xref: /openbmc/linux/include/net/ip_vs.h (revision a1e58bbd)
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 #include <linux/sysctl.h>	/* For ctl_path */
13 
14 #define IP_VS_VERSION_CODE	0x010201
15 #define NVERSION(version)			\
16 	(version >> 16) & 0xFF,			\
17 	(version >> 8) & 0xFF,			\
18 	version & 0xFF
19 
20 /*
21  *      Virtual Service Flags
22  */
23 #define IP_VS_SVC_F_PERSISTENT	0x0001		/* persistent port */
24 #define IP_VS_SVC_F_HASHED	0x0002		/* hashed entry */
25 
26 /*
27  *      Destination Server Flags
28  */
29 #define IP_VS_DEST_F_AVAILABLE	0x0001		/* server is available */
30 #define IP_VS_DEST_F_OVERLOAD	0x0002		/* server is overloaded */
31 
32 /*
33  *      IPVS sync daemon states
34  */
35 #define IP_VS_STATE_NONE	0x0000		/* daemon is stopped */
36 #define IP_VS_STATE_MASTER	0x0001		/* started as master */
37 #define IP_VS_STATE_BACKUP	0x0002		/* started as backup */
38 
39 /*
40  *      IPVS socket options
41  */
42 #define IP_VS_BASE_CTL		(64+1024+64)		/* base */
43 
44 #define IP_VS_SO_SET_NONE	IP_VS_BASE_CTL		/* just peek */
45 #define IP_VS_SO_SET_INSERT	(IP_VS_BASE_CTL+1)
46 #define IP_VS_SO_SET_ADD	(IP_VS_BASE_CTL+2)
47 #define IP_VS_SO_SET_EDIT	(IP_VS_BASE_CTL+3)
48 #define IP_VS_SO_SET_DEL	(IP_VS_BASE_CTL+4)
49 #define IP_VS_SO_SET_FLUSH	(IP_VS_BASE_CTL+5)
50 #define IP_VS_SO_SET_LIST	(IP_VS_BASE_CTL+6)
51 #define IP_VS_SO_SET_ADDDEST	(IP_VS_BASE_CTL+7)
52 #define IP_VS_SO_SET_DELDEST	(IP_VS_BASE_CTL+8)
53 #define IP_VS_SO_SET_EDITDEST	(IP_VS_BASE_CTL+9)
54 #define IP_VS_SO_SET_TIMEOUT	(IP_VS_BASE_CTL+10)
55 #define IP_VS_SO_SET_STARTDAEMON (IP_VS_BASE_CTL+11)
56 #define IP_VS_SO_SET_STOPDAEMON (IP_VS_BASE_CTL+12)
57 #define IP_VS_SO_SET_RESTORE    (IP_VS_BASE_CTL+13)
58 #define IP_VS_SO_SET_SAVE       (IP_VS_BASE_CTL+14)
59 #define IP_VS_SO_SET_ZERO	(IP_VS_BASE_CTL+15)
60 #define IP_VS_SO_SET_MAX	IP_VS_SO_SET_ZERO
61 
62 #define IP_VS_SO_GET_VERSION	IP_VS_BASE_CTL
63 #define IP_VS_SO_GET_INFO	(IP_VS_BASE_CTL+1)
64 #define IP_VS_SO_GET_SERVICES	(IP_VS_BASE_CTL+2)
65 #define IP_VS_SO_GET_SERVICE	(IP_VS_BASE_CTL+3)
66 #define IP_VS_SO_GET_DESTS	(IP_VS_BASE_CTL+4)
67 #define IP_VS_SO_GET_DEST	(IP_VS_BASE_CTL+5)	/* not used now */
68 #define IP_VS_SO_GET_TIMEOUT	(IP_VS_BASE_CTL+6)
69 #define IP_VS_SO_GET_DAEMON	(IP_VS_BASE_CTL+7)
70 #define IP_VS_SO_GET_MAX	IP_VS_SO_GET_DAEMON
71 
72 
73 /*
74  *      IPVS Connection Flags
75  */
76 #define IP_VS_CONN_F_FWD_MASK	0x0007		/* mask for the fwd methods */
77 #define IP_VS_CONN_F_MASQ	0x0000		/* masquerading/NAT */
78 #define IP_VS_CONN_F_LOCALNODE	0x0001		/* local node */
79 #define IP_VS_CONN_F_TUNNEL	0x0002		/* tunneling */
80 #define IP_VS_CONN_F_DROUTE	0x0003		/* direct routing */
81 #define IP_VS_CONN_F_BYPASS	0x0004		/* cache bypass */
82 #define IP_VS_CONN_F_SYNC	0x0020		/* entry created by sync */
83 #define IP_VS_CONN_F_HASHED	0x0040		/* hashed entry */
