3 * data structure and functionality definitions
9 #include <linux/ip_vs.h> /* definitions shared with userland */
11 #include <asm/types.h> /* for __uXX types */
13 #include <linux/list.h> /* for struct list_head */
14 #include <linux/spinlock.h> /* for struct rwlock_t */
15 #include <linux/atomic.h> /* for struct atomic_t */
16 #include <linux/compiler.h>
17 #include <linux/timer.h>
18 #include <linux/bug.h>
20 #include <net/checksum.h>
21 #include <linux/netfilter.h> /* for union nf_inet_addr */
23 #include <linux/ipv6.h> /* for struct ipv6hdr */
25 #if IS_ENABLED(CONFIG_IP_VS_IPV6)
26 #include <linux/netfilter_ipv6/ip6_tables.h>
28 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
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) && 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 can't 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;
110 __u32 len; /* IPv4 simply where L4 starts
111 IPv6 where L4 Transport Header starts */
112 __u32 thoff_reasm; /* Transport Header Offset in nfct_reasm skb */
113 __u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
116 union nf_inet_addr saddr;
117 union nf_inet_addr daddr;
120 /* Dependency to module: nf_defrag_ipv6 */
121 #if defined(CONFIG_NF_DEFRAG_IPV6) || defined(CONFIG_NF_DEFRAG_IPV6_MODULE)
122 static inline struct sk_buff *skb_nfct_reasm(const struct sk_buff *skb)
124 return skb->nfct_reasm;
126 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
127 int len, void *buffer,
128 const struct ip_vs_iphdr *ipvsh)
130 if (unlikely(ipvsh->fragoffs && skb_nfct_reasm(skb)))
131 return skb_header_pointer(skb_nfct_reasm(skb),
132 ipvsh->thoff_reasm, len, buffer);
134 return skb_header_pointer(skb, offset, len, buffer);
137 static inline struct sk_buff *skb_nfct_reasm(const struct sk_buff *skb)
141 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
142 int len, void *buffer,
143 const struct ip_vs_iphdr *ipvsh)
145 return skb_header_pointer(skb, offset, len, buffer);
150 ip_vs_fill_ip4hdr(const void *nh, struct ip_vs_iphdr *iphdr)
152 const struct iphdr *iph = nh;
154 iphdr->len = iph->ihl * 4;
156 iphdr->protocol = iph->protocol;
157 iphdr->saddr.ip = iph->saddr;
158 iphdr->daddr.ip = iph->daddr;
161 /* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
162 * IPv6 requires some extra work, as finding proper header position,
163 * depend on the IPv6 extension headers.
166 ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, struct ip_vs_iphdr *iphdr)
168 #ifdef CONFIG_IP_VS_IPV6
169 if (af == AF_INET6) {
170 const struct ipv6hdr *iph =
171 (struct ipv6hdr *)skb_network_header(skb);
172 iphdr->saddr.in6 = iph->saddr;
173 iphdr->daddr.in6 = iph->daddr;
174 /* ipv6_find_hdr() updates len, flags, thoff_reasm */
175 iphdr->thoff_reasm = 0;
178 iphdr->protocol = ipv6_find_hdr(skb, &iphdr->len, -1,
181 /* get proto from re-assembled packet and it's offset */
182 if (skb_nfct_reasm(skb))
183 iphdr->protocol = ipv6_find_hdr(skb_nfct_reasm(skb),
190 const struct iphdr *iph =
191 (struct iphdr *)skb_network_header(skb);
192 iphdr->len = iph->ihl * 4;
194 iphdr->protocol = iph->protocol;
195 iphdr->saddr.ip = iph->saddr;
196 iphdr->daddr.ip = iph->daddr;
200 /* This function is a faster version of ip_vs_fill_iph_skb().
201 * Where we only populate {s,d}addr (and avoid calling ipv6_find_hdr()).
202 * This is used by the some of the ip_vs_*_schedule() functions.
