ip_vs.h 39 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384
  1. /*
  2. * IP Virtual Server
  3. * data structure and functionality definitions
  4. */
  5. #ifndef _NET_IP_VS_H
  6. #define _NET_IP_VS_H
  7. #include <linux/ip_vs.h> /* definitions shared with userland */
  8. /* old ipvsadm versions still include this file directly */
  9. #ifdef __KERNEL__
  10. #include <asm/types.h> /* for __uXX types */
  11. #include <linux/sysctl.h> /* for ctl_path */
  12. #include <linux/list.h> /* for struct list_head */
  13. #include <linux/spinlock.h> /* for struct rwlock_t */
  14. #include <asm/atomic.h> /* for struct atomic_t */
  15. #include <linux/compiler.h>
  16. #include <linux/timer.h>
  17. #include <net/checksum.h>
  18. #include <linux/netfilter.h> /* for union nf_inet_addr */
  19. #include <linux/ip.h>
  20. #include <linux/ipv6.h> /* for struct ipv6hdr */
  21. #include <net/ipv6.h> /* for ipv6_addr_copy */
  22. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  23. #include <net/netfilter/nf_conntrack.h>
  24. #endif
  25. #include <net/net_namespace.h> /* Netw namespace */
  26. /*
  27. * Generic access of ipvs struct
  28. */
  29. static inline struct netns_ipvs *net_ipvs(struct net* net)
  30. {
  31. return net->ipvs;
  32. }
  33. /*
  34. * Get net ptr from skb in traffic cases
  35. * use skb_sknet when call is from userland (ioctl or netlink)
  36. */
  37. static inline struct net *skb_net(const struct sk_buff *skb)
  38. {
  39. #ifdef CONFIG_NET_NS
  40. #ifdef CONFIG_IP_VS_DEBUG
  41. /*
  42. * This is used for debug only.
  43. * Start with the most likely hit
  44. * End with BUG
  45. */
  46. if (likely(skb->dev && skb->dev->nd_net))
  47. return dev_net(skb->dev);
  48. if (skb_dst(skb)->dev)
  49. return dev_net(skb_dst(skb)->dev);
  50. WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
  51. __func__, __LINE__);
  52. if (likely(skb->sk && skb->sk->sk_net))
  53. return sock_net(skb->sk);
  54. pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
  55. __func__, __LINE__);
  56. BUG();
  57. #else
  58. return dev_net(skb->dev ? : skb_dst(skb)->dev);
  59. #endif
  60. #else
  61. return &init_net;
  62. #endif
  63. }
  64. static inline struct net *skb_sknet(const struct sk_buff *skb)
  65. {
  66. #ifdef CONFIG_NET_NS
  67. #ifdef CONFIG_IP_VS_DEBUG
  68. /* Start with the most likely hit */
  69. if (likely(skb->sk && skb->sk->sk_net))
  70. return sock_net(skb->sk);
  71. WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
  72. __func__, __LINE__);
  73. if (likely(skb->dev && skb->dev->nd_net))
  74. return dev_net(skb->dev);
  75. pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
  76. __func__, __LINE__);
  77. BUG();
  78. #else
  79. return sock_net(skb->sk);
  80. #endif
  81. #else
  82. return &init_net;
  83. #endif
  84. }
  85. /*
  86. * This one needed for single_open_net since net is stored directly in
  87. * private not as a struct i.e. seq_file_net cant be used.
  88. */
  89. static inline struct net *seq_file_single_net(struct seq_file *seq)
  90. {
  91. #ifdef CONFIG_NET_NS
  92. return (struct net *)seq->private;
  93. #else
  94. return &init_net;
  95. #endif
  96. }
  97. /* Connections' size value needed by ip_vs_ctl.c */
  98. extern int ip_vs_conn_tab_size;
  99. struct ip_vs_iphdr {
  100. int len;
  101. __u8 protocol;
  102. union nf_inet_addr saddr;
  103. union nf_inet_addr daddr;
  104. };
  105. static inline void
  106. ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
  107. {
  108. #ifdef CONFIG_IP_VS_IPV6
  109. if (af == AF_INET6) {
  110. const struct ipv6hdr *iph = nh;
  111. iphdr->len = sizeof(struct ipv6hdr);
  112. iphdr->protocol = iph->nexthdr;
  113. ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
  114. ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
  115. } else
  116. #endif
  117. {
  118. const struct iphdr *iph = nh;
  119. iphdr->len = iph->ihl * 4;
  120. iphdr->protocol = iph->protocol;
  121. iphdr->saddr.ip = iph->saddr;
  122. iphdr->daddr.ip = iph->daddr;
  123. }
  124. }
  125. static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
  126. const union nf_inet_addr *src)
  127. {
  128. #ifdef CONFIG_IP_VS_IPV6
  129. if (af == AF_INET6)
  130. ipv6_addr_copy(&dst->in6, &src->in6);
  131. else
  132. #endif
  133. dst->ip = src->ip;
  134. }
  135. static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
  136. const union nf_inet_addr *b)
  137. {
  138. #ifdef CONFIG_IP_VS_IPV6
  139. if (af == AF_INET6)
  140. return ipv6_addr_equal(&a->in6, &b->in6);
  141. #endif
  142. return a->ip == b->ip;
  143. }
  144. #ifdef CONFIG_IP_VS_DEBUG
  145. #include <linux/net.h>
