xref: /openbmc/linux/include/net/ip_vs.h (revision 64aae3cb9fd22f33e491c4730d363eb2229ef910)
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/ipv6.h>			/* for struct ipv6hdr */
26 #include <net/ipv6.h>			/* for ipv6_addr_copy */
27 
28 struct ip_vs_iphdr {
29 	int len;
30 	__u8 protocol;
31 	union nf_inet_addr saddr;
32 	union nf_inet_addr daddr;
33 };
34 
35 static inline void
36 ip_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 
56 static 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 
67 static 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 
77 #ifdef CONFIG_IP_VS_DEBUG
78 #include <linux/net.h>
79 
80 extern int ip_vs_get_debug_level(void);
81 #define IP_VS_DBG(level, msg...)			\
82     do {						\
83 	    if (level <= ip_vs_get_debug_level())	\
84 		    printk(KERN_DEBUG "IPVS: " msg);	\
85     } while (0)
86 #define IP_VS_DBG_RL(msg...)				\
87     do {						\
88 	    if (net_ratelimit())			\
89 		    printk(KERN_DEBUG "IPVS: " msg);	\
90     } while (0)
91 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		\
92     do {						\
93 	    if (level <= ip_vs_get_debug_level())	\
94 		pp->debug_packet(pp, skb, ofs, msg);	\
95     } while (0)
96 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	\
97     do {						\
98 	    if (level <= ip_vs_get_debug_level() &&	\
99 		net_ratelimit())			\
100 		pp->debug_packet(pp, skb, ofs, msg);	\
101     } while (0)
102 #else	/* NO DEBUGGING at ALL */
103 #define IP_VS_DBG(level, msg...)  do {} while (0)
104 #define IP_VS_DBG_RL(msg...)  do {} while (0)
105 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		do {} while (0)
106 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	do {} while (0)
107 #endif
108 
109 #define IP_VS_BUG() BUG()
110 #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg)
111 #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg)
112 #define IP_VS_WARNING(msg...) \
113 	printk(KERN_WARNING "IPVS: " msg)
114 #define IP_VS_ERR_RL(msg...)				\
115     do {						\
116 	    if (net_ratelimit())			\
117 		    printk(KERN_ERR "IPVS: " msg);	\
118     } while (0)
119 
120 #ifdef CONFIG_IP_VS_DEBUG
121 #define EnterFunction(level)						\
122     do {								\
123 	    if (level <= ip_vs_get_debug_level())			\
124 		    printk(KERN_DEBUG "Enter: %s, %s line %i\n",	\
125 			   __FUNCTION__, __FILE__, __LINE__);		\
126     } while (0)
127 #define LeaveFunction(level)                                            \
128     do {                                                                \
129 	    if (level <= ip_vs_get_debug_level())                       \
130 			printk(KERN_DEBUG "Leave: %s, %s line %i\n",    \
131 			       __FUNCTION__, __FILE__, __LINE__);       \
132     } while (0)
133 #else
134 #define EnterFunction(level)   do {} while (0)
135 #define LeaveFunction(level)   do {} while (0)
136 #endif
137 
138 #define	IP_VS_WAIT_WHILE(expr)	while (expr) { cpu_relax(); }
139 
140 
141 /*
142  *      The port number of FTP service (in network order).
143  */
144 #define FTPPORT  __constant_htons(21)
145 #define FTPDATA  __constant_htons(20)
146 
147 /*
148  *      TCP State Values
149  */
150 enum {
151 	IP_VS_TCP_S_NONE = 0,
152 	IP_VS_TCP_S_ESTABLISHED,
153 	IP_VS_TCP_S_SYN_SENT,
154 	IP_VS_TCP_S_SYN_RECV,
155 	IP_VS_TCP_S_FIN_WAIT,
156 	IP_VS_TCP_S_TIME_WAIT,
157 	IP_VS_TCP_S_CLOSE,
158 	IP_VS_TCP_S_CLOSE_WAIT,
159 	IP_VS_TCP_S_LAST_ACK,
160 	IP_VS_TCP_S_LISTEN,
161 	IP_VS_TCP_S_SYNACK,
162 	IP_VS_TCP_S_LAST
163 };
164 
165 /*
166  *	UDP State Values
167  */
168 enum {
169 	IP_VS_UDP_S_NORMAL,
170 	IP_VS_UDP_S_LAST,
171 };
172 
173 /*
174  *	ICMP State Values
175  */
176 enum {
177 	IP_VS_ICMP_S_NORMAL,
178 	IP_VS_ICMP_S_LAST,
179 };
180 
181 /*
182  *	Delta sequence info structure
183  *	Each ip_vs_conn has 2 (output AND input seq. changes).