203 * (Mostly done to avoid ABI breakage of external schedulers)
206 ip_vs_fill_iph_addr_only(int af, const struct sk_buff *skb,
207 struct ip_vs_iphdr *iphdr)
209 #ifdef CONFIG_IP_VS_IPV6
210 if (af == AF_INET6) {
211 const struct ipv6hdr *iph =
212 (struct ipv6hdr *)skb_network_header(skb);
213 iphdr->saddr.in6 = iph->saddr;
214 iphdr->daddr.in6 = iph->daddr;
218 const struct iphdr *iph =
219 (struct iphdr *)skb_network_header(skb);
220 iphdr->saddr.ip = iph->saddr;
221 iphdr->daddr.ip = iph->daddr;
225 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
226 const union nf_inet_addr *src)
228 #ifdef CONFIG_IP_VS_IPV6
236 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
237 const union nf_inet_addr *src)
239 #ifdef CONFIG_IP_VS_IPV6
240 if (af == AF_INET6) {
251 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
252 const union nf_inet_addr *b)
254 #ifdef CONFIG_IP_VS_IPV6
256 return ipv6_addr_equal(&a->in6, &b->in6);
258 return a->ip == b->ip;
261 #ifdef CONFIG_IP_VS_DEBUG
262 #include <linux/net.h>
264 extern int ip_vs_get_debug_level(void);
266 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
267 const union nf_inet_addr *addr,
271 #ifdef CONFIG_IP_VS_IPV6
273 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
277 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
281 BUG_ON(*idx > buf_len + 1);
282 return &buf[*idx - len];
285 #define IP_VS_DBG_BUF(level, msg, ...) \
287 char ip_vs_dbg_buf[160]; \
288 int ip_vs_dbg_idx = 0; \
289 if (level <= ip_vs_get_debug_level()) \
290 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
292 #define IP_VS_ERR_BUF(msg...) \
294 char ip_vs_dbg_buf[160]; \
295 int ip_vs_dbg_idx = 0; \
299 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
300 #define IP_VS_DBG_ADDR(af, addr) \
301 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
302 sizeof(ip_vs_dbg_buf), addr, \
305 #define IP_VS_DBG(level, msg, ...) \
307 if (level <= ip_vs_get_debug_level()) \
308 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
310 #define IP_VS_DBG_RL(msg, ...) \
312 if (net_ratelimit()) \
313 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
315 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
317 if (level <= ip_vs_get_debug_level()) \
318 pp->debug_packet(af, pp, skb, ofs, msg); \
320 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
322 if (level <= ip_vs_get_debug_level() && \
324 pp->debug_packet(af, pp, skb, ofs, msg); \
326 #else /* NO DEBUGGING at ALL */
327 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
328 #define IP_VS_ERR_BUF(msg...) do {} while (0)
329 #define IP_VS_DBG(level, msg...) do {} while (0)
330 #define IP_VS_DBG_RL(msg...) do {} while (0)
331 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
332 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
335 #define IP_VS_BUG() BUG()
336 #define IP_VS_ERR_RL(msg, ...) \
338 if (net_ratelimit()) \
339 pr_err(msg, ##__VA_ARGS__); \
342 #ifdef CONFIG_IP_VS_DEBUG
343 #define EnterFunction(level) \
345 if (level <= ip_vs_get_debug_level()) \
347 pr_fmt("Enter: %s, %s line %i\n"), \
348 __func__, __FILE__, __LINE__); \
350 #define LeaveFunction(level) \
352 if (level <= ip_vs_get_debug_level()) \
354 pr_fmt("Leave: %s, %s line %i\n"), \
355 __func__, __FILE__, __LINE__); \
358 #define EnterFunction(level) do {} while (0)
359 #define LeaveFunction(level) do {} while (0)
362 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
366 * The port number of FTP service (in network order).
368 #define FTPPORT cpu_to_be16(21)
369 #define FTPDATA cpu_to_be16(20)
375 IP_VS_TCP_S_NONE = 0,
376 IP_VS_TCP_S_ESTABLISHED,
377 IP_VS_TCP_S_SYN_SENT,
378 IP_VS_TCP_S_SYN_RECV,
379 IP_VS_TCP_S_FIN_WAIT,
380 IP_VS_TCP_S_TIME_WAIT,
382 IP_VS_TCP_S_CLOSE_WAIT,
383 IP_VS_TCP_S_LAST_ACK,
408 enum ip_vs_sctp_states {
410 IP_VS_SCTP_S_INIT_CLI,
411 IP_VS_SCTP_S_INIT_SER,
412 IP_VS_SCTP_S_INIT_ACK_CLI,
413 IP_VS_SCTP_S_INIT_ACK_SER,
414 IP_VS_SCTP_S_ECHO_CLI,
415 IP_VS_SCTP_S_ECHO_SER,
416 IP_VS_SCTP_S_ESTABLISHED,
417 IP_VS_SCTP_S_SHUT_CLI,
418 IP_VS_SCTP_S_SHUT_SER,
419 IP_VS_SCTP_S_SHUT_ACK_CLI,
420 IP_VS_SCTP_S_SHUT_ACK_SER,
426 * Delta sequence info structure
427 * Each ip_vs_conn has 2 (output AND input seq. changes).
428 * Only used in the VS/NAT.
