xref: /openbmc/linux/include/net/ip_vs.h (revision d5cb9783536a41df9f9cba5b0a1d78047ed787f7)
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 #define IP_VS_CONN_F_TEMPLATE	0x1000		/* template, not connection */
88 
89 /* Move it to better place one day, for now keep it unique */
90 #define NFC_IPVS_PROPERTY	0x10000
91 
92 #define IP_VS_SCHEDNAME_MAXLEN	16
93 #define IP_VS_IFNAME_MAXLEN	16
94 
95 
96 /*
97  *	The struct ip_vs_service_user and struct ip_vs_dest_user are
98  *	used to set IPVS rules through setsockopt.
99  */
100 struct ip_vs_service_user {
101 	/* virtual service addresses */
102 	u_int16_t		protocol;
103 	u_int32_t		addr;		/* virtual ip address */
104 	u_int16_t		port;
105 	u_int32_t		fwmark;		/* firwall mark of service */
106 
107 	/* virtual service options */
108 	char			sched_name[IP_VS_SCHEDNAME_MAXLEN];
109 	unsigned		flags;		/* virtual service flags */
110 	unsigned		timeout;	/* persistent timeout in sec */
111 	u_int32_t		netmask;	/* persistent netmask */
112 };
113 
114 
115 struct ip_vs_dest_user {
116 	/* destination server address */
117 	u_int32_t		addr;
118 	u_int16_t		port;
119 
120 	/* real server options */
121 	unsigned		conn_flags;	/* connection flags */
122 	int			weight;		/* destination weight */
123 
124 	/* thresholds for active connections */
125 	u_int32_t		u_threshold;	/* upper threshold */
126 	u_int32_t		l_threshold;	/* lower threshold */
127 };
128 
129 
130 /*
131  *	IPVS statistics object (for user space)
132  */
133 struct ip_vs_stats_user
134 {
135 	__u32                   conns;          /* connections scheduled */
136 	__u32                   inpkts;         /* incoming packets */
137 	__u32                   outpkts;        /* outgoing packets */
138 	__u64                   inbytes;        /* incoming bytes */
139 	__u64                   outbytes;       /* outgoing bytes */
140 
141 	__u32			cps;		/* current connection rate */
142 	__u32			inpps;		/* current in packet rate */
143 	__u32			outpps;		/* current out packet rate */
144 	__u32			inbps;		/* current in byte rate */
145 	__u32			outbps;		/* current out byte rate */
146 };
147 
148 
149 /* The argument to IP_VS_SO_GET_INFO */
150 struct ip_vs_getinfo {
151 	/* version number */
152 	unsigned int		version;
153 
154 	/* size of connection hash table */
155 	unsigned int		size;
156 
157 	/* number of virtual services */
158 	unsigned int		num_services;
159 };
160 
161 
162 /* The argument to IP_VS_SO_GET_SERVICE */
163 struct ip_vs_service_entry {
164 	/* which service: user fills in these */
165 	u_int16_t		protocol;
166 	u_int32_t		addr;		/* virtual address */
167 	u_int16_t		port;
168 	u_int32_t		fwmark;		/* firwall mark of service */
169 
170 	/* service options */
171 	char			sched_name[IP_VS_SCHEDNAME_MAXLEN];
172 	unsigned		flags;          /* virtual service flags */
173 	unsigned		timeout;	/* persistent timeout */
174 	u_int32_t		netmask;	/* persistent netmask */
175 
176 	/* number of real servers */
177 	unsigned int		num_dests;
178 
179 	/* statistics */
180 	struct ip_vs_stats_user stats;
181 };
182 
183 
184 struct ip_vs_dest_entry {
185 	u_int32_t		addr;		/* destination address */
186 	u_int16_t		port;
187 	unsigned		conn_flags;	/* connection flags */
188 	int			weight;		/* destination weight */
189 
190 	u_int32_t		u_threshold;	/* upper threshold */
191 	u_int32_t		l_threshold;	/* lower threshold */
192 
193 	u_int32_t		activeconns;	/* active connections */
194 	u_int32_t		inactconns;	/* inactive connections */
195 	u_int32_t		persistconns;	/* persistent connections */
196 
197 	/* statistics */
198 	struct ip_vs_stats_user stats;
199 };
200 
201 
202 /* The argument to IP_VS_SO_GET_DESTS */
203 struct ip_vs_get_dests {
204 	/* which service: user fills in these */
205 	u_int16_t		protocol;
206 	u_int32_t		addr;		/* virtual address */
207 	u_int16_t		port;
208 	u_int32_t		fwmark;		/* firwall mark of service */