  146. extern int ip_vs_get_debug_level(void);
  147. static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
  148. const union nf_inet_addr *addr,
  149. int *idx)
  150. {
  151. int len;
  152. #ifdef CONFIG_IP_VS_IPV6
  153. if (af == AF_INET6)
  154. len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
  155. &addr->in6) + 1;
  156. else
  157. #endif
  158. len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
  159. &addr->ip) + 1;
  160. *idx += len;
  161. BUG_ON(*idx > buf_len + 1);
  162. return &buf[*idx - len];
  163. }
  164. #define IP_VS_DBG_BUF(level, msg, ...) \
  165. do { \
  166. char ip_vs_dbg_buf[160]; \
  167. int ip_vs_dbg_idx = 0; \
  168. if (level <= ip_vs_get_debug_level()) \
  169. printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
  170. } while (0)
  171. #define IP_VS_ERR_BUF(msg...) \
  172. do { \
  173. char ip_vs_dbg_buf[160]; \
  174. int ip_vs_dbg_idx = 0; \
  175. pr_err(msg); \
  176. } while (0)
  177. /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
  178. #define IP_VS_DBG_ADDR(af, addr) \
  179. ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
  180. sizeof(ip_vs_dbg_buf), addr, \
  181. &ip_vs_dbg_idx)
  182. #define IP_VS_DBG(level, msg, ...) \
  183. do { \
  184. if (level <= ip_vs_get_debug_level()) \
  185. printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
  186. } while (0)
  187. #define IP_VS_DBG_RL(msg, ...) \
  188. do { \
  189. if (net_ratelimit()) \
  190. printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
  191. } while (0)
  192. #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
  193. do { \
  194. if (level <= ip_vs_get_debug_level()) \
  195. pp->debug_packet(af, pp, skb, ofs, msg); \
  196. } while (0)
  197. #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
  198. do { \
  199. if (level <= ip_vs_get_debug_level() && \
  200. net_ratelimit()) \
  201. pp->debug_packet(af, pp, skb, ofs, msg); \
  202. } while (0)
  203. #else /* NO DEBUGGING at ALL */
  204. #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
  205. #define IP_VS_ERR_BUF(msg...) do {} while (0)
  206. #define IP_VS_DBG(level, msg...) do {} while (0)
  207. #define IP_VS_DBG_RL(msg...) do {} while (0)
  208. #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
  209. #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
  210. #endif
  211. #define IP_VS_BUG() BUG()
  212. #define IP_VS_ERR_RL(msg, ...) \
  213. do { \
  214. if (net_ratelimit()) \
  215. pr_err(msg, ##__VA_ARGS__); \
  216. } while (0)
  217. #ifdef CONFIG_IP_VS_DEBUG
  218. #define EnterFunction(level) \
  219. do { \
  220. if (level <= ip_vs_get_debug_level()) \
  221. printk(KERN_DEBUG \
  222. pr_fmt("Enter: %s, %s line %i\n"), \
  223. __func__, __FILE__, __LINE__); \
  224. } while (0)
  225. #define LeaveFunction(level) \
  226. do { \
  227. if (level <= ip_vs_get_debug_level()) \
  228. printk(KERN_DEBUG \
  229. pr_fmt("Leave: %s, %s line %i\n"), \
  230. __func__, __FILE__, __LINE__); \
  231. } while (0)
  232. #else
  233. #define EnterFunction(level) do {} while (0)
  234. #define LeaveFunction(level) do {} while (0)
  235. #endif
  236. #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
  237. /*
  238. * The port number of FTP service (in network order).
  239. */
  240. #define FTPPORT cpu_to_be16(21)
  241. #define FTPDATA cpu_to_be16(20)
  242. /*
  243. * TCP State Values
  244. */
  245. enum {
  246. IP_VS_TCP_S_NONE = 0,
  247. IP_VS_TCP_S_ESTABLISHED,
  248. IP_VS_TCP_S_SYN_SENT,
  249. IP_VS_TCP_S_SYN_RECV,
  250. IP_VS_TCP_S_FIN_WAIT,
  251. IP_VS_TCP_S_TIME_WAIT,
  252. IP_VS_TCP_S_CLOSE,
  253. IP_VS_TCP_S_CLOSE_WAIT,
  254. IP_VS_TCP_S_LAST_ACK,
  255. IP_VS_TCP_S_LISTEN,
  256. IP_VS_TCP_S_SYNACK,
  257. IP_VS_TCP_S_LAST
  258. };
  259. /*
  260. * UDP State Values
  261. */
  262. enum {
  263. IP_VS_UDP_S_NORMAL,
  264. IP_VS_UDP_S_LAST,
  265. };
  266. /*
  267. * ICMP State Values
  268. */
  269. enum {
  270. IP_VS_ICMP_S_NORMAL,
  271. IP_VS_ICMP_S_LAST,
  272. };
  273. /*
  274. * SCTP State Values
  275. */
  276. enum ip_vs_sctp_states {
  277. IP_VS_SCTP_S_NONE,
  278. IP_VS_SCTP_S_INIT_CLI,
  279. IP_VS_SCTP_S_INIT_SER,
  280. IP_VS_SCTP_S_INIT_ACK_CLI,
  281. IP_VS_SCTP_S_INIT_ACK_SER,
  282. IP_VS_SCTP_S_ECHO_CLI,
  283. IP_VS_SCTP_S_ECHO_SER,
  284. IP_VS_SCTP_S_ESTABLISHED,
  285. IP_VS_SCTP_S_SHUT_CLI,
  286. IP_VS_SCTP_S_SHUT_SER,
  287. IP_VS_SCTP_S_SHUT_ACK_CLI,
  288. IP_VS_SCTP_S_SHUT_ACK_SER,
  289. IP_VS_SCTP_S_CLOSED,
  290. IP_VS_SCTP_S_LAST
  291. };
  292. /*
  293. * Delta sequence info structure
  294. * Each ip_vs_conn has 2 (output AND input seq. changes).
  295. * Only used in the VS/NAT.
