3 * data structure and functionality definitions
9 #include <linux/ip_vs.h> /* definitions shared with userland */
11 /* old ipvsadm versions still include this file directly */
14 #include <asm/types.h> /* for __uXX types */
16 #include <linux/sysctl.h> /* for ctl_path */
17 #include <linux/list.h> /* for struct list_head */
18 #include <linux/spinlock.h> /* for struct rwlock_t */
19 #include <asm/atomic.h> /* for struct atomic_t */
20 #include <linux/compiler.h>
21 #include <linux/timer.h>
23 #include <net/checksum.h>
24 #include <linux/netfilter.h> /* for union nf_inet_addr */
25 #include <linux/ipv6.h> /* for struct ipv6hdr */
26 #include <net/ipv6.h> /* for ipv6_addr_copy */
31 union nf_inet_addr saddr;
32 union nf_inet_addr daddr;
36 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
38 #ifdef CONFIG_IP_VS_IPV6
40 const struct ipv6hdr *iph = nh;
41 iphdr->len = sizeof(struct ipv6hdr);
42 iphdr->protocol = iph->nexthdr;
43 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
44 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
48 const struct iphdr *iph = nh;
49 iphdr->len = iph->ihl * 4;
50 iphdr->protocol = iph->protocol;
51 iphdr->saddr.ip = iph->saddr;
52 iphdr->daddr.ip = iph->daddr;
56 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
57 const union nf_inet_addr *src)
59 #ifdef CONFIG_IP_VS_IPV6
61 ipv6_addr_copy(&dst->in6, &src->in6);
67 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
68 const union nf_inet_addr *b)
70 #ifdef CONFIG_IP_VS_IPV6
72 return ipv6_addr_equal(&a->in6, &b->in6);
74 return a->ip == b->ip;
77 #ifdef CONFIG_IP_VS_DEBUG
78 #include <linux/net.h>
80 extern int ip_vs_get_debug_level(void);
82 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
83 const union nf_inet_addr *addr,
87 #ifdef CONFIG_IP_VS_IPV6
89 len = snprintf(&buf[*idx], buf_len - *idx, "[" NIP6_FMT "]",
93 len = snprintf(&buf[*idx], buf_len - *idx, NIPQUAD_FMT,
94 NIPQUAD(addr->ip)) + 1;
97 BUG_ON(*idx > buf_len + 1);
98 return &buf[*idx - len];
101 #define IP_VS_DBG_BUF(level, msg...) \
103 char ip_vs_dbg_buf[160]; \
104 int ip_vs_dbg_idx = 0; \
105 if (level <= ip_vs_get_debug_level()) \
106 printk(KERN_DEBUG "IPVS: " msg); \
108 #define IP_VS_ERR_BUF(msg...) \
110 char ip_vs_dbg_buf[160]; \
111 int ip_vs_dbg_idx = 0; \
112 printk(KERN_ERR "IPVS: " msg); \
115 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
116 #define IP_VS_DBG_ADDR(af, addr) \
117 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
118 sizeof(ip_vs_dbg_buf), addr, \
121 #define IP_VS_DBG(level, msg...) \
123 if (level <= ip_vs_get_debug_level()) \
124 printk(KERN_DEBUG "IPVS: " msg); \
126 #define IP_VS_DBG_RL(msg...) \
128 if (net_ratelimit()) \
129 printk(KERN_DEBUG "IPVS: " msg); \
131 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) \
133 if (level <= ip_vs_get_debug_level()) \
134 pp->debug_packet(pp, skb, ofs, msg); \
136 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) \
138 if (level <= ip_vs_get_debug_level() && \
140 pp->debug_packet(pp, skb, ofs, msg); \
142 #else /* NO DEBUGGING at ALL */
143 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
144 #define IP_VS_ERR_BUF(msg...) do {} while (0)
145 #define IP_VS_DBG(level, msg...) do {} while (0)
146 #define IP_VS_DBG_RL(msg...) do {} while (0)
147 #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) do {} while (0)
148 #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) do {} while (0)
151 #define IP_VS_BUG() BUG()
152 #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg)
153 #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg)
154 #define IP_VS_WARNING(msg...) \
155 printk(KERN_WARNING "IPVS: " msg)
156 #define IP_VS_ERR_RL(msg...) \
158 if (net_ratelimit()) \
159 printk(KERN_ERR "IPVS: " msg); \
162 #ifdef CONFIG_IP_VS_DEBUG
163 #define EnterFunction(level) \
165 if (level <= ip_vs_get_debug_level()) \
166 printk(KERN_DEBUG "Enter: %s, %s line %i\n", \
167 __FUNCTION__, __FILE__, __LINE__); \
169 #define LeaveFunction(level) \
171 if (level <= ip_vs_get_debug_level()) \
172 printk(KERN_DEBUG "Leave: %s, %s line %i\n", \
173 __FUNCTION__, __FILE__, __LINE__); \
176 #define EnterFunction(level) do {} while (0)
177 #define LeaveFunction(level) do {} while (0)
180 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
184 * The port number of FTP service (in network order).
