?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
ipvs: provide iph to schedulers
[linux/fpc-iii.git] / include / net / ip_vs.h
blobf5faf859876ebb8eaa19c27ca110e2c77ca9b1a7
1 /*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
6 #ifndef _NET_IP_VS_H
7 #define _NET_IP_VS_H
8
9 #include <linux/ip_vs.h> /* definitions shared with userland */
10
11 #include <asm/types.h> /* for __uXX types */
12
13 #include <linux/list.h> /* for struct list_head */
14 #include <linux/spinlock.h> /* for struct rwlock_t */
15 #include <linux/atomic.h> /* for struct atomic_t */
16 #include <linux/compiler.h>
17 #include <linux/timer.h>
18 #include <linux/bug.h>
19
20 #include <net/checksum.h>
21 #include <linux/netfilter.h> /* for union nf_inet_addr */
22 #include <linux/ip.h>
23 #include <linux/ipv6.h> /* for struct ipv6hdr */
24 #include <net/ipv6.h>
25 #if IS_ENABLED(CONFIG_IP_VS_IPV6)
26 #include <linux/netfilter_ipv6/ip6_tables.h>
27 #endif
28 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
29 #include <net/netfilter/nf_conntrack.h>
30 #endif
31 #include <net/net_namespace.h> /* Netw namespace */
32
33 /*
34 * Generic access of ipvs struct
35 */
36 static inline struct netns_ipvs *net_ipvs(struct net* net)
37 {
38 return net->ipvs;
39 }
40 /*
41 * Get net ptr from skb in traffic cases
42 * use skb_sknet when call is from userland (ioctl or netlink)
43 */
44 static inline struct net *skb_net(const struct sk_buff *skb)
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) && 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
72 static inline struct net *skb_sknet(const struct sk_buff *skb)
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 }
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 can't be used.
96 */
97 static 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 }
105
106 /* Connections' size value needed by ip_vs_ctl.c */
107 extern int ip_vs_conn_tab_size;
108
109 struct ip_vs_iphdr {
110 __u32 len; /* IPv4 simply where L4 starts
111 IPv6 where L4 Transport Header starts */
112 __u32 thoff_reasm; /* Transport Header Offset in nfct_reasm skb */
113 __u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
114 __s16 protocol;
115 __s32 flags;
116 union nf_inet_addr saddr;
117 union nf_inet_addr daddr;
118 };
119
120 /* Dependency to module: nf_defrag_ipv6 */
121 #if defined(CONFIG_NF_DEFRAG_IPV6) || defined(CONFIG_NF_DEFRAG_IPV6_MODULE)
122 static inline struct sk_buff *skb_nfct_reasm(const struct sk_buff *skb)
123 {
124 return skb->nfct_reasm;
125 }
126 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
127 int len, void *buffer,
128 const struct ip_vs_iphdr *ipvsh)
129 {
130 if (unlikely(ipvsh->fragoffs && skb_nfct_reasm(skb)))
131 return skb_header_pointer(skb_nfct_reasm(skb),
132 ipvsh->thoff_reasm, len, buffer);
133
134 return skb_header_pointer(skb, offset, len, buffer);
135 }
136 #else
137 static inline struct sk_buff *skb_nfct_reasm(const struct sk_buff *skb)
138 {
139 return NULL;
140 }
141 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
142 int len, void *buffer,
143 const struct ip_vs_iphdr *ipvsh)
144 {
145 return skb_header_pointer(skb, offset, len, buffer);
146 }
147 #endif
148
149 static inline void
150 ip_vs_fill_ip4hdr(const void *nh, struct ip_vs_iphdr *iphdr)
151 {
152 const struct iphdr *iph = nh;
153
154 iphdr->len = iph->ihl * 4;
155 iphdr->fragoffs = 0;
156 iphdr->protocol = iph->protocol;
157 iphdr->saddr.ip = iph->saddr;
158 iphdr->daddr.ip = iph->daddr;
159 }
160
161 /* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
162 * IPv6 requires some extra work, as finding proper header position,
163 * depend on the IPv6 extension headers.
164 */
165 static inline void
166 ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, struct ip_vs_iphdr *iphdr)
167 {
168 #ifdef CONFIG_IP_VS_IPV6
169 if (af == AF_INET6) {
170 const struct ipv6hdr *iph =
171 (struct ipv6hdr *)skb_network_header(skb);
172 iphdr->saddr.in6 = iph->saddr;
173 iphdr->daddr.in6 = iph->daddr;
174 /* ipv6_find_hdr() updates len, flags, thoff_reasm */
175 iphdr->thoff_reasm = 0;
176 iphdr->len = 0;
177 iphdr->flags = 0;
178 iphdr->protocol = ipv6_find_hdr(skb, &iphdr->len, -1,
179 &iphdr->fragoffs,
180 &iphdr->flags);
181 /* get proto from re-assembled packet and it's offset */
182 if (skb_nfct_reasm(skb))
183 iphdr->protocol = ipv6_find_hdr(skb_nfct_reasm(skb),
184 &iphdr->thoff_reasm,
185 -1, NULL, NULL);
186
187 } else
188 #endif
189 {
190 const struct iphdr *iph =
191 (struct iphdr *)skb_network_header(skb);
192 iphdr->len = iph->ihl * 4;
193 iphdr->fragoffs = 0;
194 iphdr->protocol = iph->protocol;
195 iphdr->saddr.ip = iph->saddr;
196 iphdr->daddr.ip = iph->daddr;
197 }
198 }
199
200 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
201 const union nf_inet_addr *src)
202 {
203 #ifdef CONFIG_IP_VS_IPV6
204 if (af == AF_INET6)
205 dst->in6 = src->in6;
