Loading...
1/*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
6#ifndef _NET_IP_VS_H
7#define _NET_IP_VS_H
8
9#include <linux/ip_vs.h> /* definitions shared with userland */
10
11#include <asm/types.h> /* for __uXX types */
12
13#include <linux/sysctl.h> /* for ctl_path */
14#include <linux/list.h> /* for struct list_head */
15#include <linux/spinlock.h> /* for struct rwlock_t */
16#include <linux/atomic.h> /* for struct atomic_t */
17#include <linux/compiler.h>
18#include <linux/timer.h>
19
20#include <net/checksum.h>
21#include <linux/netfilter.h> /* for union nf_inet_addr */
22#include <linux/ip.h>
23#include <linux/ipv6.h> /* for struct ipv6hdr */
24#include <net/ipv6.h> /* for ipv6_addr_copy */
25#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
26#include <net/netfilter/nf_conntrack.h>
27#endif
28#include <net/net_namespace.h> /* Netw namespace */
29
30/*
31 * Generic access of ipvs struct
32 */
33static inline struct netns_ipvs *net_ipvs(struct net* net)
34{
35 return net->ipvs;
36}
37/*
38 * Get net ptr from skb in traffic cases
39 * use skb_sknet when call is from userland (ioctl or netlink)
40 */
41static inline struct net *skb_net(const struct sk_buff *skb)
42{
43#ifdef CONFIG_NET_NS
44#ifdef CONFIG_IP_VS_DEBUG
45 /*
46 * This is used for debug only.
47 * Start with the most likely hit
48 * End with BUG
49 */
50 if (likely(skb->dev && skb->dev->nd_net))
51 return dev_net(skb->dev);
52 if (skb_dst(skb) && skb_dst(skb)->dev)
53 return dev_net(skb_dst(skb)->dev);
54 WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
55 __func__, __LINE__);
56 if (likely(skb->sk && skb->sk->sk_net))
57 return sock_net(skb->sk);
58 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
59 __func__, __LINE__);
60 BUG();
61#else
62 return dev_net(skb->dev ? : skb_dst(skb)->dev);
63#endif
64#else
65 return &init_net;
66#endif
67}
68
69static inline struct net *skb_sknet(const struct sk_buff *skb)
70{
71#ifdef CONFIG_NET_NS
72#ifdef CONFIG_IP_VS_DEBUG
73 /* Start with the most likely hit */
74 if (likely(skb->sk && skb->sk->sk_net))
75 return sock_net(skb->sk);
76 WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
77 __func__, __LINE__);
78 if (likely(skb->dev && skb->dev->nd_net))
79 return dev_net(skb->dev);
80 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
81 __func__, __LINE__);
82 BUG();
83#else
84 return sock_net(skb->sk);
85#endif
86#else
87 return &init_net;
88#endif
89}
90/*
91 * This one needed for single_open_net since net is stored directly in
92 * private not as a struct i.e. seq_file_net can't be used.
93 */
94static inline struct net *seq_file_single_net(struct seq_file *seq)
95{
96#ifdef CONFIG_NET_NS
97 return (struct net *)seq->private;
98#else
99 return &init_net;
100#endif
101}
102
103/* Connections' size value needed by ip_vs_ctl.c */
104extern int ip_vs_conn_tab_size;
105
106
107struct ip_vs_iphdr {
108 int len;
109 __u8 protocol;
110 union nf_inet_addr saddr;
111 union nf_inet_addr daddr;
112};
113
114static inline void
115ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
116{
117#ifdef CONFIG_IP_VS_IPV6
118 if (af == AF_INET6) {
119 const struct ipv6hdr *iph = nh;
120 iphdr->len = sizeof(struct ipv6hdr);
121 iphdr->protocol = iph->nexthdr;
122 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
123 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
124 } else
125#endif
126 {
127 const struct iphdr *iph = nh;
128 iphdr->len = iph->ihl * 4;
129 iphdr->protocol = iph->protocol;
130 iphdr->saddr.ip = iph->saddr;
131 iphdr->daddr.ip = iph->daddr;
132 }
133}
134
135static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
136 const union nf_inet_addr *src)
137{
138#ifdef CONFIG_IP_VS_IPV6
139 if (af == AF_INET6)
140 ipv6_addr_copy(&dst->in6, &src->in6);
141 else
142#endif
143 dst->ip = src->ip;
144}
145
146static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
147 const union nf_inet_addr *b)
148{
149#ifdef CONFIG_IP_VS_IPV6
150 if (af == AF_INET6)
151 return ipv6_addr_equal(&a->in6, &b->in6);
152#endif
153 return a->ip == b->ip;
154}
155
156#ifdef CONFIG_IP_VS_DEBUG
157#include <linux/net.h>
158
159extern int ip_vs_get_debug_level(void);
160
161static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
162 const union nf_inet_addr *addr,
163 int *idx)
164{
165 int len;
166#ifdef CONFIG_IP_VS_IPV6
167 if (af == AF_INET6)
168 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
169 &addr->in6) + 1;
170 else
171#endif
172 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
173 &addr->ip) + 1;
174
175 *idx += len;
176 BUG_ON(*idx > buf_len + 1);
177 return &buf[*idx - len];
178}
179
180#define IP_VS_DBG_BUF(level, msg, ...) \
181 do { \
182 char ip_vs_dbg_buf[160]; \
183 int ip_vs_dbg_idx = 0; \
184 if (level <= ip_vs_get_debug_level()) \
185 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
186 } while (0)
187#define IP_VS_ERR_BUF(msg...) \
188 do { \
189 char ip_vs_dbg_buf[160]; \
190 int ip_vs_dbg_idx = 0; \
191 pr_err(msg); \
192 } while (0)
193
194/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
195#define IP_VS_DBG_ADDR(af, addr) \
196 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
197 sizeof(ip_vs_dbg_buf), addr, \
198 &ip_vs_dbg_idx)
199
200#define IP_VS_DBG(level, msg, ...) \
201 do { \
202 if (level <= ip_vs_get_debug_level()) \
203 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
204 } while (0)
205#define IP_VS_DBG_RL(msg, ...) \
206 do { \
207 if (net_ratelimit()) \
208 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
209 } while (0)
210#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
211 do { \
212 if (level <= ip_vs_get_debug_level()) \
213 pp->debug_packet(af, pp, skb, ofs, msg); \
214 } while (0)
215#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
216 do { \
217 if (level <= ip_vs_get_debug_level() && \
218 net_ratelimit()) \
219 pp->debug_packet(af, pp, skb, ofs, msg); \
220 } while (0)
221#else /* NO DEBUGGING at ALL */
222#define IP_VS_DBG_BUF(level, msg...) do {} while (0)
223#define IP_VS_ERR_BUF(msg...) do {} while (0)
224#define IP_VS_DBG(level, msg...) do {} while (0)
225#define IP_VS_DBG_RL(msg...) do {} while (0)
226#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
227#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
228#endif
229
230#define IP_VS_BUG() BUG()
231#define IP_VS_ERR_RL(msg, ...) \
232 do { \
233 if (net_ratelimit()) \
234 pr_err(msg, ##__VA_ARGS__); \
235 } while (0)
236
237#ifdef CONFIG_IP_VS_DEBUG
238#define EnterFunction(level) \
239 do { \
240 if (level <= ip_vs_get_debug_level()) \
241 printk(KERN_DEBUG \
242 pr_fmt("Enter: %s, %s line %i\n"), \
243 __func__, __FILE__, __LINE__); \
244 } while (0)
245#define LeaveFunction(level) \
246 do { \
247 if (level <= ip_vs_get_debug_level()) \
248 printk(KERN_DEBUG \
249 pr_fmt("Leave: %s, %s line %i\n"), \
250 __func__, __FILE__, __LINE__); \
251 } while (0)
252#else
253#define EnterFunction(level) do {} while (0)
254#define LeaveFunction(level) do {} while (0)
255#endif
256
257#define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
258
259
260/*
261 * The port number of FTP service (in network order).
262 */
263#define FTPPORT cpu_to_be16(21)
264#define FTPDATA cpu_to_be16(20)
265
266/*
267 * TCP State Values
268 */
269enum {
270 IP_VS_TCP_S_NONE = 0,
271 IP_VS_TCP_S_ESTABLISHED,
272 IP_VS_TCP_S_SYN_SENT,
273 IP_VS_TCP_S_SYN_RECV,
274 IP_VS_TCP_S_FIN_WAIT,
275 IP_VS_TCP_S_TIME_WAIT,
276 IP_VS_TCP_S_CLOSE,
277 IP_VS_TCP_S_CLOSE_WAIT,
278 IP_VS_TCP_S_LAST_ACK,
279 IP_VS_TCP_S_LISTEN,
280 IP_VS_TCP_S_SYNACK,
281 IP_VS_TCP_S_LAST
282};
283
284/*
285 * UDP State Values
286 */
287enum {
288 IP_VS_UDP_S_NORMAL,
289 IP_VS_UDP_S_LAST,
290};
291
292/*
293 * ICMP State Values
294 */
295enum {
296 IP_VS_ICMP_S_NORMAL,
297 IP_VS_ICMP_S_LAST,
298};
299
300/*
301 * SCTP State Values
302 */
303enum ip_vs_sctp_states {
304 IP_VS_SCTP_S_NONE,
305 IP_VS_SCTP_S_INIT_CLI,
306 IP_VS_SCTP_S_INIT_SER,
307 IP_VS_SCTP_S_INIT_ACK_CLI,
308 IP_VS_SCTP_S_INIT_ACK_SER,
309 IP_VS_SCTP_S_ECHO_CLI,
310 IP_VS_SCTP_S_ECHO_SER,
311 IP_VS_SCTP_S_ESTABLISHED,
312 IP_VS_SCTP_S_SHUT_CLI,
313 IP_VS_SCTP_S_SHUT_SER,
314 IP_VS_SCTP_S_SHUT_ACK_CLI,
315 IP_VS_SCTP_S_SHUT_ACK_SER,
316 IP_VS_SCTP_S_CLOSED,
317 IP_VS_SCTP_S_LAST
318};
319
320/*
321 * Delta sequence info structure
322 * Each ip_vs_conn has 2 (output AND input seq. changes).
323 * Only used in the VS/NAT.
