]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blob - include/net/ip_vs.h
projects / mirror_ubuntu-zesty-kernel.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
ipvs: reorder keys in connection structure
[mirror_ubuntu-zesty-kernel.git] / include / net / ip_vs.h
1 /*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
6 #ifndef _NET_IP_VS_H
7 #define _NET_IP_VS_H
8
9 #include <linux/ip_vs.h> /* definitions shared with userland */
10
11 #include <asm/types.h> /* for __uXX types */
12
13 #include <linux/list.h> /* for struct list_head */
14 #include <linux/spinlock.h> /* for struct rwlock_t */
15 #include <linux/atomic.h> /* for struct atomic_t */
16 #include <linux/compiler.h>
17 #include <linux/timer.h>
18 #include <linux/bug.h>
19
20 #include <net/checksum.h>
21 #include <linux/netfilter.h> /* for union nf_inet_addr */
22 #include <linux/ip.h>
23 #include <linux/ipv6.h> /* for struct ipv6hdr */
24 #include <net/ipv6.h>
25 #if 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 /* This function is a faster version of ip_vs_fill_iph_skb().
201 * Where we only populate {s,d}addr (and avoid calling ipv6_find_hdr()).
202 * This is used by the some of the ip_vs_*_schedule() functions.
203 * (Mostly done to avoid ABI breakage of external schedulers)
204 */
205 static inline void
206 ip_vs_fill_iph_addr_only(int af, const struct sk_buff *skb,
207 struct ip_vs_iphdr *iphdr)
208 {
209 #ifdef CONFIG_IP_VS_IPV6
210 if (af == AF_INET6) {
211 const struct ipv6hdr *iph =
212 (struct ipv6hdr *)skb_network_header(skb);
213 iphdr->saddr.in6 = iph->saddr;
214 iphdr->daddr.in6 = iph->daddr;
215 } else
216 #endif
217 {
218 const struct iphdr *iph =
219 (struct iphdr *)skb_network_header(skb);
220 iphdr->saddr.ip = iph->saddr;
221 iphdr->daddr.ip = iph->daddr;
222 }
223 }
224
225 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
226 const union nf_inet_addr *src)
227 {
228 #ifdef CONFIG_IP_VS_IPV6
229 if (af == AF_INET6)
230 dst->in6 = src->in6;
231 else
232 #endif
233 dst->ip = src->ip;
234 }
235
236 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
237 const union nf_inet_addr *b)
238 {
239 #ifdef CONFIG_IP_VS_IPV6
240 if (af == AF_INET6)
241 return ipv6_addr_equal(&a->in6, &b->in6);
242 #endif
243 return a->ip == b->ip;
244 }
245
246 #ifdef CONFIG_IP_VS_DEBUG
247 #include <linux/net.h>
248
249 extern int ip_vs_get_debug_level(void);
250
251 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
252 const union nf_inet_addr *addr,
253 int *idx)
254 {
255 int len;
256 #ifdef CONFIG_IP_VS_IPV6
257 if (af == AF_INET6)
258 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
259 &addr->in6) + 1;
260 else
261 #endif
262 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
263 &addr->ip) + 1;
264
265 *idx += len;
266 BUG_ON(*idx > buf_len + 1);
267 return &buf[*idx - len];
268 }
269
270 #define IP_VS_DBG_BUF(level, msg, ...) \
271 do { \
272 char ip_vs_dbg_buf[160]; \
273 int ip_vs_dbg_idx = 0; \
274 if (level <= ip_vs_get_debug_level()) \
275 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
276 } while (0)
277 #define IP_VS_ERR_BUF(msg...) \
278 do { \
279 char ip_vs_dbg_buf[160]; \
280 int ip_vs_dbg_idx = 0; \
281 pr_err(msg); \
282 } while (0)
283
284 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
285 #define IP_VS_DBG_ADDR(af, addr) \
286 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
287 sizeof(ip_vs_dbg_buf), addr, \
288 &ip_vs_dbg_idx)
289
290 #define IP_VS_DBG(level, msg, ...) \
291 do { \
292 if (level <= ip_vs_get_debug_level()) \
293 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
294 } while (0)
295 #define IP_VS_DBG_RL(msg, ...) \
296 do { \
297 if (net_ratelimit()) \
298 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
299 } while (0)
300 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
301 do { \
302 if (level <= ip_vs_get_debug_level()) \
303 pp->debug_packet(af, pp, skb, ofs, msg); \
304 } while (0)
305 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
306 do { \
307 if (level <= ip_vs_get_debug_level() && \
308 net_ratelimit()) \
309 pp->debug_packet(af, pp, skb, ofs, msg); \
310 } while (0)
311 #else /* NO DEBUGGING at ALL */
312 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
313 #define IP_VS_ERR_BUF(msg...) do {} while (0)
314 #define IP_VS_DBG(level, msg...) do {} while (0)
315 #define IP_VS_DBG_RL(msg...) do {} while (0)
316 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
317 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
318 #endif
319
320 #define IP_VS_BUG() BUG()
321 #define IP_VS_ERR_RL(msg, ...) \
322 do { \
323 if (net_ratelimit()) \
324 pr_err(msg, ##__VA_ARGS__); \
325 } while (0)
326
327 #ifdef CONFIG_IP_VS_DEBUG
328 #define EnterFunction(level) \
329 do { \
330 if (level <= ip_vs_get_debug_level()) \
331 printk(KERN_DEBUG \
332 pr_fmt("Enter: %s, %s line %i\n"), \
333 __func__, __FILE__, __LINE__); \
334 } while (0)
335 #define LeaveFunction(level) \
336 do { \
337 if (level <= ip_vs_get_debug_level()) \
338 printk(KERN_DEBUG \
339 pr_fmt("Leave: %s, %s line %i\n"), \
340 __func__, __FILE__, __LINE__); \
341 } while (0)
342 #else
343 #define EnterFunction(level) do {} while (0)
344 #define LeaveFunction(level) do {} while (0)
345 #endif
346
347 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
348
349
350 /*
351 * The port number of FTP service (in network order).