209 
210 	/* number of real servers */
211 	unsigned int		num_dests;
212 
213 	/* the real servers */
214 	struct ip_vs_dest_entry	entrytable[0];
215 };
216 
217 
218 /* The argument to IP_VS_SO_GET_SERVICES */
219 struct ip_vs_get_services {
220 	/* number of virtual services */
221 	unsigned int		num_services;
222 
223 	/* service table */
224 	struct ip_vs_service_entry entrytable[0];
225 };
226 
227 
228 /* The argument to IP_VS_SO_GET_TIMEOUT */
229 struct ip_vs_timeout_user {
230 	int			tcp_timeout;
231 	int			tcp_fin_timeout;
232 	int			udp_timeout;
233 };
234 
235 
236 /* The argument to IP_VS_SO_GET_DAEMON */
237 struct ip_vs_daemon_user {
238 	/* sync daemon state (master/backup) */
239 	int			state;
240 
241 	/* multicast interface name */
242 	char			mcast_ifn[IP_VS_IFNAME_MAXLEN];
243 
244 	/* SyncID we belong to */
245 	int			syncid;
246 };
247 
248 
249 #ifdef __KERNEL__
250 
251 #include <linux/config.h>
252 #include <linux/list.h>                 /* for struct list_head */
253 #include <linux/spinlock.h>             /* for struct rwlock_t */
254 #include <linux/skbuff.h>               /* for struct sk_buff */
255 #include <linux/ip.h>                   /* for struct iphdr */
256 #include <asm/atomic.h>                 /* for struct atomic_t */
257 #include <linux/netdevice.h>		/* for struct neighbour */
258 #include <net/dst.h>			/* for struct dst_entry */
259 #include <net/udp.h>
260 #include <linux/compiler.h>
261 
262 
263 #ifdef CONFIG_IP_VS_DEBUG
264 extern int ip_vs_get_debug_level(void);
265 #define IP_VS_DBG(level, msg...)			\
266     do {						\
267 	    if (level <= ip_vs_get_debug_level())	\
268 		    printk(KERN_DEBUG "IPVS: " msg);	\
269     } while (0)
270 #define IP_VS_DBG_RL(msg...)				\
271     do {						\
272 	    if (net_ratelimit())			\
273 		    printk(KERN_DEBUG "IPVS: " msg);	\
274     } while (0)
275 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		\
276     do {						\
277 	    if (level <= ip_vs_get_debug_level())	\
278 		pp->debug_packet(pp, skb, ofs, msg);	\
279     } while (0)
280 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	\
281     do {						\
282 	    if (level <= ip_vs_get_debug_level() &&	\
283 		net_ratelimit())			\
284 		pp->debug_packet(pp, skb, ofs, msg);	\
285     } while (0)
286 #else	/* NO DEBUGGING at ALL */
287 #define IP_VS_DBG(level, msg...)  do {} while (0)
288 #define IP_VS_DBG_RL(msg...)  do {} while (0)
289 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg)		do {} while (0)
290 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg)	do {} while (0)
291 #endif
292 
293 #define IP_VS_BUG() BUG()
294 #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg)
295 #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg)
296 #define IP_VS_WARNING(msg...) \
297 	printk(KERN_WARNING "IPVS: " msg)
298 #define IP_VS_ERR_RL(msg...)				\
299     do {						\
300 	    if (net_ratelimit())			\
301 		    printk(KERN_ERR "IPVS: " msg);	\
302     } while (0)
303 
304 #ifdef CONFIG_IP_VS_DEBUG
305 #define EnterFunction(level)						\
306     do {								\
307 	    if (level <= ip_vs_get_debug_level())			\
308 		    printk(KERN_DEBUG "Enter: %s, %s line %i\n",	\
309 			   __FUNCTION__, __FILE__, __LINE__);		\
310     } while (0)
311 #define LeaveFunction(level)                                            \
312     do {                                                                \
313 	    if (level <= ip_vs_get_debug_level())                       \
314 			printk(KERN_DEBUG "Leave: %s, %s line %i\n",    \
315 			       __FUNCTION__, __FILE__, __LINE__);       \
316     } while (0)
317 #else
318 #define EnterFunction(level)   do {} while (0)
319 #define LeaveFunction(level)   do {} while (0)
320 #endif
321 
322 #define	IP_VS_WAIT_WHILE(expr)	while (expr) { cpu_relax(); }
323 
324 
325 /*
326  *      The port number of FTP service (in network order).
