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 */
net_ipvs(struct net * net)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 */
skb_net(const struct sk_buff * skb)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
skb_sknet(const struct sk_buff * skb)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 */
seq_file_single_net(struct seq_file * seq)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)
skb_nfct_reasm(const struct sk_buff * skb)122 static inline struct sk_buff *skb_nfct_reasm(const struct sk_buff *skb)
123 {
124 return skb->nfct_reasm;
125 }
frag_safe_skb_hp(const struct sk_buff * skb,int offset,int len,void * buffer,const struct ip_vs_iphdr * ipvsh)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
skb_nfct_reasm(const struct sk_buff * skb)137 static inline struct sk_buff *skb_nfct_reasm(const struct sk_buff *skb)
138 {
139 return NULL;
140 }
frag_safe_skb_hp(const struct sk_buff * skb,int offset,int len,void * buffer,const struct ip_vs_iphdr * ipvsh)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
ip_vs_fill_ip4hdr(const void * nh,struct ip_vs_iphdr * iphdr)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
ip_vs_fill_iph_skb(int af,const struct sk_buff * skb,struct ip_vs_iphdr * iphdr)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
ip_vs_fill_iph_addr_only(int af,const struct sk_buff * skb,struct ip_vs_iphdr * iphdr)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
ip_vs_addr_copy(int af,union nf_inet_addr * dst,const union nf_inet_addr * src)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
ip_vs_addr_set(int af,union nf_inet_addr * dst,const union nf_inet_addr * src)236 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
237 const union nf_inet_addr *src)
238 {
239 #ifdef CONFIG_IP_VS_IPV6
240 if (af == AF_INET6) {
241 dst->in6 = src->in6;
242 return;
243 }
244 #endif
245 dst->ip = src->ip;
246 dst->all[1] = 0;
247 dst->all[2] = 0;
248 dst->all[3] = 0;
249 }
250
ip_vs_addr_equal(int af,const union nf_inet_addr * a,const union nf_inet_addr * b)251 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
252 const union nf_inet_addr *b)
253 {
254 #ifdef CONFIG_IP_VS_IPV6
255 if (af == AF_INET6)
256 return ipv6_addr_equal(&a->in6, &b->in6);
257 #endif
258 return a->ip == b->ip;
259 }
260
261 #ifdef CONFIG_IP_VS_DEBUG
262 #include <linux/net.h>
263
264 extern int ip_vs_get_debug_level(void);
265
ip_vs_dbg_addr(int af,char * buf,size_t buf_len,const union nf_inet_addr * addr,int * idx)266 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
267 const union nf_inet_addr *addr,
268 int *idx)
269 {
270 int len;
271 #ifdef CONFIG_IP_VS_IPV6
272 if (af == AF_INET6)
273 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
274 &addr->in6) + 1;
275 else
276 #endif
277 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
278 &addr->ip) + 1;
279
280 *idx += len;
281 BUG_ON(*idx > buf_len + 1);
282 return &buf[*idx - len];
283 }
284
285 #define IP_VS_DBG_BUF(level, msg, ...) \
286 do { \
287 char ip_vs_dbg_buf[160]; \
288 int ip_vs_dbg_idx = 0; \
289 if (level <= ip_vs_get_debug_level()) \
290 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
291 } while (0)
292 #define IP_VS_ERR_BUF(msg...) \
293 do { \
294 char ip_vs_dbg_buf[160]; \
295 int ip_vs_dbg_idx = 0; \
296 pr_err(msg); \
297 } while (0)
298
299 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
300 #define IP_VS_DBG_ADDR(af, addr) \
301 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
302 sizeof(ip_vs_dbg_buf), addr, \
303 &ip_vs_dbg_idx)
304
305 #define IP_VS_DBG(level, msg, ...) \
306 do { \
307 if (level <= ip_vs_get_debug_level()) \
308 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
309 } while (0)
310 #define IP_VS_DBG_RL(msg, ...) \
311 do { \
312 if (net_ratelimit()) \
313 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
314 } while (0)
315 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
316 do { \
317 if (level <= ip_vs_get_debug_level()) \
318 pp->debug_packet(af, pp, skb, ofs, msg); \
319 } while (0)
320 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
321 do { \
322 if (level <= ip_vs_get_debug_level() && \
323 net_ratelimit()) \
324 pp->debug_packet(af, pp, skb, ofs, msg); \
325 } while (0)
326 #else /* NO DEBUGGING at ALL */
327 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
328 #define IP_VS_ERR_BUF(msg...) do {} while (0)
329 #define IP_VS_DBG(level, msg...) do {} while (0)
330 #define IP_VS_DBG_RL(msg...) do {} while (0)
331 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
332 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
333 #endif
334
335 #define IP_VS_BUG() BUG()
336 #define IP_VS_ERR_RL(msg, ...) \
337 do { \
338 if (net_ratelimit()) \
339 pr_err(msg, ##__VA_ARGS__); \
340 } while (0)
341
342 #ifdef CONFIG_IP_VS_DEBUG
343 #define EnterFunction(level) \
344 do { \
345 if (level <= ip_vs_get_debug_level()) \
346 printk(KERN_DEBUG \
347 pr_fmt("Enter: %s, %s line %i\n"), \
348 __func__, __FILE__, __LINE__); \
349 } while (0)
350 #define LeaveFunction(level) \
351 do { \
352 if (level <= ip_vs_get_debug_level()) \
353 printk(KERN_DEBUG \
354 pr_fmt("Leave: %s, %s line %i\n"), \
355 __func__, __FILE__, __LINE__); \
356 } while (0)
357 #else
358 #define EnterFunction(level) do {} while (0)
359 #define LeaveFunction(level) do {} while (0)
360 #endif
361
362
363 /*
364 * The port number of FTP service (in network order).
