]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - include/net/ip_vs.h
ipvs: move struct netns_ipvs
[mirror_ubuntu-bionic-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 11/* old ipvsadm versions still include this file directly */
1da177e4
LT
12#ifdef __KERNEL__
13
bc4768eb
JV
14#include <asm/types.h> /* for __uXX types */
15
16#include <linux/sysctl.h> /* for ctl_path */
1da177e4
LT
17#include <linux/list.h> /* for struct list_head */
18#include <linux/spinlock.h> /* for struct rwlock_t */
1da177e4 19#include <asm/atomic.h> /* for struct atomic_t */
1da177e4 20#include <linux/compiler.h>
14c85021 21#include <linux/timer.h>
1da177e4 22
14c85021 23#include <net/checksum.h>
e7ade46a 24#include <linux/netfilter.h> /* for union nf_inet_addr */
1668e010 25#include <linux/ip.h>
e7ade46a
JV
26#include <linux/ipv6.h> /* for struct ipv6hdr */
27#include <net/ipv6.h> /* for ipv6_addr_copy */
cf356d69 28#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
f4bc17cd
JA
29#include <net/netfilter/nf_conntrack.h>
30#endif
61b1ab45
HS
31#include <net/net_namespace.h> /* Netw namespace */
32
33/*
34 * Generic access of ipvs struct
35 */
36static inline struct netns_ipvs *net_ipvs(struct net* net)
37{
38 return net->ipvs;
39}
fc723250
HS
40/*
41 * Get net ptr from skb in traffic cases
42 * use skb_sknet when call is from userland (ioctl or netlink)
43 */
a0840e2e 44static inline struct net *skb_net(const struct sk_buff *skb)
fc723250
HS
45{
46#ifdef CONFIG_NET_NS
47#ifdef CONFIG_IP_VS_DEBUG
48 /*
49 * This is used for debug only.
50 * Start with the most likely hit
51 * End with BUG
52 */
53 if (likely(skb->dev && skb->dev->nd_net))
54 return dev_net(skb->dev);
55 if (skb_dst(skb)->dev)
56 return dev_net(skb_dst(skb)->dev);
57 WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
58 __func__, __LINE__);
59 if (likely(skb->sk && skb->sk->sk_net))
60 return sock_net(skb->sk);
61 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
62 __func__, __LINE__);
63 BUG();
64#else
65 return dev_net(skb->dev ? : skb_dst(skb)->dev);
66#endif
67#else
68 return &init_net;
69#endif
70}
71
a0840e2e 72static inline struct net *skb_sknet(const struct sk_buff *skb)
fc723250
HS
73{
74#ifdef CONFIG_NET_NS
75#ifdef CONFIG_IP_VS_DEBUG
76 /* Start with the most likely hit */
77 if (likely(skb->sk && skb->sk->sk_net))
78 return sock_net(skb->sk);
79 WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
80 __func__, __LINE__);
81 if (likely(skb->dev && skb->dev->nd_net))
82 return dev_net(skb->dev);
83 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
84 __func__, __LINE__);
85 BUG();
86#else
87 return sock_net(skb->sk);
88#endif
89#else
90 return &init_net;
91#endif
92}
b17fc996
HS
93/*
94 * This one needed for single_open_net since net is stored directly in
95 * private not as a struct i.e. seq_file_net cant be used.
96 */
97static inline struct net *seq_file_single_net(struct seq_file *seq)
98{
99#ifdef CONFIG_NET_NS
100 return (struct net *)seq->private;
101#else
102 return &init_net;
103#endif
104}
6f7edb48
CB
105
106/* Connections' size value needed by ip_vs_ctl.c */
107extern int ip_vs_conn_tab_size;
108
109
64aae3cb
JV
110struct ip_vs_iphdr {
111 int len;
112 __u8 protocol;
113 union nf_inet_addr saddr;
114 union nf_inet_addr daddr;
115};
116
117static inline void
118ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
119{
120#ifdef CONFIG_IP_VS_IPV6
121 if (af == AF_INET6) {
122 const struct ipv6hdr *iph = nh;
123 iphdr->len = sizeof(struct ipv6hdr);
124 iphdr->protocol = iph->nexthdr;
125 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
126 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
127 } else
128#endif
129 {
130 const struct iphdr *iph = nh;
131 iphdr->len = iph->ihl * 4;
132 iphdr->protocol = iph->protocol;
133 iphdr->saddr.ip = iph->saddr;
134 iphdr->daddr.ip = iph->daddr;
135 }
136}
137
138static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
139 const union nf_inet_addr *src)
140{
141#ifdef CONFIG_IP_VS_IPV6
142 if (af == AF_INET6)
143 ipv6_addr_copy(&dst->in6, &src->in6);
144 else
145#endif
146 dst->ip = src->ip;
147}
148
149static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
150 const union nf_inet_addr *b)
151{
152#ifdef CONFIG_IP_VS_IPV6
153 if (af == AF_INET6)
154 return ipv6_addr_equal(&a->in6, &b->in6);
155#endif
156 return a->ip == b->ip;
157}
158
1da177e4 159#ifdef CONFIG_IP_VS_DEBUG
14c85021
ACM
160#include <linux/net.h>
161
1da177e4 162extern int ip_vs_get_debug_level(void);
c842a3ad
JV
163
