xref: /linux/include/net/ip_vs.h (revision 6b2d2cec1081a979e0efd6a1e9559e5a01a3c10e)
1 /*
2  *      IP Virtual Server
3  *      data structure and functionality definitions
4  */
5 
6 #ifndef _IP_VS_H
7 #define _IP_VS_H
8 
9 #include <asm/types.h>		/* For __uXX types */
10 #include <linux/types.h>	/* For __beXX types in userland */
11 
12 #define IP_VS_VERSION_CODE	0x010201
13 #define NVERSION(version)			\
14 	(version >> 16) & 0xFF,			\
15 	(version >> 8) & 0xFF,			\
16 	version & 0xFF
17 
18 /*
19  *      Virtual Service Flags
20  */
21 #define IP_VS_SVC_F_PERSISTENT	0x0001		/* persistent port */
22 #define IP_VS_SVC_F_HASHED	0x0002		/* hashed entry */
23 
24 /*
25  *      Destination Server Flags
26  */
27 #define IP_VS_DEST_F_AVAILABLE	0x0001		/* server is available */
28 #define IP_VS_DEST_F_OVERLOAD	0x0002		/* server is overloaded */
29 
30 /*
31  *      IPVS sync daemon states
32  */
33 #define IP_VS_STATE_NONE	0x0000		/* daemon is stopped */
34 #define IP_VS_STATE_MASTER	0x0001		/* started as master */
35 #define IP_VS_STATE_BACKUP	0x0002		/* started as backup */
36 
37 /*
38  *      IPVS socket options
39  */
40 #define IP_VS_BASE_CTL		(64+1024+64)		/* base */
41 
42 #define IP_VS_SO_SET_NONE	IP_VS_BASE_CTL		/* just peek */
43 #define IP_VS_SO_SET_INSERT	(IP_VS_BASE_CTL+1)
44 #define IP_VS_SO_SET_ADD	(IP_VS_BASE_CTL+2)
45 #define IP_VS_SO_SET_EDIT	(IP_VS_BASE_CTL+3)
46 #define IP_VS_SO_SET_DEL	(IP_VS_BASE_CTL+4)
47 #define IP_VS_SO_SET_FLUSH	(IP_VS_BASE_CTL+5)
48 #define IP_VS_SO_SET_LIST	(IP_VS_BASE_CTL+6)
49 #define IP_VS_SO_SET_ADDDEST	(IP_VS_BASE_CTL+7)
50 #define IP_VS_SO_SET_DELDEST	(IP_VS_BASE_CTL+8)
51 #define IP_VS_SO_SET_EDITDEST	(IP_VS_BASE_CTL+9)
52 #define IP_VS_SO_SET_TIMEOUT	(IP_VS_BASE_CTL+10)
53 #define IP_VS_SO_SET_STARTDAEMON (IP_VS_BASE_CTL+11)
54 #define IP_VS_SO_SET_STOPDAEMON (IP_VS_BASE_CTL+12)
55 #define IP_VS_SO_SET_RESTORE    (IP_VS_BASE_CTL+13)
56 #define IP_VS_SO_SET_SAVE       (IP_VS_BASE_CTL+14)
57 #define IP_VS_SO_SET_ZERO	(IP_VS_BASE_CTL+15)
58 #define IP_VS_SO_SET_MAX	IP_VS_SO_SET_ZERO
59 
60 #define IP_VS_SO_GET_VERSION	IP_VS_BASE_CTL
61 #define IP_VS_SO_GET_INFO	(IP_VS_BASE_CTL+1)
62 #define IP_VS_SO_GET_SERVICES	(IP_VS_BASE_CTL+2)
63 #define IP_VS_SO_GET_SERVICE	(IP_VS_BASE_CTL+3)
64 #define IP_VS_SO_GET_DESTS	(IP_VS_BASE_CTL+4)
65 #define IP_VS_SO_GET_DEST	(IP_VS_BASE_CTL+5)	/* not used now */
66 #define IP_VS_SO_GET_TIMEOUT	(IP_VS_BASE_CTL+6)
67 #define IP_VS_SO_GET_DAEMON	(IP_VS_BASE_CTL+7)
68 #define IP_VS_SO_GET_MAX	IP_VS_SO_GET_DAEMON
69 
70 
71 /*
72  *      IPVS Connection Flags
73  */
74 #define IP_VS_CONN_F_FWD_MASK	0x0007		/* mask for the fwd methods */
75 #define IP_VS_CONN_F_MASQ	0x0000		/* masquerading/NAT */
76 #define IP_VS_CONN_F_LOCALNODE	0x0001		/* local node */
77 #define IP_VS_CONN_F_TUNNEL	0x0002		/* tunneling */
78 #define IP_VS_CONN_F_DROUTE	0x0003		/* direct routing */
79 #define IP_VS_CONN_F_BYPASS	0x0004		/* cache bypass */
80 #define IP_VS_CONN_F_SYNC	0x0020		/* entry created by sync */
81 #define IP_VS_CONN_F_HASHED	0x0040		/* hashed entry */
82 #define IP_VS_CONN_F_NOOUTPUT	0x0080		/* no output packets */
83 #define IP_VS_CONN_F_INACTIVE	0x0100		/* not established */
84 #define IP_VS_CONN_F_OUT_SEQ	0x0200		/* must do output seq adjust */
85 #define IP_VS_CONN_F_IN_SEQ	0x0400		/* must do input seq adjust */
86 #define IP_VS_CONN_F_SEQ_MASK	0x0600		/* in/out sequence mask */
87 #define IP_VS_CONN_F_NO_CPORT	0x0800		/* no client port set yet */
88 #define IP_VS_CONN_F_TEMPLATE	0x1000		/* template, not connection */
89 
90 /* Move it to better place one day, for now keep it unique */