431 __u32 init_seq; /* Add delta from this seq */
432 __u32 delta; /* Delta in sequence numbers */
433 __u32 previous_delta; /* Delta in sequence numbers
434 before last resized pkt */
440 struct ip_vs_counters {
441 __u32 conns; /* connections scheduled */
442 __u32 inpkts; /* incoming packets */
443 __u32 outpkts; /* outgoing packets */
444 __u64 inbytes; /* incoming bytes */
445 __u64 outbytes; /* outgoing bytes */
450 struct ip_vs_cpu_stats {
451 struct ip_vs_counters ustats;
452 struct u64_stats_sync syncp;
456 * IPVS statistics objects
458 struct ip_vs_estimator {
459 struct list_head list;
475 struct ip_vs_stats_user ustats; /* statistics */
476 struct ip_vs_estimator est; /* estimator */
477 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
478 spinlock_t lock; /* spin lock */
479 struct ip_vs_stats_user ustats0; /* reset values */
487 struct ip_vs_proto_data;
489 struct ip_vs_protocol {
490 struct ip_vs_protocol *next;
496 void (*init)(struct ip_vs_protocol *pp);
498 void (*exit)(struct ip_vs_protocol *pp);
500 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
502 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
504 int (*conn_schedule)(int af, struct sk_buff *skb,
505 struct ip_vs_proto_data *pd,
506 int *verdict, struct ip_vs_conn **cpp,
507 struct ip_vs_iphdr *iph);
510 (*conn_in_get)(int af,
511 const struct sk_buff *skb,
512 const struct ip_vs_iphdr *iph,
516 (*conn_out_get)(int af,
517 const struct sk_buff *skb,
518 const struct ip_vs_iphdr *iph,
521 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
522 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
524 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
525 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
527 int (*csum_check)(int af, struct sk_buff *skb,
528 struct ip_vs_protocol *pp);
530 const char *(*state_name)(int state);
532 void (*state_transition)(struct ip_vs_conn *cp, int direction,
533 const struct sk_buff *skb,
534 struct ip_vs_proto_data *pd);
536 int (*register_app)(struct net *net, struct ip_vs_app *inc);
538 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
540 int (*app_conn_bind)(struct ip_vs_conn *cp);
542 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
543 const struct sk_buff *skb,
547 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
551 * protocol data per netns
553 struct ip_vs_proto_data {
554 struct ip_vs_proto_data *next;
555 struct ip_vs_protocol *pp;
556 int *timeout_table; /* protocol timeout table */
557 atomic_t appcnt; /* counter of proto app incs. */
558 struct tcp_states_t *tcp_state_table;
561 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
562 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
563 unsigned short proto);
565 struct ip_vs_conn_param {
567 const union nf_inet_addr *caddr;
568 const union nf_inet_addr *vaddr;
574 const struct ip_vs_pe *pe;
580 * IP_VS structure allocated for each dynamically scheduled connection
583 struct hlist_node c_list; /* hashed list heads */
584 /* Protocol, addresses and port numbers */
588 u16 af; /* address family */
589 union nf_inet_addr caddr; /* client address */
590 union nf_inet_addr vaddr; /* virtual address */
591 union nf_inet_addr daddr; /* destination address */
592 volatile __u32 flags; /* status flags */
593 __u16 protocol; /* Which protocol (TCP/UDP) */
595 struct net *net; /* Name space */
598 /* counter and timer */
599 atomic_t refcnt; /* reference count */
600 struct timer_list timer; /* Expiration timer */
601 volatile unsigned long timeout; /* timeout */
603 /* Flags and state transition */
604 spinlock_t lock; /* lock for state transition */
605 volatile __u16 state; /* state info */
606 volatile __u16 old_state; /* old state, to be used for
607 * state transition triggerd
610 __u32 fwmark; /* Fire wall mark from skb */
611 unsigned long sync_endtime; /* jiffies + sent_retries */
613 /* Control members */
614 struct ip_vs_conn *control; /* Master control connection */
615 atomic_t n_control; /* Number of controlled ones */
616 struct ip_vs_dest *dest; /* real server */
617 atomic_t in_pkts; /* incoming packet counter */
619 /* packet transmitter for different forwarding methods. If it
620 mangles the packet, it must return NF_DROP or better NF_STOLEN,
621 otherwise this must be changed to a sk_buff **.
622 NF_ACCEPT can be returned when destination is local.
624 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
625 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
627 /* Note: we can group the following members into a structure,
628 in order to save more space, and the following members are
629 only used in VS/NAT anyway */
630 struct ip_vs_app *app; /* bound ip_vs_app object */
631 void *app_data; /* Application private data */
632 struct ip_vs_seq in_seq; /* incoming seq. struct */
633 struct ip_vs_seq out_seq; /* outgoing seq. struct */
635 const struct ip_vs_pe *pe;
639 struct rcu_head rcu_head;
643 * To save some memory in conn table when name space is disabled.
