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 int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
237 const union nf_inet_addr *b)
239 #ifdef CONFIG_IP_VS_IPV6
241 return ipv6_addr_equal(&a->in6, &b->in6);
243 return a->ip == b->ip;
246 #ifdef CONFIG_IP_VS_DEBUG
247 #include <linux/net.h>
249 extern int ip_vs_get_debug_level(void);
251 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
252 const union nf_inet_addr *addr,
256 #ifdef CONFIG_IP_VS_IPV6
258 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
262 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
266 BUG_ON(*idx > buf_len + 1);
267 return &buf[*idx - len];
270 #define IP_VS_DBG_BUF(level, msg, ...) \
272 char ip_vs_dbg_buf[160]; \
273 int ip_vs_dbg_idx = 0; \
274 if (level <= ip_vs_get_debug_level()) \
275 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
277 #define IP_VS_ERR_BUF(msg...) \
279 char ip_vs_dbg_buf[160]; \
280 int ip_vs_dbg_idx = 0; \
284 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
285 #define IP_VS_DBG_ADDR(af, addr) \
286 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
287 sizeof(ip_vs_dbg_buf), addr, \
290 #define IP_VS_DBG(level, msg, ...) \
292 if (level <= ip_vs_get_debug_level()) \
293 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
295 #define IP_VS_DBG_RL(msg, ...) \
297 if (net_ratelimit()) \
298 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
300 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
302 if (level <= ip_vs_get_debug_level()) \
303 pp->debug_packet(af, pp, skb, ofs, msg); \
305 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
307 if (level <= ip_vs_get_debug_level() && \
309 pp->debug_packet(af, pp, skb, ofs, msg); \
311 #else /* NO DEBUGGING at ALL */
312 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
313 #define IP_VS_ERR_BUF(msg...) do {} while (0)
314 #define IP_VS_DBG(level, msg...) do {} while (0)
315 #define IP_VS_DBG_RL(msg...) do {} while (0)
316 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
317 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
320 #define IP_VS_BUG() BUG()
321 #define IP_VS_ERR_RL(msg, ...) \
323 if (net_ratelimit()) \
324 pr_err(msg, ##__VA_ARGS__); \
327 #ifdef CONFIG_IP_VS_DEBUG
328 #define EnterFunction(level) \
330 if (level <= ip_vs_get_debug_level()) \
332 pr_fmt("Enter: %s, %s line %i\n"), \
333 __func__, __FILE__, __LINE__); \
335 #define LeaveFunction(level) \
337 if (level <= ip_vs_get_debug_level()) \
339 pr_fmt("Leave: %s, %s line %i\n"), \
340 __func__, __FILE__, __LINE__); \
343 #define EnterFunction(level) do {} while (0)
344 #define LeaveFunction(level) do {} while (0)
347 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
351 * The port number of FTP service (in network order).
353 #define FTPPORT cpu_to_be16(21)
354 #define FTPDATA cpu_to_be16(20)
360 IP_VS_TCP_S_NONE = 0,
361 IP_VS_TCP_S_ESTABLISHED,
362 IP_VS_TCP_S_SYN_SENT,
363 IP_VS_TCP_S_SYN_RECV,
364 IP_VS_TCP_S_FIN_WAIT,
365 IP_VS_TCP_S_TIME_WAIT,
367 IP_VS_TCP_S_CLOSE_WAIT,
368 IP_VS_TCP_S_LAST_ACK,
393 enum ip_vs_sctp_states {
395 IP_VS_SCTP_S_INIT_CLI,
396 IP_VS_SCTP_S_INIT_SER,
397 IP_VS_SCTP_S_INIT_ACK_CLI,
398 IP_VS_SCTP_S_INIT_ACK_SER,
399 IP_VS_SCTP_S_ECHO_CLI,
400 IP_VS_SCTP_S_ECHO_SER,
401 IP_VS_SCTP_S_ESTABLISHED,
402 IP_VS_SCTP_S_SHUT_CLI,
403 IP_VS_SCTP_S_SHUT_SER,
404 IP_VS_SCTP_S_SHUT_ACK_CLI,
405 IP_VS_SCTP_S_SHUT_ACK_SER,
411 * Delta sequence info structure
412 * Each ip_vs_conn has 2 (output AND input seq. changes).
