IPVS: netns, use ip_vs_proto_data as param.
[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_proto_data *pd,
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                        const struct ip_vs_iphdr *iph,
382                        unsigned int proto_off,
383                        int inverse);
384
385         struct ip_vs_conn *
386         (*conn_out_get)(int af,
387                         const struct sk_buff *skb,
388                         const struct ip_vs_iphdr *iph,
389                         unsigned int proto_off,
390                         int inverse);
391
392         int (*snat_handler)(struct sk_buff *skb,
393                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
394
395         int (*dnat_handler)(struct sk_buff *skb,
396                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
397
398         int (*csum_check)(int af, struct sk_buff *skb,
399                           struct ip_vs_protocol *pp);
400
401         const char *(*state_name)(int state);
402
403         int (*state_transition)(struct ip_vs_conn *cp, int direction,
404                                 const struct sk_buff *skb,
405                                 struct ip_vs_proto_data *pd);
406
407         int (*register_app)(struct ip_vs_app *inc);
408
409         void (*unregister_app)(struct ip_vs_app *inc);
410
411         int (*app_conn_bind)(struct ip_vs_conn *cp);
412
413         void (*debug_packet)(int af, struct ip_vs_protocol *pp,
414                              const struct sk_buff *skb,
415                              int offset,
416                              const char *msg);
417
418         void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
419 };
420
421 /*
422  * protocol data per netns
423  */
424 struct ip_vs_proto_data {
425         struct ip_vs_proto_data *next;
426         struct ip_vs_protocol   *pp;
427         int                     *timeout_table; /* protocol timeout table */
428         atomic_t                appcnt;         /* counter of proto app incs. */
429         struct tcp_states_t     *tcp_state_table;
430 };
431
432 extern struct ip_vs_protocol   *ip_vs_proto_get(unsigned short proto);
433 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
434                                                      unsigned short proto);
435
436 struct ip_vs_conn_param {
437         const union nf_inet_addr        *caddr;
438         const union nf_inet_addr        *vaddr;
439         __be16                          cport;
440         __be16                          vport;
441         __u16                           protocol;
442         u16                             af;
443
444         const struct ip_vs_pe           *pe;
445         char                            *pe_data;
446         __u8                            pe_data_len;
447 };
448
449 /*
450  *      IP_VS structure allocated for each dynamically scheduled connection
451  */
452 struct ip_vs_conn {
453         struct list_head        c_list;         /* hashed list heads */
454
455         /* Protocol, addresses and port numbers */
456         u16                      af;            /* address family */
457         union nf_inet_addr       caddr;          /* client address */
458         union nf_inet_addr       vaddr;          /* virtual address */
459         union nf_inet_addr       daddr;          /* destination address */
460         volatile __u32           flags;          /* status flags */
461         __u32                    fwmark;         /* Fire wall mark from skb */
462         __be16                   cport;
463         __be16                   vport;
464         __be16                   dport;
465         __u16                   protocol;       /* Which protocol (TCP/UDP) */
466
467         /* counter and timer */
468         atomic_t                refcnt;         /* reference count */
469         struct timer_list       timer;          /* Expiration timer */
470         volatile unsigned long  timeout;        /* timeout */
471
472         /* Flags and state transition */
473         spinlock_t              lock;           /* lock for state transition */
474         volatile __u16          state;          /* state info */
475         volatile __u16          old_state;      /* old state, to be used for
476                                                  * state transition triggerd
477                                                  * synchronization
478                                                  */
479
480         /* Control members */
481         struct ip_vs_conn       *control;       /* Master control connection */
482         atomic_t                n_control;      /* Number of controlled ones */
483         struct ip_vs_dest       *dest;          /* real server */
484         atomic_t                in_pkts;        /* incoming packet counter */
485
486         /* packet transmitter for different forwarding methods.  If it
487            mangles the packet, it must return NF_DROP or better NF_STOLEN,
488            otherwise this must be changed to a sk_buff **.