  296. */
  297. struct ip_vs_seq {
  298. __u32 init_seq; /* Add delta from this seq */
  299. __u32 delta; /* Delta in sequence numbers */
  300. __u32 previous_delta; /* Delta in sequence numbers
  301. before last resized pkt */
  302. };
  303. /*
  304. * counters per cpu
  305. */
  306. struct ip_vs_counters {
  307. __u32 conns; /* connections scheduled */
  308. __u32 inpkts; /* incoming packets */
  309. __u32 outpkts; /* outgoing packets */
  310. __u64 inbytes; /* incoming bytes */
  311. __u64 outbytes; /* outgoing bytes */
  312. };
  313. /*
  314. * Stats per cpu
  315. */
  316. struct ip_vs_cpu_stats {
  317. struct ip_vs_counters ustats;
  318. struct u64_stats_sync syncp;
  319. };
  320. /*
  321. * IPVS statistics objects
  322. */
  323. struct ip_vs_estimator {
  324. struct list_head list;
  325. u64 last_inbytes;
  326. u64 last_outbytes;
  327. u32 last_conns;
  328. u32 last_inpkts;
  329. u32 last_outpkts;
  330. u32 cps;
  331. u32 inpps;
  332. u32 outpps;
  333. u32 inbps;
  334. u32 outbps;
  335. };
  336. struct ip_vs_stats {
  337. struct ip_vs_stats_user ustats; /* statistics */
  338. struct ip_vs_estimator est; /* estimator */
  339. struct ip_vs_cpu_stats *cpustats; /* per cpu counters */
  340. spinlock_t lock; /* spin lock */
  341. struct ip_vs_stats_user ustats0; /* reset values */
  342. };
  343. /*
  344. * Helper Macros for per cpu
  345. * ipvs->tot_stats->ustats.count
  346. */
  347. #define IPVS_STAT_INC(ipvs, count) \
  348. __this_cpu_inc((ipvs)->ustats->count)
  349. #define IPVS_STAT_ADD(ipvs, count, value) \
  350. do {\
  351. write_seqcount_begin(per_cpu_ptr((ipvs)->ustats_seq, \
  352. raw_smp_processor_id())); \
  353. __this_cpu_add((ipvs)->ustats->count, value); \
  354. write_seqcount_end(per_cpu_ptr((ipvs)->ustats_seq, \
  355. raw_smp_processor_id())); \
  356. } while (0)
  357. struct dst_entry;
  358. struct iphdr;
  359. struct ip_vs_conn;
  360. struct ip_vs_app;
  361. struct sk_buff;
  362. struct ip_vs_proto_data;
  363. struct ip_vs_protocol {
  364. struct ip_vs_protocol *next;
  365. char *name;
  366. u16 protocol;
  367. u16 num_states;
  368. int dont_defrag;
  369. void (*init)(struct ip_vs_protocol *pp);
  370. void (*exit)(struct ip_vs_protocol *pp);
  371. void (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
  372. void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
  373. int (*conn_schedule)(int af, struct sk_buff *skb,
  374. struct ip_vs_proto_data *pd,
  375. int *verdict, struct ip_vs_conn **cpp);
  376. struct ip_vs_conn *
  377. (*conn_in_get)(int af,
  378. const struct sk_buff *skb,
  379. const struct ip_vs_iphdr *iph,
  380. unsigned int proto_off,
  381. int inverse);
  382. struct ip_vs_conn *
  383. (*conn_out_get)(int af,
  384. const struct sk_buff *skb,
  385. const struct ip_vs_iphdr *iph,
  386. unsigned int proto_off,
  387. int inverse);
  388. int (*snat_handler)(struct sk_buff *skb,
  389. struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
  390. int (*dnat_handler)(struct sk_buff *skb,
  391. struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
  392. int (*csum_check)(int af, struct sk_buff *skb,
  393. struct ip_vs_protocol *pp);
  394. const char *(*state_name)(int state);
  395. int (*state_transition)(struct ip_vs_conn *cp, int direction,
  396. const struct sk_buff *skb,
  397. struct ip_vs_proto_data *pd);
  398. int (*register_app)(struct net *net, struct ip_vs_app *inc);
  399. void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
  400. int (*app_conn_bind)(struct ip_vs_conn *cp);
  401. void (*debug_packet)(int af, struct ip_vs_protocol *pp,
  402. const struct sk_buff *skb,
  403. int offset,
  404. const char *msg);
  405. void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
  406. };
  407. /*
  408. * protocol data per netns
  409. */
  410. struct ip_vs_proto_data {
  411. struct ip_vs_proto_data *next;
  412. struct ip_vs_protocol *pp;
  413. int *timeout_table; /* protocol timeout table */
  414. atomic_t appcnt; /* counter of proto app incs. */
  415. struct tcp_states_t *tcp_state_table;
  416. };
  417. extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
  418. extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
  419. unsigned short proto);
  420. struct ip_vs_conn_param {
  421. struct net *net;
  422. const union nf_inet_addr *caddr;
  423. const union nf_inet_addr *vaddr;
  424. __be16 cport;
  425. __be16 vport;
  426. __u16 protocol;
  427. u16 af;
  428. const struct ip_vs_pe *pe;
  429. char *pe_data;
  430. __u8 pe_data_len;
  431. };
  432. /*
  433. * IP_VS structure allocated for each dynamically scheduled connection
  434. */
  435. struct ip_vs_conn {
  436. struct hlist_node c_list; /* hashed list heads */
  437. #ifdef CONFIG_NET_NS
  438. struct net *net; /* Name space */
  439. #endif
  440. /* Protocol, addresses and port numbers */
  441. u16 af; /* address family */
  442. __be16 cport;
  443. __be16 vport;
  444. __be16 dport;
  445. __u32 fwmark; /* Fire wall mark from skb */
  446. union nf_inet_addr caddr; /* client address */
  447. union nf_inet_addr vaddr; /* virtual address */
  448. union nf_inet_addr daddr; /* destination address */
  449. volatile __u32 flags; /* status flags */
  450. __u16 protocol; /* Which protocol (TCP/UDP) */
  451. /* counter and timer */
  452. atomic_t refcnt; /* reference count */
  453. struct timer_list timer; /* Expiration timer */
  454. volatile unsigned long timeout; /* timeout */
  455. /* Flags and state transition */
  456. spinlock_t lock; /* lock for state transition */
  457. volatile __u16 state; /* state info */
  458. volatile __u16 old_state; /* old state, to be used for
  459. * state transition triggerd
  460. * synchronization
  461. */
  462. /* Control members */
  463. struct ip_vs_conn *control; /* Master control connection */
  464. atomic_t n_control; /* Number of controlled ones */
  465. struct ip_vs_dest *dest; /* real server */
  466. atomic_t in_pkts; /* incoming packet counter */
  467. /* packet transmitter for different forwarding methods. If it
  468. mangles the packet, it must return NF_DROP or better NF_STOLEN,
  469. otherwise this must be changed to a sk_buff **.