413 * Only used in the VS/NAT.
416 __u32 init_seq; /* Add delta from this seq */
417 __u32 delta; /* Delta in sequence numbers */
418 __u32 previous_delta; /* Delta in sequence numbers
419 before last resized pkt */
425 struct ip_vs_counters {
426 __u32 conns; /* connections scheduled */
427 __u32 inpkts; /* incoming packets */
428 __u32 outpkts; /* outgoing packets */
429 __u64 inbytes; /* incoming bytes */
430 __u64 outbytes; /* outgoing bytes */
435 struct ip_vs_cpu_stats {
436 struct ip_vs_counters ustats;
437 struct u64_stats_sync syncp;
441 * IPVS statistics objects
443 struct ip_vs_estimator {
444 struct list_head list;
460 struct ip_vs_stats_user ustats; /* statistics */
461 struct ip_vs_estimator est; /* estimator */
462 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
463 spinlock_t lock; /* spin lock */
464 struct ip_vs_stats_user ustats0; /* reset values */
472 struct ip_vs_proto_data;
474 struct ip_vs_protocol {
475 struct ip_vs_protocol *next;
481 void (*init)(struct ip_vs_protocol *pp);
483 void (*exit)(struct ip_vs_protocol *pp);
485 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
487 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
489 int (*conn_schedule)(int af, struct sk_buff *skb,
490 struct ip_vs_proto_data *pd,
491 int *verdict, struct ip_vs_conn **cpp,
492 struct ip_vs_iphdr *iph);
495 (*conn_in_get)(int af,
496 const struct sk_buff *skb,
497 const struct ip_vs_iphdr *iph,
501 (*conn_out_get)(int af,
502 const struct sk_buff *skb,
503 const struct ip_vs_iphdr *iph,
506 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
507 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
509 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
510 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
512 int (*csum_check)(int af, struct sk_buff *skb,
513 struct ip_vs_protocol *pp);
515 const char *(*state_name)(int state);
517 void (*state_transition)(struct ip_vs_conn *cp, int direction,
518 const struct sk_buff *skb,
519 struct ip_vs_proto_data *pd);
521 int (*register_app)(struct net *net, struct ip_vs_app *inc);
523 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
525 int (*app_conn_bind)(struct ip_vs_conn *cp);
527 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
528 const struct sk_buff *skb,
532 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
536 * protocol data per netns
538 struct ip_vs_proto_data {
539 struct ip_vs_proto_data *next;
540 struct ip_vs_protocol *pp;
541 int *timeout_table; /* protocol timeout table */
542 atomic_t appcnt; /* counter of proto app incs. */
543 struct tcp_states_t *tcp_state_table;
546 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
547 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
548 unsigned short proto);
550 struct ip_vs_conn_param {
552 const union nf_inet_addr *caddr;
553 const union nf_inet_addr *vaddr;
559 const struct ip_vs_pe *pe;
565 * IP_VS structure allocated for each dynamically scheduled connection
568 struct hlist_node c_list; /* hashed list heads */
570 struct net *net; /* Name space */
572 /* Protocol, addresses and port numbers */
573 u16 af; /* address family */
577 __u32 fwmark; /* Fire wall mark from skb */
578 union nf_inet_addr caddr; /* client address */
579 union nf_inet_addr vaddr; /* virtual address */
580 union nf_inet_addr daddr; /* destination address */
581 volatile __u32 flags; /* status flags */
582 __u16 protocol; /* Which protocol (TCP/UDP) */
584 /* counter and timer */
585 atomic_t refcnt; /* reference count */
586 struct timer_list timer; /* Expiration timer */
587 volatile unsigned long timeout; /* timeout */
589 /* Flags and state transition */
590 spinlock_t lock; /* lock for state transition */
591 volatile __u16 state; /* state info */
592 volatile __u16 old_state; /* old state, to be used for
593 * state transition triggerd
596 unsigned long sync_endtime; /* jiffies + sent_retries */
598 /* Control members */
599 struct ip_vs_conn *control; /* Master control connection */
600 atomic_t n_control; /* Number of controlled ones */
601 struct ip_vs_dest *dest; /* real server */
602 atomic_t in_pkts; /* incoming packet counter */
604 /* packet transmitter for different forwarding methods. If it
605 mangles the packet, it must return NF_DROP or better NF_STOLEN,
606 otherwise this must be changed to a sk_buff **.