365 */
366 #define FTPPORT cpu_to_be16(21)
367 #define FTPDATA cpu_to_be16(20)
368
369 /*
370 * TCP State Values
371 */
372 enum {
373 IP_VS_TCP_S_NONE = 0,
374 IP_VS_TCP_S_ESTABLISHED,
375 IP_VS_TCP_S_SYN_SENT,
376 IP_VS_TCP_S_SYN_RECV,
377 IP_VS_TCP_S_FIN_WAIT,
378 IP_VS_TCP_S_TIME_WAIT,
379 IP_VS_TCP_S_CLOSE,
380 IP_VS_TCP_S_CLOSE_WAIT,
381 IP_VS_TCP_S_LAST_ACK,
382 IP_VS_TCP_S_LISTEN,
383 IP_VS_TCP_S_SYNACK,
384 IP_VS_TCP_S_LAST
385 };
386
387 /*
388 * UDP State Values
389 */
390 enum {
391 IP_VS_UDP_S_NORMAL,
392 IP_VS_UDP_S_LAST,
393 };
394
395 /*
396 * ICMP State Values
397 */
398 enum {
399 IP_VS_ICMP_S_NORMAL,
400 IP_VS_ICMP_S_LAST,
401 };
402
403 /*
404 * SCTP State Values
405 */
406 enum ip_vs_sctp_states {
407 IP_VS_SCTP_S_NONE,
408 IP_VS_SCTP_S_INIT_CLI,
409 IP_VS_SCTP_S_INIT_SER,
410 IP_VS_SCTP_S_INIT_ACK_CLI,
411 IP_VS_SCTP_S_INIT_ACK_SER,
412 IP_VS_SCTP_S_ECHO_CLI,
413 IP_VS_SCTP_S_ECHO_SER,
414 IP_VS_SCTP_S_ESTABLISHED,
415 IP_VS_SCTP_S_SHUT_CLI,
416 IP_VS_SCTP_S_SHUT_SER,
417 IP_VS_SCTP_S_SHUT_ACK_CLI,
418 IP_VS_SCTP_S_SHUT_ACK_SER,
419 IP_VS_SCTP_S_CLOSED,
420 IP_VS_SCTP_S_LAST
421 };
422
423 /*
424 * Delta sequence info structure
425 * Each ip_vs_conn has 2 (output AND input seq. changes).
426 * Only used in the VS/NAT.
427 */
428 struct ip_vs_seq {
429 __u32 init_seq; /* Add delta from this seq */
430 __u32 delta; /* Delta in sequence numbers */
431 __u32 previous_delta; /* Delta in sequence numbers
432 before last resized pkt */
433 };
434
435 /*
436 * counters per cpu
437 */
438 struct ip_vs_counters {
439 __u32 conns; /* connections scheduled */
440 __u32 inpkts; /* incoming packets */
441 __u32 outpkts; /* outgoing packets */
442 __u64 inbytes; /* incoming bytes */
443 __u64 outbytes; /* outgoing bytes */
444 };
445 /*
446 * Stats per cpu
447 */
448 struct ip_vs_cpu_stats {
449 struct ip_vs_counters ustats;
450 struct u64_stats_sync syncp;
451 };
452
453 /*
454 * IPVS statistics objects
455 */
456 struct ip_vs_estimator {
457 struct list_head list;
458
459 u64 last_inbytes;
460 u64 last_outbytes;
461 u32 last_conns;
462 u32 last_inpkts;
463 u32 last_outpkts;
464
465 u32 cps;
466 u32 inpps;
467 u32 outpps;
468 u32 inbps;
469 u32 outbps;
470 };
471
472 struct ip_vs_stats {
473 struct ip_vs_stats_user ustats; /* statistics */
474 struct ip_vs_estimator est; /* estimator */
475 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
476 spinlock_t lock; /* spin lock */
477 struct ip_vs_stats_user ustats0; /* reset values */
478 };
479
480 struct dst_entry;
481 struct iphdr;
482 struct ip_vs_conn;
483 struct ip_vs_app;
484 struct sk_buff;
485 struct ip_vs_proto_data;
486
487 struct ip_vs_protocol {
488 struct ip_vs_protocol *next;
489 char *name;
490 u16 protocol;
491 u16 num_states;
492 int dont_defrag;
493
494 void (*init)(struct ip_vs_protocol *pp);
495
496 void (*exit)(struct ip_vs_protocol *pp);
497
498 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
499
500 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
501
502 int (*conn_schedule)(int af, struct sk_buff *skb,
503 struct ip_vs_proto_data *pd,
504 int *verdict, struct ip_vs_conn **cpp,
505 struct ip_vs_iphdr *iph);
506
507 struct ip_vs_conn *
508 (*conn_in_get)(int af,
509 const struct sk_buff *skb,
510 const struct ip_vs_iphdr *iph,
511 int inverse);
512
513 struct ip_vs_conn *
514 (*conn_out_get)(int af,
515 const struct sk_buff *skb,
516 const struct ip_vs_iphdr *iph,
517 int inverse);
518
519 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
520 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
521
522 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
523 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
524
525 int (*csum_check)(int af, struct sk_buff *skb,
526 struct ip_vs_protocol *pp);
527
528 const char *(*state_name)(int state);
529
530 void (*state_transition)(struct ip_vs_conn *cp, int direction,
531 const struct sk_buff *skb,
532 struct ip_vs_proto_data *pd);
533
534 int (*register_app)(struct net *net, struct ip_vs_app *inc);
535
536 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
537
538 int (*app_conn_bind)(struct ip_vs_conn *cp);
539
540 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
541 const struct sk_buff *skb,
542 int offset,
543 const char *msg);
544
545 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
546 };
547
548 /*
549 * protocol data per netns
550 */
551 struct ip_vs_proto_data {
552 struct ip_vs_proto_data *next;
553 struct ip_vs_protocol *pp;
554 int *timeout_table; /* protocol timeout table */
555 atomic_t appcnt; /* counter of proto app incs. */
556 struct tcp_states_t *tcp_state_table;
557 };
558
559 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
560 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
561 unsigned short proto);
562
563 struct ip_vs_conn_param {
564 struct net *net;
565 const union nf_inet_addr *caddr;
566 const union nf_inet_addr *vaddr;
567 __be16 cport;
568 __be16 vport;
569 __u16 protocol;
570 u16 af;
571
572 const struct ip_vs_pe *pe;
573 char *pe_data;
574 __u8 pe_data_len;
575 };
576
577 /*
578 * IP_VS structure allocated for each dynamically scheduled connection
579 */
580 struct ip_vs_conn {
581 struct hlist_node c_list; /* hashed list heads */
582 /* Protocol, addresses and port numbers */
583 __be16 cport;
584 __be16 dport;
585 __be16 vport;
586 u16 af; /* address family */
587 union nf_inet_addr caddr; /* client address */
588 union nf_inet_addr vaddr; /* virtual address */