84 #define IP_VS_CONN_F_NOOUTPUT	0x0080		/* no output packets */
85 #define IP_VS_CONN_F_INACTIVE	0x0100		/* not established */
86 #define IP_VS_CONN_F_OUT_SEQ	0x0200		/* must do output seq adjust */
87 #define IP_VS_CONN_F_IN_SEQ	0x0400		/* must do input seq adjust */
88 #define IP_VS_CONN_F_SEQ_MASK	0x0600		/* in/out sequence mask */
89 #define IP_VS_CONN_F_NO_CPORT	0x0800		/* no client port set yet */
90 #define IP_VS_CONN_F_TEMPLATE	0x1000		/* template, not connection */
91 
92 /* Move it to better place one day, for now keep it unique */
93 #define NFC_IPVS_PROPERTY	0x10000
94 
95 #define IP_VS_SCHEDNAME_MAXLEN	16
96 #define IP_VS_IFNAME_MAXLEN	16
97 
98 
99 /*
100  *	The struct ip_vs_service_user and struct ip_vs_dest_user are
101  *	used to set IPVS rules through setsockopt.
102  */
103 struct ip_vs_service_user {
104 	/* virtual service addresses */
105 	u_int16_t		protocol;
106 	__be32			addr;		/* virtual ip address */
107 	__be16			port;
108 	u_int32_t		fwmark;		/* firwall mark of service */
109 
110 	/* virtual service options */
111 	char			sched_name[IP_VS_SCHEDNAME_MAXLEN];
112 	unsigned		flags;		/* virtual service flags */
113 	unsigned		timeout;	/* persistent timeout in sec */
114 	__be32			netmask;	/* persistent netmask */
115 };
116 
117 
118 struct ip_vs_dest_user {
119 	/* destination server address */
120 	__be32			addr;
121 	__be16			port;
122 
123 	/* real server options */
124 	unsigned		conn_flags;	/* connection flags */
125 	int			weight;		/* destination weight */
126 
127 	/* thresholds for active connections */
128 	u_int32_t		u_threshold;	/* upper threshold */
129 	u_int32_t		l_threshold;	/* lower threshold */
130 };
131 
132 
133 /*
134  *	IPVS statistics object (for user space)
135  */
136 struct ip_vs_stats_user
137 {
138 	__u32                   conns;          /* connections scheduled */
139 	__u32                   inpkts;         /* incoming packets */
140 	__u32                   outpkts;        /* outgoing packets */
141 	__u64                   inbytes;        /* incoming bytes */
142 	__u64                   outbytes;       /* outgoing bytes */
143 
144 	__u32			cps;		/* current connection rate */
145 	__u32			inpps;		/* current in packet rate */
146 	__u32			outpps;		/* current out packet rate */
147 	__u32			inbps;		/* current in byte rate */
148 	__u32			outbps;		/* current out byte rate */
149 };
150 
151 
152 /* The argument to IP_VS_SO_GET_INFO */
153 struct ip_vs_getinfo {
154 	/* version number */
155 	unsigned int		version;
156 
157 	/* size of connection hash table */
158 	unsigned int		size;
159 
160 	/* number of virtual services */
161 	unsigned int		num_services;
162 };
163 
164 
165 /* The argument to IP_VS_SO_GET_SERVICE */
166 struct ip_vs_service_entry {
167 	/* which service: user fills in these */
168 	u_int16_t		protocol;
169 	__be32			addr;		/* virtual address */