186 #define FTPPORT __constant_htons(21)
187 #define FTPDATA __constant_htons(20)
193 IP_VS_TCP_S_NONE = 0,
194 IP_VS_TCP_S_ESTABLISHED,
195 IP_VS_TCP_S_SYN_SENT,
196 IP_VS_TCP_S_SYN_RECV,
197 IP_VS_TCP_S_FIN_WAIT,
198 IP_VS_TCP_S_TIME_WAIT,
200 IP_VS_TCP_S_CLOSE_WAIT,
201 IP_VS_TCP_S_LAST_ACK,
224 * Delta sequence info structure
225 * Each ip_vs_conn has 2 (output AND input seq. changes).
226 * Only used in the VS/NAT.
229 __u32 init_seq; /* Add delta from this seq */
230 __u32 delta; /* Delta in sequence numbers */
231 __u32 previous_delta; /* Delta in sequence numbers
232 before last resized pkt */
237 * IPVS statistics objects
239 struct ip_vs_estimator {
240 struct list_head list;
257 __u32 conns; /* connections scheduled */
258 __u32 inpkts; /* incoming packets */
259 __u32 outpkts; /* outgoing packets */
260 __u64 inbytes; /* incoming bytes */
261 __u64 outbytes; /* outgoing bytes */
263 __u32 cps; /* current connection rate */
264 __u32 inpps; /* current in packet rate */
265 __u32 outpps; /* current out packet rate */
266 __u32 inbps; /* current in byte rate */
267 __u32 outbps; /* current out byte rate */
270 * Don't add anything before the lock, because we use memcpy() to copy
271 * the members before the lock to struct ip_vs_stats_user in
275 spinlock_t lock; /* spin lock */
277 struct ip_vs_estimator est; /* estimator */
286 struct ip_vs_protocol {
287 struct ip_vs_protocol *next;
292 atomic_t appcnt; /* counter of proto app incs */
293 int *timeout_table; /* protocol timeout table */
295 void (*init)(struct ip_vs_protocol *pp);
297 void (*exit)(struct ip_vs_protocol *pp);
299 int (*conn_schedule)(struct sk_buff *skb,
300 struct ip_vs_protocol *pp,
301 int *verdict, struct ip_vs_conn **cpp);
304 (*conn_in_get)(const struct sk_buff *skb,
305 struct ip_vs_protocol *pp,
306 const struct iphdr *iph,
307 unsigned int proto_off,
311 (*conn_out_get)(const struct sk_buff *skb,
312 struct ip_vs_protocol *pp,
313 const struct iphdr *iph,
314 unsigned int proto_off,
317 int (*snat_handler)(struct sk_buff *skb,
318 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
320 int (*dnat_handler)(struct sk_buff *skb,
321 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
323 int (*csum_check)(struct sk_buff *skb, struct ip_vs_protocol *pp);
325 const char *(*state_name)(int state);
327 int (*state_transition)(struct ip_vs_conn *cp, int direction,
328 const struct sk_buff *skb,
329 struct ip_vs_protocol *pp);
331 int (*register_app)(struct ip_vs_app *inc);
333 void (*unregister_app)(struct ip_vs_app *inc);
335 int (*app_conn_bind)(struct ip_vs_conn *cp);
337 void (*debug_packet)(struct ip_vs_protocol *pp,
338 const struct sk_buff *skb,
342 void (*timeout_change)(struct ip_vs_protocol *pp, int flags);
344 int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to);
347 extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto);
350 * IP_VS structure allocated for each dynamically scheduled connection
353 struct list_head c_list; /* hashed list heads */
355 /* Protocol, addresses and port numbers */
356 u16 af; /* address family */
357 union nf_inet_addr caddr; /* client address */
358 union nf_inet_addr vaddr; /* virtual address */
359 union nf_inet_addr daddr; /* destination address */
363 __u16 protocol; /* Which protocol (TCP/UDP) */
365 /* counter and timer */
366 atomic_t refcnt; /* reference count */
367 struct timer_list timer; /* Expiration timer */
368 volatile unsigned long timeout; /* timeout */
