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 __u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
115 union nf_inet_addr saddr;
116 union nf_inet_addr daddr;
119 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
120 int len, void *buffer,
121 const struct ip_vs_iphdr *ipvsh)
123 return skb_header_pointer(skb, offset, len, buffer);
127 ip_vs_fill_ip4hdr(const void *nh, struct ip_vs_iphdr *iphdr)
129 const struct iphdr *iph = nh;
131 iphdr->len = iph->ihl * 4;
133 iphdr->protocol = iph->protocol;
134 iphdr->saddr.ip = iph->saddr;
135 iphdr->daddr.ip = iph->daddr;
138 /* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
139 * IPv6 requires some extra work, as finding proper header position,
140 * depend on the IPv6 extension headers.
143 ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, struct ip_vs_iphdr *iphdr)
145 #ifdef CONFIG_IP_VS_IPV6
146 if (af == AF_INET6) {
147 const struct ipv6hdr *iph =
148 (struct ipv6hdr *)skb_network_header(skb);
149 iphdr->saddr.in6 = iph->saddr;
150 iphdr->daddr.in6 = iph->daddr;
151 /* ipv6_find_hdr() updates len, flags */
154 iphdr->protocol = ipv6_find_hdr(skb, &iphdr->len, -1,
160 const struct iphdr *iph =
161 (struct iphdr *)skb_network_header(skb);
162 iphdr->len = iph->ihl * 4;
164 iphdr->protocol = iph->protocol;
165 iphdr->saddr.ip = iph->saddr;
166 iphdr->daddr.ip = iph->daddr;
170 /* This function is a faster version of ip_vs_fill_iph_skb().
171 * Where we only populate {s,d}addr (and avoid calling ipv6_find_hdr()).
172 * This is used by the some of the ip_vs_*_schedule() functions.
173 * (Mostly done to avoid ABI breakage of external schedulers)
176 ip_vs_fill_iph_addr_only(int af, const struct sk_buff *skb,
177 struct ip_vs_iphdr *iphdr)
179 #ifdef CONFIG_IP_VS_IPV6
180 if (af == AF_INET6) {
181 const struct ipv6hdr *iph =
182 (struct ipv6hdr *)skb_network_header(skb);
183 iphdr->saddr.in6 = iph->saddr;
184 iphdr->daddr.in6 = iph->daddr;
188 const struct iphdr *iph =
189 (struct iphdr *)skb_network_header(skb);
190 iphdr->saddr.ip = iph->saddr;
191 iphdr->daddr.ip = iph->daddr;
195 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
196 const union nf_inet_addr *src)
198 #ifdef CONFIG_IP_VS_IPV6
206 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
207 const union nf_inet_addr *src)
209 #ifdef CONFIG_IP_VS_IPV6
210 if (af == AF_INET6) {
221 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
222 const union nf_inet_addr *b)
224 #ifdef CONFIG_IP_VS_IPV6
226 return ipv6_addr_equal(&a->in6, &b->in6);
228 return a->ip == b->ip;
231 #ifdef CONFIG_IP_VS_DEBUG
232 #include <linux/net.h>
234 extern int ip_vs_get_debug_level(void);
236 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
237 const union nf_inet_addr *addr,
241 #ifdef CONFIG_IP_VS_IPV6
243 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
247 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
251 BUG_ON(*idx > buf_len + 1);
252 return &buf[*idx - len];
255 #define IP_VS_DBG_BUF(level, msg, ...) \
257 char ip_vs_dbg_buf[160]; \
258 int ip_vs_dbg_idx = 0; \
259 if (level <= ip_vs_get_debug_level()) \
260 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
262 #define IP_VS_ERR_BUF(msg...) \
264 char ip_vs_dbg_buf[160]; \
265 int ip_vs_dbg_idx = 0; \
269 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
270 #define IP_VS_DBG_ADDR(af, addr) \
271 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
272 sizeof(ip_vs_dbg_buf), addr, \
275 #define IP_VS_DBG(level, msg, ...) \
277 if (level <= ip_vs_get_debug_level()) \
278 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
280 #define IP_VS_DBG_RL(msg, ...) \
282 if (net_ratelimit()) \
283 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
285 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
287 if (level <= ip_vs_get_debug_level()) \
288 pp->debug_packet(af, pp, skb, ofs, msg); \
290 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
292 if (level <= ip_vs_get_debug_level() && \
294 pp->debug_packet(af, pp, skb, ofs, msg); \
296 #else /* NO DEBUGGING at ALL */
297 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
298 #define IP_VS_ERR_BUF(msg...) do {} while (0)
299 #define IP_VS_DBG(level, msg...) do {} while (0)
300 #define IP_VS_DBG_RL(msg...) do {} while (0)
301 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
302 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
305 #define IP_VS_BUG() BUG()
306 #define IP_VS_ERR_RL(msg, ...) \
308 if (net_ratelimit()) \
309 pr_err(msg, ##__VA_ARGS__); \
312 #ifdef CONFIG_IP_VS_DEBUG
313 #define EnterFunction(level) \
315 if (level <= ip_vs_get_debug_level()) \
317 pr_fmt("Enter: %s, %s line %i\n"), \
318 __func__, __FILE__, __LINE__); \
320 #define LeaveFunction(level) \
322 if (level <= ip_vs_get_debug_level()) \
324 pr_fmt("Leave: %s, %s line %i\n"), \
325 __func__, __FILE__, __LINE__); \
328 #define EnterFunction(level) do {} while (0)
329 #define LeaveFunction(level) do {} while (0)
334 * The port number of FTP service (in network order).