589 union nf_inet_addr daddr; /* destination address */
590 volatile __u32 flags; /* status flags */
591 __u16 protocol; /* Which protocol (TCP/UDP) */
592 #ifdef CONFIG_NET_NS
593 struct net *net; /* Name space */
594 #endif
595
596 /* counter and timer */
597 atomic_t refcnt; /* reference count */
598 struct timer_list timer; /* Expiration timer */
599 volatile unsigned long timeout; /* timeout */
600
601 /* Flags and state transition */
602 spinlock_t lock; /* lock for state transition */
603 volatile __u16 state; /* state info */
604 volatile __u16 old_state; /* old state, to be used for
605 * state transition triggerd
606 * synchronization
607 */
608 __u32 fwmark; /* Fire wall mark from skb */
609 unsigned long sync_endtime; /* jiffies + sent_retries */
610
611 /* Control members */
612 struct ip_vs_conn *control; /* Master control connection */
613 atomic_t n_control; /* Number of controlled ones */
614 struct ip_vs_dest *dest; /* real server */
615 atomic_t in_pkts; /* incoming packet counter */
616
617 /* packet transmitter for different forwarding methods. If it
618 mangles the packet, it must return NF_DROP or better NF_STOLEN,
619 otherwise this must be changed to a sk_buff **.
620 NF_ACCEPT can be returned when destination is local.
621 */
622 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
623 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
624
625 /* Note: we can group the following members into a structure,
626 in order to save more space, and the following members are
627 only used in VS/NAT anyway */
628 struct ip_vs_app *app; /* bound ip_vs_app object */
629 void *app_data; /* Application private data */
630 struct ip_vs_seq in_seq; /* incoming seq. struct */
631 struct ip_vs_seq out_seq; /* outgoing seq. struct */
632
633 const struct ip_vs_pe *pe;
634 char *pe_data;
635 __u8 pe_data_len;
636
637 struct rcu_head rcu_head;
638 };
639
640 /*
641 * To save some memory in conn table when name space is disabled.
642 */
ip_vs_conn_net(const struct ip_vs_conn * cp)643 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
644 {
645 #ifdef CONFIG_NET_NS
646 return cp->net;
647 #else
648 return &init_net;
649 #endif
650 }
ip_vs_conn_net_set(struct ip_vs_conn * cp,struct net * net)651 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
652 {
653 #ifdef CONFIG_NET_NS
654 cp->net = net;
655 #endif
656 }
657
ip_vs_conn_net_eq(const struct ip_vs_conn * cp,struct net * net)658 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
659 struct net *net)
660 {
661 #ifdef CONFIG_NET_NS
662 return cp->net == net;
663 #else
664 return 1;
665 #endif
666 }
667
668 /*
669 * Extended internal versions of struct ip_vs_service_user and
670 * ip_vs_dest_user for IPv6 support.
671 *
672 * We need these to conveniently pass around service and destination
673 * options, but unfortunately, we also need to keep the old definitions to
674 * maintain userspace backwards compatibility for the setsockopt interface.
675 */
676 struct ip_vs_service_user_kern {
677 /* virtual service addresses */
678 u16 af;
679 u16 protocol;
680 union nf_inet_addr addr; /* virtual ip address */
681 __be16 port;
682 u32 fwmark; /* firwall mark of service */
683
684 /* virtual service options */
685 char *sched_name;
686 char *pe_name;
687 unsigned int flags; /* virtual service flags */
688 unsigned int timeout; /* persistent timeout in sec */
689 __be32 netmask; /* persistent netmask or plen */
690 };
691
692
693 struct ip_vs_dest_user_kern {
694 /* destination server address */
695 union nf_inet_addr addr;
696 __be16 port;
697
698 /* real server options */
699 unsigned int conn_flags; /* connection flags */
700 int weight; /* destination weight */
701
702 /* thresholds for active connections */
703 u32 u_threshold; /* upper threshold */
704 u32 l_threshold; /* lower threshold */
705 };
706
707
708 /*
709 * The information about the virtual service offered to the net
710 * and the forwarding entries
711 */
712 struct ip_vs_service {
713 struct hlist_node s_list; /* for normal service table */
714 struct hlist_node f_list; /* for fwmark-based service table */
715 atomic_t refcnt; /* reference counter */
716
717 u16 af; /* address family */
718 __u16 protocol; /* which protocol (TCP/UDP) */
719 union nf_inet_addr addr; /* IP address for virtual service */
720 __be16 port; /* port number for the service */
721 __u32 fwmark; /* firewall mark of the service */
722 unsigned int flags; /* service status flags */
723 unsigned int timeout; /* persistent timeout in ticks */
724 __be32 netmask; /* grouping granularity, mask/plen */
725 struct net *net;
726
727 struct list_head destinations; /* real server d-linked list */
728 __u32 num_dests; /* number of servers */
729 struct ip_vs_stats stats; /* statistics for the service */
730
731 /* for scheduling */
732 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
733 spinlock_t sched_lock; /* lock sched_data */
734 void *sched_data; /* scheduler application data */
735
736 /* alternate persistence engine */
737 struct ip_vs_pe __rcu *pe;
738
739 struct rcu_head rcu_head;
740 };
741
742 /* Information for cached dst */
743 struct ip_vs_dest_dst {
744 struct dst_entry *dst_cache; /* destination cache entry */
745 u32 dst_cookie;
746 union nf_inet_addr dst_saddr;
747 struct rcu_head rcu_head;
748 };
749
750 /* In grace period after removing */
751 #define IP_VS_DEST_STATE_REMOVING 0x01
752 /*
753 * The real server destination forwarding entry
754 * with ip address, port number, and so on.