184  *      Only used in the VS/NAT.
185  */
186 struct ip_vs_seq {
187 	__u32			init_seq;	/* Add delta from this seq */
188 	__u32			delta;		/* Delta in sequence numbers */
189 	__u32			previous_delta;	/* Delta in sequence numbers
190 						   before last resized pkt */
191 };
192 
193 
194 /*
195  *	IPVS statistics objects
196  */
197 struct ip_vs_estimator {
198 	struct list_head	list;
199 
200 	u64			last_inbytes;
201 	u64			last_outbytes;
202 	u32			last_conns;
203 	u32			last_inpkts;
204 	u32			last_outpkts;
205 
206 	u32			cps;
207 	u32			inpps;
208 	u32			outpps;
209 	u32			inbps;
210 	u32			outbps;
211 };
212 
213 struct ip_vs_stats
214 {
215 	__u32                   conns;          /* connections scheduled */
216 	__u32                   inpkts;         /* incoming packets */
217 	__u32                   outpkts;        /* outgoing packets */
218 	__u64                   inbytes;        /* incoming bytes */
219 	__u64                   outbytes;       /* outgoing bytes */
220 
221 	__u32			cps;		/* current connection rate */
222 	__u32			inpps;		/* current in packet rate */
223 	__u32			outpps;		/* current out packet rate */
224 	__u32			inbps;		/* current in byte rate */
225 	__u32			outbps;		/* current out byte rate */
226 
227 	/*
228 	 * Don't add anything before the lock, because we use memcpy() to copy
229 	 * the members before the lock to struct ip_vs_stats_user in
230 	 * ip_vs_ctl.c.
231 	 */
232 
233 	spinlock_t              lock;           /* spin lock */
234 
235 	struct ip_vs_estimator	est;		/* estimator */
236 };
237 
238 struct dst_entry;
239 struct iphdr;
240 struct ip_vs_conn;
241 struct ip_vs_app;
242 struct sk_buff;
243 
244 struct ip_vs_protocol {
245 	struct ip_vs_protocol	*next;
246 	char			*name;
247 	u16			protocol;
248 	u16			num_states;
249 	int			dont_defrag;
250 	atomic_t		appcnt;		/* counter of proto app incs */
251 	int			*timeout_table;	/* protocol timeout table */
252 
253 	void (*init)(struct ip_vs_protocol *pp);
254 
255 	void (*exit)(struct ip_vs_protocol *pp);
256 
257 	int (*conn_schedule)(struct sk_buff *skb,
258 			     struct ip_vs_protocol *pp,
259 			     int *verdict, struct ip_vs_conn **cpp);
260 
261 	struct ip_vs_conn *
262 	(*conn_in_get)(const struct sk_buff *skb,
263 		       struct ip_vs_protocol *pp,
264 		       const struct iphdr *iph,
265 		       unsigned int proto_off,
266 		       int inverse);
267 
268 	struct ip_vs_conn *
269 	(*conn_out_get)(const struct sk_buff *skb,
270 			struct ip_vs_protocol *pp,
271 			const struct iphdr *iph,
272 			unsigned int proto_off,
273 			int inverse);
274 
275 	int (*snat_handler)(struct sk_buff *skb,
276 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
277 
278 	int (*dnat_handler)(struct sk_buff *skb,
279 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
280 
281 	int (*csum_check)(struct sk_buff *skb, struct ip_vs_protocol *pp);
282 
283 	const char *(*state_name)(int state);
284 
285 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
286 				const struct sk_buff *skb,
287 				struct ip_vs_protocol *pp);
288 
289 	int (*register_app)(struct ip_vs_app *inc);
290 
291 	void (*unregister_app)(struct ip_vs_app *inc);
292 
293 	int (*app_conn_bind)(struct ip_vs_conn *cp);
294 
295 	void (*debug_packet)(struct ip_vs_protocol *pp,
296 			     const struct sk_buff *skb,
297 			     int offset,
298 			     const char *msg);
299 
300 	void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
301 
302 	int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
303 };
304 
305 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
306 
307 /*
308  *	IP_VS structure allocated for each dynamically scheduled connection
309  */