324 */
325struct ip_vs_seq {
326 __u32 init_seq; /* Add delta from this seq */
327 __u32 delta; /* Delta in sequence numbers */
328 __u32 previous_delta; /* Delta in sequence numbers
329 before last resized pkt */
330};
331
332/*
333 * counters per cpu
334 */
335struct ip_vs_counters {
336 __u32 conns; /* connections scheduled */
337 __u32 inpkts; /* incoming packets */
338 __u32 outpkts; /* outgoing packets */
339 __u64 inbytes; /* incoming bytes */
340 __u64 outbytes; /* outgoing bytes */
341};
342/*
343 * Stats per cpu
344 */
345struct ip_vs_cpu_stats {
346 struct ip_vs_counters ustats;
347 struct u64_stats_sync syncp;
348};
349
350/*
351 * IPVS statistics objects
352 */
353struct ip_vs_estimator {
354 struct list_head list;
355
356 u64 last_inbytes;
357 u64 last_outbytes;
358 u32 last_conns;
359 u32 last_inpkts;
360 u32 last_outpkts;
361
362 u32 cps;
363 u32 inpps;
364 u32 outpps;
365 u32 inbps;
366 u32 outbps;
367};
368
369struct ip_vs_stats {
370 struct ip_vs_stats_user ustats; /* statistics */
371 struct ip_vs_estimator est; /* estimator */
372 struct ip_vs_cpu_stats *cpustats; /* per cpu counters */
373 spinlock_t lock; /* spin lock */
374 struct ip_vs_stats_user ustats0; /* reset values */
375};
376
377struct dst_entry;
378struct iphdr;
379struct ip_vs_conn;
380struct ip_vs_app;
381struct sk_buff;
382struct ip_vs_proto_data;
383
384struct ip_vs_protocol {
385 struct ip_vs_protocol *next;
386 char *name;
387 u16 protocol;
388 u16 num_states;
389 int dont_defrag;
390
391 void (*init)(struct ip_vs_protocol *pp);
392
393 void (*exit)(struct ip_vs_protocol *pp);
394
395 void (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
396
397 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
398
399 int (*conn_schedule)(int af, struct sk_buff *skb,
400 struct ip_vs_proto_data *pd,
401 int *verdict, struct ip_vs_conn **cpp);
402
403 struct ip_vs_conn *
404 (*conn_in_get)(int af,
405 const struct sk_buff *skb,
406 const struct ip_vs_iphdr *iph,
407 unsigned int proto_off,
408 int inverse);
409
410 struct ip_vs_conn *
411 (*conn_out_get)(int af,
412 const struct sk_buff *skb,
413 const struct ip_vs_iphdr *iph,
414 unsigned int proto_off,
415 int inverse);
416
417 int (*snat_handler)(struct sk_buff *skb,
418 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
419
420 int (*dnat_handler)(struct sk_buff *skb,
421 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
422
423 int (*csum_check)(int af, struct sk_buff *skb,
424 struct ip_vs_protocol *pp);
425
426 const char *(*state_name)(int state);
427
428 int (*state_transition)(struct ip_vs_conn *cp, int direction,
429 const struct sk_buff *skb,
430 struct ip_vs_proto_data *pd);
431
432 int (*register_app)(struct net *net, struct ip_vs_app *inc);
433
434 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
435
436 int (*app_conn_bind)(struct ip_vs_conn *cp);
437
438 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
439 const struct sk_buff *skb,
440 int offset,
441 const char *msg);
442
443 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
444};
445
446/*
447 * protocol data per netns
448 */
449struct ip_vs_proto_data {
450 struct ip_vs_proto_data *next;
451 struct ip_vs_protocol *pp;
452 int *timeout_table; /* protocol timeout table */
453 atomic_t appcnt; /* counter of proto app incs. */
454 struct tcp_states_t *tcp_state_table;
455};
456
457extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
458extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
459 unsigned short proto);
460
461struct ip_vs_conn_param {
462 struct net *net;
463 const union nf_inet_addr *caddr;
464 const union nf_inet_addr *vaddr;
465 __be16 cport;
466 __be16 vport;
467 __u16 protocol;
468 u16 af;
469
470 const struct ip_vs_pe *pe;
471 char *pe_data;
472 __u8 pe_data_len;
473};
474
475/*
476 * IP_VS structure allocated for each dynamically scheduled connection
477 */
478struct ip_vs_conn {
479 struct hlist_node c_list; /* hashed list heads */
480#ifdef CONFIG_NET_NS
481 struct net *net; /* Name space */
482#endif
483 /* Protocol, addresses and port numbers */
484 u16 af; /* address family */
485 __be16 cport;
486 __be16 vport;
487 __be16 dport;
488 __u32 fwmark; /* Fire wall mark from skb */
489 union nf_inet_addr caddr; /* client address */
490 union nf_inet_addr vaddr; /* virtual address */
491 union nf_inet_addr daddr; /* destination address */
492 volatile __u32 flags; /* status flags */
493 __u16 protocol; /* Which protocol (TCP/UDP) */
494
495 /* counter and timer */
496 atomic_t refcnt; /* reference count */
497 struct timer_list timer; /* Expiration timer */
498 volatile unsigned long timeout; /* timeout */
499
500 /* Flags and state transition */
501 spinlock_t lock; /* lock for state transition */
502 volatile __u16 state; /* state info */
503 volatile __u16 old_state; /* old state, to be used for
504 * state transition triggerd
505 * synchronization
506 */
507
508 /* Control members */
509 struct ip_vs_conn *control; /* Master control connection */
510 atomic_t n_control; /* Number of controlled ones */
511 struct ip_vs_dest *dest; /* real server */
512 atomic_t in_pkts; /* incoming packet counter */
513
514 /* packet transmitter for different forwarding methods. If it
515 mangles the packet, it must return NF_DROP or better NF_STOLEN,
516 otherwise this must be changed to a sk_buff **.
517 NF_ACCEPT can be returned when destination is local.
518 */
519 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
520 struct ip_vs_protocol *pp);
521
522 /* Note: we can group the following members into a structure,
523 in order to save more space, and the following members are
524 only used in VS/NAT anyway */
525 struct ip_vs_app *app; /* bound ip_vs_app object */
526 void *app_data; /* Application private data */
527 struct ip_vs_seq in_seq; /* incoming seq. struct */
528 struct ip_vs_seq out_seq; /* outgoing seq. struct */
529
530 const struct ip_vs_pe *pe;
531 char *pe_data;
532 __u8 pe_data_len;
533};
534
535/*
536 * To save some memory in conn table when name space is disabled.
537 */
538static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
539{
540#ifdef CONFIG_NET_NS
541 return cp->net;
542#else
543 return &init_net;
544#endif
545}
546static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
547{
548#ifdef CONFIG_NET_NS
549 cp->net = net;
550#endif
551}
552
553static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
554 struct net *net)
555{
556#ifdef CONFIG_NET_NS
557 return cp->net == net;
558#else
559 return 1;
560#endif
561}
562
563/*
564 * Extended internal versions of struct ip_vs_service_user and
565 * ip_vs_dest_user for IPv6 support.
566 *
567 * We need these to conveniently pass around service and destination
568 * options, but unfortunately, we also need to keep the old definitions to
569 * maintain userspace backwards compatibility for the setsockopt interface.
570 */
571struct ip_vs_service_user_kern {
572 /* virtual service addresses */
573 u16 af;
574 u16 protocol;
575 union nf_inet_addr addr; /* virtual ip address */
576 u16 port;
577 u32 fwmark; /* firwall mark of service */
578
579 /* virtual service options */
580 char *sched_name;
581 char *pe_name;
582 unsigned flags; /* virtual service flags */
583 unsigned timeout; /* persistent timeout in sec */
584 u32 netmask; /* persistent netmask */
585};
586
587
588struct ip_vs_dest_user_kern {
589 /* destination server address */
590 union nf_inet_addr addr;
591 u16 port;
592
593 /* real server options */
594 unsigned conn_flags; /* connection flags */
595 int weight; /* destination weight */
596
597 /* thresholds for active connections */
598 u32 u_threshold; /* upper threshold */
599 u32 l_threshold; /* lower threshold */
600};
601
602
603/*
604 * The information about the virtual service offered to the net
605 * and the forwarding entries
606 */
607struct ip_vs_service {
608 struct list_head s_list; /* for normal service table */
609 struct list_head f_list; /* for fwmark-based service table */
610 atomic_t refcnt; /* reference counter */
611 atomic_t usecnt; /* use counter */
612
613 u16 af; /* address family */
614 __u16 protocol; /* which protocol (TCP/UDP) */
615 union nf_inet_addr addr; /* IP address for virtual service */
616 __be16 port; /* port number for the service */
617 __u32 fwmark; /* firewall mark of the service */
618 unsigned flags; /* service status flags */
619 unsigned timeout; /* persistent timeout in ticks */
620 __be32 netmask; /* grouping granularity */
621 struct net *net;
622
623 struct list_head destinations; /* real server d-linked list */
624 __u32 num_dests; /* number of servers */
625 struct ip_vs_stats stats; /* statistics for the service */
626 struct ip_vs_app *inc; /* bind conns to this app inc */
627
628 /* for scheduling */
629 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
630 rwlock_t sched_lock; /* lock sched_data */
631 void *sched_data; /* scheduler application data */
632
633 /* alternate persistence engine */
634 struct ip_vs_pe *pe;
635};
636
637
638/*
639 * The real server destination forwarding entry
640 * with ip address, port number, and so on.
641 */
642struct ip_vs_dest {
643 struct list_head n_list; /* for the dests in the service */
644 struct list_head d_list; /* for table with all the dests */
645
646 u16 af; /* address family */
647 __be16 port; /* port number of the server */
648 union nf_inet_addr addr; /* IP address of the server */
649 volatile unsigned flags; /* dest status flags */
650 atomic_t conn_flags; /* flags to copy to conn */
651 atomic_t weight; /* server weight */
652
653 atomic_t refcnt; /* reference counter */
654 struct ip_vs_stats stats; /* statistics */
655
656 /* connection counters and thresholds */
657 atomic_t activeconns; /* active connections */
658 atomic_t inactconns; /* inactive connections */
659 atomic_t persistconns; /* persistent connections */
660 __u32 u_threshold; /* upper threshold */
661 __u32 l_threshold; /* lower threshold */
662
663 /* for destination cache */
664 spinlock_t dst_lock; /* lock of dst_cache */
665 struct dst_entry *dst_cache; /* destination cache entry */
666 u32 dst_rtos; /* RT_TOS(tos) for dst */
667 u32 dst_cookie;
668 union nf_inet_addr dst_saddr;
669
670 /* for virtual service */
671 struct ip_vs_service *svc; /* service it belongs to */
672 __u16 protocol; /* which protocol (TCP/UDP) */
673 __be16 vport; /* virtual port number */
674 union nf_inet_addr vaddr; /* virtual IP address */
675 __u32 vfwmark; /* firewall mark of service */
676};
677
678
679/*
680 * The scheduler object
681 */
682struct ip_vs_scheduler {
683 struct list_head n_list; /* d-linked list head */
684 char *name; /* scheduler name */
685 atomic_t refcnt; /* reference counter */
686 struct module *module; /* THIS_MODULE/NULL */
687
688 /* scheduler initializing service */
689 int (*init_service)(struct ip_vs_service *svc);
690 /* scheduling service finish */
691 int (*done_service)(struct ip_vs_service *svc);
692 /* scheduler updating service */
693 int (*update_service)(struct ip_vs_service *svc);
694
695 /* selecting a server from the given service */
696 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
697 const struct sk_buff *skb);
698};
699
700/* The persistence engine object */
701struct ip_vs_pe {
702 struct list_head n_list; /* d-linked list head */
703 char *name; /* scheduler name */
704 atomic_t refcnt; /* reference counter */
705 struct module *module; /* THIS_MODULE/NULL */
706
707 /* get the connection template, if any */
708 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
709 bool (*ct_match)(const struct ip_vs_conn_param *p,
710 struct ip_vs_conn *ct);
711 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
712 bool inverse);
713 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
714};
715
716/*
717 * The application module object (a.k.a. app incarnation)
718 */
719struct ip_vs_app {
720 struct list_head a_list; /* member in app list */
721 int type; /* IP_VS_APP_TYPE_xxx */
722 char *name; /* application module name */
723 __u16 protocol;
724 struct module *module; /* THIS_MODULE/NULL */
725 struct list_head incs_list; /* list of incarnations */
726
727 /* members for application incarnations */
728 struct list_head p_list; /* member in proto app list */
729 struct ip_vs_app *app; /* its real application */
730 __be16 port; /* port number in net order */
731 atomic_t usecnt; /* usage counter */
732
733 /*
734 * output hook: Process packet in inout direction, diff set for TCP.