645 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
653 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
660 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
664 return cp->net == net;
671 * Extended internal versions of struct ip_vs_service_user and
672 * ip_vs_dest_user for IPv6 support.
674 * We need these to conveniently pass around service and destination
675 * options, but unfortunately, we also need to keep the old definitions to
676 * maintain userspace backwards compatibility for the setsockopt interface.
678 struct ip_vs_service_user_kern {
679 /* virtual service addresses */
682 union nf_inet_addr addr; /* virtual ip address */
684 u32 fwmark; /* firwall mark of service */
686 /* virtual service options */
689 unsigned int flags; /* virtual service flags */
690 unsigned int timeout; /* persistent timeout in sec */
691 u32 netmask; /* persistent netmask */
695 struct ip_vs_dest_user_kern {
696 /* destination server address */
697 union nf_inet_addr addr;
700 /* real server options */
701 unsigned int conn_flags; /* connection flags */
702 int weight; /* destination weight */
704 /* thresholds for active connections */
705 u32 u_threshold; /* upper threshold */
706 u32 l_threshold; /* lower threshold */
711 * The information about the virtual service offered to the net
712 * and the forwarding entries
714 struct ip_vs_service {
715 struct list_head s_list; /* for normal service table */
716 struct list_head f_list; /* for fwmark-based service table */
717 atomic_t refcnt; /* reference counter */
718 atomic_t usecnt; /* use counter */
720 u16 af; /* address family */
721 __u16 protocol; /* which protocol (TCP/UDP) */
722 union nf_inet_addr addr; /* IP address for virtual service */
723 __be16 port; /* port number for the service */
724 __u32 fwmark; /* firewall mark of the service */
725 unsigned int flags; /* service status flags */
726 unsigned int timeout; /* persistent timeout in ticks */
727 __be32 netmask; /* grouping granularity */
730 struct list_head destinations; /* real server d-linked list */
731 __u32 num_dests; /* number of servers */
732 struct ip_vs_stats stats; /* statistics for the service */
733 struct ip_vs_app *inc; /* bind conns to this app inc */
736 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
737 rwlock_t sched_lock; /* lock sched_data */
738 void *sched_data; /* scheduler application data */
740 /* alternate persistence engine */
744 /* Information for cached dst */
745 struct ip_vs_dest_dst {
746 struct dst_entry *dst_cache; /* destination cache entry */
748 union nf_inet_addr dst_saddr;
749 struct rcu_head rcu_head;
753 * The real server destination forwarding entry
754 * with ip address, port number, and so on.
757 struct list_head n_list; /* for the dests in the service */
758 struct hlist_node d_list; /* for table with all the dests */
760 u16 af; /* address family */
761 __be16 port; /* port number of the server */
762 union nf_inet_addr addr; /* IP address of the server */
763 volatile unsigned int flags; /* dest status flags */
764 atomic_t conn_flags; /* flags to copy to conn */
765 atomic_t weight; /* server weight */
767 atomic_t refcnt; /* reference counter */
768 struct ip_vs_stats stats; /* statistics */
770 /* connection counters and thresholds */
771 atomic_t activeconns; /* active connections */
772 atomic_t inactconns; /* inactive connections */
773 atomic_t persistconns; /* persistent connections */
774 __u32 u_threshold; /* upper threshold */
775 __u32 l_threshold; /* lower threshold */
777 /* for destination cache */
778 spinlock_t dst_lock; /* lock of dst_cache */
779 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
781 /* for virtual service */
782 struct ip_vs_service *svc; /* service it belongs to */
783 __u16 protocol; /* which protocol (TCP/UDP) */
784 __be16 vport; /* virtual port number */
785 union nf_inet_addr vaddr; /* virtual IP address */
786 __u32 vfwmark; /* firewall mark of service */
788 struct rcu_head rcu_head;
789 unsigned int in_rs_table:1; /* we are in rs_table */
794 * The scheduler object
796 struct ip_vs_scheduler {
797 struct list_head n_list; /* d-linked list head */
798 char *name; /* scheduler name */
799 atomic_t refcnt; /* reference counter */
800 struct module *module; /* THIS_MODULE/NULL */
802 /* scheduler initializing service */
803 int (*init_service)(struct ip_vs_service *svc);
804 /* scheduling service finish */
805 int (*done_service)(struct ip_vs_service *svc);
806 /* scheduler updating service */
807 int (*update_service)(struct ip_vs_service *svc);
809 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
810 /* dest is unlinked */
811 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
812 /* dest is updated */
813 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
815 /* selecting a server from the given service */
816 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
817 const struct sk_buff *skb);
820 /* The persistence engine object */
822 struct list_head n_list; /* d-linked list head */
823 char *name; /* scheduler name */
824 atomic_t refcnt; /* reference counter */
825 struct module *module; /* THIS_MODULE/NULL */
827 /* get the connection template, if any */
828 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
829 bool (*ct_match)(const struct ip_vs_conn_param *p,
830 struct ip_vs_conn *ct);
831 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
833 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
837 * The application module object (a.k.a. app incarnation)
840 struct list_head a_list; /* member in app list */
841 int type; /* IP_VS_APP_TYPE_xxx */
842 char *name; /* application module name */
844 struct module *module; /* THIS_MODULE/NULL */
845 struct list_head incs_list; /* list of incarnations */
847 /* members for application incarnations */
848 struct list_head p_list; /* member in proto app list */
849 struct ip_vs_app *app; /* its real application */
850 __be16 port; /* port number in net order */
851 atomic_t usecnt; /* usage counter */
852 struct rcu_head rcu_head;
855 * output hook: Process packet in inout direction, diff set for TCP.
