]> git.karo-electronics.de Git - karo-tx-linux.git/blob - include/net/ip_vs.h
Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
[karo-tx-linux.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
29 struct ip_vs_iphdr {
30         int len;
31         __u8 protocol;
32         union nf_inet_addr saddr;
33         union nf_inet_addr daddr;
34 };
35
36 static inline void
37 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
38 {
39 #ifdef CONFIG_IP_VS_IPV6
40         if (af == AF_INET6) {
41                 const struct ipv6hdr *iph = nh;
42                 iphdr->len = sizeof(struct ipv6hdr);
43                 iphdr->protocol = iph->nexthdr;
44                 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
45                 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
46         } else
47 #endif
48         {
49                 const struct iphdr *iph = nh;
50                 iphdr->len = iph->ihl * 4;
51                 iphdr->protocol = iph->protocol;
52                 iphdr->saddr.ip = iph->saddr;
53                 iphdr->daddr.ip = iph->daddr;
54         }
55 }
56
57 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
58                                    const union nf_inet_addr *src)
59 {
60 #ifdef CONFIG_IP_VS_IPV6
61         if (af == AF_INET6)
62                 ipv6_addr_copy(&dst->in6, &src->in6);
63         else
64 #endif
65         dst->ip = src->ip;
66 }
67
68 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
69                                    const union nf_inet_addr *b)
70 {
71 #ifdef CONFIG_IP_VS_IPV6
72         if (af == AF_INET6)
73                 return ipv6_addr_equal(&a->in6, &b->in6);
74 #endif
75         return a->ip == b->ip;
76 }
77
78 #ifdef CONFIG_IP_VS_DEBUG
79 #include <linux/net.h>
80
81 extern int ip_vs_get_debug_level(void);
82
83 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
84                                          const union nf_inet_addr *addr,
85                                          int *idx)
86 {
87         int len;
88 #ifdef CONFIG_IP_VS_IPV6
89         if (af == AF_INET6)
90                 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
91                                &addr->in6) + 1;
92         else
93 #endif
94                 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
95                                &addr->ip) + 1;
96
97         *idx += len;
98         BUG_ON(*idx > buf_len + 1);
99         return &buf[*idx - len];
100 }
101
102 #define IP_VS_DBG_BUF(level, msg, ...)                                  \
103         do {                                                            \
104                 char ip_vs_dbg_buf[160];                                \
105                 int ip_vs_dbg_idx = 0;                                  \
106                 if (level <= ip_vs_get_debug_level())                   \
107                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
108         } while (0)
109 #define IP_VS_ERR_BUF(msg...)                                           \
110         do {                                                            \
111                 char ip_vs_dbg_buf[160];                                \
112                 int ip_vs_dbg_idx = 0;                                  \
113                 pr_err(msg);                                            \
114         } while (0)
115
116 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
117 #define IP_VS_DBG_ADDR(af, addr)                                        \
118         ip_vs_dbg_addr(af, ip_vs_dbg_buf,                               \
119                        sizeof(ip_vs_dbg_buf), addr,                     \
120                        &ip_vs_dbg_idx)
121
122 #define IP_VS_DBG(level, msg, ...)                                      \
123         do {                                                            \
124                 if (level <= ip_vs_get_debug_level())                   \
125                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
126         } while (0)
127 #define IP_VS_DBG_RL(msg, ...)                                          \
128         do {                                                            \
129                 if (net_ratelimit())                                    \
130                         printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);  \
131         } while (0)
132 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)                         \
133         do {                                                            \
134                 if (level <= ip_vs_get_debug_level())                   \
135                         pp->debug_packet(pp, skb, ofs, msg);            \
136         } while (0)
137 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)                      \
138         do {                                                            \
139                 if (level <= ip_vs_get_debug_level() &&                 \
140                     net_ratelimit())                                    \
141                         pp->debug_packet(pp, skb, ofs, msg);            \
142         } while (0)
143 #else   /* NO DEBUGGING at ALL */