206 else
207 #endif
208 dst->ip = src->ip;
209 }
210
211 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
212 const union nf_inet_addr *src)
213 {
214 #ifdef CONFIG_IP_VS_IPV6
215 if (af == AF_INET6) {
216 dst->in6 = src->in6;
217 return;
218 }
219 #endif
220 dst->ip = src->ip;
221 dst->all[1] = 0;
222 dst->all[2] = 0;
223 dst->all[3] = 0;
224 }
225
226 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
227 const union nf_inet_addr *b)
228 {
229 #ifdef CONFIG_IP_VS_IPV6
230 if (af == AF_INET6)
231 return ipv6_addr_equal(&a->in6, &b->in6);
232 #endif
233 return a->ip == b->ip;
234 }
235
236 #ifdef CONFIG_IP_VS_DEBUG
237 #include <linux/net.h>
238
239 extern int ip_vs_get_debug_level(void);
240
241 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
242 const union nf_inet_addr *addr,
243 int *idx)
244 {
245 int len;
246 #ifdef CONFIG_IP_VS_IPV6
247 if (af == AF_INET6)
248 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
249 &addr->in6) + 1;
250 else
251 #endif
252 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
253 &addr->ip) + 1;
254
255 *idx += len;
256 BUG_ON(*idx > buf_len + 1);
257 return &buf[*idx - len];
258 }
259
260 #define IP_VS_DBG_BUF(level, msg, ...) \
261 do { \
262 char ip_vs_dbg_buf[160]; \
263 int ip_vs_dbg_idx = 0; \
264 if (level <= ip_vs_get_debug_level()) \
265 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
266 } while (0)
267 #define IP_VS_ERR_BUF(msg...) \
268 do { \
269 char ip_vs_dbg_buf[160]; \
270 int ip_vs_dbg_idx = 0; \
271 pr_err(msg); \
272 } while (0)
273
274 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
275 #define IP_VS_DBG_ADDR(af, addr) \
276 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
277 sizeof(ip_vs_dbg_buf), addr, \
278 &ip_vs_dbg_idx)
279
280 #define IP_VS_DBG(level, msg, ...) \
281 do { \
282 if (level <= ip_vs_get_debug_level()) \
283 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
284 } while (0)
285 #define IP_VS_DBG_RL(msg, ...) \
286 do { \
287 if (net_ratelimit()) \
288 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
289 } while (0)
290 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
291 do { \
292 if (level <= ip_vs_get_debug_level()) \
293 pp->debug_packet(af, pp, skb, ofs, msg); \
294 } while (0)
295 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
296 do { \
297 if (level <= ip_vs_get_debug_level() && \
298 net_ratelimit()) \
299 pp->debug_packet(af, pp, skb, ofs, msg); \
300 } while (0)
301 #else /* NO DEBUGGING at ALL */
302 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
303 #define IP_VS_ERR_BUF(msg...) do {} while (0)
304 #define IP_VS_DBG(level, msg...) do {} while (0)
305 #define IP_VS_DBG_RL(msg...) do {} while (0)
306 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
307 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
308 #endif
309
310 #define IP_VS_BUG() BUG()
311 #define IP_VS_ERR_RL(msg, ...) \
312 do { \
313 if (net_ratelimit()) \
314 pr_err(msg, ##__VA_ARGS__); \
315 } while (0)
316
317 #ifdef CONFIG_IP_VS_DEBUG
318 #define EnterFunction(level) \
319 do { \
320 if (level <= ip_vs_get_debug_level()) \
321 printk(KERN_DEBUG \
322 pr_fmt("Enter: %s, %s line %i\n"), \
323 __func__, __FILE__, __LINE__); \
324 } while (0)
325 #define LeaveFunction(level) \
326 do { \
327 if (level <= ip_vs_get_debug_level()) \
328 printk(KERN_DEBUG \
329 pr_fmt("Leave: %s, %s line %i\n"), \
330 __func__, __FILE__, __LINE__); \
331 } while (0)
332 #else
333 #define EnterFunction(level) do {} while (0)
334 #define LeaveFunction(level) do {} while (0)
335 #endif
336
337
338 /*
339 * The port number of FTP service (in network order).
340 */
341 #define FTPPORT cpu_to_be16(21)
342 #define FTPDATA cpu_to_be16(20)
343
344 /*
345 * TCP State Values
346 */
347 enum {
348 IP_VS_TCP_S_NONE = 0,
349 IP_VS_TCP_S_ESTABLISHED,
350 IP_VS_TCP_S_SYN_SENT,
351 IP_VS_TCP_S_SYN_RECV,
352 IP_VS_TCP_S_FIN_WAIT,
353 IP_VS_TCP_S_TIME_WAIT,
354 IP_VS_TCP_S_CLOSE,
355 IP_VS_TCP_S_CLOSE_WAIT,
356 IP_VS_TCP_S_LAST_ACK,
357 IP_VS_TCP_S_LISTEN,
358 IP_VS_TCP_S_SYNACK,
359 IP_VS_TCP_S_LAST
360 };
361
362 /*
363 * UDP State Values
364 */
365 enum {
366 IP_VS_UDP_S_NORMAL,
367 IP_VS_UDP_S_LAST,
368 };
369
370 /*
371 * ICMP State Values
372 */
373 enum {
374 IP_VS_ICMP_S_NORMAL,
375 IP_VS_ICMP_S_LAST,
376 };
377
378 /*
379 * SCTP State Values
380 */
381 enum ip_vs_sctp_states {
382 IP_VS_SCTP_S_NONE,
383 IP_VS_SCTP_S_INIT_CLI,
384 IP_VS_SCTP_S_INIT_SER,
385 IP_VS_SCTP_S_INIT_ACK_CLI,
386 IP_VS_SCTP_S_INIT_ACK_SER,
387 IP_VS_SCTP_S_ECHO_CLI,
388 IP_VS_SCTP_S_ECHO_SER,
389 IP_VS_SCTP_S_ESTABLISHED,
390 IP_VS_SCTP_S_SHUT_CLI,
391 IP_VS_SCTP_S_SHUT_SER,
392 IP_VS_SCTP_S_SHUT_ACK_CLI,
393 IP_VS_SCTP_S_SHUT_ACK_SER,
394 IP_VS_SCTP_S_CLOSED,
395 IP_VS_SCTP_S_LAST
396 };
397
398 /*
399 * Delta sequence info structure
400 * Each ip_vs_conn has 2 (output AND input seq. changes).
