1 /*
2 * IP Virtual Server
3 * data structure and functionality definitions
4 */
5
6 #ifndef _IP_VS_H
7 #define _IP_VS_H
8
9 #include <asm/types.h> /* For __uXX types */
10
11 #define IP_VS_VERSION_CODE 0x01000C
12 #define NVERSION(version) \
13 (version >> 16) & 0xFF, \
14 (version >> 8) & 0xFF, \
15 version & 0xFF
16
17 /*
18 * Virtual Service Flags
19 */
20 #define IP_VS_SVC_F_PERSISTENT 0x0001 /* persistent port */
21 #define IP_VS_SVC_F_HASHED 0x0002 /* hashed entry */
22
23 /*
24 * Destination Server Flags
25 */
26 #define IP_VS_DEST_F_AVAILABLE 0x0001 /* Available tag */
27
28 /*
29 * IPVS sync daemon states
30 */
31 #define IP_VS_STATE_NONE 0 /* daemon is stopped */
32 #define IP_VS_STATE_MASTER 1 /* started as master */
33 #define IP_VS_STATE_BACKUP 2 /* started as backup */
34
35 /*
36 * IPVS socket options
37 */
38 #define IP_VS_BASE_CTL (64+1024+64) /* base */
39
40 #define IP_VS_SO_SET_NONE IP_VS_BASE_CTL /* just peek */
41 #define IP_VS_SO_SET_INSERT (IP_VS_BASE_CTL+1)
42 #define IP_VS_SO_SET_ADD (IP_VS_BASE_CTL+2)
43 #define IP_VS_SO_SET_EDIT (IP_VS_BASE_CTL+3)
44 #define IP_VS_SO_SET_DEL (IP_VS_BASE_CTL+4)
45 #define IP_VS_SO_SET_FLUSH (IP_VS_BASE_CTL+5)
46 #define IP_VS_SO_SET_LIST (IP_VS_BASE_CTL+6)
47 #define IP_VS_SO_SET_ADDDEST (IP_VS_BASE_CTL+7)
48 #define IP_VS_SO_SET_DELDEST (IP_VS_BASE_CTL+8)
49 #define IP_VS_SO_SET_EDITDEST (IP_VS_BASE_CTL+9)
50 #define IP_VS_SO_SET_TIMEOUTS (IP_VS_BASE_CTL+10)
51 #define IP_VS_SO_SET_STARTDAEMON (IP_VS_BASE_CTL+11)
52 #define IP_VS_SO_SET_STOPDAEMON (IP_VS_BASE_CTL+12)
53 #define IP_VS_SO_SET_RESTORE (IP_VS_BASE_CTL+13)
54 #define IP_VS_SO_SET_SAVE (IP_VS_BASE_CTL+14)
55 #define IP_VS_SO_SET_ZERO (IP_VS_BASE_CTL+15)
56 #define IP_VS_SO_SET_MAX IP_VS_SO_SET_ZERO
57
58 #define IP_VS_SO_GET_VERSION IP_VS_BASE_CTL
59 #define IP_VS_SO_GET_INFO (IP_VS_BASE_CTL+1)
60 #define IP_VS_SO_GET_SERVICES (IP_VS_BASE_CTL+2)
61 #define IP_VS_SO_GET_SERVICE (IP_VS_BASE_CTL+3)
62 #define IP_VS_SO_GET_DESTS (IP_VS_BASE_CTL+4)
63 #define IP_VS_SO_GET_DEST (IP_VS_BASE_CTL+5) /* not used now */
64 #define IP_VS_SO_GET_TIMEOUTS (IP_VS_BASE_CTL+6)
65 #define IP_VS_SO_GET_DAEMON (IP_VS_BASE_CTL+7)
66 #define IP_VS_SO_GET_MAX IP_VS_SO_GET_DAEMON
67
68
69 /*
70 * IPVS Connection Flags
71 */
72 #define IP_VS_CONN_F_FWD_MASK 0x0007 /* mask for the fwd methods */
73 #define IP_VS_CONN_F_MASQ 0x0000 /* masquerading */
74 #define IP_VS_CONN_F_LOCALNODE 0x0001 /* local node */
75 #define IP_VS_CONN_F_TUNNEL 0x0002 /* tunneling */
76 #define IP_VS_CONN_F_DROUTE 0x0003 /* direct routing */
77 #define IP_VS_CONN_F_BYPASS 0x0004 /* cache bypass */
78 #define IP_VS_CONN_F_HASHED 0x0040 /* hashed entry */
79 #define IP_VS_CONN_F_NOOUTPUT 0x0080 /* no output packets */
80 #define IP_VS_CONN_F_INACTIVE 0x0100 /* not established */
81 #define IP_VS_CONN_F_OUT_SEQ 0x0200 /* must do output seq adjust */
82 #define IP_VS_CONN_F_IN_SEQ 0x0400 /* must do input seq adjust */
83 #define IP_VS_CONN_F_SEQ_MASK 0x0600 /* in/out sequence mask */
84 #define IP_VS_CONN_F_NO_CPORT 0x0800 /* no client port set yet */
85 #define IP_VS_CONN_F_TEMPLATE 0x1000 /* template, not connection */
86
87 /* Move it to better place one day, for now keep it unique */
88 #define NFC_IPVS_PROPERTY 0x10000
89
90 #define IP_VS_SCHEDNAME_MAXLEN 16
91 #define IP_VS_IFNAME_MAXLEN 16
92
93 struct ip_vs_rule_user {
94 /* global options */
95 int tcp_timeout; /* timeout values */
96 int tcp_fin_timeout;
97 int udp_timeout;
98 int state; /* sync daemon state */
99 char mcast_ifn[IP_VS_IFNAME_MAXLEN];
100 /* multicast interface name */
101 int syncid;
102
103 /* virtual service options */