144 #define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
145 #define IP_VS_ERR_BUF(msg...)  do {} while (0)
146 #define IP_VS_DBG(level, msg...)  do {} while (0)
147 #define IP_VS_DBG_RL(msg...)  do {} while (0)
148 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)         do {} while (0)
149 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)      do {} while (0)
150 #endif
151
152 #define IP_VS_BUG() BUG()
153 #define IP_VS_ERR(msg...) pr_err(msg)
154 #define IP_VS_INFO(msg...) pr_info(msg)
155 #define IP_VS_WARNING(msg...) pr_warning(msg)
156 #define IP_VS_ERR_RL(msg...)                                            \
157         do {                                                            \
158                 if (net_ratelimit())                                    \
159                         pr_err(msg);                                    \
160         } while (0)
161
162 #ifdef CONFIG_IP_VS_DEBUG
163 #define EnterFunction(level)                                            \
164         do {                                                            \
165                 if (level <= ip_vs_get_debug_level())                   \
166                         printk(KERN_DEBUG                               \
167                                pr_fmt("Enter: %s, %s line %i\n"),       \
168                                __func__, __FILE__, __LINE__);           \
169         } while (0)
170 #define LeaveFunction(level)                                            \
171         do {                                                            \
172                 if (level <= ip_vs_get_debug_level())                   \
173                         printk(KERN_DEBUG                               \
174                                pr_fmt("Leave: %s, %s line %i\n"),       \
175                                __func__, __FILE__, __LINE__);           \
176         } while (0)
177 #else
178 #define EnterFunction(level)   do {} while (0)
179 #define LeaveFunction(level)   do {} while (0)
180 #endif
181
182 #define IP_VS_WAIT_WHILE(expr)  while (expr) { cpu_relax(); }
183
184
185 /*
186  *      The port number of FTP service (in network order).
187  */
188 #define FTPPORT  cpu_to_be16(21)
189 #define FTPDATA  cpu_to_be16(20)
190
191 /*
192  *      TCP State Values
193  */
194 enum {
195         IP_VS_TCP_S_NONE = 0,
196         IP_VS_TCP_S_ESTABLISHED,
197         IP_VS_TCP_S_SYN_SENT,
198         IP_VS_TCP_S_SYN_RECV,
199         IP_VS_TCP_S_FIN_WAIT,
200         IP_VS_TCP_S_TIME_WAIT,
201         IP_VS_TCP_S_CLOSE,
202         IP_VS_TCP_S_CLOSE_WAIT,
203         IP_VS_TCP_S_LAST_ACK,
204         IP_VS_TCP_S_LISTEN,
205         IP_VS_TCP_S_SYNACK,
206         IP_VS_TCP_S_LAST
207 };
208
209 /*
210  *      UDP State Values
211  */
212 enum {
213         IP_VS_UDP_S_NORMAL,
214         IP_VS_UDP_S_LAST,
215 };
216
217 /*
218  *      ICMP State Values
219  */
220 enum {
221         IP_VS_ICMP_S_NORMAL,
222         IP_VS_ICMP_S_LAST,
223 };
224
225 /*
226  *      Delta sequence info structure
227  *      Each ip_vs_conn has 2 (output AND input seq. changes).
228  *      Only used in the VS/NAT.
229  */
230 struct ip_vs_seq {
231         __u32                   init_seq;       /* Add delta from this seq */
232         __u32                   delta;          /* Delta in sequence numbers */
233         __u32                   previous_delta; /* Delta in sequence numbers
234                                                    before last resized pkt */
235 };
236
237
238 /*
239  *      IPVS statistics objects
240  */
241 struct ip_vs_estimator {
242         struct list_head        list;
243
244         u64                     last_inbytes;
245         u64                     last_outbytes;
246         u32                     last_conns;
247         u32                     last_inpkts;
248         u32                     last_outpkts;
249
250         u32                     cps;
251         u32                     inpps;
252         u32                     outpps;
253         u32                     inbps;
254         u32                     outbps;
255 };
256
257 struct ip_vs_stats
258 {
259         struct ip_vs_stats_user ustats;         /* statistics */
260         struct ip_vs_estimator  est;            /* estimator */
261
262         spinlock_t              lock;           /* spin lock */
263 };
264
265 struct dst_entry;
266 struct iphdr;
267 struct ip_vs_conn;
268 struct ip_vs_app;
269 struct sk_buff;
270
271 struct ip_vs_protocol {
272         struct ip_vs_protocol   *next;
273         char                    *name;
274         u16                     protocol;
275         u16                     num_states;
276         int                     dont_defrag;
277         atomic_t                appcnt;         /* counter of proto app incs */
278         int                     *timeout_table; /* protocol timeout table */
279
280         void (*init)(struct ip_vs_protocol *pp);
281
282         void (*exit)(struct ip_vs_protocol *pp);
283
284         int (*conn_schedule)(int af, struct sk_buff *skb,
285                              struct ip_vs_protocol *pp,