352 */
353 #define FTPPORT cpu_to_be16(21)
354 #define FTPDATA cpu_to_be16(20)
355
356 /*
357 * TCP State Values
358 */
359 enum {
360 IP_VS_TCP_S_NONE = 0,
361 IP_VS_TCP_S_ESTABLISHED,
362 IP_VS_TCP_S_SYN_SENT,
363 IP_VS_TCP_S_SYN_RECV,
364 IP_VS_TCP_S_FIN_WAIT,
365 IP_VS_TCP_S_TIME_WAIT,
366 IP_VS_TCP_S_CLOSE,
367 IP_VS_TCP_S_CLOSE_WAIT,
368 IP_VS_TCP_S_LAST_ACK,
369 IP_VS_TCP_S_LISTEN,
370 IP_VS_TCP_S_SYNACK,
371 IP_VS_TCP_S_LAST
372 };
373
374 /*
375 * UDP State Values
376 */
377 enum {
378 IP_VS_UDP_S_NORMAL,
379 IP_VS_UDP_S_LAST,
380 };
381
382 /*
383 * ICMP State Values
384 */
385 enum {
386 IP_VS_ICMP_S_NORMAL,
387 IP_VS_ICMP_S_LAST,
388 };
389
390 /*
391 * SCTP State Values
392 */
393 enum ip_vs_sctp_states {
394 IP_VS_SCTP_S_NONE,
395 IP_VS_SCTP_S_INIT_CLI,
396 IP_VS_SCTP_S_INIT_SER,
397 IP_VS_SCTP_S_INIT_ACK_CLI,
398 IP_VS_SCTP_S_INIT_ACK_SER,
399 IP_VS_SCTP_S_ECHO_CLI,
400 IP_VS_SCTP_S_ECHO_SER,
401 IP_VS_SCTP_S_ESTABLISHED,
402 IP_VS_SCTP_S_SHUT_CLI,
403 IP_VS_SCTP_S_SHUT_SER,
404 IP_VS_SCTP_S_SHUT_ACK_CLI,
405 IP_VS_SCTP_S_SHUT_ACK_SER,
406 IP_VS_SCTP_S_CLOSED,
407 IP_VS_SCTP_S_LAST
408 };
409
410 /*
411 * Delta sequence info structure
412 * Each ip_vs_conn has 2 (output AND input seq. changes).
413 * Only used in the VS/NAT.
414 */
415 struct ip_vs_seq {
416 __u32 init_seq; /* Add delta from this seq */
417 __u32 delta; /* Delta in sequence numbers */
418 __u32 previous_delta; /* Delta in sequence numbers
419 before last resized pkt */
420 };
421
422 /*
423 * counters per cpu
424 */
425 struct ip_vs_counters {
426 __u32 conns; /* connections scheduled */
427 __u32 inpkts; /* incoming packets */
428 __u32 outpkts; /* outgoing packets */
429 __u64 inbytes; /* incoming bytes */
430 __u64 outbytes; /* outgoing bytes */
431 };
432 /*
433 * Stats per cpu
434 */
435 struct ip_vs_cpu_stats {
436 struct ip_vs_counters ustats;
437 struct u64_stats_sync syncp;
438 };
439
440 /*
441 * IPVS statistics objects
442 */
443 struct ip_vs_estimator {
444 struct list_head list;
445
446 u64 last_inbytes;
447 u64 last_outbytes;
448 u32 last_conns;
449 u32 last_inpkts;
450 u32 last_outpkts;
451
452 u32 cps;
453 u32 inpps;
454 u32 outpps;
455 u32 inbps;
456 u32 outbps;
457 };
458
459 struct ip_vs_stats {
460 struct ip_vs_stats_user ustats; /* statistics */
461 struct ip_vs_estimator est; /* estimator */
462 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
463 spinlock_t lock; /* spin lock */
464 struct ip_vs_stats_user ustats0; /* reset values */
465 };
466
467 struct dst_entry;
468 struct iphdr;
469 struct ip_vs_conn;
470 struct ip_vs_app;
471 struct sk_buff;
472 struct ip_vs_proto_data;
473
474 struct ip_vs_protocol {
475 struct ip_vs_protocol *next;
476 char *name;
477 u16 protocol;
478 u16 num_states;
479 int dont_defrag;
480
481 void (*init)(struct ip_vs_protocol *pp);
482
483 void (*exit)(struct ip_vs_protocol *pp);
484
485 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
486
487 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
488
489 int (*conn_schedule)(int af, struct sk_buff *skb,
490 struct ip_vs_proto_data *pd,
491 int *verdict, struct ip_vs_conn **cpp,
492 struct ip_vs_iphdr *iph);
493
494 struct ip_vs_conn *
495 (*conn_in_get)(int af,
496 const struct sk_buff *skb,
497 const struct ip_vs_iphdr *iph,
498 int inverse);
499
500 struct ip_vs_conn *
501 (*conn_out_get)(int af,
502 const struct sk_buff *skb,
503 const struct ip_vs_iphdr *iph,
504 int inverse);
505
506 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
507 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
508
509 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
510 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
511
512 int (*csum_check)(int af, struct sk_buff *skb,
513 struct ip_vs_protocol *pp);
514
515 const char *(*state_name)(int state);
516
517 void (*state_transition)(struct ip_vs_conn *cp, int direction,
518 const struct sk_buff *skb,
519 struct ip_vs_proto_data *pd);
520
521 int (*register_app)(struct net *net, struct ip_vs_app *inc);
