xref: /linux/include/net/ip_vs.h (revision 13abf8130139c2ccd4962a7e5a8902be5e6cb5a7)
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 
11 #define IP_VS_VERSION_CODE	0x010201
12 #define NVERSION(version)			\
13 	(version >> 16) & 0xFF,			\
14 	(version >> 8) & 0xFF,			\
15 	version & 0xFF
16 
17 /*
18  *      Virtual Service Flags
19  */
20 #define IP_VS_SVC_F_PERSISTENT	0x0001		/* persistent port */
21 #define IP_VS_SVC_F_HASHED	0x0002		/* hashed entry */
22 
23 /*
24  *      Destination Server Flags
25  */
26 #define IP_VS_DEST_F_AVAILABLE	0x0001		/* server is available */
27 #define IP_VS_DEST_F_OVERLOAD	0x0002		/* server is overloaded */
28 
29 /*
30  *      IPVS sync daemon states
31  */
32 #define IP_VS_STATE_NONE	0x0000		/* daemon is stopped */
33 #define IP_VS_STATE_MASTER	0x0001		/* started as master */
34 #define IP_VS_STATE_BACKUP	0x0002		/* started as backup */
35 
36 /*
37  *      IPVS socket options
38  */
39 #define IP_VS_BASE_CTL		(64+1024+64)		/* base */
40 
41 #define IP_VS_SO_SET_NONE	IP_VS_BASE_CTL		/* just peek */
42 #define IP_VS_SO_SET_INSERT	(IP_VS_BASE_CTL+1)
43 #define IP_VS_SO_SET_ADD	(IP_VS_BASE_CTL+2)
44 #define IP_VS_SO_SET_EDIT	(IP_VS_BASE_CTL+3)
45 #define IP_VS_SO_SET_DEL	(IP_VS_BASE_CTL+4)
46 #define IP_VS_SO_SET_FLUSH	(IP_VS_BASE_CTL+5)
47 #define IP_VS_SO_SET_LIST	(IP_VS_BASE_CTL+6)
48 #define IP_VS_SO_SET_ADDDEST	(IP_VS_BASE_CTL+7)
49 #define IP_VS_SO_SET_DELDEST	(IP_VS_BASE_CTL+8)
50 #define IP_VS_SO_SET_EDITDEST	(IP_VS_BASE_CTL+9)
51 #define IP_VS_SO_SET_TIMEOUT	(IP_VS_BASE_CTL+10)
52 #define IP_VS_SO_SET_STARTDAEMON (IP_VS_BASE_CTL+11)
53 #define IP_VS_SO_SET_STOPDAEMON (IP_VS_BASE_CTL+12)
54 #define IP_VS_SO_SET_RESTORE    (IP_VS_BASE_CTL+13)
55 #define IP_VS_SO_SET_SAVE       (IP_VS_BASE_CTL+14)
56 #define IP_VS_SO_SET_ZERO	(IP_VS_BASE_CTL+15)
57 #define IP_VS_SO_SET_MAX	IP_VS_SO_SET_ZERO
58 
59 #define IP_VS_SO_GET_VERSION	IP_VS_BASE_CTL
60 #define IP_VS_SO_GET_INFO	(IP_VS_BASE_CTL+1)
61 #define IP_VS_SO_GET_SERVICES	(IP_VS_BASE_CTL+2)
62 #define IP_VS_SO_GET_SERVICE	(IP_VS_BASE_CTL+3)
63 #define IP_VS_SO_GET_DESTS	(IP_VS_BASE_CTL+4)
64 #define IP_VS_SO_GET_DEST	(IP_VS_BASE_CTL+5)	/* not used now */
65 #define IP_VS_SO_GET_TIMEOUT	(IP_VS_BASE_CTL+6)
66 #define IP_VS_SO_GET_DAEMON	(IP_VS_BASE_CTL+7)
67 #define IP_VS_SO_GET_MAX	IP_VS_SO_GET_DAEMON
68 
69 
70 /*
71  *      IPVS Connection Flags
72  */
73 #define IP_VS_CONN_F_FWD_MASK	0x0007		/* mask for the fwd methods */
74 #define IP_VS_CONN_F_MASQ	0x0000		/* masquerading/NAT */
75 #define IP_VS_CONN_F_LOCALNODE	0x0001		/* local node */
76 #define IP_VS_CONN_F_TUNNEL	0x0002		/* tunneling */
77 #define IP_VS_CONN_F_DROUTE	0x0003		/* direct routing */
78 #define IP_VS_CONN_F_BYPASS	0x0004		/* cache bypass */
79 #define IP_VS_CONN_F_SYNC	0x0020		/* entry created by sync */
80 #define IP_VS_CONN_F_HASHED	0x0040		/* hashed entry */
81 #define IP_VS_CONN_F_NOOUTPUT	0x0080		/* no output packets */
82 #define IP_VS_CONN_F_INACTIVE	0x0100		/* not established */
83 #define IP_VS_CONN_F_OUT_SEQ	0x0200		/* must do output seq adjust */
84 #define IP_VS_CONN_F_IN_SEQ	0x0400		/* must do input seq adjust */
85 #define IP_VS_CONN_F_SEQ_MASK	0x0600		/* in/out sequence mask */
86 #define IP_VS_CONN_F_NO_CPORT	0x0800		/* no client port set yet */
87 
88 /* Move it to better place one day, for now keep it unique */
89 #define NFC_IPVS_PROPERTY	0x10000
90 
91 #define IP_VS_SCHEDNAME_MAXLEN	16
92 #define IP_VS_IFNAME_MAXLEN	16
93 
94 
95 /*
96  *	The struct ip_vs_service_user and struct ip_vs_dest_user are
97  *	used to set IPVS rules through setsockopt.