286                              int *verdict, struct ip_vs_conn **cpp);
287
288         struct ip_vs_conn *
289         (*conn_in_get)(int af,
290                        const struct sk_buff *skb,
291                        struct ip_vs_protocol *pp,
292                        const struct ip_vs_iphdr *iph,
293                        unsigned int proto_off,
294                        int inverse);
295
296         struct ip_vs_conn *
297         (*conn_out_get)(int af,
298                         const struct sk_buff *skb,
299                         struct ip_vs_protocol *pp,
300                         const struct ip_vs_iphdr *iph,
301                         unsigned int proto_off,
302                         int inverse);
303
304         int (*snat_handler)(struct sk_buff *skb,
305                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
306
307         int (*dnat_handler)(struct sk_buff *skb,
308                             struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
309
310         int (*csum_check)(int af, struct sk_buff *skb,
311                           struct ip_vs_protocol *pp);
312
313         const char *(*state_name)(int state);
314
315         int (*state_transition)(struct ip_vs_conn *cp, int direction,
316                                 const struct sk_buff *skb,
317                                 struct ip_vs_protocol *pp);
318
319         int (*register_app)(struct ip_vs_app *inc);
320
321         void (*unregister_app)(struct ip_vs_app *inc);
322
323         int (*app_conn_bind)(struct ip_vs_conn *cp);
324
325         void (*debug_packet)(struct ip_vs_protocol *pp,
326                              const struct sk_buff *skb,
327                              int offset,
328                              const char *msg);
329
330         void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
331
332         int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
333 };
334
335 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
336
337 /*
338  *      IP_VS structure allocated for each dynamically scheduled connection
339  */
340 struct ip_vs_conn {
341         struct list_head        c_list;         /* hashed list heads */
342
343         /* Protocol, addresses and port numbers */
344         u16                      af;            /* address family */
345         union nf_inet_addr       caddr;          /* client address */
346         union nf_inet_addr       vaddr;          /* virtual address */
347         union nf_inet_addr       daddr;          /* destination address */
348         __be16                   cport;
349         __be16                   vport;
350         __be16                   dport;
351         __u16                   protocol;       /* Which protocol (TCP/UDP) */
352
353         /* counter and timer */
354         atomic_t                refcnt;         /* reference count */
355         struct timer_list       timer;          /* Expiration timer */
356         volatile unsigned long  timeout;        /* timeout */
357
358         /* Flags and state transition */
359         spinlock_t              lock;           /* lock for state transition */
360         volatile __u16          flags;          /* status flags */
361         volatile __u16          state;          /* state info */
362         volatile __u16          old_state;      /* old state, to be used for
363                                                  * state transition triggerd
364                                                  * synchronization
365                                                  */
366
367         /* Control members */
368         struct ip_vs_conn       *control;       /* Master control connection */
369         atomic_t                n_control;      /* Number of controlled ones */
370         struct ip_vs_dest       *dest;          /* real server */
371         atomic_t                in_pkts;        /* incoming packet counter */
372
373         /* packet transmitter for different forwarding methods.  If it
374            mangles the packet, it must return NF_DROP or better NF_STOLEN,
375            otherwise this must be changed to a sk_buff **.
376          */
377         int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
378                            struct ip_vs_protocol *pp);
379
380         /* Note: we can group the following members into a structure,
381            in order to save more space, and the following members are
382            only used in VS/NAT anyway */
383         struct ip_vs_app        *app;           /* bound ip_vs_app object */
384         void                    *app_data;      /* Application private data */
385         struct ip_vs_seq        in_seq;         /* incoming seq. struct */
386         struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
387 };
388
389
390 /*
391  *      Extended internal versions of struct ip_vs_service_user and
392  *      ip_vs_dest_user for IPv6 support.
393  *
394  *      We need these to conveniently pass around service and destination
395  *      options, but unfortunately, we also need to keep the old definitions to
396  *      maintain userspace backwards compatibility for the setsockopt interface.
