xref: /openbmc/linux/include/net/ip_vs.h (revision fd589a8f)
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_RL(msg, ...)						\
154 	do {								\
155 		if (net_ratelimit())					\
156 			pr_err(msg, ##__VA_ARGS__);			\
157 	} while (0)
158 
159 #ifdef CONFIG_IP_VS_DEBUG
160 #define EnterFunction(level)						\
161 	do {								\
162 		if (level <= ip_vs_get_debug_level())			\
163 			printk(KERN_DEBUG				\
164 			       pr_fmt("Enter: %s, %s line %i\n"),	\
165 			       __func__, __FILE__, __LINE__);		\
166 	} while (0)
167 #define LeaveFunction(level)						\
168 	do {								\
169 		if (level <= ip_vs_get_debug_level())			\
170 			printk(KERN_DEBUG				\
171 			       pr_fmt("Leave: %s, %s line %i\n"),	\
172 			       __func__, __FILE__, __LINE__);		\
173 	} while (0)
174 #else
175 #define EnterFunction(level)   do {} while (0)
176 #define LeaveFunction(level)   do {} while (0)
177 #endif
178 
179 #define	IP_VS_WAIT_WHILE(expr)	while (expr) { cpu_relax(); }
180 
181 
182 /*
183  *      The port number of FTP service (in network order).
184  */
185 #define FTPPORT  cpu_to_be16(21)
186 #define FTPDATA  cpu_to_be16(20)
187 
188 /*
189  *      TCP State Values
190  */
191 enum {
192 	IP_VS_TCP_S_NONE = 0,
193 	IP_VS_TCP_S_ESTABLISHED,
194 	IP_VS_TCP_S_SYN_SENT,
195 	IP_VS_TCP_S_SYN_RECV,
196 	IP_VS_TCP_S_FIN_WAIT,
197 	IP_VS_TCP_S_TIME_WAIT,
198 	IP_VS_TCP_S_CLOSE,
199 	IP_VS_TCP_S_CLOSE_WAIT,
200 	IP_VS_TCP_S_LAST_ACK,
201 	IP_VS_TCP_S_LISTEN,
202 	IP_VS_TCP_S_SYNACK,
203 	IP_VS_TCP_S_LAST
204 };
205 
206 /*
207  *	UDP State Values
208  */
209 enum {
210 	IP_VS_UDP_S_NORMAL,
211 	IP_VS_UDP_S_LAST,
212 };
213 
214 /*
215  *	ICMP State Values
216  */
217 enum {
218 	IP_VS_ICMP_S_NORMAL,
219 	IP_VS_ICMP_S_LAST,
220 };
221 
222 /*
223  *	Delta sequence info structure
224  *	Each ip_vs_conn has 2 (output AND input seq. changes).
225  *      Only used in the VS/NAT.
226  */
227 struct ip_vs_seq {
228 	__u32			init_seq;	/* Add delta from this seq */
229 	__u32			delta;		/* Delta in sequence numbers */
230 	__u32			previous_delta;	/* Delta in sequence numbers
231 						   before last resized pkt */
232 };
233 
234 
235 /*
236  *	IPVS statistics objects
237  */
238 struct ip_vs_estimator {
239 	struct list_head	list;
240 
241 	u64			last_inbytes;
242 	u64			last_outbytes;
243 	u32			last_conns;
244 	u32			last_inpkts;
245 	u32			last_outpkts;
246 
247 	u32			cps;
248 	u32			inpps;
249 	u32			outpps;
250 	u32			inbps;
251 	u32			outbps;
252 };
253 
254 struct ip_vs_stats
255 {
256 	struct ip_vs_stats_user	ustats;         /* statistics */
257 	struct ip_vs_estimator	est;		/* estimator */
258 
259 	spinlock_t              lock;           /* spin lock */
260 };
261 
262 struct dst_entry;
263 struct iphdr;
264 struct ip_vs_conn;
265 struct ip_vs_app;
266 struct sk_buff;
267 
268 struct ip_vs_protocol {
269 	struct ip_vs_protocol	*next;
270 	char			*name;
271 	u16			protocol;
272 	u16			num_states;
273 	int			dont_defrag;
274 	atomic_t		appcnt;		/* counter of proto app incs */
275 	int			*timeout_table;	/* protocol timeout table */
276 
277 	void (*init)(struct ip_vs_protocol *pp);
278 
279 	void (*exit)(struct ip_vs_protocol *pp);
280 
281 	int (*conn_schedule)(int af, struct sk_buff *skb,
282 			     struct ip_vs_protocol *pp,
283 			     int *verdict, struct ip_vs_conn **cpp);
284 
285 	struct ip_vs_conn *
286 	(*conn_in_get)(int af,
287 		       const struct sk_buff *skb,
288 		       struct ip_vs_protocol *pp,
289 		       const struct ip_vs_iphdr *iph,
290 		       unsigned int proto_off,
291 		       int inverse);
292 
293 	struct ip_vs_conn *
294 	(*conn_out_get)(int af,
295 			const struct sk_buff *skb,
296 			struct ip_vs_protocol *pp,
297 			const struct ip_vs_iphdr *iph,
298 			unsigned int proto_off,