98  */
99 struct ip_vs_service_user {
100 	/* virtual service addresses */
101 	u_int16_t		protocol;
102 	u_int32_t		addr;		/* virtual ip address */
103 	u_int16_t		port;
104 	u_int32_t		fwmark;		/* firwall mark of service */
105 
106 	/* virtual service options */
107 	char			sched_name[IP_VS_SCHEDNAME_MAXLEN];
108 	unsigned		flags;		/* virtual service flags */
109 	unsigned		timeout;	/* persistent timeout in sec */
110 	u_int32_t		netmask;	/* persistent netmask */
111 };
112 
113 
114 struct ip_vs_dest_user {
115 	/* destination server address */
116 	u_int32_t		addr;
117 	u_int16_t		port;
118 
119 	/* real server options */
120 	unsigned		conn_flags;	/* connection flags */
121 	int			weight;		/* destination weight */
122 
123 	/* thresholds for active connections */
124 	u_int32_t		u_threshold;	/* upper threshold */
125 	u_int32_t		l_threshold;	/* lower threshold */
126 };
127 
128 
129 /*
130  *	IPVS statistics object (for user space)
131  */
132 struct ip_vs_stats_user
133 {
134 	__u32                   conns;          /* connections scheduled */
135 	__u32                   inpkts;         /* incoming packets */
136 	__u32                   outpkts;        /* outgoing packets */
137 	__u64                   inbytes;        /* incoming bytes */
138 	__u64                   outbytes;       /* outgoing bytes */
139 
140 	__u32			cps;		/* current connection rate */
141 	__u32			inpps;		/* current in packet rate */
142 	__u32			outpps;		/* current out packet rate */
143 	__u32			inbps;		/* current in byte rate */
144 	__u32			outbps;		/* current out byte rate */
145 };
146 
147 
148 /* The argument to IP_VS_SO_GET_INFO */
149 struct ip_vs_getinfo {
150 	/* version number */
151 	unsigned int		version;
152 
153 	/* size of connection hash table */
154 	unsigned int		size;
155 
156 	/* number of virtual services */
157 	unsigned int		num_services;
158 };
159 
160 
161 /* The argument to IP_VS_SO_GET_SERVICE */
162 struct ip_vs_service_entry {
163 	/* which service: user fills in these */
164 	u_int16_t		protocol;
165 	u_int32_t		addr;		/* virtual address */
166 	u_int16_t		port;
167 	u_int32_t		fwmark;		/* firwall mark of service */
168 
169 	/* service options */
170 	char			sched_name[IP_VS_SCHEDNAME_MAXLEN];
171 	unsigned		flags;          /* virtual service flags */
172 	unsigned		timeout;	/* persistent timeout */
173 	u_int32_t		netmask;	/* persistent netmask */
174 
175 	/* number of real servers */
176 	unsigned int		num_dests;
177 
178 	/* statistics */
179 	struct ip_vs_stats_user stats;
180 };
181 
182 
183 struct ip_vs_dest_entry {
184 	u_int32_t		addr;		/* destination address */
185 	u_int16_t		port;
186 	unsigned		conn_flags;	/* connection flags */
187 	int			weight;		/* destination weight */
188 
189 	u_int32_t		u_threshold;	/* upper threshold */
190 	u_int32_t		l_threshold;	/* lower threshold */
191 
192 	u_int32_t		activeconns;	/* active connections */
193 	u_int32_t		inactconns;	/* inactive connections */
194 	u_int32_t		persistconns;	/* persistent connections */
195 
196 	/* statistics */
197 	struct ip_vs_stats_user stats;
198 };
199 
200 
201 /* The argument to IP_VS_SO_GET_DESTS */
202 struct ip_vs_get_dests {
203 	/* which service: user fills in these */
204 	u_int16_t		protocol;