327  */
328 #define FTPPORT  __constant_htons(21)
329 #define FTPDATA  __constant_htons(20)
330 
331 /*
332  *      IPVS sysctl variables under the /proc/sys/net/ipv4/vs/
333  */
334 #define NET_IPV4_VS              21
335 
336 enum {
337 	NET_IPV4_VS_DEBUG_LEVEL=1,
338 	NET_IPV4_VS_AMEMTHRESH=2,
339 	NET_IPV4_VS_AMDROPRATE=3,
340 	NET_IPV4_VS_DROP_ENTRY=4,
341 	NET_IPV4_VS_DROP_PACKET=5,
342 	NET_IPV4_VS_SECURE_TCP=6,
343 	NET_IPV4_VS_TO_ES=7,
344 	NET_IPV4_VS_TO_SS=8,
345 	NET_IPV4_VS_TO_SR=9,
346 	NET_IPV4_VS_TO_FW=10,
347 	NET_IPV4_VS_TO_TW=11,
348 	NET_IPV4_VS_TO_CL=12,
349 	NET_IPV4_VS_TO_CW=13,
350 	NET_IPV4_VS_TO_LA=14,
351 	NET_IPV4_VS_TO_LI=15,
352 	NET_IPV4_VS_TO_SA=16,
353 	NET_IPV4_VS_TO_UDP=17,
354 	NET_IPV4_VS_TO_ICMP=18,
355 	NET_IPV4_VS_LBLC_EXPIRE=19,
356 	NET_IPV4_VS_LBLCR_EXPIRE=20,
357 	NET_IPV4_VS_CACHE_BYPASS=22,
358 	NET_IPV4_VS_EXPIRE_NODEST_CONN=23,
359 	NET_IPV4_VS_SYNC_THRESHOLD=24,
360 	NET_IPV4_VS_NAT_ICMP_SEND=25,
361 	NET_IPV4_VS_EXPIRE_QUIESCENT_TEMPLATE=26,
362 	NET_IPV4_VS_LAST
363 };
364 
365 /*
366  *      TCP State Values
367  */
368 enum {
369 	IP_VS_TCP_S_NONE = 0,
370 	IP_VS_TCP_S_ESTABLISHED,
371 	IP_VS_TCP_S_SYN_SENT,
372 	IP_VS_TCP_S_SYN_RECV,
373 	IP_VS_TCP_S_FIN_WAIT,
374 	IP_VS_TCP_S_TIME_WAIT,
375 	IP_VS_TCP_S_CLOSE,
376 	IP_VS_TCP_S_CLOSE_WAIT,
377 	IP_VS_TCP_S_LAST_ACK,
378 	IP_VS_TCP_S_LISTEN,
379 	IP_VS_TCP_S_SYNACK,
380 	IP_VS_TCP_S_LAST
381 };
382 
383 /*
384  *	UDP State Values
385  */
386 enum {
387 	IP_VS_UDP_S_NORMAL,
388 	IP_VS_UDP_S_LAST,
389 };
390 
391 /*
392  *	ICMP State Values
393  */
394 enum {
395 	IP_VS_ICMP_S_NORMAL,
396 	IP_VS_ICMP_S_LAST,
397 };
398 
399 /*
400  *	Delta sequence info structure
401  *	Each ip_vs_conn has 2 (output AND input seq. changes).
402  *      Only used in the VS/NAT.