91 #define NFC_IPVS_PROPERTY	0x10000
92 
93 #define IP_VS_SCHEDNAME_MAXLEN	16
94 #define IP_VS_IFNAME_MAXLEN	16
95 
96 
97 /*
98  *	The struct ip_vs_service_user and struct ip_vs_dest_user are
99  *	used to set IPVS rules through setsockopt.
100  */
101 struct ip_vs_service_user {
102 	/* virtual service addresses */
103 	u_int16_t		protocol;
104 	__be32			addr;		/* virtual ip address */
105 	__be16			port;
106 	u_int32_t		fwmark;		/* firwall mark of service */
107 
108 	/* virtual service options */
109 	char			sched_name[IP_VS_SCHEDNAME_MAXLEN];
110 	unsigned		flags;		/* virtual service flags */
111 	unsigned		timeout;	/* persistent timeout in sec */
112 	__be32			netmask;	/* persistent netmask */
113 };
114 
115 
116 struct ip_vs_dest_user {
117 	/* destination server address */
118 	__be32			addr;
119 	__be16			port;
120 
121 	/* real server options */
122 	unsigned		conn_flags;	/* connection flags */
123 	int			weight;		/* destination weight */
124 
125 	/* thresholds for active connections */
126 	u_int32_t		u_threshold;	/* upper threshold */
127 	u_int32_t		l_threshold;	/* lower threshold */
128 };
129 
130 
131 /*
132  *	IPVS statistics object (for user space)
133  */
134 struct ip_vs_stats_user
135 {
136 	__u32                   conns;          /* connections scheduled */
137 	__u32                   inpkts;         /* incoming packets */
138 	__u32                   outpkts;        /* outgoing packets */
139 	__u64                   inbytes;        /* incoming bytes */
140 	__u64                   outbytes;       /* outgoing bytes */
141 
142 	__u32			cps;		/* current connection rate */
143 	__u32			inpps;		/* current in packet rate */
144 	__u32			outpps;		/* current out packet rate */
145 	__u32			inbps;		/* current in byte rate */
146 	__u32			outbps;		/* current out byte rate */
147 };
148 
149 
150 /* The argument to IP_VS_SO_GET_INFO */
151 struct ip_vs_getinfo {
152 	/* version number */
153 	unsigned int		version;
154 
155 	/* size of connection hash table */
156 	unsigned int		size;
157 
158 	/* number of virtual services */
159 	unsigned int		num_services;
160 };
161 
162 
163 /* The argument to IP_VS_SO_GET_SERVICE */
164 struct ip_vs_service_entry {
165 	/* which service: user fills in these */
166 	u_int16_t		protocol;
167 	__be32			addr;		/* virtual address */
168 	__be16			port;
169 	u_int32_t		fwmark;		/* firwall mark of service */
170 
171 	/* service options */
172 	char			sched_name[IP_VS_SCHEDNAME_MAXLEN];
173 	unsigned		flags;          /* virtual service flags */
174 	unsigned		timeout;	/* persistent timeout */
175 	__be32			netmask;	/* persistent netmask */
176 
177 	/* number of real servers */
178 	unsigned int		num_dests;
179 
180 	/* statistics */
181 	struct ip_vs_stats_user stats;
182 };
183 
184 
185 struct ip_vs_dest_entry {
186 	__be32			addr;		/* destination address */
187 	__be16			port;
188 	unsigned		conn_flags;	/* connection flags */
189 	int			weight;		/* destination weight */
190 
191 	u_int32_t		u_threshold;	/* upper threshold */
192 	u_int32_t		l_threshold;	/* lower threshold */
193 
194 	u_int32_t		activeconns;	/* active connections */
195 	u_int32_t		inactconns;	/* inactive connections */
196 	u_int32_t		persistconns;	/* persistent connections */
197 
198 	/* statistics */
199 	struct ip_vs_stats_user stats;
200 };
201 
202 
203 /* The argument to IP_VS_SO_GET_DESTS */
204 struct ip_vs_get_dests {
205 	/* which service: user fills in these */
206 	u_int16_t		protocol;
207 	__be32			addr;		/* virtual address */
208 	__be16			port;
209 	u_int32_t		fwmark;		/* firwall mark of service */
210 
211 	/* number of real servers */