299 			int inverse);
300 
301 	int (*snat_handler)(struct sk_buff *skb,
302 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
303 
304 	int (*dnat_handler)(struct sk_buff *skb,
305 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
306 
307 	int (*csum_check)(int af, struct sk_buff *skb,
308 			  struct ip_vs_protocol *pp);
309 
310 	const char *(*state_name)(int state);
311 
312 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
313 				const struct sk_buff *skb,
314 				struct ip_vs_protocol *pp);
315 
316 	int (*register_app)(struct ip_vs_app *inc);
317 
318 	void (*unregister_app)(struct ip_vs_app *inc);
319 
320 	int (*app_conn_bind)(struct ip_vs_conn *cp);
321 
322 	void (*debug_packet)(struct ip_vs_protocol *pp,
323 			     const struct sk_buff *skb,
324 			     int offset,
325 			     const char *msg);
326 
327 	void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
328 
329 	int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
330 };
331 
332 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
333 
334 /*
335  *	IP_VS structure allocated for each dynamically scheduled connection
336  */
337 struct ip_vs_conn {
338 	struct list_head        c_list;         /* hashed list heads */
339 
340 	/* Protocol, addresses and port numbers */
341 	u16                      af;		/* address family */
342 	union nf_inet_addr       caddr;          /* client address */
343 	union nf_inet_addr       vaddr;          /* virtual address */
344 	union nf_inet_addr       daddr;          /* destination address */
345 	__be16                   cport;
346 	__be16                   vport;
347 	__be16                   dport;
348 	__u16                   protocol;       /* Which protocol (TCP/UDP) */
349 
350 	/* counter and timer */
351 	atomic_t		refcnt;		/* reference count */
352 	struct timer_list	timer;		/* Expiration timer */
353 	volatile unsigned long	timeout;	/* timeout */
354 
355 	/* Flags and state transition */
356 	spinlock_t              lock;           /* lock for state transition */
357 	volatile __u16          flags;          /* status flags */
358 	volatile __u16          state;          /* state info */
359 	volatile __u16          old_state;      /* old state, to be used for
360 						 * state transition triggerd
361 						 * synchronization
362 						 */
363 
364 	/* Control members */
365 	struct ip_vs_conn       *control;       /* Master control connection */
366 	atomic_t                n_control;      /* Number of controlled ones */
367 	struct ip_vs_dest       *dest;          /* real server */
368 	atomic_t                in_pkts;        /* incoming packet counter */
369 
370 	/* packet transmitter for different forwarding methods.  If it
371 	   mangles the packet, it must return NF_DROP or better NF_STOLEN,
372 	   otherwise this must be changed to a sk_buff **.
373 	 */
374 	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
375 			   struct ip_vs_protocol *pp);
376 
377 	/* Note: we can group the following members into a structure,
378 	   in order to save more space, and the following members are
379 	   only used in VS/NAT anyway */
380 	struct ip_vs_app        *app;           /* bound ip_vs_app object */
381 	void                    *app_data;      /* Application private data */
382 	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
383 	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
384 };
385 
386 
387 /*
388  *	Extended internal versions of struct ip_vs_service_user and
389  *	ip_vs_dest_user for IPv6 support.
390  *
391  *	We need these to conveniently pass around service and destination
392  *	options, but unfortunately, we also need to keep the old definitions to
393  *	maintain userspace backwards compatibility for the setsockopt interface.
