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 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
201 const union nf_inet_addr *src)
203 #ifdef CONFIG_IP_VS_IPV6
211 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
212 const union nf_inet_addr *src)
214 #ifdef CONFIG_IP_VS_IPV6
215 if (af == AF_INET6) {
226 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
227 const union nf_inet_addr *b)
229 #ifdef CONFIG_IP_VS_IPV6
231 return ipv6_addr_equal(&a->in6, &b->in6);
233 return a->ip == b->ip;
236 #ifdef CONFIG_IP_VS_DEBUG
237 #include <linux/net.h>
239 extern int ip_vs_get_debug_level(void);
241 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
242 const union nf_inet_addr *addr,
246 #ifdef CONFIG_IP_VS_IPV6
248 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
252 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
256 BUG_ON(*idx > buf_len + 1);
257 return &buf[*idx - len];
260 #define IP_VS_DBG_BUF(level, msg, ...) \
262 char ip_vs_dbg_buf[160]; \
263 int ip_vs_dbg_idx = 0; \
264 if (level <= ip_vs_get_debug_level()) \
265 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
267 #define IP_VS_ERR_BUF(msg...) \
269 char ip_vs_dbg_buf[160]; \
270 int ip_vs_dbg_idx = 0; \
274 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
275 #define IP_VS_DBG_ADDR(af, addr) \
276 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
277 sizeof(ip_vs_dbg_buf), addr, \
280 #define IP_VS_DBG(level, msg, ...) \
282 if (level <= ip_vs_get_debug_level()) \
283 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
285 #define IP_VS_DBG_RL(msg, ...) \
287 if (net_ratelimit()) \
288 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
290 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
292 if (level <= ip_vs_get_debug_level()) \
293 pp->debug_packet(af, pp, skb, ofs, msg); \
295 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
297 if (level <= ip_vs_get_debug_level() && \
299 pp->debug_packet(af, pp, skb, ofs, msg); \
301 #else /* NO DEBUGGING at ALL */
302 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
303 #define IP_VS_ERR_BUF(msg...) do {} while (0)
304 #define IP_VS_DBG(level, msg...) do {} while (0)
305 #define IP_VS_DBG_RL(msg...) do {} while (0)
306 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
307 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
310 #define IP_VS_BUG() BUG()
311 #define IP_VS_ERR_RL(msg, ...) \
313 if (net_ratelimit()) \
314 pr_err(msg, ##__VA_ARGS__); \
317 #ifdef CONFIG_IP_VS_DEBUG
318 #define EnterFunction(level) \
320 if (level <= ip_vs_get_debug_level()) \
322 pr_fmt("Enter: %s, %s line %i\n"), \
323 __func__, __FILE__, __LINE__); \
325 #define LeaveFunction(level) \
327 if (level <= ip_vs_get_debug_level()) \
329 pr_fmt("Leave: %s, %s line %i\n"), \
330 __func__, __FILE__, __LINE__); \
333 #define EnterFunction(level) do {} while (0)
334 #define LeaveFunction(level) do {} while (0)
339 * The port number of FTP service (in network order).
341 #define FTPPORT cpu_to_be16(21)
342 #define FTPDATA cpu_to_be16(20)
348 IP_VS_TCP_S_NONE = 0,
349 IP_VS_TCP_S_ESTABLISHED,
350 IP_VS_TCP_S_SYN_SENT,
351 IP_VS_TCP_S_SYN_RECV,
352 IP_VS_TCP_S_FIN_WAIT,
353 IP_VS_TCP_S_TIME_WAIT,
355 IP_VS_TCP_S_CLOSE_WAIT,
356 IP_VS_TCP_S_LAST_ACK,
381 enum ip_vs_sctp_states {
385 IP_VS_SCTP_S_COOKIE_SENT,
386 IP_VS_SCTP_S_COOKIE_REPLIED,
387 IP_VS_SCTP_S_COOKIE_WAIT,
389 IP_VS_SCTP_S_COOKIE_ECHOED,
390 IP_VS_SCTP_S_ESTABLISHED,
391 IP_VS_SCTP_S_SHUTDOWN_SENT,
392 IP_VS_SCTP_S_SHUTDOWN_RECEIVED,
393 IP_VS_SCTP_S_SHUTDOWN_ACK_SENT,
394 IP_VS_SCTP_S_REJECTED,
400 * Delta sequence info structure
401 * Each ip_vs_conn has 2 (output AND input seq. changes).