522
523 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
524
525 int (*app_conn_bind)(struct ip_vs_conn *cp);
526
527 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
528 const struct sk_buff *skb,
529 int offset,
530 const char *msg);
531
532 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
533 };
534
535 /*
536 * protocol data per netns
537 */
538 struct ip_vs_proto_data {
539 struct ip_vs_proto_data *next;
540 struct ip_vs_protocol *pp;
541 int *timeout_table; /* protocol timeout table */
542 atomic_t appcnt; /* counter of proto app incs. */
543 struct tcp_states_t *tcp_state_table;
544 };
545
546 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
547 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
548 unsigned short proto);
549
550 struct ip_vs_conn_param {
551 struct net *net;
552 const union nf_inet_addr *caddr;
553 const union nf_inet_addr *vaddr;
554 __be16 cport;
555 __be16 vport;
556 __u16 protocol;
557 u16 af;
558
559 const struct ip_vs_pe *pe;
560 char *pe_data;
561 __u8 pe_data_len;
562 };
563
564 /*
565 * IP_VS structure allocated for each dynamically scheduled connection
566 */
567 struct ip_vs_conn {
568 struct hlist_node c_list; /* hashed list heads */
569 /* Protocol, addresses and port numbers */
570 __be16 cport;
571 __be16 dport;
572 __be16 vport;
573 u16 af; /* address family */
574 union nf_inet_addr caddr; /* client address */
575 union nf_inet_addr vaddr; /* virtual address */
576 union nf_inet_addr daddr; /* destination address */
577 volatile __u32 flags; /* status flags */
578 __u16 protocol; /* Which protocol (TCP/UDP) */
579 #ifdef CONFIG_NET_NS
580 struct net *net; /* Name space */
581 #endif
582
583 /* counter and timer */
584 atomic_t refcnt; /* reference count */
585 struct timer_list timer; /* Expiration timer */
586 volatile unsigned long timeout; /* timeout */
587
588 /* Flags and state transition */
589 spinlock_t lock; /* lock for state transition */
590 volatile __u16 state; /* state info */
591 volatile __u16 old_state; /* old state, to be used for
592 * state transition triggerd
593 * synchronization
594 */
595 __u32 fwmark; /* Fire wall mark from skb */
596 unsigned long sync_endtime; /* jiffies + sent_retries */
597
598 /* Control members */
599 struct ip_vs_conn *control; /* Master control connection */
600 atomic_t n_control; /* Number of controlled ones */
601 struct ip_vs_dest *dest; /* real server */
602 atomic_t in_pkts; /* incoming packet counter */
603
604 /* packet transmitter for different forwarding methods. If it
605 mangles the packet, it must return NF_DROP or better NF_STOLEN,
606 otherwise this must be changed to a sk_buff **.
607 NF_ACCEPT can be returned when destination is local.
608 */
609 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
610 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
611
612 /* Note: we can group the following members into a structure,
613 in order to save more space, and the following members are
614 only used in VS/NAT anyway */
615 struct ip_vs_app *app; /* bound ip_vs_app object */
616 void *app_data; /* Application private data */
617 struct ip_vs_seq in_seq; /* incoming seq. struct */
618 struct ip_vs_seq out_seq; /* outgoing seq. struct */
619
620 const struct ip_vs_pe *pe;
621 char *pe_data;
622 __u8 pe_data_len;
623
624 struct rcu_head rcu_head;
625 };
626
627 /*
628 * To save some memory in conn table when name space is disabled.
629 */
630 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
631 {
632 #ifdef CONFIG_NET_NS
633 return cp->net;
634 #else
635 return &init_net;
636 #endif
637 }
638 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
639 {
640 #ifdef CONFIG_NET_NS
641 cp->net = net;
642 #endif
643 }
644
645 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
646 struct net *net)
647 {
648 #ifdef CONFIG_NET_NS
649 return cp->net == net;
650 #else
651 return 1;
652 #endif
653 }
654
655 /*
656 * Extended internal versions of struct ip_vs_service_user and
657 * ip_vs_dest_user for IPv6 support.
