Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * IP Virtual Server | |
3 | * data structure and functionality definitions | |
4 | */ | |
5 | ||
bc4768eb JV |
6 | #ifndef _NET_IP_VS_H |
7 | #define _NET_IP_VS_H | |
1da177e4 | 8 | |
bc4768eb | 9 | #include <linux/ip_vs.h> /* definitions shared with userland */ |
1da177e4 | 10 | |
bc4768eb | 11 | /* old ipvsadm versions still include this file directly */ |
1da177e4 LT |
12 | #ifdef __KERNEL__ |
13 | ||
bc4768eb JV |
14 | #include <asm/types.h> /* for __uXX types */ |
15 | ||
16 | #include <linux/sysctl.h> /* for ctl_path */ | |
1da177e4 LT |
17 | #include <linux/list.h> /* for struct list_head */ |
18 | #include <linux/spinlock.h> /* for struct rwlock_t */ | |
1da177e4 | 19 | #include <asm/atomic.h> /* for struct atomic_t */ |
1da177e4 | 20 | #include <linux/compiler.h> |
14c85021 | 21 | #include <linux/timer.h> |
1da177e4 | 22 | |
14c85021 | 23 | #include <net/checksum.h> |
e7ade46a | 24 | #include <linux/netfilter.h> /* for union nf_inet_addr */ |
1668e010 | 25 | #include <linux/ip.h> |
e7ade46a JV |
26 | #include <linux/ipv6.h> /* for struct ipv6hdr */ |
27 | #include <net/ipv6.h> /* for ipv6_addr_copy */ | |
1da177e4 | 28 | |
64aae3cb JV |
29 | struct ip_vs_iphdr { |
30 | int len; | |
31 | __u8 protocol; | |
32 | union nf_inet_addr saddr; | |
33 | union nf_inet_addr daddr; | |
34 | }; | |
35 | ||
36 | static inline void | |
37 | ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr) | |
38 | { | |
39 | #ifdef CONFIG_IP_VS_IPV6 | |
40 | if (af == AF_INET6) { | |
41 | const struct ipv6hdr *iph = nh; | |
42 | iphdr->len = sizeof(struct ipv6hdr); | |
43 | iphdr->protocol = iph->nexthdr; | |
44 | ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr); | |
45 | ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr); | |
46 | } else | |
47 | #endif | |
48 | { | |
49 | const struct iphdr *iph = nh; | |
50 | iphdr->len = iph->ihl * 4; | |
51 | iphdr->protocol = iph->protocol; | |
52 | iphdr->saddr.ip = iph->saddr; | |
53 | iphdr->daddr.ip = iph->daddr; | |
54 | } | |
55 | } | |
56 | ||
57 | static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst, | |
58 | const union nf_inet_addr *src) | |
59 | { | |
60 | #ifdef CONFIG_IP_VS_IPV6 | |
61 | if (af == AF_INET6) | |
62 | ipv6_addr_copy(&dst->in6, &src->in6); | |
63 | else | |
64 | #endif | |
65 | dst->ip = src->ip; | |
66 | } | |
67 | ||
68 | static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a, | |
69 | const union nf_inet_addr *b) | |
70 | { | |
71 | #ifdef CONFIG_IP_VS_IPV6 | |
72 | if (af == AF_INET6) | |
73 | return ipv6_addr_equal(&a->in6, &b->in6); | |
74 | #endif | |
75 | return a->ip == b->ip; | |
76 | } | |
77 | ||
1da177e4 | 78 | #ifdef CONFIG_IP_VS_DEBUG |
14c85021 ACM |
79 | #include <linux/net.h> |
80 | ||
1da177e4 | 81 | extern int ip_vs_get_debug_level(void); |
c842a3ad JV |
82 | |
83 | static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len, | |
84 | const union nf_inet_addr *addr, | |
85 | int *idx) | |
86 | { | |
87 | int len; | |
88 | #ifdef CONFIG_IP_VS_IPV6 | |
89 | if (af == AF_INET6) | |
90 | len = snprintf(&buf[*idx], buf_len - *idx, "[" NIP6_FMT "]", | |
91 | NIP6(addr->in6)) + 1; | |
92 | else | |
93 | #endif | |
94 | len = snprintf(&buf[*idx], buf_len - *idx, NIPQUAD_FMT, | |
95 | NIPQUAD(addr->ip)) + 1; | |
96 | ||
97 | *idx += len; | |
98 | BUG_ON(*idx > buf_len + 1); | |
99 | return &buf[*idx - len]; | |
100 | } | |
101 | ||
102 | #define IP_VS_DBG_BUF(level, msg...) \ | |
103 | do { \ | |
104 | char ip_vs_dbg_buf[160]; \ | |
105 | int ip_vs_dbg_idx = 0; \ | |
106 | if (level <= ip_vs_get_debug_level()) \ | |
107 | printk(KERN_DEBUG "IPVS: " msg); \ | |
108 | } while (0) | |
109 | #define IP_VS_ERR_BUF(msg...) \ | |
110 | do { \ | |
111 | char ip_vs_dbg_buf[160]; \ | |
112 | int ip_vs_dbg_idx = 0; \ | |
113 | printk(KERN_ERR "IPVS: " msg); \ | |
114 | } while (0) | |
115 | ||
116 | /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */ | |
117 | #define IP_VS_DBG_ADDR(af, addr) \ | |
118 | ip_vs_dbg_addr(af, ip_vs_dbg_buf, \ | |
119 | sizeof(ip_vs_dbg_buf), addr, \ | |
120 | &ip_vs_dbg_idx) | |
121 | ||
1da177e4 LT |
122 | #define IP_VS_DBG(level, msg...) \ |
123 | do { \ | |
124 | if (level <= ip_vs_get_debug_level()) \ | |
125 | printk(KERN_DEBUG "IPVS: " msg); \ | |
126 | } while (0) | |
127 | #define IP_VS_DBG_RL(msg...) \ | |
128 | do { \ | |
129 | if (net_ratelimit()) \ | |
130 | printk(KERN_DEBUG "IPVS: " msg); \ | |
131 | } while (0) | |
132 | #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) \ | |