402 * Only used in the VS/NAT.
405 __u32 init_seq; /* Add delta from this seq */
406 __u32 delta; /* Delta in sequence numbers */
407 __u32 previous_delta; /* Delta in sequence numbers
408 before last resized pkt */
414 struct ip_vs_counters {
415 __u32 conns; /* connections scheduled */
416 __u32 inpkts; /* incoming packets */
417 __u32 outpkts; /* outgoing packets */
418 __u64 inbytes; /* incoming bytes */
419 __u64 outbytes; /* outgoing bytes */
424 struct ip_vs_cpu_stats {
425 struct ip_vs_counters ustats;
426 struct u64_stats_sync syncp;
430 * IPVS statistics objects
432 struct ip_vs_estimator {
433 struct list_head list;
449 struct ip_vs_stats_user ustats; /* statistics */
450 struct ip_vs_estimator est; /* estimator */
451 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
452 spinlock_t lock; /* spin lock */
453 struct ip_vs_stats_user ustats0; /* reset values */
461 struct ip_vs_proto_data;
463 struct ip_vs_protocol {
464 struct ip_vs_protocol *next;
470 void (*init)(struct ip_vs_protocol *pp);
472 void (*exit)(struct ip_vs_protocol *pp);
474 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
476 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
478 int (*conn_schedule)(int af, struct sk_buff *skb,
479 struct ip_vs_proto_data *pd,
480 int *verdict, struct ip_vs_conn **cpp,
481 struct ip_vs_iphdr *iph);
484 (*conn_in_get)(int af,
485 const struct sk_buff *skb,
486 const struct ip_vs_iphdr *iph,
490 (*conn_out_get)(int af,
491 const struct sk_buff *skb,
492 const struct ip_vs_iphdr *iph,
495 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
496 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
498 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
499 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
501 int (*csum_check)(int af, struct sk_buff *skb,
502 struct ip_vs_protocol *pp);
504 const char *(*state_name)(int state);
506 void (*state_transition)(struct ip_vs_conn *cp, int direction,
507 const struct sk_buff *skb,
508 struct ip_vs_proto_data *pd);
510 int (*register_app)(struct net *net, struct ip_vs_app *inc);
512 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
514 int (*app_conn_bind)(struct ip_vs_conn *cp);
516 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
517 const struct sk_buff *skb,
521 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
525 * protocol data per netns
527 struct ip_vs_proto_data {
528 struct ip_vs_proto_data *next;
529 struct ip_vs_protocol *pp;
530 int *timeout_table; /* protocol timeout table */
531 atomic_t appcnt; /* counter of proto app incs. */
532 struct tcp_states_t *tcp_state_table;
535 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
536 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
537 unsigned short proto);
539 struct ip_vs_conn_param {
541 const union nf_inet_addr *caddr;
542 const union nf_inet_addr *vaddr;
548 const struct ip_vs_pe *pe;
554 * IP_VS structure allocated for each dynamically scheduled connection
557 struct hlist_node c_list; /* hashed list heads */
558 /* Protocol, addresses and port numbers */
562 u16 af; /* address family */
563 union nf_inet_addr caddr; /* client address */
564 union nf_inet_addr vaddr; /* virtual address */
565 union nf_inet_addr daddr; /* destination address */
566 volatile __u32 flags; /* status flags */
567 __u16 protocol; /* Which protocol (TCP/UDP) */
569 struct net *net; /* Name space */
572 /* counter and timer */
573 atomic_t refcnt; /* reference count */
574 struct timer_list timer; /* Expiration timer */
575 volatile unsigned long timeout; /* timeout */
577 /* Flags and state transition */
578 spinlock_t lock; /* lock for state transition */
579 volatile __u16 state; /* state info */
580 volatile __u16 old_state; /* old state, to be used for
581 * state transition triggerd
584 __u32 fwmark; /* Fire wall mark from skb */
585 unsigned long sync_endtime; /* jiffies + sent_retries */
587 /* Control members */
588 struct ip_vs_conn *control; /* Master control connection */
589 atomic_t n_control; /* Number of controlled ones */
590 struct ip_vs_dest *dest; /* real server */
591 atomic_t in_pkts; /* incoming packet counter */
593 /* packet transmitter for different forwarding methods. If it
594 mangles the packet, it must return NF_DROP or better NF_STOLEN,
595 otherwise this must be changed to a sk_buff **.