394  */
395 struct ip_vs_service_user_kern {
396 	/* virtual service addresses */
397 	u16			af;
398 	u16			protocol;
399 	union nf_inet_addr	addr;		/* virtual ip address */
400 	u16			port;
401 	u32			fwmark;		/* firwall mark of service */
402 
403 	/* virtual service options */
404 	char			*sched_name;
405 	unsigned		flags;		/* virtual service flags */
406 	unsigned		timeout;	/* persistent timeout in sec */
407 	u32			netmask;	/* persistent netmask */
408 };
409 
410 
411 struct ip_vs_dest_user_kern {
412 	/* destination server address */
413 	union nf_inet_addr	addr;
414 	u16			port;
415 
416 	/* real server options */
417 	unsigned		conn_flags;	/* connection flags */
418 	int			weight;		/* destination weight */
419 
420 	/* thresholds for active connections */
421 	u32			u_threshold;	/* upper threshold */
422 	u32			l_threshold;	/* lower threshold */
423 };
424 
425 
426 /*
427  *	The information about the virtual service offered to the net
428  *	and the forwarding entries
429  */
430 struct ip_vs_service {
431 	struct list_head	s_list;   /* for normal service table */
432 	struct list_head	f_list;   /* for fwmark-based service table */
433 	atomic_t		refcnt;   /* reference counter */
434 	atomic_t		usecnt;   /* use counter */
435 
436 	u16			af;       /* address family */
437 	__u16			protocol; /* which protocol (TCP/UDP) */
438 	union nf_inet_addr	addr;	  /* IP address for virtual service */
439 	__be16			port;	  /* port number for the service */
440 	__u32                   fwmark;   /* firewall mark of the service */
441 	unsigned		flags;	  /* service status flags */
442 	unsigned		timeout;  /* persistent timeout in ticks */
443 	__be32			netmask;  /* grouping granularity */
444 
445 	struct list_head	destinations;  /* real server d-linked list */
446 	__u32			num_dests;     /* number of servers */
447 	struct ip_vs_stats      stats;         /* statistics for the service */
448 	struct ip_vs_app	*inc;	  /* bind conns to this app inc */
449 
450 	/* for scheduling */
451 	struct ip_vs_scheduler	*scheduler;    /* bound scheduler object */
452 	rwlock_t		sched_lock;    /* lock sched_data */
453 	void			*sched_data;   /* scheduler application data */
454 };
455 
456 
457 /*
458  *	The real server destination forwarding entry
459  *	with ip address, port number, and so on.
460  */
461 struct ip_vs_dest {
462 	struct list_head	n_list;   /* for the dests in the service */
463 	struct list_head	d_list;   /* for table with all the dests */
464 
465 	u16			af;		/* address family */
466 	union nf_inet_addr	addr;		/* IP address of the server */
467 	__be16			port;		/* port number of the server */
468 	volatile unsigned	flags;		/* dest status flags */
469 	atomic_t		conn_flags;	/* flags to copy to conn */
470 	atomic_t		weight;		/* server weight */
471 
472 	atomic_t		refcnt;		/* reference counter */
473 	struct ip_vs_stats      stats;          /* statistics */
474 
475 	/* connection counters and thresholds */
476 	atomic_t		activeconns;	/* active connections */
477 	atomic_t		inactconns;	/* inactive connections */
478 	atomic_t		persistconns;	/* persistent connections */
479 	__u32			u_threshold;	/* upper threshold */
480 	__u32			l_threshold;	/* lower threshold */
481 
482 	/* for destination cache */
483 	spinlock_t		dst_lock;	/* lock of dst_cache */
484 	struct dst_entry	*dst_cache;	/* destination cache entry */
485 	u32			dst_rtos;	/* RT_TOS(tos) for dst */
486 
487 	/* for virtual service */
488 	struct ip_vs_service	*svc;		/* service it belongs to */
489 	__u16			protocol;	/* which protocol (TCP/UDP) */
490 	union nf_inet_addr	vaddr;		/* virtual IP address */
491 	__be16			vport;		/* virtual port number */
492 	__u32			vfwmark;	/* firewall mark of service */
493 };
494 
495 
496 /*
497  *	The scheduler object
498  */
499 struct ip_vs_scheduler {
500 	struct list_head	n_list;		/* d-linked list head */
501 	char			*name;		/* scheduler name */
502 	atomic_t		refcnt;		/* reference counter */
503 	struct module		*module;	/* THIS_MODULE/NULL */
504 
505 	/* scheduler initializing service */
506 	int (*init_service)(struct ip_vs_service *svc);
507 	/* scheduling service finish */
508 	int (*done_service)(struct ip_vs_service *svc);
509 	/* scheduler updating service */
510 	int (*update_service)(struct ip_vs_service *svc);
511 
512 	/* selecting a server from the given service */
513 	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
514 				       const struct sk_buff *skb);
515 };
516 
517 
518 /*
519  *	The application module object (a.k.a. app incarnation)
520  */
521 struct ip_vs_app
522 {
523 	struct list_head	a_list;		/* member in app list */
524 	int			type;		/* IP_VS_APP_TYPE_xxx */
