2 * data structure and functionality definitions
8 #include <linux/ip_vs.h> /* definitions shared with userland */
10 #include <asm/types.h> /* for __uXX types */
12 #include <linux/list.h> /* for struct list_head */
13 #include <linux/spinlock.h> /* for struct rwlock_t */
14 #include <linux/atomic.h> /* for struct atomic_t */
15 #include <linux/compiler.h>
16 #include <linux/timer.h>
17 #include <linux/bug.h>
19 #include <net/checksum.h>
20 #include <linux/netfilter.h> /* for union nf_inet_addr */
22 #include <linux/ipv6.h> /* for struct ipv6hdr */
24 #if IS_ENABLED(CONFIG_IP_VS_IPV6)
25 #include <linux/netfilter_ipv6/ip6_tables.h>
27 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
28 #include <net/netfilter/nf_conntrack.h>
30 #include <net/net_namespace.h> /* Netw namespace */
32 #define IP_VS_HDR_INVERSE 1
33 #define IP_VS_HDR_ICMP 2
35 /* 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 && dev_net(skb->dev)))
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 && sock_net(skb->sk)))
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 && sock_net(skb->sk)))
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 && dev_net(skb->dev)))
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 int hdr_flags; /* ipvs flags */
111 __u32 off; /* Where IP or IPv4 header starts */
112 __u32 len; /* IPv4 simply where L4 starts
113 * IPv6 where L4 Transport Header starts */
114 __u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
117 union nf_inet_addr saddr;
118 union nf_inet_addr daddr;
121 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
122 int len, void *buffer,
123 const struct ip_vs_iphdr *ipvsh)
125 return skb_header_pointer(skb, offset, len, buffer);
128 /* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
129 * IPv6 requires some extra work, as finding proper header position,
130 * depend on the IPv6 extension headers.
133 ip_vs_fill_iph_skb_off(int af, const struct sk_buff *skb, int offset,
134 int hdr_flags, struct ip_vs_iphdr *iphdr)
136 iphdr->hdr_flags = hdr_flags;
139 #ifdef CONFIG_IP_VS_IPV6
140 if (af == AF_INET6) {
142 const struct ipv6hdr *iph = skb_header_pointer(
143 skb, offset, sizeof(_iph), &_iph);
147 iphdr->saddr.in6 = iph->saddr;
148 iphdr->daddr.in6 = iph->daddr;
149 /* ipv6_find_hdr() updates len, flags */
152 iphdr->protocol = ipv6_find_hdr(skb, &iphdr->len, -1,
155 if (iphdr->protocol < 0)
161 const struct iphdr *iph = skb_header_pointer(
162 skb, offset, sizeof(_iph), &_iph);
166 iphdr->len = offset + iph->ihl * 4;
168 iphdr->protocol = iph->protocol;
169 iphdr->saddr.ip = iph->saddr;
170 iphdr->daddr.ip = iph->daddr;
177 ip_vs_fill_iph_skb_icmp(int af, const struct sk_buff *skb, int offset,
178 bool inverse, struct ip_vs_iphdr *iphdr)
180 int hdr_flags = IP_VS_HDR_ICMP;
183 hdr_flags |= IP_VS_HDR_INVERSE;
185 return ip_vs_fill_iph_skb_off(af, skb, offset, hdr_flags, iphdr);
189 ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, bool inverse,
190 struct ip_vs_iphdr *iphdr)
195 hdr_flags |= IP_VS_HDR_INVERSE;
197 return ip_vs_fill_iph_skb_off(af, skb, skb_network_offset(skb),
202 ip_vs_iph_inverse(const struct ip_vs_iphdr *iph)
204 return !!(iph->hdr_flags & IP_VS_HDR_INVERSE);
208 ip_vs_iph_icmp(const struct ip_vs_iphdr *iph)
210 return !!(iph->hdr_flags & IP_VS_HDR_ICMP);
213 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
214 const union nf_inet_addr *src)
216 #ifdef CONFIG_IP_VS_IPV6
224 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
225 const union nf_inet_addr *src)
227 #ifdef CONFIG_IP_VS_IPV6