104 u_int16_t protocol;
105 u_int32_t vaddr; /* virtual address */
106 u_int16_t vport;
107 u_int32_t vfwmark; /* firwall mark of virtual service*/
108 char sched_name[IP_VS_SCHEDNAME_MAXLEN];
109 unsigned vs_flags; /* virtual service flags */
110 unsigned timeout; /* persistent timeout in ticks */
111 u_int32_t netmask; /* persistent netmask */
112
113 /* destination specific options */
114 u_int32_t daddr; /* destination address */
115 u_int16_t dport;
116 unsigned conn_flags; /* destination flags */
117 int weight; /* destination weight */
118 };
119
120
121 /*
122 * IPVS statistics object (for user space)
123 */
124 struct ip_vs_stats_user
125 {
126 __u32 conns; /* connections scheduled */
127 __u32 inpkts; /* incoming packets */
128 __u32 outpkts; /* outgoing packets */
129 __u64 inbytes; /* incoming bytes */
130 __u64 outbytes; /* outgoing bytes */
131
132 __u32 cps; /* current connection rate */
133 __u32 inpps; /* current in packet rate */
134 __u32 outpps; /* current out packet rate */
135 __u32 inbps; /* current in byte rate */
136 __u32 outbps; /* current out byte rate */
137 };
138
139
140 /* The argument to IP_VS_SO_GET_INFO */
141 struct ip_vs_getinfo {
142 /* version number */
143 unsigned int version;
144
145 /* size of connection hash table */
146 unsigned int size;
147
148 /* number of virtual services */
149 unsigned int num_services;
150 };
151
152 /* The argument to IP_VS_SO_GET_SERVICE */
153 struct ip_vs_service_user {
154 /* which service: user fills this in */
155 u_int16_t protocol;
156 u_int32_t addr; /* virtual address */
157 u_int16_t port;
158 u_int32_t fwmark; /* firwall mark of virtual service */
159
160 /* service options */
161 char sched_name[IP_VS_SCHEDNAME_MAXLEN];
162 unsigned flags; /* virtual service flags */
163 unsigned timeout; /* persistent timeout in ticks */
164 u_int32_t netmask; /* persistent netmask */
165
166 /* number of real servers */
167 unsigned int num_dests;
168
169 /* statistics */
170 struct ip_vs_stats_user stats;
171 };
172
173 struct ip_vs_dest_user {
174 u_int32_t addr; /* destination address */
175 u_int16_t port;
176 unsigned flags; /* destination flags */
177 int weight; /* destination weight */
178 u_int32_t activeconns; /* active connections */
179 u_int32_t inactconns; /* inactive connections */
180
181 /* statistics */
182 struct ip_vs_stats_user stats;
183 };
184
185 /* The argument to IP_VS_SO_GET_DESTS */
186 struct ip_vs_get_dests {
187 /* which service: user fills this in */
188 u_int16_t protocol;
189 u_int32_t addr; /* virtual address */
190 u_int16_t port;
191 u_int32_t fwmark; /* firwall mark of virtual service */
192
193 /* number of real servers */
194 unsigned int num_dests;
195
196 /* the real servers */
197 struct ip_vs_dest_user entrytable[0];
198 };
199
200 /* The argument to IP_VS_SO_GET_SERVICES */
201 struct ip_vs_get_services {
202 /* number of virtual services */
203 unsigned int num_services;
204
205 /* service table */
206 struct ip_vs_service_user entrytable[0];
207 };
208
209 /* The argument to IP_VS_SO_GET_TIMEOUTS */
210 struct ip_vs_timeout_user {
211 int tcp_timeout;
212 int tcp_fin_timeout;
213 int udp_timeout;
214 };
215
216 /* The argument to IP_VS_SO_GET_DAEMON */
217 struct ip_vs_daemon_user {
218 int state; /* sync daemon state */
219 char mcast_master_ifn[IP_VS_IFNAME_MAXLEN]; /* mcast master interface name */
220 char mcast_backup_ifn[IP_VS_IFNAME_MAXLEN]; /* mcast backup interface name */
221 };
222
223
224 #ifdef __KERNEL__
225
226 #include <linux/config.h>
227 #include <linux/list.h> /* for struct list_head */
228 #include <linux/spinlock.h> /* for struct rwlock_t */
