3 * data structure and functionality definitions
9 #include <linux/ip_vs.h> /* definitions shared with userland */
11 /* old ipvsadm versions still include this file directly */
14 #include <asm/types.h> /* for __uXX types */
16 #include <linux/sysctl.h> /* for ctl_path */
17 #include <linux/list.h> /* for struct list_head */
18 #include <linux/spinlock.h> /* for struct rwlock_t */
19 #include <asm/atomic.h> /* for struct atomic_t */
20 #include <linux/compiler.h>
21 #include <linux/timer.h>
23 #include <net/checksum.h>
24 #include <linux/netfilter.h> /* for union nf_inet_addr */
26 #include <linux/ipv6.h> /* for struct ipv6hdr */
27 #include <net/ipv6.h> /* for ipv6_addr_copy */
28 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
29 #include <net/netfilter/nf_conntrack.h>
31 #include <net/net_namespace.h> /* Netw namespace */
34 * Generic access of ipvs struct
36 static inline struct netns_ipvs
*net_ipvs(struct net
* net
)
41 * Get net ptr from skb in traffic cases
42 * use skb_sknet when call is from userland (ioctl or netlink)
44 static inline struct net
*skb_net(const struct sk_buff
*skb
)
47 #ifdef CONFIG_IP_VS_DEBUG
49 * This is used for debug only.
50 * Start with the most likely hit
53 if (likely(skb
->dev
&& skb
->dev
->nd_net
))
54 return dev_net(skb
->dev
);
55 if (skb_dst(skb
)->dev
)
56 return dev_net(skb_dst(skb
)->dev
);
57 WARN(skb
->sk
, "Maybe skb_sknet should be used in %s() at line:%d\n",
59 if (likely(skb
->sk
&& skb
->sk
->sk_net
))
60 return sock_net(skb
->sk
);
61 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
65 return dev_net(skb
->dev
? : skb_dst(skb
)->dev
);
72 static inline struct net
*skb_sknet(const struct sk_buff
*skb
)
75 #ifdef CONFIG_IP_VS_DEBUG
76 /* Start with the most likely hit */
77 if (likely(skb
->sk
&& skb
->sk
->sk_net
))
78 return sock_net(skb
->sk
);
79 WARN(skb
->dev
, "Maybe skb_net should be used instead in %s() line:%d\n",
81 if (likely(skb
->dev
&& skb
->dev
->nd_net
))
82 return dev_net(skb
->dev
);
83 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
87 return sock_net(skb
->sk
);
94 * This one needed for single_open_net since net is stored directly in
95 * private not as a struct i.e. seq_file_net cant be used.
97 static inline struct net
*seq_file_single_net(struct seq_file
*seq
)
100 return (struct net
*)seq
->private;
106 /* Connections' size value needed by ip_vs_ctl.c */
107 extern int ip_vs_conn_tab_size
;
113 union nf_inet_addr saddr
;
114 union nf_inet_addr daddr
;
118 ip_vs_fill_iphdr(int af
, const void *nh
, struct ip_vs_iphdr
*iphdr
)
120 #ifdef CONFIG_IP_VS_IPV6
121 if (af
== AF_INET6
) {
122 const struct ipv6hdr
*iph
= nh
;
123 iphdr
->len
= sizeof(struct ipv6hdr
);
124 iphdr
->protocol
= iph
->nexthdr
;
125 ipv6_addr_copy(&iphdr
->saddr
.in6
, &iph
->saddr
);
126 ipv6_addr_copy(&iphdr
->daddr
.in6
, &iph
->daddr
);
130 const struct iphdr
*iph
= nh
;
131 iphdr
->len
= iph
->ihl
* 4;
132 iphdr
->protocol
= iph
->protocol
;
133 iphdr
->saddr
.ip
= iph
->saddr
;
134 iphdr
->daddr
.ip
= iph
->daddr
;
138 static inline void ip_vs_addr_copy(int af
, union nf_inet_addr
*dst
,
139 const union nf_inet_addr
*src
)
141 #ifdef CONFIG_IP_VS_IPV6
143 ipv6_addr_copy(&dst
->in6
, &src
->in6
);
149 static inline int ip_vs_addr_equal(int af
, const union nf_inet_addr
*a
,
150 const union nf_inet_addr
*b
)
152 #ifdef CONFIG_IP_VS_IPV6
154 return ipv6_addr_equal(&a
->in6
, &b
->in6
);
156 return a
->ip
== b
->ip
;
159 #ifdef CONFIG_IP_VS_DEBUG
160 #include <linux/net.h>
162 extern int ip_vs_get_debug_level(void);
164 static inline const char *ip_vs_dbg_addr(int af
, char *buf
, size_t buf_len
,
165 const union nf_inet_addr
*addr
,
169 #ifdef CONFIG_IP_VS_IPV6
171 len
= snprintf(&buf
[*idx
], buf_len
- *idx
, "[%pI6]",
175 len
= snprintf(&buf
[*idx
], buf_len
- *idx
, "%pI4",
179 BUG_ON(*idx
> buf_len
+ 1);
180 return &buf
[*idx
- len
];
183 #define IP_VS_DBG_BUF(level, msg, ...) \
185 char ip_vs_dbg_buf[160]; \
186 int ip_vs_dbg_idx = 0; \
187 if (level <= ip_vs_get_debug_level()) \
188 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
190 #define IP_VS_ERR_BUF(msg...) \
192 char ip_vs_dbg_buf[160]; \
193 int ip_vs_dbg_idx = 0; \
197 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
198 #define IP_VS_DBG_ADDR(af, addr) \