170 	__be16			port;
171 	u_int32_t		fwmark;		/* firwall mark of service */
172 
173 	/* service options */
174 	char			sched_name[IP_VS_SCHEDNAME_MAXLEN];
175 	unsigned		flags;          /* virtual service flags */
176 	unsigned		timeout;	/* persistent timeout */
177 	__be32			netmask;	/* persistent netmask */
178 
179 	/* number of real servers */
180 	unsigned int		num_dests;
181 
182 	/* statistics */
183 	struct ip_vs_stats_user stats;
184 };
185 
186 
187 struct ip_vs_dest_entry {
188 	__be32			addr;		/* destination address */
189 	__be16			port;
190 	unsigned		conn_flags;	/* connection flags */
191 	int			weight;		/* destination weight */
192 
193 	u_int32_t		u_threshold;	/* upper threshold */
194 	u_int32_t		l_threshold;	/* lower threshold */
195 
196 	u_int32_t		activeconns;	/* active connections */
197 	u_int32_t		inactconns;	/* inactive connections */
198 	u_int32_t		persistconns;	/* persistent connections */
199 
200 	/* statistics */
201 	struct ip_vs_stats_user stats;
202 };
203 
204 
205 /* The argument to IP_VS_SO_GET_DESTS */
206 struct ip_vs_get_dests {
207 	/* which service: user fills in these */
208 	u_int16_t		protocol;
209 	__be32			addr;		/* virtual address */
210 	__be16			port;
211 	u_int32_t		fwmark;		/* firwall mark of service */
212 
213 	/* number of real servers */
214 	unsigned int		num_dests;
215 
216 	/* the real servers */
217 	struct ip_vs_dest_entry	entrytable[0];
218 };
219 
220 
221 /* The argument to IP_VS_SO_GET_SERVICES */
222 struct ip_vs_get_services {
223 	/* number of virtual services */
224 	unsigned int		num_services;
225 
226 	/* service table */
227 	struct ip_vs_service_entry entrytable[0];
228 };
229 
230 
231 /* The argument to IP_VS_SO_GET_TIMEOUT */
232 struct ip_vs_timeout_user {
233 	int			tcp_timeout;
234 	int			tcp_fin_timeout;
235 	int			udp_timeout;
236 };
237 
238 
239 /* The argument to IP_VS_SO_GET_DAEMON */
240 struct ip_vs_daemon_user {
241 	/* sync daemon state (master/backup) */
242 	int			state;
243 
244 	/* multicast interface name */
245 	char			mcast_ifn[IP_VS_IFNAME_MAXLEN];
246 
247 	/* SyncID we belong to */
248 	int			syncid;
249 };
250 
251 
252 #ifdef __KERNEL__
253 
254 #include <linux/list.h>                 /* for struct list_head */
255 #include <linux/spinlock.h>             /* for struct rwlock_t */
256 #include <asm/atomic.h>                 /* for struct atomic_t */
257 #include <linux/compiler.h>
258 #include <linux/timer.h>
259 
260 #include <net/checksum.h>
261 
262 #ifdef CONFIG_IP_VS_DEBUG
263 #include <linux/net.h>
264 
265 extern int ip_vs_get_debug_level(void);
266 #define IP_VS_DBG(level, msg...)			\
267     do {						\
268 	    if (level <= ip_vs_get_debug_level())	\
269 		    printk(KERN_DEBUG "IPVS: " msg);	\
270     } while (0)
271 #define IP_VS_DBG_RL(msg...)				\
272     do {						\
273 	    if (net_ratelimit())			\
274 		    printk(KERN_DEBUG "IPVS: " msg);	\
