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