205 	u_int32_t		addr;		/* virtual address */
206 	u_int16_t		port;
207 	u_int32_t		fwmark;		/* firwall mark of service */
208 
209 	/* number of real servers */
210 	unsigned int		num_dests;
211 
212 	/* the real servers */
213 	struct ip_vs_dest_entry	entrytable[0];
214 };
215 
216 
217 /* The argument to IP_VS_SO_GET_SERVICES */
218 struct ip_vs_get_services {
219 	/* number of virtual services */
220 	unsigned int		num_services;
221 
222 	/* service table */
223 	struct ip_vs_service_entry entrytable[0];
224 };
225 
226 
227 /* The argument to IP_VS_SO_GET_TIMEOUT */
228 struct ip_vs_timeout_user {
229 	int			tcp_timeout;
230 	int			tcp_fin_timeout;
231 	int			udp_timeout;
232 };
233 
234 
235 /* The argument to IP_VS_SO_GET_DAEMON */
236 struct ip_vs_daemon_user {
237 	/* sync daemon state (master/backup) */
238 	int			state;
239 
240 	/* multicast interface name */
241 	char			mcast_ifn[IP_VS_IFNAME_MAXLEN];
242 
243 	/* SyncID we belong to */
244 	int			syncid;
245 };
246 
247 
248 #ifdef __KERNEL__
249 
250 #include <linux/config.h>
251 #include <linux/list.h>                 /* for struct list_head */
252 #include <linux/spinlock.h>             /* for struct rwlock_t */
253 #include <linux/skbuff.h>               /* for struct sk_buff */
254 #include <linux/ip.h>                   /* for struct iphdr */
255 #include <asm/atomic.h>                 /* for struct atomic_t */
256 #include <linux/netdevice.h>		/* for struct neighbour */
257 #include <net/dst.h>			/* for struct dst_entry */
258 #include <net/udp.h>
259 #include <linux/compiler.h>
260 
261 
262 #ifdef CONFIG_IP_VS_DEBUG
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  *      IPVS sysctl variables under the /proc/sys/net/ipv4/vs/
332  */
333 #define NET_IPV4_VS              21
334 
335 enum {
336 	NET_IPV4_VS_DEBUG_LEVEL=1,
337 	NET_IPV4_VS_AMEMTHRESH=2,
338 	NET_IPV4_VS_AMDROPRATE=3,
339 	NET_IPV4_VS_DROP_ENTRY=4,
340 	NET_IPV4_VS_DROP_PACKET=5,
341 	NET_IPV4_VS_SECURE_TCP=6,
342 	NET_IPV4_VS_TO_ES=7,
343 	NET_IPV4_VS_TO_SS=8,
344 	NET_IPV4_VS_TO_SR=9,
345 	NET_IPV4_VS_TO_FW=10,
346 	NET_IPV4_VS_TO_TW=11,
347 	NET_IPV4_VS_TO_CL=12,
348 	NET_IPV4_VS_TO_CW=13,
349 	NET_IPV4_VS_TO_LA=14,
350 	NET_IPV4_VS_TO_LI=15,
351 	NET_IPV4_VS_TO_SA=16,
352 	NET_IPV4_VS_TO_UDP=17,
353 	NET_IPV4_VS_TO_ICMP=18,
354 	NET_IPV4_VS_LBLC_EXPIRE=19,
355 	NET_IPV4_VS_LBLCR_EXPIRE=20,
356 	NET_IPV4_VS_CACHE_BYPASS=22,
357 	NET_IPV4_VS_EXPIRE_NODEST_CONN=23,
358 	NET_IPV4_VS_SYNC_THRESHOLD=24,
359 	NET_IPV4_VS_NAT_ICMP_SEND=25,
360 	NET_IPV4_VS_EXPIRE_QUIESCENT_TEMPLATE=26,
361 	NET_IPV4_VS_LAST
362 };
363 
364 /*
365  *      TCP State Values
366  */
367 enum {
368 	IP_VS_TCP_S_NONE = 0,
369 	IP_VS_TCP_S_ESTABLISHED,
370 	IP_VS_TCP_S_SYN_SENT,
371 	IP_VS_TCP_S_SYN_RECV,
372 	IP_VS_TCP_S_FIN_WAIT,
373 	IP_VS_TCP_S_TIME_WAIT,
374 	IP_VS_TCP_S_CLOSE,
375 	IP_VS_TCP_S_CLOSE_WAIT,
376 	IP_VS_TCP_S_LAST_ACK,
377 	IP_VS_TCP_S_LISTEN,
378 	IP_VS_TCP_S_SYNACK,
379 	IP_VS_TCP_S_LAST
380 };
381 
382 /*
383  *	UDP State Values
384  */
385 enum {
386 	IP_VS_UDP_S_NORMAL,
387 	IP_VS_UDP_S_LAST,
388 };
389 
390 /*
391  *	ICMP State Values
392  */
393 enum {
394 	IP_VS_ICMP_S_NORMAL,
395 	IP_VS_ICMP_S_LAST,
396 };
397 
398 /*
399  *	Delta sequence info structure
400  *	Each ip_vs_conn has 2 (output AND input seq. changes).