525 	char			*name;		/* application module name */
526 	__u16			protocol;
527 	struct module		*module;	/* THIS_MODULE/NULL */
528 	struct list_head	incs_list;	/* list of incarnations */
529 
530 	/* members for application incarnations */
531 	struct list_head	p_list;		/* member in proto app list */
532 	struct ip_vs_app	*app;		/* its real application */
533 	__be16			port;		/* port number in net order */
534 	atomic_t		usecnt;		/* usage counter */
535 
536 	/* output hook: return false if can't linearize. diff set for TCP.  */
537 	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
538 		       struct sk_buff *, int *diff);
539 
540 	/* input hook: return false if can't linearize. diff set for TCP. */
541 	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
542 		      struct sk_buff *, int *diff);
543 
544 	/* ip_vs_app initializer */
545 	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
546 
547 	/* ip_vs_app finish */
548 	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
549 
550 
551 	/* not used now */
552 	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
553 			 struct ip_vs_protocol *);
554 
555 	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
556 
557 	int *			timeout_table;
558 	int *			timeouts;
559 	int			timeouts_size;
560 
561 	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
562 			     int *verdict, struct ip_vs_conn **cpp);
563 
564 	struct ip_vs_conn *
565 	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
566 		       const struct iphdr *iph, unsigned int proto_off,
567 		       int inverse);
568 
569 	struct ip_vs_conn *
570 	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
571 			const struct iphdr *iph, unsigned int proto_off,
572 			int inverse);
573 
574 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
575 				const struct sk_buff *skb,
576 				struct ip_vs_app *app);
577 
578 	void (*timeout_change)(struct ip_vs_app *app, int flags);
579 };
580 
581 
582 /*
583  *      IPVS core functions
584  *      (from ip_vs_core.c)
585  */
586 extern const char *ip_vs_proto_name(unsigned proto);
587 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
588 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
589 
590 #define IP_VS_APP_TYPE_FTP	1
591 
592 /*
593  *     ip_vs_conn handling functions
594  *     (from ip_vs_conn.c)
595  */
596 
597 /*
598  *     IPVS connection entry hash table
599  */
600 #ifndef CONFIG_IP_VS_TAB_BITS
601 #define CONFIG_IP_VS_TAB_BITS   12
602 #endif
603 
604 #define IP_VS_CONN_TAB_BITS	CONFIG_IP_VS_TAB_BITS
605 #define IP_VS_CONN_TAB_SIZE     (1 << IP_VS_CONN_TAB_BITS)
606 #define IP_VS_CONN_TAB_MASK     (IP_VS_CONN_TAB_SIZE - 1)
607 
608 enum {
609 	IP_VS_DIR_INPUT = 0,
610 	IP_VS_DIR_OUTPUT,
611 	IP_VS_DIR_INPUT_ONLY,
612 	IP_VS_DIR_LAST,
613 };
614 
615 extern struct ip_vs_conn *ip_vs_conn_in_get
616 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
617  const union nf_inet_addr *d_addr, __be16 d_port);
618 
619 extern struct ip_vs_conn *ip_vs_ct_in_get
620 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
621  const union nf_inet_addr *d_addr, __be16 d_port);
622 
623 extern struct ip_vs_conn *ip_vs_conn_out_get
624 (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port,
625  const union nf_inet_addr *d_addr, __be16 d_port);
626 
627 /* put back the conn without restarting its timer */
628 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
629 {
630 	atomic_dec(&cp->refcnt);
631 }
632 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
633 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
634 
635 extern struct ip_vs_conn *
636 ip_vs_conn_new(int af, int proto, const union nf_inet_addr *caddr, __be16 cport,
637 	       const union nf_inet_addr *vaddr, __be16 vport,
638 	       const union nf_inet_addr *daddr, __be16 dport, unsigned flags,
639 	       struct ip_vs_dest *dest);
640 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
641 
642 extern const char * ip_vs_state_name(__u16 proto, int state);
643 
644 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
645 extern int ip_vs_check_template(struct ip_vs_conn *ct);
646 extern void ip_vs_random_dropentry(void);
647 extern int ip_vs_conn_init(void);
648 extern void ip_vs_conn_cleanup(void);
649 
650 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
651 {
652 	struct ip_vs_conn *ctl_cp = cp->control;
653 	if (!ctl_cp) {
654 		IP_VS_ERR_BUF("request control DEL for uncontrolled: "
655 			      "%s:%d to %s:%d\n",
656 			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
657 			      ntohs(cp->cport),
658 			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
659 			      ntohs(cp->vport));
660 