133 | do { \ | |
134 | if (level <= ip_vs_get_debug_level()) \ | |
135 | pp->debug_packet(pp, skb, ofs, msg); \ | |
136 | } while (0) | |
137 | #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) \ | |
138 | do { \ | |
139 | if (level <= ip_vs_get_debug_level() && \ | |
140 | net_ratelimit()) \ | |
141 | pp->debug_packet(pp, skb, ofs, msg); \ | |
142 | } while (0) | |
143 | #else /* NO DEBUGGING at ALL */ | |
c842a3ad JV |
144 | #define IP_VS_DBG_BUF(level, msg...) do {} while (0) |
145 | #define IP_VS_ERR_BUF(msg...) do {} while (0) | |
1da177e4 LT |
146 | #define IP_VS_DBG(level, msg...) do {} while (0) |
147 | #define IP_VS_DBG_RL(msg...) do {} while (0) | |
148 | #define IP_VS_DBG_PKT(level, pp, skb, ofs, msg) do {} while (0) | |
149 | #define IP_VS_DBG_RL_PKT(level, pp, skb, ofs, msg) do {} while (0) | |
150 | #endif | |
151 | ||
152 | #define IP_VS_BUG() BUG() | |
153 | #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg) | |
154 | #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg) | |
155 | #define IP_VS_WARNING(msg...) \ | |
156 | printk(KERN_WARNING "IPVS: " msg) | |
157 | #define IP_VS_ERR_RL(msg...) \ | |
158 | do { \ | |
159 | if (net_ratelimit()) \ | |
160 | printk(KERN_ERR "IPVS: " msg); \ | |
161 | } while (0) | |
162 | ||
163 | #ifdef CONFIG_IP_VS_DEBUG | |
164 | #define EnterFunction(level) \ | |
165 | do { \ | |
166 | if (level <= ip_vs_get_debug_level()) \ | |
167 | printk(KERN_DEBUG "Enter: %s, %s line %i\n", \ | |
d5c003b4 | 168 | __func__, __FILE__, __LINE__); \ |
1da177e4 LT |
169 | } while (0) |
170 | #define LeaveFunction(level) \ | |
171 | do { \ | |
172 | if (level <= ip_vs_get_debug_level()) \ | |
173 | printk(KERN_DEBUG "Leave: %s, %s line %i\n", \ | |
d5c003b4 | 174 | __func__, __FILE__, __LINE__); \ |
1da177e4 LT |
175 | } while (0) |
176 | #else | |
177 | #define EnterFunction(level) do {} while (0) | |
178 | #define LeaveFunction(level) do {} while (0) | |
179 | #endif | |
180 | ||
181 | #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); } | |
182 | ||
183 | ||
184 | /* | |
185 | * The port number of FTP service (in network order). | |
186 | */ | |
187 | #define FTPPORT __constant_htons(21) | |
188 | #define FTPDATA __constant_htons(20) | |
189 | ||
1da177e4 LT |
190 | /* |
191 | * TCP State Values | |
192 | */ | |
193 | enum { | |
194 | IP_VS_TCP_S_NONE = 0, | |
195 | IP_VS_TCP_S_ESTABLISHED, | |
196 | IP_VS_TCP_S_SYN_SENT, | |
197 | IP_VS_TCP_S_SYN_RECV, | |
198 | IP_VS_TCP_S_FIN_WAIT, | |
199 | IP_VS_TCP_S_TIME_WAIT, | |
200 | IP_VS_TCP_S_CLOSE, | |
201 | IP_VS_TCP_S_CLOSE_WAIT, | |
202 | IP_VS_TCP_S_LAST_ACK, | |
203 | IP_VS_TCP_S_LISTEN, | |
204 | IP_VS_TCP_S_SYNACK, | |
205 | IP_VS_TCP_S_LAST | |
206 | }; | |
207 | ||
208 | /* | |
209 | * UDP State Values | |
210 | */ | |
211 | enum { | |
212 | IP_VS_UDP_S_NORMAL, | |
213 | IP_VS_UDP_S_LAST, | |
214 | }; | |
215 | ||
216 | /* | |
217 | * ICMP State Values | |
218 | */ | |
219 | enum { | |
220 | IP_VS_ICMP_S_NORMAL, | |
221 | IP_VS_ICMP_S_LAST, | |
222 | }; | |
223 | ||
224 | /* | |
225 | * Delta sequence info structure | |
226 | * Each ip_vs_conn has 2 (output AND input seq. changes). | |
227 | * Only used in the VS/NAT. | |
228 | */ | |
229 | struct ip_vs_seq { | |
230 | __u32 init_seq; /* Add delta from this seq */ | |
231 | __u32 delta; /* Delta in sequence numbers */ | |
232 | __u32 previous_delta; /* Delta in sequence numbers | |
233 | before last resized pkt */ | |
234 | }; | |
235 | ||
236 | ||
237 | /* | |
3a14a313 | 238 | * IPVS statistics objects |
1da177e4 | 239 | */ |
3a14a313 SW |
240 | struct ip_vs_estimator { |
241 | struct list_head list; | |
242 | ||
243 | u64 last_inbytes; | |
244 | u64 last_outbytes; | |
245 | u32 last_conns; | |
246 | u32 last_inpkts; | |
247 | u32 last_outpkts; | |
248 | ||
249 | u32 cps; | |
250 | u32 inpps; | |
251 | u32 outpps; | |
252 | u32 inbps; | |
253 | u32 outbps; | |
254 | }; | |
255 | ||
1da177e4 LT |
256 | struct ip_vs_stats |
257 | { | |
e9c0ce23 SW |
258 | struct ip_vs_stats_user ustats; /* statistics */ |
259 | struct ip_vs_estimator est; /* estimator */ | |
3a14a313 | 260 | |
1da177e4 LT |
261 | spinlock_t lock; /* spin lock */ |
262 | }; | |
263 | ||
14c85021 ACM |
264 | struct dst_entry; |
265 | struct iphdr; | |
1da177e4 LT |
266 | struct ip_vs_conn; |
267 | struct ip_vs_app; | |
14c85021 | 268 | struct sk_buff; |
1da177e4 LT |
269 | |
270 | struct ip_vs_protocol { | |
271 | struct ip_vs_protocol *next; | |
272 | char *name; | |
2ad17def JA |
273 | u16 protocol; |