  470. NF_ACCEPT can be returned when destination is local.
  471. */
  472. int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
  473. struct ip_vs_protocol *pp);
  474. /* Note: we can group the following members into a structure,
  475. in order to save more space, and the following members are
  476. only used in VS/NAT anyway */
  477. struct ip_vs_app *app; /* bound ip_vs_app object */
  478. void *app_data; /* Application private data */
  479. struct ip_vs_seq in_seq; /* incoming seq. struct */
  480. struct ip_vs_seq out_seq; /* outgoing seq. struct */
  481. const struct ip_vs_pe *pe;
  482. char *pe_data;
  483. __u8 pe_data_len;
  484. };
  485. /*
  486. * To save some memory in conn table when name space is disabled.
  487. */
  488. static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
  489. {
  490. #ifdef CONFIG_NET_NS
  491. return cp->net;
  492. #else
  493. return &init_net;
  494. #endif
  495. }
  496. static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
  497. {
  498. #ifdef CONFIG_NET_NS
  499. cp->net = net;
  500. #endif
  501. }
  502. static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
  503. struct net *net)
  504. {
  505. #ifdef CONFIG_NET_NS
  506. return cp->net == net;
  507. #else
  508. return 1;
  509. #endif
  510. }
  511. /*
  512. * Extended internal versions of struct ip_vs_service_user and
  513. * ip_vs_dest_user for IPv6 support.
  514. *
  515. * We need these to conveniently pass around service and destination
  516. * options, but unfortunately, we also need to keep the old definitions to
  517. * maintain userspace backwards compatibility for the setsockopt interface.
  518. */
  519. struct ip_vs_service_user_kern {
  520. /* virtual service addresses */
  521. u16 af;
  522. u16 protocol;
  523. union nf_inet_addr addr; /* virtual ip address */
  524. u16 port;
  525. u32 fwmark; /* firwall mark of service */
  526. /* virtual service options */
  527. char *sched_name;
  528. char *pe_name;
  529. unsigned flags; /* virtual service flags */
  530. unsigned timeout; /* persistent timeout in sec */
  531. u32 netmask; /* persistent netmask */
  532. };
  533. struct ip_vs_dest_user_kern {
  534. /* destination server address */
  535. union nf_inet_addr addr;
  536. u16 port;
  537. /* real server options */
  538. unsigned conn_flags; /* connection flags */
  539. int weight; /* destination weight */
  540. /* thresholds for active connections */
  541. u32 u_threshold; /* upper threshold */
  542. u32 l_threshold; /* lower threshold */
  543. };
  544. /*
  545. * The information about the virtual service offered to the net
  546. * and the forwarding entries
  547. */
  548. struct ip_vs_service {
  549. struct list_head s_list; /* for normal service table */
  550. struct list_head f_list; /* for fwmark-based service table */
  551. atomic_t refcnt; /* reference counter */
  552. atomic_t usecnt; /* use counter */
  553. u16 af; /* address family */
  554. __u16 protocol; /* which protocol (TCP/UDP) */
  555. union nf_inet_addr addr; /* IP address for virtual service */
  556. __be16 port; /* port number for the service */
  557. __u32 fwmark; /* firewall mark of the service */
  558. unsigned flags; /* service status flags */
  559. unsigned timeout; /* persistent timeout in ticks */
  560. __be32 netmask; /* grouping granularity */
  561. struct net *net;
  562. struct list_head destinations; /* real server d-linked list */
  563. __u32 num_dests; /* number of servers */
  564. struct ip_vs_stats stats; /* statistics for the service */
  565. struct ip_vs_app *inc; /* bind conns to this app inc */
  566. /* for scheduling */
  567. struct ip_vs_scheduler *scheduler; /* bound scheduler object */
  568. rwlock_t sched_lock; /* lock sched_data */
  569. void *sched_data; /* scheduler application data */
  570. /* alternate persistence engine */
  571. struct ip_vs_pe *pe;
  572. };
  573. /*
  574. * The real server destination forwarding entry
  575. * with ip address, port number, and so on.