401  *      Only used in the VS/NAT.
402  */
403 struct ip_vs_seq {
404 	__u32			init_seq;	/* Add delta from this seq */
405 	__u32			delta;		/* Delta in sequence numbers */
406 	__u32			previous_delta;	/* Delta in sequence numbers
407 						   before last resized pkt */
408 };
409 
410 
411 /*
412  *	IPVS statistics object
413  */
414 struct ip_vs_stats
415 {
416 	__u32                   conns;          /* connections scheduled */
417 	__u32                   inpkts;         /* incoming packets */
418 	__u32                   outpkts;        /* outgoing packets */
419 	__u64                   inbytes;        /* incoming bytes */
420 	__u64                   outbytes;       /* outgoing bytes */
421 
422 	__u32			cps;		/* current connection rate */
423 	__u32			inpps;		/* current in packet rate */
424 	__u32			outpps;		/* current out packet rate */
425 	__u32			inbps;		/* current in byte rate */
426 	__u32			outbps;		/* current out byte rate */
427 
428 	spinlock_t              lock;           /* spin lock */
429 };
430 
431 struct ip_vs_conn;
432 struct ip_vs_app;
433 
434 struct ip_vs_protocol {
435 	struct ip_vs_protocol	*next;
436 	char			*name;
437 	__u16			protocol;
438 	int			dont_defrag;
439 	atomic_t		appcnt;		/* counter of proto app incs */
440 	int			*timeout_table;	/* protocol timeout table */
441 
442 	void (*init)(struct ip_vs_protocol *pp);
443 
444 	void (*exit)(struct ip_vs_protocol *pp);
445 
446 	int (*conn_schedule)(struct sk_buff *skb,
447 			     struct ip_vs_protocol *pp,
448 			     int *verdict, struct ip_vs_conn **cpp);
449 
450 	struct ip_vs_conn *
451 	(*conn_in_get)(const struct sk_buff *skb,
452 		       struct ip_vs_protocol *pp,
453 		       const struct iphdr *iph,
454 		       unsigned int proto_off,
455 		       int inverse);
456 
457 	struct ip_vs_conn *
458 	(*conn_out_get)(const struct sk_buff *skb,
459 			struct ip_vs_protocol *pp,
460 			const struct iphdr *iph,
461 			unsigned int proto_off,
462 			int inverse);
463 
464 	int (*snat_handler)(struct sk_buff **pskb,
465 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
466 
467 	int (*dnat_handler)(struct sk_buff **pskb,
468 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
469 
470 	int (*csum_check)(struct sk_buff *skb, struct ip_vs_protocol *pp);
471 
472 	const char *(*state_name)(int state);
473 
474 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
475 				const struct sk_buff *skb,
476 				struct ip_vs_protocol *pp);
477 
478 	int (*register_app)(struct ip_vs_app *inc);
479 
480 	void (*unregister_app)(struct ip_vs_app *inc);
481 
482 	int (*app_conn_bind)(struct ip_vs_conn *cp);
483 
484 	void (*debug_packet)(struct ip_vs_protocol *pp,
485 			     const struct sk_buff *skb,
486 			     int offset,
487 			     const char *msg);
488 
489 	void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
490 
491 	int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
492 };
493 
494 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
495 
496 /*
497  *	IP_VS structure allocated for each dynamically scheduled connection
498  */
499 struct ip_vs_conn {
500 	struct list_head        c_list;         /* hashed list heads */
501 
502 	/* Protocol, addresses and port numbers */