310 struct ip_vs_conn {
311 	struct list_head        c_list;         /* hashed list heads */
312 
313 	/* Protocol, addresses and port numbers */
314 	u16                      af;		/* address family */
315 	union nf_inet_addr       caddr;          /* client address */
316 	union nf_inet_addr       vaddr;          /* virtual address */
317 	union nf_inet_addr       daddr;          /* destination address */
318 	__be16                   cport;
319 	__be16                   vport;
320 	__be16                   dport;
321 	__u16                   protocol;       /* Which protocol (TCP/UDP) */
322 
323 	/* counter and timer */
324 	atomic_t		refcnt;		/* reference count */
325 	struct timer_list	timer;		/* Expiration timer */
326 	volatile unsigned long	timeout;	/* timeout */
327 
328 	/* Flags and state transition */
329 	spinlock_t              lock;           /* lock for state transition */
330 	volatile __u16          flags;          /* status flags */
331 	volatile __u16          state;          /* state info */
332 	volatile __u16          old_state;      /* old state, to be used for
333 						 * state transition triggerd
334 						 * synchronization
335 						 */
336 
337 	/* Control members */
338 	struct ip_vs_conn       *control;       /* Master control connection */
339 	atomic_t                n_control;      /* Number of controlled ones */
340 	struct ip_vs_dest       *dest;          /* real server */
341 	atomic_t                in_pkts;        /* incoming packet counter */
342 
343 	/* packet transmitter for different forwarding methods.  If it
344 	   mangles the packet, it must return NF_DROP or better NF_STOLEN,
345 	   otherwise this must be changed to a sk_buff **.
346 	 */
347 	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
348 			   struct ip_vs_protocol *pp);
349 
350 	/* Note: we can group the following members into a structure,
351 	   in order to save more space, and the following members are
352 	   only used in VS/NAT anyway */
353 	struct ip_vs_app        *app;           /* bound ip_vs_app object */
354 	void                    *app_data;      /* Application private data */
355 	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
356 	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
357 };
358 
359 
360 /*
361  *	The information about the virtual service offered to the net
362  *	and the forwarding entries
363  */
364 struct ip_vs_service {
365 	struct list_head	s_list;   /* for normal service table */
366 	struct list_head	f_list;   /* for fwmark-based service table */
367 	atomic_t		refcnt;   /* reference counter */
368 	atomic_t		usecnt;   /* use counter */
369 
370 	u16			af;       /* address family */
371 	__u16			protocol; /* which protocol (TCP/UDP) */
372 	union nf_inet_addr	addr;	  /* IP address for virtual service */
373 	__be16			port;	  /* port number for the service */
374 	__u32                   fwmark;   /* firewall mark of the service */
375 	unsigned		flags;	  /* service status flags */
376 	unsigned		timeout;  /* persistent timeout in ticks */
377 	__be32			netmask;  /* grouping granularity */
378 
379 	struct list_head	destinations;  /* real server d-linked list */
380 	__u32			num_dests;     /* number of servers */
381 	struct ip_vs_stats      stats;         /* statistics for the service */
382 	struct ip_vs_app	*inc;	  /* bind conns to this app inc */
383 
384 	/* for scheduling */
385 	struct ip_vs_scheduler	*scheduler;    /* bound scheduler object */
386 	rwlock_t		sched_lock;    /* lock sched_data */
387 	void			*sched_data;   /* scheduler application data */
388 };
389 
390 
391 /*
392  *	The real server destination forwarding entry
393  *	with ip address, port number, and so on.