274 | u16 num_states; | |
1da177e4 LT |
275 | int dont_defrag; |
276 | atomic_t appcnt; /* counter of proto app incs */ | |
277 | int *timeout_table; /* protocol timeout table */ | |
278 | ||
279 | void (*init)(struct ip_vs_protocol *pp); | |
280 | ||
281 | void (*exit)(struct ip_vs_protocol *pp); | |
282 | ||
51ef348b | 283 | int (*conn_schedule)(int af, struct sk_buff *skb, |
1da177e4 LT |
284 | struct ip_vs_protocol *pp, |
285 | int *verdict, struct ip_vs_conn **cpp); | |
286 | ||
287 | struct ip_vs_conn * | |
51ef348b JV |
288 | (*conn_in_get)(int af, |
289 | const struct sk_buff *skb, | |
1da177e4 | 290 | struct ip_vs_protocol *pp, |
51ef348b | 291 | const struct ip_vs_iphdr *iph, |
1da177e4 LT |
292 | unsigned int proto_off, |
293 | int inverse); | |
294 | ||
295 | struct ip_vs_conn * | |
51ef348b JV |
296 | (*conn_out_get)(int af, |
297 | const struct sk_buff *skb, | |
1da177e4 | 298 | struct ip_vs_protocol *pp, |
51ef348b | 299 | const struct ip_vs_iphdr *iph, |
1da177e4 LT |
300 | unsigned int proto_off, |
301 | int inverse); | |
302 | ||
3db05fea | 303 | int (*snat_handler)(struct sk_buff *skb, |
1da177e4 LT |
304 | struct ip_vs_protocol *pp, struct ip_vs_conn *cp); |
305 | ||
3db05fea | 306 | int (*dnat_handler)(struct sk_buff *skb, |
1da177e4 LT |
307 | struct ip_vs_protocol *pp, struct ip_vs_conn *cp); |
308 | ||
51ef348b JV |
309 | int (*csum_check)(int af, struct sk_buff *skb, |
310 | struct ip_vs_protocol *pp); | |
1da177e4 LT |
311 | |
312 | const char *(*state_name)(int state); | |
313 | ||
314 | int (*state_transition)(struct ip_vs_conn *cp, int direction, | |
315 | const struct sk_buff *skb, | |
316 | struct ip_vs_protocol *pp); | |
317 | ||
318 | int (*register_app)(struct ip_vs_app *inc); | |
319 | ||
320 | void (*unregister_app)(struct ip_vs_app *inc); | |
321 | ||
322 | int (*app_conn_bind)(struct ip_vs_conn *cp); | |
323 | ||
324 | void (*debug_packet)(struct ip_vs_protocol *pp, | |
325 | const struct sk_buff *skb, | |
326 | int offset, | |
327 | const char *msg); | |
328 | ||
329 | void (*timeout_change)(struct ip_vs_protocol *pp, int flags); | |
330 | ||
331 | int (*set_state_timeout)(struct ip_vs_protocol *pp, char *sname, int to); | |
332 | }; | |
333 | ||
334 | extern struct ip_vs_protocol * ip_vs_proto_get(unsigned short proto); | |
335 | ||
336 | /* | |
337 | * IP_VS structure allocated for each dynamically scheduled connection | |
338 | */ | |
339 | struct ip_vs_conn { | |
340 | struct list_head c_list; /* hashed list heads */ | |
341 | ||
342 | /* Protocol, addresses and port numbers */ | |
e7ade46a JV |
343 | u16 af; /* address family */ |
344 | union nf_inet_addr caddr; /* client address */ | |
345 | union nf_inet_addr vaddr; /* virtual address */ | |
346 | union nf_inet_addr daddr; /* destination address */ | |
014d730d AV |
347 | __be16 cport; |
348 | __be16 vport; | |
349 | __be16 dport; | |
1da177e4 LT |
350 | __u16 protocol; /* Which protocol (TCP/UDP) */ |
351 | ||
352 | /* counter and timer */ | |
353 | atomic_t refcnt; /* reference count */ | |
354 | struct timer_list timer; /* Expiration timer */ | |
355 | volatile unsigned long timeout; /* timeout */ | |
356 | ||
357 | /* Flags and state transition */ | |
358 | spinlock_t lock; /* lock for state transition */ | |
359 | volatile __u16 flags; /* status flags */ | |
360 | volatile __u16 state; /* state info */ | |
efac5276 RB |
361 | volatile __u16 old_state; /* old state, to be used for |
362 | * state transition triggerd | |
363 | * synchronization | |
364 | */ | |
1da177e4 LT |
365 | |
366 | /* Control members */ | |
367 | struct ip_vs_conn *control; /* Master control connection */ | |
368 | atomic_t n_control; /* Number of controlled ones */ | |
369 | struct ip_vs_dest *dest; /* real server */ | |
370 | atomic_t in_pkts; /* incoming packet counter */ | |
371 | ||
372 | /* packet transmitter for different forwarding methods. If it | |
373 | mangles the packet, it must return NF_DROP or better NF_STOLEN, | |
374 | otherwise this must be changed to a sk_buff **. | |
375 | */ | |
376 | int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp, | |
377 | struct ip_vs_protocol *pp); | |
378 | ||
379 | /* Note: we can group the following members into a structure, | |
380 | in order to save more space, and the following members are | |
381 | only used in VS/NAT anyway */ | |
382 | struct ip_vs_app *app; /* bound ip_vs_app object */ | |
383 | void *app_data; /* Application private data */ | |
384 | struct ip_vs_seq in_seq; /* incoming seq. struct */ | |