229 #include <linux/skbuff.h> /* for struct sk_buff */
230 #include <linux/ip.h> /* for struct iphdr */
231 #include <asm/atomic.h> /* for struct atomic_t */
232 #include <linux/netdevice.h> /* for struct neighbour; */
233 #include <net/dst.h> /* for struct dst_entry */
234 #include <net/route.h> /* for ip_route_output */
235 #include <net/tcp.h>
236 #include <net/udp.h>
237
238
239 #ifdef CONFIG_IP_VS_DEBUG
240 extern int ip_vs_get_debug_level(void);
241 #define IP_VS_DBG(level, msg...) \
242 do { \
243 if (level <= ip_vs_get_debug_level()) \
244 printk(KERN_DEBUG "IPVS: " msg); \
245 } while (0)
246 #define IP_VS_DBG_RL(msg...) \
247 do { \
248 if (net_ratelimit()) \
249 printk(KERN_DEBUG "IPVS: " msg); \
250 } while (0)
251 #else /* NO DEBUGGING at ALL */
252 #define IP_VS_DBG(level, msg...) do {} while (0)
253 #define IP_VS_DBG_RL(msg...) do {} while (0)
254 #endif
255
256 #define IP_VS_BUG() BUG()
257 #define IP_VS_ERR(msg...) printk(KERN_ERR "IPVS: " msg)
258 #define IP_VS_INFO(msg...) printk(KERN_INFO "IPVS: " msg)
259 #define IP_VS_WARNING(msg...) \
260 printk(KERN_WARNING "IPVS: " msg)
261 #define IP_VS_ERR_RL(msg...) \
262 do { \
263 if (net_ratelimit()) \
264 printk(KERN_ERR "IPVS: " msg); \
265 } while (0)
266
267 #ifdef CONFIG_IP_VS_DEBUG
268 #define EnterFunction(level) \
269 do { \
270 if (level <= ip_vs_get_debug_level()) \
271 printk(KERN_DEBUG "Enter: %s, %s line %i\n", \
272 __FUNCTION__, __FILE__, __LINE__); \
273 } while (0)
274 #define LeaveFunction(level) \
275 do { \
276 if (level <= ip_vs_get_debug_level()) \
277 printk(KERN_DEBUG "Leave: %s, %s line %i\n", \
278 __FUNCTION__, __FILE__, __LINE__); \
279 } while (0)
280 #else
281 #define EnterFunction(level) do {} while (0)
282 #define LeaveFunction(level) do {} while (0)
283 #endif
284
285
286 /*
287 * The port number of FTP service (in network order).
288 */
289 #define FTPPORT __constant_htons(21)
290 #define FTPDATA __constant_htons(20)
291
292
293 /*
294 * IPVS sysctl variables under the /proc/sys/net/ipv4/vs/
295 */
296 #define NET_IPV4_VS 21
297
298 enum {
299 NET_IPV4_VS_DEBUG_LEVEL=1,
300 NET_IPV4_VS_AMEMTHRESH=2,
301 NET_IPV4_VS_AMDROPRATE=3,
302 NET_IPV4_VS_DROP_ENTRY=4,
303 NET_IPV4_VS_DROP_PACKET=5,
304 NET_IPV4_VS_SECURE_TCP=6,
305 NET_IPV4_VS_TO_ES=7,
306 NET_IPV4_VS_TO_SS=8,
307 NET_IPV4_VS_TO_SR=9,
308 NET_IPV4_VS_TO_FW=10,
309 NET_IPV4_VS_TO_TW=11,
310 NET_IPV4_VS_TO_CL=12,
311 NET_IPV4_VS_TO_CW=13,
312 NET_IPV4_VS_TO_LA=14,
313 NET_IPV4_VS_TO_LI=15,
314 NET_IPV4_VS_TO_SA=16,
315 NET_IPV4_VS_TO_UDP=17,
316 NET_IPV4_VS_TO_ICMP=18,
317 NET_IPV4_VS_LBLC_EXPIRE=19,
318 NET_IPV4_VS_LBLCR_EXPIRE=20,
319 NET_IPV4_VS_CACHE_BYPASS=22,
320 NET_IPV4_VS_EXPIRE_NODEST_CONN=23,
321 NET_IPV4_VS_SYNC_THRESHOLD=24,
322 NET_IPV4_VS_NAT_ICMP_SEND=25,
323 NET_IPV4_VS_EXPIRE_QUIESCENT_TEMPLATE=26,
324 NET_IPV4_VS_LAST
325 };
326
327
328 /*
329 * IPVS State Values
330 */
331 enum {
332 IP_VS_S_NONE = 0,
333 IP_VS_S_ESTABLISHED,
334 IP_VS_S_SYN_SENT,
335 IP_VS_S_SYN_RECV,
336 IP_VS_S_FIN_WAIT,
337 IP_VS_S_TIME_WAIT,
338 IP_VS_S_CLOSE,
339 IP_VS_S_CLOSE_WAIT,
340 IP_VS_S_LAST_ACK,
341 IP_VS_S_LISTEN,
342 IP_VS_S_SYNACK,
343 IP_VS_S_UDP,
344 IP_VS_S_ICMP,
345 IP_VS_S_LAST
346 };
347
348
349 struct ip_vs_timeout_table {
350 atomic_t refcnt;
351 int scale;
352 int timeout[IP_VS_S_LAST+1];
353 };
354
355
356 /*
357 * Transport protocol header
358 */
359 union ip_vs_tphdr {
360 unsigned char *raw;
361 struct udphdr *uh;
362 struct tcphdr *th;
363 struct icmphdr *icmph;
364 __u16 *portp;
365 };
366
367
368 /*
369 * Delta sequence info structure
370 * Each ip_vs_conn has 2 (output AND input seq. changes).