275     } while (0)
276 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		\
277     do {						\
278 	    if (level <= ip_vs_get_debug_level())	\
279 		pp->debug_packet(pp, skb, ofs, msg);	\
280     } while (0)
281 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	\
282     do {						\
283 	    if (level <= ip_vs_get_debug_level() &&	\
284 		net_ratelimit())			\
285 		pp->debug_packet(pp, skb, ofs, msg);	\
286     } while (0)
287 #else	/* NO DEBUGGING at ALL */
288 #define IP_VS_DBG(level, msg...)  do {} while (0)
289 #define IP_VS_DBG_RL(msg...)  do {} while (0)
290 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		do {} while (0)
291 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	do {} while (0)
292 #endif
293 
294 #define IP_VS_BUG() BUG()
295 #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg)
296 #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg)
297 #define IP_VS_WARNING(msg...) \
298 	printk(KERN_WARNING "IPVS: " msg)
299 #define IP_VS_ERR_RL(msg...)				\
300     do {						\
301 	    if (net_ratelimit())			\
302 		    printk(KERN_ERR "IPVS: " msg);	\
303     } while (0)
304 
305 #ifdef CONFIG_IP_VS_DEBUG
306 #define EnterFunction(level)						\
307     do {								\
308 	    if (level <= ip_vs_get_debug_level())			\
309 		    printk(KERN_DEBUG "Enter: %s, %s line %i\n",	\
310 			   __FUNCTION__, __FILE__, __LINE__);		\
311     } while (0)
312 #define LeaveFunction(level)                                            \
313     do {                                                                \
314 	    if (level <= ip_vs_get_debug_level())                       \
315 			printk(KERN_DEBUG "Leave: %s, %s line %i\n",    \
316 			       __FUNCTION__, __FILE__, __LINE__);       \
317     } while (0)
318 #else
319 #define EnterFunction(level)   do {} while (0)
320 #define LeaveFunction(level)   do {} while (0)
321 #endif
322 
323 #define	IP_VS_WAIT_WHILE(expr)	while (expr) { cpu_relax(); }
324 
325 
326 /*
327  *      The port number of FTP service (in network order).
328  */
329 #define FTPPORT  __constant_htons(21)
330 #define FTPDATA  __constant_htons(20)
331 
332 /*
333  *      TCP State Values
334  */
335 enum {
336 	IP_VS_TCP_S_NONE = 0,
337 	IP_VS_TCP_S_ESTABLISHED,
338 	IP_VS_TCP_S_SYN_SENT,
339 	IP_VS_TCP_S_SYN_RECV,
340 	IP_VS_TCP_S_FIN_WAIT,
341 	IP_VS_TCP_S_TIME_WAIT,
342 	IP_VS_TCP_S_CLOSE,
343 	IP_VS_TCP_S_CLOSE_WAIT,
344 	IP_VS_TCP_S_LAST_ACK,
345 	IP_VS_TCP_S_LISTEN,
346 	IP_VS_TCP_S_SYNACK,
347 	IP_VS_TCP_S_LAST
348 };
349 
350 /*
351  *	UDP State Values
352  */
353 enum {
354 	IP_VS_UDP_S_NORMAL,
355 	IP_VS_UDP_S_LAST,
356 };
357 
358 /*
359  *	ICMP State Values
360  */
361 enum {
362 	IP_VS_ICMP_S_NORMAL,
363 	IP_VS_ICMP_S_LAST,
364 };
365 
366 /*
367  *	Delta sequence info structure
368  *	Each ip_vs_conn has 2 (output AND input seq. changes).
369  *      Only used in the VS/NAT.