385 | struct ip_vs_seq out_seq; /* outgoing seq. struct */ | |
386 | }; | |
387 | ||
388 | ||
c860c6b1 JV |
389 | /* |
390 | * Extended internal versions of struct ip_vs_service_user and | |
391 | * ip_vs_dest_user for IPv6 support. | |
392 | * | |
393 | * We need these to conveniently pass around service and destination | |
394 | * options, but unfortunately, we also need to keep the old definitions to | |
395 | * maintain userspace backwards compatibility for the setsockopt interface. | |
396 | */ | |
397 | struct ip_vs_service_user_kern { | |
398 | /* virtual service addresses */ | |
399 | u16 af; | |
400 | u16 protocol; | |
401 | union nf_inet_addr addr; /* virtual ip address */ | |
402 | u16 port; | |
403 | u32 fwmark; /* firwall mark of service */ | |
404 | ||
405 | /* virtual service options */ | |
406 | char *sched_name; | |
407 | unsigned flags; /* virtual service flags */ | |
408 | unsigned timeout; /* persistent timeout in sec */ | |
409 | u32 netmask; /* persistent netmask */ | |
410 | }; | |
411 | ||
412 | ||
413 | struct ip_vs_dest_user_kern { | |
414 | /* destination server address */ | |
415 | union nf_inet_addr addr; | |
416 | u16 port; | |
417 | ||
418 | /* real server options */ | |
419 | unsigned conn_flags; /* connection flags */ | |
420 | int weight; /* destination weight */ | |
421 | ||
422 | /* thresholds for active connections */ | |
423 | u32 u_threshold; /* upper threshold */ | |
424 | u32 l_threshold; /* lower threshold */ | |
425 | }; | |
426 | ||
427 | ||
1da177e4 LT |
428 | /* |
429 | * The information about the virtual service offered to the net | |
430 | * and the forwarding entries | |
431 | */ | |
432 | struct ip_vs_service { | |
433 | struct list_head s_list; /* for normal service table */ | |
434 | struct list_head f_list; /* for fwmark-based service table */ | |
435 | atomic_t refcnt; /* reference counter */ | |
436 | atomic_t usecnt; /* use counter */ | |
437 | ||
e7ade46a | 438 | u16 af; /* address family */ |
1da177e4 | 439 | __u16 protocol; /* which protocol (TCP/UDP) */ |
e7ade46a | 440 | union nf_inet_addr addr; /* IP address for virtual service */ |
014d730d | 441 | __be16 port; /* port number for the service */ |
1da177e4 LT |
442 | __u32 fwmark; /* firewall mark of the service */ |
443 | unsigned flags; /* service status flags */ | |
444 | unsigned timeout; /* persistent timeout in ticks */ | |
014d730d | 445 | __be32 netmask; /* grouping granularity */ |
1da177e4 LT |
446 | |
447 | struct list_head destinations; /* real server d-linked list */ | |
448 | __u32 num_dests; /* number of servers */ | |
449 | struct ip_vs_stats stats; /* statistics for the service */ | |
450 | struct ip_vs_app *inc; /* bind conns to this app inc */ | |
451 | ||
452 | /* for scheduling */ | |
453 | struct ip_vs_scheduler *scheduler; /* bound scheduler object */ | |
454 | rwlock_t sched_lock; /* lock sched_data */ | |
455 | void *sched_data; /* scheduler application data */ | |
456 | }; | |
457 | ||
458 | ||
459 | /* | |
460 | * The real server destination forwarding entry | |
461 | * with ip address, port number, and so on. | |
462 | */ | |
463 | struct ip_vs_dest { | |
464 | struct list_head n_list; /* for the dests in the service */ | |
465 | struct list_head d_list; /* for table with all the dests */ | |
466 | ||
e7ade46a JV |
467 | u16 af; /* address family */ |
468 | union nf_inet_addr addr; /* IP address of the server */ | |
014d730d | 469 | __be16 port; /* port number of the server */ |
1da177e4 LT |
470 | volatile unsigned flags; /* dest status flags */ |
471 | atomic_t conn_flags; /* flags to copy to conn */ | |
472 | atomic_t weight; /* server weight */ | |
473 | ||
474 | atomic_t refcnt; /* reference counter */ | |
475 | struct ip_vs_stats stats; /* statistics */ | |
476 | ||
477 | /* connection counters and thresholds */ | |
478 | atomic_t activeconns; /* active connections */ | |
479 | atomic_t inactconns; /* inactive connections */ | |
480 | atomic_t persistconns; /* persistent connections */ | |
481 | __u32 u_threshold; /* upper threshold */ | |
482 | __u32 l_threshold; /* lower threshold */ | |
483 | ||
484 | /* for destination cache */ | |
485 | spinlock_t dst_lock; /* lock of dst_cache */ | |
486 | struct dst_entry *dst_cache; /* destination cache entry */ | |
487 | u32 dst_rtos; /* RT_TOS(tos) for dst */ | |
488 | ||
489 | /* for virtual service */ | |
490 | struct ip_vs_service *svc; /* service it belongs to */ | |
491 | __u16 protocol; /* which protocol (TCP/UDP) */ | |