596 NF_ACCEPT can be returned when destination is local.
598 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
599 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
601 /* Note: we can group the following members into a structure,
602 in order to save more space, and the following members are
603 only used in VS/NAT anyway */
604 struct ip_vs_app *app; /* bound ip_vs_app object */
605 void *app_data; /* Application private data */
606 struct ip_vs_seq in_seq; /* incoming seq. struct */
607 struct ip_vs_seq out_seq; /* outgoing seq. struct */
609 const struct ip_vs_pe *pe;
613 struct rcu_head rcu_head;
617 * To save some memory in conn table when name space is disabled.
619 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
627 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
634 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
638 return cp->net == net;
645 * Extended internal versions of struct ip_vs_service_user and
646 * ip_vs_dest_user for IPv6 support.
648 * We need these to conveniently pass around service and destination
649 * options, but unfortunately, we also need to keep the old definitions to
650 * maintain userspace backwards compatibility for the setsockopt interface.
652 struct ip_vs_service_user_kern {
653 /* virtual service addresses */
656 union nf_inet_addr addr; /* virtual ip address */
658 u32 fwmark; /* firwall mark of service */
660 /* virtual service options */
663 unsigned int flags; /* virtual service flags */
664 unsigned int timeout; /* persistent timeout in sec */
665 __be32 netmask; /* persistent netmask or plen */
669 struct ip_vs_dest_user_kern {
670 /* destination server address */
671 union nf_inet_addr addr;
674 /* real server options */
675 unsigned int conn_flags; /* connection flags */
676 int weight; /* destination weight */
678 /* thresholds for active connections */
679 u32 u_threshold; /* upper threshold */
680 u32 l_threshold; /* lower threshold */
685 * The information about the virtual service offered to the net
686 * and the forwarding entries
688 struct ip_vs_service {
689 struct hlist_node s_list; /* for normal service table */
690 struct hlist_node f_list; /* for fwmark-based service table */
691 atomic_t refcnt; /* reference counter */
693 u16 af; /* address family */
694 __u16 protocol; /* which protocol (TCP/UDP) */
695 union nf_inet_addr addr; /* IP address for virtual service */
696 __be16 port; /* port number for the service */
697 __u32 fwmark; /* firewall mark of the service */
698 unsigned int flags; /* service status flags */
699 unsigned int timeout; /* persistent timeout in ticks */
700 __be32 netmask; /* grouping granularity, mask/plen */
703 struct list_head destinations; /* real server d-linked list */
704 __u32 num_dests; /* number of servers */
705 struct ip_vs_stats stats; /* statistics for the service */
708 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
709 spinlock_t sched_lock; /* lock sched_data */
710 void *sched_data; /* scheduler application data */
712 /* alternate persistence engine */
713 struct ip_vs_pe __rcu *pe;
715 struct rcu_head rcu_head;
718 /* Information for cached dst */
719 struct ip_vs_dest_dst {
720 struct dst_entry *dst_cache; /* destination cache entry */
722 union nf_inet_addr dst_saddr;
723 struct rcu_head rcu_head;
727 * The real server destination forwarding entry
728 * with ip address, port number, and so on.