199 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
200 sizeof(ip_vs_dbg_buf), addr, \
203 #define IP_VS_DBG(level, msg, ...) \
205 if (level <= ip_vs_get_debug_level()) \
206 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
208 #define IP_VS_DBG_RL(msg, ...) \
210 if (net_ratelimit()) \
211 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
213 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
215 if (level <= ip_vs_get_debug_level()) \
216 pp->debug_packet(af, pp, skb, ofs, msg); \
218 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
220 if (level <= ip_vs_get_debug_level() && \
222 pp->debug_packet(af, pp, skb, ofs, msg); \
224 #else /* NO DEBUGGING at ALL */
225 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
226 #define IP_VS_ERR_BUF(msg...) do {} while (0)
227 #define IP_VS_DBG(level, msg...) do {} while (0)
228 #define IP_VS_DBG_RL(msg...) do {} while (0)
229 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
230 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
233 #define IP_VS_BUG() BUG()
234 #define IP_VS_ERR_RL(msg, ...) \
236 if (net_ratelimit()) \
237 pr_err(msg, ##__VA_ARGS__); \
240 #ifdef CONFIG_IP_VS_DEBUG
241 #define EnterFunction(level) \
243 if (level <= ip_vs_get_debug_level()) \
245 pr_fmt("Enter: %s, %s line %i\n"), \
246 __func__, __FILE__, __LINE__); \
248 #define LeaveFunction(level) \
250 if (level <= ip_vs_get_debug_level()) \
252 pr_fmt("Leave: %s, %s line %i\n"), \
253 __func__, __FILE__, __LINE__); \
256 #define EnterFunction(level) do {} while (0)
257 #define LeaveFunction(level) do {} while (0)
260 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
264 * The port number of FTP service (in network order).
266 #define FTPPORT cpu_to_be16(21)
267 #define FTPDATA cpu_to_be16(20)
273 IP_VS_TCP_S_NONE
= 0,
274 IP_VS_TCP_S_ESTABLISHED
,
275 IP_VS_TCP_S_SYN_SENT
,
276 IP_VS_TCP_S_SYN_RECV
,
277 IP_VS_TCP_S_FIN_WAIT
,
278 IP_VS_TCP_S_TIME_WAIT
,
280 IP_VS_TCP_S_CLOSE_WAIT
,
281 IP_VS_TCP_S_LAST_ACK
,
306 enum ip_vs_sctp_states
{
308 IP_VS_SCTP_S_INIT_CLI
,
309 IP_VS_SCTP_S_INIT_SER
,
310 IP_VS_SCTP_S_INIT_ACK_CLI
,
311 IP_VS_SCTP_S_INIT_ACK_SER
,
312 IP_VS_SCTP_S_ECHO_CLI
,
313 IP_VS_SCTP_S_ECHO_SER
,
314 IP_VS_SCTP_S_ESTABLISHED
,
315 IP_VS_SCTP_S_SHUT_CLI
,
316 IP_VS_SCTP_S_SHUT_SER
,
317 IP_VS_SCTP_S_SHUT_ACK_CLI
,
318 IP_VS_SCTP_S_SHUT_ACK_SER
,
324 * Delta sequence info structure
325 * Each ip_vs_conn has 2 (output AND input seq. changes).
326 * Only used in the VS/NAT.
329 __u32 init_seq
; /* Add delta from this seq */
330 __u32 delta
; /* Delta in sequence numbers */
331 __u32 previous_delta
; /* Delta in sequence numbers
332 before last resized pkt */
338 struct ip_vs_counters
{
339 __u32 conns
; /* connections scheduled */
340 __u32 inpkts
; /* incoming packets */
341 __u32 outpkts
; /* outgoing packets */
342 __u64 inbytes
; /* incoming bytes */
343 __u64 outbytes
; /* outgoing bytes */
348 struct ip_vs_cpu_stats
{
349 struct ip_vs_counters ustats
;
350 struct u64_stats_sync syncp
;
354 * IPVS statistics objects
356 struct ip_vs_estimator
{
357 struct list_head list
;
373 struct ip_vs_stats_user ustats
; /* statistics */
374 struct ip_vs_estimator est
; /* estimator */
375 struct ip_vs_cpu_stats
*cpustats
; /* per cpu counters */
376 spinlock_t lock
; /* spin lock */
380 * Helper Macros for per cpu
381 * ipvs->tot_stats->ustats.count
383 #define IPVS_STAT_INC(ipvs, count) \
384 __this_cpu_inc((ipvs)->ustats->count)
386 #define IPVS_STAT_ADD(ipvs, count, value) \
388 write_seqcount_begin(per_cpu_ptr((ipvs)->ustats_seq, \
389 raw_smp_processor_id())); \
390 __this_cpu_add((ipvs)->ustats->count, value); \
391 write_seqcount_end(per_cpu_ptr((ipvs)->ustats_seq, \
392 raw_smp_processor_id())); \
400 struct ip_vs_proto_data
;
402 struct ip_vs_protocol
{
403 struct ip_vs_protocol
*next
;
409 void (*init
)(struct ip_vs_protocol
*pp
);
411 void (*exit
)(struct ip_vs_protocol
*pp
);
413 void (*init_netns
)(struct net
*net
, struct ip_vs_proto_data
*pd
);
415 void (*exit_netns
)(struct net
*net
, struct ip_vs_proto_data
*pd
);
417 int (*conn_schedule
)(int af
, struct sk_buff
*skb
,
418 struct ip_vs_proto_data
*pd
,
419 int *verdict
, struct ip_vs_conn
**cpp
);
422 (*conn_in_get
)(int af
,
423 const struct sk_buff