336 #define FTPPORT cpu_to_be16(21)
337 #define FTPDATA cpu_to_be16(20)
343 IP_VS_TCP_S_NONE = 0,
344 IP_VS_TCP_S_ESTABLISHED,
345 IP_VS_TCP_S_SYN_SENT,
346 IP_VS_TCP_S_SYN_RECV,
347 IP_VS_TCP_S_FIN_WAIT,
348 IP_VS_TCP_S_TIME_WAIT,
350 IP_VS_TCP_S_CLOSE_WAIT,
351 IP_VS_TCP_S_LAST_ACK,
376 enum ip_vs_sctp_states {
378 IP_VS_SCTP_S_INIT_CLI,
379 IP_VS_SCTP_S_INIT_SER,
380 IP_VS_SCTP_S_INIT_ACK_CLI,
381 IP_VS_SCTP_S_INIT_ACK_SER,
382 IP_VS_SCTP_S_ECHO_CLI,
383 IP_VS_SCTP_S_ECHO_SER,
384 IP_VS_SCTP_S_ESTABLISHED,
385 IP_VS_SCTP_S_SHUT_CLI,
386 IP_VS_SCTP_S_SHUT_SER,
387 IP_VS_SCTP_S_SHUT_ACK_CLI,
388 IP_VS_SCTP_S_SHUT_ACK_SER,
394 * Delta sequence info structure
395 * Each ip_vs_conn has 2 (output AND input seq. changes).
396 * Only used in the VS/NAT.
399 __u32 init_seq; /* Add delta from this seq */
400 __u32 delta; /* Delta in sequence numbers */
401 __u32 previous_delta; /* Delta in sequence numbers
402 before last resized pkt */
408 struct ip_vs_counters {
409 __u32 conns; /* connections scheduled */
410 __u32 inpkts; /* incoming packets */
411 __u32 outpkts; /* outgoing packets */
412 __u64 inbytes; /* incoming bytes */
413 __u64 outbytes; /* outgoing bytes */
418 struct ip_vs_cpu_stats {
419 struct ip_vs_counters ustats;
420 struct u64_stats_sync syncp;
424 * IPVS statistics objects
426 struct ip_vs_estimator {
427 struct list_head list;
443 struct ip_vs_stats_user ustats; /* statistics */
444 struct ip_vs_estimator est; /* estimator */
445 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
446 spinlock_t lock; /* spin lock */
447 struct ip_vs_stats_user ustats0; /* reset values */
455 struct ip_vs_proto_data;
457 struct ip_vs_protocol {
458 struct ip_vs_protocol *next;
464 void (*init)(struct ip_vs_protocol *pp);
466 void (*exit)(struct ip_vs_protocol *pp);
468 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
470 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
472 int (*conn_schedule)(int af, struct sk_buff *skb,
473 struct ip_vs_proto_data *pd,
474 int *verdict, struct ip_vs_conn **cpp,
475 struct ip_vs_iphdr *iph);
478 (*conn_in_get)(int af,
479 const struct sk_buff *skb,
480 const struct ip_vs_iphdr *iph,
484 (*conn_out_get)(int af,
485 const struct sk_buff *skb,
486 const struct ip_vs_iphdr *iph,
489 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
490 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
492 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
493 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
495 int (*csum_check)(int af, struct sk_buff *skb,
496 struct ip_vs_protocol *pp);
498 const char *(*state_name)(int state);
500 void (*state_transition)(struct ip_vs_conn *cp, int direction,
501 const struct sk_buff *skb,
502 struct ip_vs_proto_data *pd);
504 int (*register_app)(struct net *net, struct ip_vs_app *inc);
506 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
508 int (*app_conn_bind)(struct ip_vs_conn *cp);
510 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
511 const struct sk_buff *skb,
515 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
519 * protocol data per netns
521 struct ip_vs_proto_data {