228 if (af == AF_INET6) {
239 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
240 const union nf_inet_addr *b)
242 #ifdef CONFIG_IP_VS_IPV6
244 return ipv6_addr_equal(&a->in6, &b->in6);
246 return a->ip == b->ip;
249 #ifdef CONFIG_IP_VS_DEBUG
250 #include <linux/net.h>
252 int ip_vs_get_debug_level(void);
254 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
255 const union nf_inet_addr *addr,
259 #ifdef CONFIG_IP_VS_IPV6
261 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
265 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
269 BUG_ON(*idx > buf_len + 1);
270 return &buf[*idx - len];
273 #define IP_VS_DBG_BUF(level, msg, ...) \
275 char ip_vs_dbg_buf[160]; \
276 int ip_vs_dbg_idx = 0; \
277 if (level <= ip_vs_get_debug_level()) \
278 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
280 #define IP_VS_ERR_BUF(msg...) \
282 char ip_vs_dbg_buf[160]; \
283 int ip_vs_dbg_idx = 0; \
287 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
288 #define IP_VS_DBG_ADDR(af, addr) \
289 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
290 sizeof(ip_vs_dbg_buf), addr, \
293 #define IP_VS_DBG(level, msg, ...) \
295 if (level <= ip_vs_get_debug_level()) \
296 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
298 #define IP_VS_DBG_RL(msg, ...) \
300 if (net_ratelimit()) \
301 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
303 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
305 if (level <= ip_vs_get_debug_level()) \
306 pp->debug_packet(af, pp, skb, ofs, msg); \
308 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
310 if (level <= ip_vs_get_debug_level() && \
312 pp->debug_packet(af, pp, skb, ofs, msg); \
314 #else /* NO DEBUGGING at ALL */
315 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
316 #define IP_VS_ERR_BUF(msg...) do {} while (0)
317 #define IP_VS_DBG(level, msg...) do {} while (0)
318 #define IP_VS_DBG_RL(msg...) do {} while (0)
319 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
320 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
323 #define IP_VS_BUG() BUG()
324 #define IP_VS_ERR_RL(msg, ...) \
326 if (net_ratelimit()) \
327 pr_err(msg, ##__VA_ARGS__); \
330 #ifdef CONFIG_IP_VS_DEBUG
331 #define EnterFunction(level) \
333 if (level <= ip_vs_get_debug_level()) \
335 pr_fmt("Enter: %s, %s line %i\n"), \
336 __func__, __FILE__, __LINE__); \
338 #define LeaveFunction(level) \
340 if (level <= ip_vs_get_debug_level()) \
342 pr_fmt("Leave: %s, %s line %i\n"), \
343 __func__, __FILE__, __LINE__); \
346 #define EnterFunction(level) do {} while (0)
347 #define LeaveFunction(level) do {} while (0)
350 /* The port number of FTP service (in network order). */
351 #define FTPPORT cpu_to_be16(21)
352 #define FTPDATA cpu_to_be16(20)
354 /* TCP State Values */
356 IP_VS_TCP_S_NONE = 0,
357 IP_VS_TCP_S_ESTABLISHED,
358 IP_VS_TCP_S_SYN_SENT,
359 IP_VS_TCP_S_SYN_RECV,
360 IP_VS_TCP_S_FIN_WAIT,
361 IP_VS_TCP_S_TIME_WAIT,
363 IP_VS_TCP_S_CLOSE_WAIT,
364 IP_VS_TCP_S_LAST_ACK,
370 /* UDP State Values */
376 /* ICMP State Values */
382 /* SCTP State Values */
383 enum ip_vs_sctp_states {
387 IP_VS_SCTP_S_COOKIE_SENT,
388 IP_VS_SCTP_S_COOKIE_REPLIED,
389 IP_VS_SCTP_S_COOKIE_WAIT,
391 IP_VS_SCTP_S_COOKIE_ECHOED,
392 IP_VS_SCTP_S_ESTABLISHED,
393 IP_VS_SCTP_S_SHUTDOWN_SENT,
394 IP_VS_SCTP_S_SHUTDOWN_RECEIVED,
395 IP_VS_SCTP_S_SHUTDOWN_ACK_SENT,
396 IP_VS_SCTP_S_REJECTED,
401 /* Delta sequence info structure
402 * Each ip_vs_conn has 2 (output AND input seq. changes).
403 * Only used in the VS/NAT.