370 /* Flags and state transition */
371 spinlock_t lock; /* lock for state transition */
372 volatile __u16 flags; /* status flags */
373 volatile __u16 state; /* state info */
374 volatile __u16 old_state; /* old state, to be used for
375 * state transition triggerd
379 /* Control members */
380 struct ip_vs_conn *control; /* Master control connection */
381 atomic_t n_control; /* Number of controlled ones */
382 struct ip_vs_dest *dest; /* real server */
383 atomic_t in_pkts; /* incoming packet counter */
385 /* packet transmitter for different forwarding methods. If it
386 mangles the packet, it must return NF_DROP or better NF_STOLEN,
387 otherwise this must be changed to a sk_buff **.
389 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
390 struct ip_vs_protocol *pp);
392 /* Note: we can group the following members into a structure,
393 in order to save more space, and the following members are
394 only used in VS/NAT anyway */
395 struct ip_vs_app *app; /* bound ip_vs_app object */
396 void *app_data; /* Application private data */
397 struct ip_vs_seq in_seq; /* incoming seq. struct */
398 struct ip_vs_seq out_seq; /* outgoing seq. struct */
403 * Extended internal versions of struct ip_vs_service_user and
404 * ip_vs_dest_user for IPv6 support.
406 * We need these to conveniently pass around service and destination
407 * options, but unfortunately, we also need to keep the old definitions to
408 * maintain userspace backwards compatibility for the setsockopt interface.
410 struct ip_vs_service_user_kern {
411 /* virtual service addresses */
414 union nf_inet_addr addr; /* virtual ip address */
416 u32 fwmark; /* firwall mark of service */
418 /* virtual service options */
420 unsigned flags; /* virtual service flags */
421 unsigned timeout; /* persistent timeout in sec */
422 u32 netmask; /* persistent netmask */
426 struct ip_vs_dest_user_kern {
427 /* destination server address */
428 union nf_inet_addr addr;
431 /* real server options */
432 unsigned conn_flags; /* connection flags */
433 int weight; /* destination weight */
435 /* thresholds for active connections */
436 u32 u_threshold; /* upper threshold */
437 u32 l_threshold; /* lower threshold */
442 * The information about the virtual service offered to the net
443 * and the forwarding entries
445 struct ip_vs_service {
446 struct list_head s_list; /* for normal service table */
447 struct list_head f_list; /* for fwmark-based service table */
448 atomic_t refcnt; /* reference counter */
449 atomic_t usecnt; /* use counter */
451 u16 af; /* address family */
452 __u16 protocol; /* which protocol (TCP/UDP) */
453 union nf_inet_addr addr; /* IP address for virtual service */
454 __be16 port; /* port number for the service */
455 __u32 fwmark; /* firewall mark of the service */
456 unsigned flags; /* service status flags */
457 unsigned timeout; /* persistent timeout in ticks */
458 __be32 netmask; /* grouping granularity */
460 struct list_head destinations; /* real server d-linked list */
461 __u32 num_dests; /* number of servers */
462 struct ip_vs_stats stats; /* statistics for the service */
463 struct ip_vs_app *inc; /* bind conns to this app inc */
466 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
467 rwlock_t sched_lock; /* lock sched_data */
468 void *sched_data; /* scheduler application data */
473 * The real server destination forwarding entry
474 * with ip address, port number, and so on.