  576. */
  577. struct ip_vs_dest {
  578. struct list_head n_list; /* for the dests in the service */
  579. struct list_head d_list; /* for table with all the dests */
  580. u16 af; /* address family */
  581. __be16 port; /* port number of the server */
  582. union nf_inet_addr addr; /* IP address of the server */
  583. volatile unsigned flags; /* dest status flags */
  584. atomic_t conn_flags; /* flags to copy to conn */
  585. atomic_t weight; /* server weight */
  586. atomic_t refcnt; /* reference counter */
  587. struct ip_vs_stats stats; /* statistics */
  588. /* connection counters and thresholds */
  589. atomic_t activeconns; /* active connections */
  590. atomic_t inactconns; /* inactive connections */
  591. atomic_t persistconns; /* persistent connections */
  592. __u32 u_threshold; /* upper threshold */
  593. __u32 l_threshold; /* lower threshold */
  594. /* for destination cache */
  595. spinlock_t dst_lock; /* lock of dst_cache */
  596. struct dst_entry *dst_cache; /* destination cache entry */
  597. u32 dst_rtos; /* RT_TOS(tos) for dst */
  598. u32 dst_cookie;
  599. #ifdef CONFIG_IP_VS_IPV6
  600. struct in6_addr dst_saddr;
  601. #endif
  602. /* for virtual service */
  603. struct ip_vs_service *svc; /* service it belongs to */
  604. __u16 protocol; /* which protocol (TCP/UDP) */
  605. __be16 vport; /* virtual port number */
  606. union nf_inet_addr vaddr; /* virtual IP address */
  607. __u32 vfwmark; /* firewall mark of service */
  608. };
  609. /*
  610. * The scheduler object
  611. */
  612. struct ip_vs_scheduler {
  613. struct list_head n_list; /* d-linked list head */
  614. char *name; /* scheduler name */
  615. atomic_t refcnt; /* reference counter */
  616. struct module *module; /* THIS_MODULE/NULL */
  617. /* scheduler initializing service */
  618. int (*init_service)(struct ip_vs_service *svc);
  619. /* scheduling service finish */
  620. int (*done_service)(struct ip_vs_service *svc);
  621. /* scheduler updating service */
  622. int (*update_service)(struct ip_vs_service *svc);
  623. /* selecting a server from the given service */
  624. struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
  625. const struct sk_buff *skb);
  626. };
  627. /* The persistence engine object */
  628. struct ip_vs_pe {
  629. struct list_head n_list; /* d-linked list head */
  630. char *name; /* scheduler name */
  631. atomic_t refcnt; /* reference counter */
  632. struct module *module; /* THIS_MODULE/NULL */
  633. /* get the connection template, if any */
  634. int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
  635. bool (*ct_match)(const struct ip_vs_conn_param *p,
  636. struct ip_vs_conn *ct);
  637. u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
  638. bool inverse);
  639. int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
  640. };
  641. /*
  642. * The application module object (a.k.a. app incarnation)
  643. */
  644. struct ip_vs_app {
  645. struct list_head a_list; /* member in app list */
  646. int type; /* IP_VS_APP_TYPE_xxx */
  647. char *name; /* application module name */
  648. __u16 protocol;
  649. struct module *module; /* THIS_MODULE/NULL */
  650. struct list_head incs_list; /* list of incarnations */
  651. /* members for application incarnations */
  652. struct list_head p_list; /* member in proto app list */
  653. struct ip_vs_app *app; /* its real application */
  654. __be16 port; /* port number in net order */
  655. atomic_t usecnt; /* usage counter */
  656. /*
  657. * output hook: Process packet in inout direction, diff set for TCP.
  658. * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
  659. * 2=Mangled but checksum was not updated
  660. */
  661. int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
  662. struct sk_buff *, int *diff);
  663. /*
  664. * input hook: Process packet in outin direction, diff set for TCP.
  665. * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
  666. * 2=Mangled but checksum was not updated
  667. */
  668. int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
  669. struct sk_buff *, int *diff);
  670. /* ip_vs_app initializer */
  671. int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
  672. /* ip_vs_app finish */
  673. int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
  674. /* not used now */
  675. int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
  676. struct ip_vs_protocol *);
  677. void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
  678. int * timeout_table;
  679. int * timeouts;
  680. int timeouts_size;
  681. int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
  682. int *verdict, struct ip_vs_conn **cpp);
  683. struct ip_vs_conn *
  684. (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
  685. const struct iphdr *iph, unsigned int proto_off,
  686. int inverse);
  687. struct ip_vs_conn *
  688. (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
  689. const struct iphdr *iph, unsigned int proto_off,
  690. int inverse);
  691. int (*state_transition)(struct ip_vs_conn *cp, int direction,
  692. const struct sk_buff *skb,
  693. struct ip_vs_app *app);
  694. void (*timeout_change)(struct ip_vs_app *app, int flags);
  695. };
  696. /* IPVS in network namespace */
  697. struct netns_ipvs {
  698. int gen; /* Generation */
  699. /*
  700. * Hash table: for real service lookups
  701. */
  702. #define IP_VS_RTAB_BITS 4
  703. #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
  704. #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
  705. struct list_head rs_table[IP_VS_RTAB_SIZE];
  706. /* ip_vs_app */
  707. struct list_head app_list;
  708. struct mutex app_mutex;
  709. struct lock_class_key app_key; /* mutex debuging */
  710. /* ip_vs_proto */
  711. #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
  712. struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
  713. /* ip_vs_proto_tcp */
  714. #ifdef CONFIG_IP_VS_PROTO_TCP
  715. #define TCP_APP_TAB_BITS 4
  716. #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
  717. #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
  718. struct list_head tcp_apps[TCP_APP_TAB_SIZE];
  719. spinlock_t tcp_app_lock;
  720. #endif