403  */
404 struct ip_vs_seq {
405 	__u32			init_seq;	/* Add delta from this seq */
406 	__u32			delta;		/* Delta in sequence numbers */
407 	__u32			previous_delta;	/* Delta in sequence numbers
408 						   before last resized pkt */
409 };
410 
411 
412 /*
413  *	IPVS statistics object
414  */
415 struct ip_vs_stats
416 {
417 	__u32                   conns;          /* connections scheduled */
418 	__u32                   inpkts;         /* incoming packets */
419 	__u32                   outpkts;        /* outgoing packets */
420 	__u64                   inbytes;        /* incoming bytes */
421 	__u64                   outbytes;       /* outgoing bytes */
422 
423 	__u32			cps;		/* current connection rate */
424 	__u32			inpps;		/* current in packet rate */
425 	__u32			outpps;		/* current out packet rate */
426 	__u32			inbps;		/* current in byte rate */
427 	__u32			outbps;		/* current out byte rate */
428 
429 	spinlock_t              lock;           /* spin lock */
430 };
431 
432 struct ip_vs_conn;
433 struct ip_vs_app;
434 
435 struct ip_vs_protocol {
436 	struct ip_vs_protocol	*next;
437 	char			*name;
438 	__u16			protocol;
439 	int			dont_defrag;
440 	atomic_t		appcnt;		/* counter of proto app incs */
441 	int			*timeout_table;	/* protocol timeout table */
442 
443 	void (*init)(struct ip_vs_protocol *pp);
444 
445 	void (*exit)(struct ip_vs_protocol *pp);
446 
447 	int (*conn_schedule)(struct sk_buff *skb,
448 			     struct ip_vs_protocol *pp,
449 			     int *verdict, struct ip_vs_conn **cpp);
450 
451 	struct ip_vs_conn *
452 	(*conn_in_get)(const struct sk_buff *skb,
453 		       struct ip_vs_protocol *pp,
454 		       const struct iphdr *iph,
455 		       unsigned int proto_off,
456 		       int inverse);
457 
458 	struct ip_vs_conn *
459 	(*conn_out_get)(const struct sk_buff *skb,
460 			struct ip_vs_protocol *pp,
461 			const struct iphdr *iph,
462 			unsigned int proto_off,
463 			int inverse);
464 
465 	int (*snat_handler)(struct sk_buff **pskb,
466 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
467 
468 	int (*dnat_handler)(struct sk_buff **pskb,
469 			    struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
470 
471 	int (*csum_check)(struct sk_buff *skb, struct ip_vs_protocol *pp);
472 
473 	const char *(*state_name)(int state);
474 
475 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
476 				const struct sk_buff *skb,
477 				struct ip_vs_protocol *pp);
478 
479 	int (*register_app)(struct ip_vs_app *inc);
480 
481 	void (*unregister_app)(struct ip_vs_app *inc);
482 
483 	int (*app_conn_bind)(struct ip_vs_conn *cp);
484 
485 	void (*debug_packet)(struct ip_vs_protocol *pp,
486 			     const struct sk_buff *skb,
487 			     int offset,
488 			     const char *msg);
489 
490 	void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
491 
492 	int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
493 };
494 
495 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
496 
497 /*
498  *	IP_VS structure allocated for each dynamically scheduled connection
499  */
500 struct ip_vs_conn {
501 	struct list_head        c_list;         /* hashed list heads */
502 
503 	/* Protocol, addresses and port numbers */
504 	__u32                   caddr;          /* client address */
505 	__u32                   vaddr;          /* virtual address */
506 	__u32                   daddr;          /* destination address */
507 	__u16                   cport;
508 	__u16                   vport;
509 	__u16                   dport;
510 	__u16                   protocol;       /* Which protocol (TCP/UDP) */
511 
512 	/* counter and timer */
513 	atomic_t		refcnt;		/* reference count */
514 	struct timer_list	timer;		/* Expiration timer */
515 	volatile unsigned long	timeout;	/* timeout */
516 
517 	/* Flags and state transition */
518 	spinlock_t              lock;           /* lock for state transition */
519 	volatile __u16          flags;          /* status flags */
520 	volatile __u16          state;          /* state info */
521 
522 	/* Control members */
523 	struct ip_vs_conn       *control;       /* Master control connection */
524 	atomic_t                n_control;      /* Number of controlled ones */
525 	struct ip_vs_dest       *dest;          /* real server */
526 	atomic_t                in_pkts;        /* incoming packet counter */
527 
528 	/* packet transmitter for different forwarding methods.  If it
529 	   mangles the packet, it must return NF_DROP or better NF_STOLEN,
530 	   otherwise this must be changed to a sk_buff **.