164static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
165 const union nf_inet_addr *addr,
166 int *idx)
167{
168 int len;
169#ifdef CONFIG_IP_VS_IPV6
170 if (af == AF_INET6)
5b095d98 171 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
0c6ce78a 172 &addr->in6) + 1;
c842a3ad
JV
173 else
174#endif
3685f25d
HH
175 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
176 &addr->ip) + 1;
c842a3ad
JV
177
178 *idx += len;
179 BUG_ON(*idx > buf_len + 1);
180 return &buf[*idx - len];
181}
182
9aada7ac
HE
183#define IP_VS_DBG_BUF(level, msg, ...) \
184 do { \
185 char ip_vs_dbg_buf[160]; \
186 int ip_vs_dbg_idx = 0; \
187 if (level <= ip_vs_get_debug_level()) \
188 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
189 } while (0)
190#define IP_VS_ERR_BUF(msg...) \
191 do { \
192 char ip_vs_dbg_buf[160]; \
193 int ip_vs_dbg_idx = 0; \
194 pr_err(msg); \
195 } while (0)
c842a3ad
JV
196
197/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
9aada7ac
HE
198#define IP_VS_DBG_ADDR(af, addr) \
199 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
200 sizeof(ip_vs_dbg_buf), addr, \
201 &ip_vs_dbg_idx)
202
203#define IP_VS_DBG(level, msg, ...) \
204 do { \
205 if (level <= ip_vs_get_debug_level()) \
206 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
207 } while (0)
208#define IP_VS_DBG_RL(msg, ...) \
209 do { \
210 if (net_ratelimit()) \
211 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
212 } while (0)
0d79641a 213#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
9aada7ac
HE
214 do { \
215 if (level <= ip_vs_get_debug_level()) \
0d79641a 216 pp->debug_packet(af, pp, skb, ofs, msg); \
9aada7ac 217 } while (0)
0d79641a 218#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
9aada7ac
HE
219 do { \
220 if (level <= ip_vs_get_debug_level() && \
221 net_ratelimit()) \
0d79641a 222 pp->debug_packet(af, pp, skb, ofs, msg); \
9aada7ac 223 } while (0)
1da177e4 224#else /* NO DEBUGGING at ALL */
c842a3ad
JV
225#define IP_VS_DBG_BUF(level, msg...) do {} while (0)
226#define IP_VS_ERR_BUF(msg...) do {} while (0)
1da177e4
LT
227#define IP_VS_DBG(level, msg...) do {} while (0)
228#define IP_VS_DBG_RL(msg...) do {} while (0)
0d79641a
JA
229#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
230#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
1da177e4
LT
231#endif
232
233#define IP_VS_BUG() BUG()
1e3e238e 234#define IP_VS_ERR_RL(msg, ...) \
9aada7ac
HE
235 do { \
236 if (net_ratelimit()) \
1e3e238e 237 pr_err(msg, ##__VA_ARGS__); \
9aada7ac 238 } while (0)
1da177e4
LT
239
240#ifdef CONFIG_IP_VS_DEBUG
241#define EnterFunction(level) \
9aada7ac
HE
242 do { \
243 if (level <= ip_vs_get_debug_level()) \
244 printk(KERN_DEBUG \
245 pr_fmt("Enter: %s, %s line %i\n"), \
246 __func__, __FILE__, __LINE__); \
247 } while (0)
248#define LeaveFunction(level) \
249 do { \
250 if (level <= ip_vs_get_debug_level()) \
251 printk(KERN_DEBUG \
252 pr_fmt("Leave: %s, %s line %i\n"), \
253 __func__, __FILE__, __LINE__); \
254 } while (0)
1da177e4
LT
255#else
256#define EnterFunction(level) do {} while (0)
257#define LeaveFunction(level) do {} while (0)
258#endif
259
260#define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
261
262
263/*
264 * The port number of FTP service (in network order).
265 */
f3a7c66b
HH
266#define FTPPORT cpu_to_be16(21)
267#define FTPDATA cpu_to_be16(20)
1da177e4 268
1da177e4
LT
269/*
270 * TCP State Values
271 */
272enum {
273 IP_VS_TCP_S_NONE = 0,
274 IP_VS_TCP_S_ESTABLISHED,
275 IP_VS_TCP_S_SYN_SENT,
276 IP_VS_TCP_S_SYN_RECV,
277 IP_VS_TCP_S_FIN_WAIT,
278 IP_VS_TCP_S_TIME_WAIT,
279 IP_VS_TCP_S_CLOSE,
280 IP_VS_TCP_S_CLOSE_WAIT,
281 IP_VS_TCP_S_LAST_ACK,
282 IP_VS_TCP_S_LISTEN,
283 IP_VS_TCP_S_SYNACK,
284 IP_VS_TCP_S_LAST
285};
286
287/*
288 * UDP State Values
289 */
290enum {
291 IP_VS_UDP_S_NORMAL,
292 IP_VS_UDP_S_LAST,
293};
294
295/*
296 * ICMP State Values
297 */
298enum {
299 IP_VS_ICMP_S_NORMAL,
300 IP_VS_ICMP_S_LAST,
301};
302
2906f66a
VMR
303/*
304 * SCTP State Values
305 */
306enum ip_vs_sctp_states {
307 IP_VS_SCTP_S_NONE,
308 IP_VS_SCTP_S_INIT_CLI,
309 IP_VS_SCTP_S_INIT_SER,
310 IP_VS_SCTP_S_INIT_ACK_CLI,
311 IP_VS_SCTP_S_INIT_ACK_SER,
312 IP_VS_SCTP_S_ECHO_CLI,
313 IP_VS_SCTP_S_ECHO_SER,
314 IP_VS_SCTP_S_ESTABLISHED,
315 IP_VS_SCTP_S_SHUT_CLI,
316 IP_VS_SCTP_S_SHUT_SER,
317 IP_VS_SCTP_S_SHUT_ACK_CLI,
318 IP_VS_SCTP_S_SHUT_ACK_SER,
319 IP_VS_SCTP_S_CLOSED,
320 IP_VS_SCTP_S_LAST
321};
322
1da177e4
LT
323/*
324 * Delta sequence info structure
325 * Each ip_vs_conn has 2 (output AND input seq. changes).
326 * Only used in the VS/NAT.
