xref: /linux/include/net/ip_vs.h (revision c0c9209ddd96bc4f1d70a8b9958710671e076080)
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 
25 #ifdef CONFIG_IP_VS_DEBUG
26 #include <linux/net.h>
27 
28 extern int ip_vs_get_debug_level(void);
29 #define IP_VS_DBG(level, msg...)			\
30     do {						\
31 	    if (level <= ip_vs_get_debug_level())	\
32 		    printk(KERN_DEBUG "IPVS: " msg);	\
33     } while (0)
34 #define IP_VS_DBG_RL(msg...)				\
35     do {						\
36 	    if (net_ratelimit())			\
37 		    printk(KERN_DEBUG "IPVS: " msg);	\
38     } while (0)
39 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		\
40     do {						\
41 	    if (level <= ip_vs_get_debug_level())	\
42 		pp->debug_packet(pp, skb, ofs, msg);	\
43     } while (0)
44 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	\
45     do {						\
46 	    if (level <= ip_vs_get_debug_level() &&	\
47 		net_ratelimit())			\
48 		pp->debug_packet(pp, skb, ofs, msg);	\
49     } while (0)
50 #else	/* NO DEBUGGING at ALL */
51 #define IP_VS_DBG(level, msg...)  do {} while (0)
52 #define IP_VS_DBG_RL(msg...)  do {} while (0)
53 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		do {} while (0)
54 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	do {} while (0)
55 #endif
56 
57 #define IP_VS_BUG() BUG()
58 #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg)
59 #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg)
60 #define IP_VS_WARNING(msg...) \
61 	printk(KERN_WARNING "IPVS: " msg)
62 #define IP_VS_ERR_RL(msg...)				\
63     do {						\
64 	    if (net_ratelimit())			\
65 		    printk(KERN_ERR "IPVS: " msg);	\
66     } while (0)
67 
68 #ifdef CONFIG_IP_VS_DEBUG
69 #define EnterFunction(level)						\
70     do {								\
71 	    if (level <= ip_vs_get_debug_level())			\
72 		    printk(KERN_DEBUG "Enter: %s, %s line %i\n",	\
73 			   __FUNCTION__, __FILE__, __LINE__);		\
74     } while (0)
75 #define LeaveFunction(level)                                            \
76     do {                                                                \
77 	    if (level <= ip_vs_get_debug_level())                       \
78 			printk(KERN_DEBUG "Leave: %s, %s line %i\n",    \
79 			       __FUNCTION__, __FILE__, __LINE__);       \
80     } while (0)
81 #else
82 #define EnterFunction(level)   do {} while (0)
83 #define LeaveFunction(level)   do {} while (0)
84 #endif
85 
86 #define	IP_VS_WAIT_WHILE(expr)	while (expr) { cpu_relax(); }
87 
88 
89 /*
90  *      The port number of FTP service (in network order).
91  */
92 #define FTPPORT  __constant_htons(21)
93 #define FTPDATA  __constant_htons(20)
94 
95 /*
96  *      TCP State Values
97  */
98 enum {
99 	IP_VS_TCP_S_NONE = 0,
100 	IP_VS_TCP_S_ESTABLISHED,
101 	IP_VS_TCP_S_SYN_SENT,
102 	IP_VS_TCP_S_SYN_RECV,
103 	IP_VS_TCP_S_FIN_WAIT,
104 	IP_VS_TCP_S_TIME_WAIT,
105 	IP_VS_TCP_S_CLOSE,
106 	IP_VS_TCP_S_CLOSE_WAIT,
107 	IP_VS_TCP_S_LAST_ACK,
108 	IP_VS_TCP_S_LISTEN,
109 	IP_VS_TCP_S_SYNACK,
110 	IP_VS_TCP_S_LAST
111 };
112 
113 /*
114  *	UDP State Values
115  */
116 enum {
117 	IP_VS_UDP_S_NORMAL,
118 	IP_VS_UDP_S_LAST,
119 };
120 
121 /*
122  *	ICMP State Values
123  */
124 enum {
125 	IP_VS_ICMP_S_NORMAL,
126 	IP_VS_ICMP_S_LAST,
127 };
128 
129 /*
130  *	Delta sequence info structure
131  *	Each ip_vs_conn has 2 (output AND input seq. changes).
132  *      Only used in the VS/NAT.