397  */
398 struct ip_vs_service_user_kern {
399         /* virtual service addresses */
400         u16                     af;
401         u16                     protocol;
402         union nf_inet_addr      addr;           /* virtual ip address */
403         u16                     port;
404         u32                     fwmark;         /* firwall mark of service */
405
406         /* virtual service options */
407         char                    *sched_name;
408         unsigned                flags;          /* virtual service flags */
409         unsigned                timeout;        /* persistent timeout in sec */
410         u32                     netmask;        /* persistent netmask */
411 };
412
413
414 struct ip_vs_dest_user_kern {
415         /* destination server address */
416         union nf_inet_addr      addr;
417         u16                     port;
418
419         /* real server options */
420         unsigned                conn_flags;     /* connection flags */
421         int                     weight;         /* destination weight */
422
423         /* thresholds for active connections */
424         u32                     u_threshold;    /* upper threshold */
425         u32                     l_threshold;    /* lower threshold */
426 };
427
428
429 /*
430  *      The information about the virtual service offered to the net
431  *      and the forwarding entries
432  */
433 struct ip_vs_service {
434         struct list_head        s_list;   /* for normal service table */
435         struct list_head        f_list;   /* for fwmark-based service table */
436         atomic_t                refcnt;   /* reference counter */
437         atomic_t                usecnt;   /* use counter */
438
439         u16                     af;       /* address family */
440         __u16                   protocol; /* which protocol (TCP/UDP) */
441         union nf_inet_addr      addr;     /* IP address for virtual service */
442         __be16                  port;     /* port number for the service */
443         __u32                   fwmark;   /* firewall mark of the service */
444         unsigned                flags;    /* service status flags */
445         unsigned                timeout;  /* persistent timeout in ticks */
446         __be32                  netmask;  /* grouping granularity */
447
448         struct list_head        destinations;  /* real server d-linked list */
449         __u32                   num_dests;     /* number of servers */
450         struct ip_vs_stats      stats;         /* statistics for the service */
451         struct ip_vs_app        *inc;     /* bind conns to this app inc */
452
453         /* for scheduling */
454         struct ip_vs_scheduler  *scheduler;    /* bound scheduler object */
455         rwlock_t                sched_lock;    /* lock sched_data */
456         void                    *sched_data;   /* scheduler application data */
457 };
458
459
460 /*
461  *      The real server destination forwarding entry
462  *      with ip address, port number, and so on.
463  */
464 struct ip_vs_dest {
465         struct list_head        n_list;   /* for the dests in the service */
466         struct list_head        d_list;   /* for table with all the dests */
467
468         u16                     af;             /* address family */
469         union nf_inet_addr      addr;           /* IP address of the server */
470         __be16                  port;           /* port number of the server */
471         volatile unsigned       flags;          /* dest status flags */
472         atomic_t                conn_flags;     /* flags to copy to conn */
473         atomic_t                weight;         /* server weight */
474
475         atomic_t                refcnt;         /* reference counter */
476         struct ip_vs_stats      stats;          /* statistics */
477
478         /* connection counters and thresholds */
479         atomic_t                activeconns;    /* active connections */
480         atomic_t                inactconns;     /* inactive connections */
481         atomic_t                persistconns;   /* persistent connections */
482         __u32                   u_threshold;    /* upper threshold */
483         __u32                   l_threshold;    /* lower threshold */
484
485         /* for destination cache */
486         spinlock_t              dst_lock;       /* lock of dst_cache */
487         struct dst_entry        *dst_cache;     /* destination cache entry */
488         u32                     dst_rtos;       /* RT_TOS(tos) for dst */
489
490         /* for virtual service */
491         struct ip_vs_service    *svc;           /* service it belongs to */
492         __u16                   protocol;       /* which protocol (TCP/UDP) */
493         union nf_inet_addr      vaddr;          /* virtual IP address */
494         __be16                  vport;          /* virtual port number */
495         __u32                   vfwmark;        /* firewall mark of service */
496 };
497
498
499 /*
500  *      The scheduler object
501  */
502 struct ip_vs_scheduler {
503         struct list_head        n_list;         /* d-linked list head */
504         char                    *name;          /* scheduler name */
505         atomic_t                refcnt;         /* reference counter */
506         struct module           *module;        /* THIS_MODULE/NULL */
507