658 *
659 * We need these to conveniently pass around service and destination
660 * options, but unfortunately, we also need to keep the old definitions to
661 * maintain userspace backwards compatibility for the setsockopt interface.
662 */
663 struct ip_vs_service_user_kern {
664 /* virtual service addresses */
665 u16 af;
666 u16 protocol;
667 union nf_inet_addr addr; /* virtual ip address */
668 u16 port;
669 u32 fwmark; /* firwall mark of service */
670
671 /* virtual service options */
672 char *sched_name;
673 char *pe_name;
674 unsigned int flags; /* virtual service flags */
675 unsigned int timeout; /* persistent timeout in sec */
676 u32 netmask; /* persistent netmask */
677 };
678
679
680 struct ip_vs_dest_user_kern {
681 /* destination server address */
682 union nf_inet_addr addr;
683 u16 port;
684
685 /* real server options */
686 unsigned int conn_flags; /* connection flags */
687 int weight; /* destination weight */
688
689 /* thresholds for active connections */
690 u32 u_threshold; /* upper threshold */
691 u32 l_threshold; /* lower threshold */
692 };
693
694
695 /*
696 * The information about the virtual service offered to the net
697 * and the forwarding entries
698 */
699 struct ip_vs_service {
700 struct list_head s_list; /* for normal service table */
701 struct list_head f_list; /* for fwmark-based service table */
702 atomic_t refcnt; /* reference counter */
703 atomic_t usecnt; /* use counter */
704
705 u16 af; /* address family */
706 __u16 protocol; /* which protocol (TCP/UDP) */
707 union nf_inet_addr addr; /* IP address for virtual service */
708 __be16 port; /* port number for the service */
709 __u32 fwmark; /* firewall mark of the service */
710 unsigned int flags; /* service status flags */
711 unsigned int timeout; /* persistent timeout in ticks */
712 __be32 netmask; /* grouping granularity */
713 struct net *net;
714
715 struct list_head destinations; /* real server d-linked list */
716 __u32 num_dests; /* number of servers */
717 struct ip_vs_stats stats; /* statistics for the service */
718 struct ip_vs_app *inc; /* bind conns to this app inc */
719
720 /* for scheduling */
721 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
722 rwlock_t sched_lock; /* lock sched_data */
723 void *sched_data; /* scheduler application data */
724
725 /* alternate persistence engine */
726 struct ip_vs_pe *pe;
727 };
728
729 /* Information for cached dst */
730 struct ip_vs_dest_dst {
731 struct dst_entry *dst_cache; /* destination cache entry */
732 u32 dst_cookie;
733 union nf_inet_addr dst_saddr;
734 struct rcu_head rcu_head;
735 };
736
737 /*
738 * The real server destination forwarding entry
739 * with ip address, port number, and so on.
740 */
741 struct ip_vs_dest {
742 struct list_head n_list; /* for the dests in the service */
743 struct hlist_node d_list; /* for table with all the dests */
744
745 u16 af; /* address family */
746 __be16 port; /* port number of the server */
747 union nf_inet_addr addr; /* IP address of the server */
748 volatile unsigned int flags; /* dest status flags */
749 atomic_t conn_flags; /* flags to copy to conn */
750 atomic_t weight; /* server weight */
751
752 atomic_t refcnt; /* reference counter */
753 struct ip_vs_stats stats; /* statistics */
754
755 /* connection counters and thresholds */
756 atomic_t activeconns; /* active connections */
757 atomic_t inactconns; /* inactive connections */
758 atomic_t persistconns; /* persistent connections */
759 __u32 u_threshold; /* upper threshold */
760 __u32 l_threshold; /* lower threshold */
761
762 /* for destination cache */
763 spinlock_t dst_lock; /* lock of dst_cache */
764 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
765
766 /* for virtual service */
767 struct ip_vs_service *svc; /* service it belongs to */
768 __u16 protocol; /* which protocol (TCP/UDP) */
769 __be16 vport; /* virtual port number */
770 union nf_inet_addr vaddr; /* virtual IP address */
771 __u32 vfwmark; /* firewall mark of service */
772
773 struct rcu_head rcu_head;
774 unsigned int in_rs_table:1; /* we are in rs_table */
775 };
776
777
778 /*
779 * The scheduler object
780 */
781 struct ip_vs_scheduler {
782 struct list_head n_list; /* d-linked list head */
783 char *name; /* scheduler name */
784 atomic_t refcnt; /* reference counter */
785 struct module *module; /* THIS_MODULE/NULL */
786
787 /* scheduler initializing service */
788 int (*init_service)(struct ip_vs_service *svc);
789 /* scheduling service finish */
790 int (*done_service)(struct ip_vs_service *svc);
791 /* scheduler updating service */
792 int (*update_service)(struct ip_vs_service *svc);
793
794 /* selecting a server from the given service */
795 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
796 const struct sk_buff *skb);
797 };
798
799 /* The persistence engine object */
800 struct ip_vs_pe {
801 struct list_head n_list; /* d-linked list head */
802 char *name; /* scheduler name */
803 atomic_t refcnt; /* reference counter */
804 struct module *module; /* THIS_MODULE/NULL */
805
806 /* get the connection template, if any */
807 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
808 bool (*ct_match)(const struct ip_vs_conn_param *p,
809 struct ip_vs_conn *ct);
810 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
811 bool inverse);
812 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
813 };
814
815 /*
816 * The application module object (a.k.a. app incarnation)
817 */
818 struct ip_vs_app {
819 struct list_head a_list; /* member in app list */
820 int type; /* IP_VS_APP_TYPE_xxx */
821 char *name; /* application module name */
822 __u16 protocol;
823 struct module *module; /* THIS_MODULE/NULL */
824 struct list_head incs_list; /* list of incarnations */
825
826 /* members for application incarnations */
827 struct list_head p_list; /* member in proto app list */
828 struct ip_vs_app *app; /* its real application */
829 __be16 port; /* port number in net order */
830 atomic_t usecnt; /* usage counter */
831 struct rcu_head rcu_head;
832
833 /*
834 * output hook: Process packet in inout direction, diff set for TCP.