661 		return;
662 	}
663 
664 	IP_VS_DBG_BUF(7, "DELeting control for: "
665 		      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
666 		      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
667 		      ntohs(cp->cport),
668 		      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
669 		      ntohs(ctl_cp->cport));
670 
671 	cp->control = NULL;
672 	if (atomic_read(&ctl_cp->n_control) == 0) {
673 		IP_VS_ERR_BUF("BUG control DEL with n=0 : "
674 			      "%s:%d to %s:%d\n",
675 			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
676 			      ntohs(cp->cport),
677 			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
678 			      ntohs(cp->vport));
679 
680 		return;
681 	}
682 	atomic_dec(&ctl_cp->n_control);
683 }
684 
685 static inline void
686 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
687 {
688 	if (cp->control) {
689 		IP_VS_ERR_BUF("request control ADD for already controlled: "
690 			      "%s:%d to %s:%d\n",
691 			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
692 			      ntohs(cp->cport),
693 			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
694 			      ntohs(cp->vport));
695 
696 		ip_vs_control_del(cp);
697 	}
698 
699 	IP_VS_DBG_BUF(7, "ADDing control for: "
700 		      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
701 		      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
702 		      ntohs(cp->cport),
703 		      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
704 		      ntohs(ctl_cp->cport));
705 
706 	cp->control = ctl_cp;
707 	atomic_inc(&ctl_cp->n_control);
708 }
709 
710 
711 /*
712  *      IPVS application functions
713  *      (from ip_vs_app.c)
714  */
715 #define IP_VS_APP_MAX_PORTS  8
716 extern int register_ip_vs_app(struct ip_vs_app *app);
717 extern void unregister_ip_vs_app(struct ip_vs_app *app);
718 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
719 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
720 extern int
721 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
722 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
723 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
724 
725 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
726 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
727 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
728 			     char *o_buf, int o_len, char *n_buf, int n_len);
729 extern int ip_vs_app_init(void);
730 extern void ip_vs_app_cleanup(void);
731 
732 
733 /*
734  *	IPVS protocol functions (from ip_vs_proto.c)
735  */
736 extern int ip_vs_protocol_init(void);
737 extern void ip_vs_protocol_cleanup(void);
738 extern void ip_vs_protocol_timeout_change(int flags);
739 extern int *ip_vs_create_timeout_table(int *table, int size);
740 extern int
741 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
742 			const char *name, int to);
743 extern void
744 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
745 			  int offset, const char *msg);
746 
747 extern struct ip_vs_protocol ip_vs_protocol_tcp;
748 extern struct ip_vs_protocol ip_vs_protocol_udp;
749 extern struct ip_vs_protocol ip_vs_protocol_icmp;
750 extern struct ip_vs_protocol ip_vs_protocol_esp;
751 extern struct ip_vs_protocol ip_vs_protocol_ah;
752 
753 
754 /*
755  *      Registering/unregistering scheduler functions
756  *      (from ip_vs_sched.c)
757  */
758 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
759 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
760 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
761 				struct ip_vs_scheduler *scheduler);
762 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
763 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
764 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
765 extern struct ip_vs_conn *
766 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
767 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
768 			struct ip_vs_protocol *pp);
769 
770 
771 /*
772  *      IPVS control data and functions (from ip_vs_ctl.c)
773  */
774 extern int sysctl_ip_vs_cache_bypass;
775 extern int sysctl_ip_vs_expire_nodest_conn;
776 extern int sysctl_ip_vs_expire_quiescent_template;
777 extern int sysctl_ip_vs_sync_threshold[2];
778 extern int sysctl_ip_vs_nat_icmp_send;
779 extern struct ip_vs_stats ip_vs_stats;
780 extern const struct ctl_path net_vs_ctl_path[];
781 
782 extern struct ip_vs_service *
783 ip_vs_service_get(int af, __u32 fwmark, __u16 protocol,
784 		  const union nf_inet_addr *vaddr, __be16 vport);
785 
786 static inline void ip_vs_service_put(struct ip_vs_service *svc)
787 {
788 	atomic_dec(&svc->usecnt);
789 }
790 
791 extern struct ip_vs_dest *