  721. /* ip_vs_proto_udp */
  722. #ifdef CONFIG_IP_VS_PROTO_UDP
  723. #define UDP_APP_TAB_BITS 4
  724. #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
  725. #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
  726. struct list_head udp_apps[UDP_APP_TAB_SIZE];
  727. spinlock_t udp_app_lock;
  728. #endif
  729. /* ip_vs_proto_sctp */
  730. #ifdef CONFIG_IP_VS_PROTO_SCTP
  731. #define SCTP_APP_TAB_BITS 4
  732. #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
  733. #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
  734. /* Hash table for SCTP application incarnations */
  735. struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
  736. spinlock_t sctp_app_lock;
  737. #endif
  738. /* ip_vs_conn */
  739. atomic_t conn_count; /* connection counter */
  740. /* ip_vs_ctl */
  741. struct ip_vs_stats tot_stats; /* Statistics & est. */
  742. seqcount_t *ustats_seq; /* u64 read retry */
  743. int num_services; /* no of virtual services */
  744. /* 1/rate drop and drop-entry variables */
  745. struct delayed_work defense_work; /* Work handler */
  746. int drop_rate;
  747. int drop_counter;
  748. atomic_t dropentry;
  749. /* locks in ctl.c */
  750. spinlock_t dropentry_lock; /* drop entry handling */
  751. spinlock_t droppacket_lock; /* drop packet handling */
  752. spinlock_t securetcp_lock; /* state and timeout tables */
  753. rwlock_t rs_lock; /* real services table */
  754. /* semaphore for IPVS sockopts. And, [gs]etsockopt may sleep. */
  755. struct lock_class_key ctl_key; /* ctl_mutex debuging */
  756. /* Trash for destinations */
  757. struct list_head dest_trash;
  758. /* Service counters */
  759. atomic_t ftpsvc_counter;
  760. atomic_t nullsvc_counter;
  761. /* sys-ctl struct */
  762. struct ctl_table_header *sysctl_hdr;
  763. struct ctl_table *sysctl_tbl;
  764. /* sysctl variables */
  765. int sysctl_amemthresh;
  766. int sysctl_am_droprate;
  767. int sysctl_drop_entry;
  768. int sysctl_drop_packet;
  769. int sysctl_secure_tcp;
  770. #ifdef CONFIG_IP_VS_NFCT
  771. int sysctl_conntrack;
  772. #endif
  773. int sysctl_snat_reroute;
  774. int sysctl_sync_ver;
  775. int sysctl_cache_bypass;
  776. int sysctl_expire_nodest_conn;
  777. int sysctl_expire_quiescent_template;
  778. int sysctl_sync_threshold[2];
  779. int sysctl_nat_icmp_send;
  780. /* ip_vs_lblc */
  781. int sysctl_lblc_expiration;
  782. struct ctl_table_header *lblc_ctl_header;
  783. struct ctl_table *lblc_ctl_table;
  784. /* ip_vs_lblcr */
  785. int sysctl_lblcr_expiration;
  786. struct ctl_table_header *lblcr_ctl_header;
  787. struct ctl_table *lblcr_ctl_table;
  788. /* ip_vs_est */
  789. struct list_head est_list; /* estimator list */
  790. spinlock_t est_lock;
  791. struct timer_list est_timer; /* Estimation timer */
  792. /* ip_vs_sync */
  793. struct list_head sync_queue;
  794. spinlock_t sync_lock;
  795. struct ip_vs_sync_buff *sync_buff;
  796. spinlock_t sync_buff_lock;
  797. struct sockaddr_in sync_mcast_addr;
  798. struct task_struct *master_thread;
  799. struct task_struct *backup_thread;
  800. int send_mesg_maxlen;
  801. int recv_mesg_maxlen;
  802. volatile int sync_state;
  803. volatile int master_syncid;
  804. volatile int backup_syncid;
  805. /* multicast interface name */
  806. char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
  807. char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
  808. /* net name space ptr */
  809. struct net *net; /* Needed by timer routines */
  810. };
  811. /*
  812. * IPVS core functions
  813. * (from ip_vs_core.c)
  814. */
  815. extern const char *ip_vs_proto_name(unsigned proto);
  816. extern void ip_vs_init_hash_table(struct list_head *table, int rows);
  817. #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
  818. #define IP_VS_APP_TYPE_FTP 1
  819. /*
  820. * ip_vs_conn handling functions
  821. * (from ip_vs_conn.c)
  822. */
  823. enum {
  824. IP_VS_DIR_INPUT = 0,
  825. IP_VS_DIR_OUTPUT,
  826. IP_VS_DIR_INPUT_ONLY,
  827. IP_VS_DIR_LAST,
  828. };
  829. static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
  830. const union nf_inet_addr *caddr,
  831. __be16 cport,
  832. const union nf_inet_addr *vaddr,
  833. __be16 vport,
  834. struct ip_vs_conn_param *p)
  835. {
  836. p->net = net;
  837. p->af = af;
  838. p->protocol = protocol;
  839. p->caddr = caddr;
  840. p->cport = cport;
  841. p->vaddr = vaddr;
  842. p->vport = vport;
  843. p->pe = NULL;
  844. p->pe_data = NULL;
  845. }
  846. struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
  847. struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
  848. struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
  849. const struct ip_vs_iphdr *iph,
  850. unsigned int proto_off,
  851. int inverse);
  852. struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
  853. struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
  854. const struct ip_vs_iphdr *iph,
  855. unsigned int proto_off,
  856. int inverse);
  857. /* put back the conn without restarting its timer */
  858. static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
  859. {
  860. atomic_dec(&cp->refcnt);
  861. }
  862. extern void ip_vs_conn_put(struct ip_vs_conn *cp);
  863. extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
  864. struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
  865. const union nf_inet_addr *daddr,
  866. __be16 dport, unsigned flags,
  867. struct ip_vs_dest *dest, __u32 fwmark);
  868. extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
  869. extern const char * ip_vs_state_name(__u16 proto, int state);
  870. extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
  871. extern int ip_vs_check_template(struct ip_vs_conn *ct);
  872. extern void ip_vs_random_dropentry(struct net *net);
  873. extern int ip_vs_conn_init(void);
  874. extern void ip_vs_conn_cleanup(void);
  875. static inline void ip_vs_control_del(struct ip_vs_conn *cp)
  876. {
  877. struct ip_vs_conn *ctl_cp = cp->control;
  878. if (!ctl_cp) {
  879. IP_VS_ERR_BUF("request control DEL for uncontrolled: "
  880. "%s:%d to %s:%d\n",
  881. IP_VS_DBG_ADDR(cp->af, &cp->caddr),
  882. ntohs(cp->cport),
  883. IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