327 */
328struct ip_vs_seq {
329 __u32 init_seq; /* Add delta from this seq */
330 __u32 delta; /* Delta in sequence numbers */
331 __u32 previous_delta; /* Delta in sequence numbers
332 before last resized pkt */
333};
334
b17fc996
HS
335/*
336 * counters per cpu
337 */
338struct ip_vs_counters {
339 __u32 conns; /* connections scheduled */
340 __u32 inpkts; /* incoming packets */
341 __u32 outpkts; /* outgoing packets */
342 __u64 inbytes; /* incoming bytes */
343 __u64 outbytes; /* outgoing bytes */
344};
345/*
346 * Stats per cpu
347 */
348struct ip_vs_cpu_stats {
349 struct ip_vs_counters ustats;
350 struct u64_stats_sync syncp;
351};
1da177e4
LT
352
353/*
3a14a313 354 * IPVS statistics objects
1da177e4 355 */
3a14a313
SW
356struct ip_vs_estimator {
357 struct list_head list;
358
359 u64 last_inbytes;
360 u64 last_outbytes;
361 u32 last_conns;
362 u32 last_inpkts;
363 u32 last_outpkts;
364
365 u32 cps;
366 u32 inpps;
367 u32 outpps;
368 u32 inbps;
369 u32 outbps;
370};
371
fd2c3ef7 372struct ip_vs_stats {
b17fc996 373 struct ip_vs_stats_user ustats; /* statistics */
e9c0ce23 374 struct ip_vs_estimator est; /* estimator */
b17fc996
HS
375 struct ip_vs_cpu_stats *cpustats; /* per cpu counters */
376 spinlock_t lock; /* spin lock */
1da177e4
LT
377};
378
b17fc996
HS
379/*
380 * Helper Macros for per cpu
381 * ipvs->tot_stats->ustats.count
382 */
383#define IPVS_STAT_INC(ipvs, count) \
384 __this_cpu_inc((ipvs)->ustats->count)
385
386#define IPVS_STAT_ADD(ipvs, count, value) \
387 do {\
388 write_seqcount_begin(per_cpu_ptr((ipvs)->ustats_seq, \
389 raw_smp_processor_id())); \
390 __this_cpu_add((ipvs)->ustats->count, value); \
391 write_seqcount_end(per_cpu_ptr((ipvs)->ustats_seq, \
392 raw_smp_processor_id())); \
393 } while (0)
394
14c85021
ACM
395struct dst_entry;
396struct iphdr;
1da177e4
LT
397struct ip_vs_conn;
398struct ip_vs_app;
14c85021 399struct sk_buff;
252c6410 400struct ip_vs_proto_data;
1da177e4
LT
401
402struct ip_vs_protocol {
403 struct ip_vs_protocol *next;
404 char *name;
2ad17def
JA
405 u16 protocol;
406 u16 num_states;
1da177e4 407 int dont_defrag;
1da177e4
LT
408
409 void (*init)(struct ip_vs_protocol *pp);
410
411 void (*exit)(struct ip_vs_protocol *pp);
412
252c6410
HS
413 void (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
414
415 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
416
51ef348b 417 int (*conn_schedule)(int af, struct sk_buff *skb,
9330419d 418 struct ip_vs_proto_data *pd,
1da177e4
LT
419 int *verdict, struct ip_vs_conn **cpp);
420
421 struct ip_vs_conn *
51ef348b
JV
422 (*conn_in_get)(int af,
423 const struct sk_buff *skb,
51ef348b 424 const struct ip_vs_iphdr *iph,
1da177e4
LT
425 unsigned int proto_off,
426 int inverse);
427
428 struct ip_vs_conn *
51ef348b
JV
429 (*conn_out_get)(int af,
430 const struct sk_buff *skb,
51ef348b 431 const struct ip_vs_iphdr *iph,
1da177e4
LT
432 unsigned int proto_off,
433 int inverse);
434
3db05fea 435 int (*snat_handler)(struct sk_buff *skb,
1da177e4
LT
436 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
437
3db05fea 438 int (*dnat_handler)(struct sk_buff *skb,
1da177e4
LT
439 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
440
51ef348b
JV
441 int (*csum_check)(int af, struct sk_buff *skb,
442 struct ip_vs_protocol *pp);
1da177e4
LT
443
444 const char *(*state_name)(int state);
445
446 int (*state_transition)(struct ip_vs_conn *cp, int direction,
447 const struct sk_buff *skb,
9330419d 448 struct ip_vs_proto_data *pd);
1da177e4 449
ab8a5e84 450 int (*register_app)(struct net *net, struct ip_vs_app *inc);
1da177e4 451
ab8a5e84 452 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
1da177e4
LT
453
454 int (*app_conn_bind)(struct ip_vs_conn *cp);
455
0d79641a 456 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
1da177e4
LT
457 const struct sk_buff *skb,
458 int offset,
459 const char *msg);
460
9330419d 461 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
1da177e4
LT
462};
463
252c6410
HS
464/*
465 * protocol data per netns
466 */
467struct ip_vs_proto_data {
468 struct ip_vs_proto_data *next;
469 struct ip_vs_protocol *pp;
470 int *timeout_table; /* protocol timeout table */
471 atomic_t appcnt; /* counter of proto app incs. */
472 struct tcp_states_t *tcp_state_table;
473};
474
475extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
476extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
477 unsigned short proto);
1da177e4 478
f11017ec 479struct ip_vs_conn_param {
6e67e586 480 struct net *net;
f11017ec
SH
481 const union nf_inet_addr *caddr;
482 const union nf_inet_addr *vaddr;
483 __be16 cport;
484 __be16 vport;
485 __u16 protocol;
486 u16 af;
85999283
SH
487
488 const struct ip_vs_pe *pe;
489 char *pe_data;
490 __u8 pe_data_len;
f11017ec
SH
491};
492
1da177e4
LT
493/*
494 * IP_VS structure allocated for each dynamically scheduled connection
495 */
496struct ip_vs_conn {
731109e7 497 struct hlist_node c_list; /* hashed list heads */
6e67e586
HS
498#ifdef CONFIG_NET_NS
499 struct net *net; /* Name space */
500#endif
1da177e4 501 /* Protocol, addresses and port numbers */
6e67e586
HS
502 u16 af; /* address family */
503 __be16 cport;
504 __be16 vport;
505 __be16 dport;
506 __u32 fwmark; /* Fire wall mark from skb */
507 union nf_inet_addr caddr; /* client address */
508 union nf_inet_addr vaddr; /* virtual address */
509 union nf_inet_addr daddr; /* destination address */
510 volatile __u32 flags; /* status flags */
1da177e4
LT
511 __u16 protocol; /* Which protocol (TCP/UDP) */
512
513 /* counter and timer */
514 atomic_t refcnt; /* reference count */
515 struct timer_list timer; /* Expiration timer */
516 volatile unsigned long timeout; /* timeout */
517
518 /* Flags and state transition */
519 spinlock_t lock; /* lock for state transition */
1da177e4 520 volatile __u16 state; /* state info */
efac5276
RB
521 volatile __u16 old_state; /* old state, to be used for
522 * state transition triggerd
523 * synchronization
524 */
1da177e4
LT
525
526 /* Control members */
527 struct ip_vs_conn *control; /* Master control connection */
528 atomic_t n_control; /* Number of controlled ones */
529 struct ip_vs_dest *dest; /* real server */
530 atomic_t in_pkts; /* incoming packet counter */
531
532 /* packet transmitter for different forwarding methods. If it
533 mangles the packet, it must return NF_DROP or better NF_STOLEN,
534 otherwise this must be changed to a sk_buff **.