522 struct ip_vs_proto_data *next;
523 struct ip_vs_protocol *pp;
524 int *timeout_table; /* protocol timeout table */
525 atomic_t appcnt; /* counter of proto app incs. */
526 struct tcp_states_t *tcp_state_table;
529 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
530 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
531 unsigned short proto);
533 struct ip_vs_conn_param {
535 const union nf_inet_addr *caddr;
536 const union nf_inet_addr *vaddr;
542 const struct ip_vs_pe *pe;
548 * IP_VS structure allocated for each dynamically scheduled connection
551 struct hlist_node c_list; /* hashed list heads */
552 /* Protocol, addresses and port numbers */
556 u16 af; /* address family */
557 union nf_inet_addr caddr; /* client address */
558 union nf_inet_addr vaddr; /* virtual address */
559 union nf_inet_addr daddr; /* destination address */
560 volatile __u32 flags; /* status flags */
561 __u16 protocol; /* Which protocol (TCP/UDP) */
563 struct net *net; /* Name space */
566 /* counter and timer */
567 atomic_t refcnt; /* reference count */
568 struct timer_list timer; /* Expiration timer */
569 volatile unsigned long timeout; /* timeout */
571 /* Flags and state transition */
572 spinlock_t lock; /* lock for state transition */
573 volatile __u16 state; /* state info */
574 volatile __u16 old_state; /* old state, to be used for
575 * state transition triggerd
578 __u32 fwmark; /* Fire wall mark from skb */
579 unsigned long sync_endtime; /* jiffies + sent_retries */
581 /* Control members */
582 struct ip_vs_conn *control; /* Master control connection */
583 atomic_t n_control; /* Number of controlled ones */
584 struct ip_vs_dest *dest; /* real server */
585 atomic_t in_pkts; /* incoming packet counter */
587 /* packet transmitter for different forwarding methods. If it
588 mangles the packet, it must return NF_DROP or better NF_STOLEN,
589 otherwise this must be changed to a sk_buff **.
590 NF_ACCEPT can be returned when destination is local.
592 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
593 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
595 /* Note: we can group the following members into a structure,
596 in order to save more space, and the following members are
597 only used in VS/NAT anyway */
598 struct ip_vs_app *app; /* bound ip_vs_app object */
599 void *app_data; /* Application private data */
600 struct ip_vs_seq in_seq; /* incoming seq. struct */
601 struct ip_vs_seq out_seq; /* outgoing seq. struct */
603 const struct ip_vs_pe *pe;
607 struct rcu_head rcu_head;
611 * To save some memory in conn table when name space is disabled.
613 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
621 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
628 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
632 return cp->net == net;
639 * Extended internal versions of struct ip_vs_service_user and
640 * ip_vs_dest_user for IPv6 support.
642 * We need these to conveniently pass around service and destination
643 * options, but unfortunately, we also need to keep the old definitions to
644 * maintain userspace backwards compatibility for the setsockopt interface.