607 NF_ACCEPT can be returned when destination is local.
609 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
610 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
612 /* Note: we can group the following members into a structure,
613 in order to save more space, and the following members are
614 only used in VS/NAT anyway */
615 struct ip_vs_app *app; /* bound ip_vs_app object */
616 void *app_data; /* Application private data */
617 struct ip_vs_seq in_seq; /* incoming seq. struct */
618 struct ip_vs_seq out_seq; /* outgoing seq. struct */
620 const struct ip_vs_pe *pe;
626 * To save some memory in conn table when name space is disabled.
628 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
636 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
643 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
647 return cp->net == net;
654 * Extended internal versions of struct ip_vs_service_user and
655 * ip_vs_dest_user for IPv6 support.
657 * We need these to conveniently pass around service and destination
658 * options, but unfortunately, we also need to keep the old definitions to
659 * maintain userspace backwards compatibility for the setsockopt interface.
661 struct ip_vs_service_user_kern {
662 /* virtual service addresses */
665 union nf_inet_addr addr; /* virtual ip address */
667 u32 fwmark; /* firwall mark of service */
669 /* virtual service options */
672 unsigned int flags; /* virtual service flags */
673 unsigned int timeout; /* persistent timeout in sec */
674 u32 netmask; /* persistent netmask */
678 struct ip_vs_dest_user_kern {
679 /* destination server address */
680 union nf_inet_addr addr;
683 /* real server options */
684 unsigned int conn_flags; /* connection flags */
685 int weight; /* destination weight */
687 /* thresholds for active connections */
688 u32 u_threshold; /* upper threshold */
689 u32 l_threshold; /* lower threshold */
694 * The information about the virtual service offered to the net
695 * and the forwarding entries
697 struct ip_vs_service {
698 struct list_head s_list; /* for normal service table */
699 struct list_head f_list; /* for fwmark-based service table */
700 atomic_t refcnt; /* reference counter */
701 atomic_t usecnt; /* use counter */
703 u16 af; /* address family */
704 __u16 protocol; /* which protocol (TCP/UDP) */
705 union nf_inet_addr addr; /* IP address for virtual service */
706 __be16 port; /* port number for the service */
707 __u32 fwmark; /* firewall mark of the service */
708 unsigned int flags; /* service status flags */
709 unsigned int timeout; /* persistent timeout in ticks */
710 __be32 netmask; /* grouping granularity */
713 struct list_head destinations; /* real server d-linked list */
714 __u32 num_dests; /* number of servers */
715 struct ip_vs_stats stats; /* statistics for the service */
716 struct ip_vs_app *inc; /* bind conns to this app inc */
719 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
720 rwlock_t sched_lock; /* lock sched_data */
721 void *sched_data; /* scheduler application data */
723 /* alternate persistence engine */
727 /* Information for cached dst */
728 struct ip_vs_dest_dst {
729 struct dst_entry *dst_cache; /* destination cache entry */
731 union nf_inet_addr dst_saddr;
732 struct rcu_head rcu_head;
736 * The real server destination forwarding entry
737 * with ip address, port number, and so on.