856 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
857 * 2=Mangled but checksum was not updated
859 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
860 struct sk_buff *, int *diff);
863 * input hook: Process packet in outin direction, diff set for TCP.
864 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
865 * 2=Mangled but checksum was not updated
867 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
868 struct sk_buff *, int *diff);
870 /* ip_vs_app initializer */
871 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
873 /* ip_vs_app finish */
874 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
878 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
879 struct ip_vs_protocol *);
881 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
887 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
888 int *verdict, struct ip_vs_conn **cpp);
891 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
892 const struct iphdr *iph, int inverse);
895 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
896 const struct iphdr *iph, int inverse);
898 int (*state_transition)(struct ip_vs_conn *cp, int direction,
899 const struct sk_buff *skb,
900 struct ip_vs_app *app);
902 void (*timeout_change)(struct ip_vs_app *app, int flags);
905 struct ipvs_master_sync_state {
906 struct list_head sync_queue;
907 struct ip_vs_sync_buff *sync_buff;
909 unsigned int sync_queue_delay;
910 struct task_struct *master_thread;
911 struct delayed_work master_wakeup_work;
912 struct netns_ipvs *ipvs;
915 /* IPVS in network namespace */
917 int gen; /* Generation */
918 int enable; /* enable like nf_hooks do */
920 * Hash table: for real service lookups
922 #define IP_VS_RTAB_BITS 4
923 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
924 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
926 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
928 struct list_head app_list;
930 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
931 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
932 /* ip_vs_proto_tcp */
933 #ifdef CONFIG_IP_VS_PROTO_TCP
934 #define TCP_APP_TAB_BITS 4
935 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
936 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
937 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
939 /* ip_vs_proto_udp */
940 #ifdef CONFIG_IP_VS_PROTO_UDP
941 #define UDP_APP_TAB_BITS 4
942 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
943 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
944 struct list_head udp_apps[UDP_APP_TAB_SIZE];
946 /* ip_vs_proto_sctp */
947 #ifdef CONFIG_IP_VS_PROTO_SCTP
948 #define SCTP_APP_TAB_BITS 4
949 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
950 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
951 /* Hash table for SCTP application incarnations */
952 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
955 atomic_t conn_count; /* connection counter */
958 struct ip_vs_stats tot_stats; /* Statistics & est. */
960 int num_services; /* no of virtual services */
962 /* Trash for destinations */
963 struct list_head dest_trash;
964 /* Service counters */
965 atomic_t ftpsvc_counter;
966 atomic_t nullsvc_counter;
969 /* 1/rate drop and drop-entry variables */
970 struct delayed_work defense_work; /* Work handler */
975 spinlock_t dropentry_lock; /* drop entry handling */
976 spinlock_t droppacket_lock; /* drop packet handling */
977 spinlock_t securetcp_lock; /* state and timeout tables */
980 struct ctl_table_header *sysctl_hdr;
981 struct ctl_table *sysctl_tbl;
984 /* sysctl variables */
985 int sysctl_amemthresh;
986 int sysctl_am_droprate;
987 int sysctl_drop_entry;
988 int sysctl_drop_packet;
989 int sysctl_secure_tcp;
990 #ifdef CONFIG_IP_VS_NFCT
991 int sysctl_conntrack;
993 int sysctl_snat_reroute;
995 int sysctl_sync_ports;
996 int sysctl_sync_qlen_max;
997 int sysctl_sync_sock_size;
998 int sysctl_cache_bypass;
999 int sysctl_expire_nodest_conn;
1000 int sysctl_expire_quiescent_template;
1001 int sysctl_sync_threshold[2];
1002 unsigned int sysctl_sync_refresh_period;
1003 int sysctl_sync_retries;
1004 int sysctl_nat_icmp_send;
1005 int sysctl_pmtu_disc;
1006 int sysctl_backup_only;
1009 int sysctl_lblc_expiration;
1010 struct ctl_table_header *lblc_ctl_header;