835 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
836 * 2=Mangled but checksum was not updated
837 */
838 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
839 struct sk_buff *, int *diff);
840
841 /*
842 * input hook: Process packet in outin direction, diff set for TCP.
843 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
844 * 2=Mangled but checksum was not updated
845 */
846 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
847 struct sk_buff *, int *diff);
848
849 /* ip_vs_app initializer */
850 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
851
852 /* ip_vs_app finish */
853 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
854
855
856 /* not used now */
857 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
858 struct ip_vs_protocol *);
859
860 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
861
862 int * timeout_table;
863 int * timeouts;
864 int timeouts_size;
865
866 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
867 int *verdict, struct ip_vs_conn **cpp);
868
869 struct ip_vs_conn *
870 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
871 const struct iphdr *iph, int inverse);
872
873 struct ip_vs_conn *
874 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
875 const struct iphdr *iph, int inverse);
876
877 int (*state_transition)(struct ip_vs_conn *cp, int direction,
878 const struct sk_buff *skb,
879 struct ip_vs_app *app);
880
881 void (*timeout_change)(struct ip_vs_app *app, int flags);
882 };
883
884 struct ipvs_master_sync_state {
885 struct list_head sync_queue;
886 struct ip_vs_sync_buff *sync_buff;
887 int sync_queue_len;
888 unsigned int sync_queue_delay;
889 struct task_struct *master_thread;
890 struct delayed_work master_wakeup_work;
891 struct netns_ipvs *ipvs;
892 };
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 /* Service counters */
944 atomic_t ftpsvc_counter;
945 atomic_t nullsvc_counter;
946
947 #ifdef CONFIG_SYSCTL
948 /* 1/rate drop and drop-entry variables */
949 struct delayed_work defense_work; /* Work handler */
950 int drop_rate;
951 int drop_counter;
952 atomic_t dropentry;
953 /* locks in ctl.c */
954 spinlock_t dropentry_lock; /* drop entry handling */
955 spinlock_t droppacket_lock; /* drop packet handling */
956 spinlock_t securetcp_lock; /* state and timeout tables */
957
958 /* sys-ctl struct */
959 struct ctl_table_header *sysctl_hdr;
960 struct ctl_table *sysctl_tbl;
961 #endif
962
963 /* sysctl variables */
964 int sysctl_amemthresh;
965 int sysctl_am_droprate;
966 int sysctl_drop_entry;
967 int sysctl_drop_packet;
968 int sysctl_secure_tcp;
969 #ifdef CONFIG_IP_VS_NFCT
970 int sysctl_conntrack;
971 #endif
972 int sysctl_snat_reroute;
973 int sysctl_sync_ver;
974 int sysctl_sync_ports;
975 int sysctl_sync_qlen_max;
976 int sysctl_sync_sock_size;
977 int sysctl_cache_bypass;
978 int sysctl_expire_nodest_conn;
979 int sysctl_expire_quiescent_template;
980 int sysctl_sync_threshold[2];
981 unsigned int sysctl_sync_refresh_period;
982 int sysctl_sync_retries;
983 int sysctl_nat_icmp_send;
984 int sysctl_pmtu_disc;
985 int sysctl_backup_only;
986
987 /* ip_vs_lblc */
988 int sysctl_lblc_expiration;
989 struct ctl_table_header *lblc_ctl_header;
990 struct ctl_table *lblc_ctl_table;
991 /* ip_vs_lblcr */
992 int sysctl_lblcr_expiration;
993 struct ctl_table_header *lblcr_ctl_header;
994 struct ctl_table *lblcr_ctl_table;
995 /* ip_vs_est */
996 struct list_head est_list; /* estimator list */
997 spinlock_t est_lock;
998 struct timer_list est_timer; /* Estimation timer */
999 /* ip_vs_sync */
1000 spinlock_t sync_lock;
1001 struct ipvs_master_sync_state *ms;
1002 spinlock_t sync_buff_lock;
1003 struct task_struct **backup_threads;
1004 int threads_mask;
1005 int send_mesg_maxlen;
1006 int recv_mesg_maxlen;
1007 volatile int sync_state;
1008 volatile int master_syncid;
1009 volatile int backup_syncid;
1010 struct mutex sync_mutex;
1011 /* multicast interface name */
1012 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1013 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1014 /* net name space ptr */
1015 struct net *net; /* Needed by timer routines */
1016 };
1017
1018 #define DEFAULT_SYNC_THRESHOLD 3
1019 #define DEFAULT_SYNC_PERIOD 50
