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