1011 struct ctl_table *lblc_ctl_table;
1013 int sysctl_lblcr_expiration;
1014 struct ctl_table_header *lblcr_ctl_header;
1015 struct ctl_table *lblcr_ctl_table;
1017 struct list_head est_list; /* estimator list */
1018 spinlock_t est_lock;
1019 struct timer_list est_timer; /* Estimation timer */
1021 spinlock_t sync_lock;
1022 struct ipvs_master_sync_state *ms;
1023 spinlock_t sync_buff_lock;
1024 struct task_struct **backup_threads;
1026 int send_mesg_maxlen;
1027 int recv_mesg_maxlen;
1028 volatile int sync_state;
1029 volatile int master_syncid;
1030 volatile int backup_syncid;
1031 struct mutex sync_mutex;
1032 /* multicast interface name */
1033 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1034 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1035 /* net name space ptr */
1036 struct net *net; /* Needed by timer routines */
1039 #define DEFAULT_SYNC_THRESHOLD 3
1040 #define DEFAULT_SYNC_PERIOD 50
1041 #define DEFAULT_SYNC_VER 1
1042 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1043 #define DEFAULT_SYNC_RETRIES 0
1044 #define IPVS_SYNC_WAKEUP_RATE 8
1045 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1046 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1047 #define IPVS_SYNC_CHECK_PERIOD HZ
1048 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1049 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1051 #ifdef CONFIG_SYSCTL
1053 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1055 return ipvs->sysctl_sync_threshold[0];
1058 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1060 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
1063 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1065 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
1068 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1070 return ipvs->sysctl_sync_retries;
1073 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1075 return ipvs->sysctl_sync_ver;
1078 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1080 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
1083 static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1085 return ipvs->sysctl_sync_qlen_max;
1088 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1090 return ipvs->sysctl_sync_sock_size;
1093 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1095 return ipvs->sysctl_pmtu_disc;
1098 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1100 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1101 ipvs->sysctl_backup_only;
1106 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1108 return DEFAULT_SYNC_THRESHOLD;
1111 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1113 return DEFAULT_SYNC_PERIOD;
1116 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1118 return DEFAULT_SYNC_REFRESH_PERIOD;
1121 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1123 return DEFAULT_SYNC_RETRIES & 3;
1126 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1128 return DEFAULT_SYNC_VER;
1131 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1136 static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1138 return IPVS_SYNC_QLEN_MAX;
1141 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1146 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1151 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1159 * IPVS core functions
1160 * (from ip_vs_core.c)
1162 extern const char *ip_vs_proto_name(unsigned int proto);
1163 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
1164 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1166 #define IP_VS_APP_TYPE_FTP 1
1169 * ip_vs_conn handling functions
1170 * (from ip_vs_conn.c)
1174 IP_VS_DIR_INPUT = 0,
1176 IP_VS_DIR_INPUT_ONLY,
1180 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1181 const union nf_inet_addr *caddr,
1183 const union nf_inet_addr *vaddr,
1185 struct ip_vs_conn_param *p)
1189 p->protocol = protocol;
1198 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1199 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1201 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1202 const struct ip_vs_iphdr *iph,
1205 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1207 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1208 const struct ip_vs_iphdr *iph,
1211 /* Get reference to gain full access to conn.