133  */
134 struct ip_vs_seq {
135 	__u32			init_seq;	/* Add delta from this seq */
136 	__u32			delta;		/* Delta in sequence numbers */
137 	__u32			previous_delta;	/* Delta in sequence numbers
138 						   before last resized pkt */
139 };
140 
141 
142 /*
143  *	IPVS statistics objects
144  */
145 struct ip_vs_estimator {
146 	struct list_head	list;
147 
148 	u64			last_inbytes;
149 	u64			last_outbytes;
150 	u32			last_conns;
151 	u32			last_inpkts;
152 	u32			last_outpkts;
153 
154 	u32			cps;
155 	u32			inpps;
156 	u32			outpps;
157 	u32			inbps;
158 	u32			outbps;
159 };
160 
161 struct ip_vs_stats
162 {
163 	__u32                   conns;          /* connections scheduled */
164 	__u32                   inpkts;         /* incoming packets */
165 	__u32                   outpkts;        /* outgoing packets */
166 	__u64                   inbytes;        /* incoming bytes */
167 	__u64                   outbytes;       /* outgoing bytes */
168 
169 	__u32			cps;		/* current connection rate */
170 	__u32			inpps;		/* current in packet rate */
171 	__u32			outpps;		/* current out packet rate */
172 	__u32			inbps;		/* current in byte rate */
173 	__u32			outbps;		/* current out byte rate */
174 
175 	/*
176 	 * Don't add anything before the lock, because we use memcpy() to copy
177 	 * the members before the lock to struct ip_vs_stats_user in
178 	 * ip_vs_ctl.c.
179 	 */
180 
181 	spinlock_t              lock;           /* spin lock */
182 
183 	struct ip_vs_estimator	est;		/* estimator */
184 };
185 
186 struct dst_entry;
187 struct iphdr;
188 struct ip_vs_conn;
189 struct ip_vs_app;
190 struct sk_buff;
191 
192 struct ip_vs_protocol {
193 	struct ip_vs_protocol	*next;
194 	char			*name;
195 	u16			protocol;
196 	u16			num_states;
197 	int			dont_defrag;
198 	atomic_t		appcnt;		/* counter of proto app incs */
199 	int			*timeout_table;	/* protocol timeout table */
200 
201 	void (*init)(struct ip_vs_protocol *pp);
202 
203 	void (*exit)(struct ip_vs_protocol *pp);
204 
205 	int (*conn_schedule)(struct sk_buff *skb,
206 			     struct ip_vs_protocol *pp,
207 			     int *verdict, struct ip_vs_conn **cpp);
208 
209 	struct ip_vs_conn *
210 	(*conn_in_get)(const struct sk_buff *skb,
211 		       struct ip_vs_protocol *pp,
212 		       const struct iphdr *iph,
213 		       unsigned int proto_off,
214 		       int inverse);
215 
216 	struct ip_vs_conn *
217 	(*conn_out_get)(const struct sk_buff *skb,
218 			struct ip_vs_protocol *pp,
219 			const struct iphdr *iph,
220 			unsigned int proto_off,
221 			int inverse);
222 
223 	int (*snat_handler)(struct sk_buff *skb,
224 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
225 
226 	int (*dnat_handler)(struct sk_buff *skb,
227 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
228 
229 	int (*csum_check)(struct sk_buff *skb, struct ip_vs_protocol *pp);
230 
231 	const char *(*state_name)(int state);
232 
233 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
234 				const struct sk_buff *skb,
235 				struct ip_vs_protocol *pp);
236 
237 	int (*register_app)(struct ip_vs_app *inc);
238 
239 	void (*unregister_app)(struct ip_vs_app *inc);
240 
241 	int (*app_conn_bind)(struct ip_vs_conn *cp);
242 
243 	void (*debug_packet)(struct ip_vs_protocol *pp,
244 			     const struct sk_buff *skb,
245 			     int offset,
246 			     const char *msg);
247 
248 	void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
249 
250 	int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
251 };
252 
253 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
254 
255 /*
256  *	IP_VS structure allocated for each dynamically scheduled connection