1020 #define DEFAULT_SYNC_VER 1
1021 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1022 #define DEFAULT_SYNC_RETRIES 0
1023 #define IPVS_SYNC_WAKEUP_RATE 8
1024 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1025 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1026 #define IPVS_SYNC_CHECK_PERIOD HZ
1027 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1028 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1029
1030 #ifdef CONFIG_SYSCTL
1031
1032 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1033 {
1034 return ipvs->sysctl_sync_threshold[0];
1035 }
1036
1037 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1038 {
1039 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
1040 }
1041
1042 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1043 {
1044 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
1045 }
1046
1047 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1048 {
1049 return ipvs->sysctl_sync_retries;
1050 }
1051
1052 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1053 {
1054 return ipvs->sysctl_sync_ver;
1055 }
1056
1057 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1058 {
1059 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
1060 }
1061
1062 static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1063 {
1064 return ipvs->sysctl_sync_qlen_max;
1065 }
1066
1067 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1068 {
1069 return ipvs->sysctl_sync_sock_size;
1070 }
1071
1072 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1073 {
1074 return ipvs->sysctl_pmtu_disc;
1075 }
1076
1077 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1078 {
1079 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1080 ipvs->sysctl_backup_only;
1081 }
1082
1083 #else
1084
1085 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1086 {
1087 return DEFAULT_SYNC_THRESHOLD;
1088 }
1089
1090 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1091 {
1092 return DEFAULT_SYNC_PERIOD;
1093 }
1094
1095 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1096 {
1097 return DEFAULT_SYNC_REFRESH_PERIOD;
1098 }
1099
1100 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1101 {
1102 return DEFAULT_SYNC_RETRIES & 3;
1103 }
1104
1105 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1106 {
1107 return DEFAULT_SYNC_VER;
1108 }
1109
1110 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1111 {
1112 return 1;
1113 }
1114
1115 static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1116 {
1117 return IPVS_SYNC_QLEN_MAX;
1118 }
1119
1120 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1121 {
1122 return 0;
1123 }
1124
1125 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1126 {
1127 return 1;
1128 }
1129
1130 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1131 {
1132 return 0;
1133 }
1134
1135 #endif
1136
1137 /*
1138 * IPVS core functions
1139 * (from ip_vs_core.c)
1140 */
1141 extern const char *ip_vs_proto_name(unsigned int proto);
1142 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
1143 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1144
1145 #define IP_VS_APP_TYPE_FTP 1
1146
1147 /*
1148 * ip_vs_conn handling functions
1149 * (from ip_vs_conn.c)
1150 */
1151
1152 enum {
1153 IP_VS_DIR_INPUT = 0,
1154 IP_VS_DIR_OUTPUT,
1155 IP_VS_DIR_INPUT_ONLY,
1156 IP_VS_DIR_LAST,
1157 };
1158
1159 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1160 const union nf_inet_addr *caddr,
1161 __be16 cport,
1162 const union nf_inet_addr *vaddr,
1163 __be16 vport,
1164 struct ip_vs_conn_param *p)
1165 {
1166 p->net = net;
1167 p->af = af;
1168 p->protocol = protocol;
1169 p->caddr = caddr;
1170 p->cport = cport;
1171 p->vaddr = vaddr;
1172 p->vport = vport;
1173 p->pe = NULL;
1174 p->pe_data = NULL;
1175 }
1176
1177 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1178 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1179
1180 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1181 const struct ip_vs_iphdr *iph,
1182 int inverse);
1183
1184 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1185
1186 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1187 const struct ip_vs_iphdr *iph,
1188 int inverse);
1189
1190 /* Get reference to gain full access to conn.