*skb
,
424 const struct ip_vs_iphdr
*iph
,
425 unsigned int proto_off
,
429 (*conn_out_get
)(int af
,
430 const struct sk_buff
*skb
,
431 const struct ip_vs_iphdr
*iph
,
432 unsigned int proto_off
,
435 int (*snat_handler
)(struct sk_buff
*skb
,
436 struct ip_vs_protocol
*pp
, struct ip_vs_conn
*cp
);
438 int (*dnat_handler
)(struct sk_buff
*skb
,
439 struct ip_vs_protocol
*pp
, struct ip_vs_conn
*cp
);
441 int (*csum_check
)(int af
, struct sk_buff
*skb
,
442 struct ip_vs_protocol
*pp
);
444 const char *(*state_name
)(int state
);
446 int (*state_transition
)(struct ip_vs_conn
*cp
, int direction
,
447 const struct sk_buff
*skb
,
448 struct ip_vs_proto_data
*pd
);
450 int (*register_app
)(struct net
*net
, struct ip_vs_app
*inc
);
452 void (*unregister_app
)(struct net
*net
, struct ip_vs_app
*inc
);
454 int (*app_conn_bind
)(struct ip_vs_conn
*cp
);
456 void (*debug_packet
)(int af
, struct ip_vs_protocol
*pp
,
457 const struct sk_buff
*skb
,
461 void (*timeout_change
)(struct ip_vs_proto_data
*pd
, int flags
);
465 * protocol data per netns
467 struct ip_vs_proto_data
{
468 struct ip_vs_proto_data
*next
;
469 struct ip_vs_protocol
*pp
;
470 int *timeout_table
; /* protocol timeout table */
471 atomic_t appcnt
; /* counter of proto app incs. */
472 struct tcp_states_t
*tcp_state_table
;
475 extern struct ip_vs_protocol
*ip_vs_proto_get(unsigned short proto
);
476 extern struct ip_vs_proto_data
*ip_vs_proto_data_get(struct net
*net
,
477 unsigned short proto
);
479 struct ip_vs_conn_param
{
481 const union nf_inet_addr
*caddr
;
482 const union nf_inet_addr
*vaddr
;
488 const struct ip_vs_pe
*pe
;
494 * IP_VS structure allocated for each dynamically scheduled connection
497 struct hlist_node c_list
; /* hashed list heads */
499 struct net
*net
; /* Name space */
501 /* Protocol, addresses and port numbers */
502 u16 af
; /* address family */
506 __u32 fwmark
; /* Fire wall mark from skb */
507 union nf_inet_addr caddr
; /* client address */
508 union nf_inet_addr vaddr
; /* virtual address */
509 union nf_inet_addr daddr
; /* destination address */
510 volatile __u32 flags
; /* status flags */
511 __u16 protocol
; /* Which protocol (TCP/UDP) */
513 /* counter and timer */
514 atomic_t refcnt
; /* reference count */
515 struct timer_list timer
; /* Expiration timer */
516 volatile unsigned long timeout
; /* timeout */
518 /* Flags and state transition */
519 spinlock_t lock
; /* lock for state transition */
520 volatile __u16 state
; /* state info */
521 volatile __u16 old_state
; /* old state, to be used for
522 * state transition triggerd
526 /* Control members */
527 struct ip_vs_conn
*control
; /* Master control connection */
528 atomic_t n_control
; /* Number of controlled ones */
529 struct ip_vs_dest
*dest
; /* real server */
530 atomic_t in_pkts
; /* incoming packet counter */
532 /* packet transmitter for different forwarding methods. If it
533 mangles the packet, it must return NF_DROP or better NF_STOLEN,
534 otherwise this must be changed to a sk_buff **.
535 NF_ACCEPT can be returned when destination is local.
537 int (*packet_xmit
)(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
538 struct ip_vs_protocol
*pp
);
540 /* Note: we can group the following members into a structure,
541 in order to save more space, and the following members are
542 only used in VS/NAT anyway */
543 struct ip_vs_app
*app
; /* bound ip_vs_app object */
544 void *app_data
; /* Application private data */
545 struct ip_vs_seq in_seq
; /* incoming seq. struct */
546 struct ip_vs_seq out_seq
; /* outgoing seq. struct */
548 const struct ip_vs_pe
*pe
;
554 * To save some memory in conn table when name space is disabled.
556 static inline struct net
*ip_vs_conn_net(const struct ip_vs_conn
*cp
)
564 static inline void ip_vs_conn_net_set(struct ip_vs_conn
*cp
, struct net
*net
)
571 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn
*cp
,
575 return cp
->net
== net
;
582 * Extended internal versions of struct ip_vs_service_user and
583 * ip_vs_dest_user for IPv6 support.
585 * We need these to conveniently pass around service and destination
586 * options, but unfortunately, we also need to keep the old definitions to
587 * maintain userspace backwards compatibility for the setsockopt interface.