401 * Only used in the VS/NAT.
402 */
403 struct ip_vs_seq {
404 __u32 init_seq; /* Add delta from this seq */
405 __u32 delta; /* Delta in sequence numbers */
406 __u32 previous_delta; /* Delta in sequence numbers
407 before last resized pkt */
408 };
409
410 /*
411 * counters per cpu
412 */
413 struct ip_vs_counters {
414 __u32 conns; /* connections scheduled */
415 __u32 inpkts; /* incoming packets */
416 __u32 outpkts; /* outgoing packets */
417 __u64 inbytes; /* incoming bytes */
418 __u64 outbytes; /* outgoing bytes */
419 };
420 /*
421 * Stats per cpu
422 */
423 struct ip_vs_cpu_stats {
424 struct ip_vs_counters ustats;
425 struct u64_stats_sync syncp;
426 };
427
428 /*
429 * IPVS statistics objects
430 */
431 struct ip_vs_estimator {
432 struct list_head list;
433
434 u64 last_inbytes;
435 u64 last_outbytes;
436 u32 last_conns;
437 u32 last_inpkts;
438 u32 last_outpkts;
439
440 u32 cps;
441 u32 inpps;
442 u32 outpps;
443 u32 inbps;
444 u32 outbps;
445 };
446
447 struct ip_vs_stats {
448 struct ip_vs_stats_user ustats; /* statistics */
449 struct ip_vs_estimator est; /* estimator */
450 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
451 spinlock_t lock; /* spin lock */
452 struct ip_vs_stats_user ustats0; /* reset values */
453 };
454
455 struct dst_entry;
456 struct iphdr;
457 struct ip_vs_conn;
458 struct ip_vs_app;
459 struct sk_buff;
460 struct ip_vs_proto_data;
461
462 struct ip_vs_protocol {
463 struct ip_vs_protocol *next;
464 char *name;
465 u16 protocol;
466 u16 num_states;
467 int dont_defrag;
468
469 void (*init)(struct ip_vs_protocol *pp);
470
471 void (*exit)(struct ip_vs_protocol *pp);
472
473 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
474
475 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
476
477 int (*conn_schedule)(int af, struct sk_buff *skb,
478 struct ip_vs_proto_data *pd,
479 int *verdict, struct ip_vs_conn **cpp,
480 struct ip_vs_iphdr *iph);
481
482 struct ip_vs_conn *
483 (*conn_in_get)(int af,
484 const struct sk_buff *skb,
485 const struct ip_vs_iphdr *iph,
486 int inverse);
487
488 struct ip_vs_conn *
489 (*conn_out_get)(int af,
490 const struct sk_buff *skb,
491 const struct ip_vs_iphdr *iph,
492 int inverse);
493
494 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
495 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
496
497 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
498 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
499
500 int (*csum_check)(int af, struct sk_buff *skb,
501 struct ip_vs_protocol *pp);
502
503 const char *(*state_name)(int state);
504
505 void (*state_transition)(struct ip_vs_conn *cp, int direction,
506 const struct sk_buff *skb,
507 struct ip_vs_proto_data *pd);
508
509 int (*register_app)(struct net *net, struct ip_vs_app *inc);
510
511 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
512
513 int (*app_conn_bind)(struct ip_vs_conn *cp);
514
515 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
516 const struct sk_buff *skb,
517 int offset,
518 const char *msg);
519
520 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
521 };
522
523 /*
524 * protocol data per netns
525 */
526 struct ip_vs_proto_data {
527 struct ip_vs_proto_data *next;
528 struct ip_vs_protocol *pp;
529 int *timeout_table; /* protocol timeout table */
530 atomic_t appcnt; /* counter of proto app incs. */
531 struct tcp_states_t *tcp_state_table;
532 };
533
534 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
535 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
536 unsigned short proto);
537
538 struct ip_vs_conn_param {
539 struct net *net;
540 const union nf_inet_addr *caddr;
541 const union nf_inet_addr *vaddr;
542 __be16 cport;
543 __be16 vport;
544 __u16 protocol;
545 u16 af;
546
547 const struct ip_vs_pe *pe;
548 char *pe_data;
549 __u8 pe_data_len;
550 };
551
552 /*
553 * IP_VS structure allocated for each dynamically scheduled connection
554 */
555 struct ip_vs_conn {
556 struct hlist_node c_list; /* hashed list heads */
557 /* Protocol, addresses and port numbers */
558 __be16 cport;
559 __be16 dport;
560 __be16 vport;
561 u16 af; /* address family */
562 union nf_inet_addr caddr; /* client address */
563 union nf_inet_addr vaddr; /* virtual address */
564 union nf_inet_addr daddr; /* destination address */
565 volatile __u32 flags; /* status flags */
566 __u16 protocol; /* Which protocol (TCP/UDP) */
567 #ifdef CONFIG_NET_NS
568 struct net *net; /* Name space */
569 #endif
570
571 /* counter and timer */
572 atomic_t refcnt; /* reference count */
573 struct timer_list timer; /* Expiration timer */
574 volatile unsigned long timeout; /* timeout */
575
576 /* Flags and state transition */
577 spinlock_t lock; /* lock for state transition */
578 volatile __u16 state; /* state info */
579 volatile __u16 old_state; /* old state, to be used for
580 * state transition triggerd
581 * synchronization
582 */
583 __u32 fwmark; /* Fire wall mark from skb */
584 unsigned long sync_endtime; /* jiffies + sent_retries */
585
586 /* Control members */
587 struct ip_vs_conn *control; /* Master control connection */
588 atomic_t n_control; /* Number of controlled ones */
589 struct ip_vs_dest *dest; /* real server */
590 atomic_t in_pkts; /* incoming packet counter */
591
592 /* packet transmitter for different forwarding methods. If it
593 mangles the packet, it must return NF_DROP or better NF_STOLEN,
594 otherwise this must be changed to a sk_buff **.