508         /* scheduler initializing service */
509         int (*init_service)(struct ip_vs_service *svc);
510         /* scheduling service finish */
511         int (*done_service)(struct ip_vs_service *svc);
512         /* scheduler updating service */
513         int (*update_service)(struct ip_vs_service *svc);
514
515         /* selecting a server from the given service */
516         struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
517                                        const struct sk_buff *skb);
518 };
519
520
521 /*
522  *      The application module object (a.k.a. app incarnation)
523  */
524 struct ip_vs_app
525 {
526         struct list_head        a_list;         /* member in app list */
527         int                     type;           /* IP_VS_APP_TYPE_xxx */
528         char                    *name;          /* application module name */
529         __u16                   protocol;
530         struct module           *module;        /* THIS_MODULE/NULL */
531         struct list_head        incs_list;      /* list of incarnations */
532
533         /* members for application incarnations */
534         struct list_head        p_list;         /* member in proto app list */
535         struct ip_vs_app        *app;           /* its real application */
536         __be16                  port;           /* port number in net order */
537         atomic_t                usecnt;         /* usage counter */
538
539         /* output hook: return false if can't linearize. diff set for TCP.  */
540         int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
541                        struct sk_buff *, int *diff);
542
543         /* input hook: return false if can't linearize. diff set for TCP. */
544         int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
545                       struct sk_buff *, int *diff);
546
547         /* ip_vs_app initializer */
548         int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
549
550         /* ip_vs_app finish */
551         int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
552
553
554         /* not used now */
555         int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
556                          struct ip_vs_protocol *);
557
558         void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
559
560         int *                   timeout_table;
561         int *                   timeouts;
562         int                     timeouts_size;
563
564         int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
565                              int *verdict, struct ip_vs_conn **cpp);
566
567         struct ip_vs_conn *
568         (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
569                        const struct iphdr *iph, unsigned int proto_off,
570                        int inverse);
571
572         struct ip_vs_conn *
573         (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
574                         const struct iphdr *iph, unsigned int proto_off,
575                         int inverse);
576
577         int (*state_transition)(struct ip_vs_conn *cp, int direction,
578                                 const struct sk_buff *skb,
579                                 struct ip_vs_app *app);
580
581         void (*timeout_change)(struct ip_vs_app *app, int flags);
582 };
583
584
585 /*
586  *      IPVS core functions
587  *      (from ip_vs_core.c)
588  */
589 extern const char *ip_vs_proto_name(unsigned proto);
590 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
591 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
592
593 #define IP_VS_APP_TYPE_FTP      1
594
595 /*
596  *     ip_vs_conn handling functions
597  *     (from ip_vs_conn.c)
598  */
599
600 /*
601  *     IPVS connection entry hash table
602  */
603 #ifndef CONFIG_IP_VS_TAB_BITS
604 #define CONFIG_IP_VS_TAB_BITS   12
605 #endif
606
607 #define IP_VS_CONN_TAB_BITS     CONFIG_IP_VS_TAB_BITS
608 #define IP_VS_CONN_TAB_SIZE     (1 << IP_VS_CONN_TAB_BITS)
609 #define IP_VS_CONN_TAB_MASK     (IP_VS_CONN_TAB_SIZE - 1)
610
611 enum {
612         IP_VS_DIR_INPUT = 0,
613         IP_VS_DIR_OUTPUT,
614         IP_VS_DIR_INPUT_ONLY,
615         IP_VS_DIR_LAST,
616 };
617
618 extern struct ip_vs_conn *ip_vs_conn_in_get
619 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
620  const union nf_inet_addr *d_addr, __be16 d_port);
621
622 extern struct ip_vs_conn *ip_vs_ct_in_get
623 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
624  const union nf_inet_addr *d_addr, __be16 d_port);
625
626 extern struct ip_vs_conn *ip_vs_conn_out_get
627 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
628  const union nf_inet_addr *d_addr, __be16 d_port);
629
630 /* put back the conn without restarting its timer */
631 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
632 {
633         atomic_dec(&cp->refcnt);
634 }
635 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
636 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
637
638 extern struct ip_vs_conn *
639 ip_vs_conn_new(int af, int proto, const union nf_inet_addr *caddr, __be16 cport,
640                const union nf_inet_addr *vaddr, __be16 vport,
641                const union nf_inet_addr *daddr, __be16 dport, unsigned flags,
642                struct ip_vs_dest *dest);
643 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
644
645 extern const char * ip_vs_state_name(__u16 proto, int state);
646