589 struct ip_vs_service_user_kern
{
590 /* virtual service addresses */
593 union nf_inet_addr addr
; /* virtual ip address */
595 u32 fwmark
; /* firwall mark of service */
597 /* virtual service options */
600 unsigned flags
; /* virtual service flags */
601 unsigned timeout
; /* persistent timeout in sec */
602 u32 netmask
; /* persistent netmask */
606 struct ip_vs_dest_user_kern
{
607 /* destination server address */
608 union nf_inet_addr addr
;
611 /* real server options */
612 unsigned conn_flags
; /* connection flags */
613 int weight
; /* destination weight */
615 /* thresholds for active connections */
616 u32 u_threshold
; /* upper threshold */
617 u32 l_threshold
; /* lower threshold */
622 * The information about the virtual service offered to the net
623 * and the forwarding entries
625 struct ip_vs_service
{
626 struct list_head s_list
; /* for normal service table */
627 struct list_head f_list
; /* for fwmark-based service table */
628 atomic_t refcnt
; /* reference counter */
629 atomic_t usecnt
; /* use counter */
631 u16 af
; /* address family */
632 __u16 protocol
; /* which protocol (TCP/UDP) */
633 union nf_inet_addr addr
; /* IP address for virtual service */
634 __be16 port
; /* port number for the service */
635 __u32 fwmark
; /* firewall mark of the service */
636 unsigned flags
; /* service status flags */
637 unsigned timeout
; /* persistent timeout in ticks */
638 __be32 netmask
; /* grouping granularity */
641 struct list_head destinations
; /* real server d-linked list */
642 __u32 num_dests
; /* number of servers */
643 struct ip_vs_stats stats
; /* statistics for the service */
644 struct ip_vs_app
*inc
; /* bind conns to this app inc */
647 struct ip_vs_scheduler
*scheduler
; /* bound scheduler object */
648 rwlock_t sched_lock
; /* lock sched_data */
649 void *sched_data
; /* scheduler application data */
651 /* alternate persistence engine */
657 * The real server destination forwarding entry
658 * with ip address, port number, and so on.
661 struct list_head n_list
; /* for the dests in the service */
662 struct list_head d_list
; /* for table with all the dests */
664 u16 af
; /* address family */
665 __be16 port
; /* port number of the server */
666 union nf_inet_addr addr
; /* IP address of the server */
667 volatile unsigned flags
; /* dest status flags */
668 atomic_t conn_flags
; /* flags to copy to conn */
669 atomic_t weight
; /* server weight */
671 atomic_t refcnt
; /* reference counter */
672 struct ip_vs_stats stats
; /* statistics */
674 /* connection counters and thresholds */
675 atomic_t activeconns
; /* active connections */
676 atomic_t inactconns
; /* inactive connections */
677 atomic_t persistconns
; /* persistent connections */
678 __u32 u_threshold
; /* upper threshold */
679 __u32 l_threshold
; /* lower threshold */
681 /* for destination cache */
682 spinlock_t dst_lock
; /* lock of dst_cache */
683 struct dst_entry
*dst_cache
; /* destination cache entry */
684 u32 dst_rtos
; /* RT_TOS(tos) for dst */
686 #ifdef CONFIG_IP_VS_IPV6
687 struct in6_addr dst_saddr
;
690 /* for virtual service */
691 struct ip_vs_service
*svc
; /* service it belongs to */
692 __u16 protocol
; /* which protocol (TCP/UDP) */
693 __be16 vport
; /* virtual port number */
694 union nf_inet_addr vaddr
; /* virtual IP address */
695 __u32 vfwmark
; /* firewall mark of service */
700 * The scheduler object
702 struct ip_vs_scheduler
{
703 struct list_head n_list
; /* d-linked list head */
704 char *name
; /* scheduler name */
705 atomic_t refcnt
; /* reference counter */
706 struct module
*module
; /* THIS_MODULE/NULL */
708 /* scheduler initializing service */
709 int (*init_service
)(struct ip_vs_service
*svc
);
710 /* scheduling service finish */
711 int (*done_service
)(struct ip_vs_service
*svc
);
712 /* scheduler updating service */
713 int (*update_service
)(struct ip_vs_service
*svc
);
715 /* selecting a server from the given service */
716 struct ip_vs_dest
* (*schedule
)(struct ip_vs_service
*svc
,
717 const struct sk_buff
*skb
);
720 /* The persistence engine object */
722 struct list_head n_list
; /* d-linked list head */
723 char *name
; /* scheduler name */
724 atomic_t refcnt
; /* reference counter */
725 struct module
*module
; /* THIS_MODULE/NULL */
727 /* get the connection template, if any */
728 int (*fill_param
)(struct ip_vs_conn_param
*p
, struct sk_buff
*skb
);
729 bool (*ct_match
)(const struct ip_vs_conn_param
*p
,
730 struct ip_vs_conn
*ct
);
731 u32 (*hashkey_raw
)(const struct ip_vs_conn_param
*p
, u32 initval
,
733 int (*show_pe_data
)(const struct ip_vs_conn
*cp
, char *buf
);
737 * The application module object (a.k.a. app incarnation)
740 struct list_head a_list
; /* member in app list */
741 int type
; /* IP_VS_APP_TYPE_xxx */
742 char *name
; /* application module name */
744 struct module
*module
; /* THIS_MODULE/NULL */
745 struct list_head incs_list
; /* list of incarnations */
747 /* members for application incarnations */
748 struct list_head p_list
; /* member in proto app list */
749 struct ip_vs_app
*app
; /* its real application */
750 __be16 port
; /* port number in net order */
751 atomic_t usecnt
; /* usage counter */
754 * output hook: Process packet in inout direction, diff set for TCP.
755 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
756 * 2=Mangled but checksum was not updated
758 int (*pkt_out
)(struct ip_vs_app
*, struct ip_vs_conn
*,
759 struct sk_buff
*, int *diff
);
762 * input hook: Process packet in outin direction, diff set for TCP.