1212 * By default, RCU read-side critical sections have access only to
1213 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1215 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1217 return atomic_inc_not_zero(&cp->refcnt);
1220 /* put back the conn without restarting its timer */
1221 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1223 smp_mb__before_atomic_dec();
1224 atomic_dec(&cp->refcnt);
1226 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1227 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1229 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1230 const union nf_inet_addr *daddr,
1231 __be16 dport, unsigned int flags,
1232 struct ip_vs_dest *dest, __u32 fwmark);
1233 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1235 extern const char * ip_vs_state_name(__u16 proto, int state);
1237 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1238 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1239 extern void ip_vs_random_dropentry(struct net *net);
1240 extern int ip_vs_conn_init(void);
1241 extern void ip_vs_conn_cleanup(void);
1243 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1245 struct ip_vs_conn *ctl_cp = cp->control;
1247 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1249 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1251 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1257 IP_VS_DBG_BUF(7, "DELeting control for: "
1258 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1259 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1261 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1262 ntohs(ctl_cp->cport));
1265 if (atomic_read(&ctl_cp->n_control) == 0) {
1266 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1268 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1270 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1275 atomic_dec(&ctl_cp->n_control);
1279 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1282 IP_VS_ERR_BUF("request control ADD for already controlled: "
1284 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1286 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1289 ip_vs_control_del(cp);
1292 IP_VS_DBG_BUF(7, "ADDing control for: "
1293 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1294 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1296 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1297 ntohs(ctl_cp->cport));
1299 cp->control = ctl_cp;
1300 atomic_inc(&ctl_cp->n_control);
1304 * IPVS netns init & cleanup functions
1306 extern int ip_vs_estimator_net_init(struct net *net);
1307 extern int ip_vs_control_net_init(struct net *net);
1308 extern int ip_vs_protocol_net_init(struct net *net);
1309 extern int ip_vs_app_net_init(struct net *net);
1310 extern int ip_vs_conn_net_init(struct net *net);
1311 extern int ip_vs_sync_net_init(struct net *net);
1312 extern void ip_vs_conn_net_cleanup(struct net *net);
1313 extern void ip_vs_app_net_cleanup(struct net *net);
1314 extern void ip_vs_protocol_net_cleanup(struct net *net);
1315 extern void ip_vs_control_net_cleanup(struct net *net);
1316 extern void ip_vs_estimator_net_cleanup(struct net *net);
1317 extern void ip_vs_sync_net_cleanup(struct net *net);
1318 extern void ip_vs_service_net_cleanup(struct net *net);
1321 * IPVS application functions
1322 * (from ip_vs_app.c)
1324 #define IP_VS_APP_MAX_PORTS 8
1325 extern struct ip_vs_app *register_ip_vs_app(struct net *net,
1326 struct ip_vs_app *app);
1327 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1328 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1329 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1330 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1331 __u16 proto, __u16 port);
1332 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1333 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1335 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1336 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1338 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1339 void ip_vs_unbind_pe(struct ip_vs_service *svc);
1340 int register_ip_vs_pe(struct ip_vs_pe *pe);
1341 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1342 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1343 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1346 * Use a #define to avoid all of module.h just for these trivial ops
1348 #define ip_vs_pe_get(pe) \
1349 if (pe && pe->module) \
1350 __module_get(pe->module);
1352 #define ip_vs_pe_put(pe) \
1353 if (pe && pe->module) \
1354 module_put(pe->module);
1357 * IPVS protocol functions (from ip_vs_proto.c)
1359 extern int ip_vs_protocol_init(void);
1360 extern void ip_vs_protocol_cleanup(void);
1361 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1362 extern int *ip_vs_create_timeout_table(int *table, int size);
1364 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1365 const char *name, int to);
1367 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1368 const struct sk_buff *skb,
1369 int offset, const char *msg);
1371 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1372 extern struct ip_vs_protocol ip_vs_protocol_udp;
1373 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1374 extern struct ip_vs_protocol ip_vs_protocol_esp;
1375 extern struct ip_vs_protocol ip_vs_protocol_ah;
1376 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1379 * Registering/unregistering scheduler functions
1380 * (from ip_vs_sched.c)
1382 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1383 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1384 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1385 struct ip_vs_scheduler *scheduler);
1386 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1387 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1388 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1389 extern struct ip_vs_conn *
1390 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1391 struct ip_vs_proto_data *pd, int *ignored,
1392 struct ip_vs_iphdr *iph);
1393 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1394 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1396 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1400 * IPVS control data and functions (from ip_vs_ctl.c)
1402 extern struct ip_vs_stats ip_vs_stats;
1403 extern int sysctl_ip_vs_sync_ver;
1405 extern struct ip_vs_service *
1406 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1407 const union nf_inet_addr *vaddr, __be16 vport);
1409 static inline void ip_vs_service_put(struct ip_vs_service *svc)