e7ade46a | 492 | union nf_inet_addr vaddr; /* virtual IP address */ |
014d730d | 493 | __be16 vport; /* virtual port number */ |
1da177e4 LT |
494 | __u32 vfwmark; /* firewall mark of service */ |
495 | }; | |
496 | ||
497 | ||
498 | /* | |
499 | * The scheduler object | |
500 | */ | |
501 | struct ip_vs_scheduler { | |
502 | struct list_head n_list; /* d-linked list head */ | |
503 | char *name; /* scheduler name */ | |
504 | atomic_t refcnt; /* reference counter */ | |
505 | struct module *module; /* THIS_MODULE/NULL */ | |
b14198f6 JV |
506 | #ifdef CONFIG_IP_VS_IPV6 |
507 | int supports_ipv6; /* scheduler has IPv6 support */ | |
508 | #endif | |
1da177e4 LT |
509 | |
510 | /* scheduler initializing service */ | |
511 | int (*init_service)(struct ip_vs_service *svc); | |
512 | /* scheduling service finish */ | |
513 | int (*done_service)(struct ip_vs_service *svc); | |
514 | /* scheduler updating service */ | |
515 | int (*update_service)(struct ip_vs_service *svc); | |
516 | ||
517 | /* selecting a server from the given service */ | |
518 | struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc, | |
519 | const struct sk_buff *skb); | |
520 | }; | |
521 | ||
522 | ||
523 | /* | |
524 | * The application module object (a.k.a. app incarnation) | |
525 | */ | |
526 | struct ip_vs_app | |
527 | { | |
528 | struct list_head a_list; /* member in app list */ | |
529 | int type; /* IP_VS_APP_TYPE_xxx */ | |
530 | char *name; /* application module name */ | |
531 | __u16 protocol; | |
532 | struct module *module; /* THIS_MODULE/NULL */ | |
533 | struct list_head incs_list; /* list of incarnations */ | |
534 | ||
535 | /* members for application incarnations */ | |
536 | struct list_head p_list; /* member in proto app list */ | |
537 | struct ip_vs_app *app; /* its real application */ | |
014d730d | 538 | __be16 port; /* port number in net order */ |
1da177e4 LT |
539 | atomic_t usecnt; /* usage counter */ |
540 | ||
541 | /* output hook: return false if can't linearize. diff set for TCP. */ | |
542 | int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *, | |
3db05fea | 543 | struct sk_buff *, int *diff); |
1da177e4 LT |
544 | |
545 | /* input hook: return false if can't linearize. diff set for TCP. */ | |
546 | int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *, | |
3db05fea | 547 | struct sk_buff *, int *diff); |
1da177e4 LT |
548 | |
549 | /* ip_vs_app initializer */ | |
550 | int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *); | |
551 | ||
552 | /* ip_vs_app finish */ | |
553 | int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *); | |
554 | ||
555 | ||
556 | /* not used now */ | |
557 | int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *, | |
558 | struct ip_vs_protocol *); | |
559 | ||
560 | void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *); | |
561 | ||
562 | int * timeout_table; | |
563 | int * timeouts; | |
564 | int timeouts_size; | |
565 | ||
566 | int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app, | |
567 | int *verdict, struct ip_vs_conn **cpp); | |
568 | ||
569 | struct ip_vs_conn * | |
570 | (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app, | |
571 | const struct iphdr *iph, unsigned int proto_off, | |
572 | int inverse); | |
573 | ||
574 | struct ip_vs_conn * | |
575 | (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app, | |
576 | const struct iphdr *iph, unsigned int proto_off, | |
577 | int inverse); | |
578 | ||
579 | int (*state_transition)(struct ip_vs_conn *cp, int direction, | |
580 | const struct sk_buff *skb, | |
581 | struct ip_vs_app *app); | |
582 | ||
583 | void (*timeout_change)(struct ip_vs_app *app, int flags); | |
584 | }; | |
585 | ||
586 | ||
587 | /* | |
588 | * IPVS core functions | |
589 | * (from ip_vs_core.c) | |
590 | */ | |
591 | extern const char *ip_vs_proto_name(unsigned proto); | |
592 | extern void ip_vs_init_hash_table(struct list_head *table, int rows); | |
afdd6140 | 593 | #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t))) |
1da177e4 | 594 | |
1da177e4 LT |
595 | #define IP_VS_APP_TYPE_FTP 1 |
596 | ||
597 | /* | |
598 | * ip_vs_conn handling functions | |
599 | * (from ip_vs_conn.c) | |
600 | */ | |
601 | ||
602 | /* | |
603 | * IPVS connection entry hash table | |
604 | */ | |
605 | #ifndef CONFIG_IP_VS_TAB_BITS | |
606 | #define CONFIG_IP_VS_TAB_BITS 12 | |
607 | #endif | |
2206a3f5 | 608 | |
1da177e4 | 609 | #define IP_VS_CONN_TAB_BITS CONFIG_IP_VS_TAB_BITS |
1da177e4 LT |
610 | #define IP_VS_CONN_TAB_SIZE (1 << IP_VS_CONN_TAB_BITS) |