406 __u32 init_seq; /* Add delta from this seq */
407 __u32 delta; /* Delta in sequence numbers */
408 __u32 previous_delta; /* Delta in sequence numbers
409 * before last resized pkt */
412 /* counters per cpu */
413 struct ip_vs_counters {
414 __u64 conns; /* connections scheduled */
415 __u64 inpkts; /* incoming packets */
416 __u64 outpkts; /* outgoing packets */
417 __u64 inbytes; /* incoming bytes */
418 __u64 outbytes; /* outgoing bytes */
421 struct ip_vs_cpu_stats {
422 struct ip_vs_counters cnt;
423 struct u64_stats_sync syncp;
426 /* IPVS statistics objects */
427 struct ip_vs_estimator {
428 struct list_head list;
444 * IPVS statistics object, 64-bit kernel version of struct ip_vs_stats_user
446 struct ip_vs_kstats {
447 u64 conns; /* connections scheduled */
448 u64 inpkts; /* incoming packets */
449 u64 outpkts; /* outgoing packets */
450 u64 inbytes; /* incoming bytes */
451 u64 outbytes; /* outgoing bytes */
453 u64 cps; /* current connection rate */
454 u64 inpps; /* current in packet rate */
455 u64 outpps; /* current out packet rate */
456 u64 inbps; /* current in byte rate */
457 u64 outbps; /* current out byte rate */
461 struct ip_vs_kstats kstats; /* kernel statistics */
462 struct ip_vs_estimator est; /* estimator */
463 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
464 spinlock_t lock; /* spin lock */
465 struct ip_vs_kstats kstats0; /* reset values */
473 struct ip_vs_proto_data;
475 struct ip_vs_protocol {
476 struct ip_vs_protocol *next;
482 void (*init)(struct ip_vs_protocol *pp);
484 void (*exit)(struct ip_vs_protocol *pp);
486 int (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
488 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
490 int (*conn_schedule)(int af, struct sk_buff *skb,
491 struct ip_vs_proto_data *pd,
492 int *verdict, struct ip_vs_conn **cpp,
493 struct ip_vs_iphdr *iph);
496 (*conn_in_get)(int af,
497 const struct sk_buff *skb,
498 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);
505 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
506 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
508 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
509 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
511 int (*csum_check)(int af, struct sk_buff *skb,
512 struct ip_vs_protocol *pp);
514 const char *(*state_name)(int state);
516 void (*state_transition)(struct ip_vs_conn *cp, int direction,
517 const struct sk_buff *skb,
518 struct ip_vs_proto_data *pd);
520 int (*register_app)(struct net *net, struct ip_vs_app *inc);
522 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
524 int (*app_conn_bind)(struct ip_vs_conn *cp);
526 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
527 const struct sk_buff *skb,
531 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
534 /* protocol data per netns */
535 struct ip_vs_proto_data {
536 struct ip_vs_proto_data *next;
537 struct ip_vs_protocol *pp;
538 int *timeout_table; /* protocol timeout table */
539 atomic_t appcnt; /* counter of proto app incs. */
540 struct tcp_states_t *tcp_state_table;
543 struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
544 struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
545 unsigned short proto);
547 struct ip_vs_conn_param {
549 const union nf_inet_addr *caddr;
550 const union nf_inet_addr *vaddr;
556 const struct ip_vs_pe *pe;
561 /* IP_VS structure allocated for each dynamically scheduled connection */
563 struct hlist_node c_list; /* hashed list heads */
564 /* Protocol, addresses and port numbers */
568 u16 af; /* address family */
569 union nf_inet_addr caddr; /* client address */
570 union nf_inet_addr vaddr; /* virtual address */
571 union nf_inet_addr daddr; /* destination address */
572 volatile __u32 flags; /* status flags */
573 __u16 protocol; /* Which protocol (TCP/UDP) */
574 __u16 daf; /* Address family of the dest */
576 struct net *net; /* Name space */
579 /* counter and timer */
580 atomic_t refcnt; /* reference count */
581 struct timer_list timer; /* Expiration timer */
582 volatile unsigned long timeout; /* timeout */
584 /* Flags and state transition */
585 spinlock_t lock; /* lock for state transition */
586 volatile __u16 state; /* state info */
587 volatile __u16 old_state; /* old state, to be used for
588 * state transition triggerd
591 __u32 fwmark; /* Fire wall mark from skb */
592 unsigned long sync_endtime; /* jiffies + sent_retries */
594 /* Control members */
595 struct ip_vs_conn *control; /* Master control connection */
596 atomic_t n_control; /* Number of controlled ones */
597 struct ip_vs_dest *dest; /* real server */
598 atomic_t in_pkts; /* incoming packet counter */
600 /* Packet transmitter for different forwarding methods. If it
601 * mangles the packet, it must return NF_DROP or better NF_STOLEN,
602 * otherwise this must be changed to a sk_buff **.