257  */
258 struct ip_vs_conn {
259 	struct list_head        c_list;         /* hashed list heads */
260 
261 	/* Protocol, addresses and port numbers */
262 	__be32                   caddr;          /* client address */
263 	__be32                   vaddr;          /* virtual address */
264 	__be32                   daddr;          /* destination address */
265 	__be16                   cport;
266 	__be16                   vport;
267 	__be16                   dport;
268 	__u16                   protocol;       /* Which protocol (TCP/UDP) */
269 
270 	/* counter and timer */
271 	atomic_t		refcnt;		/* reference count */
272 	struct timer_list	timer;		/* Expiration timer */
273 	volatile unsigned long	timeout;	/* timeout */
274 
275 	/* Flags and state transition */
276 	spinlock_t              lock;           /* lock for state transition */
277 	volatile __u16          flags;          /* status flags */
278 	volatile __u16          state;          /* state info */
279 	volatile __u16          old_state;      /* old state, to be used for
280 						 * state transition triggerd
281 						 * synchronization
282 						 */
283 
284 	/* Control members */
285 	struct ip_vs_conn       *control;       /* Master control connection */
286 	atomic_t                n_control;      /* Number of controlled ones */
287 	struct ip_vs_dest       *dest;          /* real server */
288 	atomic_t                in_pkts;        /* incoming packet counter */
289 
290 	/* packet transmitter for different forwarding methods.  If it
291 	   mangles the packet, it must return NF_DROP or better NF_STOLEN,
292 	   otherwise this must be changed to a sk_buff **.
293 	 */
294 	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
295 			   struct ip_vs_protocol *pp);
296 
297 	/* Note: we can group the following members into a structure,
298 	   in order to save more space, and the following members are
299 	   only used in VS/NAT anyway */
300 	struct ip_vs_app        *app;           /* bound ip_vs_app object */
301 	void                    *app_data;      /* Application private data */
302 	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
303 	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
304 };
305 
306 
307 /*
308  *	The information about the virtual service offered to the net
309  *	and the forwarding entries
310  */
311 struct ip_vs_service {
312 	struct list_head	s_list;   /* for normal service table */
313 	struct list_head	f_list;   /* for fwmark-based service table */
314 	atomic_t		refcnt;   /* reference counter */
315 	atomic_t		usecnt;   /* use counter */
316 
317 	__u16			protocol; /* which protocol (TCP/UDP) */
318 	__be32			addr;	  /* IP address for virtual service */
319 	__be16			port;	  /* port number for the service */
320 	__u32                   fwmark;   /* firewall mark of the service */
321 	unsigned		flags;	  /* service status flags */
322 	unsigned		timeout;  /* persistent timeout in ticks */
323 	__be32			netmask;  /* grouping granularity */
324 
325 	struct list_head	destinations;  /* real server d-linked list */
326 	__u32			num_dests;     /* number of servers */
327 	struct ip_vs_stats      stats;         /* statistics for the service */
328 	struct ip_vs_app	*inc;	  /* bind conns to this app inc */
329 
330 	/* for scheduling */
331 	struct ip_vs_scheduler	*scheduler;    /* bound scheduler object */
332 	rwlock_t		sched_lock;    /* lock sched_data */
333 	void			*sched_data;   /* scheduler application data */
334 };
335 
336 
337 /*
338  *	The real server destination forwarding entry
339  *	with ip address, port number, and so on.