792 ip_vs_lookup_real_service(int af, __u16 protocol,
793 			  const union nf_inet_addr *daddr, __be16 dport);
794 
795 extern int ip_vs_use_count_inc(void);
796 extern void ip_vs_use_count_dec(void);
797 extern int ip_vs_control_init(void);
798 extern void ip_vs_control_cleanup(void);
799 extern struct ip_vs_dest *
800 ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport,
801 		const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol);
802 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
803 
804 
805 /*
806  *      IPVS sync daemon data and function prototypes
807  *      (from ip_vs_sync.c)
808  */
809 extern volatile int ip_vs_sync_state;
810 extern volatile int ip_vs_master_syncid;
811 extern volatile int ip_vs_backup_syncid;
812 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
813 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
814 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
815 extern int stop_sync_thread(int state);
816 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
817 
818 
819 /*
820  *      IPVS rate estimator prototypes (from ip_vs_est.c)
821  */
822 extern int ip_vs_estimator_init(void);
823 extern void ip_vs_estimator_cleanup(void);
824 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
825 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
826 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
827 
828 /*
829  *	Various IPVS packet transmitters (from ip_vs_xmit.c)
830  */
831 extern int ip_vs_null_xmit
832 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
833 extern int ip_vs_bypass_xmit
834 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
835 extern int ip_vs_nat_xmit
836 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
837 extern int ip_vs_tunnel_xmit
838 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
839 extern int ip_vs_dr_xmit
840 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
841 extern int ip_vs_icmp_xmit
842 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
843 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
844 
845 #ifdef CONFIG_IP_VS_IPV6
846 extern int ip_vs_bypass_xmit_v6
847 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
848 extern int ip_vs_nat_xmit_v6
849 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
850 extern int ip_vs_tunnel_xmit_v6
851 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
852 extern int ip_vs_dr_xmit_v6
853 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
854 extern int ip_vs_icmp_xmit_v6
855 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
856  int offset);
857 #endif
858 
859 /*
860  *	This is a simple mechanism to ignore packets when
861  *	we are loaded. Just set ip_vs_drop_rate to 'n' and
862  *	we start to drop 1/rate of the packets
863  */
864 extern int ip_vs_drop_rate;
865 extern int ip_vs_drop_counter;
866 
867 static __inline__ int ip_vs_todrop(void)
868 {
869 	if (!ip_vs_drop_rate) return 0;
870 	if (--ip_vs_drop_counter > 0) return 0;
871 	ip_vs_drop_counter = ip_vs_drop_rate;
872 	return 1;
873 }
874 
875 /*
876  *      ip_vs_fwd_tag returns the forwarding tag of the connection
877  */
878 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
879 
880 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
881 {
882 	char fwd;
883 
884 	switch (IP_VS_FWD_METHOD(cp)) {
885 	case IP_VS_CONN_F_MASQ:
886 		fwd = 'M'; break;
887 	case IP_VS_CONN_F_LOCALNODE:
888 		fwd = 'L'; break;
889 	case IP_VS_CONN_F_TUNNEL:
890 		fwd = 'T'; break;
891 	case IP_VS_CONN_F_DROUTE:
892 		fwd = 'R'; break;
893 	case IP_VS_CONN_F_BYPASS:
894 		fwd = 'B'; break;
895 	default:
896 		fwd = '?'; break;
897 	}
898 	return fwd;
899 }
900 
901 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
902 			   struct ip_vs_conn *cp, int dir);
903 
904 #ifdef CONFIG_IP_VS_IPV6
905 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
906 			      struct ip_vs_conn *cp, int dir);
907 #endif
908 
909 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
910 
911 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
912 {
913 	__be32 diff[2] = { ~old, new };
914 
915 	return csum_partial(diff, sizeof(diff), oldsum);
916 }
917 
918 #ifdef CONFIG_IP_VS_IPV6
919 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
920 					__wsum oldsum)
921 {
922 	__be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
923 			    new[3],  new[2],  new[1],  new[0] };
924 
925 	return csum_partial(diff, sizeof(diff), oldsum);
926 }
927 #endif
928 
929 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
930 {
931 	__be16 diff[2] = { ~old, new };
932 
933 	return csum_partial(diff, sizeof(diff), oldsum);
934 }
935 
936 #endif /* __KERNEL__ */
937 
938 #endif	/* _NET_IP_VS_H */
939