489            NF_ACCEPT can be returned when destination is local.
490          */
491         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
492                            struct ip_vs_protocol *pp);
493
494         /* Note: we can group the following members into a structure,
495            in order to save more space, and the following members are
496            only used in VS/NAT anyway */
497         struct ip_vs_app        *app;           /* bound ip_vs_app object */
498         void                    *app_data;      /* Application private data */
499         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
500         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
501
502         const struct ip_vs_pe   *pe;
503         char                    *pe_data;
504         __u8                    pe_data_len;
505 };
506
507
508 /*
509  *      Extended internal versions of struct ip_vs_service_user and
510  *      ip_vs_dest_user for IPv6 support.
511  *
512  *      We need these to conveniently pass around service and destination
513  *      options, but unfortunately, we also need to keep the old definitions to
514  *      maintain userspace backwards compatibility for the setsockopt interface.
515  */
516 struct ip_vs_service_user_kern {
517         /* virtual service addresses */
518         u16                     af;
519         u16                     protocol;
520         union nf_inet_addr      addr;           /* virtual ip address */
521         u16                     port;
522         u32                     fwmark;         /* firwall mark of service */
523
524         /* virtual service options */
525         char                    *sched_name;
526         char                    *pe_name;
527         unsigned                flags;          /* virtual service flags */
528         unsigned                timeout;        /* persistent timeout in sec */
529         u32                     netmask;        /* persistent netmask */
530 };
531
532
533 struct ip_vs_dest_user_kern {
534         /* destination server address */
535         union nf_inet_addr      addr;
536         u16                     port;
537
538         /* real server options */
539         unsigned                conn_flags;     /* connection flags */
540         int                     weight;         /* destination weight */
541
542         /* thresholds for active connections */
543         u32                     u_threshold;    /* upper threshold */
544         u32                     l_threshold;    /* lower threshold */
545 };
546
547
548 /*
549  *      The information about the virtual service offered to the net
550  *      and the forwarding entries
551  */
552 struct ip_vs_service {
553         struct list_head        s_list;   /* for normal service table */
554         struct list_head        f_list;   /* for fwmark-based service table */
555         atomic_t                refcnt;   /* reference counter */
556         atomic_t                usecnt;   /* use counter */
557
558         u16                     af;       /* address family */
559         __u16                   protocol; /* which protocol (TCP/UDP) */
560         union nf_inet_addr      addr;     /* IP address for virtual service */
561         __be16                  port;     /* port number for the service */
562         __u32                   fwmark;   /* firewall mark of the service */
563         unsigned                flags;    /* service status flags */
564         unsigned                timeout;  /* persistent timeout in ticks */
565         __be32                  netmask;  /* grouping granularity */
566         struct net              *net;
567
568         struct list_head        destinations;  /* real server d-linked list */
569         __u32                   num_dests;     /* number of servers */
570         struct ip_vs_stats      stats;         /* statistics for the service */
571         struct ip_vs_app        *inc;     /* bind conns to this app inc */
572
573         /* for scheduling */
574         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
575         rwlock_t                sched_lock;    /* lock sched_data */
576         void                    *sched_data;   /* scheduler application data */
577
578         /* alternate persistence engine */
579         struct ip_vs_pe         *pe;
580 };
581
582
583 /*
584  *      The real server destination forwarding entry
585  *      with ip address, port number, and so on.