340  */
341 struct ip_vs_dest {
342 	struct list_head	n_list;   /* for the dests in the service */
343 	struct list_head	d_list;   /* for table with all the dests */
344 
345 	__be32			addr;		/* IP address of the server */
346 	__be16			port;		/* port number of the server */
347 	volatile unsigned	flags;		/* dest status flags */
348 	atomic_t		conn_flags;	/* flags to copy to conn */
349 	atomic_t		weight;		/* server weight */
350 
351 	atomic_t		refcnt;		/* reference counter */
352 	struct ip_vs_stats      stats;          /* statistics */
353 
354 	/* connection counters and thresholds */
355 	atomic_t		activeconns;	/* active connections */
356 	atomic_t		inactconns;	/* inactive connections */
357 	atomic_t		persistconns;	/* persistent connections */
358 	__u32			u_threshold;	/* upper threshold */
359 	__u32			l_threshold;	/* lower threshold */
360 
361 	/* for destination cache */
362 	spinlock_t		dst_lock;	/* lock of dst_cache */
363 	struct dst_entry	*dst_cache;	/* destination cache entry */
364 	u32			dst_rtos;	/* RT_TOS(tos) for dst */
365 
366 	/* for virtual service */
367 	struct ip_vs_service	*svc;		/* service it belongs to */
368 	__u16			protocol;	/* which protocol (TCP/UDP) */
369 	__be32			vaddr;		/* virtual IP address */
370 	__be16			vport;		/* virtual port number */
371 	__u32			vfwmark;	/* firewall mark of service */
372 };
373 
374 
375 /*
376  *	The scheduler object
377  */
378 struct ip_vs_scheduler {
379 	struct list_head	n_list;		/* d-linked list head */
380 	char			*name;		/* scheduler name */
381 	atomic_t		refcnt;		/* reference counter */
382 	struct module		*module;	/* THIS_MODULE/NULL */
383 
384 	/* scheduler initializing service */
385 	int (*init_service)(struct ip_vs_service *svc);
386 	/* scheduling service finish */
387 	int (*done_service)(struct ip_vs_service *svc);
388 	/* scheduler updating service */
389 	int (*update_service)(struct ip_vs_service *svc);
390 
391 	/* selecting a server from the given service */
392 	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
393 				       const struct sk_buff *skb);
394 };
395 
396 
397 /*
398  *	The application module object (a.k.a. app incarnation)
399  */
400 struct ip_vs_app
401 {
402 	struct list_head	a_list;		/* member in app list */
403 	int			type;		/* IP_VS_APP_TYPE_xxx */
404 	char			*name;		/* application module name */
405 	__u16			protocol;
406 	struct module		*module;	/* THIS_MODULE/NULL */
407 	struct list_head	incs_list;	/* list of incarnations */
408 
409 	/* members for application incarnations */
410 	struct list_head	p_list;		/* member in proto app list */
411 	struct ip_vs_app	*app;		/* its real application */
412 	__be16			port;		/* port number in net order */
413 	atomic_t		usecnt;		/* usage counter */
414 
415 	/* output hook: return false if can't linearize. diff set for TCP.  */
416 	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
417 		       struct sk_buff *, int *diff);
418 
419 	/* input hook: return false if can't linearize. diff set for TCP. */
420 	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
421 		      struct sk_buff *, int *diff);
422 
423 	/* ip_vs_app initializer */
424 	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
425 
426 	/* ip_vs_app finish */
427 	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
428 
429 
430 	/* not used now */
431 	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
432 			 struct ip_vs_protocol *);
433 
434 	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
435 
436 	int *			timeout_table;
437 	int *			timeouts;
438 	int			timeouts_size;
439 
440 	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
441 			     int *verdict, struct ip_vs_conn **cpp);
442 
443 	struct ip_vs_conn *
444 	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
445 		       const struct iphdr *iph, unsigned int proto_off,
446 		       int inverse);
447 
448 	struct ip_vs_conn *
449 	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
450 			const struct iphdr *iph, unsigned int proto_off,
451 			int inverse);
452 
453 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
454 				const struct sk_buff *skb,
455 				struct ip_vs_app *app);
456 
457 	void (*timeout_change)(struct ip_vs_app *app, int flags);
458 };
459 
460 
461 /*
462  *      IPVS core functions
463  *      (from ip_vs_core.c)
464  */
465 extern const char *ip_vs_proto_name(unsigned proto);
466 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
467 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
468 
469 #define IP_VS_APP_TYPE_FTP	1
470 
471 /*
472  *     ip_vs_conn handling functions
473  *     (from ip_vs_conn.c)
474  */
475 
476 /*
477  *     IPVS connection entry hash table
478  */
479 #ifndef CONFIG_IP_VS_TAB_BITS
480 #define CONFIG_IP_VS_TAB_BITS   12
481 #endif
482 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
483 #if CONFIG_IP_VS_TAB_BITS < 8
484 #define IP_VS_CONN_TAB_BITS	8
485 #endif
