]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - include/net/ip_vs.h
ipvs: wakeup master thread
[mirror_ubuntu-artful-kernel.git] / include / net / ip_vs.h
CommitLineData
1da177e4
LT
1/*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
bc4768eb
JV
6#ifndef _NET_IP_VS_H
7#define _NET_IP_VS_H
1da177e4 8
bc4768eb 9#include <linux/ip_vs.h> /* definitions shared with userland */
1da177e4 10
bc4768eb
JV
11#include <asm/types.h> /* for __uXX types */
12
1da177e4
LT
13#include <linux/list.h> /* for struct list_head */
14#include <linux/spinlock.h> /* for struct rwlock_t */
60063497 15#include <linux/atomic.h> /* for struct atomic_t */
1da177e4 16#include <linux/compiler.h>
14c85021 17#include <linux/timer.h>
187f1882 18#include <linux/bug.h>
1da177e4 19
14c85021 20#include <net/checksum.h>
e7ade46a 21#include <linux/netfilter.h> /* for union nf_inet_addr */
1668e010 22#include <linux/ip.h>
e7ade46a 23#include <linux/ipv6.h> /* for struct ipv6hdr */
4e3fd7a0 24#include <net/ipv6.h>
cf356d69 25#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
f4bc17cd
JA
26#include <net/netfilter/nf_conntrack.h>
27#endif
61b1ab45
HS
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}
fc723250
HS
37/*
38 * Get net ptr from skb in traffic cases
39 * use skb_sknet when call is from userland (ioctl or netlink)
40 */
a0840e2e 41static inline struct net *skb_net(const struct sk_buff *skb)
fc723250
HS
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);
a09d1977 52 if (skb_dst(skb) && skb_dst(skb)->dev)
fc723250
HS
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
a0840e2e 69static inline struct net *skb_sknet(const struct sk_buff *skb)
fc723250
HS
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}
b17fc996
HS
90/*
91 * This one needed for single_open_net since net is stored directly in
25985edc 92 * private not as a struct i.e. seq_file_net can't be used.
b17fc996
HS
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}
6f7edb48
CB
102
103/* Connections' size value needed by ip_vs_ctl.c */
104extern int ip_vs_conn_tab_size;
105
106
64aae3cb
JV
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;
4e3fd7a0
AD
122 iphdr->saddr.in6 = iph->saddr;
123 iphdr->daddr.in6 = iph->daddr;
64aae3cb
JV
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)
4e3fd7a0 140 dst->in6 = src->in6;
64aae3cb
JV
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
1da177e4 156#ifdef CONFIG_IP_VS_DEBUG
14c85021
ACM
157#include <linux/net.h>
158
1da177e4 159extern int ip_vs_get_debug_level(void);
c842a3ad
JV
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)
5b095d98 168 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
0c6ce78a 169 &addr->in6) + 1;
c842a3ad
JV
170 else
171#endif
3685f25d
HH
172 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
173 &addr->ip) + 1;
c842a3ad
JV
174
175 *idx += len;
176 BUG_ON(*idx > buf_len + 1);
177 return &buf[*idx - len];
178}
179
9aada7ac
HE
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)
c842a3ad
JV
193
194/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
9aada7ac
HE
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)
0d79641a 210#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
9aada7ac
HE
211 do { \
212 if (level <= ip_vs_get_debug_level()) \
0d79641a 213 pp->debug_packet(af, pp, skb, ofs, msg); \
9aada7ac 214 } while (0)
0d79641a 215#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
9aada7ac
HE
216 do { \
217 if (level <= ip_vs_get_debug_level() && \
218 net_ratelimit()) \
0d79641a 219 pp->debug_packet(af, pp, skb, ofs, msg); \
9aada7ac 220 } while (0)
1da177e4 221#else /* NO DEBUGGING at ALL */
c842a3ad
JV
222#define IP_VS_DBG_BUF(level, msg...) do {} while (0)
223#define IP_VS_ERR_BUF(msg...) do {} while (0)
1da177e4
LT
224#define IP_VS_DBG(level, msg...) do {} while (0)
225#define IP_VS_DBG_RL(msg...) do {} while (0)
0d79641a
JA
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)
1da177e4
LT
228#endif
229
230#define IP_VS_BUG() BUG()
1e3e238e 231#define IP_VS_ERR_RL(msg, ...) \
9aada7ac
HE
232 do { \
233 if (net_ratelimit()) \
1e3e238e 234 pr_err(msg, ##__VA_ARGS__); \
9aada7ac 235 } while (0)
1da177e4
LT
236
237#ifdef CONFIG_IP_VS_DEBUG
238#define EnterFunction(level) \
9aada7ac
HE