503 	__u32                   caddr;          /* client address */
504 	__u32                   vaddr;          /* virtual address */
505 	__u32                   daddr;          /* destination address */
506 	__u16                   cport;
507 	__u16                   vport;
508 	__u16                   dport;
509 	__u16                   protocol;       /* Which protocol (TCP/UDP) */
510 
511 	/* counter and timer */
512 	atomic_t		refcnt;		/* reference count */
513 	struct timer_list	timer;		/* Expiration timer */
514 	volatile unsigned long	timeout;	/* timeout */
515 
516 	/* Flags and state transition */
517 	spinlock_t              lock;           /* lock for state transition */
518 	volatile __u16          flags;          /* status flags */
519 	volatile __u16          state;          /* state info */
520 
521 	/* Control members */
522 	struct ip_vs_conn       *control;       /* Master control connection */
523 	atomic_t                n_control;      /* Number of controlled ones */
524 	struct ip_vs_dest       *dest;          /* real server */
525 	atomic_t                in_pkts;        /* incoming packet counter */
526 
527 	/* packet transmitter for different forwarding methods.  If it
528 	   mangles the packet, it must return NF_DROP or better NF_STOLEN,
529 	   otherwise this must be changed to a sk_buff **.
530 	 */
531 	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
532 			   struct ip_vs_protocol *pp);
533 
534 	/* Note: we can group the following members into a structure,
535 	   in order to save more space, and the following members are
536 	   only used in VS/NAT anyway */
537 	struct ip_vs_app        *app;           /* bound ip_vs_app object */
538 	void                    *app_data;      /* Application private data */
539 	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
540 	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
541 };
542 
543 
544 /*
545  *	The information about the virtual service offered to the net
546  *	and the forwarding entries
547  */
548 struct ip_vs_service {
549 	struct list_head	s_list;   /* for normal service table */
550 	struct list_head	f_list;   /* for fwmark-based service table */
551 	atomic_t		refcnt;   /* reference counter */
552 	atomic_t		usecnt;   /* use counter */
553 
554 	__u16			protocol; /* which protocol (TCP/UDP) */
555 	__u32			addr;	  /* IP address for virtual service */
556 	__u16			port;	  /* port number for the service */
557 	__u32                   fwmark;   /* firewall mark of the service */
558 	unsigned		flags;	  /* service status flags */
559 	unsigned		timeout;  /* persistent timeout in ticks */
560 	__u32			netmask;  /* grouping granularity */
561 
562 	struct list_head	destinations;  /* real server d-linked list */
563 	__u32			num_dests;     /* number of servers */
564 	struct ip_vs_stats      stats;         /* statistics for the service */
565 	struct ip_vs_app	*inc;	  /* bind conns to this app inc */
566 
567 	/* for scheduling */
568 	struct ip_vs_scheduler	*scheduler;    /* bound scheduler object */
569 	rwlock_t		sched_lock;    /* lock sched_data */
570 	void			*sched_data;   /* scheduler application data */
571 };
572 
573 
574 /*
575  *	The real server destination forwarding entry
576  *	with ip address, port number, and so on.
