]> bbs.cooldavid.org Git - net-next-2.6.git/blame - include/net/ip_vs.h
git://bbs.cooldavid.org / net-next-2.6.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
IPVS: Add and bind IPv6 xmit functions
[net-next-2.6.git] / include / net / ip_vs.h
CommitLineData
1da177e4
LT		 1/*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
bc4768eb
JV		 6#ifndef _NET_IP_VS_H
7#define _NET_IP_VS_H
1da177e4 8
bc4768eb 9#include <linux/ip_vs.h> /* definitions shared with userland */
1da177e4 10
bc4768eb 11/* old ipvsadm versions still include this file directly */
1da177e4
LT		 12#ifdef __KERNEL__
13
bc4768eb
JV		 14#include <asm/types.h> /* for __uXX types */
15
16#include <linux/sysctl.h> /* for ctl_path */
1da177e4
LT		 17#include <linux/list.h> /* for struct list_head */
18#include <linux/spinlock.h> /* for struct rwlock_t */
1da177e4 19#include <asm/atomic.h> /* for struct atomic_t */
1da177e4 20#include <linux/compiler.h>
14c85021 21#include <linux/timer.h>
1da177e4 22
14c85021 23#include <net/checksum.h>
e7ade46a
JV		 24#include <linux/netfilter.h> /* for union nf_inet_addr */
25#include <linux/ipv6.h> /* for struct ipv6hdr */
26#include <net/ipv6.h> /* for ipv6_addr_copy */
1da177e4 27
64aae3cb
JV		 28struct ip_vs_iphdr {
29 int len;
30 __u8 protocol;
31 union nf_inet_addr saddr;
32 union nf_inet_addr daddr;
33};
34
35static inline void
36ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
37{
38#ifdef CONFIG_IP_VS_IPV6
39 if (af == AF_INET6) {
40 const struct ipv6hdr *iph = nh;
41 iphdr->len = sizeof(struct ipv6hdr);
42 iphdr->protocol = iph->nexthdr;
43 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
44 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
45 } else
46#endif
47 {
48 const struct iphdr *iph = nh;
49 iphdr->len = iph->ihl * 4;
50 iphdr->protocol = iph->protocol;
51 iphdr->saddr.ip = iph->saddr;
52 iphdr->daddr.ip = iph->daddr;
53 }
54}
55
56static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
57 const union nf_inet_addr *src)
58{
59#ifdef CONFIG_IP_VS_IPV6
60 if (af == AF_INET6)
61 ipv6_addr_copy(&dst->in6, &src->in6);
62 else
63#endif
64 dst->ip = src->ip;
65}
66
67static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
68 const union nf_inet_addr *b)
69{
70#ifdef CONFIG_IP_VS_IPV6
71 if (af == AF_INET6)
72 return ipv6_addr_equal(&a->in6, &b->in6);
73#endif
74 return a->ip == b->ip;
75}
76
1da177e4 77#ifdef CONFIG_IP_VS_DEBUG
14c85021
ACM		 78#include <linux/net.h>
79
1da177e4 80extern int ip_vs_get_debug_level(void);
c842a3ad
JV		 81
82static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
83 const union nf_inet_addr *addr,
84 int *idx)
85{
86 int len;
87#ifdef CONFIG_IP_VS_IPV6
88 if (af == AF_INET6)
89 len = snprintf(&buf[*idx], buf_len - *idx, "[" NIP6_FMT "]",
90 NIP6(addr->in6)) + 1;
91 else
92#endif
93 len = snprintf(&buf[*idx], buf_len - *idx, NIPQUAD_FMT,
94 NIPQUAD(addr->ip)) + 1;
95
96 *idx += len;
97 BUG_ON(*idx > buf_len + 1);
98 return &buf[*idx - len];
99}
100
101#define IP_VS_DBG_BUF(level, msg...) \
102 do { \
103 char ip_vs_dbg_buf[160]; \
104 int ip_vs_dbg_idx = 0; \
105 if (level <= ip_vs_get_debug_level()) \
106 printk(KERN_DEBUG "IPVS: " msg); \
107 } while (0)
108#define IP_VS_ERR_BUF(msg...) \