370  */
371 struct ip_vs_seq {
372 	__u32			init_seq;	/* Add delta from this seq */
373 	__u32			delta;		/* Delta in sequence numbers */
374 	__u32			previous_delta;	/* Delta in sequence numbers
375 						   before last resized pkt */
376 };
377 
378 
379 /*
380  *	IPVS statistics object
381  */
382 struct ip_vs_stats
383 {
384 	__u32                   conns;          /* connections scheduled */
385 	__u32                   inpkts;         /* incoming packets */
386 	__u32                   outpkts;        /* outgoing packets */
387 	__u64                   inbytes;        /* incoming bytes */
388 	__u64                   outbytes;       /* outgoing bytes */
389 
390 	__u32			cps;		/* current connection rate */
391 	__u32			inpps;		/* current in packet rate */
392 	__u32			outpps;		/* current out packet rate */
393 	__u32			inbps;		/* current in byte rate */
394 	__u32			outbps;		/* current out byte rate */
395 
396 	spinlock_t              lock;           /* spin lock */
397 };
398 
399 struct dst_entry;
400 struct iphdr;
401 struct ip_vs_conn;
402 struct ip_vs_app;
403 struct sk_buff;
404 
405 struct ip_vs_protocol {
406 	struct ip_vs_protocol	*next;
407 	char			*name;
408 	__u16			protocol;
409 	int			dont_defrag;
410 	atomic_t		appcnt;		/* counter of proto app incs */
411 	int			*timeout_table;	/* protocol timeout table */
412 
413 	void (*init)(struct ip_vs_protocol *pp);
414 
415 	void (*exit)(struct ip_vs_protocol *pp);
416 
417 	int (*conn_schedule)(struct sk_buff *skb,
418 			     struct ip_vs_protocol *pp,
419 			     int *verdict, struct ip_vs_conn **cpp);
420 
421 	struct ip_vs_conn *
422 	(*conn_in_get)(const struct sk_buff *skb,
423 		       struct ip_vs_protocol *pp,
424 		       const struct iphdr *iph,
425 		       unsigned int proto_off,
426 		       int inverse);
427 
428 	struct ip_vs_conn *
429 	(*conn_out_get)(const struct sk_buff *skb,
430 			struct ip_vs_protocol *pp,
431 			const struct iphdr *iph,
432 			unsigned int proto_off,
433 			int inverse);
434 
435 	int (*snat_handler)(struct sk_buff *skb,
436 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
437 
438 	int (*dnat_handler)(struct sk_buff *skb,
439 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
440 
441 	int (*csum_check)(struct sk_buff *skb, struct ip_vs_protocol *pp);
442 
443 	const char *(*state_name)(int state);
444 
445 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
446 				const struct sk_buff *skb,
447 				struct ip_vs_protocol *pp);
448 
449 	int (*register_app)(struct ip_vs_app *inc);
450 
451 	void (*unregister_app)(struct ip_vs_app *inc);
452 
453 	int (*app_conn_bind)(struct ip_vs_conn *cp);
454 
455 	void (*debug_packet)(struct ip_vs_protocol *pp,
456 			     const struct sk_buff *skb,
457 			     int offset,
458 			     const char *msg);
459 
460 	void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
461 
462 	int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
463 };
464 
465 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
466 
467 /*
468  *	IP_VS structure allocated for each dynamically scheduled connection
469  */
470 struct ip_vs_conn {
471 	struct list_head        c_list;         /* hashed list heads */
472 
473 	/* Protocol, addresses and port numbers */
474 	__be32                   caddr;          /* client address */
475 	__be32                   vaddr;          /* virtual address */
476 	__be32                   daddr;          /* destination address */
477 	__be16                   cport;
478 	__be16                   vport;
479 	__be16                   dport;
480 	__u16                   protocol;       /* Which protocol (TCP/UDP) */
481 
482 	/* counter and timer */
483 	atomic_t		refcnt;		/* reference count */
484 	struct timer_list	timer;		/* Expiration timer */
485 	volatile unsigned long	timeout;	/* timeout */
486 
487 	/* Flags and state transition */
488 	spinlock_t              lock;           /* lock for state transition */
489 	volatile __u16          flags;          /* status flags */
490 	volatile __u16          state;          /* state info */
491 	volatile __u16          old_state;      /* old state, to be used for
492 						 * state transition triggerd
493 						 * synchronization
494 						 */
495 
496 	/* Control members */
497 	struct ip_vs_conn       *control;       /* Master control connection */
498 	atomic_t                n_control;      /* Number of controlled ones */
499 	struct ip_vs_dest       *dest;          /* real server */
500 	atomic_t                in_pkts;        /* incoming packet counter */
501 
502 	/* packet transmitter for different forwarding methods.  If it
503 	   mangles the packet, it must return NF_DROP or better NF_STOLEN,
504 	   otherwise this must be changed to a sk_buff **.