740 struct list_head n_list; /* for the dests in the service */
741 struct list_head d_list; /* for table with all the dests */
743 u16 af; /* address family */
744 __be16 port; /* port number of the server */
745 union nf_inet_addr addr; /* IP address of the server */
746 volatile unsigned int flags; /* dest status flags */
747 atomic_t conn_flags; /* flags to copy to conn */
748 atomic_t weight; /* server weight */
750 atomic_t refcnt; /* reference counter */
751 struct ip_vs_stats stats; /* statistics */
753 /* connection counters and thresholds */
754 atomic_t activeconns; /* active connections */
755 atomic_t inactconns; /* inactive connections */
756 atomic_t persistconns; /* persistent connections */
757 __u32 u_threshold; /* upper threshold */
758 __u32 l_threshold; /* lower threshold */
760 /* for destination cache */
761 spinlock_t dst_lock; /* lock of dst_cache */
762 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
764 /* for virtual service */
765 struct ip_vs_service *svc; /* service it belongs to */
766 __u16 protocol; /* which protocol (TCP/UDP) */
767 __be16 vport; /* virtual port number */
768 union nf_inet_addr vaddr; /* virtual IP address */
769 __u32 vfwmark; /* firewall mark of service */
774 * The scheduler object
776 struct ip_vs_scheduler {
777 struct list_head n_list; /* d-linked list head */
778 char *name; /* scheduler name */
779 atomic_t refcnt; /* reference counter */
780 struct module *module; /* THIS_MODULE/NULL */
782 /* scheduler initializing service */
783 int (*init_service)(struct ip_vs_service *svc);
784 /* scheduling service finish */
785 int (*done_service)(struct ip_vs_service *svc);
786 /* scheduler updating service */
787 int (*update_service)(struct ip_vs_service *svc);
789 /* selecting a server from the given service */
790 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
791 const struct sk_buff *skb);
794 /* The persistence engine object */
796 struct list_head n_list; /* d-linked list head */
797 char *name; /* scheduler name */
798 atomic_t refcnt; /* reference counter */
799 struct module *module; /* THIS_MODULE/NULL */
801 /* get the connection template, if any */
802 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
803 bool (*ct_match)(const struct ip_vs_conn_param *p,
804 struct ip_vs_conn *ct);
805 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
807 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
811 * The application module object (a.k.a. app incarnation)
814 struct list_head a_list; /* member in app list */
815 int type; /* IP_VS_APP_TYPE_xxx */
816 char *name; /* application module name */
818 struct module *module; /* THIS_MODULE/NULL */
819 struct list_head incs_list; /* list of incarnations */
821 /* members for application incarnations */
822 struct list_head p_list; /* member in proto app list */
823 struct ip_vs_app *app; /* its real application */
824 __be16 port; /* port number in net order */
825 atomic_t usecnt; /* usage counter */
826 struct rcu_head rcu_head;
829 * output hook: Process packet in inout direction, diff set for TCP.
830 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
831 * 2=Mangled but checksum was not updated
833 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
834 struct sk_buff *, int *diff);
837 * input hook: Process packet in outin direction, diff set for TCP.