586  */
587 struct ip_vs_dest {
588         struct list_head        n_list;   /* for the dests in the service */
589         struct list_head        d_list;   /* for table with all the dests */
590
591         u16                     af;             /* address family */
592         union nf_inet_addr      addr;           /* IP address of the server */
593         __be16                  port;           /* port number of the server */
594         volatile unsigned       flags;          /* dest status flags */
595         atomic_t                conn_flags;     /* flags to copy to conn */
596         atomic_t                weight;         /* server weight */
597
598         atomic_t                refcnt;         /* reference counter */
599         struct ip_vs_stats      stats;          /* statistics */
600
601         /* connection counters and thresholds */
602         atomic_t                activeconns;    /* active connections */
603         atomic_t                inactconns;     /* inactive connections */
604         atomic_t                persistconns;   /* persistent connections */
605         __u32                   u_threshold;    /* upper threshold */
606         __u32                   l_threshold;    /* lower threshold */
607
608         /* for destination cache */
609         spinlock_t              dst_lock;       /* lock of dst_cache */
610         struct dst_entry        *dst_cache;     /* destination cache entry */
611         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
612         u32                     dst_cookie;
613 #ifdef CONFIG_IP_VS_IPV6
614         struct in6_addr         dst_saddr;
615 #endif
616
617         /* for virtual service */
618         struct ip_vs_service    *svc;           /* service it belongs to */
619         __u16                   protocol;       /* which protocol (TCP/UDP) */
620         union nf_inet_addr      vaddr;          /* virtual IP address */
621         __be16                  vport;          /* virtual port number */
622         __u32                   vfwmark;        /* firewall mark of service */
623 };
624
625
626 /*
627  *      The scheduler object
628  */
629 struct ip_vs_scheduler {
630         struct list_head        n_list;         /* d-linked list head */
631         char                    *name;          /* scheduler name */
632         atomic_t                refcnt;         /* reference counter */
633         struct module           *module;        /* THIS_MODULE/NULL */
634
635         /* scheduler initializing service */
636         int (*init_service)(struct ip_vs_service *svc);
637         /* scheduling service finish */
638         int (*done_service)(struct ip_vs_service *svc);
639         /* scheduler updating service */
640         int (*update_service)(struct ip_vs_service *svc);
641
642         /* selecting a server from the given service */
643         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
644                                        const struct sk_buff *skb);
645 };
646
647 /* The persistence engine object */
648 struct ip_vs_pe {
649         struct list_head        n_list;         /* d-linked list head */
650         char                    *name;          /* scheduler name */
651         atomic_t                refcnt;         /* reference counter */
652         struct module           *module;        /* THIS_MODULE/NULL */
653
654         /* get the connection template, if any */
655         int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
656         bool (*ct_match)(const struct ip_vs_conn_param *p,
657                          struct ip_vs_conn *ct);
658         u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
659                            bool inverse);
660         int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
661 };
662
663 /*
664  *      The application module object (a.k.a. app incarnation)
665  */
666 struct ip_vs_app {
667         struct list_head        a_list;         /* member in app list */
668         int                     type;           /* IP_VS_APP_TYPE_xxx */
669         char                    *name;          /* application module name */
670         __u16                   protocol;
671         struct module           *module;        /* THIS_MODULE/NULL */
672         struct list_head        incs_list;      /* list of incarnations */
673
674         /* members for application incarnations */
675         struct list_head        p_list;         /* member in proto app list */
676         struct ip_vs_app        *app;           /* its real application */
677         __be16                  port;           /* port number in net order */
678         atomic_t                usecnt;         /* usage counter */
679
680         /*
681          * output hook: Process packet in inout direction, diff set for TCP.
682          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
683          *         2=Mangled but checksum was not updated
684          */
685         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
686                        struct sk_buff *, int *diff);
687
688         /*
689          * input hook: Process packet in outin direction, diff set for TCP.
690          * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
691          *         2=Mangled but checksum was not updated
692          */
693         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
694                       struct sk_buff *, int *diff);
695
696         /* ip_vs_app initializer */
697         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
698
699         /* ip_vs_app finish */
700         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
701
702
703         /* not used now */
704         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
705                          struct ip_vs_protocol *);
706