735 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
736 * 2=Mangled but checksum was not updated
737 */
738 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
739 struct sk_buff *, int *diff);
740
741 /*
742 * input hook: Process packet in outin direction, diff set for TCP.
743 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
744 * 2=Mangled but checksum was not updated
745 */
746 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
747 struct sk_buff *, int *diff);
748
749 /* ip_vs_app initializer */
750 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
751
752 /* ip_vs_app finish */
753 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
754
755
756 /* not used now */
757 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
758 struct ip_vs_protocol *);
759
760 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
761
762 int * timeout_table;
763 int * timeouts;
764 int timeouts_size;
765
766 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
767 int *verdict, struct ip_vs_conn **cpp);
768
769 struct ip_vs_conn *
770 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
771 const struct iphdr *iph, unsigned int proto_off,
772 int inverse);
773
774 struct ip_vs_conn *
775 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
776 const struct iphdr *iph, unsigned int proto_off,
777 int inverse);
778
779 int (*state_transition)(struct ip_vs_conn *cp, int direction,
780 const struct sk_buff *skb,
781 struct ip_vs_app *app);
782
783 void (*timeout_change)(struct ip_vs_app *app, int flags);
784};
785
786/* IPVS in network namespace */
787struct netns_ipvs {
788 int gen; /* Generation */
789 int enable; /* enable like nf_hooks do */
790 /*
791 * Hash table: for real service lookups
792 */
793 #define IP_VS_RTAB_BITS 4
794 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
795 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
796
797 struct list_head rs_table[IP_VS_RTAB_SIZE];
798 /* ip_vs_app */
799 struct list_head app_list;
800 /* ip_vs_ftp */
801 struct ip_vs_app *ftp_app;
802 /* ip_vs_proto */
803 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
804 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
805 /* ip_vs_proto_tcp */
806#ifdef CONFIG_IP_VS_PROTO_TCP
807 #define TCP_APP_TAB_BITS 4
808 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
809 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
810 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
811 spinlock_t tcp_app_lock;
812#endif
813 /* ip_vs_proto_udp */
814#ifdef CONFIG_IP_VS_PROTO_UDP
815 #define UDP_APP_TAB_BITS 4
816 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
817 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
818 struct list_head udp_apps[UDP_APP_TAB_SIZE];
819 spinlock_t udp_app_lock;
820#endif
821 /* ip_vs_proto_sctp */
822#ifdef CONFIG_IP_VS_PROTO_SCTP
823 #define SCTP_APP_TAB_BITS 4
824 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
825 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
826 /* Hash table for SCTP application incarnations */
827 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
828 spinlock_t sctp_app_lock;
829#endif
830 /* ip_vs_conn */
831 atomic_t conn_count; /* connection counter */
832
833 /* ip_vs_ctl */
834 struct ip_vs_stats tot_stats; /* Statistics & est. */
835
836 int num_services; /* no of virtual services */
837
838 rwlock_t rs_lock; /* real services table */
839 /* Trash for destinations */
840 struct list_head dest_trash;
841 /* Service counters */
842 atomic_t ftpsvc_counter;
843 atomic_t nullsvc_counter;
844
845#ifdef CONFIG_SYSCTL
846 /* 1/rate drop and drop-entry variables */
847 struct delayed_work defense_work; /* Work handler */
848 int drop_rate;
849 int drop_counter;
850 atomic_t dropentry;
851 /* locks in ctl.c */
852 spinlock_t dropentry_lock; /* drop entry handling */
853 spinlock_t droppacket_lock; /* drop packet handling */
854 spinlock_t securetcp_lock; /* state and timeout tables */
855
856 /* sys-ctl struct */
857 struct ctl_table_header *sysctl_hdr;
858 struct ctl_table *sysctl_tbl;
859#endif
860
861 /* sysctl variables */
862 int sysctl_amemthresh;
863 int sysctl_am_droprate;
864 int sysctl_drop_entry;
865 int sysctl_drop_packet;
866 int sysctl_secure_tcp;
867#ifdef CONFIG_IP_VS_NFCT
868 int sysctl_conntrack;
869#endif
870 int sysctl_snat_reroute;
871 int sysctl_sync_ver;
872 int sysctl_cache_bypass;
873 int sysctl_expire_nodest_conn;
874 int sysctl_expire_quiescent_template;
875 int sysctl_sync_threshold[2];
876 int sysctl_nat_icmp_send;
877
878 /* ip_vs_lblc */
879 int sysctl_lblc_expiration;
880 struct ctl_table_header *lblc_ctl_header;
881 struct ctl_table *lblc_ctl_table;
882 /* ip_vs_lblcr */
883 int sysctl_lblcr_expiration;
884 struct ctl_table_header *lblcr_ctl_header;
885 struct ctl_table *lblcr_ctl_table;
886 /* ip_vs_est */
887 struct list_head est_list; /* estimator list */
888 spinlock_t est_lock;
889 struct timer_list est_timer; /* Estimation timer */
890 /* ip_vs_sync */
891 struct list_head sync_queue;
892 spinlock_t sync_lock;
893 struct ip_vs_sync_buff *sync_buff;
894 spinlock_t sync_buff_lock;
895 struct sockaddr_in sync_mcast_addr;
896 struct task_struct *master_thread;
897 struct task_struct *backup_thread;
898 int send_mesg_maxlen;
899 int recv_mesg_maxlen;
900 volatile int sync_state;
901 volatile int master_syncid;
902 volatile int backup_syncid;
903 struct mutex sync_mutex;
904 /* multicast interface name */
905 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
906 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
907 /* net name space ptr */
908 struct net *net; /* Needed by timer routines */
909};
910
911#define DEFAULT_SYNC_THRESHOLD 3
912#define DEFAULT_SYNC_PERIOD 50
913#define DEFAULT_SYNC_VER 1
914
915#ifdef CONFIG_SYSCTL
916
917static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
918{
919 return ipvs->sysctl_sync_threshold[0];
920}
921
922static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
923{
924 return ipvs->sysctl_sync_threshold[1];
925}
926
927static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
928{
929 return ipvs->sysctl_sync_ver;
930}
931
932#else
933
934static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
935{
936 return DEFAULT_SYNC_THRESHOLD;
937}
938
939static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
940{
941 return DEFAULT_SYNC_PERIOD;
942}
943
944static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
945{
946 return DEFAULT_SYNC_VER;
947}
948
949#endif
950
951/*
952 * IPVS core functions
953 * (from ip_vs_core.c)
954 */
955extern const char *ip_vs_proto_name(unsigned proto);
956extern void ip_vs_init_hash_table(struct list_head *table, int rows);
957#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
958
959#define IP_VS_APP_TYPE_FTP 1
960
961/*
962 * ip_vs_conn handling functions
963 * (from ip_vs_conn.c)
964 */
965
966enum {
967 IP_VS_DIR_INPUT = 0,
968 IP_VS_DIR_OUTPUT,
969 IP_VS_DIR_INPUT_ONLY,
970 IP_VS_DIR_LAST,
971};
972
973static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
974 const union nf_inet_addr *caddr,
975 __be16 cport,
976 const union nf_inet_addr *vaddr,
977 __be16 vport,
978 struct ip_vs_conn_param *p)
979{
980 p->net = net;
981 p->af = af;
982 p->protocol = protocol;
983 p->caddr = caddr;
984 p->cport = cport;
985 p->vaddr = vaddr;
986 p->vport = vport;
987 p->pe = NULL;
988 p->pe_data = NULL;
989}
990
991struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
992struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
993
994struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
995 const struct ip_vs_iphdr *iph,
996 unsigned int proto_off,
997 int inverse);
998
999struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1000
1001struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1002 const struct ip_vs_iphdr *iph,
1003 unsigned int proto_off,
1004 int inverse);
1005
1006/* put back the conn without restarting its timer */
1007static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1008{
1009 atomic_dec(&cp->refcnt);
1010}
1011extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1012extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1013
1014struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1015 const union nf_inet_addr *daddr,
1016 __be16 dport, unsigned flags,
1017 struct ip_vs_dest *dest, __u32 fwmark);
1018extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1019
1020extern const char * ip_vs_state_name(__u16 proto, int state);
1021
1022extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1023extern int ip_vs_check_template(struct ip_vs_conn *ct);
1024extern void ip_vs_random_dropentry(struct net *net);
1025extern int ip_vs_conn_init(void);
1026extern void ip_vs_conn_cleanup(void);
1027
1028static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1029{
1030 struct ip_vs_conn *ctl_cp = cp->control;
1031 if (!ctl_cp) {
1032 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1033 "%s:%d to %s:%d\n",
1034 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1035 ntohs(cp->cport),
1036 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1037 ntohs(cp->vport));
1038
1039 return;
1040 }
1041
1042 IP_VS_DBG_BUF(7, "DELeting control for: "
1043 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1044 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1045 ntohs(cp->cport),
1046 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1047 ntohs(ctl_cp->cport));
1048
1049 cp->control = NULL;
1050 if (atomic_read(&ctl_cp->n_control) == 0) {
1051 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1052 "%s:%d to %s:%d\n",
1053 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1054 ntohs(cp->cport),
1055 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1056 ntohs(cp->vport));
1057
1058 return;
1059 }
1060 atomic_dec(&ctl_cp->n_control);
1061}
1062
1063static inline void
1064ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1065{
1066 if (cp->control) {
1067 IP_VS_ERR_BUF("request control ADD for already controlled: "
1068 "%s:%d to %s:%d\n",
1069 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1070 ntohs(cp->cport),
1071 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1072 ntohs(cp->vport));
1073
1074 ip_vs_control_del(cp);
1075 }
1076
1077 IP_VS_DBG_BUF(7, "ADDing control for: "
1078 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1079 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1080 ntohs(cp->cport),
1081 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1082 ntohs(ctl_cp->cport));
1083
1084 cp->control = ctl_cp;
1085 atomic_inc(&ctl_cp->n_control);
1086}
1087
1088/*
1089 * IPVS netns init & cleanup functions
1090 */
1091extern int ip_vs_estimator_net_init(struct net *net);
1092extern int ip_vs_control_net_init(struct net *net);
1093extern int ip_vs_protocol_net_init(struct net *net);
1094extern int ip_vs_app_net_init(struct net *net);
1095extern int ip_vs_conn_net_init(struct net *net);
1096extern int ip_vs_sync_net_init(struct net *net);
1097extern void ip_vs_conn_net_cleanup(struct net *net);
1098extern void ip_vs_app_net_cleanup(struct net *net);
1099extern void ip_vs_protocol_net_cleanup(struct net *net);
1100extern void ip_vs_control_net_cleanup(struct net *net);
1101extern void ip_vs_estimator_net_cleanup(struct net *net);
1102extern void ip_vs_sync_net_cleanup(struct net *net);
1103extern void ip_vs_service_net_cleanup(struct net *net);
1104
1105/*
1106 * IPVS application functions
1107 * (from ip_vs_app.c)
1108 */