838 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
839 * 2=Mangled but checksum was not updated
841 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
842 struct sk_buff *, int *diff);
844 /* ip_vs_app initializer */
845 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
847 /* ip_vs_app finish */
848 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
852 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
853 struct ip_vs_protocol *);
855 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
861 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
862 int *verdict, struct ip_vs_conn **cpp);
865 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
866 const struct iphdr *iph, int inverse);
869 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
870 const struct iphdr *iph, int inverse);
872 int (*state_transition)(struct ip_vs_conn *cp, int direction,
873 const struct sk_buff *skb,
874 struct ip_vs_app *app);
876 void (*timeout_change)(struct ip_vs_app *app, int flags);
879 struct ipvs_master_sync_state {
880 struct list_head sync_queue;
881 struct ip_vs_sync_buff *sync_buff;
883 unsigned int sync_queue_delay;
884 struct task_struct *master_thread;
885 struct delayed_work master_wakeup_work;
886 struct netns_ipvs *ipvs;
889 /* IPVS in network namespace */
891 int gen; /* Generation */
892 int enable; /* enable like nf_hooks do */
894 * Hash table: for real service lookups
896 #define IP_VS_RTAB_BITS 4
897 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
898 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
900 struct list_head rs_table[IP_VS_RTAB_SIZE];
902 struct list_head app_list;
904 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
905 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
906 /* ip_vs_proto_tcp */
907 #ifdef CONFIG_IP_VS_PROTO_TCP
908 #define TCP_APP_TAB_BITS 4
909 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
910 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
911 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
913 /* ip_vs_proto_udp */
914 #ifdef CONFIG_IP_VS_PROTO_UDP
915 #define UDP_APP_TAB_BITS 4
916 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
917 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
918 struct list_head udp_apps[UDP_APP_TAB_SIZE];
920 /* ip_vs_proto_sctp */
921 #ifdef CONFIG_IP_VS_PROTO_SCTP
922 #define SCTP_APP_TAB_BITS 4
923 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
924 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
925 /* Hash table for SCTP application incarnations */
926 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
929 atomic_t conn_count; /* connection counter */
932 struct ip_vs_stats tot_stats; /* Statistics & est. */
934 int num_services; /* no of virtual services */
936 rwlock_t rs_lock; /* real services table */
937 /* Trash for destinations */
938 struct list_head dest_trash;
939 /* Service counters */
940 atomic_t ftpsvc_counter;
941 atomic_t nullsvc_counter;
944 /* 1/rate drop and drop-entry variables */
945 struct delayed_work defense_work; /* Work handler */
950 spinlock_t dropentry_lock; /* drop entry handling */
951 spinlock_t droppacket_lock; /* drop packet handling */
952 spinlock_t securetcp_lock; /* state and timeout tables */
955 struct ctl_table_header *sysctl_hdr;
956 struct ctl_table *sysctl_tbl;
959 /* sysctl variables */
960 int sysctl_amemthresh;
961 int sysctl_am_droprate;
962 int sysctl_drop_entry;
963 int sysctl_drop_packet;
964 int sysctl_secure_tcp;
965 #ifdef CONFIG_IP_VS_NFCT
966 int sysctl_conntrack;
968 int sysctl_snat_reroute;
970 int sysctl_sync_ports;
971 int sysctl_sync_qlen_max;
972 int sysctl_sync_sock_size;
973 int sysctl_cache_bypass;
974 int sysctl_expire_nodest_conn;
975 int sysctl_expire_quiescent_template;
976 int sysctl_sync_threshold[2];
977 unsigned int sysctl_sync_refresh_period;
978 int sysctl_sync_retries;
979 int sysctl_nat_icmp_send;
980 int sysctl_pmtu_disc;
981 int sysctl_backup_only;
984 int sysctl_lblc_expiration;
985 struct ctl_table_header *lblc_ctl_header;
986 struct ctl_table *lblc_ctl_table;
988 int sysctl_lblcr_expiration;
989 struct ctl_table_header *lblcr_ctl_header;
990 struct ctl_table *lblcr_ctl_table;
992 struct list_head est_list; /* estimator list */
994 struct timer_list est_timer; /* Estimation timer */
996 spinlock_t sync_lock;
997 struct ipvs_master_sync_state *ms;
998 spinlock_t sync_buff_lock;
999 struct task_struct **backup_threads;
1001 int send_mesg_maxlen;
1002 int recv_mesg_maxlen;
1003 volatile int sync_state;
1004 volatile int master_syncid;
1005 volatile int backup_syncid;
1006 struct mutex sync_mutex;
1007 /* multicast interface name */
1008 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1009 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1010 /* net name space ptr */
1011 struct net *net; /* Needed by timer routines */
1014 #define DEFAULT_SYNC_THRESHOLD 3
1015 #define DEFAULT_SYNC_PERIOD 50
1016 #define DEFAULT_SYNC_VER 1
1017 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1018 #define DEFAULT_SYNC_RETRIES 0
1019 #define IPVS_SYNC_WAKEUP_RATE 8
1020 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1021 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1022 #define IPVS_SYNC_CHECK_PERIOD HZ