531 	 */
532 	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
533 			   struct ip_vs_protocol *pp);
534 
535 	/* Note: we can group the following members into a structure,
536 	   in order to save more space, and the following members are
537 	   only used in VS/NAT anyway */
538 	struct ip_vs_app        *app;           /* bound ip_vs_app object */
539 	void                    *app_data;      /* Application private data */
540 	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
541 	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
542 };
543 
544 
545 /*
546  *	The information about the virtual service offered to the net
547  *	and the forwarding entries
548  */
549 struct ip_vs_service {
550 	struct list_head	s_list;   /* for normal service table */
551 	struct list_head	f_list;   /* for fwmark-based service table */
552 	atomic_t		refcnt;   /* reference counter */
553 	atomic_t		usecnt;   /* use counter */
554 
555 	__u16			protocol; /* which protocol (TCP/UDP) */
556 	__u32			addr;	  /* IP address for virtual service */
557 	__u16			port;	  /* port number for the service */
558 	__u32                   fwmark;   /* firewall mark of the service */
559 	unsigned		flags;	  /* service status flags */
560 	unsigned		timeout;  /* persistent timeout in ticks */
561 	__u32			netmask;  /* grouping granularity */
562 
563 	struct list_head	destinations;  /* real server d-linked list */
564 	__u32			num_dests;     /* number of servers */
565 	struct ip_vs_stats      stats;         /* statistics for the service */
566 	struct ip_vs_app	*inc;	  /* bind conns to this app inc */
567 
568 	/* for scheduling */
569 	struct ip_vs_scheduler	*scheduler;    /* bound scheduler object */
570 	rwlock_t		sched_lock;    /* lock sched_data */
571 	void			*sched_data;   /* scheduler application data */
572 };
573 
574 
575 /*
576  *	The real server destination forwarding entry
577  *	with ip address, port number, and so on.
578  */
579 struct ip_vs_dest {
580 	struct list_head	n_list;   /* for the dests in the service */
581 	struct list_head	d_list;   /* for table with all the dests */
582 
583 	__u32			addr;		/* IP address of the server */
584 	__u16			port;		/* port number of the server */
585 	volatile unsigned	flags;		/* dest status flags */
586 	atomic_t		conn_flags;	/* flags to copy to conn */
587 	atomic_t		weight;		/* server weight */
588 
589 	atomic_t		refcnt;		/* reference counter */
590 	struct ip_vs_stats      stats;          /* statistics */
591 
592 	/* connection counters and thresholds */
593 	atomic_t		activeconns;	/* active connections */
594 	atomic_t		inactconns;	/* inactive connections */
595 	atomic_t		persistconns;	/* persistent connections */
596 	__u32			u_threshold;	/* upper threshold */
597 	__u32			l_threshold;	/* lower threshold */
598 
599 	/* for destination cache */
600 	spinlock_t		dst_lock;	/* lock of dst_cache */
601 	struct dst_entry	*dst_cache;	/* destination cache entry */
602 	u32			dst_rtos;	/* RT_TOS(tos) for dst */
603 
604 	/* for virtual service */
605 	struct ip_vs_service	*svc;		/* service it belongs to */
606 	__u16			protocol;	/* which protocol (TCP/UDP) */
607 	__u32			vaddr;		/* virtual IP address */
608 	__u16			vport;		/* virtual port number */
609 	__u32			vfwmark;	/* firewall mark of service */
610 };
611 
612 
613 /*
614  *	The scheduler object
615  */
616 struct ip_vs_scheduler {
617 	struct list_head	n_list;		/* d-linked list head */
618 	char			*name;		/* scheduler name */
619 	atomic_t		refcnt;		/* reference counter */
620 	struct module		*module;	/* THIS_MODULE/NULL */
621 
622 	/* scheduler initializing service */
623 	int (*init_service)(struct ip_vs_service *svc);
624 	/* scheduling service finish */
625 	int (*done_service)(struct ip_vs_service *svc);
626 	/* scheduler updating service */
627 	int (*update_service)(struct ip_vs_service *svc);
628 
629 	/* selecting a server from the given service */
630 	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
631 				       const struct sk_buff *skb);
632 };
633 
634 
635 /*
636  *	The application module object (a.k.a. app incarnation)
637  */
638 struct ip_vs_app
639 {