212 	unsigned int		num_dests;
213 
214 	/* the real servers */
215 	struct ip_vs_dest_entry	entrytable[0];
216 };
217 
218 
219 /* The argument to IP_VS_SO_GET_SERVICES */
220 struct ip_vs_get_services {
221 	/* number of virtual services */
222 	unsigned int		num_services;
223 
224 	/* service table */
225 	struct ip_vs_service_entry entrytable[0];
226 };
227 
228 
229 /* The argument to IP_VS_SO_GET_TIMEOUT */
230 struct ip_vs_timeout_user {
231 	int			tcp_timeout;
232 	int			tcp_fin_timeout;
233 	int			udp_timeout;
234 };
235 
236 
237 /* The argument to IP_VS_SO_GET_DAEMON */
238 struct ip_vs_daemon_user {
239 	/* sync daemon state (master/backup) */
240 	int			state;
241 
242 	/* multicast interface name */
243 	char			mcast_ifn[IP_VS_IFNAME_MAXLEN];
244 
245 	/* SyncID we belong to */
246 	int			syncid;
247 };
248 
249 
250 #ifdef __KERNEL__
251 
252 #include <linux/list.h>                 /* for struct list_head */
253 #include <linux/spinlock.h>             /* for struct rwlock_t */
254 #include <asm/atomic.h>                 /* for struct atomic_t */
255 #include <linux/compiler.h>
256 #include <linux/timer.h>
257 
258 #include <net/checksum.h>
259 
260 #ifdef CONFIG_IP_VS_DEBUG
261 #include <linux/net.h>
262 
263 extern int ip_vs_get_debug_level(void);
264 #define IP_VS_DBG(level, msg...)			\
265     do {						\
266 	    if (level <= ip_vs_get_debug_level())	\
267 		    printk(KERN_DEBUG "IPVS: " msg);	\
268     } while (0)
269 #define IP_VS_DBG_RL(msg...)				\
270     do {						\
271 	    if (net_ratelimit())			\
272 		    printk(KERN_DEBUG "IPVS: " msg);	\
273     } while (0)
274 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		\
275     do {						\
276 	    if (level <= ip_vs_get_debug_level())	\
277 		pp->debug_packet(pp, skb, ofs, msg);	\
278     } while (0)
279 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	\
280     do {						\
281 	    if (level <= ip_vs_get_debug_level() &&	\
282 		net_ratelimit())			\
283 		pp->debug_packet(pp, skb, ofs, msg);	\
284     } while (0)
285 #else	/* NO DEBUGGING at ALL */
286 #define IP_VS_DBG(level, msg...)  do {} while (0)
287 #define IP_VS_DBG_RL(msg...)  do {} while (0)
288 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		do {} while (0)
289 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	do {} while (0)
290 #endif
291 
292 #define IP_VS_BUG() BUG()
293 #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg)
294 #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg)
295 #define IP_VS_WARNING(msg...) \
296 	printk(KERN_WARNING "IPVS: " msg)
297 #define IP_VS_ERR_RL(msg...)				\
298     do {						\
299 	    if (net_ratelimit())			\
300 		    printk(KERN_ERR "IPVS: " msg);	\
301     } while (0)
302 
303 #ifdef CONFIG_IP_VS_DEBUG
304 #define EnterFunction(level)						\
305     do {								\
306 	    if (level <= ip_vs_get_debug_level())			\
307 		    printk(KERN_DEBUG "Enter: %s, %s line %i\n",	\
308 			   __FUNCTION__, __FILE__, __LINE__);		\
309     } while (0)
310 #define LeaveFunction(level)                                            \
311     do {                                                                \
312 	    if (level <= ip_vs_get_debug_level())                       \
313 			printk(KERN_DEBUG "Leave: %s, %s line %i\n",    \
314 			       __FUNCTION__, __FILE__, __LINE__);       \
315     } while (0)
316 #else
317 #define EnterFunction(level)   do {} while (0)
318 #define LeaveFunction(level)   do {} while (0)
319 #endif
320 
321 #define	IP_VS_WAIT_WHILE(expr)	while (expr) { cpu_relax(); }
322 
323 
324 /*
325  *      The port number of FTP service (in network order).