731 struct list_head n_list; /* for the dests in the service */
732 struct hlist_node d_list; /* for table with all the dests */
734 u16 af; /* address family */
735 __be16 port; /* port number of the server */
736 union nf_inet_addr addr; /* IP address of the server */
737 volatile unsigned int flags; /* dest status flags */
738 atomic_t conn_flags; /* flags to copy to conn */
739 atomic_t weight; /* server weight */
741 atomic_t refcnt; /* reference counter */
742 struct ip_vs_stats stats; /* statistics */
743 unsigned long idle_start; /* start time, jiffies */
745 /* connection counters and thresholds */
746 atomic_t activeconns; /* active connections */
747 atomic_t inactconns; /* inactive connections */
748 atomic_t persistconns; /* persistent connections */
749 __u32 u_threshold; /* upper threshold */
750 __u32 l_threshold; /* lower threshold */
752 /* for destination cache */
753 spinlock_t dst_lock; /* lock of dst_cache */
754 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
756 /* for virtual service */
757 struct ip_vs_service __rcu *svc; /* service it belongs to */
758 __u16 protocol; /* which protocol (TCP/UDP) */
759 __be16 vport; /* virtual port number */
760 union nf_inet_addr vaddr; /* virtual IP address */
761 __u32 vfwmark; /* firewall mark of service */
763 struct list_head t_list; /* in dest_trash */
764 unsigned int in_rs_table:1; /* we are in rs_table */
769 * The scheduler object
771 struct ip_vs_scheduler {
772 struct list_head n_list; /* d-linked list head */
773 char *name; /* scheduler name */
774 atomic_t refcnt; /* reference counter */
775 struct module *module; /* THIS_MODULE/NULL */
777 /* scheduler initializing service */
778 int (*init_service)(struct ip_vs_service *svc);
779 /* scheduling service finish */
780 void (*done_service)(struct ip_vs_service *svc);
782 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
783 /* dest is unlinked */
784 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
785 /* dest is updated */
786 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
788 /* selecting a server from the given service */
789 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
790 const struct sk_buff *skb,
791 struct ip_vs_iphdr *iph);
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;
882 unsigned long sync_queue_len;
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 /* How much time to keep dests in trash */
890 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
892 /* IPVS in network namespace */
894 int gen; /* Generation */
895 int enable; /* enable like nf_hooks do */
897 * Hash table: for real service lookups
899 #define IP_VS_RTAB_BITS 4
900 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
901 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
903 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
905 struct list_head app_list;
907 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
908 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
909 /* ip_vs_proto_tcp */
910 #ifdef CONFIG_IP_VS_PROTO_TCP
911 #define TCP_APP_TAB_BITS 4
912 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
913 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
914 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
916 /* ip_vs_proto_udp */
917 #ifdef CONFIG_IP_VS_PROTO_UDP
918 #define UDP_APP_TAB_BITS 4
919 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
920 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
921 struct list_head udp_apps[UDP_APP_TAB_SIZE];
923 /* ip_vs_proto_sctp */
924 #ifdef CONFIG_IP_VS_PROTO_SCTP
925 #define SCTP_APP_TAB_BITS 4
926 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
927 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
928 /* Hash table for SCTP application incarnations */
929 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
932 atomic_t conn_count; /* connection counter */
935 struct ip_vs_stats tot_stats; /* Statistics & est. */
937 int num_services; /* no of virtual services */
939 /* Trash for destinations */
940 struct list_head dest_trash;
941 spinlock_t dest_trash_lock;
942 struct timer_list dest_trash_timer; /* expiration timer */
943 /* Service counters */
944 atomic_t ftpsvc_counter;
945 atomic_t nullsvc_counter;
948 /* 1/rate drop and drop-entry variables */
949 struct delayed_work defense_work; /* Work handler */
954 spinlock_t dropentry_lock; /* drop entry handling */
955 spinlock_t droppacket_lock; /* drop packet handling */
956 spinlock_t securetcp_lock; /* state and timeout tables */
959 struct ctl_table_header *sysctl_hdr;
960 struct ctl_table *sysctl_tbl;
963 /* sysctl variables */
964 int sysctl_amemthresh;