647 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
648 extern int ip_vs_check_template(struct ip_vs_conn *ct);
649 extern void ip_vs_random_dropentry(void);
650 extern int ip_vs_conn_init(void);
651 extern void ip_vs_conn_cleanup(void);
652
653 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
654 {
655         struct ip_vs_conn *ctl_cp = cp->control;
656         if (!ctl_cp) {
657                 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
658                               "%s:%d to %s:%d\n",
659                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
660                               ntohs(cp->cport),
661                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
662                               ntohs(cp->vport));
663
664                 return;
665         }
666
667         IP_VS_DBG_BUF(7, "DELeting control for: "
668                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
669                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
670                       ntohs(cp->cport),
671                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
672                       ntohs(ctl_cp->cport));
673
674         cp->control = NULL;
675         if (atomic_read(&ctl_cp->n_control) == 0) {
676                 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
677                               "%s:%d to %s:%d\n",
678                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
679                               ntohs(cp->cport),
680                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
681                               ntohs(cp->vport));
682
683                 return;
684         }
685         atomic_dec(&ctl_cp->n_control);
686 }
687
688 static inline void
689 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
690 {
691         if (cp->control) {
692                 IP_VS_ERR_BUF("request control ADD for already controlled: "
693                               "%s:%d to %s:%d\n",
694                               IP_VS_DBG_ADDR(cp->af, &cp->caddr),
695                               ntohs(cp->cport),
696                               IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
697                               ntohs(cp->vport));
698
699                 ip_vs_control_del(cp);
700         }
701
702         IP_VS_DBG_BUF(7, "ADDing control for: "
703                       "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
704                       IP_VS_DBG_ADDR(cp->af, &cp->caddr),
705                       ntohs(cp->cport),
706                       IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
707                       ntohs(ctl_cp->cport));
708
709         cp->control = ctl_cp;
710         atomic_inc(&ctl_cp->n_control);
711 }
712
713
714 /*
715  *      IPVS application functions
716  *      (from ip_vs_app.c)
717  */
718 #define IP_VS_APP_MAX_PORTS  8
719 extern int register_ip_vs_app(struct ip_vs_app *app);
720 extern void unregister_ip_vs_app(struct ip_vs_app *app);
721 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
722 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
723 extern int
724 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
725 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
726 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
727
728 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
729 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
730 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
731                              char *o_buf, int o_len, char *n_buf, int n_len);
732 extern int ip_vs_app_init(void);
733 extern void ip_vs_app_cleanup(void);
734
735
736 /*
737  *      IPVS protocol functions (from ip_vs_proto.c)
738  */
739 extern int ip_vs_protocol_init(void);
740 extern void ip_vs_protocol_cleanup(void);
741 extern void ip_vs_protocol_timeout_change(int flags);
742 extern int *ip_vs_create_timeout_table(int *table, int size);
743 extern int
744 ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to);
745 extern void
746 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
747                           int offset, const char *msg);
748
749 extern struct ip_vs_protocol ip_vs_protocol_tcp;
750 extern struct ip_vs_protocol ip_vs_protocol_udp;
751 extern struct ip_vs_protocol ip_vs_protocol_icmp;
752 extern struct ip_vs_protocol ip_vs_protocol_esp;
753 extern struct ip_vs_protocol ip_vs_protocol_ah;
754
755
756 /*
757  *      Registering/unregistering scheduler functions
758  *      (from ip_vs_sched.c)
759  */
760 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
761 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
762 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
763                                 struct ip_vs_scheduler *scheduler);
764 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
765 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
766 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
767 extern struct ip_vs_conn *
768 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
769 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
770                         struct ip_vs_protocol *pp);
771
772
773 /*
774  *      IPVS control data and functions (from ip_vs_ctl.c)
775  */
776 extern int sysctl_ip_vs_cache_bypass;
777 extern int sysctl_ip_vs_expire_nodest_conn;
778 extern int sysctl_ip_vs_expire_quiescent_template;
779 extern int sysctl_ip_vs_sync_threshold[2];