1109#define IP_VS_APP_MAX_PORTS 8
1110extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1111extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1112extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1113extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1114extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1115 __u16 proto, __u16 port);
1116extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1117extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1118
1119extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1120extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1121
1122void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1123void ip_vs_unbind_pe(struct ip_vs_service *svc);
1124int register_ip_vs_pe(struct ip_vs_pe *pe);
1125int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1126struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1127struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1128
1129static inline void ip_vs_pe_get(const struct ip_vs_pe *pe)
1130{
1131 if (pe && pe->module)
1132 __module_get(pe->module);
1133}
1134
1135static inline void ip_vs_pe_put(const struct ip_vs_pe *pe)
1136{
1137 if (pe && pe->module)
1138 module_put(pe->module);
1139}
1140
1141/*
1142 * IPVS protocol functions (from ip_vs_proto.c)
1143 */
1144extern int ip_vs_protocol_init(void);
1145extern void ip_vs_protocol_cleanup(void);
1146extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1147extern int *ip_vs_create_timeout_table(int *table, int size);
1148extern int
1149ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1150 const char *name, int to);
1151extern void
1152ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1153 const struct sk_buff *skb,
1154 int offset, const char *msg);
1155
1156extern struct ip_vs_protocol ip_vs_protocol_tcp;
1157extern struct ip_vs_protocol ip_vs_protocol_udp;
1158extern struct ip_vs_protocol ip_vs_protocol_icmp;
1159extern struct ip_vs_protocol ip_vs_protocol_esp;
1160extern struct ip_vs_protocol ip_vs_protocol_ah;
1161extern struct ip_vs_protocol ip_vs_protocol_sctp;
1162
1163/*
1164 * Registering/unregistering scheduler functions
1165 * (from ip_vs_sched.c)
1166 */
1167extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1168extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1169extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1170 struct ip_vs_scheduler *scheduler);
1171extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1172extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1173extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1174extern struct ip_vs_conn *
1175ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1176 struct ip_vs_proto_data *pd, int *ignored);
1177extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1178 struct ip_vs_proto_data *pd);
1179
1180extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1181
1182
1183/*
1184 * IPVS control data and functions (from ip_vs_ctl.c)
1185 */
1186extern struct ip_vs_stats ip_vs_stats;
1187extern const struct ctl_path net_vs_ctl_path[];
1188extern int sysctl_ip_vs_sync_ver;
1189
1190extern void ip_vs_sync_switch_mode(struct net *net, int mode);
1191extern struct ip_vs_service *
1192ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1193 const union nf_inet_addr *vaddr, __be16 vport);
1194
1195static inline void ip_vs_service_put(struct ip_vs_service *svc)
1196{
1197 atomic_dec(&svc->usecnt);
1198}
1199
1200extern struct ip_vs_dest *
1201ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
1202 const union nf_inet_addr *daddr, __be16 dport);
1203
1204extern int ip_vs_use_count_inc(void);
1205extern void ip_vs_use_count_dec(void);
1206extern int ip_vs_control_init(void);
1207extern void ip_vs_control_cleanup(void);
1208extern struct ip_vs_dest *
1209ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1210 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1211 __u16 protocol, __u32 fwmark);
1212extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1213
1214
1215/*
1216 * IPVS sync daemon data and function prototypes
1217 * (from ip_vs_sync.c)
1218 */
1219extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1220 __u8 syncid);
1221extern int stop_sync_thread(struct net *net, int state);
1222extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
1223
1224
1225/*
1226 * IPVS rate estimator prototypes (from ip_vs_est.c)
1227 */
1228extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1229extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1230extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1231extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1232 struct ip_vs_stats *stats);
1233
1234/*
1235 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1236 */
1237extern int ip_vs_null_xmit
1238(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1239extern int ip_vs_bypass_xmit
1240(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1241extern int ip_vs_nat_xmit
1242(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1243extern int ip_vs_tunnel_xmit
1244(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1245extern int ip_vs_dr_xmit
1246(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1247extern int ip_vs_icmp_xmit
1248(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1249 int offset, unsigned int hooknum);
1250extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1251
1252#ifdef CONFIG_IP_VS_IPV6
1253extern int ip_vs_bypass_xmit_v6
1254(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1255extern int ip_vs_nat_xmit_v6
1256(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1257extern int ip_vs_tunnel_xmit_v6
1258(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1259extern int ip_vs_dr_xmit_v6
1260(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1261extern int ip_vs_icmp_xmit_v6
1262(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1263 int offset, unsigned int hooknum);
1264#endif
1265
1266#ifdef CONFIG_SYSCTL
1267/*
1268 * This is a simple mechanism to ignore packets when
1269 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1270 * we start to drop 1/rate of the packets
1271 */
1272
1273static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1274{
1275 if (!ipvs->drop_rate)
1276 return 0;
1277 if (--ipvs->drop_counter > 0)
1278 return 0;
1279 ipvs->drop_counter = ipvs->drop_rate;
1280 return 1;
1281}
1282#else
1283static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1284#endif
1285
1286/*
1287 * ip_vs_fwd_tag returns the forwarding tag of the connection
1288 */
1289#define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1290
1291static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1292{
1293 char fwd;
1294
1295 switch (IP_VS_FWD_METHOD(cp)) {
1296 case IP_VS_CONN_F_MASQ:
1297 fwd = 'M'; break;
1298 case IP_VS_CONN_F_LOCALNODE:
1299 fwd = 'L'; break;
1300 case IP_VS_CONN_F_TUNNEL:
1301 fwd = 'T'; break;
1302 case IP_VS_CONN_F_DROUTE:
1303 fwd = 'R'; break;
1304 case IP_VS_CONN_F_BYPASS:
1305 fwd = 'B'; break;
1306 default:
1307 fwd = '?'; break;
1308 }
1309 return fwd;
1310}
1311
1312extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1313 struct ip_vs_conn *cp, int dir);
1314
1315#ifdef CONFIG_IP_VS_IPV6
1316extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1317 struct ip_vs_conn *cp, int dir);
1318#endif
1319
1320extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1321
1322static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1323{
1324 __be32 diff[2] = { ~old, new };
1325
1326 return csum_partial(diff, sizeof(diff), oldsum);
1327}
1328
1329#ifdef CONFIG_IP_VS_IPV6
1330static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1331 __wsum oldsum)
1332{
1333 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1334 new[3], new[2], new[1], new[0] };
1335
1336 return csum_partial(diff, sizeof(diff), oldsum);
1337}
1338#endif
1339
1340static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1341{
1342 __be16 diff[2] = { ~old, new };
1343
1344 return csum_partial(diff, sizeof(diff), oldsum);
1345}
1346
1347/*
1348 * Forget current conntrack (unconfirmed) and attach notrack entry
1349 */
1350static inline void ip_vs_notrack(struct sk_buff *skb)
1351{
1352#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1353 enum ip_conntrack_info ctinfo;
1354 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1355
1356 if (!ct || !nf_ct_is_untracked(ct)) {
1357 nf_reset(skb);
1358 skb->nfct = &nf_ct_untracked_get()->ct_general;
1359 skb->nfctinfo = IP_CT_NEW;
1360 nf_conntrack_get(skb->nfct);
1361 }
1362#endif
1363}
1364
1365#ifdef CONFIG_IP_VS_NFCT
1366/*
1367 * Netfilter connection tracking
1368 * (from ip_vs_nfct.c)
1369 */
1370static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1371{
1372#ifdef CONFIG_SYSCTL
1373 return ipvs->sysctl_conntrack;
1374#else
1375 return 0;
1376#endif
1377}
1378
1379extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1380 int outin);
1381extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1382extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1383 struct ip_vs_conn *cp, u_int8_t proto,
1384 const __be16 port, int from_rs);
1385extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1386
1387#else
1388
1389static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1390{
1391 return 0;
1392}
1393
1394static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1395 struct ip_vs_conn *cp, int outin)
1396{
1397}
1398
1399static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1400 struct ip_vs_conn *cp)
1401{
1402 return NF_ACCEPT;
1403}
1404
1405static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1406{
1407}
1408/* CONFIG_IP_VS_NFCT */
1409#endif
1410
1411static inline unsigned int
1412ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1413{
1414 /*
1415 * We think the overhead of processing active connections is 256
1416 * times higher than that of inactive connections in average. (This
1417 * 256 times might not be accurate, we will change it later) We
1418 * use the following formula to estimate the overhead now:
1419 * dest->activeconns*256 + dest->inactconns
1420 */
1421 return (atomic_read(&dest->activeconns) << 8) +
1422 atomic_read(&dest->inactconns);
1423}
1424
1425#endif /* _NET_IP_VS_H */
1/*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
6#ifndef _NET_IP_VS_H
7#define _NET_IP_VS_H
8
9#include <linux/ip_vs.h> /* definitions shared with userland */
10
11#include <asm/types.h> /* for __uXX types */
12
13#include <linux/list.h> /* for struct list_head */
14#include <linux/spinlock.h> /* for struct rwlock_t */
15#include <linux/atomic.h> /* for struct atomic_t */
16#include <linux/compiler.h>
17#include <linux/timer.h>
18#include <linux/bug.h>
19
20#include <net/checksum.h>
21#include <linux/netfilter.h> /* for union nf_inet_addr */
22#include <linux/ip.h>
23#include <linux/ipv6.h> /* for struct ipv6hdr */
24#include <net/ipv6.h>
25#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
26#include <net/netfilter/nf_conntrack.h>
27#endif
28#include <net/net_namespace.h> /* Netw namespace */
29
30/*
31 * Generic access of ipvs struct
32 */
33static inline struct netns_ipvs *net_ipvs(struct net* net)
34{
35 return net->ipvs;
36}
37/*
38 * Get net ptr from skb in traffic cases
39 * use skb_sknet when call is from userland (ioctl or netlink)
40 */
41static inline struct net *skb_net(const struct sk_buff *skb)
42{
43#ifdef CONFIG_NET_NS
44#ifdef CONFIG_IP_VS_DEBUG
45 /*
46 * This is used for debug only.