965 int sysctl_am_droprate;
966 int sysctl_drop_entry;
967 int sysctl_drop_packet;
968 int sysctl_secure_tcp;
969 #ifdef CONFIG_IP_VS_NFCT
970 int sysctl_conntrack;
972 int sysctl_snat_reroute;
974 int sysctl_sync_ports;
975 int sysctl_sync_persist_mode;
976 unsigned long sysctl_sync_qlen_max;
977 int sysctl_sync_sock_size;
978 int sysctl_cache_bypass;
979 int sysctl_expire_nodest_conn;
980 int sysctl_sloppy_tcp;
981 int sysctl_sloppy_sctp;
982 int sysctl_expire_quiescent_template;
983 int sysctl_sync_threshold[2];
984 unsigned int sysctl_sync_refresh_period;
985 int sysctl_sync_retries;
986 int sysctl_nat_icmp_send;
987 int sysctl_pmtu_disc;
988 int sysctl_backup_only;
991 int sysctl_lblc_expiration;
992 struct ctl_table_header *lblc_ctl_header;
993 struct ctl_table *lblc_ctl_table;
995 int sysctl_lblcr_expiration;
996 struct ctl_table_header *lblcr_ctl_header;
997 struct ctl_table *lblcr_ctl_table;
999 struct list_head est_list; /* estimator list */
1000 spinlock_t est_lock;
1001 struct timer_list est_timer; /* Estimation timer */
1003 spinlock_t sync_lock;
1004 struct ipvs_master_sync_state *ms;
1005 spinlock_t sync_buff_lock;
1006 struct task_struct **backup_threads;
1008 int send_mesg_maxlen;
1009 int recv_mesg_maxlen;
1010 volatile int sync_state;
1011 volatile int master_syncid;
1012 volatile int backup_syncid;
1013 struct mutex sync_mutex;
1014 /* multicast interface name */
1015 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1016 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
1017 /* net name space ptr */
1018 struct net *net; /* Needed by timer routines */
1021 #define DEFAULT_SYNC_THRESHOLD 3
1022 #define DEFAULT_SYNC_PERIOD 50
1023 #define DEFAULT_SYNC_VER 1
1024 #define DEFAULT_SLOPPY_TCP 0
1025 #define DEFAULT_SLOPPY_SCTP 0
1026 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1027 #define DEFAULT_SYNC_RETRIES 0
1028 #define IPVS_SYNC_WAKEUP_RATE 8
1029 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1030 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1031 #define IPVS_SYNC_CHECK_PERIOD HZ
1032 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1033 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1035 #ifdef CONFIG_SYSCTL
1037 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1039 return ipvs->sysctl_sync_threshold[0];
1042 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1044 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
1047 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1049 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
1052 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1054 return ipvs->sysctl_sync_retries;
1057 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1059 return ipvs->sysctl_sync_ver;
1062 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1064 return ipvs->sysctl_sloppy_tcp;
1067 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1069 return ipvs->sysctl_sloppy_sctp;
1072 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1074 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
1077 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1079 return ipvs->sysctl_sync_persist_mode;
1082 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1084 return ipvs->sysctl_sync_qlen_max;
1087 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1089 return ipvs->sysctl_sync_sock_size;
1092 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1094 return ipvs->sysctl_pmtu_disc;
1097 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1099 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1100 ipvs->sysctl_backup_only;
1105 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1107 return DEFAULT_SYNC_THRESHOLD;
1110 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1112 return DEFAULT_SYNC_PERIOD;
1115 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1117 return DEFAULT_SYNC_REFRESH_PERIOD;
1120 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1122 return DEFAULT_SYNC_RETRIES & 3;
1125 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1127 return DEFAULT_SYNC_VER;
1130 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1132 return DEFAULT_SLOPPY_TCP;
1135 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1137 return DEFAULT_SLOPPY_SCTP;
1140 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1145 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1150 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1152 return IPVS_SYNC_QLEN_MAX;
1155 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1160 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1165 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1173 * IPVS core functions