1023 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1024 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1026 #ifdef CONFIG_SYSCTL
1028 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1030 return ipvs->sysctl_sync_threshold[0];
1033 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1035 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
1038 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1040 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
1043 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1045 return ipvs->sysctl_sync_retries;
1048 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1050 return ipvs->sysctl_sync_ver;
1053 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1055 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
1058 static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1060 return ipvs->sysctl_sync_qlen_max;
1063 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1065 return ipvs->sysctl_sync_sock_size;
1068 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1070 return ipvs->sysctl_pmtu_disc;
1073 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1075 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1076 ipvs->sysctl_backup_only;
1081 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1083 return DEFAULT_SYNC_THRESHOLD;
1086 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1088 return DEFAULT_SYNC_PERIOD;
1091 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1093 return DEFAULT_SYNC_REFRESH_PERIOD;
1096 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1098 return DEFAULT_SYNC_RETRIES & 3;
1101 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1103 return DEFAULT_SYNC_VER;
1106 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1111 static inline int sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1113 return IPVS_SYNC_QLEN_MAX;
1116 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1121 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1126 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1134 * IPVS core functions
1135 * (from ip_vs_core.c)
1137 extern const char *ip_vs_proto_name(unsigned int proto);
1138 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
1139 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1141 #define IP_VS_APP_TYPE_FTP 1
1144 * ip_vs_conn handling functions
1145 * (from ip_vs_conn.c)
1149 IP_VS_DIR_INPUT = 0,
1151 IP_VS_DIR_INPUT_ONLY,
1155 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1156 const union nf_inet_addr *caddr,
1158 const union nf_inet_addr *vaddr,
1160 struct ip_vs_conn_param *p)
1164 p->protocol = protocol;
1173 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1174 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1176 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1177 const struct ip_vs_iphdr *iph,
1180 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1182 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1183 const struct ip_vs_iphdr *iph,
1186 /* put back the conn without restarting its timer */
1187 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1189 atomic_dec(&cp->refcnt);
1191 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1192 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1194 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1195 const union nf_inet_addr *daddr,
1196 __be16 dport, unsigned int flags,
1197 struct ip_vs_dest *dest, __u32 fwmark);
1198 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1200 extern const char * ip_vs_state_name(__u16 proto, int state);
1202 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1203 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1204 extern void ip_vs_random_dropentry(struct net *net);
1205 extern int ip_vs_conn_init(void);
1206 extern void ip_vs_conn_cleanup(void);
1208 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1210 struct ip_vs_conn *ctl_cp = cp->control;
1212 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1214 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1216 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1222 IP_VS_DBG_BUF(7, "DELeting control for: "
1223 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1224 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1226 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1227 ntohs(ctl_cp->cport));
1230 if (atomic_read(&ctl_cp->n_control) == 0) {
1231 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1233 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1235 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1240 atomic_dec(&ctl_cp->n_control);
1244 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1247 IP_VS_ERR_BUF("request control ADD for already controlled: "
1249 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1251 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1254 ip_vs_control_del(cp);
1257 IP_VS_DBG_BUF(7, "ADDing 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));
1264 cp->control = ctl_cp;
1265 atomic_inc(&ctl_cp->n_control);
1269 * IPVS netns init & cleanup functions
1271 extern int ip_vs_estimator_net_init(struct net *net);