  884. ntohs(cp->vport));
  885. return;
  886. }
  887. IP_VS_DBG_BUF(7, "DELeting control for: "
  888. "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
  889. IP_VS_DBG_ADDR(cp->af, &cp->caddr),
  890. ntohs(cp->cport),
  891. IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
  892. ntohs(ctl_cp->cport));
  893. cp->control = NULL;
  894. if (atomic_read(&ctl_cp->n_control) == 0) {
  895. IP_VS_ERR_BUF("BUG control DEL with n=0 : "
  896. "%s:%d to %s:%d\n",
  897. IP_VS_DBG_ADDR(cp->af, &cp->caddr),
  898. ntohs(cp->cport),
  899. IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
  900. ntohs(cp->vport));
  901. return;
  902. }
  903. atomic_dec(&ctl_cp->n_control);
  904. }
  905. static inline void
  906. ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
  907. {
  908. if (cp->control) {
  909. IP_VS_ERR_BUF("request control ADD for already controlled: "
  910. "%s:%d to %s:%d\n",
  911. IP_VS_DBG_ADDR(cp->af, &cp->caddr),
  912. ntohs(cp->cport),
  913. IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
  914. ntohs(cp->vport));
  915. ip_vs_control_del(cp);
  916. }
  917. IP_VS_DBG_BUF(7, "ADDing control for: "
  918. "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
  919. IP_VS_DBG_ADDR(cp->af, &cp->caddr),
  920. ntohs(cp->cport),
  921. IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
  922. ntohs(ctl_cp->cport));
  923. cp->control = ctl_cp;
  924. atomic_inc(&ctl_cp->n_control);
  925. }
  926. /*
  927. * IPVS application functions
  928. * (from ip_vs_app.c)
  929. */
  930. #define IP_VS_APP_MAX_PORTS 8
  931. extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
  932. extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
  933. extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  934. extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
  935. extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
  936. __u16 proto, __u16 port);
  937. extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
  938. extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
  939. extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
  940. extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
  941. extern int ip_vs_app_init(void);
  942. extern void ip_vs_app_cleanup(void);
  943. void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
  944. void ip_vs_unbind_pe(struct ip_vs_service *svc);
  945. int register_ip_vs_pe(struct ip_vs_pe *pe);
  946. int unregister_ip_vs_pe(struct ip_vs_pe *pe);
  947. struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
  948. struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
  949. static inline void ip_vs_pe_get(const struct ip_vs_pe *pe)
  950. {
  951. if (pe && pe->module)
  952. __module_get(pe->module);
  953. }
  954. static inline void ip_vs_pe_put(const struct ip_vs_pe *pe)
  955. {
  956. if (pe && pe->module)
  957. module_put(pe->module);
  958. }
  959. /*
  960. * IPVS protocol functions (from ip_vs_proto.c)
  961. */
  962. extern int ip_vs_protocol_init(void);
  963. extern void ip_vs_protocol_cleanup(void);
  964. extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
  965. extern int *ip_vs_create_timeout_table(int *table, int size);
  966. extern int
  967. ip_vs_set_state_timeout(int *table, int num, const char *const *names,
  968. const char *name, int to);
  969. extern void
  970. ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
  971. const struct sk_buff *skb,
  972. int offset, const char *msg);
  973. extern struct ip_vs_protocol ip_vs_protocol_tcp;
  974. extern struct ip_vs_protocol ip_vs_protocol_udp;
  975. extern struct ip_vs_protocol ip_vs_protocol_icmp;
  976. extern struct ip_vs_protocol ip_vs_protocol_esp;
  977. extern struct ip_vs_protocol ip_vs_protocol_ah;
  978. extern struct ip_vs_protocol ip_vs_protocol_sctp;
  979. /*
  980. * Registering/unregistering scheduler functions
  981. * (from ip_vs_sched.c)
  982. */
  983. extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
  984. extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
  985. extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
  986. struct ip_vs_scheduler *scheduler);
  987. extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
  988. extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
  989. extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
  990. extern struct ip_vs_conn *
  991. ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
  992. struct ip_vs_proto_data *pd, int *ignored);
  993. extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
  994. struct ip_vs_proto_data *pd);
  995. extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
  996. /*
  997. * IPVS control data and functions (from ip_vs_ctl.c)
  998. */
  999. extern struct ip_vs_stats ip_vs_stats;
  1000. extern const struct ctl_path net_vs_ctl_path[];
  1001. extern int sysctl_ip_vs_sync_ver;
  1002. extern void ip_vs_sync_switch_mode(struct net *net, int mode);
  1003. extern struct ip_vs_service *
  1004. ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
  1005. const union nf_inet_addr *vaddr, __be16 vport);
  1006. static inline void ip_vs_service_put(struct ip_vs_service *svc)
  1007. {
  1008. atomic_dec(&svc->usecnt);
  1009. }
  1010. extern struct ip_vs_dest *
  1011. ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
  1012. const union nf_inet_addr *daddr, __be16 dport);
  1013. extern int ip_vs_use_count_inc(void);
  1014. extern void ip_vs_use_count_dec(void);
  1015. extern int ip_vs_control_init(void);
  1016. extern void ip_vs_control_cleanup(void);
  1017. extern struct ip_vs_dest *
  1018. ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
  1019. __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
  1020. __u16 protocol, __u32 fwmark);
  1021. extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
  1022. /*
  1023. * IPVS sync daemon data and function prototypes
  1024. * (from ip_vs_sync.c)
  1025. */
  1026. extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
  1027. __u8 syncid);
  1028. extern int stop_sync_thread(struct net *net, int state);
  1029. extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
  1030. extern int ip_vs_sync_init(void);