371 * Only used in the VS/NAT.
372 */
373 struct ip_vs_seq {
374 __u32 init_seq; /* Add delta from this seq */
375 __u32 delta; /* Delta in sequence numbers */
376 __u32 previous_delta; /* Delta in sequence numbers
377 before last resized pkt */
378 };
379
380
381 /*
382 * IPVS statistics object
383 */
384 struct ip_vs_stats
385 {
386 __u32 conns; /* connections scheduled */
387 __u32 inpkts; /* incoming packets */
388 __u32 outpkts; /* outgoing packets */
389 __u64 inbytes; /* incoming bytes */
390 __u64 outbytes; /* outgoing bytes */
391
392 __u32 cps; /* current connection rate */
393 __u32 inpps; /* current in packet rate */
394 __u32 outpps; /* current out packet rate */
395 __u32 inbps; /* current in byte rate */
396 __u32 outbps; /* current out byte rate */
397
398 spinlock_t lock; /* spin lock */
399 };
400
401
402 /*
403 * IP_VS structure allocated for each dynamically scheduled connection
404 */
405 struct ip_vs_conn {
406 struct list_head c_list; /* hashed list heads */
407
408 /* Protocol, addresses and port numbers */
409 __u32 caddr; /* client address */
410 __u32 vaddr; /* virtual address */
411 __u32 daddr; /* destination address */
412 __u16 cport;
413 __u16 vport;
414 __u16 dport;
415 __u16 protocol; /* Which protocol (TCP/UDP) */
416
417 /* counter and timer */
418 atomic_t refcnt; /* reference count */
419 struct timer_list timer; /* Expiration timer */
420 volatile unsigned long timeout; /* timeout */
421 struct ip_vs_timeout_table *timeout_table;
422
423 /* Flags and state transition */
424 spinlock_t lock; /* lock for state transition */
425 volatile __u16 flags; /* status flags */
426 volatile __u16 state; /* state info */
427
428 /* Control members */
429 struct ip_vs_conn *control; /* Master control connection */
430 atomic_t n_control; /* Number of controlled ones */
431 struct ip_vs_dest *dest; /* real server */
432 atomic_t in_pkts; /* incoming packet counter */
433
434 /* packet transmitter for different forwarding methods */
435 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp);
436
437 /* Note: we can group the following members into a structure,
438 in order to save more space, and the following members are
439 only used in VS/NAT anyway */
440 struct ip_vs_app *app; /* bound ip_vs_app object */
441 void *app_data; /* Application private data */
442 struct ip_vs_seq in_seq; /* incoming seq. struct */
443 struct ip_vs_seq out_seq; /* outgoing seq. struct */
444 };
445
446
447 /*
448 * The information about the virtual service offered to the net
449 * and the forwarding entries
450 */
451 struct ip_vs_service {
452 struct list_head s_list; /* for normal service table */
453 struct list_head f_list; /* for fwmark-based service table */
454 atomic_t refcnt; /* reference counter */
455 atomic_t usecnt; /* use counter */
456
457 __u16 protocol; /* which protocol (TCP/UDP) */
458 __u32 addr; /* IP address for virtual service */
459 __u16 port; /* port number for the service */
460 __u32 fwmark; /* firewall mark of the service */
461 unsigned flags; /* service status flags */
462 unsigned timeout; /* persistent timeout in ticks */
463 __u32 netmask; /* grouping granularity */
464
465 struct list_head destinations; /* real server d-linked list */
466 __u32 num_dests; /* number of servers */
467 struct ip_vs_stats stats; /* statistics for the service */
468
469 /* for scheduling */
470 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
471 rwlock_t sched_lock; /* lock sched_data */
472 void *sched_data; /* scheduler application data */
473 };
474
475
476 /*
477 * The real server destination forwarding entry
478 * with ip address, port number, and so on.