1272 extern int ip_vs_control_net_init(struct net *net);
1273 extern int ip_vs_protocol_net_init(struct net *net);
1274 extern int ip_vs_app_net_init(struct net *net);
1275 extern int ip_vs_conn_net_init(struct net *net);
1276 extern int ip_vs_sync_net_init(struct net *net);
1277 extern void ip_vs_conn_net_cleanup(struct net *net);
1278 extern void ip_vs_app_net_cleanup(struct net *net);
1279 extern void ip_vs_protocol_net_cleanup(struct net *net);
1280 extern void ip_vs_control_net_cleanup(struct net *net);
1281 extern void ip_vs_estimator_net_cleanup(struct net *net);
1282 extern void ip_vs_sync_net_cleanup(struct net *net);
1283 extern void ip_vs_service_net_cleanup(struct net *net);
1286 * IPVS application functions
1287 * (from ip_vs_app.c)
1289 #define IP_VS_APP_MAX_PORTS 8
1290 extern struct ip_vs_app *register_ip_vs_app(struct net *net,
1291 struct ip_vs_app *app);
1292 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1293 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1294 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1295 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1296 __u16 proto, __u16 port);
1297 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1298 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1300 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1301 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1303 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1304 void ip_vs_unbind_pe(struct ip_vs_service *svc);
1305 int register_ip_vs_pe(struct ip_vs_pe *pe);
1306 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1307 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1308 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1311 * Use a #define to avoid all of module.h just for these trivial ops
1313 #define ip_vs_pe_get(pe) \
1314 if (pe && pe->module) \
1315 __module_get(pe->module);
1317 #define ip_vs_pe_put(pe) \
1318 if (pe && pe->module) \
1319 module_put(pe->module);
1322 * IPVS protocol functions (from ip_vs_proto.c)
1324 extern int ip_vs_protocol_init(void);
1325 extern void ip_vs_protocol_cleanup(void);
1326 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1327 extern int *ip_vs_create_timeout_table(int *table, int size);
1329 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1330 const char *name, int to);
1332 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1333 const struct sk_buff *skb,
1334 int offset, const char *msg);
1336 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1337 extern struct ip_vs_protocol ip_vs_protocol_udp;
1338 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1339 extern struct ip_vs_protocol ip_vs_protocol_esp;
1340 extern struct ip_vs_protocol ip_vs_protocol_ah;
1341 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1344 * Registering/unregistering scheduler functions
1345 * (from ip_vs_sched.c)
1347 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1348 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1349 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1350 struct ip_vs_scheduler *scheduler);
1351 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1352 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1353 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1354 extern struct ip_vs_conn *
1355 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1356 struct ip_vs_proto_data *pd, int *ignored,
1357 struct ip_vs_iphdr *iph);
1358 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1359 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1361 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1365 * IPVS control data and functions (from ip_vs_ctl.c)
1367 extern struct ip_vs_stats ip_vs_stats;
1368 extern int sysctl_ip_vs_sync_ver;
1370 extern struct ip_vs_service *
1371 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1372 const union nf_inet_addr *vaddr, __be16 vport);
1374 static inline void ip_vs_service_put(struct ip_vs_service *svc)
1376 atomic_dec(&svc->usecnt);
1379 extern struct ip_vs_dest *
1380 ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
1381 const union nf_inet_addr *daddr, __be16 dport);
1383 extern int ip_vs_use_count_inc(void);
1384 extern void ip_vs_use_count_dec(void);
1385 extern int ip_vs_register_nl_ioctl(void);
1386 extern void ip_vs_unregister_nl_ioctl(void);
1387 extern int ip_vs_control_init(void);
1388 extern void ip_vs_control_cleanup(void);
1389 extern struct ip_vs_dest *
1390 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1391 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1392 __u16 protocol, __u32 fwmark, __u32 flags);
1393 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1397 * IPVS sync daemon data and function prototypes
1398 * (from ip_vs_sync.c)
1400 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1402 extern int stop_sync_thread(struct net *net, int state);
1403 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1407 * IPVS rate estimator prototypes (from ip_vs_est.c)
1409 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1410 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1411 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1412 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1413 struct ip_vs_stats *stats);