fc604767 535 NF_ACCEPT can be returned when destination is local.
1da177e4
LT
536 */
537 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
538 struct ip_vs_protocol *pp);
539
540 /* Note: we can group the following members into a structure,
541 in order to save more space, and the following members are
542 only used in VS/NAT anyway */
543 struct ip_vs_app *app; /* bound ip_vs_app object */
544 void *app_data; /* Application private data */
545 struct ip_vs_seq in_seq; /* incoming seq. struct */
546 struct ip_vs_seq out_seq; /* outgoing seq. struct */
85999283 547
e9e5eee8 548 const struct ip_vs_pe *pe;
85999283
SH
549 char *pe_data;
550 __u8 pe_data_len;
1da177e4
LT
551};
552
6e67e586
HS
553/*
554 * To save some memory in conn table when name space is disabled.
555 */
556static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
557{
558#ifdef CONFIG_NET_NS
559 return cp->net;
560#else
561 return &init_net;
562#endif
563}
564static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
565{
566#ifdef CONFIG_NET_NS
567 cp->net = net;
568#endif
569}
570
571static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
572 struct net *net)
573{
574#ifdef CONFIG_NET_NS
575 return cp->net == net;
576#else
577 return 1;
578#endif
579}
1da177e4 580
c860c6b1
JV
581/*
582 * Extended internal versions of struct ip_vs_service_user and
583 * ip_vs_dest_user for IPv6 support.
584 *
585 * We need these to conveniently pass around service and destination
586 * options, but unfortunately, we also need to keep the old definitions to
587 * maintain userspace backwards compatibility for the setsockopt interface.
588 */
589struct ip_vs_service_user_kern {
590 /* virtual service addresses */
591 u16 af;
592 u16 protocol;
593 union nf_inet_addr addr; /* virtual ip address */
594 u16 port;
595 u32 fwmark; /* firwall mark of service */
596
597 /* virtual service options */
598 char *sched_name;
0d1e71b0 599 char *pe_name;
c860c6b1
JV
600 unsigned flags; /* virtual service flags */
601 unsigned timeout; /* persistent timeout in sec */
602 u32 netmask; /* persistent netmask */
603};
604
605
606struct ip_vs_dest_user_kern {
607 /* destination server address */
608 union nf_inet_addr addr;
609 u16 port;
610
611 /* real server options */
612 unsigned conn_flags; /* connection flags */
613 int weight; /* destination weight */
614
615 /* thresholds for active connections */
616 u32 u_threshold; /* upper threshold */
617 u32 l_threshold; /* lower threshold */
618};
619
620
1da177e4
LT
621/*
622 * The information about the virtual service offered to the net
623 * and the forwarding entries
624 */
625struct ip_vs_service {
626 struct list_head s_list; /* for normal service table */
627 struct list_head f_list; /* for fwmark-based service table */
628 atomic_t refcnt; /* reference counter */
629 atomic_t usecnt; /* use counter */
630
e7ade46a 631 u16 af; /* address family */
1da177e4 632 __u16 protocol; /* which protocol (TCP/UDP) */
e7ade46a 633 union nf_inet_addr addr; /* IP address for virtual service */
014d730d 634 __be16 port; /* port number for the service */
1da177e4
LT
635 __u32 fwmark; /* firewall mark of the service */
636 unsigned flags; /* service status flags */
637 unsigned timeout; /* persistent timeout in ticks */
014d730d 638 __be32 netmask; /* grouping granularity */
fc723250 639 struct net *net;
1da177e4
LT
640
641 struct list_head destinations; /* real server d-linked list */
642 __u32 num_dests; /* number of servers */
643 struct ip_vs_stats stats; /* statistics for the service */
644 struct ip_vs_app *inc; /* bind conns to this app inc */
645
646 /* for scheduling */
647 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
648 rwlock_t sched_lock; /* lock sched_data */
649 void *sched_data; /* scheduler application data */
85999283
SH
650
651 /* alternate persistence engine */
652 struct ip_vs_pe *pe;
1da177e4
LT
653};
654
655
656/*
657 * The real server destination forwarding entry
658 * with ip address, port number, and so on.