109 do { \
110 char ip_vs_dbg_buf[160]; \
111 int ip_vs_dbg_idx = 0; \
112 printk(KERN_ERR "IPVS: " msg); \
113 } while (0)
114
115/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
116#define IP_VS_DBG_ADDR(af, addr) \
117 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
118 sizeof(ip_vs_dbg_buf), addr, \
119 &ip_vs_dbg_idx)
120
1da177e4
LT		 121#define IP_VS_DBG(level, msg...) \
122 do { \
123 if (level <= ip_vs_get_debug_level()) \
124 printk(KERN_DEBUG "IPVS: " msg); \
125 } while (0)
126#define IP_VS_DBG_RL(msg...) \
127 do { \
128 if (net_ratelimit()) \
129 printk(KERN_DEBUG "IPVS: " msg); \
130 } while (0)
131#define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) \
132 do { \
133 if (level <= ip_vs_get_debug_level()) \
134 pp->debug_packet(pp, skb, ofs, msg); \
135 } while (0)
136#define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) \
137 do { \
138 if (level <= ip_vs_get_debug_level() && \
139 net_ratelimit()) \
140 pp->debug_packet(pp, skb, ofs, msg); \
141 } while (0)
142#else /* NO DEBUGGING at ALL */
c842a3ad
JV		 143#define IP_VS_DBG_BUF(level, msg...) do {} while (0)
144#define IP_VS_ERR_BUF(msg...) do {} while (0)
1da177e4
LT		 145#define IP_VS_DBG(level, msg...) do {} while (0)
146#define IP_VS_DBG_RL(msg...) do {} while (0)
147#define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) do {} while (0)
148#define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) do {} while (0)
149#endif
150
151#define IP_VS_BUG() BUG()
152#define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg)
153#define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg)
154#define IP_VS_WARNING(msg...) \
155 printk(KERN_WARNING "IPVS: " msg)
156#define IP_VS_ERR_RL(msg...) \
157 do { \
158 if (net_ratelimit()) \
159 printk(KERN_ERR "IPVS: " 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 "Enter: %s, %s line %i\n", \
167 __FUNCTION__, __FILE__, __LINE__); \
168 } while (0)
169#define LeaveFunction(level) \
170 do { \
171 if (level <= ip_vs_get_debug_level()) \
172 printk(KERN_DEBUG "Leave: %s, %s line %i\n", \
173 __FUNCTION__, __FILE__, __LINE__); \
174 } while (0)
175#else
176#define EnterFunction(level) do {} while (0)
177#define LeaveFunction(level) do {} while (0)
178#endif
179
180#define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
181
182
183/*
184 * The port number of FTP service (in network order).
185 */
186#define FTPPORT __constant_htons(21)
187#define FTPDATA __constant_htons(20)
188
1da177e4
LT		 189/*
190 * TCP State Values
191 */
192enum {
193 IP_VS_TCP_S_NONE = 0,
194 IP_VS_TCP_S_ESTABLISHED,
195 IP_VS_TCP_S_SYN_SENT,
196 IP_VS_TCP_S_SYN_RECV,
197 IP_VS_TCP_S_FIN_WAIT,
198 IP_VS_TCP_S_TIME_WAIT,
199 IP_VS_TCP_S_CLOSE,
200 IP_VS_TCP_S_CLOSE_WAIT,
201 IP_VS_TCP_S_LAST_ACK,
202 IP_VS_TCP_S_LISTEN,
203 IP_VS_TCP_S_SYNACK,
204 IP_VS_TCP_S_LAST
205};
206
207/*
208 * UDP State Values
209 */
210enum {
211 IP_VS_UDP_S_NORMAL,
212 IP_VS_UDP_S_LAST,
213};
214
215/*
216 * ICMP State Values
217 */
218enum {
219 IP_VS_ICMP_S_NORMAL,
220 IP_VS_ICMP_S_LAST,
221};
222
223/*
224 * Delta sequence info structure
225 * Each ip_vs_conn has 2 (output AND input seq. changes).
226 * Only used in the VS/NAT.