707         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
708
709         int *                   timeout_table;
710         int *                   timeouts;
711         int                     timeouts_size;
712
713         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
714                              int *verdict, struct ip_vs_conn **cpp);
715
716         struct ip_vs_conn *
717         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
718                        const struct iphdr *iph, unsigned int proto_off,
719                        int inverse);
720
721         struct ip_vs_conn *
722         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
723                         const struct iphdr *iph, unsigned int proto_off,
724                         int inverse);
725
726         int (*state_transition)(struct ip_vs_conn *cp, int direction,
727                                 const struct sk_buff *skb,
728                                 struct ip_vs_app *app);
729
730         void (*timeout_change)(struct ip_vs_app *app, int flags);
731 };
732
733
734 /*
735  *      IPVS core functions
736  *      (from ip_vs_core.c)
737  */
738 extern const char *ip_vs_proto_name(unsigned proto);
739 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
740 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
741
742 #define IP_VS_APP_TYPE_FTP      1
743
744 /*
745  *     ip_vs_conn handling functions
746  *     (from ip_vs_conn.c)
747  */
748
749 enum {
750         IP_VS_DIR_INPUT = 0,
751         IP_VS_DIR_OUTPUT,
752         IP_VS_DIR_INPUT_ONLY,
753         IP_VS_DIR_LAST,
754 };
755
756 static inline void ip_vs_conn_fill_param(int af, int protocol,
757                                          const union nf_inet_addr *caddr,
758                                          __be16 cport,
759                                          const union nf_inet_addr *vaddr,
760                                          __be16 vport,
761                                          struct ip_vs_conn_param *p)
762 {
763         p->af = af;
764         p->protocol = protocol;
765         p->caddr = caddr;
766         p->cport = cport;
767         p->vaddr = vaddr;
768         p->vport = vport;
769         p->pe = NULL;
770         p->pe_data = NULL;
771 }
772
773 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
774 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
775
776 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
777                                             const struct ip_vs_iphdr *iph,
778                                             unsigned int proto_off,
779                                             int inverse);
780
781 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
782
783 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
784                                              const struct ip_vs_iphdr *iph,
785                                              unsigned int proto_off,
786                                              int inverse);
787
788 /* put back the conn without restarting its timer */
789 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
790 {
791         atomic_dec(&cp->refcnt);
792 }
793 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
794 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
795
796 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
797                                   const union nf_inet_addr *daddr,
798                                   __be16 dport, unsigned flags,
799                                   struct ip_vs_dest *dest, __u32 fwmark);
800 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
801
802 extern const char * ip_vs_state_name(__u16 proto, int state);
803
804 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
805 extern int ip_vs_check_template(struct ip_vs_conn *ct);
806 extern void ip_vs_random_dropentry(void);
807 extern int ip_vs_conn_init(void);
808 extern void ip_vs_conn_cleanup(void);
809
810 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
811 {
812         struct ip_vs_conn *ctl_cp = cp->control;
813         if (!ctl_cp) {
814                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
815                               "%s:%d to %s:%d\n",
816                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
817                               ntohs(cp->cport),
818                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
819                               ntohs(cp->vport));
820
821                 return;
822         }
823
824         IP_VS_DBG_BUF(7, "DELeting control for: "
825                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
826                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
827                       ntohs(cp->cport),
828                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
829                       ntohs(ctl_cp->cport));
830
831         cp->control = NULL;
832         if (atomic_read(&ctl_cp->n_control) == 0) {
833                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
834                               "%s:%d to %s:%d\n",
835                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
836                               ntohs(cp->cport),
837                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
838                               ntohs(cp->vport));
839
840                 return;
841         }
842         atomic_dec(&ctl_cp->n_control);
843 }
844
845 static inline void
846 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
847 {
848         if (cp->control) {
849                 IP_VS_ERR_BUF("request control ADD for already controlled: "
850                               "%s:%d to %s:%d\n",
851                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
852                               ntohs(cp->cport),
853                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