47 * Start with the most likely hit
48 * End with BUG
49 */
50 if (likely(skb->dev && skb->dev->nd_net))
51 return dev_net(skb->dev);
52 if (skb_dst(skb) && skb_dst(skb)->dev)
53 return dev_net(skb_dst(skb)->dev);
54 WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
55 __func__, __LINE__);
56 if (likely(skb->sk && skb->sk->sk_net))
57 return sock_net(skb->sk);
58 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
59 __func__, __LINE__);
60 BUG();
61#else
62 return dev_net(skb->dev ? : skb_dst(skb)->dev);
63#endif
64#else
65 return &init_net;
66#endif
67}
68
69static inline struct net *skb_sknet(const struct sk_buff *skb)
70{
71#ifdef CONFIG_NET_NS
72#ifdef CONFIG_IP_VS_DEBUG
73 /* Start with the most likely hit */
74 if (likely(skb->sk && skb->sk->sk_net))
75 return sock_net(skb->sk);
76 WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
77 __func__, __LINE__);
78 if (likely(skb->dev && skb->dev->nd_net))
79 return dev_net(skb->dev);
80 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
81 __func__, __LINE__);
82 BUG();
83#else
84 return sock_net(skb->sk);
85#endif
86#else
87 return &init_net;
88#endif
89}
90/*
91 * This one needed for single_open_net since net is stored directly in
92 * private not as a struct i.e. seq_file_net can't be used.
93 */
94static inline struct net *seq_file_single_net(struct seq_file *seq)
95{
96#ifdef CONFIG_NET_NS
97 return (struct net *)seq->private;
98#else
99 return &init_net;
100#endif
101}
102
103/* Connections' size value needed by ip_vs_ctl.c */
104extern int ip_vs_conn_tab_size;
105
106
107struct ip_vs_iphdr {
108 int len;
109 __u8 protocol;
110 union nf_inet_addr saddr;
111 union nf_inet_addr daddr;
112};
113
114static inline void
115ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
116{
117#ifdef CONFIG_IP_VS_IPV6
118 if (af == AF_INET6) {
119 const struct ipv6hdr *iph = nh;
120 iphdr->len = sizeof(struct ipv6hdr);
121 iphdr->protocol = iph->nexthdr;
122 iphdr->saddr.in6 = iph->saddr;
123 iphdr->daddr.in6 = iph->daddr;
124 } else
125#endif
126 {
127 const struct iphdr *iph = nh;
128 iphdr->len = iph->ihl * 4;
129 iphdr->protocol = iph->protocol;
130 iphdr->saddr.ip = iph->saddr;
131 iphdr->daddr.ip = iph->daddr;
132 }
133}
134
135static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
136 const union nf_inet_addr *src)
137{
138#ifdef CONFIG_IP_VS_IPV6
139 if (af == AF_INET6)
140 dst->in6 = src->in6;
141 else
142#endif
143 dst->ip = src->ip;
144}
145
146static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
147 const union nf_inet_addr *b)
148{
149#ifdef CONFIG_IP_VS_IPV6
150 if (af == AF_INET6)
151 return ipv6_addr_equal(&a->in6, &b->in6);
152#endif
153 return a->ip == b->ip;
154}
155
156#ifdef CONFIG_IP_VS_DEBUG
157#include <linux/net.h>
158
159extern int ip_vs_get_debug_level(void);
160
161static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
162 const union nf_inet_addr *addr,
163 int *idx)
164{
165 int len;
166#ifdef CONFIG_IP_VS_IPV6
167 if (af == AF_INET6)
168 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
169 &addr->in6) + 1;
170 else
171#endif
172 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
173 &addr->ip) + 1;
174
175 *idx += len;
176 BUG_ON(*idx > buf_len + 1);
177 return &buf[*idx - len];
178}
179
180#define IP_VS_DBG_BUF(level, msg, ...) \
181 do { \
182 char ip_vs_dbg_buf[160]; \
183 int ip_vs_dbg_idx = 0; \
184 if (level <= ip_vs_get_debug_level()) \
185 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
186 } while (0)
187#define IP_VS_ERR_BUF(msg...) \
188 do { \
189 char ip_vs_dbg_buf[160]; \
190 int ip_vs_dbg_idx = 0; \
191 pr_err(msg); \
192 } while (0)
193
194/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
195#define IP_VS_DBG_ADDR(af, addr) \
196 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
197 sizeof(ip_vs_dbg_buf), addr, \
198 &ip_vs_dbg_idx)
199
200#define IP_VS_DBG(level, msg, ...) \
201 do { \
202 if (level <= ip_vs_get_debug_level()) \
203 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
204 } while (0)
205#define IP_VS_DBG_RL(msg, ...) \
206 do { \
207 if (net_ratelimit()) \
208 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
209 } while (0)
210#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
211 do { \
212 if (level <= ip_vs_get_debug_level()) \
213 pp->debug_packet(af, pp, skb, ofs, msg); \
214 } while (0)
215#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
216 do { \
217 if (level <= ip_vs_get_debug_level() && \
218 net_ratelimit()) \
219 pp->debug_packet(af, pp, skb, ofs, msg); \
220 } while (0)
221#else /* NO DEBUGGING at ALL */
222#define IP_VS_DBG_BUF(level, msg...) do {} while (0)
223#define IP_VS_ERR_BUF(msg...) do {} while (0)
224#define IP_VS_DBG(level, msg...) do {} while (0)
225#define IP_VS_DBG_RL(msg...) do {} while (0)
226#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
227#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
228#endif
229
230#define IP_VS_BUG() BUG()
231#define IP_VS_ERR_RL(msg, ...) \
232 do { \
233 if (net_ratelimit()) \
234 pr_err(msg, ##__VA_ARGS__); \
235 } while (0)
236
237#ifdef CONFIG_IP_VS_DEBUG
238#define EnterFunction(level) \
239 do { \
240 if (level <= ip_vs_get_debug_level()) \
241 printk(KERN_DEBUG \
242 pr_fmt("Enter: %s, %s line %i\n"), \
243 __func__, __FILE__, __LINE__); \
244 } while (0)
245#define LeaveFunction(level) \
246 do { \
247 if (level <= ip_vs_get_debug_level()) \
248 printk(KERN_DEBUG \
249 pr_fmt("Leave: %s, %s line %i\n"), \
250 __func__, __FILE__, __LINE__); \
251 } while (0)
252#else
253#define EnterFunction(level) do {} while (0)
254#define LeaveFunction(level) do {} while (0)
255#endif
256
257#define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
258
259
260/*
261 * The port number of FTP service (in network order).
262 */
263#define FTPPORT cpu_to_be16(21)
264#define FTPDATA cpu_to_be16(20)
265
266/*
267 * TCP State Values
268 */
269enum {
270 IP_VS_TCP_S_NONE = 0,
271 IP_VS_TCP_S_ESTABLISHED,
272 IP_VS_TCP_S_SYN_SENT,
273 IP_VS_TCP_S_SYN_RECV,
274 IP_VS_TCP_S_FIN_WAIT,
275 IP_VS_TCP_S_TIME_WAIT,
276 IP_VS_TCP_S_CLOSE,
277 IP_VS_TCP_S_CLOSE_WAIT,
278 IP_VS_TCP_S_LAST_ACK,
279 IP_VS_TCP_S_LISTEN,
280 IP_VS_TCP_S_SYNACK,
281 IP_VS_TCP_S_LAST
282};
283
284/*
285 * UDP State Values
286 */
287enum {
288 IP_VS_UDP_S_NORMAL,
289 IP_VS_UDP_S_LAST,
290};
291
292/*
293 * ICMP State Values
294 */
295enum {
296 IP_VS_ICMP_S_NORMAL,
297 IP_VS_ICMP_S_LAST,
298};
299
300/*
301 * SCTP State Values
302 */
303enum ip_vs_sctp_states {
304 IP_VS_SCTP_S_NONE,
305 IP_VS_SCTP_S_INIT_CLI,
306 IP_VS_SCTP_S_INIT_SER,
307 IP_VS_SCTP_S_INIT_ACK_CLI,
308 IP_VS_SCTP_S_INIT_ACK_SER,
309 IP_VS_SCTP_S_ECHO_CLI,
310 IP_VS_SCTP_S_ECHO_SER,
311 IP_VS_SCTP_S_ESTABLISHED,
312 IP_VS_SCTP_S_SHUT_CLI,
313 IP_VS_SCTP_S_SHUT_SER,
314 IP_VS_SCTP_S_SHUT_ACK_CLI,
315 IP_VS_SCTP_S_SHUT_ACK_SER,
316 IP_VS_SCTP_S_CLOSED,
317 IP_VS_SCTP_S_LAST
318};
319
320/*
321 * Delta sequence info structure
322 * Each ip_vs_conn has 2 (output AND input seq. changes).
323 * Only used in the VS/NAT.