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)
1da177e4
LT
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 */
f3a7c66b
HH
263#define FTPPORT cpu_to_be16(21)
264#define FTPDATA cpu_to_be16(20)
1da177e4 265
1da177e4
LT
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
2906f66a
VMR
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
1da177e4
LT
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
b17fc996
HS
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};
1da177e4
LT
349
350/*
3a14a313 351 * IPVS statistics objects
1da177e4 352 */
3a14a313
SW
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
fd2c3ef7 369struct ip_vs_stats {
b17fc996 370 struct ip_vs_stats_user ustats; /* statistics */
e9c0ce23 371 struct ip_vs_estimator est; /* estimator */
b17fc996
HS
372 struct ip_vs_cpu_stats *cpustats; /* per cpu counters */
373 spinlock_t lock; /* spin lock */
55a3d4e1 374 struct ip_vs_stats_user ustats0; /* reset values */
1da177e4
LT
375};
376
14c85021
ACM
377struct dst_entry;
378struct iphdr;
1da177e4
LT
379struct ip_vs_conn;
380struct ip_vs_app;
14c85021 381struct sk_buff;
252c6410 382struct ip_vs_proto_data;
1da177e4
LT
383
384struct ip_vs_protocol {
385 struct ip_vs_protocol *next;
386 char *name;
2ad17def
JA
387 u16 protocol;
388 u16 num_states;
1da177e4 389 int dont_defrag;
1da177e4
LT
390
391 void (*init)(struct ip_vs_protocol *pp);
392
393 void (*exit)(struct ip_vs_protocol *pp);
394
582b8e3e 395 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
252c6410
HS
396
397 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
398
51ef348b 399 int (*conn_schedule)(int af, struct sk_buff *skb,
9330419d 400 struct ip_vs_proto_data *pd,
1da177e4
LT
401 int *verdict, struct ip_vs_conn **cpp);
402
403 struct ip_vs_conn *
51ef348b
JV
404 (*conn_in_get)(int af,
405 const struct sk_buff *skb,
51ef348b 406 const struct ip_vs_iphdr *iph,
1da177e4
LT
407 unsigned int proto_off,
408 int inverse);
409
410 struct ip_vs_conn *
51ef348b
JV
411 (*conn_out_get)(int af,
412 const struct sk_buff *skb,
51ef348b 413 const struct ip_vs_iphdr *iph,
1da177e4
LT
414 unsigned int proto_off,
415 int inverse);
416
3db05fea 417 int (*snat_handler)(struct sk_buff *skb,
1da177e4
LT
418 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
419
3db05fea 420 int (*dnat_handler)(struct sk_buff *skb,
1da177e4
LT
421 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
422
51ef348b
JV
423 int (*csum_check)(int af, struct sk_buff *skb,
424 struct ip_vs_protocol *pp);
1da177e4
LT
425
426 const char *(*state_name)(int state);
427
4a516f11
SH
428 void (*state_transition)(struct ip_vs_conn *cp, int direction,
429 const struct sk_buff *skb,
430 struct ip_vs_proto_data *pd);
1da177e4 431
ab8a5e84 432 int (*register_app)(struct net *net, struct ip_vs_app *inc);
1da177e4 433
ab8a5e84 434 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
1da177e4
LT
435
436 int (*app_conn_bind)(struct ip_vs_conn *cp);
437
0d79641a 438 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
1da177e4
LT
439 const struct sk_buff *skb,
440 int offset,
441 const char *msg);
442
9330419d 443 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
1da177e4
LT
444};
445
252c6410
HS
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);
1da177e4 460
f11017ec 461struct ip_vs_conn_param {
6e67e586 462 struct net *net;
f11017ec
SH
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;
85999283
SH
469
470 const struct ip_vs_pe *pe;
471 char *pe_data;
472 __u8 pe_data_len;
f11017ec
SH
473};
474
1da177e4
LT
475/*
476 * IP_VS structure allocated for each dynamically scheduled connection
477 */
478struct ip_vs_conn {
731109e7 479 struct hlist_node c_list; /* hashed list heads */
6e67e586
HS
480#ifdef CONFIG_NET_NS
481 struct net *net; /* Name space */
482#endif
1da177e4 483 /* Protocol, addresses and port numbers */
6e67e586
HS
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 */
1da177e4
LT
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 */
1da177e4 502 volatile __u16 state; /* state info */
efac5276
RB
503 volatile __u16 old_state; /* old state, to be used for
504 * state transition triggerd
505 * synchronization
506 */
1da177e4
LT
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 **.