227 */
228struct ip_vs_seq {
229 __u32 init_seq; /* Add delta from this seq */
230 __u32 delta; /* Delta in sequence numbers */
231 __u32 previous_delta; /* Delta in sequence numbers
232 before last resized pkt */
233};
234
235
236/*
3a14a313 237 * IPVS statistics objects
1da177e4 238 */
3a14a313
SW		 239struct ip_vs_estimator {
240 struct list_head list;
241
242 u64 last_inbytes;
243 u64 last_outbytes;
244 u32 last_conns;
245 u32 last_inpkts;
246 u32 last_outpkts;
247
248 u32 cps;
249 u32 inpps;
250 u32 outpps;
251 u32 inbps;
252 u32 outbps;
253};
254
1da177e4
LT		 255struct ip_vs_stats
256{
257 __u32 conns; /* connections scheduled */
258 __u32 inpkts; /* incoming packets */
259 __u32 outpkts; /* outgoing packets */
260 __u64 inbytes; /* incoming bytes */
261 __u64 outbytes; /* outgoing bytes */
262
263 __u32 cps; /* current connection rate */
264 __u32 inpps; /* current in packet rate */
265 __u32 outpps; /* current out packet rate */
266 __u32 inbps; /* current in byte rate */
267 __u32 outbps; /* current out byte rate */
268
3a14a313
SW		 269 /*
270 * Don't add anything before the lock, because we use memcpy() to copy
271 * the members before the lock to struct ip_vs_stats_user in
272 * ip_vs_ctl.c.
273 */
274
1da177e4 275 spinlock_t lock; /* spin lock */
3a14a313
SW		 276
277 struct ip_vs_estimator est; /* estimator */
1da177e4
LT		 278};
279
14c85021
ACM		 280struct dst_entry;
281struct iphdr;
1da177e4
LT		 282struct ip_vs_conn;
283struct ip_vs_app;
14c85021 284struct sk_buff;
1da177e4
LT		 285
286struct ip_vs_protocol {
287 struct ip_vs_protocol *next;
288 char *name;
2ad17def
JA		 289 u16 protocol;
290 u16 num_states;
1da177e4
LT		 291 int dont_defrag;
292 atomic_t appcnt; /* counter of proto app incs */
293 int *timeout_table; /* protocol timeout table */
294
295 void (*init)(struct ip_vs_protocol *pp);
296
297 void (*exit)(struct ip_vs_protocol *pp);
298
51ef348b 299 int (*conn_schedule)(int af, struct sk_buff *skb,
1da177e4
LT		 300 struct ip_vs_protocol *pp,
301 int *verdict, struct ip_vs_conn **cpp);
302
303 struct ip_vs_conn *
51ef348b
JV		 304 (*conn_in_get)(int af,
305 const struct sk_buff *skb,
1da177e4 306 struct ip_vs_protocol *pp,
51ef348b 307 const struct ip_vs_iphdr *iph,
1da177e4
LT		 308 unsigned int proto_off,
309 int inverse);
310
311 struct ip_vs_conn *
51ef348b
JV		 312 (*conn_out_get)(int af,
313 const struct sk_buff *skb,
1da177e4 314 struct ip_vs_protocol *pp,
51ef348b 315 const struct ip_vs_iphdr *iph,
1da177e4
LT		 316 unsigned int proto_off,
317 int inverse);
318
3db05fea 319 int (*snat_handler)(struct sk_buff *skb,
1da177e4
LT		 320 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
321
3db05fea 322 int (*dnat_handler)(struct sk_buff *skb,
1da177e4
LT		 323 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
324
51ef348b
JV		 325 int (*csum_check)(int af, struct sk_buff *skb,
326 struct ip_vs_protocol *pp);
1da177e4
LT		 327
328 const char *(*state_name)(int state);
329
330 int (*state_transition)(struct ip_vs_conn *cp, int direction,
331 const struct sk_buff *skb,
332 struct ip_vs_protocol *pp);
333
334 int (*register_app)(struct ip_vs_app *inc);
335
336 void (*unregister_app)(struct ip_vs_app *inc);
337
338 int (*app_conn_bind)(struct ip_vs_conn *cp);
339
340 void (*debug_packet)(struct ip_vs_protocol *pp,
341 const struct sk_buff *skb,
342 int offset,
343 const char *msg);
344
345 void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
346
347 int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
348};
349
350extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
351
352/*
353 * IP_VS structure allocated for each dynamically scheduled connection
354 */
355struct ip_vs_conn {
356 struct list_head c_list; /* hashed list heads */
357
358 /* Protocol, addresses and port numbers */
e7ade46a
JV		 359 u16 af; /* address family */
360 union nf_inet_addr caddr; /* client address */
361 union nf_inet_addr vaddr; /* virtual address */
362 union nf_inet_addr daddr; /* destination address */
014d730d
AV		 363 __be16 cport;
364 __be16 vport;
365 __be16 dport;
1da177e4