577  */
578 struct ip_vs_dest {
579 	struct list_head	n_list;   /* for the dests in the service */
580 	struct list_head	d_list;   /* for table with all the dests */
581 
582 	__u32			addr;		/* IP address of the server */
583 	__u16			port;		/* port number of the server */
584 	volatile unsigned	flags;		/* dest status flags */
585 	atomic_t		conn_flags;	/* flags to copy to conn */
586 	atomic_t		weight;		/* server weight */
587 
588 	atomic_t		refcnt;		/* reference counter */
589 	struct ip_vs_stats      stats;          /* statistics */
590 
591 	/* connection counters and thresholds */
592 	atomic_t		activeconns;	/* active connections */
593 	atomic_t		inactconns;	/* inactive connections */
594 	atomic_t		persistconns;	/* persistent connections */
595 	__u32			u_threshold;	/* upper threshold */
596 	__u32			l_threshold;	/* lower threshold */
597 
598 	/* for destination cache */
599 	spinlock_t		dst_lock;	/* lock of dst_cache */
600 	struct dst_entry	*dst_cache;	/* destination cache entry */
601 	u32			dst_rtos;	/* RT_TOS(tos) for dst */
602 
603 	/* for virtual service */
604 	struct ip_vs_service	*svc;		/* service it belongs to */
605 	__u16			protocol;	/* which protocol (TCP/UDP) */
606 	__u32			vaddr;		/* virtual IP address */
607 	__u16			vport;		/* virtual port number */
608 	__u32			vfwmark;	/* firewall mark of service */
609 };
610 
611 
612 /*
613  *	The scheduler object
614  */
615 struct ip_vs_scheduler {
616 	struct list_head	n_list;		/* d-linked list head */
617 	char			*name;		/* scheduler name */
618 	atomic_t		refcnt;		/* reference counter */
619 	struct module		*module;	/* THIS_MODULE/NULL */
620 
621 	/* scheduler initializing service */
622 	int (*init_service)(struct ip_vs_service *svc);
623 	/* scheduling service finish */
624 	int (*done_service)(struct ip_vs_service *svc);
625 	/* scheduler updating service */
626 	int (*update_service)(struct ip_vs_service *svc);
627 
628 	/* selecting a server from the given service */
629 	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
630 				       const struct sk_buff *skb);
631 };
632 
633 
634 /*
635  *	The application module object (a.k.a. app incarnation)
636  */
637 struct ip_vs_app
638 {
639 	struct list_head	a_list;		/* member in app list */
640 	int			type;		/* IP_VS_APP_TYPE_xxx */
641 	char			*name;		/* application module name */
642 	__u16			protocol;
643 	struct module		*module;	/* THIS_MODULE/NULL */
644 	struct list_head	incs_list;	/* list of incarnations */
645 
646 	/* members for application incarnations */
647 	struct list_head	p_list;		/* member in proto app list */
648 	struct ip_vs_app	*app;		/* its real application */
649 	__u16			port;		/* port number in net order */
650 	atomic_t		usecnt;		/* usage counter */
651 
652 	/* output hook: return false if can't linearize. diff set for TCP.  */
653 	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
654 		       struct sk_buff **, int *diff);
655 
656 	/* input hook: return false if can't linearize. diff set for TCP. */
657 	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
658 		      struct sk_buff **, int *diff);
659 
660 	/* ip_vs_app initializer */
661 	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
662 
663 	/* ip_vs_app finish */
664 	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
665 
666 
667 	/* not used now */
668 	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
669 			 struct ip_vs_protocol *);
670 
671 	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
672 
673 	int *			timeout_table;
674 	int *			timeouts;
675 	int			timeouts_size;
676 
677 	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
678 			     int *verdict, struct ip_vs_conn **cpp);
679 
680 	struct ip_vs_conn *
681 	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
682 		       const struct iphdr *iph, unsigned int proto_off,
683 		       int inverse);
684 
685 	struct ip_vs_conn *
686 	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
687 			const struct iphdr *iph, unsigned int proto_off,
688 			int inverse);
689 
690 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
691 				const struct sk_buff *skb,
692 				struct ip_vs_app *app);
693 
694 	void (*timeout_change)(struct ip_vs_app *app, int flags);
695 };
696 
697 
698 /*
699  *      IPVS core functions
700  *      (from ip_vs_core.c)
701  */
702 extern const char *ip_vs_proto_name(unsigned proto);
703 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
704 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table(t, sizeof(t)/sizeof(t[0]))
705 
706 #define IP_VS_APP_TYPE_UNSPEC	0
707 #define IP_VS_APP_TYPE_FTP	1
708 
709 /*
710  *     ip_vs_conn handling functions
711  *     (from ip_vs_conn.c)
712  */
713 
714 /*
715  *     IPVS connection entry hash table
716  */
717 #ifndef CONFIG_IP_VS_TAB_BITS
718 #define CONFIG_IP_VS_TAB_BITS   12
719 #endif