326  */
327 #define FTPPORT  __constant_htons(21)
328 #define FTPDATA  __constant_htons(20)
329 
330 /*
331  *      TCP State Values
332  */
333 enum {
334 	IP_VS_TCP_S_NONE = 0,
335 	IP_VS_TCP_S_ESTABLISHED,
336 	IP_VS_TCP_S_SYN_SENT,
337 	IP_VS_TCP_S_SYN_RECV,
338 	IP_VS_TCP_S_FIN_WAIT,
339 	IP_VS_TCP_S_TIME_WAIT,
340 	IP_VS_TCP_S_CLOSE,
341 	IP_VS_TCP_S_CLOSE_WAIT,
342 	IP_VS_TCP_S_LAST_ACK,
343 	IP_VS_TCP_S_LISTEN,
344 	IP_VS_TCP_S_SYNACK,
345 	IP_VS_TCP_S_LAST
346 };
347 
348 /*
349  *	UDP State Values
350  */
351 enum {
352 	IP_VS_UDP_S_NORMAL,
353 	IP_VS_UDP_S_LAST,
354 };
355 
356 /*
357  *	ICMP State Values
358  */
359 enum {
360 	IP_VS_ICMP_S_NORMAL,
361 	IP_VS_ICMP_S_LAST,
362 };
363 
364 /*
365  *	Delta sequence info structure
366  *	Each ip_vs_conn has 2 (output AND input seq. changes).
367  *      Only used in the VS/NAT.
368  */
369 struct ip_vs_seq {
370 	__u32			init_seq;	/* Add delta from this seq */
371 	__u32			delta;		/* Delta in sequence numbers */
372 	__u32			previous_delta;	/* Delta in sequence numbers
373 						   before last resized pkt */
374 };
375 
376 
377 /*
378  *	IPVS statistics object
379  */
380 struct ip_vs_stats
381 {
382 	__u32                   conns;          /* connections scheduled */
383 	__u32                   inpkts;         /* incoming packets */
384 	__u32                   outpkts;        /* outgoing packets */
385 	__u64                   inbytes;        /* incoming bytes */
386 	__u64                   outbytes;       /* outgoing bytes */
387 
388 	__u32			cps;		/* current connection rate */
389 	__u32			inpps;		/* current in packet rate */
390 	__u32			outpps;		/* current out packet rate */
391 	__u32			inbps;		/* current in byte rate */
392 	__u32			outbps;		/* current out byte rate */
393 
394 	spinlock_t              lock;           /* spin lock */
395 };
396 
397 struct dst_entry;
398 struct iphdr;
399 struct ip_vs_conn;
400 struct ip_vs_app;
401 struct sk_buff;
402 
403 struct ip_vs_protocol {
404 	struct ip_vs_protocol	*next;
405 	char			*name;
406 	__u16			protocol;
407 	int			dont_defrag;
408 	atomic_t		appcnt;		/* counter of proto app incs */
409 	int			*timeout_table;	/* protocol timeout table */
410 
411 	void (*init)(struct ip_vs_protocol *pp);
412 
413 	void (*exit)(struct ip_vs_protocol *pp);
414 
415 	int (*conn_schedule)(struct sk_buff *skb,
416 			     struct ip_vs_protocol *pp,
417 			     int *verdict, struct ip_vs_conn **cpp);
418 
419 	struct ip_vs_conn *
420 	(*conn_in_get)(const struct sk_buff *skb,
421 		       struct ip_vs_protocol *pp,
422 		       const struct iphdr *iph,
423 		       unsigned int proto_off,
424 		       int inverse);
425 
426 	struct ip_vs_conn *
427 	(*conn_out_get)(const struct sk_buff *skb,
428 			struct ip_vs_protocol *pp,
429 			const struct iphdr *iph,
430 			unsigned int proto_off,
431 			int inverse);
432 
433 	int (*snat_handler)(struct sk_buff *skb,
434 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
435 
436 	int (*dnat_handler)(struct sk_buff *skb,
437 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
438 
439 	int (*csum_check)(struct sk_buff *skb, struct ip_vs_protocol *pp);
440 
441 	const char *(*state_name)(int state);
442 
443 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
444 				const struct sk_buff *skb,
445 				struct ip_vs_protocol *pp);
446 
447 	int (*register_app)(struct ip_vs_app *inc);
448 
449 	void (*unregister_app)(struct ip_vs_app *inc);
450 
451 	int (*app_conn_bind)(struct ip_vs_conn *cp);
452 
453 	void (*debug_packet)(struct ip_vs_protocol *pp,
454 			     const struct sk_buff *skb,
455 			     int offset,
456 			     const char *msg);
457 
458 	void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
459 
460 	int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