1411 atomic_dec(&svc->usecnt);
1415 ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
1416 const union nf_inet_addr *daddr, __be16 dport);
1418 extern int ip_vs_use_count_inc(void);
1419 extern void ip_vs_use_count_dec(void);
1420 extern int ip_vs_register_nl_ioctl(void);
1421 extern void ip_vs_unregister_nl_ioctl(void);
1422 extern int ip_vs_control_init(void);
1423 extern void ip_vs_control_cleanup(void);
1424 extern struct ip_vs_dest *
1425 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1426 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1427 __u16 protocol, __u32 fwmark, __u32 flags);
1428 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1430 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1432 atomic_inc(&dest->refcnt);
1435 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1437 smp_mb__before_atomic_dec();
1438 atomic_dec(&dest->refcnt);
1442 * IPVS sync daemon data and function prototypes
1443 * (from ip_vs_sync.c)
1445 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1447 extern int stop_sync_thread(struct net *net, int state);
1448 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1452 * IPVS rate estimator prototypes (from ip_vs_est.c)
1454 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1455 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1456 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1457 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1458 struct ip_vs_stats *stats);
1461 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1463 extern int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1464 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1465 extern int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1466 struct ip_vs_protocol *pp,
1467 struct ip_vs_iphdr *iph);
1468 extern int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1469 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1470 extern int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1471 struct ip_vs_protocol *pp,
1472 struct ip_vs_iphdr *iph);
1473 extern int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1474 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1475 extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1476 struct ip_vs_protocol *pp, int offset,
1477 unsigned int hooknum, struct ip_vs_iphdr *iph);
1478 extern void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1480 #ifdef CONFIG_IP_VS_IPV6
1481 extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1482 struct ip_vs_protocol *pp,
1483 struct ip_vs_iphdr *iph);
1484 extern int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1485 struct ip_vs_protocol *pp,
1486 struct ip_vs_iphdr *iph);
1487 extern int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1488 struct ip_vs_protocol *pp,
1489 struct ip_vs_iphdr *iph);
1490 extern int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1491 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1492 extern int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1493 struct ip_vs_protocol *pp, int offset,
1494 unsigned int hooknum, struct ip_vs_iphdr *iph);
1497 #ifdef CONFIG_SYSCTL
1499 * This is a simple mechanism to ignore packets when
1500 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1501 * we start to drop 1/rate of the packets
1504 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1506 if (!ipvs->drop_rate)
1508 if (--ipvs->drop_counter > 0)
1510 ipvs->drop_counter = ipvs->drop_rate;
1514 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1518 * ip_vs_fwd_tag returns the forwarding tag of the connection
1520 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1522 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1526 switch (IP_VS_FWD_METHOD(cp)) {
1527 case IP_VS_CONN_F_MASQ:
1529 case IP_VS_CONN_F_LOCALNODE:
1531 case IP_VS_CONN_F_TUNNEL:
1533 case IP_VS_CONN_F_DROUTE:
1535 case IP_VS_CONN_F_BYPASS:
1543 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1544 struct ip_vs_conn *cp, int dir);
1546 #ifdef CONFIG_IP_VS_IPV6
1547 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1548 struct ip_vs_conn *cp, int dir);
1551 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1553 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1555 __be32 diff[2] = { ~old, new };
1557 return csum_partial(diff, sizeof(diff), oldsum);
1560 #ifdef CONFIG_IP_VS_IPV6
1561 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1564 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1565 new[3], new[2], new[1], new[0] };
1567 return csum_partial(diff, sizeof(diff), oldsum);
1571 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1573 __be16 diff[2] = { ~old, new };
1575 return csum_partial(diff, sizeof(diff), oldsum);
1579 * Forget current conntrack (unconfirmed) and attach notrack entry
1581 static inline void ip_vs_notrack(struct sk_buff *skb)
1583 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1584 enum ip_conntrack_info ctinfo;
1585 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1587 if (!ct || !nf_ct_is_untracked(ct)) {
1588 nf_conntrack_put(skb->nfct);
1589 skb->nfct = &nf_ct_untracked_get()->ct_general;
1590 skb->nfctinfo = IP_CT_NEW;
1591 nf_conntrack_get(skb->nfct);
1596 #ifdef CONFIG_IP_VS_NFCT
1598 * Netfilter connection tracking
1599 * (from ip_vs_nfct.c)
1601 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1603 #ifdef CONFIG_SYSCTL
1604 return ipvs->sysctl_conntrack;
1610 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1612 extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1613 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1614 struct ip_vs_conn *cp, u_int8_t proto,
1615 const __be16 port, int from_rs);
1616 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1620 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1625 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1626 struct ip_vs_conn *cp, int outin)
1630 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1635 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1638 /* CONFIG_IP_VS_NFCT */
1641 static inline unsigned int
1642 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1645 * We think the overhead of processing active connections is 256
1646 * times higher than that of inactive connections in average. (This
1647 * 256 times might not be accurate, we will change it later) We
1648 * use the following formula to estimate the overhead now:
1649 * dest->activeconns*256 + dest->inactconns
1651 return (atomic_read(&dest->activeconns) << 8) +
1652 atomic_read(&dest->inactconns);
1655 #endif /* _NET_IP_VS_H */