720 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
721 #if CONFIG_IP_VS_TAB_BITS < 8
722 #define IP_VS_CONN_TAB_BITS	8
723 #endif
724 #if CONFIG_IP_VS_TAB_BITS > 20
725 #define IP_VS_CONN_TAB_BITS	20
726 #endif
727 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
728 #define IP_VS_CONN_TAB_BITS	CONFIG_IP_VS_TAB_BITS
729 #endif
730 #define IP_VS_CONN_TAB_SIZE     (1 << IP_VS_CONN_TAB_BITS)
731 #define IP_VS_CONN_TAB_MASK     (IP_VS_CONN_TAB_SIZE - 1)
732 
733 enum {
734 	IP_VS_DIR_INPUT = 0,
735 	IP_VS_DIR_OUTPUT,
736 	IP_VS_DIR_INPUT_ONLY,
737 	IP_VS_DIR_LAST,
738 };
739 
740 extern struct ip_vs_conn *ip_vs_conn_in_get
741 (int protocol, __u32 s_addr, __u16 s_port, __u32 d_addr, __u16 d_port);
742 extern struct ip_vs_conn *ip_vs_conn_out_get
743 (int protocol, __u32 s_addr, __u16 s_port, __u32 d_addr, __u16 d_port);
744 
745 /* put back the conn without restarting its timer */
746 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
747 {
748 	atomic_dec(&cp->refcnt);
749 }
750 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
751 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __u16 cport);
752 
753 extern struct ip_vs_conn *
754 ip_vs_conn_new(int proto, __u32 caddr, __u16 cport, __u32 vaddr, __u16 vport,
755 	       __u32 daddr, __u16 dport, unsigned flags,
756 	       struct ip_vs_dest *dest);
757 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
758 
759 extern const char * ip_vs_state_name(__u16 proto, int state);
760 
761 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
762 extern int ip_vs_check_template(struct ip_vs_conn *ct);
763 extern void ip_vs_secure_tcp_set(int on);
764 extern void ip_vs_random_dropentry(void);
765 extern int ip_vs_conn_init(void);
766 extern void ip_vs_conn_cleanup(void);
767 
768 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
769 {
770 	struct ip_vs_conn *ctl_cp = cp->control;
771 	if (!ctl_cp) {
772 		IP_VS_ERR("request control DEL for uncontrolled: "
773 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
774 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
775 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
776 		return;
777 	}
778 
779 	IP_VS_DBG(7, "DELeting control for: "
780 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
781 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
782 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
783 
784 	cp->control = NULL;
785 	if (atomic_read(&ctl_cp->n_control) == 0) {
786 		IP_VS_ERR("BUG control DEL with n=0 : "
787 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
788 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
789 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
790 		return;
791 	}
792 	atomic_dec(&ctl_cp->n_control);
793 }
794 
795 static inline void
796 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
797 {
798 	if (cp->control) {
799 		IP_VS_ERR("request control ADD for already controlled: "
800 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
801 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
802 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
803 		ip_vs_control_del(cp);
804 	}
805 
806 	IP_VS_DBG(7, "ADDing control for: "
807 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
808 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
809 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
810 
811 	cp->control = ctl_cp;
812 	atomic_inc(&ctl_cp->n_control);
813 }
814 
815 
816 /*
817  *      IPVS application functions
818  *      (from ip_vs_app.c)
819  */
820 #define IP_VS_APP_MAX_PORTS  8
821 extern int register_ip_vs_app(struct ip_vs_app *app);
822 extern void unregister_ip_vs_app(struct ip_vs_app *app);
823 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
824 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
825 extern int
826 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
827 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
828 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
829 
830 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff **pskb);
831 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff **pskb);
832 extern int ip_vs_skb_replace(struct sk_buff *skb, int pri,
833 			     char *o_buf, int o_len, char *n_buf, int n_len);
834 extern int ip_vs_app_init(void);
835 extern void ip_vs_app_cleanup(void);
836 
837 
838 /*
839  *	IPVS protocol functions (from ip_vs_proto.c)
840  */
841 extern int ip_vs_protocol_init(void);
842 extern void ip_vs_protocol_cleanup(void);
843 extern void ip_vs_protocol_timeout_change(int flags);
844 extern int *ip_vs_create_timeout_table(int *table, int size);
845 extern int
846 ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to);
847 extern void
848 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
849 			  int offset, const char *msg);
850 
851 extern struct ip_vs_protocol ip_vs_protocol_tcp;