755 */
756 struct ip_vs_dest {
757 struct list_head n_list; /* for the dests in the service */
758 struct hlist_node d_list; /* for table with all the dests */
759
760 u16 af; /* address family */
761 __be16 port; /* port number of the server */
762 union nf_inet_addr addr; /* IP address of the server */
763 volatile unsigned int flags; /* dest status flags */
764 atomic_t conn_flags; /* flags to copy to conn */
765 atomic_t weight; /* server weight */
766
767 atomic_t refcnt; /* reference counter */
768 struct ip_vs_stats stats; /* statistics */
769 unsigned long state; /* state flags */
770
771 /* connection counters and thresholds */
772 atomic_t activeconns; /* active connections */
773 atomic_t inactconns; /* inactive connections */
774 atomic_t persistconns; /* persistent connections */
775 __u32 u_threshold; /* upper threshold */
776 __u32 l_threshold; /* lower threshold */
777
778 /* for destination cache */
779 spinlock_t dst_lock; /* lock of dst_cache */
780 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
781
782 /* for virtual service */
783 struct ip_vs_service *svc; /* service it belongs to */
784 __u16 protocol; /* which protocol (TCP/UDP) */
785 __be16 vport; /* virtual port number */
786 union nf_inet_addr vaddr; /* virtual IP address */
787 __u32 vfwmark; /* firewall mark of service */
788
789 struct list_head t_list; /* in dest_trash */
790 struct rcu_head rcu_head;
791 unsigned int in_rs_table:1; /* we are in rs_table */
792 };
793
794
795 /*
796 * The scheduler object
797 */
798 struct ip_vs_scheduler {
799 struct list_head n_list; /* d-linked list head */
800 char *name; /* scheduler name */
801 atomic_t refcnt; /* reference counter */
802 struct module *module; /* THIS_MODULE/NULL */
803
804 /* scheduler initializing service */
805 int (*init_service)(struct ip_vs_service *svc);
806 /* scheduling service finish */
807 void (*done_service)(struct ip_vs_service *svc);
808 /* dest is linked */
809 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
810 /* dest is unlinked */
811 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
812 /* dest is updated */
813 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
814
815 /* selecting a server from the given service */
816 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
817 const struct sk_buff *skb);
818 };
819
820 /* The persistence engine object */
821 struct ip_vs_pe {
822 struct list_head n_list; /* d-linked list head */
823 char *name; /* scheduler name */
824 atomic_t refcnt; /* reference counter */
825 struct module *module; /* THIS_MODULE/NULL */
826
827 /* get the connection template, if any */
828 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
829 bool (*ct_match)(const struct ip_vs_conn_param *p,
830 struct ip_vs_conn *ct);
831 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
832 bool inverse);
833 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
834 };
835
836 /*
837 * The application module object (a.k.a. app incarnation)
838 */
839 struct ip_vs_app {
840 struct list_head a_list; /* member in app list */
841 int type; /* IP_VS_APP_TYPE_xxx */
842 char *name; /* application module name */
843 __u16 protocol;
844 struct module *module; /* THIS_MODULE/NULL */
845 struct list_head incs_list; /* list of incarnations */
846
847 /* members for application incarnations */
848 struct list_head p_list; /* member in proto app list */
849 struct ip_vs_app *app; /* its real application */
850 __be16 port; /* port number in net order */
851 atomic_t usecnt; /* usage counter */
852 struct rcu_head rcu_head;
853
854 /*
855 * output hook: Process packet in inout direction, diff set for TCP.
856 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
857 * 2=Mangled but checksum was not updated
858 */
859 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
860 struct sk_buff *, int *diff);
861
862 /*
863 * input hook: Process packet in outin direction, diff set for TCP.