fc604767 517 NF_ACCEPT can be returned when destination is local.
1da177e4
LT
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 */
85999283 529
e9e5eee8 530 const struct ip_vs_pe *pe;
85999283
SH
531 char *pe_data;
532 __u8 pe_data_len;
1da177e4
LT
533};
534
6e67e586
HS
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}
1da177e4 562
c860c6b1
JV
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;
0d1e71b0 581 char *pe_name;
95c96174
ED
582 unsigned int flags; /* virtual service flags */
583 unsigned int timeout; /* persistent timeout in sec */
c860c6b1
JV
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 */
95c96174 594 unsigned int conn_flags; /* connection flags */
c860c6b1
JV
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
1da177e4
LT
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
e7ade46a 613 u16 af; /* address family */
1da177e4 614 __u16 protocol; /* which protocol (TCP/UDP) */
e7ade46a 615 union nf_inet_addr addr; /* IP address for virtual service */
014d730d 616 __be16 port; /* port number for the service */
1da177e4 617 __u32 fwmark; /* firewall mark of the service */
95c96174
ED
618 unsigned int flags; /* service status flags */
619 unsigned int timeout; /* persistent timeout in ticks */
014d730d 620 __be32 netmask; /* grouping granularity */
fc723250 621 struct net *net;
1da177e4
LT
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 */
85999283
SH
632
633 /* alternate persistence engine */
634 struct ip_vs_pe *pe;
1da177e4
LT
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
e7ade46a 646 u16 af; /* address family */
014d730d 647 __be16 port; /* port number of the server */
f2431e6e 648 union nf_inet_addr addr; /* IP address of the server */
95c96174 649 volatile unsigned int flags; /* dest status flags */
1da177e4
LT
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 */
714f095f 667 u32 dst_cookie;
c92f5ca2 668 union nf_inet_addr dst_saddr;
1da177e4
LT
669
670 /* for virtual service */
671 struct ip_vs_service *svc; /* service it belongs to */
672 __u16 protocol; /* which protocol (TCP/UDP) */
014d730d 673 __be16 vport; /* virtual port number */
f2431e6e 674 union nf_inet_addr vaddr; /* virtual IP address */
1da177e4
LT
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
85999283
SH
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);
a3c918ac 713 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
85999283 714};
1da177e4
LT
715
716/*
717 * The application module object (a.k.a. app incarnation)
718 */
fd2c3ef7 719struct ip_vs_app {
1da177e4
LT
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 */
014d730d 730 __be16 port; /* port number in net order */
1da177e4
LT
731 atomic_t usecnt; /* usage counter */
732
8b27b10f
JA
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 */
1da177e4 738 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
3db05fea 739 struct sk_buff *, int *diff);
1da177e4 740
8b27b10f
JA
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 */
1da177e4 746 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
3db05fea 747 struct sk_buff *, int *diff);
1da177e4
LT
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
2553d064
JA
786/* IPVS in network namespace */
787struct netns_ipvs {
788 int gen; /* Generation */
7a4f0761 789 int enable; /* enable like nf_hooks do */
2553d064
JA
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;
c74c0bfe
HS
800 /* ip_vs_ftp */
801 struct ip_vs_app *ftp_app;
2553d064
JA
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 */
2a0751af 834 struct ip_vs_stats tot_stats; /* Statistics & est. */
2553d064
JA
835
836 int num_services; /* no of virtual services */
f2247fbd
SH
837
838 rwlock_t rs_lock; /* real services table */
f2247fbd
SH
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
2553d064
JA
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 */
2553d064
JA
855
856 /* sys-ctl struct */
857 struct ctl_table_header *sysctl_hdr;
858 struct ctl_table *sysctl_tbl;
f2247fbd
SH
859#endif
860
2553d064
JA