  1031. extern void ip_vs_sync_cleanup(void);
  1032. /*
  1033. * IPVS rate estimator prototypes (from ip_vs_est.c)
  1034. */
  1035. extern int ip_vs_estimator_init(void);
  1036. extern void ip_vs_estimator_cleanup(void);
  1037. extern void ip_vs_new_estimator(struct net *net, struct ip_vs_stats *stats);
  1038. extern void ip_vs_kill_estimator(struct net *net, struct ip_vs_stats *stats);
  1039. extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
  1040. /*
  1041. * Various IPVS packet transmitters (from ip_vs_xmit.c)
  1042. */
  1043. extern int ip_vs_null_xmit
  1044. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1045. extern int ip_vs_bypass_xmit
  1046. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1047. extern int ip_vs_nat_xmit
  1048. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1049. extern int ip_vs_tunnel_xmit
  1050. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1051. extern int ip_vs_dr_xmit
  1052. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1053. extern int ip_vs_icmp_xmit
  1054. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp, int offset);
  1055. extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
  1056. #ifdef CONFIG_IP_VS_IPV6
  1057. extern int ip_vs_bypass_xmit_v6
  1058. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1059. extern int ip_vs_nat_xmit_v6
  1060. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1061. extern int ip_vs_tunnel_xmit_v6
  1062. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1063. extern int ip_vs_dr_xmit_v6
  1064. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
  1065. extern int ip_vs_icmp_xmit_v6
  1066. (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
  1067. int offset);
  1068. #endif
  1069. /*
  1070. * This is a simple mechanism to ignore packets when
  1071. * we are loaded. Just set ip_vs_drop_rate to 'n' and
  1072. * we start to drop 1/rate of the packets
  1073. */
  1074. static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
  1075. {
  1076. if (!ipvs->drop_rate)
  1077. return 0;
  1078. if (--ipvs->drop_counter > 0)
  1079. return 0;
  1080. ipvs->drop_counter = ipvs->drop_rate;
  1081. return 1;
  1082. }
  1083. /*
  1084. * ip_vs_fwd_tag returns the forwarding tag of the connection
  1085. */
  1086. #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
  1087. static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
  1088. {
  1089. char fwd;
  1090. switch (IP_VS_FWD_METHOD(cp)) {
  1091. case IP_VS_CONN_F_MASQ:
  1092. fwd = 'M'; break;
  1093. case IP_VS_CONN_F_LOCALNODE:
  1094. fwd = 'L'; break;
  1095. case IP_VS_CONN_F_TUNNEL:
  1096. fwd = 'T'; break;
  1097. case IP_VS_CONN_F_DROUTE:
  1098. fwd = 'R'; break;
  1099. case IP_VS_CONN_F_BYPASS:
  1100. fwd = 'B'; break;
  1101. default:
  1102. fwd = '?'; break;
  1103. }
  1104. return fwd;
  1105. }
  1106. extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
  1107. struct ip_vs_conn *cp, int dir);
  1108. #ifdef CONFIG_IP_VS_IPV6
  1109. extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
  1110. struct ip_vs_conn *cp, int dir);
  1111. #endif
  1112. extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
  1113. static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
  1114. {
  1115. __be32 diff[2] = { ~old, new };
  1116. return csum_partial(diff, sizeof(diff), oldsum);
  1117. }
  1118. #ifdef CONFIG_IP_VS_IPV6
  1119. static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
  1120. __wsum oldsum)
  1121. {
  1122. __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
  1123. new[3], new[2], new[1], new[0] };
  1124. return csum_partial(diff, sizeof(diff), oldsum);
  1125. }
  1126. #endif
  1127. static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
  1128. {
  1129. __be16 diff[2] = { ~old, new };
  1130. return csum_partial(diff, sizeof(diff), oldsum);
  1131. }
  1132. /*
  1133. * Forget current conntrack (unconfirmed) and attach notrack entry
  1134. */
  1135. static inline void ip_vs_notrack(struct sk_buff *skb)
  1136. {
  1137. #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
  1138. enum ip_conntrack_info ctinfo;
  1139. struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
  1140. if (!ct || !nf_ct_is_untracked(ct)) {
  1141. nf_reset(skb);
  1142. skb->nfct = &nf_ct_untracked_get()->ct_general;
  1143. skb->nfctinfo = IP_CT_NEW;
  1144. nf_conntrack_get(skb->nfct);
  1145. }
  1146. #endif
  1147. }
  1148. #ifdef CONFIG_IP_VS_NFCT
  1149. /*
  1150. * Netfilter connection tracking
  1151. * (from ip_vs_nfct.c)
  1152. */
  1153. static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
  1154. {
  1155. return ipvs->sysctl_conntrack;
  1156. }
  1157. extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
  1158. int outin);
  1159. extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
  1160. extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
  1161. struct ip_vs_conn *cp, u_int8_t proto,
  1162. const __be16 port, int from_rs);
  1163. extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
  1164. #else
  1165. static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
  1166. {
  1167. return 0;
  1168. }
  1169. static inline void ip_vs_update_conntrack(struct sk_buff *skb,
  1170. struct ip_vs_conn *cp, int outin)
  1171. {
  1172. }
  1173. static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
  1174. struct ip_vs_conn *cp)
  1175. {
  1176. return NF_ACCEPT;
  1177. }
  1178. static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
  1179. {
  1180. }
  1181. /* CONFIG_IP_VS_NFCT */
  1182. #endif
  1183. static inline unsigned int
  1184. ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
  1185. {
  1186. /*
  1187. * We think the overhead of processing active connections is 256
  1188. * times higher than that of inactive connections in average. (This
  1189. * 256 times might not be accurate, we will change it later) We
  1190. * use the following formula to estimate the overhead now:
  1191. * dest->activeconns*256 + dest->inactconns
  1192. */
  1193. return (atomic_read(&dest->activeconns) << 8) +
  1194. atomic_read(&dest->inactconns);
  1195. }
  1196. #endif /* __KERNEL__ */
  1197. #endif /* _NET_IP_VS_H */