1416 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1418 extern int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1419 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1420 extern int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1421 struct ip_vs_protocol *pp,
1422 struct ip_vs_iphdr *iph);
1423 extern int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1424 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1425 extern int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1426 struct ip_vs_protocol *pp,
1427 struct ip_vs_iphdr *iph);
1428 extern int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1429 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1430 extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1431 struct ip_vs_protocol *pp, int offset,
1432 unsigned int hooknum, struct ip_vs_iphdr *iph);
1433 extern void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1435 #ifdef CONFIG_IP_VS_IPV6
1436 extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1437 struct ip_vs_protocol *pp,
1438 struct ip_vs_iphdr *iph);
1439 extern int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1440 struct ip_vs_protocol *pp,
1441 struct ip_vs_iphdr *iph);
1442 extern int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1443 struct ip_vs_protocol *pp,
1444 struct ip_vs_iphdr *iph);
1445 extern int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1446 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1447 extern int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1448 struct ip_vs_protocol *pp, int offset,
1449 unsigned int hooknum, struct ip_vs_iphdr *iph);
1452 #ifdef CONFIG_SYSCTL
1454 * This is a simple mechanism to ignore packets when
1455 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1456 * we start to drop 1/rate of the packets
1459 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1461 if (!ipvs->drop_rate)
1463 if (--ipvs->drop_counter > 0)
1465 ipvs->drop_counter = ipvs->drop_rate;
1469 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1473 * ip_vs_fwd_tag returns the forwarding tag of the connection
1475 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1477 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1481 switch (IP_VS_FWD_METHOD(cp)) {
1482 case IP_VS_CONN_F_MASQ:
1484 case IP_VS_CONN_F_LOCALNODE:
1486 case IP_VS_CONN_F_TUNNEL:
1488 case IP_VS_CONN_F_DROUTE:
1490 case IP_VS_CONN_F_BYPASS:
1498 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1499 struct ip_vs_conn *cp, int dir);
1501 #ifdef CONFIG_IP_VS_IPV6
1502 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1503 struct ip_vs_conn *cp, int dir);
1506 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1508 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1510 __be32 diff[2] = { ~old, new };
1512 return csum_partial(diff, sizeof(diff), oldsum);
1515 #ifdef CONFIG_IP_VS_IPV6
1516 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1519 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1520 new[3], new[2], new[1], new[0] };
1522 return csum_partial(diff, sizeof(diff), oldsum);
1526 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1528 __be16 diff[2] = { ~old, new };
1530 return csum_partial(diff, sizeof(diff), oldsum);
1534 * Forget current conntrack (unconfirmed) and attach notrack entry
1536 static inline void ip_vs_notrack(struct sk_buff *skb)
1538 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1539 enum ip_conntrack_info ctinfo;
1540 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1542 if (!ct || !nf_ct_is_untracked(ct)) {
1543 nf_conntrack_put(skb->nfct);
1544 skb->nfct = &nf_ct_untracked_get()->ct_general;
1545 skb->nfctinfo = IP_CT_NEW;
1546 nf_conntrack_get(skb->nfct);
1551 #ifdef CONFIG_IP_VS_NFCT
1553 * Netfilter connection tracking
1554 * (from ip_vs_nfct.c)
1556 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1558 #ifdef CONFIG_SYSCTL
1559 return ipvs->sysctl_conntrack;
1565 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1567 extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1568 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1569 struct ip_vs_conn *cp, u_int8_t proto,
1570 const __be16 port, int from_rs);
1571 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1575 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1580 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1581 struct ip_vs_conn *cp, int outin)
1585 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1590 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1593 /* CONFIG_IP_VS_NFCT */
1596 static inline unsigned int
1597 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1600 * We think the overhead of processing active connections is 256
1601 * times higher than that of inactive connections in average. (This
1602 * 256 times might not be accurate, we will change it later) We
1603 * use the following formula to estimate the overhead now:
1604 * dest->activeconns*256 + dest->inactconns
1606 return (atomic_read(&dest->activeconns) << 8) +
1607 atomic_read(&dest->inactconns);
1610 #endif /* _NET_IP_VS_H */