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;
1c003b15
PNA
872 int sysctl_sync_qlen_max;
873 int sysctl_sync_sock_size;
2553d064
JA
874 int sysctl_cache_bypass;
875 int sysctl_expire_nodest_conn;
876 int sysctl_expire_quiescent_template;
877 int sysctl_sync_threshold[2];
878 int sysctl_nat_icmp_send;
879
880 /* ip_vs_lblc */
881 int sysctl_lblc_expiration;
882 struct ctl_table_header *lblc_ctl_header;
883 struct ctl_table *lblc_ctl_table;
884 /* ip_vs_lblcr */
885 int sysctl_lblcr_expiration;
886 struct ctl_table_header *lblcr_ctl_header;
887 struct ctl_table *lblcr_ctl_table;
888 /* ip_vs_est */
889 struct list_head est_list; /* estimator list */
890 spinlock_t est_lock;
891 struct timer_list est_timer; /* Estimation timer */
892 /* ip_vs_sync */
893 struct list_head sync_queue;
1c003b15
PNA
894 int sync_queue_len;
895 unsigned int sync_queue_delay;
896 struct delayed_work master_wakeup_work;
2553d064
JA
897 spinlock_t sync_lock;
898 struct ip_vs_sync_buff *sync_buff;
899 spinlock_t sync_buff_lock;
900 struct sockaddr_in sync_mcast_addr;
901 struct task_struct *master_thread;
902 struct task_struct *backup_thread;
903 int send_mesg_maxlen;
904 int recv_mesg_maxlen;
905 volatile int sync_state;
906 volatile int master_syncid;
907 volatile int backup_syncid;
ae1d48b2 908 struct mutex sync_mutex;
2553d064
JA
909 /* multicast interface name */
910 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
911 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
912 /* net name space ptr */
913 struct net *net; /* Needed by timer routines */
914};
1da177e4 915
59e0350e
SH
916#define DEFAULT_SYNC_THRESHOLD 3
917#define DEFAULT_SYNC_PERIOD 50
7532e8d4 918#define DEFAULT_SYNC_VER 1
1c003b15
PNA
919#define IPVS_SYNC_WAKEUP_RATE 8
920#define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
921#define IPVS_SYNC_SEND_DELAY (HZ / 50)
922#define IPVS_SYNC_CHECK_PERIOD HZ
59e0350e
SH
923
924#ifdef CONFIG_SYSCTL
925
926static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
927{
928 return ipvs->sysctl_sync_threshold[0];
929}
930
931static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
932{
933 return ipvs->sysctl_sync_threshold[1];
934}
935
7532e8d4
SH
936static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
937{
938 return ipvs->sysctl_sync_ver;
939}
940
1c003b15
PNA
941static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
942{
943 return ipvs->sysctl_sync_qlen_max;
944}
945
946static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
947{
948 return ipvs->sysctl_sync_sock_size;
949}
950
59e0350e
SH
951#else
952
953static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
954{
955 return DEFAULT_SYNC_THRESHOLD;
956}
957
958static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
959{
960 return DEFAULT_SYNC_PERIOD;
961}
962
7532e8d4
SH
963static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
964{
965 return DEFAULT_SYNC_VER;
966}
967
1c003b15
PNA
968static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
969{
970 return IPVS_SYNC_QLEN_MAX;
971}
972
973static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
974{
975 return 0;
976}
977
59e0350e
SH
978#endif
979
1da177e4
LT
980/*
981 * IPVS core functions
982 * (from ip_vs_core.c)
983 */
95c96174 984extern const char *ip_vs_proto_name(unsigned int proto);
1da177e4 985extern void ip_vs_init_hash_table(struct list_head *table, int rows);
afdd6140 986#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1da177e4 987
1da177e4
LT
988#define IP_VS_APP_TYPE_FTP 1
989
990/*
991 * ip_vs_conn handling functions
992 * (from ip_vs_conn.c)
993 */
994
1da177e4
LT
995enum {
996 IP_VS_DIR_INPUT = 0,
997 IP_VS_DIR_OUTPUT,
998 IP_VS_DIR_INPUT_ONLY,
999 IP_VS_DIR_LAST,
1000};
1001
6e67e586 1002static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
f11017ec
SH
1003 const union nf_inet_addr *caddr,
1004 __be16 cport,
1005 const union nf_inet_addr *vaddr,
1006 __be16 vport,
1007 struct ip_vs_conn_param *p)
1008{
6e67e586 1009 p->net = net;
f11017ec
SH
1010 p->af = af;