LT		 366 __u16 protocol; /* Which protocol (TCP/UDP) */
367
368 /* counter and timer */
369 atomic_t refcnt; /* reference count */
370 struct timer_list timer; /* Expiration timer */
371 volatile unsigned long timeout; /* timeout */
372
373 /* Flags and state transition */
374 spinlock_t lock; /* lock for state transition */
375 volatile __u16 flags; /* status flags */
376 volatile __u16 state; /* state info */
efac5276
RB		 377 volatile __u16 old_state; /* old state, to be used for
378 * state transition triggerd
379 * synchronization
380 */
1da177e4
LT		 381
382 /* Control members */
383 struct ip_vs_conn *control; /* Master control connection */
384 atomic_t n_control; /* Number of controlled ones */
385 struct ip_vs_dest *dest; /* real server */
386 atomic_t in_pkts; /* incoming packet counter */
387
388 /* packet transmitter for different forwarding methods. If it
389 mangles the packet, it must return NF_DROP or better NF_STOLEN,
390 otherwise this must be changed to a sk_buff **.
391 */
392 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
393 struct ip_vs_protocol *pp);
394
395 /* Note: we can group the following members into a structure,
396 in order to save more space, and the following members are
397 only used in VS/NAT anyway */
398 struct ip_vs_app *app; /* bound ip_vs_app object */
399 void *app_data; /* Application private data */
400 struct ip_vs_seq in_seq; /* incoming seq. struct */
401 struct ip_vs_seq out_seq; /* outgoing seq. struct */
402};
403
404
c860c6b1
JV		 405/*
406 * Extended internal versions of struct ip_vs_service_user and
407 * ip_vs_dest_user for IPv6 support.
408 *
409 * We need these to conveniently pass around service and destination
410 * options, but unfortunately, we also need to keep the old definitions to
411 * maintain userspace backwards compatibility for the setsockopt interface.
412 */
413struct ip_vs_service_user_kern {
414 /* virtual service addresses */
415 u16 af;
416 u16 protocol;
417 union nf_inet_addr addr; /* virtual ip address */
418 u16 port;
419 u32 fwmark; /* firwall mark of service */
420
421 /* virtual service options */
422 char *sched_name;
423 unsigned flags; /* virtual service flags */
424 unsigned timeout; /* persistent timeout in sec */
425 u32 netmask; /* persistent netmask */
426};
427
428
429struct ip_vs_dest_user_kern {
430 /* destination server address */
431 union nf_inet_addr addr;
432 u16 port;
433
434 /* real server options */
435 unsigned conn_flags; /* connection flags */
436 int weight; /* destination weight */
437
438 /* thresholds for active connections */
439 u32 u_threshold; /* upper threshold */
440 u32 l_threshold; /* lower threshold */
441};
442
443
1da177e4
LT		 444/*
445 * The information about the virtual service offered to the net
446 * and the forwarding entries
447 */
448struct ip_vs_service {
449 struct list_head s_list; /* for normal service table */
450 struct list_head f_list; /* for fwmark-based service table */
451 atomic_t refcnt; /* reference counter */
452 atomic_t usecnt; /* use counter */
453
e7ade46a 454 u16 af; /* address family */
1da177e4 455 __u16 protocol; /* which protocol (TCP/UDP) */
e7ade46a 456 union nf_inet_addr addr; /* IP address for virtual service */
014d730d 457 __be16 port; /* port number for the service */
1da177e4
LT		 458 __u32 fwmark; /* firewall mark of the service */
459 unsigned flags; /* service status flags */
460 unsigned timeout; /* persistent timeout in ticks */
014d730d 461 __be32 netmask; /* grouping granularity */
1da177e4
LT		 462
463 struct list_head destinations; /* real server d-linked list */
464 __u32 num_dests; /* number of servers */
465 struct ip_vs_stats stats; /* statistics for the service */
466 struct ip_vs_app *inc; /* bind conns to this app inc */
467
468 /* for scheduling */
469 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
470 rwlock_t sched_lock; /* lock sched_data */
471 void *sched_data; /* scheduler application data */
472};
473
474
475/*
476 * The real server destination forwarding entry
477 * with ip address, port number, and so on.