1174 * (from ip_vs_core.c)
1176 extern const char *ip_vs_proto_name(unsigned int proto);
1177 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
1178 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1180 #define IP_VS_APP_TYPE_FTP 1
1183 * ip_vs_conn handling functions
1184 * (from ip_vs_conn.c)
1188 IP_VS_DIR_INPUT = 0,
1190 IP_VS_DIR_INPUT_ONLY,
1194 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1195 const union nf_inet_addr *caddr,
1197 const union nf_inet_addr *vaddr,
1199 struct ip_vs_conn_param *p)
1203 p->protocol = protocol;
1212 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1213 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1215 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1216 const struct ip_vs_iphdr *iph,
1219 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1221 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1222 const struct ip_vs_iphdr *iph,
1225 /* Get reference to gain full access to conn.
1226 * By default, RCU read-side critical sections have access only to
1227 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1229 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1231 return atomic_inc_not_zero(&cp->refcnt);
1234 /* put back the conn without restarting its timer */
1235 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1237 smp_mb__before_atomic_dec();
1238 atomic_dec(&cp->refcnt);
1240 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1241 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1243 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1244 const union nf_inet_addr *daddr,
1245 __be16 dport, unsigned int flags,
1246 struct ip_vs_dest *dest, __u32 fwmark);
1247 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1249 extern const char * ip_vs_state_name(__u16 proto, int state);
1251 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1252 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1253 extern void ip_vs_random_dropentry(struct net *net);
1254 extern int ip_vs_conn_init(void);
1255 extern void ip_vs_conn_cleanup(void);
1257 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1259 struct ip_vs_conn *ctl_cp = cp->control;
1261 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1263 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1265 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1271 IP_VS_DBG_BUF(7, "DELeting control for: "
1272 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1273 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1275 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1276 ntohs(ctl_cp->cport));
1279 if (atomic_read(&ctl_cp->n_control) == 0) {
1280 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1282 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1284 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1289 atomic_dec(&ctl_cp->n_control);
1293 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1296 IP_VS_ERR_BUF("request control ADD for already controlled: "
1298 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1300 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1303 ip_vs_control_del(cp);
1306 IP_VS_DBG_BUF(7, "ADDing control for: "
1307 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1308 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1310 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1311 ntohs(ctl_cp->cport));
1313 cp->control = ctl_cp;
1314 atomic_inc(&ctl_cp->n_control);
1318 * IPVS netns init & cleanup functions
1320 extern int ip_vs_estimator_net_init(struct net *net);
1321 extern int ip_vs_control_net_init(struct net *net);
1322 extern int ip_vs_protocol_net_init(struct net *net);
1323 extern int ip_vs_app_net_init(struct net *net);
1324 extern int ip_vs_conn_net_init(struct net *net);
1325 extern int ip_vs_sync_net_init(struct net *net);
1326 extern void ip_vs_conn_net_cleanup(struct net *net);
1327 extern void ip_vs_app_net_cleanup(struct net *net);
1328 extern void ip_vs_protocol_net_cleanup(struct net *net);
1329 extern void ip_vs_control_net_cleanup(struct net *net);
1330 extern void ip_vs_estimator_net_cleanup(struct net *net);
1331 extern void ip_vs_sync_net_cleanup(struct net *net);
1332 extern void ip_vs_service_net_cleanup(struct net *net);
1335 * IPVS application functions
1336 * (from ip_vs_app.c)
1338 #define IP_VS_APP_MAX_PORTS 8
1339 extern struct ip_vs_app *register_ip_vs_app(struct net *net,
1340 struct ip_vs_app *app);
1341 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1342 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1343 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1344 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1345 __u16 proto, __u16 port);