864 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
865 * 2=Mangled but checksum was not updated
866 */
867 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
868 struct sk_buff *, int *diff);
869
870 /* ip_vs_app initializer */
871 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
872
873 /* ip_vs_app finish */
874 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
875
876
877 /* not used now */
878 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
879 struct ip_vs_protocol *);
880
881 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
882
883 int * timeout_table;
884 int * timeouts;
885 int timeouts_size;
886
887 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
888 int *verdict, struct ip_vs_conn **cpp);
889
890 struct ip_vs_conn *
891 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
892 const struct iphdr *iph, int inverse);
893
894 struct ip_vs_conn *
895 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
896 const struct iphdr *iph, int inverse);
897
898 int (*state_transition)(struct ip_vs_conn *cp, int direction,
899 const struct sk_buff *skb,
900 struct ip_vs_app *app);
901
902 void (*timeout_change)(struct ip_vs_app *app, int flags);
903 };
904
905 struct ipvs_master_sync_state {
906 struct list_head sync_queue;
907 struct ip_vs_sync_buff *sync_buff;
908 int sync_queue_len;
909 unsigned int sync_queue_delay;
910 struct task_struct *master_thread;
911 struct delayed_work master_wakeup_work;
912 struct netns_ipvs *ipvs;
913 };
914
915 /* How much time to keep dests in trash */
916 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
917
918 /* IPVS in network namespace */
919 struct netns_ipvs {
920 int gen; /* Generation */
921 int enable; /* enable like nf_hooks do */
922 /*
923 * Hash table: for real service lookups
924 */
925 #define IP_VS_RTAB_BITS 4
926 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
927 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
928
929 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
930 /* ip_vs_app */
931 struct list_head app_list;
932 /* ip_vs_proto */
933 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
934 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
935 /* ip_vs_proto_tcp */
936 #ifdef CONFIG_IP_VS_PROTO_TCP
937 #define TCP_APP_TAB_BITS 4
938 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
939 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
940 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
941 #endif
942 /* ip_vs_proto_udp */
943 #ifdef CONFIG_IP_VS_PROTO_UDP
944 #define UDP_APP_TAB_BITS 4
945 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
946 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
947 struct list_head udp_apps[UDP_APP_TAB_SIZE];
948 #endif
949 /* ip_vs_proto_sctp */
950 #ifdef CONFIG_IP_VS_PROTO_SCTP
951 #define SCTP_APP_TAB_BITS 4
952 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
953 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
954 /* Hash table for SCTP application incarnations */
955 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
956 #endif
957 /* ip_vs_conn */
958 atomic_t conn_count; /* connection counter */
959
960 /* ip_vs_ctl */
961 struct ip_vs_stats tot_stats; /* Statistics & est. */
962
963 int num_services; /* no of virtual services */
964
965 /* Trash for destinations */
966 struct list_head dest_trash;
967 spinlock_t dest_trash_lock;
968 struct timer_list dest_trash_timer; /* expiration timer */
969 /* Service counters */
970 atomic_t ftpsvc_counter;
971 atomic_t nullsvc_counter;
972
973 #ifdef CONFIG_SYSCTL
974 /* 1/rate drop and drop-entry variables */
975 struct delayed_work defense_work; /* Work handler */
976 int drop_rate;
977 int drop_counter;
978 atomic_t dropentry;
979 /* locks in ctl.c */
980 spinlock_t dropentry_lock; /* drop entry handling */
981 spinlock_t droppacket_lock; /* drop packet handling */
982 spinlock_t securetcp_lock; /* state and timeout tables */
983
984 /* sys-ctl struct */
985 struct ctl_table_header *sysctl_hdr;
986 struct ctl_table *sysctl_tbl;
987 #endif
988
989 /* sysctl variables */
990 int sysctl_amemthresh;
991 int sysctl_am_droprate;
992 int sysctl_drop_entry;
993 int sysctl_drop_packet;
994 int sysctl_secure_tcp;
995 #ifdef CONFIG_IP_VS_NFCT
996 int sysctl_conntrack;
997 #endif
998 int sysctl_snat_reroute;
999 int sysctl_sync_ver;
1000 int sysctl_sync_ports;
1001 int sysctl_sync_qlen_max;
1002 int sysctl_sync_sock_size;
1003 int sysctl_cache_bypass;
1004 int sysctl_expire_nodest_conn;
1005 int sysctl_expire_quiescent_template;
1006 int sysctl_sync_threshold[2];
1007 unsigned int sysctl_sync_refresh_period;
1008 int sysctl_sync_retries;
1009 int sysctl_nat_icmp_send;
1010 int sysctl_pmtu_disc;
1011 int sysctl_backup_only;
1012
1013 /* ip_vs_lblc */
1014 int sysctl_lblc_expiration;
1015 struct ctl_table_header *lblc_ctl_header;
1016 struct ctl_table *lblc_ctl_table;
1017 /* ip_vs_lblcr */
1018 int sysctl_lblcr_expiration;
1019 struct ctl_table_header *lblcr_ctl_header;
1020 struct ctl_table *lblcr_ctl_table;
1021 /* ip_vs_est */
1022 struct list_head est_list; /* estimator list */
1023 spinlock_t est_lock;
1024 struct timer_list est_timer; /* Estimation timer */
1025 /* ip_vs_sync */
1026 spinlock_t sync_lock;
1027 struct ipvs_master_sync_state *ms;
1028 spinlock_t sync_buff_lock;
1029 struct task_struct **backup_threads;
1030 int threads_mask;
1031 int send_mesg_maxlen;
1032 int recv_mesg_maxlen;
1033 volatile int sync_state;
1034 volatile int master_syncid;
1035 volatile int backup_syncid;
1036 struct mutex sync_mutex;
1037 /* multicast interface name */
1038 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1039 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1040 /* net name space ptr */
1041 struct net *net; /* Needed by timer routines */
1042 };
1043
1044 #define DEFAULT_SYNC_THRESHOLD 3
1045 #define DEFAULT_SYNC_PERIOD 50
1046 #define DEFAULT_SYNC_VER 1
1047 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1048 #define DEFAULT_SYNC_RETRIES 0
1049 #define IPVS_SYNC_WAKEUP_RATE 8
1050 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1051 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1052 #define IPVS_SYNC_CHECK_PERIOD HZ
1053 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1054 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1055
1056 #ifdef CONFIG_SYSCTL
1057
sysctl_sync_threshold(struct netns_ipvs * ipvs)1058 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1059 {
1060 return ipvs->sysctl_sync_threshold[0];
1061 }
1062
sysctl_sync_period(struct netns_ipvs * ipvs)1063 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1064 {
1065 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