852 extern struct ip_vs_protocol ip_vs_protocol_udp;
853 extern struct ip_vs_protocol ip_vs_protocol_icmp;
854 extern struct ip_vs_protocol ip_vs_protocol_esp;
855 extern struct ip_vs_protocol ip_vs_protocol_ah;
856 
857 
858 /*
859  *      Registering/unregistering scheduler functions
860  *      (from ip_vs_sched.c)
861  */
862 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
863 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
864 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
865 				struct ip_vs_scheduler *scheduler);
866 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
867 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
868 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
869 extern struct ip_vs_conn *
870 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
871 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
872 			struct ip_vs_protocol *pp);
873 
874 
875 /*
876  *      IPVS control data and functions (from ip_vs_ctl.c)
877  */
878 extern int sysctl_ip_vs_cache_bypass;
879 extern int sysctl_ip_vs_expire_nodest_conn;
880 extern int sysctl_ip_vs_expire_quiescent_template;
881 extern int sysctl_ip_vs_sync_threshold[2];
882 extern int sysctl_ip_vs_nat_icmp_send;
883 extern struct ip_vs_stats ip_vs_stats;
884 
885 extern struct ip_vs_service *
886 ip_vs_service_get(__u32 fwmark, __u16 protocol, __u32 vaddr, __u16 vport);
887 
888 static inline void ip_vs_service_put(struct ip_vs_service *svc)
889 {
890 	atomic_dec(&svc->usecnt);
891 }
892 
893 extern struct ip_vs_dest *
894 ip_vs_lookup_real_service(__u16 protocol, __u32 daddr, __u16 dport);
895 extern int ip_vs_use_count_inc(void);
896 extern void ip_vs_use_count_dec(void);
897 extern int ip_vs_control_init(void);
898 extern void ip_vs_control_cleanup(void);
899 
900 
901 /*
902  *      IPVS sync daemon data and function prototypes
903  *      (from ip_vs_sync.c)
904  */
905 extern volatile int ip_vs_sync_state;
906 extern volatile int ip_vs_master_syncid;
907 extern volatile int ip_vs_backup_syncid;
908 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
909 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
910 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
911 extern int stop_sync_thread(int state);
912 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
913 
914 
915 /*
916  *      IPVS rate estimator prototypes (from ip_vs_est.c)
917  */
918 extern int ip_vs_new_estimator(struct ip_vs_stats *stats);
919 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
920 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
921 
922 /*
923  *	Various IPVS packet transmitters (from ip_vs_xmit.c)
924  */
925 extern int ip_vs_null_xmit
926 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
927 extern int ip_vs_bypass_xmit
928 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
929 extern int ip_vs_nat_xmit
930 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
931 extern int ip_vs_tunnel_xmit
932 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
933 extern int ip_vs_dr_xmit
934 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
935 extern int ip_vs_icmp_xmit
936 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
937 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
938 
939 
940 /*
941  *	This is a simple mechanism to ignore packets when
942  *	we are loaded. Just set ip_vs_drop_rate to 'n' and
943  *	we start to drop 1/rate of the packets
944  */
945 extern int ip_vs_drop_rate;
946 extern int ip_vs_drop_counter;
947 
948 static __inline__ int ip_vs_todrop(void)
949 {
950 	if (!ip_vs_drop_rate) return 0;
951 	if (--ip_vs_drop_counter > 0) return 0;
952 	ip_vs_drop_counter = ip_vs_drop_rate;
953 	return 1;
954 }
955 
956 /*
957  *      ip_vs_fwd_tag returns the forwarding tag of the connection
958  */
959 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
960 
961 extern __inline__ char ip_vs_fwd_tag(struct ip_vs_conn *cp)
962 {
963 	char fwd;
964 
965 	switch (IP_VS_FWD_METHOD(cp)) {
966 	case IP_VS_CONN_F_MASQ:
967 		fwd = 'M'; break;
968 	case IP_VS_CONN_F_LOCALNODE:
969 		fwd = 'L'; break;
970 	case IP_VS_CONN_F_TUNNEL:
971 		fwd = 'T'; break;
972 	case IP_VS_CONN_F_DROUTE:
973 		fwd = 'R'; break;
974 	case IP_VS_CONN_F_BYPASS:
975 		fwd = 'B'; break;
976 	default:
977 		fwd = '?'; break;
978 	}
979 	return fwd;
980 }
981 
982 extern int ip_vs_make_skb_writable(struct sk_buff **pskb, int len);
983 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
984 		struct ip_vs_conn *cp, int dir);
985 
986 extern u16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
987 
988 static inline u16 ip_vs_check_diff(u32 old, u32 new, u16 oldsum)
989 {
990 	u32 diff[2] = { old, new };
991 
992 	return csum_fold(csum_partial((char *) diff, sizeof(diff),
993 				      oldsum ^ 0xFFFF));
994 }
995 
996 #endif /* __KERNEL__ */
997 
998 #endif	/* _IP_VS_H */
999