611 | #define IP_VS_CONN_TAB_MASK (IP_VS_CONN_TAB_SIZE - 1) | |
612 | ||
613 | enum { | |
614 | IP_VS_DIR_INPUT = 0, | |
615 | IP_VS_DIR_OUTPUT, | |
616 | IP_VS_DIR_INPUT_ONLY, | |
617 | IP_VS_DIR_LAST, | |
618 | }; | |
619 | ||
620 | extern struct ip_vs_conn *ip_vs_conn_in_get | |
28364a59 JV |
621 | (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port, |
622 | const union nf_inet_addr *d_addr, __be16 d_port); | |
623 | ||
87375ab4 | 624 | extern struct ip_vs_conn *ip_vs_ct_in_get |
28364a59 JV |
625 | (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port, |
626 | const union nf_inet_addr *d_addr, __be16 d_port); | |
627 | ||
1da177e4 | 628 | extern struct ip_vs_conn *ip_vs_conn_out_get |
28364a59 JV |
629 | (int af, int protocol, const union nf_inet_addr *s_addr, __be16 s_port, |
630 | const union nf_inet_addr *d_addr, __be16 d_port); | |
1da177e4 LT |
631 | |
632 | /* put back the conn without restarting its timer */ | |
633 | static inline void __ip_vs_conn_put(struct ip_vs_conn *cp) | |
634 | { | |
635 | atomic_dec(&cp->refcnt); | |
636 | } | |
637 | extern void ip_vs_conn_put(struct ip_vs_conn *cp); | |
014d730d | 638 | extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport); |
1da177e4 LT |
639 | |
640 | extern struct ip_vs_conn * | |
28364a59 JV |
641 | ip_vs_conn_new(int af, int proto, const union nf_inet_addr *caddr, __be16 cport, |
642 | const union nf_inet_addr *vaddr, __be16 vport, | |
643 | const union nf_inet_addr *daddr, __be16 dport, unsigned flags, | |
1da177e4 LT |
644 | struct ip_vs_dest *dest); |
645 | extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp); | |
646 | ||
647 | extern const char * ip_vs_state_name(__u16 proto, int state); | |
648 | ||
649 | extern void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp); | |
650 | extern int ip_vs_check_template(struct ip_vs_conn *ct); | |
1da177e4 LT |
651 | extern void ip_vs_random_dropentry(void); |
652 | extern int ip_vs_conn_init(void); | |
653 | extern void ip_vs_conn_cleanup(void); | |
654 | ||
655 | static inline void ip_vs_control_del(struct ip_vs_conn *cp) | |
656 | { | |
657 | struct ip_vs_conn *ctl_cp = cp->control; | |
658 | if (!ctl_cp) { | |
cfc78c5a JV |
659 | IP_VS_ERR_BUF("request control DEL for uncontrolled: " |
660 | "%s:%d to %s:%d\n", | |
661 | IP_VS_DBG_ADDR(cp->af, &cp->caddr), | |
662 | ntohs(cp->cport), | |
663 | IP_VS_DBG_ADDR(cp->af, &cp->vaddr), | |
664 | ntohs(cp->vport)); | |
665 | ||
1da177e4 LT |
666 | return; |
667 | } | |
668 | ||
cfc78c5a JV |
669 | IP_VS_DBG_BUF(7, "DELeting control for: " |
670 | "cp.dst=%s:%d ctl_cp.dst=%s:%d\n", | |
671 | IP_VS_DBG_ADDR(cp->af, &cp->caddr), | |
672 | ntohs(cp->cport), | |
673 | IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr), | |
674 | ntohs(ctl_cp->cport)); | |
1da177e4 LT |
675 | |
676 | cp->control = NULL; | |
677 | if (atomic_read(&ctl_cp->n_control) == 0) { | |
cfc78c5a JV |
678 | IP_VS_ERR_BUF("BUG control DEL with n=0 : " |
679 | "%s:%d to %s:%d\n", | |
680 | IP_VS_DBG_ADDR(cp->af, &cp->caddr), | |
681 | ntohs(cp->cport), | |
682 | IP_VS_DBG_ADDR(cp->af, &cp->vaddr), | |
683 | ntohs(cp->vport)); | |
684 | ||
1da177e4 LT |
685 | return; |
686 | } | |
687 | atomic_dec(&ctl_cp->n_control); | |
688 | } | |
689 | ||
690 | static inline void | |
691 | ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp) | |
692 | { | |
693 | if (cp->control) { | |
cfc78c5a JV |
694 | IP_VS_ERR_BUF("request control ADD for already controlled: " |
695 | "%s:%d to %s:%d\n", | |
696 | IP_VS_DBG_ADDR(cp->af, &cp->caddr), | |
697 | ntohs(cp->cport), | |
698 | IP_VS_DBG_ADDR(cp->af, &cp->vaddr), | |
699 | ntohs(cp->vport)); | |
700 | ||
1da177e4 LT |
701 | ip_vs_control_del(cp); |
702 | } | |
703 | ||
cfc78c5a JV |
704 | IP_VS_DBG_BUF(7, "ADDing control for: " |
705 | "cp.dst=%s:%d ctl_cp.dst=%s:%d\n", | |
706 | IP_VS_DBG_ADDR(cp->af, &cp->caddr), | |
707 | ntohs(cp->cport), | |
708 | IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr), | |
709 | ntohs(ctl_cp->cport)); | |
1da177e4 LT |
710 | |
711 | cp->control = ctl_cp; | |
712 | atomic_inc(&ctl_cp->n_control); | |
713 | } | |
714 | ||
715 | ||
716 | /* | |
717 | * IPVS application functions | |
718 | * (from ip_vs_app.c) | |
719 | */ | |
720 | #define IP_VS_APP_MAX_PORTS 8 | |
721 | extern int register_ip_vs_app(struct ip_vs_app *app); | |
722 | extern void unregister_ip_vs_app(struct ip_vs_app *app); | |
723 | extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
724 | extern void ip_vs_unbind_app(struct ip_vs_conn *cp); | |