1066 }
1067
sysctl_sync_refresh_period(struct netns_ipvs * ipvs)1068 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1069 {
1070 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
1071 }
1072
sysctl_sync_retries(struct netns_ipvs * ipvs)1073 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1074 {
1075 return ipvs->sysctl_sync_retries;
1076 }
1077
sysctl_sync_ver(struct netns_ipvs * ipvs)1078 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1079 {
1080 return ipvs->sysctl_sync_ver;
1081 }
1082
sysctl_sync_ports(struct netns_ipvs * ipvs)1083 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1084 {
1085 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
1086 }
1087
sysctl_sync_qlen_max(struct netns_ipvs * ipvs)1088 static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1089 {
1090 return ipvs->sysctl_sync_qlen_max;
1091 }
1092
sysctl_sync_sock_size(struct netns_ipvs * ipvs)1093 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1094 {
1095 return ipvs->sysctl_sync_sock_size;
1096 }
1097
sysctl_pmtu_disc(struct netns_ipvs * ipvs)1098 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1099 {
1100 return ipvs->sysctl_pmtu_disc;
1101 }
1102
sysctl_backup_only(struct netns_ipvs * ipvs)1103 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1104 {
1105 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1106 ipvs->sysctl_backup_only;
1107 }
1108
1109 #else
1110
sysctl_sync_threshold(struct netns_ipvs * ipvs)1111 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1112 {
1113 return DEFAULT_SYNC_THRESHOLD;
1114 }
1115
sysctl_sync_period(struct netns_ipvs * ipvs)1116 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1117 {
1118 return DEFAULT_SYNC_PERIOD;
1119 }
1120
sysctl_sync_refresh_period(struct netns_ipvs * ipvs)1121 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1122 {
1123 return DEFAULT_SYNC_REFRESH_PERIOD;
1124 }
1125
sysctl_sync_retries(struct netns_ipvs * ipvs)1126 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1127 {
1128 return DEFAULT_SYNC_RETRIES & 3;
1129 }
1130
sysctl_sync_ver(struct netns_ipvs * ipvs)1131 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1132 {
1133 return DEFAULT_SYNC_VER;
1134 }
1135
sysctl_sync_ports(struct netns_ipvs * ipvs)1136 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1137 {
1138 return 1;
1139 }
1140
sysctl_sync_qlen_max(struct netns_ipvs * ipvs)1141 static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1142 {
1143 return IPVS_SYNC_QLEN_MAX;
1144 }
1145
sysctl_sync_sock_size(struct netns_ipvs * ipvs)1146 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1147 {
1148 return 0;
1149 }
1150
sysctl_pmtu_disc(struct netns_ipvs * ipvs)1151 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1152 {
1153 return 1;
1154 }
1155
sysctl_backup_only(struct netns_ipvs * ipvs)1156 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1157 {
1158 return 0;
1159 }
1160
1161 #endif
1162
1163 /*
1164 * IPVS core functions
1165 * (from ip_vs_core.c)
1166 */
1167 extern const char *ip_vs_proto_name(unsigned int proto);
1168 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
1169 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1170
1171 #define IP_VS_APP_TYPE_FTP 1
1172
1173 /*
1174 * ip_vs_conn handling functions
1175 * (from ip_vs_conn.c)
1176 */
1177
1178 enum {
1179 IP_VS_DIR_INPUT = 0,
1180 IP_VS_DIR_OUTPUT,
1181 IP_VS_DIR_INPUT_ONLY,
1182 IP_VS_DIR_LAST,
1183 };
1184
ip_vs_conn_fill_param(struct net * net,int af,int protocol,const union nf_inet_addr * caddr,__be16 cport,const union nf_inet_addr * vaddr,__be16 vport,struct ip_vs_conn_param * p)1185 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1186 const union nf_inet_addr *caddr,
1187 __be16 cport,
1188 const union nf_inet_addr *vaddr,
1189 __be16 vport,
1190 struct ip_vs_conn_param *p)
1191 {
1192 p->net = net;
1193 p->af = af;
1194 p->protocol = protocol;
1195 p->caddr = caddr;
1196 p->cport = cport;
1197 p->vaddr = vaddr;
1198 p->vport = vport;
1199 p->pe = NULL;
1200 p->pe_data = NULL;
1201 }
1202
1203 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1204 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1205
1206 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1207 const struct ip_vs_iphdr *iph,
1208 int inverse);
1209
1210 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1211
1212 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1213 const struct ip_vs_iphdr *iph,
1214 int inverse);
1215
1216 /* Get reference to gain full access to conn.
1217 * By default, RCU read-side critical sections have access only to
1218 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1219 */
__ip_vs_conn_get(struct ip_vs_conn * cp)1220 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1221 {
1222 return atomic_inc_not_zero(&cp->refcnt);
1223 }
1224
1225 /* put back the conn without restarting its timer */
__ip_vs_conn_put(struct ip_vs_conn * cp)1226 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1227 {
1228 smp_mb__before_atomic_dec();
1229 atomic_dec(&cp->refcnt);
1230 }
1231 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1232 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1233
1234 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1235 const union nf_inet_addr *daddr,
1236 __be16 dport, unsigned int flags,
1237 struct ip_vs_dest *dest, __u32 fwmark);
1238 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1239
1240 extern const char * ip_vs_state_name(__u16 proto, int state);
1241
1242 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1243 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1244 extern void ip_vs_random_dropentry(struct net *net);
1245 extern int ip_vs_conn_init(void);
1246 extern void ip_vs_conn_cleanup(void);
1247
ip_vs_control_del(struct ip_vs_conn * cp)1248 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1249 {
1250 struct ip_vs_conn *ctl_cp = cp->control;
1251 if (!ctl_cp) {
1252 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1253 "%s:%d to %s:%d\n",
1254 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1255 ntohs(cp->cport),
1256 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1257 ntohs(cp->vport));
1258
1259 return;
1260 }
1261
1262 IP_VS_DBG_BUF(7, "DELeting control for: "
1263 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1264 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1265 ntohs(cp->cport),
1266 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1267 ntohs(ctl_cp->cport));
1268
1269 cp->control = NULL;
1270 if (atomic_read(&ctl_cp->n_control) == 0) {
1271 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1272 "%s:%d to %s:%d\n",
1273 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1274 ntohs(cp->cport),