477 struct list_head n_list; /* for the dests in the service */
478 struct list_head d_list; /* for table with all the dests */
480 u16 af; /* address family */
481 union nf_inet_addr addr; /* IP address of the server */
482 __be16 port; /* port number of the server */
483 volatile unsigned flags; /* dest status flags */
484 atomic_t conn_flags; /* flags to copy to conn */
485 atomic_t weight; /* server weight */
487 atomic_t refcnt; /* reference counter */
488 struct ip_vs_stats stats; /* statistics */
490 /* connection counters and thresholds */
491 atomic_t activeconns; /* active connections */
492 atomic_t inactconns; /* inactive connections */
493 atomic_t persistconns; /* persistent connections */
494 __u32 u_threshold; /* upper threshold */
495 __u32 l_threshold; /* lower threshold */
497 /* for destination cache */
498 spinlock_t dst_lock; /* lock of dst_cache */
499 struct dst_entry *dst_cache; /* destination cache entry */
500 u32 dst_rtos; /* RT_TOS(tos) for dst */
502 /* for virtual service */
503 struct ip_vs_service *svc; /* service it belongs to */
504 __u16 protocol; /* which protocol (TCP/UDP) */
505 union nf_inet_addr vaddr; /* virtual IP address */
506 __be16 vport; /* virtual port number */
507 __u32 vfwmark; /* firewall mark of service */
512 * The scheduler object
514 struct ip_vs_scheduler {
515 struct list_head n_list; /* d-linked list head */
516 char *name; /* scheduler name */
517 atomic_t refcnt; /* reference counter */
518 struct module *module; /* THIS_MODULE/NULL */
520 /* scheduler initializing service */
521 int (*init_service)(struct ip_vs_service *svc);
522 /* scheduling service finish */
523 int (*done_service)(struct ip_vs_service *svc);
524 /* scheduler updating service */
525 int (*update_service)(struct ip_vs_service *svc);
527 /* selecting a server from the given service */
528 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
529 const struct sk_buff *skb);
534 * The application module object (a.k.a. app incarnation)
538 struct list_head a_list; /* member in app list */
539 int type; /* IP_VS_APP_TYPE_xxx */
540 char *name; /* application module name */
542 struct module *module; /* THIS_MODULE/NULL */
543 struct list_head incs_list; /* list of incarnations */
545 /* members for application incarnations */
546 struct list_head p_list; /* member in proto app list */
547 struct ip_vs_app *app; /* its real application */
548 __be16 port; /* port number in net order */
549 atomic_t usecnt; /* usage counter */
551 /* output hook: return false if can't linearize. diff set for TCP. */
552 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
553 struct sk_buff *, int *diff);
555 /* input hook: return false if can't linearize. diff set for TCP. */
556 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
557 struct sk_buff *, int *diff);
559 /* ip_vs_app initializer */
560 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
562 /* ip_vs_app finish */
563 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
567 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
568 struct ip_vs_protocol *);
570 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
576 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
577 int *verdict, struct ip_vs_conn **cpp);
580 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
581 const struct iphdr *iph, unsigned int proto_off,
585 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
586 const struct iphdr *iph, unsigned int proto_off,
589 int (*state_transition)(struct ip_vs_conn *cp, int direction,
590 const struct sk_buff *skb,
591 struct ip_vs_app *app);
593 void (*timeout_change)(struct ip_vs_app *app, int flags);
598 * IPVS core functions
599 * (from ip_vs_core.c)
601 extern const char *ip_vs_proto_name(unsigned proto);
602 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
603 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
605 #define IP_VS_APP_TYPE_FTP 1
608 * ip_vs_conn handling functions
609 * (from ip_vs_conn.c)
613 * IPVS connection entry hash table
615 #ifndef CONFIG_IP_VS_TAB_BITS
616 #define CONFIG_IP_VS_TAB_BITS 12
618 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
619 #if CONFIG_IP_VS_TAB_BITS < 8
620 #define IP_VS_CONN_TAB_BITS 8
622 #if CONFIG_IP_VS_TAB_BITS > 20
623 #define IP_VS_CONN_TAB_BITS 20
625 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