505 	 */
506 	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
507 			   struct ip_vs_protocol *pp);
508 
509 	/* Note: we can group the following members into a structure,
510 	   in order to save more space, and the following members are
511 	   only used in VS/NAT anyway */
512 	struct ip_vs_app        *app;           /* bound ip_vs_app object */
513 	void                    *app_data;      /* Application private data */
514 	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
515 	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
516 };
517 
518 
519 /*
520  *	The information about the virtual service offered to the net
521  *	and the forwarding entries
522  */
523 struct ip_vs_service {
524 	struct list_head	s_list;   /* for normal service table */
525 	struct list_head	f_list;   /* for fwmark-based service table */
526 	atomic_t		refcnt;   /* reference counter */
527 	atomic_t		usecnt;   /* use counter */
528 
529 	__u16			protocol; /* which protocol (TCP/UDP) */
530 	__be32			addr;	  /* IP address for virtual service */
531 	__be16			port;	  /* port number for the service */
532 	__u32                   fwmark;   /* firewall mark of the service */
533 	unsigned		flags;	  /* service status flags */
534 	unsigned		timeout;  /* persistent timeout in ticks */
535 	__be32			netmask;  /* grouping granularity */
536 
537 	struct list_head	destinations;  /* real server d-linked list */
538 	__u32			num_dests;     /* number of servers */
539 	struct ip_vs_stats      stats;         /* statistics for the service */
540 	struct ip_vs_app	*inc;	  /* bind conns to this app inc */
541 
542 	/* for scheduling */
543 	struct ip_vs_scheduler	*scheduler;    /* bound scheduler object */
544 	rwlock_t		sched_lock;    /* lock sched_data */
545 	void			*sched_data;   /* scheduler application data */
546 };
547 
548 
549 /*
550  *	The real server destination forwarding entry
551  *	with ip address, port number, and so on.
552  */
553 struct ip_vs_dest {
554 	struct list_head	n_list;   /* for the dests in the service */
555 	struct list_head	d_list;   /* for table with all the dests */
556 
557 	__be32			addr;		/* IP address of the server */
558 	__be16			port;		/* port number of the server */
559 	volatile unsigned	flags;		/* dest status flags */
560 	atomic_t		conn_flags;	/* flags to copy to conn */
561 	atomic_t		weight;		/* server weight */
562 
563 	atomic_t		refcnt;		/* reference counter */
564 	struct ip_vs_stats      stats;          /* statistics */
565 
566 	/* connection counters and thresholds */
567 	atomic_t		activeconns;	/* active connections */
568 	atomic_t		inactconns;	/* inactive connections */
569 	atomic_t		persistconns;	/* persistent connections */
570 	__u32			u_threshold;	/* upper threshold */
571 	__u32			l_threshold;	/* lower threshold */
572 
573 	/* for destination cache */
574 	spinlock_t		dst_lock;	/* lock of dst_cache */
575 	struct dst_entry	*dst_cache;	/* destination cache entry */
576 	u32			dst_rtos;	/* RT_TOS(tos) for dst */
577 
578 	/* for virtual service */
579 	struct ip_vs_service	*svc;		/* service it belongs to */
580 	__u16			protocol;	/* which protocol (TCP/UDP) */
581 	__be32			vaddr;		/* virtual IP address */
582 	__be16			vport;		/* virtual port number */
583 	__u32			vfwmark;	/* firewall mark of service */
584 };
585 
586 
587 /*
588  *	The scheduler object
589  */
590 struct ip_vs_scheduler {
591 	struct list_head	n_list;		/* d-linked list head */
592 	char			*name;		/* scheduler name */
593 	atomic_t		refcnt;		/* reference counter */
594 	struct module		*module;	/* THIS_MODULE/NULL */
595 
596 	/* scheduler initializing service */
597 	int (*init_service)(struct ip_vs_service *svc);
598 	/* scheduling service finish */
599 	int (*done_service)(struct ip_vs_service *svc);
600 	/* scheduler updating service */
601 	int (*update_service)(struct ip_vs_service *svc);
602 
603 	/* selecting a server from the given service */
604 	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
605 				       const struct sk_buff *skb);
606 };
607 
608 
609 /*
610  *	The application module object (a.k.a. app incarnation)