780 extern int sysctl_ip_vs_nat_icmp_send;
781 extern struct ip_vs_stats ip_vs_stats;
782 extern const struct ctl_path net_vs_ctl_path[];
783
784 extern struct ip_vs_service *
785 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
786                   const union nf_inet_addr *vaddr, __be16 vport);
787
788 static inline void ip_vs_service_put(struct ip_vs_service *svc)
789 {
790         atomic_dec(&svc->usecnt);
791 }
792
793 extern struct ip_vs_dest *
794 ip_vs_lookup_real_service(int af, __u16 protocol,
795                           const union nf_inet_addr *daddr, __be16 dport);
796
797 extern int ip_vs_use_count_inc(void);
798 extern void ip_vs_use_count_dec(void);
799 extern int ip_vs_control_init(void);
800 extern void ip_vs_control_cleanup(void);
801 extern struct ip_vs_dest *
802 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
803                 const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol);
804 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
805
806
807 /*
808  *      IPVS sync daemon data and function prototypes
809  *      (from ip_vs_sync.c)
810  */
811 extern volatile int ip_vs_sync_state;
812 extern volatile int ip_vs_master_syncid;
813 extern volatile int ip_vs_backup_syncid;
814 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
815 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
816 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
817 extern int stop_sync_thread(int state);
818 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
819
820
821 /*
822  *      IPVS rate estimator prototypes (from ip_vs_est.c)
823  */
824 extern int ip_vs_estimator_init(void);
825 extern void ip_vs_estimator_cleanup(void);
826 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
827 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
828 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
829
830 /*
831  *      Various IPVS packet transmitters (from ip_vs_xmit.c)
832  */
833 extern int ip_vs_null_xmit
834 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
835 extern int ip_vs_bypass_xmit
836 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
837 extern int ip_vs_nat_xmit
838 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
839 extern int ip_vs_tunnel_xmit
840 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
841 extern int ip_vs_dr_xmit
842 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
843 extern int ip_vs_icmp_xmit
844 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
845 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
846
847 #ifdef CONFIG_IP_VS_IPV6
848 extern int ip_vs_bypass_xmit_v6
849 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
850 extern int ip_vs_nat_xmit_v6
851 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
852 extern int ip_vs_tunnel_xmit_v6
853 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
854 extern int ip_vs_dr_xmit_v6
855 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
856 extern int ip_vs_icmp_xmit_v6
857 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
858  int offset);
859 #endif
860
861 /*
862  *      This is a simple mechanism to ignore packets when
863  *      we are loaded. Just set ip_vs_drop_rate to 'n' and
864  *      we start to drop 1/rate of the packets
865  */
866 extern int ip_vs_drop_rate;
867 extern int ip_vs_drop_counter;
868
869 static __inline__ int ip_vs_todrop(void)
870 {
871         if (!ip_vs_drop_rate) return 0;
872         if (--ip_vs_drop_counter > 0) return 0;
873         ip_vs_drop_counter = ip_vs_drop_rate;
874         return 1;
875 }
876
877 /*
878  *      ip_vs_fwd_tag returns the forwarding tag of the connection
879  */
880 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
881
882 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
883 {
884         char fwd;
885
886         switch (IP_VS_FWD_METHOD(cp)) {
887         case IP_VS_CONN_F_MASQ:
888                 fwd = 'M'; break;
889         case IP_VS_CONN_F_LOCALNODE:
890                 fwd = 'L'; break;
891         case IP_VS_CONN_F_TUNNEL:
892                 fwd = 'T'; break;
893         case IP_VS_CONN_F_DROUTE:
894                 fwd = 'R'; break;
895         case IP_VS_CONN_F_BYPASS:
896                 fwd = 'B'; break;
897         default:
898                 fwd = '?'; break;
899         }
900         return fwd;
901 }
902
903 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
904                            struct ip_vs_conn *cp, int dir);
905
906 #ifdef CONFIG_IP_VS_IPV6
907 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
908                               struct ip_vs_conn *cp, int dir);
909 #endif
910
911 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
912
913 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
914 {
915         __be32 diff[2] = { ~old, new };
916
917         return csum_partial(diff, sizeof(diff), oldsum);
918 }
919
920 #ifdef CONFIG_IP_VS_IPV6
921 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
922                                         __wsum oldsum)
923 {
924         __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
925                             new[3],  new[2],  new[1],  new[0] };
926
927         return csum_partial(diff, sizeof(diff), oldsum);
928 }
929 #endif
930
931 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
932 {
933         __be16 diff[2] = { ~old, new };
934
935         return csum_partial(diff, sizeof(diff), oldsum);
936 }
937
938 #endif /* __KERNEL__ */
939
940 #endif  /* _NET_IP_VS_H */