626 #define IP_VS_CONN_TAB_BITS CONFIG_IP_VS_TAB_BITS
628 #define IP_VS_CONN_TAB_SIZE (1 << IP_VS_CONN_TAB_BITS)
629 #define IP_VS_CONN_TAB_MASK (IP_VS_CONN_TAB_SIZE - 1)
634 IP_VS_DIR_INPUT_ONLY,
638 extern struct ip_vs_conn *ip_vs_conn_in_get
639 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
640 extern struct ip_vs_conn *ip_vs_ct_in_get
641 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
642 extern struct ip_vs_conn *ip_vs_conn_out_get
643 (int protocol, __be32 s_addr, __be16 s_port, __be32 d_addr, __be16 d_port);
645 /* put back the conn without restarting its timer */
646 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
648 atomic_dec(&cp->refcnt);
650 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
651 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
653 extern struct ip_vs_conn *
654 ip_vs_conn_new(int proto, __be32 caddr, __be16 cport, __be32 vaddr, __be16 vport,
655 __be32 daddr, __be16 dport, unsigned flags,
656 struct ip_vs_dest *dest);
657 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
659 extern const char * ip_vs_state_name(__u16 proto, int state);
661 extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
662 extern int ip_vs_check_template(struct ip_vs_conn *ct);
663 extern void ip_vs_random_dropentry(void);
664 extern int ip_vs_conn_init(void);
665 extern void ip_vs_conn_cleanup(void);
667 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
669 struct ip_vs_conn *ctl_cp = cp->control;
671 IP_VS_ERR("request control DEL for uncontrolled: "
672 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
673 NIPQUAD(cp->caddr),ntohs(cp->cport),
674 NIPQUAD(cp->vaddr),ntohs(cp->vport));
678 IP_VS_DBG(7, "DELeting control for: "
679 "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
680 NIPQUAD(cp->caddr),ntohs(cp->cport),
681 NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
684 if (atomic_read(&ctl_cp->n_control) == 0) {
685 IP_VS_ERR("BUG control DEL with n=0 : "
686 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
687 NIPQUAD(cp->caddr),ntohs(cp->cport),
688 NIPQUAD(cp->vaddr),ntohs(cp->vport));
691 atomic_dec(&ctl_cp->n_control);
695 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
698 IP_VS_ERR("request control ADD for already controlled: "
699 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
700 NIPQUAD(cp->caddr),ntohs(cp->cport),
701 NIPQUAD(cp->vaddr),ntohs(cp->vport));
702 ip_vs_control_del(cp);
705 IP_VS_DBG(7, "ADDing control for: "
706 "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
707 NIPQUAD(cp->caddr),ntohs(cp->cport),
708 NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
710 cp->control = ctl_cp;
711 atomic_inc(&ctl_cp->n_control);
716 * IPVS application functions
719 #define IP_VS_APP_MAX_PORTS 8
720 extern int register_ip_vs_app(struct ip_vs_app *app);
721 extern void unregister_ip_vs_app(struct ip_vs_app *app);
722 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
723 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
725 register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port);
726 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
727 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
729 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
730 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
731 extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri,
732 char *o_buf, int o_len, char *n_buf, int n_len);
733 extern int ip_vs_app_init(void);
734 extern void ip_vs_app_cleanup(void);
738 * IPVS protocol functions (from ip_vs_proto.c)
740 extern int ip_vs_protocol_init(void);
741 extern void ip_vs_protocol_cleanup(void);
742 extern void ip_vs_protocol_timeout_change(int flags);
743 extern int *ip_vs_create_timeout_table(int *table, int size);
745 ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to);
747 ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb,
748 int offset, const char *msg);
750 extern struct ip_vs_protocol ip_vs_protocol_tcp;
751 extern struct ip_vs_protocol ip_vs_protocol_udp;
752 extern struct ip_vs_protocol ip_vs_protocol_icmp;
753 extern struct ip_vs_protocol ip_vs_protocol_esp;
754 extern struct ip_vs_protocol ip_vs_protocol_ah;
758 * Registering/unregistering scheduler functions
759 * (from ip_vs_sched.c)
761 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