595 NF_ACCEPT can be returned when destination is local.
596 */
597 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
598 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
599
600 /* Note: we can group the following members into a structure,
601 in order to save more space, and the following members are
602 only used in VS/NAT anyway */
603 struct ip_vs_app *app; /* bound ip_vs_app object */
604 void *app_data; /* Application private data */
605 struct ip_vs_seq in_seq; /* incoming seq. struct */
606 struct ip_vs_seq out_seq; /* outgoing seq. struct */
607
608 const struct ip_vs_pe *pe;
609 char *pe_data;
610 __u8 pe_data_len;
611
612 struct rcu_head rcu_head;
613 };
614
615 /*
616 * To save some memory in conn table when name space is disabled.
617 */
618 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
619 {
620 #ifdef CONFIG_NET_NS
621 return cp->net;
622 #else
623 return &init_net;
624 #endif
625 }
626 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
627 {
628 #ifdef CONFIG_NET_NS
629 cp->net = net;
630 #endif
631 }
632
633 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
634 struct net *net)
635 {
636 #ifdef CONFIG_NET_NS
637 return cp->net == net;
638 #else
639 return 1;
640 #endif
641 }
642
643 /*
644 * Extended internal versions of struct ip_vs_service_user and
645 * ip_vs_dest_user for IPv6 support.
646 *
647 * We need these to conveniently pass around service and destination
648 * options, but unfortunately, we also need to keep the old definitions to
649 * maintain userspace backwards compatibility for the setsockopt interface.
650 */
651 struct ip_vs_service_user_kern {
652 /* virtual service addresses */
653 u16 af;
654 u16 protocol;
655 union nf_inet_addr addr; /* virtual ip address */
656 __be16 port;
657 u32 fwmark; /* firwall mark of service */
658
659 /* virtual service options */
660 char *sched_name;
661 char *pe_name;
662 unsigned int flags; /* virtual service flags */
663 unsigned int timeout; /* persistent timeout in sec */
664 __be32 netmask; /* persistent netmask or plen */
665 };
666
667
668 struct ip_vs_dest_user_kern {
669 /* destination server address */
670 union nf_inet_addr addr;
671 __be16 port;
672
673 /* real server options */
674 unsigned int conn_flags; /* connection flags */
675 int weight; /* destination weight */
676
677 /* thresholds for active connections */
678 u32 u_threshold; /* upper threshold */
679 u32 l_threshold; /* lower threshold */
680 };
681
682
683 /*
684 * The information about the virtual service offered to the net
685 * and the forwarding entries
686 */
687 struct ip_vs_service {
688 struct hlist_node s_list; /* for normal service table */
689 struct hlist_node f_list; /* for fwmark-based service table */
690 atomic_t refcnt; /* reference counter */
691
692 u16 af; /* address family */
693 __u16 protocol; /* which protocol (TCP/UDP) */
694 union nf_inet_addr addr; /* IP address for virtual service */
695 __be16 port; /* port number for the service */
696 __u32 fwmark; /* firewall mark of the service */
697 unsigned int flags; /* service status flags */
698 unsigned int timeout; /* persistent timeout in ticks */
699 __be32 netmask; /* grouping granularity, mask/plen */
700 struct net *net;
701
702 struct list_head destinations; /* real server d-linked list */
703 __u32 num_dests; /* number of servers */
704 struct ip_vs_stats stats; /* statistics for the service */
705
706 /* for scheduling */
707 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
708 spinlock_t sched_lock; /* lock sched_data */
709 void *sched_data; /* scheduler application data */
710
711 /* alternate persistence engine */
712 struct ip_vs_pe __rcu *pe;
713
714 struct rcu_head rcu_head;
715 };
716
717 /* Information for cached dst */
718 struct ip_vs_dest_dst {
719 struct dst_entry *dst_cache; /* destination cache entry */
720 u32 dst_cookie;
721 union nf_inet_addr dst_saddr;
722 struct rcu_head rcu_head;
723 };
724
725 /* In grace period after removing */
726 #define IP_VS_DEST_STATE_REMOVING 0x01
727 /*
728 * The real server destination forwarding entry
729 * with ip address, port number, and so on.
730 */
731 struct ip_vs_dest {
732 struct list_head n_list; /* for the dests in the service */
733 struct hlist_node d_list; /* for table with all the dests */
734
735 u16 af; /* address family */
736 __be16 port; /* port number of the server */
737 union nf_inet_addr addr; /* IP address of the server */
738 volatile unsigned int flags; /* dest status flags */
739 atomic_t conn_flags; /* flags to copy to conn */
740 atomic_t weight; /* server weight */
741
742 atomic_t refcnt; /* reference counter */
743 struct ip_vs_stats stats; /* statistics */
744 unsigned long state; /* state flags */
745
746 /* connection counters and thresholds */
747 atomic_t activeconns; /* active connections */
748 atomic_t inactconns; /* inactive connections */
749 atomic_t persistconns; /* persistent connections */
750 __u32 u_threshold; /* upper threshold */
751 __u32 l_threshold; /* lower threshold */
752
753 /* for destination cache */
754 spinlock_t dst_lock; /* lock of dst_cache */
755 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
756
757 /* for virtual service */
758 struct ip_vs_service *svc; /* service it belongs to */
759 __u16 protocol; /* which protocol (TCP/UDP) */
760 __be16 vport; /* virtual port number */
761 union nf_inet_addr vaddr; /* virtual IP address */
762 __u32 vfwmark; /* firewall mark of service */
763
764 struct list_head t_list; /* in dest_trash */
765 struct rcu_head rcu_head;
766 unsigned int in_rs_table:1; /* we are in rs_table */
767 };
768
769
770 /*
771 * The scheduler object
772 */
773 struct ip_vs_scheduler {
774 struct list_head n_list; /* d-linked list head */
775 char *name; /* scheduler name */
776 atomic_t refcnt; /* reference counter */
777 struct module *module; /* THIS_MODULE/NULL */
778
779 /* scheduler initializing service */
780 int (*init_service)(struct ip_vs_service *svc);
781 /* scheduling service finish */
782 void (*done_service)(struct ip_vs_service *svc);
783 /* dest is linked */
784 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
785 /* dest is unlinked */
786 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
787 /* dest is updated */
788 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
789
790 /* selecting a server from the given service */
791 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
792 const struct sk_buff *skb,
793 struct ip_vs_iphdr *iph);
794 };
795
796 /* The persistence engine object */
797 struct ip_vs_pe {
798 struct list_head n_list; /* d-linked list head */
799 char *name; /* scheduler name */
800 atomic_t refcnt; /* reference counter */
801 struct module *module; /* THIS_MODULE/NULL */
802
803 /* get the connection template, if any */
804 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
805 bool (*ct_match)(const struct ip_vs_conn_param *p,
806 struct ip_vs_conn *ct);
807 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
808 bool inverse);
809 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
810 };
811
812 /*
813 * The application module object (a.k.a. app incarnation)
814 */
815 struct ip_vs_app {
816 struct list_head a_list; /* member in app list */
817 int type; /* IP_VS_APP_TYPE_xxx */
818 char *name; /* application module name */
819 __u16 protocol;
820 struct module *module; /* THIS_MODULE/NULL */
821 struct list_head incs_list; /* list of incarnations */
822
823 /* members for application incarnations */
824 struct list_head p_list; /* member in proto app list */
825 struct ip_vs_app *app; /* its real application */
826 __be16 port; /* port number in net order */
827 atomic_t usecnt; /* usage counter */
828 struct rcu_head rcu_head;
829
830 /*
831 * output hook: Process packet in inout direction, diff set for TCP.
