dir);
1399#endif
1400
1401extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1402
1403static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1404{
1405        __be32 diff[2] = { ~old, new };
1406
1407        return csum_partial(diff, sizeof(diff), oldsum);
1408}
1409
1410#ifdef CONFIG_IP_VS_IPV6
1411static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1412                                        __wsum oldsum)
1413{
1414        __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1415                            new[3],  new[2],  new[1],  new[0] };
1416
1417        return csum_partial(diff, sizeof(diff), oldsum);
1418}
1419#endif
1420
1421static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1422{
1423        __be16 diff[2] = { ~old, new };
1424
1425        return csum_partial(diff, sizeof(diff), oldsum);
1426}
1427
1428/*
1429 * Forget current conntrack (unconfirmed) and attach notrack entry
1430 */
1431static inline void ip_vs_notrack(struct sk_buff *skb)
1432{
1433#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1434        enum ip_conntrack_info ctinfo;
1435        struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1436
1437        if (!ct || !nf_ct_is_untracked(ct)) {
1438                nf_conntrack_put(skb->nfct);
1439                skb->nfct = &nf_ct_untracked_get()->ct_general;
1440                skb->nfctinfo = IP_CT_NEW;
1441                nf_conntrack_get(skb->nfct);
1442        }
1443#endif
1444}
1445
1446#ifdef CONFIG_IP_VS_NFCT
1447/*
1448 *      Netfilter connection tracking
1449 *      (from ip_vs_nfct.c)
1450 */
1451static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1452{
1453#ifdef CONFIG_SYSCTL
1454        return ipvs->sysctl_conntrack;
1455#else
1456        return 0;
1457#endif
1458}
1459
1460extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1461                                   int outin);
1462extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1463extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1464                                      struct ip_vs_conn *cp, u_int8_t proto,
1465                                      const __be16 port, int from_rs);
1466extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1467
1468#else
1469
1470static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1471{
1472        return 0;
1473}
1474
1475static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1476                                          struct ip_vs_conn *cp, int outin)
1477{
1478}
1479
1480static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1481{
1482        return NF_ACCEPT;
1483}
1484
1485static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1486{
1487}
1488/* CONFIG_IP_VS_NFCT */
1489#endif
1490
1491static inline unsigned int
1492ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1493{
1494        /*
1495         * We think the overhead of processing active connections is 256
1496         * times higher than that of inactive connections in average. (This
1497         * 256 times might not be accurate, we will change it later) We
1498         * use the following formula to estimate the overhead now:
1499         *                dest->activeconns*256 + dest->inactconns
1500         */
1501        return (atomic_read(&dest->activeconns) << 8) +
1502                atomic_read(&dest->inactconns);
1503}
1504
1505#endif  /* _NET_IP_VS_H */
1506