725 | extern int | |
726 | register_ip_vs_app_inc(struct ip_vs_app *app, __u16 proto, __u16 port); | |
727 | extern int ip_vs_app_inc_get(struct ip_vs_app *inc); | |
728 | extern void ip_vs_app_inc_put(struct ip_vs_app *inc); | |
729 | ||
3db05fea HX |
730 | extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb); |
731 | extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb); | |
dd0fc66f | 732 | extern int ip_vs_skb_replace(struct sk_buff *skb, gfp_t pri, |
1da177e4 LT |
733 | char *o_buf, int o_len, char *n_buf, int n_len); |
734 | extern int ip_vs_app_init(void); | |
735 | extern void ip_vs_app_cleanup(void); | |
736 | ||
737 | ||
738 | /* | |
739 | * IPVS protocol functions (from ip_vs_proto.c) | |
740 | */ | |
741 | extern int ip_vs_protocol_init(void); | |
742 | extern void ip_vs_protocol_cleanup(void); | |
743 | extern void ip_vs_protocol_timeout_change(int flags); | |
744 | extern int *ip_vs_create_timeout_table(int *table, int size); | |
745 | extern int | |
746 | ip_vs_set_state_timeout(int *table, int num, char **names, char *name, int to); | |
747 | extern void | |
748 | ip_vs_tcpudp_debug_packet(struct ip_vs_protocol *pp, const struct sk_buff *skb, | |
749 | int offset, const char *msg); | |
750 | ||
751 | extern struct ip_vs_protocol ip_vs_protocol_tcp; | |
752 | extern struct ip_vs_protocol ip_vs_protocol_udp; | |
753 | extern struct ip_vs_protocol ip_vs_protocol_icmp; | |
754 | extern struct ip_vs_protocol ip_vs_protocol_esp; | |
755 | extern struct ip_vs_protocol ip_vs_protocol_ah; | |
756 | ||
757 | ||
758 | /* | |
759 | * Registering/unregistering scheduler functions | |
760 | * (from ip_vs_sched.c) | |
761 | */ | |
762 | extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler); | |
763 | extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler); | |
764 | extern int ip_vs_bind_scheduler(struct ip_vs_service *svc, | |
765 | struct ip_vs_scheduler *scheduler); | |
766 | extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc); | |
767 | extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name); | |
768 | extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler); | |
769 | extern struct ip_vs_conn * | |
770 | ip_vs_schedule(struct ip_vs_service *svc, const struct sk_buff *skb); | |
771 | extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb, | |
772 | struct ip_vs_protocol *pp); | |
773 | ||
774 | ||
775 | /* | |
776 | * IPVS control data and functions (from ip_vs_ctl.c) | |
777 | */ | |
778 | extern int sysctl_ip_vs_cache_bypass; | |
779 | extern int sysctl_ip_vs_expire_nodest_conn; | |
780 | extern int sysctl_ip_vs_expire_quiescent_template; | |
781 | extern int sysctl_ip_vs_sync_threshold[2]; | |
782 | extern int sysctl_ip_vs_nat_icmp_send; | |
783 | extern struct ip_vs_stats ip_vs_stats; | |
5587da55 | 784 | extern const struct ctl_path net_vs_ctl_path[]; |
1da177e4 LT |
785 | |
786 | extern struct ip_vs_service * | |
3c2e0505 JV |
787 | ip_vs_service_get(int af, __u32 fwmark, __u16 protocol, |
788 | const union nf_inet_addr *vaddr, __be16 vport); | |
1da177e4 LT |
789 | |
790 | static inline void ip_vs_service_put(struct ip_vs_service *svc) | |
791 | { | |
792 | atomic_dec(&svc->usecnt); | |
793 | } | |
794 | ||
795 | extern struct ip_vs_dest * | |
7937df15 JV |
796 | ip_vs_lookup_real_service(int af, __u16 protocol, |
797 | const union nf_inet_addr *daddr, __be16 dport); | |
798 | ||
1da177e4 LT |
799 | extern int ip_vs_use_count_inc(void); |
800 | extern void ip_vs_use_count_dec(void); | |
801 | extern int ip_vs_control_init(void); | |
802 | extern void ip_vs_control_cleanup(void); | |
1e356f9c | 803 | extern struct ip_vs_dest * |
7937df15 JV |
804 | ip_vs_find_dest(int af, const union nf_inet_addr *daddr, __be16 dport, |
805 | const union nf_inet_addr *vaddr, __be16 vport, __u16 protocol); | |
1e356f9c | 806 | extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp); |
1da177e4 LT |
807 | |
808 | ||
809 | /* | |
810 | * IPVS sync daemon data and function prototypes | |
811 | * (from ip_vs_sync.c) | |
812 | */ | |
813 | extern volatile int ip_vs_sync_state; | |
814 | extern volatile int ip_vs_master_syncid; | |
815 | extern volatile int ip_vs_backup_syncid; | |
816 | extern char ip_vs_master_mcast_ifn[IP_VS_IFNAME_MAXLEN]; | |
817 | extern char ip_vs_backup_mcast_ifn[IP_VS_IFNAME_MAXLEN]; | |
818 | extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid); | |
819 | extern int stop_sync_thread(int state); | |
820 | extern void ip_vs_sync_conn(struct ip_vs_conn *cp); | |
821 | ||
822 | ||