478 */
479struct ip_vs_dest {
480 struct list_head n_list; /* for the dests in the service */
481 struct list_head d_list; /* for table with all the dests */
482
e7ade46a
JV		 483 u16 af; /* address family */
484 union nf_inet_addr addr; /* IP address of the server */
014d730d 485 __be16 port; /* port number of the server */
1da177e4
LT		 486 volatile unsigned flags; /* dest status flags */
487 atomic_t conn_flags; /* flags to copy to conn */
488 atomic_t weight; /* server weight */
489
490 atomic_t refcnt; /* reference counter */
491 struct ip_vs_stats stats; /* statistics */
492
493 /* connection counters and thresholds */
494 atomic_t activeconns; /* active connections */
495 atomic_t inactconns; /* inactive connections */
496 atomic_t persistconns; /* persistent connections */
497 __u32 u_threshold; /* upper threshold */
498 __u32 l_threshold; /* lower threshold */
499
500 /* for destination cache */
501 spinlock_t dst_lock; /* lock of dst_cache */
502 struct dst_entry *dst_cache; /* destination cache entry */
503 u32 dst_rtos; /* RT_TOS(tos) for dst */
504
505 /* for virtual service */
506 struct ip_vs_service *svc; /* service it belongs to */
507 __u16 protocol; /* which protocol (TCP/UDP) */
e7ade46a 508 union nf_inet_addr vaddr; /* virtual IP address */
014d730d 509 __be16 vport; /* virtual port number */
1da177e4
LT		 510 __u32 vfwmark; /* firewall mark of service */
511};
512
513
514/*
515 * The scheduler object
516 */
517struct ip_vs_scheduler {
518 struct list_head n_list; /* d-linked list head */
519 char *name; /* scheduler name */
520 atomic_t refcnt; /* reference counter */
521 struct module *module; /* THIS_MODULE/NULL */
b14198f6
JV		 522#ifdef CONFIG_IP_VS_IPV6
523 int supports_ipv6; /* scheduler has IPv6 support */
524#endif
1da177e4
LT		 525
526 /* scheduler initializing service */
527 int (*init_service)(struct ip_vs_service *svc);
528 /* scheduling service finish */
529 int (*done_service)(struct ip_vs_service *svc);
530 /* scheduler updating service */
531 int (*update_service)(struct ip_vs_service *svc);
532
533 /* selecting a server from the given service */
534 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
535 const struct sk_buff *skb);
536};
537
538
539/*
540 * The application module object (a.k.a. app incarnation)
541 */
542struct ip_vs_app
543{
544 struct list_head a_list; /* member in app list */
545 int type; /* IP_VS_APP_TYPE_xxx */
546 char *name; /* application module name */
547 __u16 protocol;
548 struct module *module; /* THIS_MODULE/NULL */
549 struct list_head incs_list; /* list of incarnations */
550
551 /* members for application incarnations */
552 struct list_head p_list; /* member in proto app list */
553 struct ip_vs_app *app; /* its real application */
014d730d 554 __be16 port; /* port number in net order */
1da177e4
LT		 555 atomic_t usecnt; /* usage counter */
556
557 /* output hook: return false if can't linearize. diff set for TCP. */
558 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
3db05fea 559 struct sk_buff *, int *diff);
1da177e4
LT		 560
561 /* input hook: return false if can't linearize. diff set for TCP. */
562 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
3db05fea 563 struct sk_buff *, int *diff);
1da177e4
LT		 564
565 /* ip_vs_app initializer */
566 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
567
568 /* ip_vs_app finish */
569 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
570
571
572 /* not used now */
573 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
574 struct ip_vs_protocol *);
575
576 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
577
578 int * timeout_table;
579 int * timeouts;
580 int timeouts_size;
581
582 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
583 int *verdict, struct ip_vs_conn **cpp);
584
585 struct ip_vs_conn *
586 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
587 const struct iphdr *iph, unsigned int proto_off,
588 int inverse);
589
590 struct ip_vs_conn *
591 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
592 const struct iphdr *iph, unsigned int proto_off,