461 };
462 
463 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
464 
465 /*
466  *	IP_VS structure allocated for each dynamically scheduled connection
467  */
468 struct ip_vs_conn {
469 	struct list_head        c_list;         /* hashed list heads */
470 
471 	/* Protocol, addresses and port numbers */
472 	__be32                   caddr;          /* client address */
473 	__be32                   vaddr;          /* virtual address */
474 	__be32                   daddr;          /* destination address */
475 	__be16                   cport;
476 	__be16                   vport;
477 	__be16                   dport;
478 	__u16                   protocol;       /* Which protocol (TCP/UDP) */
479 
480 	/* counter and timer */
481 	atomic_t		refcnt;		/* reference count */
482 	struct timer_list	timer;		/* Expiration timer */
483 	volatile unsigned long	timeout;	/* timeout */
484 
485 	/* Flags and state transition */
486 	spinlock_t              lock;           /* lock for state transition */
487 	volatile __u16          flags;          /* status flags */
488 	volatile __u16          state;          /* state info */
489 	volatile __u16          old_state;      /* old state, to be used for
490 						 * state transition triggerd
491 						 * synchronization
492 						 */
493 
494 	/* Control members */
495 	struct ip_vs_conn       *control;       /* Master control connection */
496 	atomic_t                n_control;      /* Number of controlled ones */
497 	struct ip_vs_dest       *dest;          /* real server */
498 	atomic_t                in_pkts;        /* incoming packet counter */
499 
500 	/* packet transmitter for different forwarding methods.  If it
501 	   mangles the packet, it must return NF_DROP or better NF_STOLEN,
502 	   otherwise this must be changed to a sk_buff **.
503 	 */
504 	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
505 			   struct ip_vs_protocol *pp);
506 
507 	/* Note: we can group the following members into a structure,
508 	   in order to save more space, and the following members are
509 	   only used in VS/NAT anyway */
510 	struct ip_vs_app        *app;           /* bound ip_vs_app object */
511 	void                    *app_data;      /* Application private data */
512 	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
513 	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
514 };
515 
516 
517 /*
518  *	The information about the virtual service offered to the net
519  *	and the forwarding entries
520  */
521 struct ip_vs_service {
522 	struct list_head	s_list;   /* for normal service table */
523 	struct list_head	f_list;   /* for fwmark-based service table */
524 	atomic_t		refcnt;   /* reference counter */
525 	atomic_t		usecnt;   /* use counter */
526 
527 	__u16			protocol; /* which protocol (TCP/UDP) */
528 	__be32			addr;	  /* IP address for virtual service */
529 	__be16			port;	  /* port number for the service */
530 	__u32                   fwmark;   /* firewall mark of the service */
531 	unsigned		flags;	  /* service status flags */
532 	unsigned		timeout;  /* persistent timeout in ticks */
533 	__be32			netmask;  /* grouping granularity */
534 
535 	struct list_head	destinations;  /* real server d-linked list */
536 	__u32			num_dests;     /* number of servers */
537 	struct ip_vs_stats      stats;         /* statistics for the service */
538 	struct ip_vs_app	*inc;	  /* bind conns to this app inc */
539 
540 	/* for scheduling */
541 	struct ip_vs_scheduler	*scheduler;    /* bound scheduler object */
542 	rwlock_t		sched_lock;    /* lock sched_data */
543 	void			*sched_data;   /* scheduler application data */
544 };
545 
546 
547 /*
548  *	The real server destination forwarding entry
549  *	with ip address, port number, and so on.