646 struct ip_vs_service_user_kern {
647 /* virtual service addresses */
650 union nf_inet_addr addr; /* virtual ip address */
652 u32 fwmark; /* firwall mark of service */
654 /* virtual service options */
657 unsigned int flags; /* virtual service flags */
658 unsigned int timeout; /* persistent timeout in sec */
659 __be32 netmask; /* persistent netmask or plen */
663 struct ip_vs_dest_user_kern {
664 /* destination server address */
665 union nf_inet_addr addr;
668 /* real server options */
669 unsigned int conn_flags; /* connection flags */
670 int weight; /* destination weight */
672 /* thresholds for active connections */
673 u32 u_threshold; /* upper threshold */
674 u32 l_threshold; /* lower threshold */
679 * The information about the virtual service offered to the net
680 * and the forwarding entries
682 struct ip_vs_service {
683 struct hlist_node s_list; /* for normal service table */
684 struct hlist_node f_list; /* for fwmark-based service table */
685 atomic_t refcnt; /* reference counter */
687 u16 af; /* address family */
688 __u16 protocol; /* which protocol (TCP/UDP) */
689 union nf_inet_addr addr; /* IP address for virtual service */
690 __be16 port; /* port number for the service */
691 __u32 fwmark; /* firewall mark of the service */
692 unsigned int flags; /* service status flags */
693 unsigned int timeout; /* persistent timeout in ticks */
694 __be32 netmask; /* grouping granularity, mask/plen */
697 struct list_head destinations; /* real server d-linked list */
698 __u32 num_dests; /* number of servers */
699 struct ip_vs_stats stats; /* statistics for the service */
702 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
703 spinlock_t sched_lock; /* lock sched_data */
704 void *sched_data; /* scheduler application data */
706 /* alternate persistence engine */
707 struct ip_vs_pe __rcu *pe;
709 struct rcu_head rcu_head;
712 /* Information for cached dst */
713 struct ip_vs_dest_dst {
714 struct dst_entry *dst_cache; /* destination cache entry */
716 union nf_inet_addr dst_saddr;
717 struct rcu_head rcu_head;
720 /* In grace period after removing */
721 #define IP_VS_DEST_STATE_REMOVING 0x01
723 * The real server destination forwarding entry
724 * with ip address, port number, and so on.
727 struct list_head n_list; /* for the dests in the service */
728 struct hlist_node d_list; /* for table with all the dests */
730 u16 af; /* address family */
731 __be16 port; /* port number of the server */
732 union nf_inet_addr addr; /* IP address of the server */
733 volatile unsigned int flags; /* dest status flags */
734 atomic_t conn_flags; /* flags to copy to conn */
735 atomic_t weight; /* server weight */
737 atomic_t refcnt; /* reference counter */
738 struct ip_vs_stats stats; /* statistics */
739 unsigned long state; /* state flags */
741 /* connection counters and thresholds */
742 atomic_t activeconns; /* active connections */
743 atomic_t inactconns; /* inactive connections */
744 atomic_t persistconns; /* persistent connections */
745 __u32 u_threshold; /* upper threshold */
746 __u32 l_threshold; /* lower threshold */
748 /* for destination cache */
749 spinlock_t dst_lock; /* lock of dst_cache */
750 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
752 /* for virtual service */
753 struct ip_vs_service *svc; /* service it belongs to */
754 __u16 protocol; /* which protocol (TCP/UDP) */
755 __be16 vport; /* virtual port number */
756 union nf_inet_addr vaddr; /* virtual IP address */
757 __u32 vfwmark; /* firewall mark of service */
759 struct list_head t_list; /* in dest_trash */
760 struct rcu_head rcu_head;
761 unsigned int in_rs_table:1; /* we are in rs_table */
766 * The scheduler object
768 struct ip_vs_scheduler {
769 struct list_head n_list; /* d-linked list head */
770 char *name; /* scheduler name */
771 atomic_t refcnt; /* reference counter */
772 struct module *module; /* THIS_MODULE/NULL */
774 /* scheduler initializing service */
775 int (*init_service)(struct ip_vs_service *svc);
776 /* scheduling service finish */
777 void (*done_service)(struct ip_vs_service *svc);
779 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
780 /* dest is unlinked */
781 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
782 /* dest is updated */
783 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
785 /* selecting a server from the given service */
786 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
787 const struct sk_buff *skb);
790 /* The persistence engine object */
792 struct list_head n_list; /* d-linked list head */
793 char *name; /* scheduler name */
794 atomic_t refcnt; /* reference counter */
795 struct module *module; /* THIS_MODULE/NULL */
797 /* get the connection template, if any */
798 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
799 bool (*ct_match)(const struct ip_vs_conn_param *p,
800 struct ip_vs_conn *ct);
801 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
803 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
807 * The application module object (a.k.a. app incarnation)
810 struct list_head a_list; /* member in app list */
811 int type; /* IP_VS_APP_TYPE_xxx */
812 char *name; /* application module name */
814 struct module *module; /* THIS_MODULE/NULL */
815 struct list_head incs_list; /* list of incarnations */
817 /* members for application incarnations */
818 struct list_head p_list; /* member in proto app list */
819 struct ip_vs_app *app; /* its real application */
820 __be16 port; /* port number in net order */
821 atomic_t usecnt; /* usage counter */
822 struct rcu_head rcu_head;
825 * output hook: Process packet in inout direction, diff set for TCP.