763 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
764 * 2=Mangled but checksum was not updated
766 int (*pkt_in
)(struct ip_vs_app
*, struct ip_vs_conn
*,
767 struct sk_buff
*, int *diff
);
769 /* ip_vs_app initializer */
770 int (*init_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
772 /* ip_vs_app finish */
773 int (*done_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
777 int (*bind_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*,
778 struct ip_vs_protocol
*);
780 void (*unbind_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
786 int (*conn_schedule
)(struct sk_buff
*skb
, struct ip_vs_app
*app
,
787 int *verdict
, struct ip_vs_conn
**cpp
);
790 (*conn_in_get
)(const struct sk_buff
*skb
, struct ip_vs_app
*app
,
791 const struct iphdr
*iph
, unsigned int proto_off
,
795 (*conn_out_get
)(const struct sk_buff
*skb
, struct ip_vs_app
*app
,
796 const struct iphdr
*iph
, unsigned int proto_off
,
799 int (*state_transition
)(struct ip_vs_conn
*cp
, int direction
,
800 const struct sk_buff
*skb
,
801 struct ip_vs_app
*app
);
803 void (*timeout_change
)(struct ip_vs_app
*app
, int flags
);
806 /* IPVS in network namespace */
808 int gen
; /* Generation */
810 * Hash table: for real service lookups
812 #define IP_VS_RTAB_BITS 4
813 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
814 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
816 struct list_head rs_table
[IP_VS_RTAB_SIZE
];
818 struct list_head app_list
;
819 struct mutex app_mutex
;
820 struct lock_class_key app_key
; /* mutex debuging */
823 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
824 struct ip_vs_proto_data
*proto_data_table
[IP_VS_PROTO_TAB_SIZE
];
825 /* ip_vs_proto_tcp */
826 #ifdef CONFIG_IP_VS_PROTO_TCP
827 #define TCP_APP_TAB_BITS 4
828 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
829 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
830 struct list_head tcp_apps
[TCP_APP_TAB_SIZE
];
831 spinlock_t tcp_app_lock
;
833 /* ip_vs_proto_udp */
834 #ifdef CONFIG_IP_VS_PROTO_UDP
835 #define UDP_APP_TAB_BITS 4
836 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
837 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
838 struct list_head udp_apps
[UDP_APP_TAB_SIZE
];
839 spinlock_t udp_app_lock
;
841 /* ip_vs_proto_sctp */
842 #ifdef CONFIG_IP_VS_PROTO_SCTP
843 #define SCTP_APP_TAB_BITS 4
844 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
845 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
846 /* Hash table for SCTP application incarnations */
847 struct list_head sctp_apps
[SCTP_APP_TAB_SIZE
];
848 spinlock_t sctp_app_lock
;
851 atomic_t conn_count
; /* connection counter */
854 struct ip_vs_stats tot_stats
; /* Statistics & est. */
855 seqcount_t
*ustats_seq
; /* u64 read retry */
857 int num_services
; /* no of virtual services */
858 /* 1/rate drop and drop-entry variables */
859 struct delayed_work defense_work
; /* Work handler */
864 spinlock_t dropentry_lock
; /* drop entry handling */
865 spinlock_t droppacket_lock
; /* drop packet handling */
866 spinlock_t securetcp_lock
; /* state and timeout tables */
867 rwlock_t rs_lock
; /* real services table */
868 /* semaphore for IPVS sockopts. And, [gs]etsockopt may sleep. */
869 struct lock_class_key ctl_key
; /* ctl_mutex debuging */
870 /* Trash for destinations */
871 struct list_head dest_trash
;
872 /* Service counters */
873 atomic_t ftpsvc_counter
;
874 atomic_t nullsvc_counter
;
877 struct ctl_table_header
*sysctl_hdr
;
878 struct ctl_table
*sysctl_tbl
;
879 /* sysctl variables */
880 int sysctl_amemthresh
;
881 int sysctl_am_droprate
;
882 int sysctl_drop_entry
;
883 int sysctl_drop_packet
;
884 int sysctl_secure_tcp
;
885 #ifdef CONFIG_IP_VS_NFCT
886 int sysctl_conntrack
;
888 int sysctl_snat_reroute
;
890 int sysctl_cache_bypass
;
891 int sysctl_expire_nodest_conn
;
892 int sysctl_expire_quiescent_template
;
893 int sysctl_sync_threshold
[2];
894 int sysctl_nat_icmp_send
;
897 int sysctl_lblc_expiration
;
898 struct ctl_table_header
*lblc_ctl_header
;
899 struct ctl_table
*lblc_ctl_table
;
901 int sysctl_lblcr_expiration
;
902 struct ctl_table_header
*lblcr_ctl_header
;
903 struct ctl_table
*lblcr_ctl_table
;
905 struct list_head est_list
; /* estimator list */
907 struct timer_list est_timer
; /* Estimation timer */
909 struct list_head sync_queue
;
910 spinlock_t sync_lock
;
911 struct ip_vs_sync_buff
*sync_buff
;
912 spinlock_t sync_buff_lock
;
913 struct sockaddr_in sync_mcast_addr
;
914 struct task_struct
*master_thread
;
915 struct task_struct
*backup_thread
;
916 int send_mesg_maxlen
;
917 int recv_mesg_maxlen