550  */
551 struct ip_vs_dest {
552 	struct list_head	n_list;   /* for the dests in the service */
553 	struct list_head	d_list;   /* for table with all the dests */
554 
555 	__be32			addr;		/* IP address of the server */
556 	__be16			port;		/* port number of the server */
557 	volatile unsigned	flags;		/* dest status flags */
558 	atomic_t		conn_flags;	/* flags to copy to conn */
559 	atomic_t		weight;		/* server weight */
560 
561 	atomic_t		refcnt;		/* reference counter */
562 	struct ip_vs_stats      stats;          /* statistics */
563 
564 	/* connection counters and thresholds */
565 	atomic_t		activeconns;	/* active connections */
566 	atomic_t		inactconns;	/* inactive connections */
567 	atomic_t		persistconns;	/* persistent connections */
568 	__u32			u_threshold;	/* upper threshold */
569 	__u32			l_threshold;	/* lower threshold */
570 
571 	/* for destination cache */
572 	spinlock_t		dst_lock;	/* lock of dst_cache */
573 	struct dst_entry	*dst_cache;	/* destination cache entry */
574 	u32			dst_rtos;	/* RT_TOS(tos) for dst */
575 
576 	/* for virtual service */
577 	struct ip_vs_service	*svc;		/* service it belongs to */
578 	__u16			protocol;	/* which protocol (TCP/UDP) */
579 	__be32			vaddr;		/* virtual IP address */
580 	__be16			vport;		/* virtual port number */
581 	__u32			vfwmark;	/* firewall mark of service */
582 };
583 
584 
585 /*
586  *	The scheduler object
587  */
588 struct ip_vs_scheduler {
589 	struct list_head	n_list;		/* d-linked list head */
590 	char			*name;		/* scheduler name */
591 	atomic_t		refcnt;		/* reference counter */
592 	struct module		*module;	/* THIS_MODULE/NULL */
593 
594 	/* scheduler initializing service */
595 	int (*init_service)(struct ip_vs_service *svc);
596 	/* scheduling service finish */
597 	int (*done_service)(struct ip_vs_service *svc);
598 	/* scheduler updating service */
599 	int (*update_service)(struct ip_vs_service *svc);
600 
601 	/* selecting a server from the given service */
602 	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
603 				       const struct sk_buff *skb);
604 };
605 
606 
607 /*
608  *	The application module object (a.k.a. app incarnation)
609  */
610 struct ip_vs_app
611 {
612 	struct list_head	a_list;		/* member in app list */
613 	int			type;		/* IP_VS_APP_TYPE_xxx */
614 	char			*name;		/* application module name */
615 	__u16			protocol;
616 	struct module		*module;	/* THIS_MODULE/NULL */
617 	struct list_head	incs_list;	/* list of incarnations */
618 
619 	/* members for application incarnations */
620 	struct list_head	p_list;		/* member in proto app list */
621 	struct ip_vs_app	*app;		/* its real application */
622 	__be16			port;		/* port number in net order */
623 	atomic_t		usecnt;		/* usage counter */
624 
625 	/* output hook: return false if can't linearize. diff set for TCP.  */
626 	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
627 		       struct sk_buff *, int *diff);
628 
629 	/* input hook: return false if can't linearize. diff set for TCP. */
630 	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
631 		      struct sk_buff *, int *diff);
632 
633 	/* ip_vs_app initializer */
634 	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
635 
636 	/* ip_vs_app finish */
637 	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
638 
639 
640 	/* not used now */
641 	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
642 			 struct ip_vs_protocol *);
643 
644 	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
645 
646 	int *			timeout_table;
647 	int *			timeouts;
648 	int			timeouts_size;
649 
650 	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
651 			     int *verdict, struct ip_vs_conn **cpp);
652 
653 	struct ip_vs_conn *
654 	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
655 		       const struct iphdr *iph, unsigned int proto_off,
656 		       int inverse);
657 
658 	struct ip_vs_conn *
659 	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
660 			const struct iphdr *iph, unsigned int proto_off,
661 			int inverse);
662 
663 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
664 				const struct sk_buff *skb,
665 				struct ip_vs_app *app);
666 
667 	void (*timeout_change)(struct ip_vs_app *app, int flags);
668 };
669 
670 
671 /*
672  *      IPVS core functions
673  *      (from ip_vs_core.c)
674  */
675 extern const char *ip_vs_proto_name(unsigned proto);
676 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
677 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table(t, sizeof(t)/sizeof(t[0]))
678 
679 #define IP_VS_APP_TYPE_UNSPEC	0
680 #define IP_VS_APP_TYPE_FTP	1
681 
682 /*
683  *     ip_vs_conn handling functions
684  *     (from ip_vs_conn.c)
685  */
686 
687 /*
688  *     IPVS connection entry hash table
689  */
690 #ifndef CONFIG_IP_VS_TAB_BITS
691 #define CONFIG_IP_VS_TAB_BITS   12
692 #endif
693 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
694 #if CONFIG_IP_VS_TAB_BITS < 8