1011 p->protocol = protocol;
1012 p->caddr = caddr;
1013 p->cport = cport;
1014 p->vaddr = vaddr;
1015 p->vport = vport;
85999283
SH
1016 p->pe = NULL;
1017 p->pe_data = NULL;
f11017ec 1018}
28364a59 1019
f11017ec
SH
1020struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1021struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
28364a59 1022
5c0d2374 1023struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
5c0d2374
SH
1024 const struct ip_vs_iphdr *iph,
1025 unsigned int proto_off,
1026 int inverse);
1027
f11017ec 1028struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1da177e4 1029
5c0d2374 1030struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
5c0d2374
SH
1031 const struct ip_vs_iphdr *iph,
1032 unsigned int proto_off,
1033 int inverse);
1034
1da177e4
LT
1035/* put back the conn without restarting its timer */
1036static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1037{
1038 atomic_dec(&cp->refcnt);
1039}
1040extern void ip_vs_conn_put(struct ip_vs_conn *cp);
014d730d 1041extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1da177e4 1042
f11017ec
SH
1043struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1044 const union nf_inet_addr *daddr,
95c96174 1045 __be16 dport, unsigned int flags,
0e051e68 1046 struct ip_vs_dest *dest, __u32 fwmark);
1da177e4
LT
1047extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1048
1049extern const char * ip_vs_state_name(__u16 proto, int state);
1050
4a85b96c 1051extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1da177e4 1052extern int ip_vs_check_template(struct ip_vs_conn *ct);
f6340ee0 1053extern void ip_vs_random_dropentry(struct net *net);
1da177e4
LT
1054extern int ip_vs_conn_init(void);
1055extern void ip_vs_conn_cleanup(void);
1056
1057static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1058{
1059 struct ip_vs_conn *ctl_cp = cp->control;
1060 if (!ctl_cp) {
cfc78c5a
JV
1061 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1062 "%s:%d to %s:%d\n",
1063 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1064 ntohs(cp->cport),
1065 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1066 ntohs(cp->vport));
1067
1da177e4
LT
1068 return;
1069 }
1070
cfc78c5a
JV
1071 IP_VS_DBG_BUF(7, "DELeting control for: "
1072 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1073 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1074 ntohs(cp->cport),
1075 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1076 ntohs(ctl_cp->cport));
1da177e4
LT
1077
1078 cp->control = NULL;
1079 if (atomic_read(&ctl_cp->n_control) == 0) {
cfc78c5a
JV
1080 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1081 "%s:%d to %s:%d\n",
1082 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1083 ntohs(cp->cport),
1084 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1085 ntohs(cp->vport));
1086
1da177e4
LT
1087 return;
1088 }
1089 atomic_dec(&ctl_cp->n_control);
1090}
1091
1092static inline void
1093ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1094{
1095 if (cp->control) {
cfc78c5a
JV
1096 IP_VS_ERR_BUF("request control ADD for already controlled: "
1097 "%s:%d to %s:%d\n",
1098 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1099 ntohs(cp->cport),
1100 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1101 ntohs(cp->vport));
1102
1da177e4
LT
1103 ip_vs_control_del(cp);
1104 }
1105
cfc78c5a
JV
1106 IP_VS_DBG_BUF(7, "ADDing control for: "
1107 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1108 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1109 ntohs(cp->cport),
1110 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1111 ntohs(ctl_cp->cport));
1da177e4
LT
1112
1113 cp->control = ctl_cp;
1114 atomic_inc(&ctl_cp->n_control);
1115}
1116
7a4f0761
HS
1117/*
1118 * IPVS netns init & cleanup functions
1119 */
503cf15a
HS
1120extern int ip_vs_estimator_net_init(struct net *net);
1121extern int ip_vs_control_net_init(struct net *net);
1122extern int ip_vs_protocol_net_init(struct net *net);
1123extern int ip_vs_app_net_init(struct net *net);