479 */
480 struct ip_vs_dest {
481 struct list_head n_list; /* for the dests in the service */
482 struct list_head d_list; /* for table with all the dests */
483
484 __u32 addr; /* IP address of real server */
485 __u16 port; /* port number of the service */
486 unsigned flags; /* dest status flags */
487 atomic_t weight; /* server weight */
488 atomic_t conn_flags; /* flags to copy to conn */
489 atomic_t activeconns; /* active connections */
490 atomic_t inactconns; /* inactive connections */
491 atomic_t refcnt; /* reference counter */
492 struct ip_vs_stats stats; /* statistics */
493
494 /* for destination cache */
495 spinlock_t dst_lock; /* lock dst_cache */
496 struct dst_entry *dst_cache; /* destination cache entry */
497 u32 dst_rtos; /* RT_TOS(tos) for dst */
498
499 /* for virtual service */
500 struct ip_vs_service *svc; /* service that it belongs to */
501 __u16 protocol; /* which protocol (TCP/UDP) */
502 __u32 vaddr; /* IP address for virtual service */
503 __u16 vport; /* port number for the service */
504 __u32 vfwmark; /* firewall mark of the service */
505 };
506
507
508 /*
509 * The scheduler object
510 */
511 struct ip_vs_scheduler {
512 struct list_head n_list; /* d-linked list head */
513 char *name; /* scheduler name */
514 atomic_t refcnt; /* reference counter */
515 struct module *module; /* THIS_MODULE/NULL */
516
517 /* scheduler initializing service */
518 int (*init_service)(struct ip_vs_service *svc);
519 /* scheduling service finish */
520 int (*done_service)(struct ip_vs_service *svc);
521 /* scheduler updating service */
522 int (*update_service)(struct ip_vs_service *svc);
523
524 /* selecting a server from the given service */
525 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
526 struct iphdr *iph);
527 };
528
529
530 /*
531 * The application module object
532 */
533 struct ip_vs_app
534 {
535 struct list_head n_list; /* d-linked list head */
536 char *name; /* name of application module */
537 unsigned type; /* type = proto<<16 | port
538 (host byte order)*/
539 struct module *module; /* THIS_MODULE/NULL */
540
541 /* ip_vs_app initializer */
542 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
543 /* ip_vs_app finish */
544 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
545 /* output hook */
546 int (*pkt_out)(struct ip_vs_app *,
547 struct ip_vs_conn *, struct sk_buff *);
548 /* input hook */
549 int (*pkt_in)(struct ip_vs_app *,
550 struct ip_vs_conn *, struct sk_buff *);
551 };
552
553
554 /*
555 * IPVS core functions
556 * (from ip_vs_core.c)
557 */
558 extern const char *ip_vs_proto_name(unsigned proto);
559 extern unsigned int check_for_ip_vs_out(struct sk_buff **skb_p,
560 int (*okfn)(struct sk_buff *));
561
562
563 /*
564 * ip_vs_conn handling functions
565 * (from ip_vs_conn.c)
566 */
567
568 /*
569 * IPVS connection entry hash table
570 */
571 #ifndef CONFIG_IP_VS_TAB_BITS
572 #define CONFIG_IP_VS_TAB_BITS 12
573 #endif
574 /* make sure that IP_VS_CONN_TAB_BITS is located in [8, 20] */
575 #if CONFIG_IP_VS_TAB_BITS < 8
576 #define IP_VS_CONN_TAB_BITS 8
577 #endif
578 #if CONFIG_IP_VS_TAB_BITS > 20
579 #define IP_VS_CONN_TAB_BITS 20
580 #endif
581 #if 8 <= CONFIG_IP_VS_TAB_BITS && CONFIG_IP_VS_TAB_BITS <= 20
582 #define IP_VS_CONN_TAB_BITS CONFIG_IP_VS_TAB_BITS
583 #endif
584 #define IP_VS_CONN_TAB_SIZE (1 << IP_VS_CONN_TAB_BITS)
585 #define IP_VS_CONN_TAB_MASK (IP_VS_CONN_TAB_SIZE - 1)
586
587 #define VS_STATE_INPUT 0
588 #define VS_STATE_OUTPUT 4
589 #define VS_STATE_INPUT_ONLY 8
590
591 extern struct ip_vs_timeout_table vs_timeout_table;
592 extern struct ip_vs_timeout_table vs_timeout_table_dos;
593