826 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
827 * 2=Mangled but checksum was not updated
829 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
830 struct sk_buff *, int *diff);
833 * input hook: Process packet in outin direction, diff set for TCP.
834 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
835 * 2=Mangled but checksum was not updated
837 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
838 struct sk_buff *, int *diff);
840 /* ip_vs_app initializer */
841 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
843 /* ip_vs_app finish */
844 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
848 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
849 struct ip_vs_protocol *);
851 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
857 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
858 int *verdict, struct ip_vs_conn **cpp);
861 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
862 const struct iphdr *iph, int inverse);
865 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
866 const struct iphdr *iph, int inverse);
868 int (*state_transition)(struct ip_vs_conn *cp, int direction,
869 const struct sk_buff *skb,
870 struct ip_vs_app *app);
872 void (*timeout_change)(struct ip_vs_app *app, int flags);
875 struct ipvs_master_sync_state {
876 struct list_head sync_queue;
877 struct ip_vs_sync_buff *sync_buff;
879 unsigned int sync_queue_delay;
880 struct task_struct *master_thread;
881 struct delayed_work master_wakeup_work;
882 struct netns_ipvs *ipvs;
885 /* How much time to keep dests in trash */
886 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
888 /* IPVS in network namespace */
890 int gen; /* Generation */
891 int enable; /* enable like nf_hooks do */
893 * Hash table: for real service lookups
895 #define IP_VS_RTAB_BITS 4
896 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
897 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
899 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
901 struct list_head app_list;
903 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
904 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
905 /* ip_vs_proto_tcp */
906 #ifdef CONFIG_IP_VS_PROTO_TCP
907 #define TCP_APP_TAB_BITS 4
908 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
909 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
910 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
912 /* ip_vs_proto_udp */
913 #ifdef CONFIG_IP_VS_PROTO_UDP
914 #define UDP_APP_TAB_BITS 4
915 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
916 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
917 struct list_head udp_apps[UDP_APP_TAB_SIZE];
919 /* ip_vs_proto_sctp */
920 #ifdef CONFIG_IP_VS_PROTO_SCTP
921 #define SCTP_APP_TAB_BITS 4
922 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
923 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
924 /* Hash table for SCTP application incarnations */
925 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
928 atomic_t conn_count; /* connection counter */
931 struct ip_vs_stats tot_stats; /* Statistics & est. */
933 int num_services; /* no of virtual services */
935 /* Trash for destinations */
936 struct list_head dest_trash;
937 spinlock_t dest_trash_lock;
938 struct timer_list dest_trash_timer; /* expiration timer */
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 /* Get reference to gain full access to conn.
1187 * By default, RCU read-side critical sections have access only to
1188 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1190 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1192 return atomic_inc_not_zero(&cp->refcnt);
1195 /* put back the conn without restarting its timer */
1196 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1198 smp_mb__before_atomic_dec();
1199 atomic_dec(&cp->refcnt);
1201 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1202 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1204 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1205 const union nf_inet_addr *daddr,
1206 __be16 dport, unsigned int flags,
1207 struct ip_vs_dest *dest, __u32 fwmark);
1208 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1210 extern const char * ip_vs_state_name(__u16 proto, int state);
1212 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1213 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1214 extern void ip_vs_random_dropentry(struct net *net);
1215 extern int ip_vs_conn_init(void);
1216 extern void ip_vs_conn_cleanup(void);
1218 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1220 struct ip_vs_conn *ctl_cp = cp->control;
1222 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1224 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1226 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1232 IP_VS_DBG_BUF(7, "DELeting control for: "
1233 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1234 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1236 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1237 ntohs(ctl_cp->cport));
1240 if (atomic_read(&ctl_cp->n_control) == 0) {
1241 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1243 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1245 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1250 atomic_dec(&ctl_cp->n_control);
1254 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1257 IP_VS_ERR_BUF("request control ADD for already controlled: "