324 */
325struct ip_vs_seq {
326 __u32 init_seq; /* Add delta from this seq */
327 __u32 delta; /* Delta in sequence numbers */
328 __u32 previous_delta; /* Delta in sequence numbers
329 before last resized pkt */
330};
331
332/*
333 * counters per cpu
334 */
335struct ip_vs_counters {
336 __u32 conns; /* connections scheduled */
337 __u32 inpkts; /* incoming packets */
338 __u32 outpkts; /* outgoing packets */
339 __u64 inbytes; /* incoming bytes */
340 __u64 outbytes; /* outgoing bytes */
341};
342/*
343 * Stats per cpu
344 */
345struct ip_vs_cpu_stats {
346 struct ip_vs_counters ustats;
347 struct u64_stats_sync syncp;
348};
349
350/*
351 * IPVS statistics objects
352 */
353struct ip_vs_estimator {
354 struct list_head list;
355
356 u64 last_inbytes;
357 u64 last_outbytes;
358 u32 last_conns;
359 u32 last_inpkts;
360 u32 last_outpkts;
361
362 u32 cps;
363 u32 inpps;
364 u32 outpps;
365 u32 inbps;
366 u32 outbps;
367};
368
369struct ip_vs_stats {
370 struct ip_vs_stats_user ustats; /* statistics */
371 struct ip_vs_estimator est; /* estimator */
372 struct ip_vs_cpu_stats *cpustats; /* per cpu counters */
373 spinlock_t lock; /* spin lock */
374 struct ip_vs_stats_user ustats0; /* reset values */
375};
376
377struct dst_entry;
378struct iphdr;
379struct ip_vs_conn;
380struct ip_vs_app;
381struct sk_buff;
382struct ip_vs_proto_data;
383
384struct ip_vs_protocol {
385 struct ip_vs_protocol *next;
386 char *name;
387 u16 protocol;
388 u16 num_states;
389 int dont_defrag;
390
391 void (*init)(struct ip_vs_protocol *pp);
392
393 void (*exit)(struct ip_vs_protocol *pp);
394
395 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
396
397 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
398
399 int (*conn_schedule)(int af, struct sk_buff *skb,
400 struct ip_vs_proto_data *pd,
401 int *verdict, struct ip_vs_conn **cpp);
402
403 struct ip_vs_conn *
404 (*conn_in_get)(int af,
405 const struct sk_buff *skb,
406 const struct ip_vs_iphdr *iph,
407 unsigned int proto_off,
408 int inverse);
409
410 struct ip_vs_conn *
411 (*conn_out_get)(int af,
412 const struct sk_buff *skb,
413 const struct ip_vs_iphdr *iph,
414 unsigned int proto_off,
415 int inverse);
416
417 int (*snat_handler)(struct sk_buff *skb,
418 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
419
420 int (*dnat_handler)(struct sk_buff *skb,
421 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
422
423 int (*csum_check)(int af, struct sk_buff *skb,
424 struct ip_vs_protocol *pp);
425
426 const char *(*state_name)(int state);
427
428 void (*state_transition)(struct ip_vs_conn *cp, int direction,
429 const struct sk_buff *skb,
430 struct ip_vs_proto_data *pd);
431
432 int (*register_app)(struct net *net, struct ip_vs_app *inc);
433
434 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
435
436 int (*app_conn_bind)(struct ip_vs_conn *cp);
437
438 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
439 const struct sk_buff *skb,
440 int offset,
441 const char *msg);
442
443 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
444};
445
446/*
447 * protocol data per netns
448 */
449struct ip_vs_proto_data {
450 struct ip_vs_proto_data *next;
451 struct ip_vs_protocol *pp;
452 int *timeout_table; /* protocol timeout table */
453 atomic_t appcnt; /* counter of proto app incs. */
454 struct tcp_states_t *tcp_state_table;
455};
456
457extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
458extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
459 unsigned short proto);
460
461struct ip_vs_conn_param {
462 struct net *net;
463 const union nf_inet_addr *caddr;
464 const union nf_inet_addr *vaddr;
465 __be16 cport;
466 __be16 vport;
467 __u16 protocol;
468 u16 af;
469
470 const struct ip_vs_pe *pe;
471 char *pe_data;
472 __u8 pe_data_len;
473};
474
475/*
476 * IP_VS structure allocated for each dynamically scheduled connection
477 */
478struct ip_vs_conn {
479 struct hlist_node c_list; /* hashed list heads */
480#ifdef CONFIG_NET_NS
481 struct net *net; /* Name space */
482#endif
483 /* Protocol, addresses and port numbers */
484 u16 af; /* address family */
485 __be16 cport;
486 __be16 vport;
487 __be16 dport;
488 __u32 fwmark; /* Fire wall mark from skb */
489 union nf_inet_addr caddr; /* client address */
490 union nf_inet_addr vaddr; /* virtual address */
491 union nf_inet_addr daddr; /* destination address */
492 volatile __u32 flags; /* status flags */
493 __u16 protocol; /* Which protocol (TCP/UDP) */
494
495 /* counter and timer */
496 atomic_t refcnt; /* reference count */
497 struct timer_list timer; /* Expiration timer */
498 volatile unsigned long timeout; /* timeout */
499
500 /* Flags and state transition */
501 spinlock_t lock; /* lock for state transition */
502 volatile __u16 state; /* state info */
503 volatile __u16 old_state; /* old state, to be used for
504 * state transition triggerd
505 * synchronization
506 */
507 unsigned long sync_endtime; /* jiffies + sent_retries */
508
509 /* Control members */
510 struct ip_vs_conn *control; /* Master control connection */
511 atomic_t n_control; /* Number of controlled ones */
512 struct ip_vs_dest *dest; /* real server */
513 atomic_t in_pkts; /* incoming packet counter */
514
515 /* packet transmitter for different forwarding methods. If it
516 mangles the packet, it must return NF_DROP or better NF_STOLEN,
517 otherwise this must be changed to a sk_buff **.
518 NF_ACCEPT can be returned when destination is local.
519 */
520 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
521 struct ip_vs_protocol *pp);
522
523 /* Note: we can group the following members into a structure,
524 in order to save more space, and the following members are
525 only used in VS/NAT anyway */
526 struct ip_vs_app *app; /* bound ip_vs_app object */
527 void *app_data; /* Application private data */
528 struct ip_vs_seq in_seq; /* incoming seq. struct */
529 struct ip_vs_seq out_seq; /* outgoing seq. struct */
530
531 const struct ip_vs_pe *pe;
532 char *pe_data;
533 __u8 pe_data_len;
534};
535
536/*
537 * To save some memory in conn table when name space is disabled.
538 */
539static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
540{
541#ifdef CONFIG_NET_NS
542 return cp->net;
543#else
544 return &init_net;
545#endif
546}
547static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
548{
549#ifdef CONFIG_NET_NS
550 cp->net = net;
551#endif
552}
553
554static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
555 struct net *net)
556{
557#ifdef CONFIG_NET_NS
558 return cp->net == net;
559#else
560 return 1;
561#endif
562}
563
564/*
565 * Extended internal versions of struct ip_vs_service_user and
566 * ip_vs_dest_user for IPv6 support.
567 *
568 * We need these to conveniently pass around service and destination
569 * options, but unfortunately, we also need to keep the old definitions to
570 * maintain userspace backwards compatibility for the setsockopt interface.
571 */
572struct ip_vs_service_user_kern {
573 /* virtual service addresses */
574 u16 af;
575 u16 protocol;
576 union nf_inet_addr addr; /* virtual ip address */
577 u16 port;
578 u32 fwmark; /* firwall mark of service */
579
580 /* virtual service options */
581 char *sched_name;
582 char *pe_name;
583 unsigned int flags; /* virtual service flags */
584 unsigned int timeout; /* persistent timeout in sec */
585 u32 netmask; /* persistent netmask */
586};
587
588
589struct ip_vs_dest_user_kern {
590 /* destination server address */
591 union nf_inet_addr addr;
592 u16 port;
593
594 /* real server options */
595 unsigned int conn_flags; /* connection flags */
596 int weight; /* destination weight */
597
598 /* thresholds for active connections */
599 u32 u_threshold; /* upper threshold */
600 u32 l_threshold; /* lower threshold */
601};
602
603
604/*
605 * The information about the virtual service offered to the net
606 * and the forwarding entries
607 */
608struct ip_vs_service {
609 struct list_head s_list; /* for normal service table */
610 struct list_head f_list; /* for fwmark-based service table */
611 atomic_t refcnt; /* reference counter */
612 atomic_t usecnt; /* use counter */
613
614 u16 af; /* address family */
615 __u16 protocol; /* which protocol (TCP/UDP) */
616 union nf_inet_addr addr; /* IP address for virtual service */
617 __be16 port; /* port number for the service */
618 __u32 fwmark; /* firewall mark of the service */
619 unsigned int flags; /* service status flags */
620 unsigned int timeout; /* persistent timeout in ticks */
621 __be32 netmask; /* grouping granularity */
622 struct net *net;
623
624 struct list_head destinations; /* real server d-linked list */
625 __u32 num_dests; /* number of servers */
626 struct ip_vs_stats stats; /* statistics for the service */
627 struct ip_vs_app *inc; /* bind conns to this app inc */
628
629 /* for scheduling */
630 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
631 rwlock_t sched_lock; /* lock sched_data */
632 void *sched_data; /* scheduler application data */
633
634 /* alternate persistence engine */
635 struct ip_vs_pe *pe;
636};
637
638
639/*
640 * The real server destination forwarding entry
641 * with ip address, port number, and so on.
642 */
643struct ip_vs_dest {
644 struct list_head n_list; /* for the dests in the service */
645 struct list_head d_list; /* for table with all the dests */
646
647 u16 af; /* address family */
648 __be16 port; /* port number of the server */
649 union nf_inet_addr addr; /* IP address of the server */
650 volatile unsigned int flags; /* dest status flags */
651 atomic_t conn_flags; /* flags to copy to conn */
652 atomic_t weight; /* server weight */
653
654 atomic_t refcnt; /* reference counter */
655 struct ip_vs_stats stats; /* statistics */
656
657 /* connection counters and thresholds */
658 atomic_t activeconns; /* active connections */
659 atomic_t inactconns; /* inactive connections */
660 atomic_t persistconns; /* persistent connections */
661 __u32 u_threshold; /* upper threshold */
662 __u32 l_threshold; /* lower threshold */
663
664 /* for destination cache */
665 spinlock_t dst_lock; /* lock of dst_cache */
666 struct dst_entry *dst_cache; /* destination cache entry */
667 u32 dst_rtos; /* RT_TOS(tos) for dst */
668 u32 dst_cookie;
669 union nf_inet_addr dst_saddr;
670
671 /* for virtual service */
672 struct ip_vs_service *svc; /* service it belongs to */
673 __u16 protocol; /* which protocol (TCP/UDP) */
674 __be16 vport; /* virtual port number */
675 union nf_inet_addr vaddr; /* virtual IP address */
676 __u32 vfwmark; /* firewall mark of service */
677};
678
679
680/*
681 * The scheduler object
682 */
683struct ip_vs_scheduler {
684 struct list_head n_list; /* d-linked list head */
685 char *name; /* scheduler name */
686 atomic_t refcnt; /* reference counter */
687 struct module *module; /* THIS_MODULE/NULL */
688
689 /* scheduler initializing service */
690 int (*init_service)(struct ip_vs_service *svc);
691 /* scheduling service finish */
692 int (*done_service)(struct ip_vs_service *svc);
693 /* scheduler updating service */
694 int (*update_service)(struct ip_vs_service *svc);
695
696 /* selecting a server from the given service */
697 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
698 const struct sk_buff *skb);
699};
700
701/* The persistence engine object */
702struct ip_vs_pe {
703 struct list_head n_list; /* d-linked list head */
704 char *name; /* scheduler name */
705 atomic_t refcnt; /* reference counter */
706 struct module *module; /* THIS_MODULE/NULL */
707
708 /* get the connection template, if any */
709 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
710 bool (*ct_match)(const struct ip_vs_conn_param *p,
711 struct ip_vs_conn *ct);
712 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
713 bool inverse);
714 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
715};
716
717/*
718 * The application module object (a.k.a. app incarnation)
719 */
720struct ip_vs_app {
721 struct list_head a_list; /* member in app list */
722 int type; /* IP_VS_APP_TYPE_xxx */
723 char *name; /* application module name */
724 __u16 protocol;
725 struct module *module; /* THIS_MODULE/NULL */
726 struct list_head incs_list; /* list of incarnations */
727
728 /* members for application incarnations */
729 struct list_head p_list; /* member in proto app list */
730 struct ip_vs_app *app; /* its real application */
731 __be16 port; /* port number in net order */
732 atomic_t usecnt; /* usage counter */
733
734 /*
735 * output hook: Process packet in inout direction, diff set for TCP.