695 #define IP_VS_CONN_TAB_BITS	8
696 #endif
697 #if CONFIG_IP_VS_TAB_BITS > 20
698 #define IP_VS_CONN_TAB_BITS	20
699 #endif
700 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
701 #define IP_VS_CONN_TAB_BITS	CONFIG_IP_VS_TAB_BITS
702 #endif
703 #define IP_VS_CONN_TAB_SIZE     (1 << IP_VS_CONN_TAB_BITS)
704 #define IP_VS_CONN_TAB_MASK     (IP_VS_CONN_TAB_SIZE - 1)
705 
706 enum {
707 	IP_VS_DIR_INPUT = 0,
708 	IP_VS_DIR_OUTPUT,
709 	IP_VS_DIR_INPUT_ONLY,
710 	IP_VS_DIR_LAST,
711 };
712 
713 extern struct ip_vs_conn *ip_vs_conn_in_get
714 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
715 extern struct ip_vs_conn *ip_vs_ct_in_get
716 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
717 extern struct ip_vs_conn *ip_vs_conn_out_get
718 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
719 
720 /* put back the conn without restarting its timer */
721 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
722 {
723 	atomic_dec(&cp->refcnt);
724 }
725 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
726 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
727 
728 extern struct ip_vs_conn *
729 ip_vs_conn_new(int proto, __be32 caddr, __be16 cport, __be32 vaddr, __be16 vport,
730 	       __be32 daddr, __be16 dport, unsigned flags,
731 	       struct ip_vs_dest *dest);
732 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
733 
734 extern const char * ip_vs_state_name(__u16 proto, int state);
735 
736 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
737 extern int ip_vs_check_template(struct ip_vs_conn *ct);
738 extern void ip_vs_secure_tcp_set(int on);
739 extern void ip_vs_random_dropentry(void);
740 extern int ip_vs_conn_init(void);
741 extern void ip_vs_conn_cleanup(void);
742 
743 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
744 {
745 	struct ip_vs_conn *ctl_cp = cp->control;
746 	if (!ctl_cp) {
747 		IP_VS_ERR("request control DEL for uncontrolled: "
748 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
749 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
750 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
751 		return;
752 	}
753 
754 	IP_VS_DBG(7, "DELeting control for: "
755 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
756 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
757 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
758 
759 	cp->control = NULL;
760 	if (atomic_read(&ctl_cp->n_control) == 0) {
761 		IP_VS_ERR("BUG control DEL with n=0 : "
762 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
763 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
764 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
765 		return;
766 	}
767 	atomic_dec(&ctl_cp->n_control);
768 }
769 
770 static inline void
771 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
772 {
773 	if (cp->control) {
774 		IP_VS_ERR("request control ADD for already controlled: "
775 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
776 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
777 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
778 		ip_vs_control_del(cp);
779 	}
780 
781 	IP_VS_DBG(7, "ADDing control for: "
782 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
783 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
784 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
785 
786 	cp->control = ctl_cp;
787 	atomic_inc(&ctl_cp->n_control);
788 }
789 
790 
791 /*
792  *      IPVS application functions
793  *      (from ip_vs_app.c)
794  */
795 #define IP_VS_APP_MAX_PORTS  8
796 extern int register_ip_vs_app(struct ip_vs_app *app);
797 extern void unregister_ip_vs_app(struct ip_vs_app *app);
798 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
799 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
800 extern int
801 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
802 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
803 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
804 
805 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
806 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
807 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
808 			     char *o_buf, int o_len, char *n_buf, int n_len);
809 extern int ip_vs_app_init(void);
810 extern void ip_vs_app_cleanup(void);
811 
812 
813 /*
814  *	IPVS protocol functions (from ip_vs_proto.c)
815  */
816 extern int ip_vs_protocol_init(void);
817 extern void ip_vs_protocol_cleanup(void);
818 extern void ip_vs_protocol_timeout_change(int flags);
819 extern int *ip_vs_create_timeout_table(int *table, int size);
820 extern int
821 ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to);
822 extern void
823 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
824 			  int offset, const char *msg);
825 
826 extern struct ip_vs_protocol ip_vs_protocol_tcp;
827 extern struct ip_vs_protocol ip_vs_protocol_udp;
828 extern struct ip_vs_protocol ip_vs_protocol_icmp;
829 extern struct ip_vs_protocol ip_vs_protocol_esp;