394  */
395 struct ip_vs_dest {
396 	struct list_head	n_list;   /* for the dests in the service */
397 	struct list_head	d_list;   /* for table with all the dests */
398 
399 	u16			af;		/* address family */
400 	union nf_inet_addr	addr;		/* IP address of the server */
401 	__be16			port;		/* port number of the server */
402 	volatile unsigned	flags;		/* dest status flags */
403 	atomic_t		conn_flags;	/* flags to copy to conn */
404 	atomic_t		weight;		/* server weight */
405 
406 	atomic_t		refcnt;		/* reference counter */
407 	struct ip_vs_stats      stats;          /* statistics */
408 
409 	/* connection counters and thresholds */
410 	atomic_t		activeconns;	/* active connections */
411 	atomic_t		inactconns;	/* inactive connections */
412 	atomic_t		persistconns;	/* persistent connections */
413 	__u32			u_threshold;	/* upper threshold */
414 	__u32			l_threshold;	/* lower threshold */
415 
416 	/* for destination cache */
417 	spinlock_t		dst_lock;	/* lock of dst_cache */
418 	struct dst_entry	*dst_cache;	/* destination cache entry */
419 	u32			dst_rtos;	/* RT_TOS(tos) for dst */
420 
421 	/* for virtual service */
422 	struct ip_vs_service	*svc;		/* service it belongs to */
423 	__u16			protocol;	/* which protocol (TCP/UDP) */
424 	union nf_inet_addr	vaddr;		/* virtual IP address */
425 	__be16			vport;		/* virtual port number */
426 	__u32			vfwmark;	/* firewall mark of service */
427 };
428 
429 
430 /*
431  *	The scheduler object
432  */
433 struct ip_vs_scheduler {
434 	struct list_head	n_list;		/* d-linked list head */
435 	char			*name;		/* scheduler name */
436 	atomic_t		refcnt;		/* reference counter */
437 	struct module		*module;	/* THIS_MODULE/NULL */
438 
439 	/* scheduler initializing service */
440 	int (*init_service)(struct ip_vs_service *svc);
441 	/* scheduling service finish */
442 	int (*done_service)(struct ip_vs_service *svc);
443 	/* scheduler updating service */
444 	int (*update_service)(struct ip_vs_service *svc);
445 
446 	/* selecting a server from the given service */
447 	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
448 				       const struct sk_buff *skb);
449 };
450 
451 
452 /*
453  *	The application module object (a.k.a. app incarnation)
454  */
455 struct ip_vs_app
456 {
457 	struct list_head	a_list;		/* member in app list */
458 	int			type;		/* IP_VS_APP_TYPE_xxx */
459 	char			*name;		/* application module name */
460 	__u16			protocol;
461 	struct module		*module;	/* THIS_MODULE/NULL */
462 	struct list_head	incs_list;	/* list of incarnations */
463 
464 	/* members for application incarnations */
465 	struct list_head	p_list;		/* member in proto app list */
466 	struct ip_vs_app	*app;		/* its real application */
467 	__be16			port;		/* port number in net order */
468 	atomic_t		usecnt;		/* usage counter */
469 
470 	/* output hook: return false if can't linearize. diff set for TCP.  */
471 	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
472 		       struct sk_buff *, int *diff);
473 
474 	/* input hook: return false if can't linearize. diff set for TCP. */
475 	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
476 		      struct sk_buff *, int *diff);
477 
478 	/* ip_vs_app initializer */
479 	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
480 
481 	/* ip_vs_app finish */
482 	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
483 
484 
485 	/* not used now */
486 	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
487 			 struct ip_vs_protocol *);
488 
489 	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
490 
491 	int *			timeout_table;
492 	int *			timeouts;
493 	int			timeouts_size;
494 
495 	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
496 			     int *verdict, struct ip_vs_conn **cpp);
497 
498 	struct ip_vs_conn *
499 	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
500 		       const struct iphdr *iph, unsigned int proto_off,
501 		       int inverse);
502 
503 	struct ip_vs_conn *
504 	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
505 			const struct iphdr *iph, unsigned int proto_off,
506 			int inverse);
507 
508 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
509 				const struct sk_buff *skb,
510 				struct ip_vs_app *app);
511 
512 	void (*timeout_change)(struct ip_vs_app *app, int flags);
513 };
514 
515 
516 /*
517  *      IPVS core functions
518  *      (from ip_vs_core.c)
519  */
520 extern const char *ip_vs_proto_name(unsigned proto);
521 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
522 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
523 
524 #define IP_VS_APP_TYPE_FTP	1
525 
526 /*
527  *     ip_vs_conn handling functions
528  *     (from ip_vs_conn.c)
529  */
530 
531 /*
532  *     IPVS connection entry hash table
533  */
534 #ifndef CONFIG_IP_VS_TAB_BITS
535 #define CONFIG_IP_VS_TAB_BITS   12
536 #endif
537 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
538 #if CONFIG_IP_VS_TAB_BITS < 8
539 #define IP_VS_CONN_TAB_BITS	8
540 #endif