1275 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1276 ntohs(cp->vport));
1277
1278 return;
1279 }
1280 atomic_dec(&ctl_cp->n_control);
1281 }
1282
1283 static inline void
ip_vs_control_add(struct ip_vs_conn * cp,struct ip_vs_conn * ctl_cp)1284 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1285 {
1286 if (cp->control) {
1287 IP_VS_ERR_BUF("request control ADD for already controlled: "
1288 "%s:%d to %s:%d\n",
1289 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1290 ntohs(cp->cport),
1291 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1292 ntohs(cp->vport));
1293
1294 ip_vs_control_del(cp);
1295 }
1296
1297 IP_VS_DBG_BUF(7, "ADDing control for: "
1298 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1299 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1300 ntohs(cp->cport),
1301 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1302 ntohs(ctl_cp->cport));
1303
1304 cp->control = ctl_cp;
1305 atomic_inc(&ctl_cp->n_control);
1306 }
1307
1308 /*
1309 * IPVS netns init & cleanup functions
1310 */
1311 extern int ip_vs_estimator_net_init(struct net *net);
1312 extern int ip_vs_control_net_init(struct net *net);
1313 extern int ip_vs_protocol_net_init(struct net *net);
1314 extern int ip_vs_app_net_init(struct net *net);
1315 extern int ip_vs_conn_net_init(struct net *net);
1316 extern int ip_vs_sync_net_init(struct net *net);
1317 extern void ip_vs_conn_net_cleanup(struct net *net);
1318 extern void ip_vs_app_net_cleanup(struct net *net);
1319 extern void ip_vs_protocol_net_cleanup(struct net *net);
1320 extern void ip_vs_control_net_cleanup(struct net *net);
1321 extern void ip_vs_estimator_net_cleanup(struct net *net);
1322 extern void ip_vs_sync_net_cleanup(struct net *net);
1323 extern void ip_vs_service_net_cleanup(struct net *net);
1324
1325 /*
1326 * IPVS application functions
1327 * (from ip_vs_app.c)
1328 */
1329 #define IP_VS_APP_MAX_PORTS 8
1330 extern struct ip_vs_app *register_ip_vs_app(struct net *net,
1331 struct ip_vs_app *app);
1332 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1333 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1334 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1335 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1336 __u16 proto, __u16 port);
1337 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1338 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1339
1340 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1341 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1342
1343 int register_ip_vs_pe(struct ip_vs_pe *pe);
1344 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1345 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1346 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1347
1348 /*
1349 * Use a #define to avoid all of module.h just for these trivial ops
1350 */
1351 #define ip_vs_pe_get(pe) \
1352 if (pe && pe->module) \
1353 __module_get(pe->module);
1354
1355 #define ip_vs_pe_put(pe) \
1356 if (pe && pe->module) \
1357 module_put(pe->module);
1358
1359 /*
1360 * IPVS protocol functions (from ip_vs_proto.c)
1361 */
1362 extern int ip_vs_protocol_init(void);
1363 extern void ip_vs_protocol_cleanup(void);
1364 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1365 extern int *ip_vs_create_timeout_table(int *table, int size);
1366 extern int
1367 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1368 const char *name, int to);
1369 extern void
1370 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1371 const struct sk_buff *skb,
1372 int offset, const char *msg);
1373
1374 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1375 extern struct ip_vs_protocol ip_vs_protocol_udp;
1376 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1377 extern struct ip_vs_protocol ip_vs_protocol_esp;
1378 extern struct ip_vs_protocol ip_vs_protocol_ah;
1379 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1380
1381 /*
1382 * Registering/unregistering scheduler functions
1383 * (from ip_vs_sched.c)
1384 */
1385 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1386 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1387 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1388 struct ip_vs_scheduler *scheduler);
1389 extern void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1390 struct ip_vs_scheduler *sched);
1391 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1392 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1393 extern struct ip_vs_conn *
1394 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1395 struct ip_vs_proto_data *pd, int *ignored,
1396 struct ip_vs_iphdr *iph);
1397 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1398 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1399
1400 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1401
1402
1403 /*
1404 * IPVS control data and functions (from ip_vs_ctl.c)
1405 */
1406 extern struct ip_vs_stats ip_vs_stats;
1407 extern int sysctl_ip_vs_sync_ver;
1408
1409 extern struct ip_vs_service *
1410 ip_vs_service_find(struct net *net, int af, __u32 fwmark, __u16 protocol,
1411 const union nf_inet_addr *vaddr, __be16 vport);
1412
1413 extern bool
1414 ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
1415 const union nf_inet_addr *daddr, __be16 dport);
1416
1417 extern int ip_vs_use_count_inc(void);
1418 extern void ip_vs_use_count_dec(void);
1419 extern int ip_vs_register_nl_ioctl(void);
1420 extern void ip_vs_unregister_nl_ioctl(void);
1421 extern int ip_vs_control_init(void);
1422 extern void ip_vs_control_cleanup(void);
1423 extern struct ip_vs_dest *
1424 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1425 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1426 __u16 protocol, __u32 fwmark, __u32 flags);
1427 extern void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1428
ip_vs_dest_hold(struct ip_vs_dest * dest)1429 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1430 {
1431 atomic_inc(&dest->refcnt);
1432 }
1433
ip_vs_dest_put(struct ip_vs_dest * dest)1434 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1435 {
1436 smp_mb__before_atomic_dec();
1437 atomic_dec(&dest->refcnt);
1438 }
1439
1440 /*
1441 * IPVS sync daemon data and function prototypes
1442 * (from ip_vs_sync.c)
1443 */
1444 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1445 __u8 syncid);
1446 extern int stop_sync_thread(struct net *net, int state);
1447 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1448
1449
1450 /*
1451 * IPVS rate estimator prototypes (from ip_vs_est.c)
1452 */
1453 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1454 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1455 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1456 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1457 struct ip_vs_stats *stats);