830 extern struct ip_vs_protocol ip_vs_protocol_ah;
831 
832 
833 /*
834  *      Registering/unregistering scheduler functions
835  *      (from ip_vs_sched.c)
836  */
837 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
838 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
839 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
840 				struct ip_vs_scheduler *scheduler);
841 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
842 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
843 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
844 extern struct ip_vs_conn *
845 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
846 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
847 			struct ip_vs_protocol *pp);
848 
849 
850 /*
851  *      IPVS control data and functions (from ip_vs_ctl.c)
852  */
853 extern int sysctl_ip_vs_cache_bypass;
854 extern int sysctl_ip_vs_expire_nodest_conn;
855 extern int sysctl_ip_vs_expire_quiescent_template;
856 extern int sysctl_ip_vs_sync_threshold[2];
857 extern int sysctl_ip_vs_nat_icmp_send;
858 extern struct ip_vs_stats ip_vs_stats;
859 
860 extern struct ip_vs_service *
861 ip_vs_service_get(__u32 fwmark, __u16 protocol, __be32 vaddr, __be16 vport);
862 
863 static inline void ip_vs_service_put(struct ip_vs_service *svc)
864 {
865 	atomic_dec(&svc->usecnt);
866 }
867 
868 extern struct ip_vs_dest *
869 ip_vs_lookup_real_service(__u16 protocol, __be32 daddr, __be16 dport);
870 extern int ip_vs_use_count_inc(void);
871 extern void ip_vs_use_count_dec(void);
872 extern int ip_vs_control_init(void);
873 extern void ip_vs_control_cleanup(void);
874 extern struct ip_vs_dest *
875 ip_vs_find_dest(__be32 daddr, __be16 dport,
876 		 __be32 vaddr, __be16 vport, __u16 protocol);
877 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
878 
879 
880 /*
881  *      IPVS sync daemon data and function prototypes
882  *      (from ip_vs_sync.c)
883  */
884 extern volatile int ip_vs_sync_state;
885 extern volatile int ip_vs_master_syncid;
886 extern volatile int ip_vs_backup_syncid;
887 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
888 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
889 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
890 extern int stop_sync_thread(int state);
891 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
892 
893 
894 /*
895  *      IPVS rate estimator prototypes (from ip_vs_est.c)
896  */
897 extern int ip_vs_new_estimator(struct ip_vs_stats *stats);
898 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
899 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
900 
901 /*
902  *	Various IPVS packet transmitters (from ip_vs_xmit.c)
903  */
904 extern int ip_vs_null_xmit
905 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
906 extern int ip_vs_bypass_xmit
907 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
908 extern int ip_vs_nat_xmit
909 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
910 extern int ip_vs_tunnel_xmit
911 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
912 extern int ip_vs_dr_xmit
913 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
914 extern int ip_vs_icmp_xmit
915 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
916 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
917 
918 
919 /*
920  *	This is a simple mechanism to ignore packets when
921  *	we are loaded. Just set ip_vs_drop_rate to 'n' and
922  *	we start to drop 1/rate of the packets
923  */
924 extern int ip_vs_drop_rate;
925 extern int ip_vs_drop_counter;
926 
927 static __inline__ int ip_vs_todrop(void)
928 {
929 	if (!ip_vs_drop_rate) return 0;
930 	if (--ip_vs_drop_counter > 0) return 0;
931 	ip_vs_drop_counter = ip_vs_drop_rate;
932 	return 1;
933 }
934 
935 /*
936  *      ip_vs_fwd_tag returns the forwarding tag of the connection
937  */
938 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
939 
940 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
941 {
942 	char fwd;
943 
944 	switch (IP_VS_FWD_METHOD(cp)) {
945 	case IP_VS_CONN_F_MASQ:
946 		fwd = 'M'; break;
947 	case IP_VS_CONN_F_LOCALNODE:
948 		fwd = 'L'; break;
949 	case IP_VS_CONN_F_TUNNEL:
950 		fwd = 'T'; break;
951 	case IP_VS_CONN_F_DROUTE:
952 		fwd = 'R'; break;
953 	case IP_VS_CONN_F_BYPASS:
954 		fwd = 'B'; break;
955 	default:
956 		fwd = '?'; break;
957 	}
958 	return fwd;
959 }
960 
961 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
962 		struct ip_vs_conn *cp, int dir);
963 
964 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
965 
966 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
967 {
968 	__be32 diff[2] = { ~old, new };
969 
970 	return csum_partial((char *) diff, sizeof(diff), oldsum);
971 }
972 
973 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
974 {
975 	__be16 diff[2] = { ~old, new };
976 
977 	return csum_partial((char *) diff, sizeof(diff), oldsum);
978 }
979 
980 #endif /* __KERNEL__ */
981 
982 #endif	/* _IP_VS_H */
983