;
918 volatile int sync_state
;
919 volatile int master_syncid
;
920 volatile int backup_syncid
;
921 /* multicast interface name */
922 char master_mcast_ifn
[IP_VS_IFNAME_MAXLEN
];
923 char backup_mcast_ifn
[IP_VS_IFNAME_MAXLEN
];
924 /* net name space ptr */
925 struct net
*net
; /* Needed by timer routines */
929 * IPVS core functions
930 * (from ip_vs_core.c)
932 extern const char *ip_vs_proto_name(unsigned proto
);
933 extern void ip_vs_init_hash_table(struct list_head
*table
, int rows
);
934 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
936 #define IP_VS_APP_TYPE_FTP 1
939 * ip_vs_conn handling functions
940 * (from ip_vs_conn.c)
946 IP_VS_DIR_INPUT_ONLY
,
950 static inline void ip_vs_conn_fill_param(struct net
*net
, int af
, int protocol
,
951 const union nf_inet_addr
*caddr
,
953 const union nf_inet_addr
*vaddr
,
955 struct ip_vs_conn_param
*p
)
959 p
->protocol
= protocol
;
968 struct ip_vs_conn
*ip_vs_conn_in_get(const struct ip_vs_conn_param
*p
);
969 struct ip_vs_conn
*ip_vs_ct_in_get(const struct ip_vs_conn_param
*p
);
971 struct ip_vs_conn
* ip_vs_conn_in_get_proto(int af
, const struct sk_buff
*skb
,
972 const struct ip_vs_iphdr
*iph
,
973 unsigned int proto_off
,
976 struct ip_vs_conn
*ip_vs_conn_out_get(const struct ip_vs_conn_param
*p
);
978 struct ip_vs_conn
* ip_vs_conn_out_get_proto(int af
, const struct sk_buff
*skb
,
979 const struct ip_vs_iphdr
*iph
,
980 unsigned int proto_off
,
983 /* put back the conn without restarting its timer */
984 static inline void __ip_vs_conn_put(struct ip_vs_conn
*cp
)
986 atomic_dec(&cp
->refcnt
);
988 extern void ip_vs_conn_put(struct ip_vs_conn
*cp
);
989 extern void ip_vs_conn_fill_cport(struct ip_vs_conn
*cp
, __be16 cport
);
991 struct ip_vs_conn
*ip_vs_conn_new(const struct ip_vs_conn_param
*p
,
992 const union nf_inet_addr
*daddr
,
993 __be16 dport
, unsigned flags
,
994 struct ip_vs_dest
*dest
, __u32 fwmark
);
995 extern void ip_vs_conn_expire_now(struct ip_vs_conn
*cp
);
997 extern const char * ip_vs_state_name(__u16 proto
, int state
);
999 extern void ip_vs_tcp_conn_listen(struct net
*net
, struct ip_vs_conn
*cp
);
1000 extern int ip_vs_check_template(struct ip_vs_conn
*ct
);
1001 extern void ip_vs_random_dropentry(struct net
*net
);
1002 extern int ip_vs_conn_init(void);
1003 extern void ip_vs_conn_cleanup(void);
1005 static inline void ip_vs_control_del(struct ip_vs_conn
*cp
)
1007 struct ip_vs_conn
*ctl_cp
= cp
->control
;
1009 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1011 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1013 IP_VS_DBG_ADDR(cp
->af
, &cp
->vaddr
),
1019 IP_VS_DBG_BUF(7, "DELeting control for: "
1020 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1021 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1023 IP_VS_DBG_ADDR(cp
->af
, &ctl_cp
->caddr
),
1024 ntohs(ctl_cp
->cport
));
1027 if (atomic_read(&ctl_cp
->n_control
) == 0) {
1028 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1030 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1032 IP_VS_DBG_ADDR(cp
->af
, &cp
->vaddr
),
1037 atomic_dec(&ctl_cp
->n_control
);
1041 ip_vs_control_add(struct ip_vs_conn
*cp
, struct ip_vs_conn
*ctl_cp
)
1044 IP_VS_ERR_BUF("request control ADD for already controlled: "
1046 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1048 IP_VS_DBG_ADDR(cp
->af
, &cp
->vaddr
),
1051 ip_vs_control_del(cp
);
1054 IP_VS_DBG_BUF(7, "ADDing control for: "
1055 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1056 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1058 IP_VS_DBG_ADDR(cp
->af
, &ctl_cp
->caddr
),
1059 ntohs(ctl_cp
->cport
));
1061 cp
->control
= ctl_cp
;
1062 atomic_inc(&ctl_cp
->n_control
);
1067 * IPVS application functions
1068 * (from ip_vs_app.c)
1070 #define IP_VS_APP_MAX_PORTS 8
1071 extern int register_ip_vs_app(struct net
*net
, struct ip_vs_app
*app
);
1072 extern void unregister_ip_vs_app(struct net
*net
, struct ip_vs_app
*app
);
1073 extern int ip_vs_bind_app(struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1074 extern void ip_vs_unbind_app(struct ip_vs_conn
*cp
);
1075 extern int register_ip_vs_app_inc(struct net
*net
, struct ip_vs_app
*app
,
1076 __u16 proto
, __u16 port
);
1077 extern int ip_vs_app_inc_get(struct ip_vs_app
*inc
);
1078 extern void ip_vs_app_inc_put(struct ip_vs_app
*inc
);
1080 extern int ip_vs_app_pkt_out(struct ip_vs_conn
*, struct sk_buff
*skb
);
1081 extern int ip_vs_app_pkt_in(struct ip_vs_conn
*, struct sk_buff
*skb
);
1082 extern int ip_vs_app_init(void);
1083 extern void ip_vs_app_cleanup(void);
1085 void ip_vs_bind_pe(struct ip_vs_service
*svc