594 extern struct ip_vs_conn *ip_vs_conn_in_get
595 (int protocol, __u32 s_addr, __u16 s_port, __u32 d_addr, __u16 d_port);
596 extern struct ip_vs_conn *ip_vs_ct_in_get
597 (int protocol, __u32 s_addr, __u16 s_port, __u32 d_addr, __u16 d_port);
598 extern struct ip_vs_conn *ip_vs_conn_out_get
599 (int protocol, __u32 s_addr, __u16 s_port, __u32 d_addr, __u16 d_port);
600
601 /* put back the conn without restarting its timer */
__ip_vs_conn_put(struct ip_vs_conn * cp)602 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
603 {
604 atomic_dec(&cp->refcnt);
605 }
606 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
607
608 extern struct ip_vs_conn *
609 ip_vs_conn_new(int proto, __u32 caddr, __u16 cport, __u32 vaddr, __u16 vport,
610 __u32 daddr, __u16 dport, unsigned flags,
611 struct ip_vs_dest *dest);
612 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
613
614 extern const char * ip_vs_state_name(int state);
615 extern int ip_vs_set_state(struct ip_vs_conn *cp, int state_off,
616 struct iphdr *iph, void *tp);
617 extern int ip_vs_conn_listen(struct ip_vs_conn *cp);
618 extern int ip_vs_check_template(struct ip_vs_conn *ct);
619 extern void ip_vs_secure_tcp_set(int on);
620 extern void ip_vs_random_dropentry(void);
621 extern int ip_vs_conn_init(void);
622 extern void ip_vs_conn_cleanup(void);
623
ip_vs_control_del(struct ip_vs_conn * cp)624 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
625 {
626 struct ip_vs_conn *ctl_cp = cp->control;
627 if (!ctl_cp) {
628 IP_VS_ERR("request control DEL for uncontrolled: "
629 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
630 NIPQUAD(cp->caddr),ntohs(cp->cport),
631 NIPQUAD(cp->vaddr),ntohs(cp->vport));
632 return;
633 }
634
635 IP_VS_DBG(7, "DELeting control for: "
636 "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
637 NIPQUAD(cp->caddr),ntohs(cp->cport),
638 NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
639
640 cp->control = NULL;
641 if (atomic_read(&ctl_cp->n_control) == 0) {
642 IP_VS_ERR("BUG control DEL with n=0 : "
643 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
644 NIPQUAD(cp->caddr),ntohs(cp->cport),
645 NIPQUAD(cp->vaddr),ntohs(cp->vport));
646 return;
647 }
648 atomic_dec(&ctl_cp->n_control);
649 }
650
651 static inline void
ip_vs_control_add(struct ip_vs_conn * cp,struct ip_vs_conn * ctl_cp)652 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
653 {
654 if (cp->control) {
655 IP_VS_ERR("request control ADD for already controlled: "
656 "%d.%d.%d.%d:%d to %d.%d.%d.%d:%d\n",
657 NIPQUAD(cp->caddr),ntohs(cp->cport),
658 NIPQUAD(cp->vaddr),ntohs(cp->vport));
659 ip_vs_control_del(cp);
660 }
661
662 IP_VS_DBG(7, "ADDing control for: "
663 "cp.dst=%d.%d.%d.%d:%d ctl_cp.dst=%d.%d.%d.%d:%d\n",
664 NIPQUAD(cp->caddr),ntohs(cp->cport),
665 NIPQUAD(ctl_cp->caddr),ntohs(ctl_cp->cport));
666
667 cp->control = ctl_cp;
668 atomic_inc(&ctl_cp->n_control);
669 }
670
671
672 /*
673 * IPVS application functions
674 * (from ip_vs_app.c)
675 */
676 #define IP_VS_APP_MAX_PORTS 8
677 extern int register_ip_vs_app(struct ip_vs_app *mapp,
678 unsigned short proto, __u16 port);
679 extern int unregister_ip_vs_app(struct ip_vs_app *mapp);
680 extern struct ip_vs_app * ip_vs_bind_app(struct ip_vs_conn *cp);
681 extern int ip_vs_unbind_app(struct ip_vs_conn *cp);
682 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
683 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
684 extern int ip_vs_skb_replace(struct sk_buff *skb, int pri,
685 char *o_buf, int o_len, char *n_buf, int n_len);
686 extern int ip_vs_app_init(void);
687 extern void ip_vs_app_cleanup(void);
688
689
690 /*
691 * Registering/unregistering scheduler functions