1259 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1261 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1264 ip_vs_control_del(cp);
1267 IP_VS_DBG_BUF(7, "ADDing control for: "
1268 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1269 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1271 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1272 ntohs(ctl_cp->cport));
1274 cp->control = ctl_cp;
1275 atomic_inc(&ctl_cp->n_control);
1279 * IPVS netns init & cleanup functions
1281 extern int ip_vs_estimator_net_init(struct net *net);
1282 extern int ip_vs_control_net_init(struct net *net);
1283 extern int ip_vs_protocol_net_init(struct net *net);
1284 extern int ip_vs_app_net_init(struct net *net);
1285 extern int ip_vs_conn_net_init(struct net *net);
1286 extern int ip_vs_sync_net_init(struct net *net);
1287 extern void ip_vs_conn_net_cleanup(struct net *net);
1288 extern void ip_vs_app_net_cleanup(struct net *net);
1289 extern void ip_vs_protocol_net_cleanup(struct net *net);
1290 extern void ip_vs_control_net_cleanup(struct net *net);
1291 extern void ip_vs_estimator_net_cleanup(struct net *net);
1292 extern void ip_vs_sync_net_cleanup(struct net *net);
1293 extern void ip_vs_service_net_cleanup(struct net *net);
1296 * IPVS application functions
1297 * (from ip_vs_app.c)
1299 #define IP_VS_APP_MAX_PORTS 8
1300 extern struct ip_vs_app *register_ip_vs_app(struct net *net,
1301 struct ip_vs_app *app);
1302 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1303 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1304 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1305 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1306 __u16 proto, __u16 port);
1307 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1308 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1310 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1311 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1313 int register_ip_vs_pe(struct ip_vs_pe *pe);
1314 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1315 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1316 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1319 * Use a #define to avoid all of module.h just for these trivial ops
1321 #define ip_vs_pe_get(pe) \
1322 if (pe && pe->module) \
1323 __module_get(pe->module);
1325 #define ip_vs_pe_put(pe) \
1326 if (pe && pe->module) \
1327 module_put(pe->module);
1330 * IPVS protocol functions (from ip_vs_proto.c)
1332 extern int ip_vs_protocol_init(void);
1333 extern void ip_vs_protocol_cleanup(void);
1334 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1335 extern int *ip_vs_create_timeout_table(int *table, int size);
1337 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1338 const char *name, int to);
1340 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1341 const struct sk_buff *skb,
1342 int offset, const char *msg);
1344 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1345 extern struct ip_vs_protocol ip_vs_protocol_udp;
1346 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1347 extern struct ip_vs_protocol ip_vs_protocol_esp;
1348 extern struct ip_vs_protocol ip_vs_protocol_ah;
1349 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1352 * Registering/unregistering scheduler functions
1353 * (from ip_vs_sched.c)
1355 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1356 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1357 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1358 struct ip_vs_scheduler *scheduler);
1359 extern void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1360 struct ip_vs_scheduler *sched);
1361 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1362 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1363 extern struct ip_vs_conn *
1364 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1365 struct ip_vs_proto_data *pd, int *ignored,
1366 struct ip_vs_iphdr *iph);
1367 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1368 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1370 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1374 * IPVS control data and functions (from ip_vs_ctl.c)
1376 extern struct ip_vs_stats ip_vs_stats;
1377 extern int sysctl_ip_vs_sync_ver;
1379 extern struct ip_vs_service *
1380 ip_vs_service_find(struct net *net, int af, __u32 fwmark, __u16 protocol,
1381 const union nf_inet_addr *vaddr, __be16 vport);
1384 ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
1385 const union nf_inet_addr *daddr, __be16 dport);
1387 extern int ip_vs_use_count_inc(void);
1388 extern void ip_vs_use_count_dec(void);
1389 extern int ip_vs_register_nl_ioctl(void);
1390 extern void ip_vs_unregister_nl_ioctl(void);
1391 extern int ip_vs_control_init(void);
1392 extern void ip_vs_control_cleanup(void);
1393 extern struct ip_vs_dest *
1394 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1395 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1396 __u16 protocol, __u32 fwmark, __u32 flags);
1397 extern void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1399 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1401 atomic_inc(&dest->refcnt);
1404 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1406 smp_mb__before_atomic_dec();
1407 atomic_dec(&dest->refcnt);
1411 * IPVS sync daemon data and function prototypes
1412 * (from ip_vs_sync.c)
1414 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1416 extern int stop_sync_thread(struct net *net, int state);