736 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
737 * 2=Mangled but checksum was not updated
738 */
739 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
740 struct sk_buff *, int *diff);
741
742 /*
743 * input hook: Process packet in outin direction, diff set for TCP.
744 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
745 * 2=Mangled but checksum was not updated
746 */
747 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
748 struct sk_buff *, int *diff);
749
750 /* ip_vs_app initializer */
751 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
752
753 /* ip_vs_app finish */
754 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
755
756
757 /* not used now */
758 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
759 struct ip_vs_protocol *);
760
761 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
762
763 int * timeout_table;
764 int * timeouts;
765 int timeouts_size;
766
767 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
768 int *verdict, struct ip_vs_conn **cpp);
769
770 struct ip_vs_conn *
771 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
772 const struct iphdr *iph, unsigned int proto_off,
773 int inverse);
774
775 struct ip_vs_conn *
776 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
777 const struct iphdr *iph, unsigned int proto_off,
778 int inverse);
779
780 int (*state_transition)(struct ip_vs_conn *cp, int direction,
781 const struct sk_buff *skb,
782 struct ip_vs_app *app);
783
784 void (*timeout_change)(struct ip_vs_app *app, int flags);
785};
786
787struct ipvs_master_sync_state {
788 struct list_head sync_queue;
789 struct ip_vs_sync_buff *sync_buff;
790 int sync_queue_len;
791 unsigned int sync_queue_delay;
792 struct task_struct *master_thread;
793 struct delayed_work master_wakeup_work;
794 struct netns_ipvs *ipvs;
795};
796
797/* IPVS in network namespace */
798struct netns_ipvs {
799 int gen; /* Generation */
800 int enable; /* enable like nf_hooks do */
801 /*
802 * Hash table: for real service lookups
803 */
804 #define IP_VS_RTAB_BITS 4
805 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
806 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
807
808 struct list_head rs_table[IP_VS_RTAB_SIZE];
809 /* ip_vs_app */
810 struct list_head app_list;
811 /* ip_vs_ftp */
812 struct ip_vs_app *ftp_app;
813 /* ip_vs_proto */
814 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
815 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
816 /* ip_vs_proto_tcp */
817#ifdef CONFIG_IP_VS_PROTO_TCP
818 #define TCP_APP_TAB_BITS 4
819 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
820 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
821 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
822 spinlock_t tcp_app_lock;
823#endif
824 /* ip_vs_proto_udp */
825#ifdef CONFIG_IP_VS_PROTO_UDP
826 #define UDP_APP_TAB_BITS 4
827 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
828 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
829 struct list_head udp_apps[UDP_APP_TAB_SIZE];
830 spinlock_t udp_app_lock;
831#endif
832 /* ip_vs_proto_sctp */
833#ifdef CONFIG_IP_VS_PROTO_SCTP
834 #define SCTP_APP_TAB_BITS 4
835 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
836 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
837 /* Hash table for SCTP application incarnations */
838 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
839 spinlock_t sctp_app_lock;
840#endif
841 /* ip_vs_conn */
842 atomic_t conn_count; /* connection counter */
843
844 /* ip_vs_ctl */
845 struct ip_vs_stats tot_stats; /* Statistics & est. */
846
847 int num_services; /* no of virtual services */
848
849 rwlock_t rs_lock; /* real services table */
850 /* Trash for destinations */
851 struct list_head dest_trash;
852 /* Service counters */
853 atomic_t ftpsvc_counter;
854 atomic_t nullsvc_counter;
855
856#ifdef CONFIG_SYSCTL
857 /* 1/rate drop and drop-entry variables */
858 struct delayed_work defense_work; /* Work handler */
859 int drop_rate;
860 int drop_counter;
861 atomic_t dropentry;
862 /* locks in ctl.c */
863 spinlock_t dropentry_lock; /* drop entry handling */
864 spinlock_t droppacket_lock; /* drop packet handling */
865 spinlock_t securetcp_lock; /* state and timeout tables */
866
867 /* sys-ctl struct */
868 struct ctl_table_header *sysctl_hdr;
869 struct ctl_table *sysctl_tbl;
870#endif
871
872 /* sysctl variables */
873 int sysctl_amemthresh;
874 int sysctl_am_droprate;
875 int sysctl_drop_entry;
876 int sysctl_drop_packet;
877 int sysctl_secure_tcp;
878#ifdef CONFIG_IP_VS_NFCT
879 int sysctl_conntrack;
880#endif
881 int sysctl_snat_reroute;
882 int sysctl_sync_ver;
883 int sysctl_sync_ports;
884 int sysctl_sync_qlen_max;
885 int sysctl_sync_sock_size;
886 int sysctl_cache_bypass;
887 int sysctl_expire_nodest_conn;
888 int sysctl_expire_quiescent_template;
889 int sysctl_sync_threshold[2];
890 unsigned int sysctl_sync_refresh_period;
891 int sysctl_sync_retries;
892 int sysctl_nat_icmp_send;
893
894 /* ip_vs_lblc */
895 int sysctl_lblc_expiration;
896 struct ctl_table_header *lblc_ctl_header;
897 struct ctl_table *lblc_ctl_table;
898 /* ip_vs_lblcr */
899 int sysctl_lblcr_expiration;
900 struct ctl_table_header *lblcr_ctl_header;
901 struct ctl_table *lblcr_ctl_table;
902 /* ip_vs_est */
903 struct list_head est_list; /* estimator list */
904 spinlock_t est_lock;
905 struct timer_list est_timer; /* Estimation timer */
906 /* ip_vs_sync */
907 spinlock_t sync_lock;
908 struct ipvs_master_sync_state *ms;
909 spinlock_t sync_buff_lock;
910 struct task_struct **backup_threads;
911 int threads_mask;
912 int send_mesg_maxlen;
913 int recv_mesg_maxlen;
914 volatile int sync_state;
915 volatile int master_syncid;
916 volatile int backup_syncid;
917 struct mutex sync_mutex;
918 /* multicast interface name */
919 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
920 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
921 /* net name space ptr */
922 struct net *net; /* Needed by timer routines */
923};
924
925#define DEFAULT_SYNC_THRESHOLD 3
926#define DEFAULT_SYNC_PERIOD 50
927#define DEFAULT_SYNC_VER 1
928#define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
929#define DEFAULT_SYNC_RETRIES 0
930#define IPVS_SYNC_WAKEUP_RATE 8
931#define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
932#define IPVS_SYNC_SEND_DELAY (HZ / 50)
933#define IPVS_SYNC_CHECK_PERIOD HZ
934#define IPVS_SYNC_FLUSH_TIME (HZ * 2)
935#define IPVS_SYNC_PORTS_MAX (1 << 6)
936
937#ifdef CONFIG_SYSCTL
938
939static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
940{
941 return ipvs->sysctl_sync_threshold[0];
942}
943
944static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
945{
946 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
947}
948
949static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
950{
951 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
952}
953
954static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
955{
956 return ipvs->sysctl_sync_retries;
957}
958
959static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
960{
961 return ipvs->sysctl_sync_ver;
962}
963
964static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
965{
966 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
967}
968
969static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
970{
971 return ipvs->sysctl_sync_qlen_max;
972}
973
974static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
975{
976 return ipvs->sysctl_sync_sock_size;
977}
978
979#else
980
981static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
982{
983 return DEFAULT_SYNC_THRESHOLD;
984}
985
986static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
987{
988 return DEFAULT_SYNC_PERIOD;
989}
990
991static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
992{
993 return DEFAULT_SYNC_REFRESH_PERIOD;
994}
995
996static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
997{
998 return DEFAULT_SYNC_RETRIES & 3;
999}
1000
1001static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1002{
1003 return DEFAULT_SYNC_VER;
1004}
1005
1006static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1007{
1008 return 1;
1009}
1010
1011static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1012{
1013 return IPVS_SYNC_QLEN_MAX;
1014}
1015
1016static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1017{
1018 return 0;
1019}
1020
1021#endif
1022
1023/*
1024 * IPVS core functions
1025 * (from ip_vs_core.c)
1026 */
1027extern const char *ip_vs_proto_name(unsigned int proto);
1028extern void ip_vs_init_hash_table(struct list_head *table, int rows);
1029#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1030
1031#define IP_VS_APP_TYPE_FTP 1
1032
1033/*
1034 * ip_vs_conn handling functions
1035 * (from ip_vs_conn.c)
1036 */
1037
1038enum {
1039 IP_VS_DIR_INPUT = 0,
1040 IP_VS_DIR_OUTPUT,
1041 IP_VS_DIR_INPUT_ONLY,
1042 IP_VS_DIR_LAST,
1043};
1044
1045static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1046 const union nf_inet_addr *caddr,
1047 __be16 cport,
1048 const union nf_inet_addr *vaddr,
1049 __be16 vport,
1050 struct ip_vs_conn_param *p)
1051{
1052 p->net = net;
1053 p->af = af;
1054 p->protocol = protocol;
1055 p->caddr = caddr;
1056 p->cport = cport;
1057 p->vaddr = vaddr;
1058 p->vport = vport;
1059 p->pe = NULL;
1060 p->pe_data = NULL;
1061}
1062
1063struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1064struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1065
1066struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1067 const struct ip_vs_iphdr *iph,
1068 unsigned int proto_off,
1069 int inverse);
1070
1071struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1072
1073struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1074 const struct ip_vs_iphdr *iph,
1075 unsigned int proto_off,
1076 int inverse);
1077
1078/* put back the conn without restarting its timer */
1079static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1080{
1081 atomic_dec(&cp->refcnt);
1082}
1083extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1084extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1085
1086struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1087 const union nf_inet_addr *daddr,
1088 __be16 dport, unsigned int flags,
1089 struct ip_vs_dest *dest, __u32 fwmark);
1090extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1091
1092extern const char * ip_vs_state_name(__u16 proto, int state);
1093
1094extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1095extern int ip_vs_check_template(struct ip_vs_conn *ct);
1096extern void ip_vs_random_dropentry(struct net *net);
1097extern int ip_vs_conn_init(void);
1098extern void ip_vs_conn_cleanup(void);
1099
1100static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1101{
1102 struct ip_vs_conn *ctl_cp = cp->control;
1103 if (!ctl_cp) {
1104 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1105 "%s:%d to %s:%d\n",
1106 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1107 ntohs(cp->cport),
1108 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1109 ntohs(cp->vport));
1110
1111 return;
1112 }
1113
1114 IP_VS_DBG_BUF(7, "DELeting control for: "
1115 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1116 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1117 ntohs(cp->cport),
1118 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1119 ntohs(ctl_cp->cport));
1120
1121 cp->control = NULL;
1122 if (atomic_read(&ctl_cp->n_control) == 0) {
1123 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1124 "%s:%d to %s:%d\n",
1125 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1126 ntohs(cp->cport),
1127 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1128 ntohs(cp->vport));
1129
1130 return;
1131 }
1132 atomic_dec(&ctl_cp->n_control);
1133}
1134
1135static inline void