1191 * By default, RCU read-side critical sections have access only to
1192 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1193 */
1194 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1195 {
1196 return atomic_inc_not_zero(&cp->refcnt);
1197 }
1198
1199 /* put back the conn without restarting its timer */
1200 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1201 {
1202 smp_mb__before_atomic_dec();
1203 atomic_dec(&cp->refcnt);
1204 }
1205 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1206 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1207
1208 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1209 const union nf_inet_addr *daddr,
1210 __be16 dport, unsigned int flags,
1211 struct ip_vs_dest *dest, __u32 fwmark);
1212 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1213
1214 extern const char * ip_vs_state_name(__u16 proto, int state);
1215
1216 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1217 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1218 extern void ip_vs_random_dropentry(struct net *net);
1219 extern int ip_vs_conn_init(void);
1220 extern void ip_vs_conn_cleanup(void);
1221
1222 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1223 {
1224 struct ip_vs_conn *ctl_cp = cp->control;
1225 if (!ctl_cp) {
1226 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1227 "%s:%d to %s:%d\n",
1228 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1229 ntohs(cp->cport),
1230 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1231 ntohs(cp->vport));
1232
1233 return;
1234 }
1235
1236 IP_VS_DBG_BUF(7, "DELeting control for: "
1237 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1238 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1239 ntohs(cp->cport),
1240 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1241 ntohs(ctl_cp->cport));
1242
1243 cp->control = NULL;
1244 if (atomic_read(&ctl_cp->n_control) == 0) {
1245 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1246 "%s:%d to %s:%d\n",
1247 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1248 ntohs(cp->cport),
1249 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1250 ntohs(cp->vport));
1251
1252 return;
1253 }
1254 atomic_dec(&ctl_cp->n_control);
1255 }
1256
1257 static inline void
1258 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1259 {
1260 if (cp->control) {
1261 IP_VS_ERR_BUF("request control ADD for already controlled: "
1262 "%s:%d to %s:%d\n",
1263 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1264 ntohs(cp->cport),
1265 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1266 ntohs(cp->vport));
1267
1268 ip_vs_control_del(cp);
1269 }
1270
1271 IP_VS_DBG_BUF(7, "ADDing control for: "
1272 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1273 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1274 ntohs(cp->cport),
1275 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1276 ntohs(ctl_cp->cport));
1277
1278 cp->control = ctl_cp;
1279 atomic_inc(&ctl_cp->n_control);
1280 }
1281
1282 /*
1283 * IPVS netns init & cleanup functions
1284 */
1285 extern int ip_vs_estimator_net_init(struct net *net);
1286 extern int ip_vs_control_net_init(struct net *net);
1287 extern int ip_vs_protocol_net_init(struct net *net);
1288 extern int ip_vs_app_net_init(struct net *net);
1289 extern int ip_vs_conn_net_init(struct net *net);
1290 extern int ip_vs_sync_net_init(struct net *net);
1291 extern void ip_vs_conn_net_cleanup(struct net *net);
1292 extern void ip_vs_app_net_cleanup(struct net *net);
1293 extern void ip_vs_protocol_net_cleanup(struct net *net);
1294 extern void ip_vs_control_net_cleanup(struct net *net);
1295 extern void ip_vs_estimator_net_cleanup(struct net *net);
1296 extern void ip_vs_sync_net_cleanup(struct net *net);
1297 extern void ip_vs_service_net_cleanup(struct net *net);
1298
1299 /*
1300 * IPVS application functions
1301 * (from ip_vs_app.c)
1302 */
1303 #define IP_VS_APP_MAX_PORTS 8
1304 extern struct ip_vs_app *register_ip_vs_app(struct net *net,
1305 struct ip_vs_app *app);
1306 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1307 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1308 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1309 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1310 __u16 proto, __u16 port);
1311 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1312 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1313
1314 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1315 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1316
1317 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1318 void ip_vs_unbind_pe(struct ip_vs_service *svc);
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 int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1366 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1367 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1368 extern struct ip_vs_conn *
1369 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1370 struct ip_vs_proto_data *pd, int *ignored,
1371 struct ip_vs_iphdr *iph);
1372 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1373 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1374
1375 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1376
1377
1378 /*
1379 * IPVS control data and functions (from ip_vs_ctl.c)
1380 */
1381 extern struct ip_vs_stats ip_vs_stats;
1382 extern int sysctl_ip_vs_sync_ver;
1383
1384 extern struct ip_vs_service *
1385 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1386 const union nf_inet_addr *vaddr, __be16 vport);
1387
1388 static inline void ip_vs_service_put(struct ip_vs_service *svc)
1389 {
1390 atomic_dec(&svc->usecnt);
1391 }
1392
1393 extern bool
1394 ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
1395 const union nf_inet_addr *daddr, __be16 dport);
1396
1397 extern int ip_vs_use_count_inc(void);
1398 extern void ip_vs_use_count_dec(void);
1399 extern int ip_vs_register_nl_ioctl(void);
1400 extern void ip_vs_unregister_nl_ioctl(void);
1401 extern int ip_vs_control_init(void);
1402 extern void ip_vs_control_cleanup(void);
1403 extern struct ip_vs_dest *
1404 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1405 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1406 __u16 protocol, __u32 fwmark, __u32 flags);
1407 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1408
1409
1410 /*
1411 * IPVS sync daemon data and function prototypes
1412 * (from ip_vs_sync.c)
1413 */