, struct ip_vs_pe
*pe
);
1086 void ip_vs_unbind_pe(struct ip_vs_service
*svc
);
1087 int register_ip_vs_pe(struct ip_vs_pe
*pe
);
1088 int unregister_ip_vs_pe(struct ip_vs_pe
*pe
);
1089 struct ip_vs_pe
*ip_vs_pe_getbyname(const char *name
);
1090 struct ip_vs_pe
*__ip_vs_pe_getbyname(const char *pe_name
);
1092 static inline void ip_vs_pe_get(const struct ip_vs_pe
*pe
)
1094 if (pe
&& pe
->module
)
1095 __module_get(pe
->module
);
1098 static inline void ip_vs_pe_put(const struct ip_vs_pe
*pe
)
1100 if (pe
&& pe
->module
)
1101 module_put(pe
->module
);
1105 * IPVS protocol functions (from ip_vs_proto.c)
1107 extern int ip_vs_protocol_init(void);
1108 extern void ip_vs_protocol_cleanup(void);
1109 extern void ip_vs_protocol_timeout_change(struct netns_ipvs
*ipvs
, int flags
);
1110 extern int *ip_vs_create_timeout_table(int *table
, int size
);
1112 ip_vs_set_state_timeout(int *table
, int num
, const char *const *names
,
1113 const char *name
, int to
);
1115 ip_vs_tcpudp_debug_packet(int af
, struct ip_vs_protocol
*pp
,
1116 const struct sk_buff
*skb
,
1117 int offset
, const char *msg
);
1119 extern struct ip_vs_protocol ip_vs_protocol_tcp
;
1120 extern struct ip_vs_protocol ip_vs_protocol_udp
;
1121 extern struct ip_vs_protocol ip_vs_protocol_icmp
;
1122 extern struct ip_vs_protocol ip_vs_protocol_esp
;
1123 extern struct ip_vs_protocol ip_vs_protocol_ah
;
1124 extern struct ip_vs_protocol ip_vs_protocol_sctp
;
1127 * Registering/unregistering scheduler functions
1128 * (from ip_vs_sched.c)
1130 extern int register_ip_vs_scheduler(struct ip_vs_scheduler
*scheduler
);
1131 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler
*scheduler
);
1132 extern int ip_vs_bind_scheduler(struct ip_vs_service
*svc
,
1133 struct ip_vs_scheduler
*scheduler
);
1134 extern int ip_vs_unbind_scheduler(struct ip_vs_service
*svc
);
1135 extern struct ip_vs_scheduler
*ip_vs_scheduler_get(const char *sched_name
);
1136 extern void ip_vs_scheduler_put(struct ip_vs_scheduler
*scheduler
);
1137 extern struct ip_vs_conn
*
1138 ip_vs_schedule(struct ip_vs_service
*svc
, struct sk_buff
*skb
,
1139 struct ip_vs_proto_data
*pd
, int *ignored
);
1140 extern int ip_vs_leave(struct ip_vs_service
*svc
, struct sk_buff
*skb
,
1141 struct ip_vs_proto_data
*pd
);
1143 extern void ip_vs_scheduler_err(struct ip_vs_service
*svc
, const char *msg
);
1147 * IPVS control data and functions (from ip_vs_ctl.c)
1149 extern struct ip_vs_stats ip_vs_stats
;
1150 extern const struct ctl_path net_vs_ctl_path
[];
1151 extern int sysctl_ip_vs_sync_ver
;
1153 extern void ip_vs_sync_switch_mode(struct net
*net
, int mode
);
1154 extern struct ip_vs_service
*
1155 ip_vs_service_get(struct net
*net
, int af
, __u32 fwmark
, __u16 protocol
,
1156 const union nf_inet_addr
*vaddr
, __be16 vport
);
1158 static inline void ip_vs_service_put(struct ip_vs_service
*svc
)
1160 atomic_dec(&svc
->usecnt
);
1163 extern struct ip_vs_dest
*
1164 ip_vs_lookup_real_service(struct net
*net
, int af
, __u16 protocol
,
1165 const union nf_inet_addr
*daddr
, __be16 dport
);
1167 extern int ip_vs_use_count_inc(void);
1168 extern void ip_vs_use_count_dec(void);
1169 extern int ip_vs_control_init(void);
1170 extern void ip_vs_control_cleanup(void);
1171 extern struct ip_vs_dest
*
1172 ip_vs_find_dest(struct net
*net
, int af
, const union nf_inet_addr
*daddr
,
1173 __be16 dport
, const union nf_inet_addr
*vaddr
, __be16 vport
,
1174 __u16 protocol
, __u32 fwmark
);
1175 extern struct ip_vs_dest
*ip_vs_try_bind_dest(struct ip_vs_conn
*cp
);
1179 * IPVS sync daemon data and function prototypes
1180 * (from ip_vs_sync.c)
1182 extern int start_sync_thread(struct net
*net
, int state
, char *mcast_ifn
,
1184 extern int stop_sync_thread(struct net
*net
, int state
);
1185 extern void ip_vs_sync_conn(struct net
*net
, struct ip_vs_conn
*cp
);
1186 extern int ip_vs_sync_init(void);
1187 extern void ip_vs_sync_cleanup(void);
1191 * IPVS rate estimator prototypes (from ip_vs_est.c)
1193 extern int ip_vs_estimator_init(void);
1194 extern void ip_vs_estimator_cleanup(void);
1195 extern void ip_vs_new_estimator(struct net
*net
, struct ip_vs_stats
*stats
);
1196 extern void ip_vs_kill_estimator(struct net
*net
, struct ip_vs_stats
*stats
);
1197 extern void ip_vs_zero_estimator(struct ip_vs_stats
*stats
);
1200 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1202 extern int ip_vs_null_xmit
1203 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1204 extern int ip_vs_bypass_xmit
1205 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1206 extern int ip_vs_nat_xmit