832 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
833 * 2=Mangled but checksum was not updated
834 */
835 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
836 struct sk_buff *, int *diff);
837
838 /*
839 * input hook: Process packet in outin direction, diff set for TCP.
840 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
841 * 2=Mangled but checksum was not updated
842 */
843 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
844 struct sk_buff *, int *diff);
845
846 /* ip_vs_app initializer */
847 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
848
849 /* ip_vs_app finish */
850 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
851
852
853 /* not used now */
854 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
855 struct ip_vs_protocol *);
856
857 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
858
859 int * timeout_table;
860 int * timeouts;
861 int timeouts_size;
862
863 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
864 int *verdict, struct ip_vs_conn **cpp);
865
866 struct ip_vs_conn *
867 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
868 const struct iphdr *iph, int inverse);
869
870 struct ip_vs_conn *
871 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
872 const struct iphdr *iph, int inverse);
873
874 int (*state_transition)(struct ip_vs_conn *cp, int direction,
875 const struct sk_buff *skb,
876 struct ip_vs_app *app);
877
878 void (*timeout_change)(struct ip_vs_app *app, int flags);
879 };
880
881 struct ipvs_master_sync_state {
882 struct list_head sync_queue;
883 struct ip_vs_sync_buff *sync_buff;
884 unsigned long sync_queue_len;
885 unsigned int sync_queue_delay;
886 struct task_struct *master_thread;
887 struct delayed_work master_wakeup_work;
888 struct netns_ipvs *ipvs;
889 };
890
891 /* How much time to keep dests in trash */
892 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
893
894 /* IPVS in network namespace */
895 struct netns_ipvs {
896 int gen; /* Generation */
897 int enable; /* enable like nf_hooks do */
898 /*
899 * Hash table: for real service lookups
900 */
901 #define IP_VS_RTAB_BITS 4
902 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
903 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
904
905 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
906 /* ip_vs_app */
907 struct list_head app_list;
908 /* ip_vs_proto */
909 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
910 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
911 /* ip_vs_proto_tcp */
912 #ifdef CONFIG_IP_VS_PROTO_TCP
913 #define TCP_APP_TAB_BITS 4
914 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
915 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
916 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
917 #endif
918 /* ip_vs_proto_udp */
919 #ifdef CONFIG_IP_VS_PROTO_UDP
920 #define UDP_APP_TAB_BITS 4
921 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
922 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
923 struct list_head udp_apps[UDP_APP_TAB_SIZE];
924 #endif
925 /* ip_vs_proto_sctp */
926 #ifdef CONFIG_IP_VS_PROTO_SCTP
927 #define SCTP_APP_TAB_BITS 4
928 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
929 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
930 /* Hash table for SCTP application incarnations */
931 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
932 #endif
933 /* ip_vs_conn */
934 atomic_t conn_count; /* connection counter */
935
936 /* ip_vs_ctl */
937 struct ip_vs_stats tot_stats; /* Statistics & est. */
938
939 int num_services; /* no of virtual services */
940
941 /* Trash for destinations */
942 struct list_head dest_trash;
943 spinlock_t dest_trash_lock;
944 struct timer_list dest_trash_timer; /* expiration timer */
945 /* Service counters */
946 atomic_t ftpsvc_counter;
947 atomic_t nullsvc_counter;
948
949 #ifdef CONFIG_SYSCTL
950 /* 1/rate drop and drop-entry variables */
951 struct delayed_work defense_work; /* Work handler */
952 int drop_rate;
953 int drop_counter;
954 atomic_t dropentry;
955 /* locks in ctl.c */
956 spinlock_t dropentry_lock; /* drop entry handling */
957 spinlock_t droppacket_lock; /* drop packet handling */
958 spinlock_t securetcp_lock; /* state and timeout tables */
959
960 /* sys-ctl struct */
961 struct ctl_table_header *sysctl_hdr;
962 struct ctl_table *sysctl_tbl;
963 #endif
964
965 /* sysctl variables */
966 int sysctl_amemthresh;
967 int sysctl_am_droprate;
968 int sysctl_drop_entry;
969 int sysctl_drop_packet;
970 int sysctl_secure_tcp;
971 #ifdef CONFIG_IP_VS_NFCT
972 int sysctl_conntrack;
973 #endif
974 int sysctl_snat_reroute;
975 int sysctl_sync_ver;
976 int sysctl_sync_ports;
977 unsigned long sysctl_sync_qlen_max;
978 int sysctl_sync_sock_size;
979 int sysctl_cache_bypass;
980 int sysctl_expire_nodest_conn;
981 int sysctl_expire_quiescent_template;
982 int sysctl_sync_threshold[2];
983 unsigned int sysctl_sync_refresh_period;
984 int sysctl_sync_retries;
985 int sysctl_nat_icmp_send;
986 int sysctl_pmtu_disc;
987 int sysctl_backup_only;
988
989 /* ip_vs_lblc */
990 int sysctl_lblc_expiration;
991 struct ctl_table_header *lblc_ctl_header;
992 struct ctl_table *lblc_ctl_table;
993 /* ip_vs_lblcr */