1346 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1347 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1349 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1350 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1352 int register_ip_vs_pe(struct ip_vs_pe *pe);
1353 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1354 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1355 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1358 * Use a #define to avoid all of module.h just for these trivial ops
1360 #define ip_vs_pe_get(pe) \
1361 if (pe && pe->module) \
1362 __module_get(pe->module);
1364 #define ip_vs_pe_put(pe) \
1365 if (pe && pe->module) \
1366 module_put(pe->module);
1369 * IPVS protocol functions (from ip_vs_proto.c)
1371 extern int ip_vs_protocol_init(void);
1372 extern void ip_vs_protocol_cleanup(void);
1373 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1374 extern int *ip_vs_create_timeout_table(int *table, int size);
1376 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1377 const char *name, int to);
1379 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1380 const struct sk_buff *skb,
1381 int offset, const char *msg);
1383 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1384 extern struct ip_vs_protocol ip_vs_protocol_udp;
1385 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1386 extern struct ip_vs_protocol ip_vs_protocol_esp;
1387 extern struct ip_vs_protocol ip_vs_protocol_ah;
1388 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1391 * Registering/unregistering scheduler functions
1392 * (from ip_vs_sched.c)
1394 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1395 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1396 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1397 struct ip_vs_scheduler *scheduler);
1398 extern void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1399 struct ip_vs_scheduler *sched);
1400 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1401 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1402 extern struct ip_vs_conn *
1403 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1404 struct ip_vs_proto_data *pd, int *ignored,
1405 struct ip_vs_iphdr *iph);
1406 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1407 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1409 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1413 * IPVS control data and functions (from ip_vs_ctl.c)
1415 extern struct ip_vs_stats ip_vs_stats;
1416 extern int sysctl_ip_vs_sync_ver;
1418 extern struct ip_vs_service *
1419 ip_vs_service_find(struct net *net, int af, __u32 fwmark, __u16 protocol,
1420 const union nf_inet_addr *vaddr, __be16 vport);
1423 ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
1424 const union nf_inet_addr *daddr, __be16 dport);
1426 extern int ip_vs_use_count_inc(void);
1427 extern void ip_vs_use_count_dec(void);
1428 extern int ip_vs_register_nl_ioctl(void);
1429 extern void ip_vs_unregister_nl_ioctl(void);
1430 extern int ip_vs_control_init(void);
1431 extern void ip_vs_control_cleanup(void);
1432 extern struct ip_vs_dest *
1433 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1434 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1435 __u16 protocol, __u32 fwmark, __u32 flags);
1436 extern void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1438 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1440 atomic_inc(&dest->refcnt);
1443 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1445 smp_mb__before_atomic_dec();
1446 atomic_dec(&dest->refcnt);
1450 * IPVS sync daemon data and function prototypes
1451 * (from ip_vs_sync.c)
1453 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1455 extern int stop_sync_thread(struct net *net, int state);
1456 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1460 * IPVS rate estimator prototypes (from ip_vs_est.c)
1462 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1463 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1464 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1465 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1466 struct ip_vs_stats *stats);
1469 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1471 extern int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1472 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1473 extern int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1474 struct ip_vs_protocol *pp,
1475 struct ip_vs_iphdr *iph);