854                               ntohs(cp->vport));
855
856                 ip_vs_control_del(cp);
857         }
858
859         IP_VS_DBG_BUF(7, "ADDing control for: "
860                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
861                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
862                       ntohs(cp->cport),
863                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
864                       ntohs(ctl_cp->cport));
865
866         cp->control = ctl_cp;
867         atomic_inc(&ctl_cp->n_control);
868 }
869
870
871 /*
872  *      IPVS application functions
873  *      (from ip_vs_app.c)
874  */
875 #define IP_VS_APP_MAX_PORTS  8
876 extern int register_ip_vs_app(struct ip_vs_app *app);
877 extern void unregister_ip_vs_app(struct ip_vs_app *app);
878 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
879 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
880 extern int
881 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
882 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
883 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
884
885 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
886 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
887 extern int ip_vs_app_init(void);
888 extern void ip_vs_app_cleanup(void);
889
890 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
891 void ip_vs_unbind_pe(struct ip_vs_service *svc);
892 int register_ip_vs_pe(struct ip_vs_pe *pe);
893 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
894 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
895 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
896
897 static inline void ip_vs_pe_get(const struct ip_vs_pe *pe)
898 {
899         if (pe && pe->module)
900                 __module_get(pe->module);
901 }
902
903 static inline void ip_vs_pe_put(const struct ip_vs_pe *pe)
904 {
905         if (pe && pe->module)
906                 module_put(pe->module);
907 }
908
909 /*
910  *      IPVS protocol functions (from ip_vs_proto.c)
911  */
912 extern int ip_vs_protocol_init(void);
913 extern void ip_vs_protocol_cleanup(void);
914 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
915 extern int *ip_vs_create_timeout_table(int *table, int size);
916 extern int
917 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
918                         const char *name, int to);
919 extern void
920 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
921                           const struct sk_buff *skb,
922                           int offset, const char *msg);
923
924 extern struct ip_vs_protocol ip_vs_protocol_tcp;
925 extern struct ip_vs_protocol ip_vs_protocol_udp;
926 extern struct ip_vs_protocol ip_vs_protocol_icmp;
927 extern struct ip_vs_protocol ip_vs_protocol_esp;
928 extern struct ip_vs_protocol ip_vs_protocol_ah;
929 extern struct ip_vs_protocol ip_vs_protocol_sctp;
930
931 /*
932  *      Registering/unregistering scheduler functions
933  *      (from ip_vs_sched.c)
934  */
935 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
936 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
937 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
938                                 struct ip_vs_scheduler *scheduler);
939 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
940 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
941 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
942 extern struct ip_vs_conn *
943 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
944                struct ip_vs_proto_data *pd, int *ignored);
945 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
946                         struct ip_vs_proto_data *pd);
947
948
949 /*
950  *      IPVS control data and functions (from ip_vs_ctl.c)
951  */
952 extern int sysctl_ip_vs_cache_bypass;
953 extern int sysctl_ip_vs_expire_nodest_conn;
954 extern int sysctl_ip_vs_expire_quiescent_template;
955 extern int sysctl_ip_vs_sync_threshold[2];
956 extern int sysctl_ip_vs_nat_icmp_send;
957 extern int sysctl_ip_vs_conntrack;
958 extern int sysctl_ip_vs_snat_reroute;
959 extern struct ip_vs_stats ip_vs_stats;
960 extern const struct ctl_path net_vs_ctl_path[];
961 extern int sysctl_ip_vs_sync_ver;
962
963 extern void ip_vs_sync_switch_mode(int mode);
964 extern struct ip_vs_service *
965 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
966                   const union nf_inet_addr *vaddr, __be16 vport);
967
968 static inline void ip_vs_service_put(struct ip_vs_service *svc)
969 {
970         atomic_dec(&svc->usecnt);
971 }
972
973 extern struct ip_vs_dest *
974 ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
975                           const union nf_inet_addr *daddr, __be16 dport);
976
977 extern int ip_vs_use_count_inc(void);
978 extern void ip_vs_use_count_dec(void);
979 extern int ip_vs_control_init(void);
980 extern void ip_vs_control_cleanup(void);
981 extern struct ip_vs_dest *
982 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
983                 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
984                 __u16 protocol, __u32 fwmark);
985 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
986
987
988 /*
989  *      IPVS sync daemon data and function prototypes
990  *      (from ip_vs_sync.c)
991  */
992 extern volatile int ip_vs_sync_state;
993 extern volatile int ip_vs_master_syncid;
994 extern volatile int ip_vs_backup_syncid;
995 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
996 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
997 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
998 extern int stop_sync_thread(int state);
999 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
1000 extern int ip_vs_sync_init(void);
1001 extern void ip_vs_sync_cleanup(void);
1002
1003
1004 /*
1005  *      IPVS rate estimator prototypes (from ip_vs_est.c)