1458
1459 /*
1460 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1461 */
1462 extern int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1463 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1464 extern int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1465 struct ip_vs_protocol *pp,
1466 struct ip_vs_iphdr *iph);
1467 extern int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1468 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1469 extern int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1470 struct ip_vs_protocol *pp,
1471 struct ip_vs_iphdr *iph);
1472 extern int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1473 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1474 extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1475 struct ip_vs_protocol *pp, int offset,
1476 unsigned int hooknum, struct ip_vs_iphdr *iph);
1477 extern void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1478
1479 #ifdef CONFIG_IP_VS_IPV6
1480 extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1481 struct ip_vs_protocol *pp,
1482 struct ip_vs_iphdr *iph);
1483 extern int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1484 struct ip_vs_protocol *pp,
1485 struct ip_vs_iphdr *iph);
1486 extern int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1487 struct ip_vs_protocol *pp,
1488 struct ip_vs_iphdr *iph);
1489 extern int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1490 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1491 extern int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1492 struct ip_vs_protocol *pp, int offset,
1493 unsigned int hooknum, struct ip_vs_iphdr *iph);
1494 #endif
1495
1496 #ifdef CONFIG_SYSCTL
1497 /*
1498 * This is a simple mechanism to ignore packets when
1499 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1500 * we start to drop 1/rate of the packets
1501 */
1502
ip_vs_todrop(struct netns_ipvs * ipvs)1503 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1504 {
1505 if (!ipvs->drop_rate)
1506 return 0;
1507 if (--ipvs->drop_counter > 0)
1508 return 0;
1509 ipvs->drop_counter = ipvs->drop_rate;
1510 return 1;
1511 }
1512 #else
ip_vs_todrop(struct netns_ipvs * ipvs)1513 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1514 #endif
1515
1516 /*
1517 * ip_vs_fwd_tag returns the forwarding tag of the connection
1518 */
1519 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1520
ip_vs_fwd_tag(struct ip_vs_conn * cp)1521 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1522 {
1523 char fwd;
1524
1525 switch (IP_VS_FWD_METHOD(cp)) {
1526 case IP_VS_CONN_F_MASQ:
1527 fwd = 'M'; break;
1528 case IP_VS_CONN_F_LOCALNODE:
1529 fwd = 'L'; break;
1530 case IP_VS_CONN_F_TUNNEL:
1531 fwd = 'T'; break;
1532 case IP_VS_CONN_F_DROUTE:
1533 fwd = 'R'; break;
1534 case IP_VS_CONN_F_BYPASS:
1535 fwd = 'B'; break;
1536 default:
1537 fwd = '?'; break;
1538 }
1539 return fwd;
1540 }
1541
1542 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1543 struct ip_vs_conn *cp, int dir);
1544
1545 #ifdef CONFIG_IP_VS_IPV6
1546 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1547 struct ip_vs_conn *cp, int dir);
1548 #endif
1549
1550 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1551
ip_vs_check_diff4(__be32 old,__be32 new,__wsum oldsum)1552 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1553 {
1554 __be32 diff[2] = { ~old, new };
1555
1556 return csum_partial(diff, sizeof(diff), oldsum);
1557 }
1558
1559 #ifdef CONFIG_IP_VS_IPV6
ip_vs_check_diff16(const __be32 * old,const __be32 * new,__wsum oldsum)1560 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1561 __wsum oldsum)
1562 {
1563 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1564 new[3], new[2], new[1], new[0] };
1565
1566 return csum_partial(diff, sizeof(diff), oldsum);
1567 }
1568 #endif
1569
ip_vs_check_diff2(__be16 old,__be16 new,__wsum oldsum)1570 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1571 {
1572 __be16 diff[2] = { ~old, new };
1573
1574 return csum_partial(diff, sizeof(diff), oldsum);
1575 }
1576
1577 /*
1578 * Forget current conntrack (unconfirmed) and attach notrack entry
1579 */
ip_vs_notrack(struct sk_buff * skb)1580 static inline void ip_vs_notrack(struct sk_buff *skb)
1581 {
1582 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1583 enum ip_conntrack_info ctinfo;
1584 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1585
1586 if (!ct || !nf_ct_is_untracked(ct)) {
1587 nf_conntrack_put(skb->nfct);
1588 skb->nfct = &nf_ct_untracked_get()->ct_general;
1589 skb->nfctinfo = IP_CT_NEW;
1590 nf_conntrack_get(skb->nfct);
1591 }
1592 #endif
1593 }
1594
1595 #ifdef CONFIG_IP_VS_NFCT
1596 /*
1597 * Netfilter connection tracking
1598 * (from ip_vs_nfct.c)
1599 */
ip_vs_conntrack_enabled(struct netns_ipvs * ipvs)1600 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1601 {
1602 #ifdef CONFIG_SYSCTL
1603 return ipvs->sysctl_conntrack;
1604 #else
1605 return 0;
1606 #endif
1607 }
1608
1609 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1610 int outin);
1611 extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1612 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1613 struct ip_vs_conn *cp, u_int8_t proto,
1614 const __be16 port, int from_rs);
1615 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1616
1617 #else
1618
ip_vs_conntrack_enabled(struct netns_ipvs * ipvs)1619 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1620 {
1621 return 0;
1622 }
1623
ip_vs_update_conntrack(struct sk_buff * skb,struct ip_vs_conn * cp,int outin)1624 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1625 struct ip_vs_conn *cp, int outin)
1626 {
1627 }
1628
ip_vs_confirm_conntrack(struct sk_buff * skb)1629 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1630 {
1631 return NF_ACCEPT;
1632 }
1633
ip_vs_conn_drop_conntrack(struct ip_vs_conn * cp)1634 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1635 {
1636 }
1637 /* CONFIG_IP_VS_NFCT */
1638 #endif
1639
1640 static inline unsigned int
ip_vs_dest_conn_overhead(struct ip_vs_dest * dest)1641 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1642 {
1643 /*
1644 * We think the overhead of processing active connections is 256
1645 * times higher than that of inactive connections in average. (This
1646 * 256 times might not be accurate, we will change it later) We
1647 * use the following formula to estimate the overhead now:
1648 * dest->activeconns*256 + dest->inactconns
1649 */
1650 return (atomic_read(&dest->activeconns) << 8) +
1651 atomic_read(&dest->inactconns);
1652 }
1653
1654 #endif /* _NET_IP_VS_H */
1655