1124extern int ip_vs_conn_net_init(struct net *net);
1125extern int ip_vs_sync_net_init(struct net *net);
1126extern void ip_vs_conn_net_cleanup(struct net *net);
1127extern void ip_vs_app_net_cleanup(struct net *net);
1128extern void ip_vs_protocol_net_cleanup(struct net *net);
1129extern void ip_vs_control_net_cleanup(struct net *net);
1130extern void ip_vs_estimator_net_cleanup(struct net *net);
1131extern void ip_vs_sync_net_cleanup(struct net *net);
1132extern void ip_vs_service_net_cleanup(struct net *net);
1da177e4
LT
1133
1134/*
1135 * IPVS application functions
1136 * (from ip_vs_app.c)
1137 */
1138#define IP_VS_APP_MAX_PORTS 8
ab8a5e84
HS
1139extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1140extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1da177e4
LT
1141extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1142extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
ab8a5e84
HS
1143extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1144 __u16 proto, __u16 port);
1da177e4
LT
1145extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1146extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1147
3db05fea
HX
1148extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1149extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1da177e4 1150
8be67a66
SH
1151void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1152void ip_vs_unbind_pe(struct ip_vs_service *svc);
1153int register_ip_vs_pe(struct ip_vs_pe *pe);
1154int unregister_ip_vs_pe(struct ip_vs_pe *pe);
e9e5eee8 1155struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
fe5e7a1e 1156struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
e9e5eee8 1157
69e7dae4
PG
1158/*
1159 * Use a #define to avoid all of module.h just for these trivial ops
1160 */
1161#define ip_vs_pe_get(pe) \
1162 if (pe && pe->module) \
e9e5eee8 1163 __module_get(pe->module);
e9e5eee8 1164
69e7dae4
PG
1165#define ip_vs_pe_put(pe) \
1166 if (pe && pe->module) \
e9e5eee8 1167 module_put(pe->module);
1da177e4
LT
1168
1169/*
1170 * IPVS protocol functions (from ip_vs_proto.c)
1171 */
1172extern int ip_vs_protocol_init(void);
1173extern void ip_vs_protocol_cleanup(void);
9330419d 1174extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1da177e4
LT
1175extern int *ip_vs_create_timeout_table(int *table, int size);
1176extern int
36cbd3dc
JE
1177ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1178 const char *name, int to);
1da177e4 1179extern void
0d79641a
JA
1180ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1181 const struct sk_buff *skb,
1da177e4
LT
1182 int offset, const char *msg);
1183
1184extern struct ip_vs_protocol ip_vs_protocol_tcp;
1185extern struct ip_vs_protocol ip_vs_protocol_udp;
1186extern struct ip_vs_protocol ip_vs_protocol_icmp;
1187extern struct ip_vs_protocol ip_vs_protocol_esp;
1188extern struct ip_vs_protocol ip_vs_protocol_ah;
2906f66a 1189extern struct ip_vs_protocol ip_vs_protocol_sctp;
1da177e4
LT
1190
1191/*
1192 * Registering/unregistering scheduler functions
1193 * (from ip_vs_sched.c)
1194 */
1195extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1196extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1197extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1198 struct ip_vs_scheduler *scheduler);
1199extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1200extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1201extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1202extern struct ip_vs_conn *
190ecd27 1203ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
9330419d 1204 struct ip_vs_proto_data *pd, int *ignored);
1da177e4 1205extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
9330419d 1206 struct ip_vs_proto_data *pd);
1da177e4 1207
41ac51ee
PS
1208extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1209
1da177e4
LT
1210
1211/*
1212 * IPVS control data and functions (from ip_vs_ctl.c)
1213 */
1da177e4 1214extern struct ip_vs_stats ip_vs_stats;
b880c1f0 1215extern int sysctl_ip_vs_sync_ver;