640 	struct list_head	a_list;		/* member in app list */
641 	int			type;		/* IP_VS_APP_TYPE_xxx */
642 	char			*name;		/* application module name */
643 	__u16			protocol;
644 	struct module		*module;	/* THIS_MODULE/NULL */
645 	struct list_head	incs_list;	/* list of incarnations */
646 
647 	/* members for application incarnations */
648 	struct list_head	p_list;		/* member in proto app list */
649 	struct ip_vs_app	*app;		/* its real application */
650 	__u16			port;		/* port number in net order */
651 	atomic_t		usecnt;		/* usage counter */
652 
653 	/* output hook: return false if can't linearize. diff set for TCP.  */
654 	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
655 		       struct sk_buff **, int *diff);
656 
657 	/* input hook: return false if can't linearize. diff set for TCP. */
658 	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
659 		      struct sk_buff **, int *diff);
660 
661 	/* ip_vs_app initializer */
662 	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
663 
664 	/* ip_vs_app finish */
665 	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
666 
667 
668 	/* not used now */
669 	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
670 			 struct ip_vs_protocol *);
671 
672 	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
673 
674 	int *			timeout_table;
675 	int *			timeouts;
676 	int			timeouts_size;
677 
678 	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
679 			     int *verdict, struct ip_vs_conn **cpp);
680 
681 	struct ip_vs_conn *
682 	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
683 		       const struct iphdr *iph, unsigned int proto_off,
684 		       int inverse);
685 
686 	struct ip_vs_conn *
687 	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
688 			const struct iphdr *iph, unsigned int proto_off,
689 			int inverse);
690 
691 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
692 				const struct sk_buff *skb,
693 				struct ip_vs_app *app);
694 
695 	void (*timeout_change)(struct ip_vs_app *app, int flags);
696 };
697 
698 
699 /*
700  *      IPVS core functions
701  *      (from ip_vs_core.c)
702  */
703 extern const char *ip_vs_proto_name(unsigned proto);
704 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
705 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table(t, sizeof(t)/sizeof(t[0]))
706 
707 #define IP_VS_APP_TYPE_UNSPEC	0
708 #define IP_VS_APP_TYPE_FTP	1
709 
710 /*
711  *     ip_vs_conn handling functions
712  *     (from ip_vs_conn.c)
713  */
714 
715 /*
716  *     IPVS connection entry hash table
717  */
718 #ifndef CONFIG_IP_VS_TAB_BITS
719 #define CONFIG_IP_VS_TAB_BITS   12
720 #endif
721 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
722 #if CONFIG_IP_VS_TAB_BITS < 8
723 #define IP_VS_CONN_TAB_BITS	8
724 #endif
725 #if CONFIG_IP_VS_TAB_BITS > 20
726 #define IP_VS_CONN_TAB_BITS	20
727 #endif
728 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
729 #define IP_VS_CONN_TAB_BITS	CONFIG_IP_VS_TAB_BITS
730 #endif
731 #define IP_VS_CONN_TAB_SIZE     (1 << IP_VS_CONN_TAB_BITS)
732 #define IP_VS_CONN_TAB_MASK     (IP_VS_CONN_TAB_SIZE - 1)
733 
734 enum {
735 	IP_VS_DIR_INPUT = 0,
736 	IP_VS_DIR_OUTPUT,
737 	IP_VS_DIR_INPUT_ONLY,
738 	IP_VS_DIR_LAST,
739 };
740 
741 extern struct ip_vs_conn *ip_vs_conn_in_get
742 (int protocol, __u32 s_addr, __u16 s_port, __u32 d_addr, __u16 d_port);
743 extern struct ip_vs_conn *ip_vs_ct_in_get
744 (int protocol, __u32 s_addr, __u16 s_port, __u32 d_addr, __u16 d_port);
745 extern struct ip_vs_conn *ip_vs_conn_out_get
746 (int protocol, __u32 s_addr, __u16 s_port, __u32 d_addr, __u16 d_port);
747 
748 /* put back the conn without restarting its timer */
749 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
750 {
751 	atomic_dec(&cp->refcnt);
752 }
753 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
754 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __u16 cport);
755 
756 extern struct ip_vs_conn *
757 ip_vs_conn_new(int proto, __u32 caddr, __u16 cport, __u32 vaddr, __u16 vport,
758 	       __u32 daddr, __u16 dport, unsigned flags,
759 	       struct ip_vs_dest *dest);
760 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