541 #if CONFIG_IP_VS_TAB_BITS > 20
542 #define IP_VS_CONN_TAB_BITS	20
543 #endif
544 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
545 #define IP_VS_CONN_TAB_BITS	CONFIG_IP_VS_TAB_BITS
546 #endif
547 #define IP_VS_CONN_TAB_SIZE     (1 << IP_VS_CONN_TAB_BITS)
548 #define IP_VS_CONN_TAB_MASK     (IP_VS_CONN_TAB_SIZE - 1)
549 
550 enum {
551 	IP_VS_DIR_INPUT = 0,
552 	IP_VS_DIR_OUTPUT,
553 	IP_VS_DIR_INPUT_ONLY,
554 	IP_VS_DIR_LAST,
555 };
556 
557 extern struct ip_vs_conn *ip_vs_conn_in_get
558 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
559 extern struct ip_vs_conn *ip_vs_ct_in_get
560 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
561 extern struct ip_vs_conn *ip_vs_conn_out_get
562 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
563 
564 /* put back the conn without restarting its timer */
565 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
566 {
567 	atomic_dec(&cp->refcnt);
568 }
569 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
570 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
571 
572 extern struct ip_vs_conn *
573 ip_vs_conn_new(int proto, __be32 caddr, __be16 cport, __be32 vaddr, __be16 vport,
574 	       __be32 daddr, __be16 dport, unsigned flags,
575 	       struct ip_vs_dest *dest);
576 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
577 
578 extern const char * ip_vs_state_name(__u16 proto, int state);
579 
580 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
581 extern int ip_vs_check_template(struct ip_vs_conn *ct);
582 extern void ip_vs_random_dropentry(void);
583 extern int ip_vs_conn_init(void);
584 extern void ip_vs_conn_cleanup(void);
585 
586 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
587 {
588 	struct ip_vs_conn *ctl_cp = cp->control;
589 	if (!ctl_cp) {
590 		IP_VS_ERR("request control DEL for uncontrolled: "
591 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
592 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
593 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
594 		return;
595 	}
596 
597 	IP_VS_DBG(7, "DELeting control for: "
598 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
599 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
600 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
601 
602 	cp->control = NULL;
603 	if (atomic_read(&ctl_cp->n_control) == 0) {
604 		IP_VS_ERR("BUG control DEL with n=0 : "
605 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
606 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
607 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
608 		return;
609 	}
610 	atomic_dec(&ctl_cp->n_control);
611 }
612 
613 static inline void
614 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
615 {
616 	if (cp->control) {
617 		IP_VS_ERR("request control ADD for already controlled: "
618 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
619 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
620 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
621 		ip_vs_control_del(cp);
622 	}
623 
624 	IP_VS_DBG(7, "ADDing control for: "
625 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
626 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
627 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
628 
629 	cp->control = ctl_cp;
630 	atomic_inc(&ctl_cp->n_control);
631 }
632 
633 
634 /*
635  *      IPVS application functions
636  *      (from ip_vs_app.c)
637  */
638 #define IP_VS_APP_MAX_PORTS  8
639 extern int register_ip_vs_app(struct ip_vs_app *app);
640 extern void unregister_ip_vs_app(struct ip_vs_app *app);
641 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
642 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
643 extern int
644 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
645 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
646 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
647 
648 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
649 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
650 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
651 			     char *o_buf, int o_len, char *n_buf, int n_len);
652 extern int ip_vs_app_init(void);
653 extern void ip_vs_app_cleanup(void);
654 
655 
656 /*
657  *	IPVS protocol functions (from ip_vs_proto.c)
658  */
659 extern int ip_vs_protocol_init(void);
660 extern void ip_vs_protocol_cleanup(void);
661 extern void ip_vs_protocol_timeout_change(int flags);
662 extern int *ip_vs_create_timeout_table(int *table, int size);
663 extern int
664 ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to);
665 extern void
666 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
667 			  int offset, const char *msg);
668 
669 extern struct ip_vs_protocol ip_vs_protocol_tcp;
670 extern struct ip_vs_protocol ip_vs_protocol_udp;
671 extern struct ip_vs_protocol ip_vs_protocol_icmp;
672 extern struct ip_vs_protocol ip_vs_protocol_esp;
673 extern struct ip_vs_protocol ip_vs_protocol_ah;
674 
675 
676 /*
677  *      Registering/unregistering scheduler functions