1136ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1137{
1138 if (cp->control) {
1139 IP_VS_ERR_BUF("request control ADD for already controlled: "
1140 "%s:%d to %s:%d\n",
1141 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1142 ntohs(cp->cport),
1143 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1144 ntohs(cp->vport));
1145
1146 ip_vs_control_del(cp);
1147 }
1148
1149 IP_VS_DBG_BUF(7, "ADDing control for: "
1150 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1151 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1152 ntohs(cp->cport),
1153 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1154 ntohs(ctl_cp->cport));
1155
1156 cp->control = ctl_cp;
1157 atomic_inc(&ctl_cp->n_control);
1158}
1159
1160/*
1161 * IPVS netns init & cleanup functions
1162 */
1163extern int ip_vs_estimator_net_init(struct net *net);
1164extern int ip_vs_control_net_init(struct net *net);
1165extern int ip_vs_protocol_net_init(struct net *net);
1166extern int ip_vs_app_net_init(struct net *net);
1167extern int ip_vs_conn_net_init(struct net *net);
1168extern int ip_vs_sync_net_init(struct net *net);
1169extern void ip_vs_conn_net_cleanup(struct net *net);
1170extern void ip_vs_app_net_cleanup(struct net *net);
1171extern void ip_vs_protocol_net_cleanup(struct net *net);
1172extern void ip_vs_control_net_cleanup(struct net *net);
1173extern void ip_vs_estimator_net_cleanup(struct net *net);
1174extern void ip_vs_sync_net_cleanup(struct net *net);
1175extern void ip_vs_service_net_cleanup(struct net *net);
1176
1177/*
1178 * IPVS application functions
1179 * (from ip_vs_app.c)
1180 */
1181#define IP_VS_APP_MAX_PORTS 8
1182extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1183extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1184extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1185extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1186extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1187 __u16 proto, __u16 port);
1188extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1189extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1190
1191extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1192extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1193
1194void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1195void ip_vs_unbind_pe(struct ip_vs_service *svc);
1196int register_ip_vs_pe(struct ip_vs_pe *pe);
1197int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1198struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1199struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1200
1201/*
1202 * Use a #define to avoid all of module.h just for these trivial ops
1203 */
1204#define ip_vs_pe_get(pe) \
1205 if (pe && pe->module) \
1206 __module_get(pe->module);
1207
1208#define ip_vs_pe_put(pe) \
1209 if (pe && pe->module) \
1210 module_put(pe->module);
1211
1212/*
1213 * IPVS protocol functions (from ip_vs_proto.c)
1214 */
1215extern int ip_vs_protocol_init(void);
1216extern void ip_vs_protocol_cleanup(void);
1217extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1218extern int *ip_vs_create_timeout_table(int *table, int size);
1219extern int
1220ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1221 const char *name, int to);
1222extern void
1223ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1224 const struct sk_buff *skb,
1225 int offset, const char *msg);
1226
1227extern struct ip_vs_protocol ip_vs_protocol_tcp;
1228extern struct ip_vs_protocol ip_vs_protocol_udp;
1229extern struct ip_vs_protocol ip_vs_protocol_icmp;
1230extern struct ip_vs_protocol ip_vs_protocol_esp;
1231extern struct ip_vs_protocol ip_vs_protocol_ah;
1232extern struct ip_vs_protocol ip_vs_protocol_sctp;
1233
1234/*
1235 * Registering/unregistering scheduler functions
1236 * (from ip_vs_sched.c)
1237 */
1238extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1239extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1240extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1241 struct ip_vs_scheduler *scheduler);
1242extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1243extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1244extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1245extern struct ip_vs_conn *
1246ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1247 struct ip_vs_proto_data *pd, int *ignored);
1248extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1249 struct ip_vs_proto_data *pd);
1250
1251extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1252
1253
1254/*
1255 * IPVS control data and functions (from ip_vs_ctl.c)
1256 */
1257extern struct ip_vs_stats ip_vs_stats;
1258extern int sysctl_ip_vs_sync_ver;
1259
1260extern struct ip_vs_service *
1261ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1262 const union nf_inet_addr *vaddr, __be16 vport);
1263
1264static inline void ip_vs_service_put(struct ip_vs_service *svc)
1265{
1266 atomic_dec(&svc->usecnt);
1267}
1268
1269extern struct ip_vs_dest *
1270ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
1271 const union nf_inet_addr *daddr, __be16 dport);
1272
1273extern int ip_vs_use_count_inc(void);
1274extern void ip_vs_use_count_dec(void);
1275extern int ip_vs_register_nl_ioctl(void);
1276extern void ip_vs_unregister_nl_ioctl(void);
1277extern int ip_vs_control_init(void);
1278extern void ip_vs_control_cleanup(void);
1279extern struct ip_vs_dest *
1280ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1281 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1282 __u16 protocol, __u32 fwmark, __u32 flags);
1283extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1284
1285
1286/*
1287 * IPVS sync daemon data and function prototypes
1288 * (from ip_vs_sync.c)
1289 */
1290extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1291 __u8 syncid);
1292extern int stop_sync_thread(struct net *net, int state);
1293extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1294
1295
1296/*
1297 * IPVS rate estimator prototypes (from ip_vs_est.c)
1298 */
1299extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1300extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1301extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1302extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1303 struct ip_vs_stats *stats);
1304
1305/*
1306 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1307 */
1308extern int ip_vs_null_xmit
1309(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1310extern int ip_vs_bypass_xmit
1311(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1312extern int ip_vs_nat_xmit
1313(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1314extern int ip_vs_tunnel_xmit
1315(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1316extern int ip_vs_dr_xmit
1317(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1318extern int ip_vs_icmp_xmit
1319(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1320 int offset, unsigned int hooknum);
1321extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1322
1323#ifdef CONFIG_IP_VS_IPV6
1324extern int ip_vs_bypass_xmit_v6
1325(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1326extern int ip_vs_nat_xmit_v6
1327(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1328extern int ip_vs_tunnel_xmit_v6
1329(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1330extern int ip_vs_dr_xmit_v6
1331(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1332extern int ip_vs_icmp_xmit_v6
1333(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1334 int offset, unsigned int hooknum);
1335#endif
1336
1337#ifdef CONFIG_SYSCTL
1338/*
1339 * This is a simple mechanism to ignore packets when
1340 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1341 * we start to drop 1/rate of the packets
1342 */
1343
1344static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1345{
1346 if (!ipvs->drop_rate)
1347 return 0;
1348 if (--ipvs->drop_counter > 0)
1349 return 0;
1350 ipvs->drop_counter = ipvs->drop_rate;
1351 return 1;
1352}
1353#else
1354static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1355#endif
1356
1357/*
1358 * ip_vs_fwd_tag returns the forwarding tag of the connection
1359 */
1360#define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1361
1362static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1363{
1364 char fwd;
1365
1366 switch (IP_VS_FWD_METHOD(cp)) {
1367 case IP_VS_CONN_F_MASQ:
1368 fwd = 'M'; break;
1369 case IP_VS_CONN_F_LOCALNODE:
1370 fwd = 'L'; break;
1371 case IP_VS_CONN_F_TUNNEL:
1372 fwd = 'T'; break;
1373 case IP_VS_CONN_F_DROUTE:
1374 fwd = 'R'; break;
1375 case IP_VS_CONN_F_BYPASS:
1376 fwd = 'B'; break;
1377 default:
1378 fwd = '?'; break;
1379 }
1380 return fwd;
1381}
1382
1383extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1384 struct ip_vs_conn *cp, int dir);
1385
1386#ifdef CONFIG_IP_VS_IPV6
1387extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1388 struct ip_vs_conn *cp, int dir);
1389#endif
1390
1391extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1392
1393static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1394{
1395 __be32 diff[2] = { ~old, new };
1396
1397 return csum_partial(diff, sizeof(diff), oldsum);
1398}
1399
1400#ifdef CONFIG_IP_VS_IPV6
1401static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1402 __wsum oldsum)
1403{
1404 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1405 new[3], new[2], new[1], new[0] };
1406
1407 return csum_partial(diff, sizeof(diff), oldsum);
1408}
1409#endif
1410
1411static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1412{
1413 __be16 diff[2] = { ~old, new };
1414
1415 return csum_partial(diff, sizeof(diff), oldsum);
1416}
1417
1418/*
1419 * Forget current conntrack (unconfirmed) and attach notrack entry
1420 */
1421static inline void ip_vs_notrack(struct sk_buff *skb)
1422{
1423#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1424 enum ip_conntrack_info ctinfo;
1425 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1426
1427 if (!ct || !nf_ct_is_untracked(ct)) {
1428 nf_conntrack_put(skb->nfct);
1429 skb->nfct = &nf_ct_untracked_get()->ct_general;
1430 skb->nfctinfo = IP_CT_NEW;
1431 nf_conntrack_get(skb->nfct);
1432 }
1433#endif
1434}
1435
1436#ifdef CONFIG_IP_VS_NFCT
1437/*
1438 * Netfilter connection tracking
1439 * (from ip_vs_nfct.c)
1440 */
1441static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1442{
1443#ifdef CONFIG_SYSCTL
1444 return ipvs->sysctl_conntrack;
1445#else
1446 return 0;
1447#endif
1448}
1449
1450extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1451 int outin);
1452extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1453extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1454 struct ip_vs_conn *cp, u_int8_t proto,
1455 const __be16 port, int from_rs);
1456extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1457
1458#else
1459
1460static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1461{
1462 return 0;
1463}
1464
1465static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1466 struct ip_vs_conn *cp, int outin)
1467{
1468}
1469
1470static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1471{
1472 return NF_ACCEPT;
1473}
1474
1475static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1476{
1477}
1478/* CONFIG_IP_VS_NFCT */
1479#endif
1480
1481static inline unsigned int
1482ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1483{
1484 /*
1485 * We think the overhead of processing active connections is 256
1486 * times higher than that of inactive connections in average. (This
1487 * 256 times might not be accurate, we will change it later) We
1488 * use the following formula to estimate the overhead now:
1489 * dest->activeconns*256 + dest->inactconns
1490 */
1491 return (atomic_read(&dest->activeconns) << 8) +
1492 atomic_read(&dest->inactconns);
1493}
1494
1495#endif /* _NET_IP_VS_H */