1476 extern int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1477 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1478 extern int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1479 struct ip_vs_protocol *pp,
1480 struct ip_vs_iphdr *iph);
1481 extern int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1482 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1483 extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1484 struct ip_vs_protocol *pp, int offset,
1485 unsigned int hooknum, struct ip_vs_iphdr *iph);
1486 extern void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1488 #ifdef CONFIG_IP_VS_IPV6
1489 extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1490 struct ip_vs_protocol *pp,
1491 struct ip_vs_iphdr *iph);
1492 extern int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1493 struct ip_vs_protocol *pp,
1494 struct ip_vs_iphdr *iph);
1495 extern int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1496 struct ip_vs_protocol *pp,
1497 struct ip_vs_iphdr *iph);
1498 extern int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1499 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1500 extern int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1501 struct ip_vs_protocol *pp, int offset,
1502 unsigned int hooknum, struct ip_vs_iphdr *iph);
1505 #ifdef CONFIG_SYSCTL
1507 * This is a simple mechanism to ignore packets when
1508 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1509 * we start to drop 1/rate of the packets
1512 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1514 if (!ipvs->drop_rate)
1516 if (--ipvs->drop_counter > 0)
1518 ipvs->drop_counter = ipvs->drop_rate;
1522 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1526 * ip_vs_fwd_tag returns the forwarding tag of the connection
1528 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1530 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1534 switch (IP_VS_FWD_METHOD(cp)) {
1535 case IP_VS_CONN_F_MASQ:
1537 case IP_VS_CONN_F_LOCALNODE:
1539 case IP_VS_CONN_F_TUNNEL:
1541 case IP_VS_CONN_F_DROUTE:
1543 case IP_VS_CONN_F_BYPASS:
1551 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1552 struct ip_vs_conn *cp, int dir);
1554 #ifdef CONFIG_IP_VS_IPV6
1555 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1556 struct ip_vs_conn *cp, int dir);
1559 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1561 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1563 __be32 diff[2] = { ~old, new };
1565 return csum_partial(diff, sizeof(diff), oldsum);
1568 #ifdef CONFIG_IP_VS_IPV6
1569 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1572 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1573 new[3], new[2], new[1], new[0] };
1575 return csum_partial(diff, sizeof(diff), oldsum);
1579 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1581 __be16 diff[2] = { ~old, new };
1583 return csum_partial(diff, sizeof(diff), oldsum);
1587 * Forget current conntrack (unconfirmed) and attach notrack entry
1589 static inline void ip_vs_notrack(struct sk_buff *skb)
1591 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1592 enum ip_conntrack_info ctinfo;
1593 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1595 if (!ct || !nf_ct_is_untracked(ct)) {
1596 nf_conntrack_put(skb->nfct);
1597 skb->nfct = &nf_ct_untracked_get()->ct_general;
1598 skb->nfctinfo = IP_CT_NEW;
1599 nf_conntrack_get(skb->nfct);
1604 #ifdef CONFIG_IP_VS_NFCT
1606 * Netfilter connection tracking
1607 * (from ip_vs_nfct.c)
1609 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1611 #ifdef CONFIG_SYSCTL
1612 return ipvs->sysctl_conntrack;
1618 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1620 extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1621 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1622 struct ip_vs_conn *cp, u_int8_t proto,
1623 const __be16 port, int from_rs);
1624 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1628 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1633 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1634 struct ip_vs_conn *cp, int outin)
1638 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1643 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1646 /* CONFIG_IP_VS_NFCT */
1650 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1653 * We think the overhead of processing active connections is 256
1654 * times higher than that of inactive connections in average. (This
1655 * 256 times might not be accurate, we will change it later) We
1656 * use the following formula to estimate the overhead now:
1657 * dest->activeconns*256 + dest->inactconns
1659 return (atomic_read(&dest->activeconns) << 8) +
1660 atomic_read(&dest->inactconns);
1663 #endif /* _NET_IP_VS_H */