1414 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1415 __u8 syncid);
1416 extern int stop_sync_thread(struct net *net, int state);
1417 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1418
1419
1420 /*
1421 * IPVS rate estimator prototypes (from ip_vs_est.c)
1422 */
1423 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1424 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1425 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1426 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1427 struct ip_vs_stats *stats);
1428
1429 /*
1430 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1431 */
1432 extern int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1433 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1434 extern int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1435 struct ip_vs_protocol *pp,
1436 struct ip_vs_iphdr *iph);
1437 extern int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1438 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1439 extern int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1440 struct ip_vs_protocol *pp,
1441 struct ip_vs_iphdr *iph);
1442 extern int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1443 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1444 extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1445 struct ip_vs_protocol *pp, int offset,
1446 unsigned int hooknum, struct ip_vs_iphdr *iph);
1447 extern void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1448
1449 #ifdef CONFIG_IP_VS_IPV6
1450 extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1451 struct ip_vs_protocol *pp,
1452 struct ip_vs_iphdr *iph);
1453 extern int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1454 struct ip_vs_protocol *pp,
1455 struct ip_vs_iphdr *iph);
1456 extern int ip_vs_tunnel_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_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1460 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1461 extern int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1462 struct ip_vs_protocol *pp, int offset,
1463 unsigned int hooknum, struct ip_vs_iphdr *iph);
1464 #endif
1465
1466 #ifdef CONFIG_SYSCTL
1467 /*
1468 * This is a simple mechanism to ignore packets when
1469 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1470 * we start to drop 1/rate of the packets
1471 */
1472
1473 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1474 {
1475 if (!ipvs->drop_rate)
1476 return 0;
1477 if (--ipvs->drop_counter > 0)
1478 return 0;
1479 ipvs->drop_counter = ipvs->drop_rate;
1480 return 1;
1481 }
1482 #else
1483 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1484 #endif
1485
1486 /*
1487 * ip_vs_fwd_tag returns the forwarding tag of the connection
1488 */
1489 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1490
1491 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1492 {
1493 char fwd;
1494
1495 switch (IP_VS_FWD_METHOD(cp)) {
1496 case IP_VS_CONN_F_MASQ:
1497 fwd = 'M'; break;
1498 case IP_VS_CONN_F_LOCALNODE:
1499 fwd = 'L'; break;
1500 case IP_VS_CONN_F_TUNNEL:
1501 fwd = 'T'; break;
1502 case IP_VS_CONN_F_DROUTE:
1503 fwd = 'R'; break;
1504 case IP_VS_CONN_F_BYPASS:
1505 fwd = 'B'; break;
1506 default:
1507 fwd = '?'; break;
1508 }
1509 return fwd;
1510 }
1511
1512 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1513 struct ip_vs_conn *cp, int dir);
1514
1515 #ifdef CONFIG_IP_VS_IPV6
1516 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1517 struct ip_vs_conn *cp, int dir);
1518 #endif
1519
1520 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1521
1522 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1523 {
1524 __be32 diff[2] = { ~old, new };
1525
1526 return csum_partial(diff, sizeof(diff), oldsum);
1527 }
1528
1529 #ifdef CONFIG_IP_VS_IPV6
1530 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1531 __wsum oldsum)
1532 {
1533 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1534 new[3], new[2], new[1], new[0] };
1535
1536 return csum_partial(diff, sizeof(diff), oldsum);
1537 }
1538 #endif
1539
1540 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1541 {
1542 __be16 diff[2] = { ~old, new };
1543
1544 return csum_partial(diff, sizeof(diff), oldsum);
1545 }
1546
1547 /*
1548 * Forget current conntrack (unconfirmed) and attach notrack entry
1549 */
1550 static inline void ip_vs_notrack(struct sk_buff *skb)
1551 {
1552 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1553 enum ip_conntrack_info ctinfo;
1554 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1555
1556 if (!ct || !nf_ct_is_untracked(ct)) {
1557 nf_conntrack_put(skb->nfct);
1558 skb->nfct = &nf_ct_untracked_get()->ct_general;
1559 skb->nfctinfo = IP_CT_NEW;
1560 nf_conntrack_get(skb->nfct);
1561 }
1562 #endif
1563 }
1564
1565 #ifdef CONFIG_IP_VS_NFCT
1566 /*
1567 * Netfilter connection tracking
1568 * (from ip_vs_nfct.c)
1569 */
1570 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1571 {
1572 #ifdef CONFIG_SYSCTL
1573 return ipvs->sysctl_conntrack;
1574 #else
1575 return 0;
1576 #endif
1577 }
1578
1579 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1580 int outin);
1581 extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1582 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1583 struct ip_vs_conn *cp, u_int8_t proto,
1584 const __be16 port, int from_rs);
1585 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1586
1587 #else
1588
1589 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1590 {
1591 return 0;
1592 }
1593
1594 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1595 struct ip_vs_conn *cp, int outin)
1596 {
1597 }
1598
1599 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1600 {
1601 return NF_ACCEPT;
1602 }
1603
1604 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1605 {
1606 }
1607 /* CONFIG_IP_VS_NFCT */
1608 #endif
1609
1610 static inline unsigned int
1611 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1612 {
1613 /*
1614 * We think the overhead of processing active connections is 256
1615 * times higher than that of inactive connections in average. (This
1616 * 256 times might not be accurate, we will change it later) We
1617 * use the following formula to estimate the overhead now:
1618 * dest->activeconns*256 + dest->inactconns
1619 */
1620 return (atomic_read(&dest->activeconns) << 8) +
1621 atomic_read(&dest->inactconns);
1622 }
1623
1624 #endif /* _NET_IP_VS_H */
ubuntu-zesty-kernel mirror