994 int sysctl_lblcr_expiration;
995 struct ctl_table_header *lblcr_ctl_header;
996 struct ctl_table *lblcr_ctl_table;
997 /* ip_vs_est */
998 struct list_head est_list; /* estimator list */
999 spinlock_t est_lock;
1000 struct timer_list est_timer; /* Estimation timer */
1001 /* ip_vs_sync */
1002 spinlock_t sync_lock;
1003 struct ipvs_master_sync_state *ms;
1004 spinlock_t sync_buff_lock;
1005 struct task_struct **backup_threads;
1006 int threads_mask;
1007 int send_mesg_maxlen;
1008 int recv_mesg_maxlen;
1009 volatile int sync_state;
1010 volatile int master_syncid;
1011 volatile int backup_syncid;
1012 struct mutex sync_mutex;
1013 /* multicast interface name */
1014 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1015 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1016 /* net name space ptr */
1017 struct net *net; /* Needed by timer routines */
1018 };
1019
1020 #define DEFAULT_SYNC_THRESHOLD 3
1021 #define DEFAULT_SYNC_PERIOD 50
1022 #define DEFAULT_SYNC_VER 1
1023 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1024 #define DEFAULT_SYNC_RETRIES 0
1025 #define IPVS_SYNC_WAKEUP_RATE 8
1026 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1027 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1028 #define IPVS_SYNC_CHECK_PERIOD HZ
1029 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1030 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1031
1032 #ifdef CONFIG_SYSCTL
1033
1034 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1035 {
1036 return ipvs->sysctl_sync_threshold[0];
1037 }
1038
1039 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1040 {
1041 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
1042 }
1043
1044 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1045 {
1046 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
1047 }
1048
1049 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1050 {
1051 return ipvs->sysctl_sync_retries;
1052 }
1053
1054 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1055 {
1056 return ipvs->sysctl_sync_ver;
1057 }
1058
1059 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1060 {
1061 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
1062 }
1063
1064 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1065 {
1066 return ipvs->sysctl_sync_qlen_max;
1067 }
1068
1069 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1070 {
1071 return ipvs->sysctl_sync_sock_size;
1072 }
1073
1074 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1075 {
1076 return ipvs->sysctl_pmtu_disc;
1077 }
1078
1079 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1080 {
1081 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1082 ipvs->sysctl_backup_only;
1083 }
1084
1085 #else
1086
1087 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1088 {
1089 return DEFAULT_SYNC_THRESHOLD;
1090 }
1091
1092 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1093 {
1094 return DEFAULT_SYNC_PERIOD;
1095 }
1096
1097 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1098 {
1099 return DEFAULT_SYNC_REFRESH_PERIOD;
1100 }
1101
1102 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1103 {
1104 return DEFAULT_SYNC_RETRIES & 3;
1105 }
1106
1107 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1108 {
1109 return DEFAULT_SYNC_VER;
1110 }
1111
1112 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1113 {
1114 return 1;
1115 }
1116
1117 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1118 {
1119 return IPVS_SYNC_QLEN_MAX;
1120 }
1121
1122 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1123 {
1124 return 0;
1125 }
1126
1127 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1128 {
1129 return 1;
1130 }
1131
1132 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1133 {
1134 return 0;
1135 }
1136
1137 #endif
1138
1139 /*
1140 * IPVS core functions
1141 * (from ip_vs_core.c)
1142 */
1143 extern const char *ip_vs_proto_name(unsigned int proto);
1144 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
1145 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1146
1147 #define IP_VS_APP_TYPE_FTP 1
1148
1149 /*
1150 * ip_vs_conn handling functions
1151 * (from ip_vs_conn.c)
1152 */
1153
1154 enum {
1155 IP_VS_DIR_INPUT = 0,
1156 IP_VS_DIR_OUTPUT,
1157 IP_VS_DIR_INPUT_ONLY,
1158 IP_VS_DIR_LAST,
1159 };
1160
1161 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1162 const union nf_inet_addr *caddr,
1163 __be16 cport,
1164 const union nf_inet_addr *vaddr,
1165 __be16 vport,
1166 struct ip_vs_conn_param *p)
1167 {
1168 p->net = net;
1169 p->af = af;
1170 p->protocol = protocol;
1171 p->caddr = caddr;
1172 p->cport = cport;
1173 p->vaddr = vaddr;
1174 p->vport = vport;
1175 p->pe = NULL;
1176 p->pe_data = NULL;
1177 }
1178
1179 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1180 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1181
1182 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1183 const struct ip_vs_iphdr *iph,
1184 int inverse);
1185
1186 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1187
1188 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1189 const struct ip_vs_iphdr *iph,
1190 int inverse);
1191
1192 /* Get reference to gain full access to conn.