1207 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1208 extern int ip_vs_tunnel_xmit
1209 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1210 extern int ip_vs_dr_xmit
1211 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1212 extern int ip_vs_icmp_xmit
1213 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
, int offset
);
1214 extern void ip_vs_dst_reset(struct ip_vs_dest
*dest
);
1216 #ifdef CONFIG_IP_VS_IPV6
1217 extern int ip_vs_bypass_xmit_v6
1218 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1219 extern int ip_vs_nat_xmit_v6
1220 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1221 extern int ip_vs_tunnel_xmit_v6
1222 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1223 extern int ip_vs_dr_xmit_v6
1224 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1225 extern int ip_vs_icmp_xmit_v6
1226 (struct sk_buff
*skb
, struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
,
1231 * This is a simple mechanism to ignore packets when
1232 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1233 * we start to drop 1/rate of the packets
1236 static inline int ip_vs_todrop(struct netns_ipvs
*ipvs
)
1238 if (!ipvs
->drop_rate
)
1240 if (--ipvs
->drop_counter
> 0)
1242 ipvs
->drop_counter
= ipvs
->drop_rate
;
1247 * ip_vs_fwd_tag returns the forwarding tag of the connection
1249 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1251 static inline char ip_vs_fwd_tag(struct ip_vs_conn
*cp
)
1255 switch (IP_VS_FWD_METHOD(cp
)) {
1256 case IP_VS_CONN_F_MASQ
:
1258 case IP_VS_CONN_F_LOCALNODE
:
1260 case IP_VS_CONN_F_TUNNEL
:
1262 case IP_VS_CONN_F_DROUTE
:
1264 case IP_VS_CONN_F_BYPASS
:
1272 extern void ip_vs_nat_icmp(struct sk_buff
*skb
, struct ip_vs_protocol
*pp
,
1273 struct ip_vs_conn
*cp
, int dir
);
1275 #ifdef CONFIG_IP_VS_IPV6
1276 extern void ip_vs_nat_icmp_v6(struct sk_buff
*skb
, struct ip_vs_protocol
*pp
,
1277 struct ip_vs_conn
*cp
, int dir
);
1280 extern __sum16
ip_vs_checksum_complete(struct sk_buff
*skb
, int offset
);
1282 static inline __wsum
ip_vs_check_diff4(__be32 old
, __be32
new, __wsum oldsum
)
1284 __be32 diff
[2] = { ~old
, new };
1286 return csum_partial(diff
, sizeof(diff
), oldsum
);
1289 #ifdef CONFIG_IP_VS_IPV6
1290 static inline __wsum
ip_vs_check_diff16(const __be32
*old
, const __be32
*new,
1293 __be32 diff
[8] = { ~old
[3], ~old
[2], ~old
[1], ~old
[0],
1294 new[3], new[2], new[1], new[0] };
1296 return csum_partial(diff
, sizeof(diff
), oldsum
);
1300 static inline __wsum
ip_vs_check_diff2(__be16 old
, __be16
new, __wsum oldsum
)
1302 __be16 diff
[2] = { ~old
, new };
1304 return csum_partial(diff
, sizeof(diff
), oldsum
);
1308 * Forget current conntrack (unconfirmed) and attach notrack entry
1310 static inline void ip_vs_notrack(struct sk_buff
*skb
)
1312 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1313 enum ip_conntrack_info ctinfo
;
1314 struct nf_conn
*ct
= nf_ct_get(skb
, &ctinfo
);
1316 if (!ct
|| !nf_ct_is_untracked(ct
)) {
1318 skb
->nfct
= &nf_ct_untracked_get()->ct_general
;
1319 skb
->nfctinfo
= IP_CT_NEW
;
1320 nf_conntrack_get(skb
->nfct
);
1325 #ifdef CONFIG_IP_VS_NFCT
1327 * Netfilter connection tracking
1328 * (from ip_vs_nfct.c)
1330 static inline int ip_vs_conntrack_enabled(struct netns_ipvs
*ipvs
)
1332 return ipvs
->sysctl_conntrack
;
1335 extern void ip_vs_update_conntrack(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1337 extern int ip_vs_confirm_conntrack(struct sk_buff
*skb
, struct ip_vs_conn
*cp
);
1338 extern void ip_vs_nfct_expect_related(struct sk_buff
*skb
, struct nf_conn
*ct
,
1339 struct ip_vs_conn
*cp
, u_int8_t proto
,
1340 const __be16 port
, int from_rs
);
1341 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn
*cp
);
1345 static inline int ip_vs_conntrack_enabled(struct netns_ipvs
*ipvs
)
1350 static inline void ip_vs_update_conntrack(struct sk_buff
*skb
,
1351 struct ip_vs_conn
*cp
, int outin
)
1355 static inline int ip_vs_confirm_conntrack(struct sk_buff
*skb
,
1356 struct ip_vs_conn
*cp
)
1361 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn
*cp
)
1364 /* CONFIG_IP_VS_NFCT */
1367 static inline unsigned int
1368 ip_vs_dest_conn_overhead(struct ip_vs_dest
*dest
)
1371 * We think the overhead of processing active connections is 256
1372 * times higher than that of inactive connections in average. (This
1373 * 256 times might not be accurate, we will change it later) We
1374 * use the following formula to estimate the overhead now:
1375 * dest->activeconns*256 + dest->inactconns
1377 return (atomic_read(&dest
->activeconns
) << 8) +
1378 atomic_read(&dest
->inactconns
);
1381 #endif /* __KERNEL__ */
1383 #endif /* _NET_IP_VS_H */