486 #if CONFIG_IP_VS_TAB_BITS > 20
487 #define IP_VS_CONN_TAB_BITS	20
488 #endif
489 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
490 #define IP_VS_CONN_TAB_BITS	CONFIG_IP_VS_TAB_BITS
491 #endif
492 #define IP_VS_CONN_TAB_SIZE     (1 << IP_VS_CONN_TAB_BITS)
493 #define IP_VS_CONN_TAB_MASK     (IP_VS_CONN_TAB_SIZE - 1)
494 
495 enum {
496 	IP_VS_DIR_INPUT = 0,
497 	IP_VS_DIR_OUTPUT,
498 	IP_VS_DIR_INPUT_ONLY,
499 	IP_VS_DIR_LAST,
500 };
501 
502 extern struct ip_vs_conn *ip_vs_conn_in_get
503 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
504 extern struct ip_vs_conn *ip_vs_ct_in_get
505 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
506 extern struct ip_vs_conn *ip_vs_conn_out_get
507 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
508 
509 /* put back the conn without restarting its timer */
510 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
511 {
512 	atomic_dec(&cp->refcnt);
513 }
514 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
515 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
516 
517 extern struct ip_vs_conn *
518 ip_vs_conn_new(int proto, __be32 caddr, __be16 cport, __be32 vaddr, __be16 vport,
519 	       __be32 daddr, __be16 dport, unsigned flags,
520 	       struct ip_vs_dest *dest);
521 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
522 
523 extern const char * ip_vs_state_name(__u16 proto, int state);
524 
525 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
526 extern int ip_vs_check_template(struct ip_vs_conn *ct);
527 extern void ip_vs_random_dropentry(void);
528 extern int ip_vs_conn_init(void);
529 extern void ip_vs_conn_cleanup(void);
530 
531 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
532 {
533 	struct ip_vs_conn *ctl_cp = cp->control;
534 	if (!ctl_cp) {
535 		IP_VS_ERR("request control DEL for uncontrolled: "
536 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
537 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
538 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
539 		return;
540 	}
541 
542 	IP_VS_DBG(7, "DELeting control for: "
543 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
544 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
545 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
546 
547 	cp->control = NULL;
548 	if (atomic_read(&ctl_cp->n_control) == 0) {
549 		IP_VS_ERR("BUG control DEL with n=0 : "
550 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
551 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
552 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
553 		return;
554 	}
555 	atomic_dec(&ctl_cp->n_control);
556 }
557 
558 static inline void
559 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
560 {
561 	if (cp->control) {
562 		IP_VS_ERR("request control ADD for already controlled: "
563 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
564 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
565 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
566 		ip_vs_control_del(cp);
567 	}
568 
569 	IP_VS_DBG(7, "ADDing control for: "
570 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
571 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
572 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
573 
574 	cp->control = ctl_cp;
575 	atomic_inc(&ctl_cp->n_control);
576 }
577 
578 
579 /*
580  *      IPVS application functions
581  *      (from ip_vs_app.c)
582  */
583 #define IP_VS_APP_MAX_PORTS  8
584 extern int register_ip_vs_app(struct ip_vs_app *app);
585 extern void unregister_ip_vs_app(struct ip_vs_app *app);
586 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
587 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
588 extern int
589 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
590 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
591 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
592 
593 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
594 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
595 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
596 			     char *o_buf, int o_len, char *n_buf, int n_len);
597 extern int ip_vs_app_init(void);
598 extern void ip_vs_app_cleanup(void);
599 
600 
601 /*
602  *	IPVS protocol functions (from ip_vs_proto.c)
603  */
604 extern int ip_vs_protocol_init(void);
605 extern void ip_vs_protocol_cleanup(void);
606 extern void ip_vs_protocol_timeout_change(int flags);
607 extern int *ip_vs_create_timeout_table(int *table, int size);
608 extern int
609 ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to);
610 extern void
611 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
612 			  int offset, const char *msg);
613 
614 extern struct ip_vs_protocol ip_vs_protocol_tcp;
615 extern struct ip_vs_protocol ip_vs_protocol_udp;
616 extern struct ip_vs_protocol ip_vs_protocol_icmp;
617 extern struct ip_vs_protocol ip_vs_protocol_esp;
618 extern struct ip_vs_protocol ip_vs_protocol_ah;
619 
620 