761 
762 extern const char * ip_vs_state_name(__u16 proto, int state);
763 
764 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
765 extern int ip_vs_check_template(struct ip_vs_conn *ct);
766 extern void ip_vs_secure_tcp_set(int on);
767 extern void ip_vs_random_dropentry(void);
768 extern int ip_vs_conn_init(void);
769 extern void ip_vs_conn_cleanup(void);
770 
771 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
772 {
773 	struct ip_vs_conn *ctl_cp = cp->control;
774 	if (!ctl_cp) {
775 		IP_VS_ERR("request control DEL for uncontrolled: "
776 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
777 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
778 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
779 		return;
780 	}
781 
782 	IP_VS_DBG(7, "DELeting control for: "
783 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
784 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
785 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
786 
787 	cp->control = NULL;
788 	if (atomic_read(&ctl_cp->n_control) == 0) {
789 		IP_VS_ERR("BUG control DEL with n=0 : "
790 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
791 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
792 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
793 		return;
794 	}
795 	atomic_dec(&ctl_cp->n_control);
796 }
797 
798 static inline void
799 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
800 {
801 	if (cp->control) {
802 		IP_VS_ERR("request control ADD for already controlled: "
803 			  "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
804 			  NIPQUAD(cp->caddr),ntohs(cp->cport),
805 			  NIPQUAD(cp->vaddr),ntohs(cp->vport));
806 		ip_vs_control_del(cp);
807 	}
808 
809 	IP_VS_DBG(7, "ADDing control for: "
810 		  "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
811 		  NIPQUAD(cp->caddr),ntohs(cp->cport),
812 		  NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
813 
814 	cp->control = ctl_cp;
815 	atomic_inc(&ctl_cp->n_control);
816 }
817 
818 
819 /*
820  *      IPVS application functions
821  *      (from ip_vs_app.c)
822  */
823 #define IP_VS_APP_MAX_PORTS  8
824 extern int register_ip_vs_app(struct ip_vs_app *app);
825 extern void unregister_ip_vs_app(struct ip_vs_app *app);
826 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
827 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
828 extern int
829 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
830 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
831 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
832 
833 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff **pskb);
834 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff **pskb);
835 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
836 			     char *o_buf, int o_len, char *n_buf, int n_len);
837 extern int ip_vs_app_init(void);
838 extern void ip_vs_app_cleanup(void);
839 
840 
841 /*
842  *	IPVS protocol functions (from ip_vs_proto.c)
843  */
844 extern int ip_vs_protocol_init(void);
845 extern void ip_vs_protocol_cleanup(void);
846 extern void ip_vs_protocol_timeout_change(int flags);
847 extern int *ip_vs_create_timeout_table(int *table, int size);
848 extern int
849 ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to);
850 extern void
851 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
852 			  int offset, const char *msg);
853 
854 extern struct ip_vs_protocol ip_vs_protocol_tcp;
855 extern struct ip_vs_protocol ip_vs_protocol_udp;
856 extern struct ip_vs_protocol ip_vs_protocol_icmp;
857 extern struct ip_vs_protocol ip_vs_protocol_esp;
858 extern struct ip_vs_protocol ip_vs_protocol_ah;
859 
860 
861 /*
862  *      Registering/unregistering scheduler functions
863  *      (from ip_vs_sched.c)
864  */
865 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
866 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
867 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
868 				struct ip_vs_scheduler *scheduler);
869 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
870 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
871 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