593 int inverse);
594
595 int (*state_transition)(struct ip_vs_conn *cp, int direction,
596 const struct sk_buff *skb,
597 struct ip_vs_app *app);
598
599 void (*timeout_change)(struct ip_vs_app *app, int flags);
600};
601
602
603/*
604 * IPVS core functions
605 * (from ip_vs_core.c)
606 */
607extern const char *ip_vs_proto_name(unsigned proto);
608extern void ip_vs_init_hash_table(struct list_head *table, int rows);
afdd6140 609#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1da177e4 610
1da177e4
LT		 611#define IP_VS_APP_TYPE_FTP 1
612
613/*
614 * ip_vs_conn handling functions
615 * (from ip_vs_conn.c)
616 */
617
618/*
619 * IPVS connection entry hash table
620 */
621#ifndef CONFIG_IP_VS_TAB_BITS
622#define CONFIG_IP_VS_TAB_BITS 12
623#endif
624/* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
625#if CONFIG_IP_VS_TAB_BITS < 8
626#define IP_VS_CONN_TAB_BITS 8
627#endif
628#if CONFIG_IP_VS_TAB_BITS > 20
629#define IP_VS_CONN_TAB_BITS 20
630#endif
631#if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
632#define IP_VS_CONN_TAB_BITS CONFIG_IP_VS_TAB_BITS
633#endif
634#define IP_VS_CONN_TAB_SIZE (1 << IP_VS_CONN_TAB_BITS)
635#define IP_VS_CONN_TAB_MASK (IP_VS_CONN_TAB_SIZE - 1)
636
637enum {
638 IP_VS_DIR_INPUT = 0,
639 IP_VS_DIR_OUTPUT,
640 IP_VS_DIR_INPUT_ONLY,
641 IP_VS_DIR_LAST,
642};
643
644extern struct ip_vs_conn *ip_vs_conn_in_get
28364a59
JV		 645(int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
646 const union nf_inet_addr *d_addr, __be16 d_port);
647
87375ab4 648extern struct ip_vs_conn *ip_vs_ct_in_get
28364a59
JV		 649(int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
650 const union nf_inet_addr *d_addr, __be16 d_port);
651
1da177e4 652extern struct ip_vs_conn *ip_vs_conn_out_get
28364a59
JV		 653(int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
654 const union nf_inet_addr *d_addr, __be16 d_port);
1da177e4
LT		 655
656/* put back the conn without restarting its timer */
657static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
658{
659 atomic_dec(&cp->refcnt);
660}
661extern void ip_vs_conn_put(struct ip_vs_conn *cp);
014d730d 662extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1da177e4
LT		 663
664extern struct ip_vs_conn *
28364a59
JV		 665ip_vs_conn_new(int af, int proto, const union nf_inet_addr *caddr, __be16 cport,
666 const union nf_inet_addr *vaddr, __be16 vport,
667 const union nf_inet_addr *daddr, __be16 dport, unsigned flags,
1da177e4
LT		 668 struct ip_vs_dest *dest);
669extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
670
671extern const char * ip_vs_state_name(__u16 proto, int state);
672
673extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
674extern int ip_vs_check_template(struct ip_vs_conn *ct);
1da177e4
LT		 675extern void ip_vs_random_dropentry(void);
676extern int ip_vs_conn_init(void);
677extern void ip_vs_conn_cleanup(void);
678
679static inline void ip_vs_control_del(struct ip_vs_conn *cp)
680{
681 struct ip_vs_conn *ctl_cp = cp->control;
682 if (!ctl_cp) {
683 IP_VS_ERR("request control DEL for uncontrolled: "
684 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
685 NIPQUAD(cp->caddr),ntohs(cp->cport),
686 NIPQUAD(cp->vaddr),ntohs(cp->vport));
687 return;
688 }
689
690 IP_VS_DBG(7, "DELeting control for: "
691 "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
692 NIPQUAD(cp->caddr),ntohs(cp->cport),
693 NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
694
695 cp->control = NULL;
696 if (atomic_read(&ctl_cp->n_control) == 0) {
697 IP_VS_ERR("BUG control DEL with n=0 : "
698 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
699 NIPQUAD(cp->caddr),ntohs(cp->cport),
700 NIPQUAD(cp->vaddr),ntohs(cp->vport));
701 return;
702 }
703 atomic_dec(&ctl_cp->n_control);
704}
705
706static inline void
707ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
708{
709 if (cp->control) {
710 IP_VS_ERR("request control ADD for already controlled: "
711 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
712 NIPQUAD(cp->caddr),ntohs(cp->cport),
713 NIPQUAD(cp->vaddr),ntohs(cp->vport));
714 ip_vs_control_del(cp);