621 /*
622  *      Registering/unregistering scheduler functions
623  *      (from ip_vs_sched.c)
624  */
625 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
626 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
627 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
628 				struct ip_vs_scheduler *scheduler);
629 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
630 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
631 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
632 extern struct ip_vs_conn *
633 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
634 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
635 			struct ip_vs_protocol *pp);
636 
637 
638 /*
639  *      IPVS control data and functions (from ip_vs_ctl.c)
640  */
641 extern int sysctl_ip_vs_cache_bypass;
642 extern int sysctl_ip_vs_expire_nodest_conn;
643 extern int sysctl_ip_vs_expire_quiescent_template;
644 extern int sysctl_ip_vs_sync_threshold[2];
645 extern int sysctl_ip_vs_nat_icmp_send;
646 extern struct ip_vs_stats ip_vs_stats;
647 extern const struct ctl_path net_vs_ctl_path[];
648 
649 extern struct ip_vs_service *
650 ip_vs_service_get(__u32 fwmark, __u16 protocol, __be32 vaddr, __be16 vport);
651 
652 static inline void ip_vs_service_put(struct ip_vs_service *svc)
653 {
654 	atomic_dec(&svc->usecnt);
655 }
656 
657 extern struct ip_vs_dest *
658 ip_vs_lookup_real_service(__u16 protocol, __be32 daddr, __be16 dport);
659 extern int ip_vs_use_count_inc(void);
660 extern void ip_vs_use_count_dec(void);
661 extern int ip_vs_control_init(void);
662 extern void ip_vs_control_cleanup(void);
663 extern struct ip_vs_dest *
664 ip_vs_find_dest(__be32 daddr, __be16 dport,
665 		 __be32 vaddr, __be16 vport, __u16 protocol);
666 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
667 
668 
669 /*
670  *      IPVS sync daemon data and function prototypes
671  *      (from ip_vs_sync.c)
672  */
673 extern volatile int ip_vs_sync_state;
674 extern volatile int ip_vs_master_syncid;
675 extern volatile int ip_vs_backup_syncid;
676 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
677 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
678 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
679 extern int stop_sync_thread(int state);
680 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
681 
682 
683 /*
684  *      IPVS rate estimator prototypes (from ip_vs_est.c)
685  */
686 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
687 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
688 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
689 
690 /*
691  *	Various IPVS packet transmitters (from ip_vs_xmit.c)
692  */
693 extern int ip_vs_null_xmit
694 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
695 extern int ip_vs_bypass_xmit
696 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
697 extern int ip_vs_nat_xmit
698 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
699 extern int ip_vs_tunnel_xmit
700 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
701 extern int ip_vs_dr_xmit
702 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
703 extern int ip_vs_icmp_xmit
704 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
705 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
706 
707 
708 /*
709  *	This is a simple mechanism to ignore packets when
710  *	we are loaded. Just set ip_vs_drop_rate to 'n' and
711  *	we start to drop 1/rate of the packets
712  */
713 extern int ip_vs_drop_rate;
714 extern int ip_vs_drop_counter;
715 
716 static __inline__ int ip_vs_todrop(void)
717 {
718 	if (!ip_vs_drop_rate) return 0;
719 	if (--ip_vs_drop_counter > 0) return 0;
720 	ip_vs_drop_counter = ip_vs_drop_rate;
721 	return 1;
722 }
723 
724 /*
725  *      ip_vs_fwd_tag returns the forwarding tag of the connection
726  */
727 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
728 
729 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
730 {
731 	char fwd;
732 
733 	switch (IP_VS_FWD_METHOD(cp)) {
734 	case IP_VS_CONN_F_MASQ:
735 		fwd = 'M'; break;
736 	case IP_VS_CONN_F_LOCALNODE:
737 		fwd = 'L'; break;
738 	case IP_VS_CONN_F_TUNNEL:
739 		fwd = 'T'; break;
740 	case IP_VS_CONN_F_DROUTE:
741 		fwd = 'R'; break;
742 	case IP_VS_CONN_F_BYPASS:
743 		fwd = 'B'; break;
744 	default:
745 		fwd = '?'; break;
746 	}
747 	return fwd;
748 }
749 
750 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
751 		struct ip_vs_conn *cp, int dir);
752 
753 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
754 
755 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
756 {
757 	__be32 diff[2] = { ~old, new };
758 
759 	return csum_partial((char *) diff, sizeof(diff), oldsum);
760 }
761 
762 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
763 {
764 	__be16 diff[2] = { ~old, new };
765 
766 	return csum_partial((char *) diff, sizeof(diff), oldsum);
767 }
768 
769 #endif /* __KERNEL__ */
770 
771 #endif	/* _NET_IP_VS_H */
772