659 */
660struct ip_vs_dest {
661 struct list_head n_list; /* for the dests in the service */
662 struct list_head d_list; /* for table with all the dests */
663
e7ade46a 664 u16 af; /* address family */
014d730d 665 __be16 port; /* port number of the server */
f2431e6e 666 union nf_inet_addr addr; /* IP address of the server */
1da177e4
LT
667 volatile unsigned flags; /* dest status flags */
668 atomic_t conn_flags; /* flags to copy to conn */
669 atomic_t weight; /* server weight */
670
671 atomic_t refcnt; /* reference counter */
672 struct ip_vs_stats stats; /* statistics */
673
674 /* connection counters and thresholds */
675 atomic_t activeconns; /* active connections */
676 atomic_t inactconns; /* inactive connections */
677 atomic_t persistconns; /* persistent connections */
678 __u32 u_threshold; /* upper threshold */
679 __u32 l_threshold; /* lower threshold */
680
681 /* for destination cache */
682 spinlock_t dst_lock; /* lock of dst_cache */
683 struct dst_entry *dst_cache; /* destination cache entry */
684 u32 dst_rtos; /* RT_TOS(tos) for dst */
714f095f
HS
685 u32 dst_cookie;
686#ifdef CONFIG_IP_VS_IPV6
687 struct in6_addr dst_saddr;
688#endif
1da177e4
LT
689
690 /* for virtual service */
691 struct ip_vs_service *svc; /* service it belongs to */
692 __u16 protocol; /* which protocol (TCP/UDP) */
014d730d 693 __be16 vport; /* virtual port number */
f2431e6e 694 union nf_inet_addr vaddr; /* virtual IP address */
1da177e4
LT
695 __u32 vfwmark; /* firewall mark of service */
696};
697
698
699/*
700 * The scheduler object
701 */
702struct ip_vs_scheduler {
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 /* scheduler initializing service */
709 int (*init_service)(struct ip_vs_service *svc);
710 /* scheduling service finish */
711 int (*done_service)(struct ip_vs_service *svc);
712 /* scheduler updating service */
713 int (*update_service)(struct ip_vs_service *svc);
714
715 /* selecting a server from the given service */
716 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
717 const struct sk_buff *skb);
718};
719
85999283
SH
720/* The persistence engine object */
721struct ip_vs_pe {
722 struct list_head n_list; /* d-linked list head */
723 char *name; /* scheduler name */
724 atomic_t refcnt; /* reference counter */
725 struct module *module; /* THIS_MODULE/NULL */
726
727 /* get the connection template, if any */
728 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
729 bool (*ct_match)(const struct ip_vs_conn_param *p,
730 struct ip_vs_conn *ct);
731 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
732 bool inverse);
a3c918ac 733 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
85999283 734};
1da177e4
LT
735
736/*
737 * The application module object (a.k.a. app incarnation)
738 */
fd2c3ef7 739struct ip_vs_app {
1da177e4
LT
740 struct list_head a_list; /* member in app list */
741 int type; /* IP_VS_APP_TYPE_xxx */
742 char *name; /* application module name */
743 __u16 protocol;
744 struct module *module; /* THIS_MODULE/NULL */
745 struct list_head incs_list; /* list of incarnations */
746
747 /* members for application incarnations */
748 struct list_head p_list; /* member in proto app list */
749 struct ip_vs_app *app; /* its real application */
014d730d 750 __be16 port; /* port number in net order */
1da177e4
LT
751 atomic_t usecnt; /* usage counter */
752
8b27b10f
JA
753 /*
754 * output hook: Process packet in inout direction, diff set for TCP.
755 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
756 * 2=Mangled but checksum was not updated
757 */
1da177e4 758 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
3db05fea 759 struct sk_buff *, int *diff);
1da177e4 760
8b27b10f
JA
761 /*
762 * input hook: Process packet in outin direction, diff set for TCP.
763 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
764 * 2=Mangled but checksum was not updated
765 */
1da177e4 766 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
3db05fea 767 struct sk_buff *, int *diff);
1da177e4
LT
768
769 /* ip_vs_app initializer */
770 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
771
772 /* ip_vs_app finish */
773 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
774
775
776 /* not used now */
777 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
778 struct ip_vs_protocol *);
779
780 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
781
782 int * timeout_table;
783 int * timeouts;
784 int timeouts_size;
785
786 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
787 int *verdict, struct ip_vs_conn **cpp);
788
789 struct ip_vs_conn *
790 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
791 const struct iphdr *iph, unsigned int proto_off,
792 int inverse);
793
794 struct ip_vs_conn *
795 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
796 const struct iphdr *iph, unsigned int proto_off,
797 int inverse);
798
799 int (*state_transition)(struct ip_vs_conn *cp, int direction,
800 const struct sk_buff *skb,
801 struct ip_vs_app *app);
802
803 void (*timeout_change)(struct ip_vs_app *app, int flags);
804};
805
2553d064
JA
806/* IPVS in network namespace */
807struct netns_ipvs {
808 int gen; /* Generation */
809 /*
810 * Hash table: for real service lookups
811 */
812 #define IP_VS_RTAB_BITS 4
813 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
814 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
815
816 struct list_head rs_table[IP_VS_RTAB_SIZE];
817 /* ip_vs_app */
818 struct list_head app_list;
819 struct mutex app_mutex;
820 struct lock_class_key app_key; /* mutex debuging */
821
822 /* ip_vs_proto */
823 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
824 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
825 /* ip_vs_proto_tcp */
826#ifdef CONFIG_IP_VS_PROTO_TCP
827 #define TCP_APP_TAB_BITS 4
828 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
829 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
830 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
831 spinlock_t tcp_app_lock;
832#endif
833 /* ip_vs_proto_udp */
834#ifdef CONFIG_IP_VS_PROTO_UDP
835 #define UDP_APP_TAB_BITS 4
836 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
837 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
838 struct list_head udp_apps[UDP_APP_TAB_SIZE];