762 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
763 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
764 struct ip_vs_scheduler *scheduler);
765 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
766 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
767 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
768 extern struct ip_vs_conn *
769 ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb);
770 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
771 struct ip_vs_protocol *pp);
775 * IPVS control data and functions (from ip_vs_ctl.c)
777 extern int sysctl_ip_vs_cache_bypass;
778 extern int sysctl_ip_vs_expire_nodest_conn;
779 extern int sysctl_ip_vs_expire_quiescent_template;
780 extern int sysctl_ip_vs_sync_threshold[2];
781 extern int sysctl_ip_vs_nat_icmp_send;
782 extern struct ip_vs_stats ip_vs_stats;
783 extern const struct ctl_path net_vs_ctl_path[];
785 extern struct ip_vs_service *
786 ip_vs_service_get(__u32 fwmark, __u16 protocol, __be32 vaddr, __be16 vport);
788 static inline void ip_vs_service_put(struct ip_vs_service *svc)
790 atomic_dec(&svc->usecnt);
793 extern struct ip_vs_dest *
794 ip_vs_lookup_real_service(__u16 protocol, __be32 daddr, __be16 dport);
795 extern int ip_vs_use_count_inc(void);
796 extern void ip_vs_use_count_dec(void);
797 extern int ip_vs_control_init(void);
798 extern void ip_vs_control_cleanup(void);
799 extern struct ip_vs_dest *
800 ip_vs_find_dest(__be32 daddr, __be16 dport,
801 __be32 vaddr, __be16 vport, __u16 protocol);
802 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
806 * IPVS sync daemon data and function prototypes
807 * (from ip_vs_sync.c)
809 extern volatile int ip_vs_sync_state;
810 extern volatile int ip_vs_master_syncid;
811 extern volatile int ip_vs_backup_syncid;
812 extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
813 extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
814 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
815 extern int stop_sync_thread(int state);
816 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
820 * IPVS rate estimator prototypes (from ip_vs_est.c)
822 extern int ip_vs_estimator_init(void);
823 extern void ip_vs_estimator_cleanup(void);
824 extern void ip_vs_new_estimator(struct ip_vs_stats *stats);
825 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
826 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
829 * Various IPVS packet transmitters (from ip_vs_xmit.c)
831 extern int ip_vs_null_xmit
832 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
833 extern int ip_vs_bypass_xmit
834 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
835 extern int ip_vs_nat_xmit
836 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
837 extern int ip_vs_tunnel_xmit
838 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
839 extern int ip_vs_dr_xmit
840 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
841 extern int ip_vs_icmp_xmit
842 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
843 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
847 * This is a simple mechanism to ignore packets when
848 * we are loaded. Just set ip_vs_drop_rate to 'n' and
849 * we start to drop 1/rate of the packets
851 extern int ip_vs_drop_rate;
852 extern int ip_vs_drop_counter;
854 static __inline__ int ip_vs_todrop(void)
856 if (!ip_vs_drop_rate) return 0;
857 if (--ip_vs_drop_counter > 0) return 0;
858 ip_vs_drop_counter = ip_vs_drop_rate;
863 * ip_vs_fwd_tag returns the forwarding tag of the connection
865 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
867 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
871 switch (IP_VS_FWD_METHOD(cp)) {
872 case IP_VS_CONN_F_MASQ:
874 case IP_VS_CONN_F_LOCALNODE:
876 case IP_VS_CONN_F_TUNNEL:
878 case IP_VS_CONN_F_DROUTE:
880 case IP_VS_CONN_F_BYPASS:
888 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
889 struct ip_vs_conn *cp, int dir);
891 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
893 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
895 __be32 diff[2] = { ~old, new };
897 return csum_partial((char *) diff, sizeof(diff), oldsum);
900 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
902 __be16 diff[2] = { ~old, new };
904 return csum_partial((char *) diff, sizeof(diff), oldsum);
907 #endif /* __KERNEL__ */
909 #endif /* _NET_IP_VS_H */