1006  */
1007 extern int ip_vs_estimator_init(void);
1008 extern void ip_vs_estimator_cleanup(void);
1009 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
1010 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
1011 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1012
1013 /*
1014  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
1015  */
1016 extern int ip_vs_null_xmit
1017 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1018 extern int ip_vs_bypass_xmit
1019 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1020 extern int ip_vs_nat_xmit
1021 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1022 extern int ip_vs_tunnel_xmit
1023 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1024 extern int ip_vs_dr_xmit
1025 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1026 extern int ip_vs_icmp_xmit
1027 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
1028 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1029
1030 #ifdef CONFIG_IP_VS_IPV6
1031 extern int ip_vs_bypass_xmit_v6
1032 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1033 extern int ip_vs_nat_xmit_v6
1034 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1035 extern int ip_vs_tunnel_xmit_v6
1036 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1037 extern int ip_vs_dr_xmit_v6
1038 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1039 extern int ip_vs_icmp_xmit_v6
1040 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1041  int offset);
1042 #endif
1043
1044 /*
1045  *      This is a simple mechanism to ignore packets when
1046  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
1047  *      we start to drop 1/rate of the packets
1048  */
1049 extern int ip_vs_drop_rate;
1050 extern int ip_vs_drop_counter;
1051
1052 static __inline__ int ip_vs_todrop(void)
1053 {
1054         if (!ip_vs_drop_rate) return 0;
1055         if (--ip_vs_drop_counter > 0) return 0;
1056         ip_vs_drop_counter = ip_vs_drop_rate;
1057         return 1;
1058 }
1059
1060 /*
1061  *      ip_vs_fwd_tag returns the forwarding tag of the connection
1062  */
1063 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
1064
1065 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1066 {
1067         char fwd;
1068
1069         switch (IP_VS_FWD_METHOD(cp)) {
1070         case IP_VS_CONN_F_MASQ:
1071                 fwd = 'M'; break;
1072         case IP_VS_CONN_F_LOCALNODE:
1073                 fwd = 'L'; break;
1074         case IP_VS_CONN_F_TUNNEL:
1075                 fwd = 'T'; break;
1076         case IP_VS_CONN_F_DROUTE:
1077                 fwd = 'R'; break;
1078         case IP_VS_CONN_F_BYPASS:
1079                 fwd = 'B'; break;
1080         default:
1081                 fwd = '?'; break;
1082         }
1083         return fwd;
1084 }
1085
1086 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1087                            struct ip_vs_conn *cp, int dir);
1088
1089 #ifdef CONFIG_IP_VS_IPV6
1090 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1091                               struct ip_vs_conn *cp, int dir);
1092 #endif
1093
1094 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1095
1096 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1097 {
1098         __be32 diff[2] = { ~old, new };
1099
1100         return csum_partial(diff, sizeof(diff), oldsum);
1101 }
1102
1103 #ifdef CONFIG_IP_VS_IPV6
1104 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1105                                         __wsum oldsum)
1106 {
1107         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1108                             new[3],  new[2],  new[1],  new[0] };
1109
1110         return csum_partial(diff, sizeof(diff), oldsum);
1111 }
1112 #endif
1113
1114 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1115 {
1116         __be16 diff[2] = { ~old, new };
1117
1118         return csum_partial(diff, sizeof(diff), oldsum);
1119 }
1120
1121 /*
1122  * Forget current conntrack (unconfirmed) and attach notrack entry
1123  */
1124 static inline void ip_vs_notrack(struct sk_buff *skb)
1125 {
1126 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1127         enum ip_conntrack_info ctinfo;
1128         struct nf_conn *ct = ct = nf_ct_get(skb, &ctinfo);
1129
1130         if (!ct || !nf_ct_is_untracked(ct)) {
1131                 nf_reset(skb);
1132                 skb->nfct = &nf_ct_untracked_get()->ct_general;
1133                 skb->nfctinfo = IP_CT_NEW;
1134                 nf_conntrack_get(skb->nfct);
1135         }
1136 #endif
1137 }
1138
1139 #ifdef CONFIG_IP_VS_NFCT
1140 /*
1141  *      Netfilter connection tracking
1142  *      (from ip_vs_nfct.c)
1143  */
1144 static inline int ip_vs_conntrack_enabled(void)
1145 {
1146         return sysctl_ip_vs_conntrack;
1147 }
1148
1149 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1150                                    int outin);
1151 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1152 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1153                                       struct ip_vs_conn *cp, u_int8_t proto,
1154                                       const __be16 port, int from_rs);
1155 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1156
1157 #else
1158
1159 static inline int ip_vs_conntrack_enabled(void)
1160 {
1161         return 0;
1162 }
1163
1164 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1165                                           struct ip_vs_conn *cp, int outin)
1166 {
1167 }
1168
1169 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1170                                           struct ip_vs_conn *cp)
1171 {
1172         return NF_ACCEPT;
1173 }
1174
1175 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1176 {
1177 }
1178 /* CONFIG_IP_VS_NFCT */
1179 #endif
1180
1181 #endif /* __KERNEL__ */
1182
1183 #endif  /* _NET_IP_VS_H */