611  */
612 struct ip_vs_app
613 {
614 	struct list_head	a_list;		/* member in app list */
615 	int			type;		/* IP_VS_APP_TYPE_xxx */
616 	char			*name;		/* application module name */
617 	__u16			protocol;
618 	struct module		*module;	/* THIS_MODULE/NULL */
619 	struct list_head	incs_list;	/* list of incarnations */
620 
621 	/* members for application incarnations */
622 	struct list_head	p_list;		/* member in proto app list */
623 	struct ip_vs_app	*app;		/* its real application */
624 	__be16			port;		/* port number in net order */
625 	atomic_t		usecnt;		/* usage counter */
626 
627 	/* output hook: return false if can't linearize. diff set for TCP.  */
628 	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
629 		       struct sk_buff *, int *diff);
630 
631 	/* input hook: return false if can't linearize. diff set for TCP. */
632 	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
633 		      struct sk_buff *, int *diff);
634 
635 	/* ip_vs_app initializer */
636 	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
637 
638 	/* ip_vs_app finish */
639 	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
640 
641 
642 	/* not used now */
643 	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
644 			 struct ip_vs_protocol *);
645 
646 	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
647 
648 	int *			timeout_table;
649 	int *			timeouts;
650 	int			timeouts_size;
651 
652 	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
653 			     int *verdict, struct ip_vs_conn **cpp);
654 
655 	struct ip_vs_conn *
656 	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
657 		       const struct iphdr *iph, unsigned int proto_off,
658 		       int inverse);
659 
660 	struct ip_vs_conn *
661 	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
662 			const struct iphdr *iph, unsigned int proto_off,
663 			int inverse);
664 
665 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
666 				const struct sk_buff *skb,
667 				struct ip_vs_app *app);
668 
669 	void (*timeout_change)(struct ip_vs_app *app, int flags);
670 };
671 
672 
673 /*
674  *      IPVS core functions
675  *      (from ip_vs_core.c)
676  */
677 extern const char *ip_vs_proto_name(unsigned proto);
678 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
679 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table(t, sizeof(t)/sizeof(t[0]))
680 
681 #define IP_VS_APP_TYPE_FTP	1
682 
683 /*
684  *     ip_vs_conn handling functions
685  *     (from ip_vs_conn.c)
686  */
687 
688 /*
689  *     IPVS connection entry hash table
690  */
691 #ifndef CONFIG_IP_VS_TAB_BITS
692 #define CONFIG_IP_VS_TAB_BITS   12
693 #endif
694 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
695 #if CONFIG_IP_VS_TAB_BITS < 8
696 #define IP_VS_CONN_TAB_BITS	8
697 #endif
698 #if CONFIG_IP_VS_TAB_BITS > 20
699 #define IP_VS_CONN_TAB_BITS	20
700 #endif
701 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
702 #define IP_VS_CONN_TAB_BITS	CONFIG_IP_VS_TAB_BITS
703 #endif
704 #define IP_VS_CONN_TAB_SIZE     (1 << IP_VS_CONN_TAB_BITS)
705 #define IP_VS_CONN_TAB_MASK     (IP_VS_CONN_TAB_SIZE - 1)
706 
707 enum {
708 	IP_VS_DIR_INPUT = 0,
709 	IP_VS_DIR_OUTPUT,
710 	IP_VS_DIR_INPUT_ONLY,
711 	IP_VS_DIR_LAST,
712 };
713 
714 extern struct ip_vs_conn *ip_vs_conn_in_get
715 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
716 extern struct ip_vs_conn *ip_vs_ct_in_get
717 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
718 extern struct ip_vs_conn *ip_vs_conn_out_get
719 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
720 
721 /* put back the conn without restarting its timer */
722 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
723 {
724 	atomic_dec(&cp->refcnt);
725 }
726 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
727 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
728 
729 extern struct ip_vs_conn *
730 ip_vs_conn_new(int proto, __be32 caddr, __be16 cport, __be32 vaddr, __be16 vport,
731 	       __be32 daddr, __be16 dport, unsigned flags,
732 	       struct ip_vs_dest *dest);
733 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
734 