872 extern struct ip_vs_conn *
873 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
874 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
875 			struct ip_vs_protocol *pp);
876 
877 
878 /*
879  *      IPVS control data and functions (from ip_vs_ctl.c)
880  */
881 extern int sysctl_ip_vs_cache_bypass;
882 extern int sysctl_ip_vs_expire_nodest_conn;
883 extern int sysctl_ip_vs_expire_quiescent_template;
884 extern int sysctl_ip_vs_sync_threshold[2];
885 extern int sysctl_ip_vs_nat_icmp_send;
886 extern struct ip_vs_stats ip_vs_stats;
887 
888 extern struct ip_vs_service *
889 ip_vs_service_get(__u32 fwmark, __u16 protocol, __u32 vaddr, __u16 vport);
890 
891 static inline void ip_vs_service_put(struct ip_vs_service *svc)
892 {
893 	atomic_dec(&svc->usecnt);
894 }
895 
896 extern struct ip_vs_dest *
897 ip_vs_lookup_real_service(__u16 protocol, __u32 daddr, __u16 dport);
898 extern int ip_vs_use_count_inc(void);
899 extern void ip_vs_use_count_dec(void);
900 extern int ip_vs_control_init(void);
901 extern void ip_vs_control_cleanup(void);
902 
903 
904 /*
905  *      IPVS sync daemon data and function prototypes
906  *      (from ip_vs_sync.c)
907  */
908 extern volatile int ip_vs_sync_state;
909 extern volatile int ip_vs_master_syncid;
910 extern volatile int ip_vs_backup_syncid;
911 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
912 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
913 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
914 extern int stop_sync_thread(int state);
915 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
916 
917 
918 /*
919  *      IPVS rate estimator prototypes (from ip_vs_est.c)
920  */
921 extern int ip_vs_new_estimator(struct ip_vs_stats *stats);
922 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
923 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
924 
925 /*
926  *	Various IPVS packet transmitters (from ip_vs_xmit.c)
927  */
928 extern int ip_vs_null_xmit
929 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
930 extern int ip_vs_bypass_xmit
931 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
932 extern int ip_vs_nat_xmit
933 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
934 extern int ip_vs_tunnel_xmit
935 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
936 extern int ip_vs_dr_xmit
937 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
938 extern int ip_vs_icmp_xmit
939 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
940 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
941 
942 
943 /*
944  *	This is a simple mechanism to ignore packets when
945  *	we are loaded. Just set ip_vs_drop_rate to 'n' and
946  *	we start to drop 1/rate of the packets
947  */
948 extern int ip_vs_drop_rate;
949 extern int ip_vs_drop_counter;
950 
951 static __inline__ int ip_vs_todrop(void)
952 {
953 	if (!ip_vs_drop_rate) return 0;
954 	if (--ip_vs_drop_counter > 0) return 0;
955 	ip_vs_drop_counter = ip_vs_drop_rate;
956 	return 1;
957 }
958 
959 /*
960  *      ip_vs_fwd_tag returns the forwarding tag of the connection
961  */
962 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
963 
964 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
965 {
966 	char fwd;
967 
968 	switch (IP_VS_FWD_METHOD(cp)) {
969 	case IP_VS_CONN_F_MASQ:
970 		fwd = 'M'; break;
971 	case IP_VS_CONN_F_LOCALNODE:
972 		fwd = 'L'; break;
973 	case IP_VS_CONN_F_TUNNEL:
974 		fwd = 'T'; break;
975 	case IP_VS_CONN_F_DROUTE:
976 		fwd = 'R'; break;
977 	case IP_VS_CONN_F_BYPASS:
978 		fwd = 'B'; break;
979 	default:
980 		fwd = '?'; break;
981 	}
982 	return fwd;
983 }
984 
985 extern int ip_vs_make_skb_writable(struct sk_buff **pskb, int len);
986 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
987 		struct ip_vs_conn *cp, int dir);
988 
989 extern u16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
990 
991 static inline u16 ip_vs_check_diff(u32 old, u32 new, u16 oldsum)
992 {
993 	u32 diff[2] = { old, new };
994 
995 	return csum_fold(csum_partial((char *) diff, sizeof(diff),
996 				      oldsum ^ 0xFFFF));
997 }
998 
999 #endif /* __KERNEL__ */
1000 
1001 #endif	/* _IP_VS_H */
1002