839 spinlock_t udp_app_lock;
840#endif
841 /* ip_vs_proto_sctp */
842#ifdef CONFIG_IP_VS_PROTO_SCTP
843 #define SCTP_APP_TAB_BITS 4
844 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
845 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
846 /* Hash table for SCTP application incarnations */
847 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
848 spinlock_t sctp_app_lock;
849#endif
850 /* ip_vs_conn */
851 atomic_t conn_count; /* connection counter */
852
853 /* ip_vs_ctl */
854 struct ip_vs_stats *tot_stats; /* Statistics & est. */
855 struct ip_vs_cpu_stats __percpu *cpustats; /* Stats per cpu */
856 seqcount_t *ustats_seq; /* u64 read retry */
857
858 int num_services; /* no of virtual services */
859 /* 1/rate drop and drop-entry variables */
860 struct delayed_work defense_work; /* Work handler */
861 int drop_rate;
862 int drop_counter;
863 atomic_t dropentry;
864 /* locks in ctl.c */
865 spinlock_t dropentry_lock; /* drop entry handling */
866 spinlock_t droppacket_lock; /* drop packet handling */
867 spinlock_t securetcp_lock; /* state and timeout tables */
868 rwlock_t rs_lock; /* real services table */
869 /* semaphore for IPVS sockopts. And, [gs]etsockopt may sleep. */
870 struct lock_class_key ctl_key; /* ctl_mutex debuging */
871 /* Trash for destinations */
872 struct list_head dest_trash;
873 /* Service counters */
874 atomic_t ftpsvc_counter;
875 atomic_t nullsvc_counter;
876
877 /* sys-ctl struct */
878 struct ctl_table_header *sysctl_hdr;
879 struct ctl_table *sysctl_tbl;
880 /* sysctl variables */
881 int sysctl_amemthresh;
882 int sysctl_am_droprate;
883 int sysctl_drop_entry;
884 int sysctl_drop_packet;
885 int sysctl_secure_tcp;
886#ifdef CONFIG_IP_VS_NFCT
887 int sysctl_conntrack;
888#endif
889 int sysctl_snat_reroute;
890 int sysctl_sync_ver;
891 int sysctl_cache_bypass;
892 int sysctl_expire_nodest_conn;
893 int sysctl_expire_quiescent_template;
894 int sysctl_sync_threshold[2];
895 int sysctl_nat_icmp_send;
896
897 /* ip_vs_lblc */
898 int sysctl_lblc_expiration;
899 struct ctl_table_header *lblc_ctl_header;
900 struct ctl_table *lblc_ctl_table;
901 /* ip_vs_lblcr */
902 int sysctl_lblcr_expiration;
903 struct ctl_table_header *lblcr_ctl_header;
904 struct ctl_table *lblcr_ctl_table;
905 /* ip_vs_est */
906 struct list_head est_list; /* estimator list */
907 spinlock_t est_lock;
908 struct timer_list est_timer; /* Estimation timer */
909 /* ip_vs_sync */
910 struct list_head sync_queue;
911 spinlock_t sync_lock;
912 struct ip_vs_sync_buff *sync_buff;
913 spinlock_t sync_buff_lock;
914 struct sockaddr_in sync_mcast_addr;
915 struct task_struct *master_thread;
916 struct task_struct *backup_thread;
917 int send_mesg_maxlen;
918 int recv_mesg_maxlen;
919 volatile int sync_state;
920 volatile int master_syncid;
921 volatile int backup_syncid;
922 /* multicast interface name */
923 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
924 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
925 /* net name space ptr */
926 struct net *net; /* Needed by timer routines */
927};
1da177e4
LT
928
929/*
930 * IPVS core functions
931 * (from ip_vs_core.c)
932 */
933extern const char *ip_vs_proto_name(unsigned proto);
934extern void ip_vs_init_hash_table(struct list_head *table, int rows);
afdd6140 935#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1da177e4 936
1da177e4
LT
937#define IP_VS_APP_TYPE_FTP 1
938
939/*
940 * ip_vs_conn handling functions
941 * (from ip_vs_conn.c)
942 */
943
1da177e4
LT
944enum {
945 IP_VS_DIR_INPUT = 0,
946 IP_VS_DIR_OUTPUT,
947 IP_VS_DIR_INPUT_ONLY,
948 IP_VS_DIR_LAST,
949};
950
6e67e586 951static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
f11017ec
SH
952 const union nf_inet_addr *caddr,
953 __be16 cport,
954 const union nf_inet_addr *vaddr,
955 __be16 vport,
956 struct ip_vs_conn_param *p)
957{
6e67e586 958 p->net = net;
f11017ec
SH
959 p->af = af;
960 p->protocol = protocol;
961 p->caddr = caddr;
962 p->cport = cport;
963 p->vaddr = vaddr;
964 p->vport = vport;
85999283
SH
965 p->pe = NULL;
966 p->pe_data = NULL;
f11017ec 967}
28364a59 968
f11017ec
SH
969struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
970struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
28364a59 971
5c0d2374 972struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
5c0d2374
SH
973 const struct ip_vs_iphdr *iph,
974 unsigned int proto_off,
975 int inverse);
976
f11017ec 977struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1da177e4 978
5c0d2374 979struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
5c0d2374
SH
980 const struct ip_vs_iphdr *iph,
981 unsigned int proto_off,
982 int inverse);
983
1da177e4
LT
984/* put back the conn without restarting its timer */
985static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
986{
987 atomic_dec(&cp->refcnt);
988}
989extern void ip_vs_conn_put(struct ip_vs_conn *cp);
014d730d 990extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1da177e4 991
f11017ec
SH
992struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
993 const union nf_inet_addr *daddr,
994 __be16 dport, unsigned flags,
0e051e68 995 struct ip_vs_dest *dest, __u32 fwmark);
1da177e4
LT
996extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
997
998extern const char * ip_vs_state_name(__u16 proto, int state);
999
4a85b96c 1000extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1da177e4 1001extern int ip_vs_check_template(struct ip_vs_conn *ct);
f6340ee0 1002extern void ip_vs_random_dropentry(struct net *net);
1da177e4
LT
1003extern int ip_vs_conn_init(void);
1004extern void ip_vs_conn_cleanup(void);
1005
1006static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1007{
1008 struct ip_vs_conn *ctl_cp = cp->control;
1009 if (!ctl_cp) {
cfc78c5a
JV
1010 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1011 "%s:%d to %s:%d\n",
1012 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1013 ntohs(cp->cport),
1014 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1015 ntohs(cp->vport));
1016
1da177e4
LT
1017 return;
1018 }
1019
cfc78c5a
JV
1020 IP_VS_DBG_BUF(7, "DELeting control for: "
1021 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1022 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1023 ntohs(cp->cport),