1193 * By default, RCU read-side critical sections have access only to
1194 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1195 */
1196 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1197 {
1198 return atomic_inc_not_zero(&cp->refcnt);
1199 }
1200
1201 /* put back the conn without restarting its timer */
1202 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1203 {
1204 smp_mb__before_atomic_dec();
1205 atomic_dec(&cp->refcnt);
1206 }
1207 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1208 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1209
1210 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1211 const union nf_inet_addr *daddr,
1212 __be16 dport, unsigned int flags,
1213 struct ip_vs_dest *dest, __u32 fwmark);
1214 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1215
1216 extern const char * ip_vs_state_name(__u16 proto, int state);
1217
1218 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1219 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1220 extern void ip_vs_random_dropentry(struct net *net);
1221 extern int ip_vs_conn_init(void);
1222 extern void ip_vs_conn_cleanup(void);
1223
1224 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1225 {
1226 struct ip_vs_conn *ctl_cp = cp->control;
1227 if (!ctl_cp) {
1228 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1229 "%s:%d to %s:%d\n",
1230 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1231 ntohs(cp->cport),
1232 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1233 ntohs(cp->vport));
1234
1235 return;
1236 }
1237
1238 IP_VS_DBG_BUF(7, "DELeting control for: "
1239 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1240 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1241 ntohs(cp->cport),
1242 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1243 ntohs(ctl_cp->cport));
1244
1245 cp->control = NULL;
1246 if (atomic_read(&ctl_cp->n_control) == 0) {
1247 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1248 "%s:%d to %s:%d\n",
1249 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1250 ntohs(cp->cport),
1251 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1252 ntohs(cp->vport));
1253
1254 return;
1255 }
1256 atomic_dec(&ctl_cp->n_control);
1257 }
1258
1259 static inline void
1260 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1261 {
1262 if (cp->control) {
1263 IP_VS_ERR_BUF("request control ADD for already controlled: "
1264 "%s:%d to %s:%d\n",
1265 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1266 ntohs(cp->cport),
1267 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1268 ntohs(cp->vport));
1269
1270 ip_vs_control_del(cp);
1271 }
1272
1273 IP_VS_DBG_BUF(7, "ADDing control for: "
1274 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1275 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1276 ntohs(cp->cport),
1277 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1278 ntohs(ctl_cp->cport));
1279
1280 cp->control = ctl_cp;
1281 atomic_inc(&ctl_cp->n_control);
1282 }
1283
1284 /*
1285 * IPVS netns init & cleanup functions
1286 */
1287 extern int ip_vs_estimator_net_init(struct net *net);
1288 extern int ip_vs_control_net_init(struct net *net);
1289 extern int ip_vs_protocol_net_init(struct net *net);
1290 extern int ip_vs_app_net_init(struct net *net);
1291 extern int ip_vs_conn_net_init(struct net *net);
1292 extern int ip_vs_sync_net_init(struct net *net);
1293 extern void ip_vs_conn_net_cleanup(struct net *net);
1294 extern void ip_vs_app_net_cleanup(struct net *net);
1295 extern void ip_vs_protocol_net_cleanup(struct net *net);
1296 extern void ip_vs_control_net_cleanup(struct net *net);
1297 extern void ip_vs_estimator_net_cleanup(struct net *net);
1298 extern void ip_vs_sync_net_cleanup(struct net *net);
1299 extern void ip_vs_service_net_cleanup(struct net *net);
1300
1301 /*
1302 * IPVS application functions
1303 * (from ip_vs_app.c)
1304 */
1305 #define IP_VS_APP_MAX_PORTS 8
1306 extern struct ip_vs_app *register_ip_vs_app(struct net *net,
1307 struct ip_vs_app *app);
1308 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1309 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1310 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1311 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1312 __u16 proto, __u16 port);
1313 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1314 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1315
1316 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1317 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1318
1319 int register_ip_vs_pe(struct ip_vs_pe *pe);
1320 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1321 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1322 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1323
1324 /*
1325 * Use a #define to avoid all of module.h just for these trivial ops
1326 */
1327 #define ip_vs_pe_get(pe) \
1328 if (pe && pe->module) \
1329 __module_get(pe->module);
1330
1331 #define ip_vs_pe_put(pe) \
1332 if (pe && pe->module) \
1333 module_put(pe->module);
1334
1335 /*
1336 * IPVS protocol functions (from ip_vs_proto.c)
1337 */
1338 extern int ip_vs_protocol_init(void);
1339 extern void ip_vs_protocol_cleanup(void);
1340 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1341 extern int *ip_vs_create_timeout_table(int *table, int size);
1342 extern int
1343 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1344 const char *name, int to);
1345 extern void
1346 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1347 const struct sk_buff *skb,
1348 int offset, const char *msg);
1349
1350 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1351 extern struct ip_vs_protocol ip_vs_protocol_udp;
1352 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1353 extern struct ip_vs_protocol ip_vs_protocol_esp;
1354 extern struct ip_vs_protocol ip_vs_protocol_ah;
1355 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1356
1357 /*
1358 * Registering/unregistering scheduler functions
1359 * (from ip_vs_sched.c)
1360 */
1361 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1362 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1363 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1364 struct ip_vs_scheduler *scheduler);
1365 extern void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1366 struct ip_vs_scheduler *sched);
1367 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1368 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1369 extern struct ip_vs_conn *
1370 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1371 struct ip_vs_proto_data *pd, int *ignored,
1372 struct ip_vs_iphdr *iph);
1373 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1374 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1375
1376 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1377
1378
1379 /*
1380 * IPVS control data and functions (from ip_vs_ctl.c)
1381 */
1382 extern struct ip_vs_stats ip_vs_stats;
1383 extern int sysctl_ip_vs_sync_ver;
1384
1385 extern struct ip_vs_service *
1386 ip_vs_service_find(struct net *net, int af, __u32 fwmark, __u16 protocol,
1387 const union nf_inet_addr *vaddr, __be16 vport);
1388
1389 extern bool
1390 ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
1391 const union nf_inet_addr *daddr, __be16 dport);
1392
1393 extern int ip_vs_use_count_inc(void);