1417 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1421 * IPVS rate estimator prototypes (from ip_vs_est.c)
1423 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1424 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1425 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1426 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1427 struct ip_vs_stats *stats);
1430 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1432 extern int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1433 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1434 extern int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1435 struct ip_vs_protocol *pp,
1436 struct ip_vs_iphdr *iph);
1437 extern int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1438 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1439 extern int ip_vs_tunnel_xmit(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_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1443 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1444 extern int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1445 struct ip_vs_protocol *pp, int offset,
1446 unsigned int hooknum, struct ip_vs_iphdr *iph);
1447 extern void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1449 #ifdef CONFIG_IP_VS_IPV6
1450 extern int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1451 struct ip_vs_protocol *pp,
1452 struct ip_vs_iphdr *iph);
1453 extern int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1454 struct ip_vs_protocol *pp,
1455 struct ip_vs_iphdr *iph);
1456 extern int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1457 struct ip_vs_protocol *pp,
1458 struct ip_vs_iphdr *iph);
1459 extern int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1460 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1461 extern int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1462 struct ip_vs_protocol *pp, int offset,
1463 unsigned int hooknum, struct ip_vs_iphdr *iph);
1466 #ifdef CONFIG_SYSCTL
1468 * This is a simple mechanism to ignore packets when
1469 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1470 * we start to drop 1/rate of the packets
1473 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1475 if (!ipvs->drop_rate)
1477 if (--ipvs->drop_counter > 0)
1479 ipvs->drop_counter = ipvs->drop_rate;
1483 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1487 * ip_vs_fwd_tag returns the forwarding tag of the connection
1489 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1491 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1495 switch (IP_VS_FWD_METHOD(cp)) {
1496 case IP_VS_CONN_F_MASQ:
1498 case IP_VS_CONN_F_LOCALNODE:
1500 case IP_VS_CONN_F_TUNNEL:
1502 case IP_VS_CONN_F_DROUTE:
1504 case IP_VS_CONN_F_BYPASS:
1512 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1513 struct ip_vs_conn *cp, int dir);
1515 #ifdef CONFIG_IP_VS_IPV6
1516 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1517 struct ip_vs_conn *cp, int dir);
1520 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1522 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1524 __be32 diff[2] = { ~old, new };
1526 return csum_partial(diff, sizeof(diff), oldsum);
1529 #ifdef CONFIG_IP_VS_IPV6
1530 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1533 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1534 new[3], new[2], new[1], new[0] };
1536 return csum_partial(diff, sizeof(diff), oldsum);
1540 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1542 __be16 diff[2] = { ~old, new };
1544 return csum_partial(diff, sizeof(diff), oldsum);
1548 * Forget current conntrack (unconfirmed) and attach notrack entry
1550 static inline void ip_vs_notrack(struct sk_buff *skb)
1552 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1553 enum ip_conntrack_info ctinfo;
1554 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1556 if (!ct || !nf_ct_is_untracked(ct)) {
1557 nf_conntrack_put(skb->nfct);
1558 skb->nfct = &nf_ct_untracked_get()->ct_general;
1559 skb->nfctinfo = IP_CT_NEW;
1560 nf_conntrack_get(skb->nfct);
1565 #ifdef CONFIG_IP_VS_NFCT
1567 * Netfilter connection tracking
1568 * (from ip_vs_nfct.c)
1570 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1572 #ifdef CONFIG_SYSCTL
1573 return ipvs->sysctl_conntrack;
1579 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1581 extern int ip_vs_confirm_conntrack(struct sk_buff *skb);
1582 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1583 struct ip_vs_conn *cp, u_int8_t proto,
1584 const __be16 port, int from_rs);
1585 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1589 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1594 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1595 struct ip_vs_conn *cp, int outin)
1599 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1604 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1607 /* CONFIG_IP_VS_NFCT */
1610 static inline unsigned int
1611 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1614 * We think the overhead of processing active connections is 256
1615 * times higher than that of inactive connections in average. (This
1616 * 256 times might not be accurate, we will change it later) We
1617 * use the following formula to estimate the overhead now:
1618 * dest->activeconns*256 + dest->inactconns
1620 return (atomic_read(&dest->activeconns) << 8) +
1621 atomic_read(&dest->inactconns);
1624 #endif /* _NET_IP_VS_H */