692 * (from ip_vs_sched.c)
693 */
694 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
695 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
696 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
697 struct ip_vs_scheduler *scheduler);
698 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
699 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
700 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
701
702
703 /*
704 * IPVS control data and functions
705 * (from ip_vs_ctl.c)
706 */
707 extern int sysctl_ip_vs_cache_bypass;
708 extern int sysctl_ip_vs_expire_nodest_conn;
709 extern int sysctl_ip_vs_expire_quiescent_template;
710 extern int sysctl_ip_vs_sync_threshold;
711 extern int sysctl_ip_vs_nat_icmp_send;
712 extern struct ip_vs_stats ip_vs_stats;
713
714 extern struct ip_vs_service *ip_vs_service_get(__u32 fwmark,
715 __u16 protocol,
716 __u32 vaddr, __u16 vport);
ip_vs_service_put(struct ip_vs_service * svc)717 static inline void ip_vs_service_put(struct ip_vs_service *svc)
718 {
719 atomic_dec(&svc->usecnt);
720 }
721
722 extern struct ip_vs_dest *
723 ip_vs_lookup_real_service(__u16 protocol, __u32 daddr, __u16 dport);
724 extern void ip_vs_random_dropentry(void);
725 extern int ip_vs_control_init(void);
726 extern void ip_vs_control_cleanup(void);
727
728
729 /*
730 * IPVS sync daemon data and function prototypes
731 * (from ip_vs_sync.c)
732 */
733 extern volatile int ip_vs_sync_state;
734 extern char ip_vs_mcast_master_ifn[IP_VS_IFNAME_MAXLEN];
735 extern char ip_vs_mcast_backup_ifn[IP_VS_IFNAME_MAXLEN];
736 extern int start_sync_thread(int state, char *mcast_ifn, __u8 syncid);
737 extern int stop_sync_thread(int state);
738 extern void ip_vs_sync_conn(struct ip_vs_conn *cp);
739
740
741 /*
742 * IPVS rate estimator prototypes (from ip_vs_est.c)
743 */
744 extern int ip_vs_new_estimator(struct ip_vs_stats *stats);
745 extern void ip_vs_kill_estimator(struct ip_vs_stats *stats);
746 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
747
748
749 /*
750 * This is a simple mechanism to ignore packets when
751 * we are loaded. Just set ip_vs_drop_rate to 'n' and
752 * we start to drop 1/rate of the packets
753 */
754 extern int ip_vs_drop_rate;
755 extern int ip_vs_drop_counter;
756
ip_vs_todrop(void)757 static __inline__ int ip_vs_todrop(void)
758 {
759 if (!ip_vs_drop_rate) return 0;
760 if (--ip_vs_drop_counter > 0) return 0;
761 ip_vs_drop_counter = ip_vs_drop_rate;
762 return 1;
763 }
764
765
766 /*
767 * ip_vs_fwd_tag returns the forwarding tag of the connection
768 */
769 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
770
ip_vs_fwd_tag(struct ip_vs_conn * cp)771 extern __inline__ char ip_vs_fwd_tag(struct ip_vs_conn *cp)
772 {
773 char fwd;
774
775 switch (IP_VS_FWD_METHOD(cp)) {
776 case IP_VS_CONN_F_MASQ:
777 fwd = 'M'; break;
778 case IP_VS_CONN_F_LOCALNODE:
779 fwd = 'L'; break;
780 case IP_VS_CONN_F_TUNNEL:
781 fwd = 'T'; break;
782 case IP_VS_CONN_F_DROUTE:
783 fwd = 'R'; break;
784 case IP_VS_CONN_F_BYPASS:
785 fwd = 'B'; break;
786 default:
787 fwd = '?'; break;
788 }
789 return fwd;
790 }
791
792
793 /*
794 * transport layer header checking
795 */
ip_vs_header_check(struct sk_buff * skb,int proto,int ihl)796 extern inline int ip_vs_header_check(struct sk_buff *skb, int proto, int ihl)
797 {
798 int len;
799
800 switch (proto) {
801 case IPPROTO_TCP:
802 len = ihl + sizeof(struct tcphdr);
803 /* we don't care about TCP options */
804 break;
805 case IPPROTO_UDP:
806 len = ihl + sizeof(struct udphdr);
807 break;
808 default:
809 len = 0;
810 }
811
812 /* guarantee protocol header available in skb data area */