823 | /* | |
824 | * IPVS rate estimator prototypes (from ip_vs_est.c) | |
825 | */ | |
a919cf4b SW |
826 | extern int ip_vs_estimator_init(void); |
827 | extern void ip_vs_estimator_cleanup(void); | |
3a14a313 | 828 | extern void ip_vs_new_estimator(struct ip_vs_stats *stats); |
1da177e4 LT |
829 | extern void ip_vs_kill_estimator(struct ip_vs_stats *stats); |
830 | extern void ip_vs_zero_estimator(struct ip_vs_stats *stats); | |
831 | ||
832 | /* | |
833 | * Various IPVS packet transmitters (from ip_vs_xmit.c) | |
834 | */ | |
835 | extern int ip_vs_null_xmit | |
836 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
837 | extern int ip_vs_bypass_xmit | |
838 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
839 | extern int ip_vs_nat_xmit | |
840 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
841 | extern int ip_vs_tunnel_xmit | |
842 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
843 | extern int ip_vs_dr_xmit | |
844 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
845 | extern int ip_vs_icmp_xmit | |
846 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset); | |
847 | extern void ip_vs_dst_reset(struct ip_vs_dest *dest); | |
848 | ||
b3cdd2a7 JV |
849 | #ifdef CONFIG_IP_VS_IPV6 |
850 | extern int ip_vs_bypass_xmit_v6 | |
851 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
852 | extern int ip_vs_nat_xmit_v6 | |
853 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
854 | extern int ip_vs_tunnel_xmit_v6 | |
855 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
856 | extern int ip_vs_dr_xmit_v6 | |
857 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp); | |
858 | extern int ip_vs_icmp_xmit_v6 | |
859 | (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, | |
860 | int offset); | |
861 | #endif | |
1da177e4 LT |
862 | |
863 | /* | |
864 | * This is a simple mechanism to ignore packets when | |
865 | * we are loaded. Just set ip_vs_drop_rate to 'n' and | |
866 | * we start to drop 1/rate of the packets | |
867 | */ | |
868 | extern int ip_vs_drop_rate; | |
869 | extern int ip_vs_drop_counter; | |
870 | ||
871 | static __inline__ int ip_vs_todrop(void) | |
872 | { | |
873 | if (!ip_vs_drop_rate) return 0; | |
874 | if (--ip_vs_drop_counter > 0) return 0; | |
875 | ip_vs_drop_counter = ip_vs_drop_rate; | |
876 | return 1; | |
877 | } | |
878 | ||
879 | /* | |
880 | * ip_vs_fwd_tag returns the forwarding tag of the connection | |
881 | */ | |
882 | #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK) | |
883 | ||
732db659 | 884 | static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp) |
1da177e4 LT |
885 | { |
886 | char fwd; | |
887 | ||
888 | switch (IP_VS_FWD_METHOD(cp)) { | |
889 | case IP_VS_CONN_F_MASQ: | |
890 | fwd = 'M'; break; | |
891 | case IP_VS_CONN_F_LOCALNODE: | |
892 | fwd = 'L'; break; | |
893 | case IP_VS_CONN_F_TUNNEL: | |
894 | fwd = 'T'; break; | |
895 | case IP_VS_CONN_F_DROUTE: | |
896 | fwd = 'R'; break; | |
897 | case IP_VS_CONN_F_BYPASS: | |
898 | fwd = 'B'; break; | |
899 | default: | |
900 | fwd = '?'; break; | |
901 | } | |
902 | return fwd; | |
903 | } | |
904 | ||
1da177e4 | 905 | extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp, |
b3cdd2a7 JV |
906 | struct ip_vs_conn *cp, int dir); |
907 | ||
908 | #ifdef CONFIG_IP_VS_IPV6 | |
909 | extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp, | |
910 | struct ip_vs_conn *cp, int dir); | |
911 | #endif | |
1da177e4 | 912 | |
b1550f22 | 913 | extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset); |
1da177e4 | 914 | |
f9214b26 | 915 | static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum) |
1da177e4 | 916 | { |
f9214b26 | 917 | __be32 diff[2] = { ~old, new }; |
1da177e4 | 918 | |
f9214b26 AV |
919 | return csum_partial((char *) diff, sizeof(diff), oldsum); |
920 | } | |
921 | ||
0bbdd42b JV |
922 | #ifdef CONFIG_IP_VS_IPV6 |
923 | static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new, | |
924 | __wsum oldsum) | |
925 | { | |
926 | __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0], | |
927 | new[3], new[2], new[1], new[0] }; | |
928 | ||
929 | return csum_partial((char *) diff, sizeof(diff), oldsum); | |
930 | } | |
931 | #endif | |
932 | ||
f9214b26 AV |
933 | static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum) |
934 | { | |
935 | __be16 diff[2] = { ~old, new }; | |
936 | ||
937 | return csum_partial((char *) diff, sizeof(diff), oldsum); | |
1da177e4 LT |
938 | } |
939 | ||
940 | #endif /* __KERNEL__ */ | |
941 | ||
bc4768eb | 942 | #endif /* _NET_IP_VS_H */ |