715 }
716
717 IP_VS_DBG(7, "ADDing control for: "
718 "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
719 NIPQUAD(cp->caddr),ntohs(cp->cport),
720 NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
721
722 cp->control = ctl_cp;
723 atomic_inc(&ctl_cp->n_control);
724}
725
726
727/*
728 * IPVS application functions
729 * (from ip_vs_app.c)
730 */
731#define IP_VS_APP_MAX_PORTS 8
732extern int register_ip_vs_app(struct ip_vs_app *app);
733extern void unregister_ip_vs_app(struct ip_vs_app *app);
734extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
735extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
736extern int
737register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
738extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
739extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
740
3db05fea
HX		 741extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
742extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
dd0fc66f 743extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
1da177e4
LT		 744 char *o_buf, int o_len, char *n_buf, int n_len);
745extern int ip_vs_app_init(void);
746extern void ip_vs_app_cleanup(void);
747
748
749/*
750 * IPVS protocol functions (from ip_vs_proto.c)
751 */
752extern int ip_vs_protocol_init(void);
753extern void ip_vs_protocol_cleanup(void);
754extern void ip_vs_protocol_timeout_change(int flags);
755extern int *ip_vs_create_timeout_table(int *table, int size);
756extern int
757ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to);
758extern void
759ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
760 int offset, const char *msg);
761
762extern struct ip_vs_protocol ip_vs_protocol_tcp;
763extern struct ip_vs_protocol ip_vs_protocol_udp;
764extern struct ip_vs_protocol ip_vs_protocol_icmp;
765extern struct ip_vs_protocol ip_vs_protocol_esp;
766extern struct ip_vs_protocol ip_vs_protocol_ah;
767
768
769/*
770 * Registering/unregistering scheduler functions
771 * (from ip_vs_sched.c)
772 */
773extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
774extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
775extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
776 struct ip_vs_scheduler *scheduler);
777extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
778extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
779extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
780extern struct ip_vs_conn *
781ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
782extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
783 struct ip_vs_protocol *pp);
784
785
786/*
787 * IPVS control data and functions (from ip_vs_ctl.c)
788 */
789extern int sysctl_ip_vs_cache_bypass;
790extern int sysctl_ip_vs_expire_nodest_conn;
791extern int sysctl_ip_vs_expire_quiescent_template;
792extern int sysctl_ip_vs_sync_threshold[2];
793extern int sysctl_ip_vs_nat_icmp_send;
794extern struct ip_vs_stats ip_vs_stats;
5587da55 795extern const struct ctl_path net_vs_ctl_path[];
1da177e4
LT		 796
797extern struct ip_vs_service *
3c2e0505
JV		 798ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
799 const union nf_inet_addr *vaddr, __be16 vport);
1da177e4
LT		 800
801static inline void ip_vs_service_put(struct ip_vs_service *svc)
802{
803 atomic_dec(&svc->usecnt);
804}
805
806extern struct ip_vs_dest *
014d730d 807ip_vs_lookup_real_service(__u16 protocol, __be32 daddr, __be16 dport);
1da177e4
LT		 808extern int ip_vs_use_count_inc(void);
809extern void ip_vs_use_count_dec(void);
810extern int ip_vs_control_init(void);
811extern void ip_vs_control_cleanup(void);
1e356f9c
RB		 812extern struct ip_vs_dest *
813ip_vs_find_dest(__be32 daddr, __be16 dport,
814 __be32 vaddr, __be16 vport, __u16 protocol);
815extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1da177e4
LT		 816
817
818/*
819 * IPVS sync daemon data and function prototypes
820 * (from ip_vs_sync.c)
821 */
822extern volatile int ip_vs_sync_state;
823extern volatile int ip_vs_master_syncid;
824extern volatile int ip_vs_backup_syncid;
825extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
826extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
827extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
828extern int stop_sync_thread(int state);
829extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
830
831
832/*
833 * IPVS rate estimator prototypes (from ip_vs_est.c)
834 */
a919cf4b
SW		 835extern int ip_vs_estimator_init(void);
836extern void ip_vs_estimator_cleanup(void);
3a14a313 837extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
1da177e4
LT		 838extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
839extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
840
841/*
842 * Various IPVS packet transmitters (from ip_vs_xmit.c)
843 */
844extern int ip_vs_null_xmit
845(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
846extern int ip_vs_bypass_xmit
847(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
848extern int ip_vs_nat_xmit
849(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
850extern int ip_vs_tunnel_xmit
851(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
852extern int ip_vs_dr_xmit
853(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
854extern int ip_vs_icmp_xmit
855(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
856extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
857
b3cdd2a7
JV		 858#ifdef CONFIG_IP_VS_IPV6
859extern int ip_vs_bypass_xmit_v6
860(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
861extern int ip_vs_nat_xmit_v6
862(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
863extern int ip_vs_tunnel_xmit_v6
864(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
865extern int ip_vs_dr_xmit_v6
866(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
867extern int ip_vs_icmp_xmit_v6
868(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
869 int offset);
870#endif
1da177e4
LT		 871
872/*
873 * This is a simple mechanism to ignore packets when
874 * we are loaded. Just set ip_vs_drop_rate to 'n' and
875 * we start to drop 1/rate of the packets
876 */
877extern int ip_vs_drop_rate;
878extern int ip_vs_drop_counter;
879
880static __inline__ int ip_vs_todrop(void)
881{
882 if (!ip_vs_drop_rate) return 0;
883 if (--ip_vs_drop_counter > 0) return 0;
884 ip_vs_drop_counter = ip_vs_drop_rate;
885 return 1;
886}
887
888/*
889 * ip_vs_fwd_tag returns the forwarding tag of the connection
890 */
891#define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
892
732db659 893static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1da177e4
LT		 894{
895 char fwd;
896
897 switch (IP_VS_FWD_METHOD(cp)) {
898 case IP_VS_CONN_F_MASQ:
899 fwd = 'M'; break;
900 case IP_VS_CONN_F_LOCALNODE:
901 fwd = 'L'; break;
902 case IP_VS_CONN_F_TUNNEL:
903 fwd = 'T'; break;
904 case IP_VS_CONN_F_DROUTE:
905 fwd = 'R'; break;
906 case IP_VS_CONN_F_BYPASS:
907 fwd = 'B'; break;
908 default:
909 fwd = '?'; break;
910 }
911 return fwd;
912}
913
1da177e4 914extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
b3cdd2a7
JV		 915 struct ip_vs_conn *cp, int dir);
916
917#ifdef CONFIG_IP_VS_IPV6
918extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
919 struct ip_vs_conn *cp, int dir);
920#endif
1da177e4 921
b1550f22 922extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1da177e4 923
f9214b26 924static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1da177e4 925{
f9214b26 926 __be32 diff[2] = { ~old, new };
1da177e4 927
f9214b26
AV		 928 return csum_partial((char *) diff, sizeof(diff), oldsum);
929}
930
0bbdd42b
JV		 931#ifdef CONFIG_IP_VS_IPV6
932static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
933 __wsum oldsum)
934{
935 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
936 new[3], new[2], new[1], new[0] };
937
938 return csum_partial((char *) diff, sizeof(diff), oldsum);
939}
940#endif
941
f9214b26
AV		 942static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
943{
944 __be16 diff[2] = { ~old, new };
945
946 return csum_partial((char *) diff, sizeof(diff), oldsum);
1da177e4
LT		 947}
948
949#endif /* __KERNEL__ */
950
bc4768eb 951#endif /* _NET_IP_VS_H */
Clone of davem's net-next-2.6 tree