813 if (!pskb_may_pull(skb, len))
814 return -1;
815 else
816 return 0;
817 }
818
819
820 /*
821 * Destination cache
822 */
823 static inline void
__ip_vs_dst_set(struct ip_vs_dest * dest,u32 rtos,struct dst_entry * dst)824 __ip_vs_dst_set(struct ip_vs_dest *dest, u32 rtos, struct dst_entry *dst)
825 {
826 struct dst_entry *old_dst;
827
828 old_dst = dest->dst_cache;
829 dest->dst_cache = dst;
830 dest->dst_rtos = rtos;
831 dst_release(old_dst);
832 }
833
834 static inline void
__ip_vs_dst_reset(struct ip_vs_dest * dest)835 __ip_vs_dst_reset(struct ip_vs_dest *dest)
836 {
837 struct dst_entry *old_dst;
838
839 old_dst = dest->dst_cache;
840 dest->dst_cache = NULL;
841 dst_release(old_dst);
842 }
843
844 static inline struct dst_entry *
__ip_vs_dst_check(struct ip_vs_dest * dest,u32 rtos,u32 cookie)845 __ip_vs_dst_check(struct ip_vs_dest *dest, u32 rtos, u32 cookie)
846 {
847 struct dst_entry *dst = dest->dst_cache;
848
849 if (!dst)
850 return NULL;
851 if ((dst->obsolete || rtos != dest->dst_rtos) &&
852 dst->ops->check(dst, cookie) == NULL) {
853 dest->dst_cache = 0;
854 return NULL;
855 }
856 dst_hold(dst);
857 return dst;
858 }
859
860 static inline struct rtable *
__ip_vs_get_out_rt(struct ip_vs_conn * cp,u32 rtos)861 __ip_vs_get_out_rt(struct ip_vs_conn *cp, u32 rtos)
862 {
863 struct rtable *rt; /* Route to the other host */
864 struct ip_vs_dest *dest = cp->dest;
865
866 if (dest) {
867 spin_lock(&dest->dst_lock);
868 if (!(rt = (struct rtable *)
869 __ip_vs_dst_check(dest, rtos, 0))) {
870 if (ip_route_output(&rt, dest->addr, 0, rtos, 0)) {
871 spin_unlock(&dest->dst_lock);
872 IP_VS_DBG_RL("ip_route_output error, "
873 "dest: %u.%u.%u.%u\n",
874 NIPQUAD(dest->addr));
875 return NULL;
876 }
877 __ip_vs_dst_set(dest, rtos, dst_clone(&rt->u.dst));
878 IP_VS_DBG(10, "new dst %u.%u.%u.%u, refcnt=%d, rtos=%X\n",
879 NIPQUAD(dest->addr),
880 atomic_read(&rt->u.dst.__refcnt), rtos);
881 }
882 spin_unlock(&dest->dst_lock);
883 } else {
884 if (ip_route_output(&rt, cp->daddr, 0, rtos, 0)) {
885 IP_VS_DBG_RL("ip_route_output error, dest: "
886 "%u.%u.%u.%u\n", NIPQUAD(cp->daddr));
887 return NULL;
888 }
889 }
890
891 return rt;
892 }
893
ip_vs_check_diff(u32 old,u32 new,u16 oldsum)894 static inline u16 ip_vs_check_diff(u32 old, u32 new, u16 oldsum)
895 {
896 u32 diff[2] = { old, new };
897
898 return csum_fold(csum_partial((char *) diff, sizeof(diff),
899 oldsum ^ 0xFFFF));
900 }
901
ip_vs_fast_check_update(union ip_vs_tphdr * h,u32 oldip,u32 newip,u16 oldport,u16 newport,u8 protocol)902 static inline void ip_vs_fast_check_update(union ip_vs_tphdr *h,
903 u32 oldip, u32 newip, u16 oldport, u16 newport, u8 protocol)
904 {
905 u16 *checkp;
906
907 if (protocol == IPPROTO_TCP)
908 checkp = &h->th->check;
909 else
910 checkp = &h->uh->check;
911 *checkp = ip_vs_check_diff(~oldip, newip,
912 ip_vs_check_diff(oldport ^ 0xFFFF,
913 newport, *checkp));
914 if (!*checkp && protocol == IPPROTO_UDP)
915 *checkp = 0xFFFF;
916 }
917
918 static inline int
ip_vs_skb_cow(struct sk_buff * skb,unsigned int headroom,struct iphdr ** iph_p,unsigned char ** t_p)919 ip_vs_skb_cow(struct sk_buff *skb, unsigned int headroom,
920 struct iphdr **iph_p, unsigned char **t_p)
921 {
922 int delta = (headroom > 16 ? headroom : 16) - skb_headroom(skb);
923
924 if (delta < 0)
925 delta = 0;
926
927 if (delta || skb_cloned(skb)) {
928 if (pskb_expand_head(skb, (delta+15)&~15, 0, GFP_ATOMIC))
929 return -ENOMEM;
930
931 /* skb data changed, update pointers */
932 *iph_p = skb->nh.iph;
933 *t_p = (char*) (*iph_p) + (*iph_p)->ihl * 4;
934 }
935 return 0;
936 }
937
938 #endif /* __KERNEL__ */
939
940 #endif /* _IP_VS_H */
941