1394 extern void ip_vs_use_count_dec(void);
1395 extern int ip_vs_register_nl_ioctl(void);
1396 extern void ip_vs_unregister_nl_ioctl(void);
1397 extern int ip_vs_control_init(void);
1398 extern void ip_vs_control_cleanup(void);
1399 extern struct ip_vs_dest *
1400 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1401 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1402 __u16 protocol, __u32 fwmark, __u32 flags);
1403 extern void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1404
1405 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1406 {
1407 atomic_inc(&dest->refcnt);
1408 }
1409
1410 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1411 {
1412 smp_mb__before_atomic_dec();
1413 atomic_dec(&dest->refcnt);
1414 }
1415
1416 /*
1417 * IPVS sync daemon data and function prototypes
1418 * (from ip_vs_sync.c)
1419 */
1420 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1421 __u8 syncid);
1422 extern int stop_sync_thread(struct net *net, int state);
1423 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1424
1425
1426 /*
1427 * IPVS rate estimator prototypes (from ip_vs_est.c)
1428 */
1429 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1430 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1431 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1432 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1433 struct ip_vs_stats *stats);
1434
1435 /*
1436 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1437 */
1438 extern int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1439 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1440 extern int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1441 struct ip_vs_protocol *pp,
1442 struct ip_vs_iphdr *iph);
1443 extern int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1444 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1445 extern int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1446 struct ip_vs_protocol *pp,
1447 struct ip_vs_iphdr *iph);
1448 extern int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1449 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1450 extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1451 struct ip_vs_protocol *pp, int offset,
1452 unsigned int hooknum, struct ip_vs_iphdr *iph);
1453 extern void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1454
1455 #ifdef CONFIG_IP_VS_IPV6
1456 extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1457 struct ip_vs_protocol *pp,
1458 struct ip_vs_iphdr *iph);
1459 extern int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1460 struct ip_vs_protocol *pp,
1461 struct ip_vs_iphdr *iph);
1462 extern int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1463 struct ip_vs_protocol *pp,
1464 struct ip_vs_iphdr *iph);
1465 extern int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1466 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1467 extern int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1468 struct ip_vs_protocol *pp, int offset,
1469 unsigned int hooknum, struct ip_vs_iphdr *iph);
1470 #endif
1471
1472 #ifdef CONFIG_SYSCTL
1473 /*
1474 * This is a simple mechanism to ignore packets when
1475 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1476 * we start to drop 1/rate of the packets
1477 */
1478
1479 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1480 {
1481 if (!ipvs->drop_rate)
1482 return 0;
1483 if (--ipvs->drop_counter > 0)
1484 return 0;
1485 ipvs->drop_counter = ipvs->drop_rate;
1486 return 1;
1487 }
1488 #else
1489 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1490 #endif
1491
1492 /*
1493 * ip_vs_fwd_tag returns the forwarding tag of the connection
1494 */
1495 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1496
1497 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1498 {
1499 char fwd;
1500
1501 switch (IP_VS_FWD_METHOD(cp)) {
1502 case IP_VS_CONN_F_MASQ:
1503 fwd = 'M'; break;
1504 case IP_VS_CONN_F_LOCALNODE:
1505 fwd = 'L'; break;
1506 case IP_VS_CONN_F_TUNNEL:
1507 fwd = 'T'; break;
1508 case IP_VS_CONN_F_DROUTE:
1509 fwd = 'R'; break;
1510 case IP_VS_CONN_F_BYPASS:
1511 fwd = 'B'; break;
1512 default:
1513 fwd = '?'; break;
1514 }
1515 return fwd;
1516 }
1517
1518 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1519 struct ip_vs_conn *cp, int dir);
1520
1521 #ifdef CONFIG_IP_VS_IPV6
1522 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1523 struct ip_vs_conn *cp, int dir);
1524 #endif
1525
1526 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1527
1528 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1529 {
1530 __be32 diff[2] = { ~old, new };
1531
1532 return csum_partial(diff, sizeof(diff), oldsum);
1533 }
1534
1535 #ifdef CONFIG_IP_VS_IPV6
1536 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1537 __wsum oldsum)
1538 {
1539 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1540 new[3], new[2], new[1], new[0] };
1541
1542 return csum_partial(diff, sizeof(diff), oldsum);
1543 }
1544 #endif
1545
1546 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1547 {
1548 __be16 diff[2] = { ~old, new };
1549
1550 return csum_partial(diff, sizeof(diff), oldsum);
1551 }
1552
1553 /*
1554 * Forget current conntrack (unconfirmed) and attach notrack entry
1555 */
1556 static inline void ip_vs_notrack(struct sk_buff *skb)
1557 {
1558 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1559 enum ip_conntrack_info ctinfo;
1560 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1561
1562 if (!ct || !nf_ct_is_untracked(ct)) {
1563 nf_conntrack_put(skb->nfct);
1564 skb->nfct = &nf_ct_untracked_get()->ct_general;
1565 skb->nfctinfo = IP_CT_NEW;
1566 nf_conntrack_get(skb->nfct);
1567 }
1568 #endif
1569 }
1570
1571 #ifdef CONFIG_IP_VS_NFCT
1572 /*
1573 * Netfilter connection tracking
1574 * (from ip_vs_nfct.c)
1575 */
1576 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1577 {
1578 #ifdef CONFIG_SYSCTL
1579 return ipvs->sysctl_conntrack;
1580 #else
1581 return 0;
1582 #endif
1583 }
1584
1585 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1586 int outin);
1587 extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1588 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1589 struct ip_vs_conn *cp, u_int8_t proto,
1590 const __be16 port, int from_rs);
1591 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1592
1593 #else
1594
1595 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1596 {
1597 return 0;
1598 }
1599
1600 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1601 struct ip_vs_conn *cp, int outin)
1602 {
1603 }
1604
1605 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1606 {
1607 return NF_ACCEPT;
1608 }
1609
1610 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1611 {
1612 }
1613 /* CONFIG_IP_VS_NFCT */
1614 #endif
1615
1616 static inline unsigned int
1617 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1618 {
1619 /*
1620 * We think the overhead of processing active connections is 256
1621 * times higher than that of inactive connections in average. (This
1622 * 256 times might not be accurate, we will change it later) We
1623 * use the following formula to estimate the overhead now:
1624 * dest->activeconns*256 + dest->inactconns
1625 */
1626 return (atomic_read(&dest->activeconns) << 8) +
1627 atomic_read(&dest->inactconns);
1628 }
1629
1630 #endif /* _NET_IP_VS_H */
Linux with added FPC-III support