1da177e4 1216
f131315f 1217extern void ip_vs_sync_switch_mode(struct net *net, int mode);
1da177e4 1218extern struct ip_vs_service *
fc723250 1219ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
3c2e0505 1220 const union nf_inet_addr *vaddr, __be16 vport);
1da177e4
LT
1221
1222static inline void ip_vs_service_put(struct ip_vs_service *svc)
1223{
1224 atomic_dec(&svc->usecnt);
1225}
1226
1227extern struct ip_vs_dest *
fc723250 1228ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
7937df15
JV
1229 const union nf_inet_addr *daddr, __be16 dport);
1230
1da177e4
LT
1231extern int ip_vs_use_count_inc(void);
1232extern void ip_vs_use_count_dec(void);
8537de8a
HS
1233extern int ip_vs_register_nl_ioctl(void);
1234extern void ip_vs_unregister_nl_ioctl(void);
1da177e4
LT
1235extern int ip_vs_control_init(void);
1236extern void ip_vs_control_cleanup(void);
1e356f9c 1237extern struct ip_vs_dest *
fc723250
HS
1238ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1239 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
52793dbe 1240 __u16 protocol, __u32 fwmark, __u32 flags);
1e356f9c 1241extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1da177e4
LT
1242
1243
1244/*
1245 * IPVS sync daemon data and function prototypes
1246 * (from ip_vs_sync.c)
1247 */
f131315f
HS
1248extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1249 __u8 syncid);
1250extern int stop_sync_thread(struct net *net, int state);
1251extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
1da177e4
LT
1252
1253
1254/*
1255 * IPVS rate estimator prototypes (from ip_vs_est.c)
1256 */
6ef757f9
JA
1257extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1258extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1da177e4 1259extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
ea9f22cc
JA
1260extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1261 struct ip_vs_stats *stats);
1da177e4
LT
1262
1263/*
1264 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1265 */
1266extern int ip_vs_null_xmit
1267(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1268extern int ip_vs_bypass_xmit
1269(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1270extern int ip_vs_nat_xmit
1271(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1272extern int ip_vs_tunnel_xmit
1273(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1274extern int ip_vs_dr_xmit
1275(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1276extern int ip_vs_icmp_xmit
c92f5ca2
JA
1277(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1278 int offset, unsigned int hooknum);
1da177e4
LT
1279extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1280
b3cdd2a7
JV
1281#ifdef CONFIG_IP_VS_IPV6
1282extern int ip_vs_bypass_xmit_v6
1283(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1284extern int ip_vs_nat_xmit_v6
1285(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1286extern int ip_vs_tunnel_xmit_v6
1287(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1288extern int ip_vs_dr_xmit_v6
1289(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1290extern int ip_vs_icmp_xmit_v6
1291(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
c92f5ca2 1292 int offset, unsigned int hooknum);
b3cdd2a7 1293#endif
1da177e4 1294
3a1bbf18 1295#ifdef CONFIG_SYSCTL
1da177e4
LT
1296/*
1297 * This is a simple mechanism to ignore packets when
1298 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1299 * we start to drop 1/rate of the packets
1300 */
1da177e4 1301
a0840e2e 1302static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1da177e4 1303{
a0840e2e
HS
1304 if (!ipvs->drop_rate)
1305 return 0;
1306 if (--ipvs->drop_counter > 0)
1307 return 0;
1308 ipvs->drop_counter = ipvs->drop_rate;
1da177e4
LT
1309 return 1;
1310}
3a1bbf18
SH
1311#else
1312static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1313#endif
1da177e4
LT
1314
1315/*
1316 * ip_vs_fwd_tag returns the forwarding tag of the connection
1317 */
1318#define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1319