678  *      (from ip_vs_sched.c)
679  */
680 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
681 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
682 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
683 				struct ip_vs_scheduler *scheduler);
684 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
685 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
686 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
687 extern struct ip_vs_conn *
688 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
689 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
690 			struct ip_vs_protocol *pp);
691 
692 
693 /*
694  *      IPVS control data and functions (from ip_vs_ctl.c)
695  */
696 extern int sysctl_ip_vs_cache_bypass;
697 extern int sysctl_ip_vs_expire_nodest_conn;
698 extern int sysctl_ip_vs_expire_quiescent_template;
699 extern int sysctl_ip_vs_sync_threshold[2];
700 extern int sysctl_ip_vs_nat_icmp_send;
701 extern struct ip_vs_stats ip_vs_stats;
702 extern const struct ctl_path net_vs_ctl_path[];
703 
704 extern struct ip_vs_service *
705 ip_vs_service_get(__u32 fwmark, __u16 protocol, __be32 vaddr, __be16 vport);
706 
707 static inline void ip_vs_service_put(struct ip_vs_service *svc)
708 {
709 	atomic_dec(&svc->usecnt);
710 }
711 
712 extern struct ip_vs_dest *
713 ip_vs_lookup_real_service(__u16 protocol, __be32 daddr, __be16 dport);
714 extern int ip_vs_use_count_inc(void);
715 extern void ip_vs_use_count_dec(void);
716 extern int ip_vs_control_init(void);
717 extern void ip_vs_control_cleanup(void);
718 extern struct ip_vs_dest *
719 ip_vs_find_dest(__be32 daddr, __be16 dport,
720 		 __be32 vaddr, __be16 vport, __u16 protocol);
721 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
722 
723 
724 /*
725  *      IPVS sync daemon data and function prototypes
726  *      (from ip_vs_sync.c)
727  */
728 extern volatile int ip_vs_sync_state;
729 extern volatile int ip_vs_master_syncid;
730 extern volatile int ip_vs_backup_syncid;
731 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
732 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
733 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
734 extern int stop_sync_thread(int state);
735 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
736 
737 
738 /*
739  *      IPVS rate estimator prototypes (from ip_vs_est.c)
740  */
741 extern int ip_vs_estimator_init(void);
742 extern void ip_vs_estimator_cleanup(void);
743 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
744 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
745 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
746 
747 /*
748  *	Various IPVS packet transmitters (from ip_vs_xmit.c)
749  */
750 extern int ip_vs_null_xmit
751 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
752 extern int ip_vs_bypass_xmit
753 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
754 extern int ip_vs_nat_xmit
755 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
756 extern int ip_vs_tunnel_xmit
757 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
758 extern int ip_vs_dr_xmit
759 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
760 extern int ip_vs_icmp_xmit
761 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
762 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
763 
764 
765 /*
766  *	This is a simple mechanism to ignore packets when
767  *	we are loaded. Just set ip_vs_drop_rate to 'n' and
768  *	we start to drop 1/rate of the packets
769  */
770 extern int ip_vs_drop_rate;
771 extern int ip_vs_drop_counter;
772 
773 static __inline__ int ip_vs_todrop(void)
774 {
775 	if (!ip_vs_drop_rate) return 0;
776 	if (--ip_vs_drop_counter > 0) return 0;
777 	ip_vs_drop_counter = ip_vs_drop_rate;
778 	return 1;
779 }
780 
781 /*
782  *      ip_vs_fwd_tag returns the forwarding tag of the connection
783  */
784 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
785 
786 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
787 {
788 	char fwd;
789 
790 	switch (IP_VS_FWD_METHOD(cp)) {
791 	case IP_VS_CONN_F_MASQ:
792 		fwd = 'M'; break;
793 	case IP_VS_CONN_F_LOCALNODE:
794 		fwd = 'L'; break;
795 	case IP_VS_CONN_F_TUNNEL:
796 		fwd = 'T'; break;
797 	case IP_VS_CONN_F_DROUTE:
798 		fwd = 'R'; break;
799 	case IP_VS_CONN_F_BYPASS:
800 		fwd = 'B'; break;
801 	default:
802 		fwd = '?'; break;
803 	}
804 	return fwd;
805 }
806 
807 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
808 		struct ip_vs_conn *cp, int dir);
809 
810 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
811 
812 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
813 {
814 	__be32 diff[2] = { ~old, new };
815 
816 	return csum_partial((char *) diff, sizeof(diff), oldsum);
817 }
818 
819 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
820 {
821 	__be16 diff[2] = { ~old, new };
822 
823 	return csum_partial((char *) diff, sizeof(diff), oldsum);
824 }
825 
826 #endif /* __KERNEL__ */
827 
828 #endif	/* _NET_IP_VS_H */
829