1024 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1025 ntohs(ctl_cp->cport));
1da177e4
LT
1026
1027 cp->control = NULL;
1028 if (atomic_read(&ctl_cp->n_control) == 0) {
cfc78c5a
JV
1029 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1030 "%s:%d to %s:%d\n",
1031 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1032 ntohs(cp->cport),
1033 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1034 ntohs(cp->vport));
1035
1da177e4
LT
1036 return;
1037 }
1038 atomic_dec(&ctl_cp->n_control);
1039}
1040
1041static inline void
1042ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1043{
1044 if (cp->control) {
cfc78c5a
JV
1045 IP_VS_ERR_BUF("request control ADD for already controlled: "
1046 "%s:%d to %s:%d\n",
1047 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1048 ntohs(cp->cport),
1049 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1050 ntohs(cp->vport));
1051
1da177e4
LT
1052 ip_vs_control_del(cp);
1053 }
1054
cfc78c5a
JV
1055 IP_VS_DBG_BUF(7, "ADDing control for: "
1056 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1057 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1058 ntohs(cp->cport),
1059 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1060 ntohs(ctl_cp->cport));
1da177e4
LT
1061
1062 cp->control = ctl_cp;
1063 atomic_inc(&ctl_cp->n_control);
1064}
1065
1066
1067/*
1068 * IPVS application functions
1069 * (from ip_vs_app.c)
1070 */
1071#define IP_VS_APP_MAX_PORTS 8
ab8a5e84
HS
1072extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1073extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1da177e4
LT
1074extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1075extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
ab8a5e84
HS
1076extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1077 __u16 proto, __u16 port);
1da177e4
LT
1078extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1079extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1080
3db05fea
HX
1081extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1082extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1da177e4
LT
1083extern int ip_vs_app_init(void);
1084extern void ip_vs_app_cleanup(void);
1085
8be67a66
SH
1086void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1087void ip_vs_unbind_pe(struct ip_vs_service *svc);
1088int register_ip_vs_pe(struct ip_vs_pe *pe);
1089int unregister_ip_vs_pe(struct ip_vs_pe *pe);
e9e5eee8 1090struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
fe5e7a1e 1091struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
e9e5eee8
SH
1092
1093static inline void ip_vs_pe_get(const struct ip_vs_pe *pe)
1094{
1095 if (pe && pe->module)
1096 __module_get(pe->module);
1097}
1098
1099static inline void ip_vs_pe_put(const struct ip_vs_pe *pe)
1100{
1101 if (pe && pe->module)
1102 module_put(pe->module);
1103}
1da177e4
LT
1104
1105/*
1106 * IPVS protocol functions (from ip_vs_proto.c)
1107 */
1108extern int ip_vs_protocol_init(void);
1109extern void ip_vs_protocol_cleanup(void);
9330419d 1110extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1da177e4
LT
1111extern int *ip_vs_create_timeout_table(int *table, int size);
1112extern int
36cbd3dc
JE
1113ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1114 const char *name, int to);
1da177e4 1115extern void
0d79641a
JA
1116ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1117 const struct sk_buff *skb,
1da177e4
LT
1118 int offset, const char *msg);
1119
1120extern struct ip_vs_protocol ip_vs_protocol_tcp;
1121extern struct ip_vs_protocol ip_vs_protocol_udp;
1122extern struct ip_vs_protocol ip_vs_protocol_icmp;
1123extern struct ip_vs_protocol ip_vs_protocol_esp;
1124extern struct ip_vs_protocol ip_vs_protocol_ah;
2906f66a 1125extern struct ip_vs_protocol ip_vs_protocol_sctp;
1da177e4
LT
1126
1127/*
1128 * Registering/unregistering scheduler functions
1129 * (from ip_vs_sched.c)
1130 */
1131extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1132extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1133extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1134 struct ip_vs_scheduler *scheduler);
1135extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1136extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1137extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1138extern struct ip_vs_conn *
190ecd27 1139ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
9330419d 1140 struct ip_vs_proto_data *pd, int *ignored);
1da177e4 1141extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
9330419d 1142 struct ip_vs_proto_data *pd);
1da177e4 1143
41ac51ee
PS
1144extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1145
1da177e4
LT
1146
1147/*
1148 * IPVS control data and functions (from ip_vs_ctl.c)
1149 */
1da177e4 1150extern struct ip_vs_stats ip_vs_stats;
5587da55 1151extern const struct ctl_path net_vs_ctl_path[];
b880c1f0 1152extern int sysctl_ip_vs_sync_ver;
1da177e4 1153
f131315f 1154extern void ip_vs_sync_switch_mode(struct net *net, int mode);
1da177e4 1155extern struct ip_vs_service *
fc723250 1156ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
3c2e0505 1157 const union nf_inet_addr *vaddr, __be16 vport);
1da177e4
LT
1158
1159static inline void ip_vs_service_put(struct ip_vs_service *svc)
1160{
1161 atomic_dec(&svc->usecnt);
1162}
1163
1164extern struct ip_vs_dest *
fc723250 1165ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
7937df15
JV
1166 const union nf_inet_addr *daddr, __be16 dport);
1167
1da177e4
LT
1168extern int ip_vs_use_count_inc(void);
1169extern void ip_vs_use_count_dec(void);
1170extern int ip_vs_control_init(void);
1171extern void ip_vs_control_cleanup(void);
1e356f9c 1172extern struct ip_vs_dest *
fc723250
HS
1173ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1174 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1175 __u16 protocol, __u32 fwmark);
1e356f9c 1176extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1da177e4
LT
1177
1178
1179/*
1180 * IPVS sync daemon data and function prototypes
1181 * (from ip_vs_sync.c)
1182 */
f131315f
HS
1183extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1184 __u8 syncid);
1185extern int stop_sync_thread(struct net *net, int state);
1186extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
61b1ab45
HS
1187extern int ip_vs_sync_init(void);