732db659 1320static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1da177e4
LT
1321{
1322 char fwd;
1323
1324 switch (IP_VS_FWD_METHOD(cp)) {
1325 case IP_VS_CONN_F_MASQ:
1326 fwd = 'M'; break;
1327 case IP_VS_CONN_F_LOCALNODE:
1328 fwd = 'L'; break;
1329 case IP_VS_CONN_F_TUNNEL:
1330 fwd = 'T'; break;
1331 case IP_VS_CONN_F_DROUTE:
1332 fwd = 'R'; break;
1333 case IP_VS_CONN_F_BYPASS:
1334 fwd = 'B'; break;
1335 default:
1336 fwd = '?'; break;
1337 }
1338 return fwd;
1339}
1340
1da177e4 1341extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
b3cdd2a7
JV
1342 struct ip_vs_conn *cp, int dir);
1343
1344#ifdef CONFIG_IP_VS_IPV6
1345extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1346 struct ip_vs_conn *cp, int dir);
1347#endif
1da177e4 1348
b1550f22 1349extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1da177e4 1350
f9214b26 1351static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1da177e4 1352{
f9214b26 1353 __be32 diff[2] = { ~old, new };
1da177e4 1354
07f0757a 1355 return csum_partial(diff, sizeof(diff), oldsum);
f9214b26
AV
1356}
1357
0bbdd42b
JV
1358#ifdef CONFIG_IP_VS_IPV6
1359static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1360 __wsum oldsum)
1361{
1362 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1363 new[3], new[2], new[1], new[0] };
1364
07f0757a 1365 return csum_partial(diff, sizeof(diff), oldsum);
0bbdd42b
JV
1366}
1367#endif
1368
f9214b26
AV
1369static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1370{
1371 __be16 diff[2] = { ~old, new };
1372
07f0757a 1373 return csum_partial(diff, sizeof(diff), oldsum);
1da177e4
LT
1374}
1375
cf356d69
JA
1376/*
1377 * Forget current conntrack (unconfirmed) and attach notrack entry
1378 */
1379static inline void ip_vs_notrack(struct sk_buff *skb)
1380{
1381#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1382 enum ip_conntrack_info ctinfo;
06b69390 1383 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
cf356d69
JA
1384
1385 if (!ct || !nf_ct_is_untracked(ct)) {
1386 nf_reset(skb);
1387 skb->nfct = &nf_ct_untracked_get()->ct_general;
1388 skb->nfctinfo = IP_CT_NEW;
1389 nf_conntrack_get(skb->nfct);
1390 }
1391#endif
1392}
1393
f4bc17cd
JA
1394#ifdef CONFIG_IP_VS_NFCT
1395/*
1396 * Netfilter connection tracking
1397 * (from ip_vs_nfct.c)
1398 */
a0840e2e 1399static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
f4bc17cd 1400{
a4e2f5a7 1401#ifdef CONFIG_SYSCTL
a0840e2e 1402 return ipvs->sysctl_conntrack;
a4e2f5a7
SH
1403#else
1404 return 0;
1405#endif
f4bc17cd
JA
1406}
1407
6523ce15
JA
1408extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1409 int outin);
3c2de2ae 1410extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
f4bc17cd
JA
1411extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1412 struct ip_vs_conn *cp, u_int8_t proto,
1413 const __be16 port, int from_rs);
1414extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1415
1416#else
1417
a0840e2e 1418static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
f4bc17cd
JA
1419{
1420 return 0;
1421}
1422
1423static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1424 struct ip_vs_conn *cp, int outin)
1425{
1426}
1427
e23ebf0f 1428static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
f4bc17cd
JA
1429{
1430 return NF_ACCEPT;
1431}
1432
1433static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1434{
1435}
1436/* CONFIG_IP_VS_NFCT */
1437#endif
6523ce15 1438
b552f7e3
CG
1439static inline unsigned int
1440ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1441{
1442 /*
1443 * We think the overhead of processing active connections is 256
1444 * times higher than that of inactive connections in average. (This
1445 * 256 times might not be accurate, we will change it later) We
1446 * use the following formula to estimate the overhead now:
1447 * dest->activeconns*256 + dest->inactconns
1448 */
1449 return (atomic_read(&dest->activeconns) << 8) +
1450 atomic_read(&dest->inactconns);
1451}
1452
bc4768eb 1453#endif /* _NET_IP_VS_H */