735 extern const char * ip_vs_state_name(__u16 proto, int state);
736 
737 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
738 extern int ip_vs_check_template(struct ip_vs_conn *ct);
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 extern struct ctl_path net_vs_ctl_path[];
860 
861 extern struct ip_vs_service *
862 ip_vs_service_get(__u32 fwmark, __u16 protocol, __be32 vaddr, __be16 vport);
863 
864 static inline void ip_vs_service_put(struct ip_vs_service *svc)
865 {
866 	atomic_dec(&svc->usecnt);
867 }
868 
869 extern struct ip_vs_dest *
870 ip_vs_lookup_real_service(__u16 protocol, __be32 daddr, __be16 dport);
871 extern int ip_vs_use_count_inc(void);
872 extern void ip_vs_use_count_dec(void);
873 extern int ip_vs_control_init(void);
874 extern void ip_vs_control_cleanup(void);
875 extern struct ip_vs_dest *
876 ip_vs_find_dest(__be32 daddr, __be16 dport,
877 		 __be32 vaddr, __be16 vport, __u16 protocol);
878 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
879 
880 
881 /*
882  *      IPVS sync daemon data and function prototypes
883  *      (from ip_vs_sync.c)
884  */
885 extern volatile int ip_vs_sync_state;
886 extern volatile int ip_vs_master_syncid;
887 extern volatile int ip_vs_backup_syncid;
888 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
889 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
890 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
891 extern int stop_sync_thread(int state);
892 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
893 
894 
895 /*
896  *      IPVS rate estimator prototypes (from ip_vs_est.c)
897  */
898 extern int ip_vs_new_estimator(struct ip_vs_stats *stats);
899 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
900 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
901 
902 /*
903  *	Various IPVS packet transmitters (from ip_vs_xmit.c)
904  */
905 extern int ip_vs_null_xmit
906 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
907 extern int ip_vs_bypass_xmit
908 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
909 extern int ip_vs_nat_xmit
910 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
911 extern int ip_vs_tunnel_xmit
912 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
913 extern int ip_vs_dr_xmit
914 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
915 extern int ip_vs_icmp_xmit
916 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
917 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
918 
919 
920 /*
921  *	This is a simple mechanism to ignore packets when
922  *	we are loaded. Just set ip_vs_drop_rate to 'n' and
923  *	we start to drop 1/rate of the packets
924  */
925 extern int ip_vs_drop_rate;
926 extern int ip_vs_drop_counter;
927 
928 static __inline__ int ip_vs_todrop(void)
929 {
930 	if (!ip_vs_drop_rate) return 0;
931 	if (--ip_vs_drop_counter > 0) return 0;
932 	ip_vs_drop_counter = ip_vs_drop_rate;
933 	return 1;
934 }
935 
936 /*
937  *      ip_vs_fwd_tag returns the forwarding tag of the connection
938  */
939 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
940 
941 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
942 {
943 	char fwd;
944 
945 	switch (IP_VS_FWD_METHOD(cp)) {
946 	case IP_VS_CONN_F_MASQ:
947 		fwd = 'M'; break;
948 	case IP_VS_CONN_F_LOCALNODE:
949 		fwd = 'L'; break;
950 	case IP_VS_CONN_F_TUNNEL:
951 		fwd = 'T'; break;
952 	case IP_VS_CONN_F_DROUTE:
953 		fwd = 'R'; break;
954 	case IP_VS_CONN_F_BYPASS:
955 		fwd = 'B'; break;
956 	default:
957 		fwd = '?'; break;
958 	}
959 	return fwd;
960 }
961 
962 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
963 		struct ip_vs_conn *cp, int dir);
964 
965 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
966 
967 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
968 {
969 	__be32 diff[2] = { ~old, new };
970 
971 	return csum_partial((char *) diff, sizeof(diff), oldsum);
972 }
973 
974 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
975 {
976 	__be16 diff[2] = { ~old, new };
977 
978 	return csum_partial((char *) diff, sizeof(diff), oldsum);
979 }
980 
981 #endif /* __KERNEL__ */
982 
983 #endif	/* _IP_VS_H */
984