1188extern void ip_vs_sync_cleanup(void);
1da177e4
LT
1189
1190
1191/*
1192 * IPVS rate estimator prototypes (from ip_vs_est.c)
1193 */
a919cf4b
SW
1194extern int ip_vs_estimator_init(void);
1195extern void ip_vs_estimator_cleanup(void);
29c2026f
HS
1196extern void ip_vs_new_estimator(struct net *net, struct ip_vs_stats *stats);
1197extern void ip_vs_kill_estimator(struct net *net, struct ip_vs_stats *stats);
1da177e4
LT
1198extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1199
1200/*
1201 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1202 */
1203extern int ip_vs_null_xmit
1204(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1205extern int ip_vs_bypass_xmit
1206(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1207extern int ip_vs_nat_xmit
1208(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1209extern int ip_vs_tunnel_xmit
1210(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1211extern int ip_vs_dr_xmit
1212(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1213extern int ip_vs_icmp_xmit
1214(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
1215extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1216
b3cdd2a7
JV
1217#ifdef CONFIG_IP_VS_IPV6
1218extern int ip_vs_bypass_xmit_v6
1219(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1220extern int ip_vs_nat_xmit_v6
1221(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1222extern int ip_vs_tunnel_xmit_v6
1223(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1224extern int ip_vs_dr_xmit_v6
1225(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1226extern int ip_vs_icmp_xmit_v6
1227(struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1228 int offset);
1229#endif
1da177e4
LT
1230
1231/*
1232 * This is a simple mechanism to ignore packets when
1233 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1234 * we start to drop 1/rate of the packets
1235 */
1da177e4 1236
a0840e2e 1237static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1da177e4 1238{
a0840e2e
HS
1239 if (!ipvs->drop_rate)
1240 return 0;
1241 if (--ipvs->drop_counter > 0)
1242 return 0;
1243 ipvs->drop_counter = ipvs->drop_rate;
1da177e4
LT
1244 return 1;
1245}
1246
1247/*
1248 * ip_vs_fwd_tag returns the forwarding tag of the connection
1249 */
1250#define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1251
732db659 1252static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1da177e4
LT
1253{
1254 char fwd;
1255
1256 switch (IP_VS_FWD_METHOD(cp)) {
1257 case IP_VS_CONN_F_MASQ:
1258 fwd = 'M'; break;
1259 case IP_VS_CONN_F_LOCALNODE:
1260 fwd = 'L'; break;
1261 case IP_VS_CONN_F_TUNNEL:
1262 fwd = 'T'; break;
1263 case IP_VS_CONN_F_DROUTE:
1264 fwd = 'R'; break;
1265 case IP_VS_CONN_F_BYPASS:
1266 fwd = 'B'; break;
1267 default:
1268 fwd = '?'; break;
1269 }
1270 return fwd;
1271}
1272
1da177e4 1273extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
b3cdd2a7
JV
1274 struct ip_vs_conn *cp, int dir);
1275
1276#ifdef CONFIG_IP_VS_IPV6
1277extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1278 struct ip_vs_conn *cp, int dir);
1279#endif
1da177e4 1280
b1550f22 1281extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1da177e4 1282
f9214b26 1283static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1da177e4 1284{
f9214b26 1285 __be32 diff[2] = { ~old, new };
1da177e4 1286
07f0757a 1287 return csum_partial(diff, sizeof(diff), oldsum);
f9214b26
AV
1288}
1289
0bbdd42b
JV
1290#ifdef CONFIG_IP_VS_IPV6
1291static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1292 __wsum oldsum)
1293{
1294 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1295 new[3], new[2], new[1], new[0] };
1296
07f0757a 1297 return csum_partial(diff, sizeof(diff), oldsum);
0bbdd42b
JV
1298}
1299#endif
1300
f9214b26
AV
1301static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1302{
1303 __be16 diff[2] = { ~old, new };
1304
07f0757a 1305 return csum_partial(diff, sizeof(diff), oldsum);
1da177e4
LT
1306}
1307
cf356d69
JA
1308/*
1309 * Forget current conntrack (unconfirmed) and attach notrack entry
1310 */
1311static inline void ip_vs_notrack(struct sk_buff *skb)
1312{
1313#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1314 enum ip_conntrack_info ctinfo;
06b69390 1315 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
cf356d69
JA
1316
1317 if (!ct || !nf_ct_is_untracked(ct)) {
1318 nf_reset(skb);
1319 skb->nfct = &nf_ct_untracked_get()->ct_general;
1320 skb->nfctinfo = IP_CT_NEW;
1321 nf_conntrack_get(skb->nfct);
1322 }
1323#endif
1324}
1325
f4bc17cd
JA
1326#ifdef CONFIG_IP_VS_NFCT
1327/*
1328 * Netfilter connection tracking
1329 * (from ip_vs_nfct.c)
1330 */
a0840e2e 1331static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
f4bc17cd 1332{
a0840e2e 1333 return ipvs->sysctl_conntrack;
f4bc17cd
JA
1334}
1335
6523ce15
JA
1336extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1337 int outin);
f4bc17cd
JA
1338extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1339extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1340 struct ip_vs_conn *cp, u_int8_t proto,
1341 const __be16 port, int from_rs);
1342extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1343
1344#else
1345
a0840e2e 1346static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
f4bc17cd
JA
1347{
1348 return 0;
1349}
1350
1351static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1352 struct ip_vs_conn *cp, int outin)
1353{
1354}
1355
1356static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1357 struct ip_vs_conn *cp)
1358{
1359 return NF_ACCEPT;
1360}
1361
1362static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1363{
1364}
1365/* CONFIG_IP_VS_NFCT */
1366#endif
6523ce15 1367
b552f7e3
CG
1368static inline unsigned int
1369ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1370{
1371 /*
1372 * We think the overhead of processing active connections is 256
1373 * times higher than that of inactive connections in average. (This
1374 * 256 times might not be accurate, we will change it later) We
1375 * use the following formula to estimate the overhead now:
1376 * dest->activeconns*256 + dest->inactconns
1377 */
1378 return (atomic_read(&dest->activeconns) << 8) +
1379 atomic_read(&dest->inactconns);
1380}
1381
1da177e4
LT
1382#endif /* __KERNEL__ */
1383
bc4768eb 1384#endif /* _NET_IP_VS_H */