603 * NF_ACCEPT can be returned when destination is local.
605 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
606 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
608 /* Note: we can group the following members into a structure,
609 * in order to save more space, and the following members are
610 * only used in VS/NAT anyway
612 struct ip_vs_app *app; /* bound ip_vs_app object */
613 void *app_data; /* Application private data */
614 struct ip_vs_seq in_seq; /* incoming seq. struct */
615 struct ip_vs_seq out_seq; /* outgoing seq. struct */
617 const struct ip_vs_pe *pe;
621 struct rcu_head rcu_head;
624 /* To save some memory in conn table when name space is disabled. */
625 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
634 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
641 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
645 return cp->net == net;
651 /* Extended internal versions of struct ip_vs_service_user and ip_vs_dest_user
654 * We need these to conveniently pass around service and destination
655 * options, but unfortunately, we also need to keep the old definitions to
656 * maintain userspace backwards compatibility for the setsockopt interface.
658 struct ip_vs_service_user_kern {
659 /* virtual service addresses */
662 union nf_inet_addr addr; /* virtual ip address */
664 u32 fwmark; /* firwall mark of service */
666 /* virtual service options */
669 unsigned int flags; /* virtual service flags */
670 unsigned int timeout; /* persistent timeout in sec */
671 __be32 netmask; /* persistent netmask or plen */
675 struct ip_vs_dest_user_kern {
676 /* destination server address */
677 union nf_inet_addr addr;
680 /* real server options */
681 unsigned int conn_flags; /* connection flags */
682 int weight; /* destination weight */
684 /* thresholds for active connections */
685 u32 u_threshold; /* upper threshold */
686 u32 l_threshold; /* lower threshold */
688 /* Address family of addr */
694 * The information about the virtual service offered to the net and the
695 * forwarding entries.
697 struct ip_vs_service {
698 struct hlist_node s_list; /* for normal service table */
699 struct hlist_node f_list; /* for fwmark-based service table */
700 atomic_t refcnt; /* reference counter */
702 u16 af; /* address family */
703 __u16 protocol; /* which protocol (TCP/UDP) */
704 union nf_inet_addr addr; /* IP address for virtual service */
705 __be16 port; /* port number for the service */
706 __u32 fwmark; /* firewall mark of the service */
707 unsigned int flags; /* service status flags */
708 unsigned int timeout; /* persistent timeout in ticks */
709 __be32 netmask; /* grouping granularity, mask/plen */
712 struct list_head destinations; /* real server d-linked list */
713 __u32 num_dests; /* number of servers */
714 struct ip_vs_stats stats; /* statistics for the service */
717 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
718 spinlock_t sched_lock; /* lock sched_data */
719 void *sched_data; /* scheduler application data */
721 /* alternate persistence engine */
722 struct ip_vs_pe __rcu *pe;
724 struct rcu_head rcu_head;
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;
735 /* The real server destination forwarding entry with ip address, port number,
739 struct list_head n_list; /* for the dests in the service */
740 struct hlist_node d_list; /* for table with all the dests */
742 u16 af; /* address family */
743 __be16 port; /* port number of the server */
744 union nf_inet_addr addr; /* IP address of the server */
745 volatile unsigned int flags; /* dest status flags */
746 atomic_t conn_flags; /* flags to copy to conn */
747 atomic_t weight; /* server weight */
749 atomic_t refcnt; /* reference counter */
750 struct ip_vs_stats stats; /* statistics */
751 unsigned long idle_start; /* start time, jiffies */
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 __rcu *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 */
771 struct list_head t_list; /* in dest_trash */
772 unsigned int in_rs_table:1; /* we are in rs_table */
775 /* 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 void (*done_service)(struct ip_vs_service *svc);
787 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
788 /* dest is unlinked */
789 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
790 /* dest is updated */
791 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
793 /* selecting a server from the given service */
794 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
795 const struct sk_buff *skb,
796 struct ip_vs_iphdr *iph);
799 /* The persistence engine object */
801 struct list_head n_list; /* d-linked list head */
802 char *name; /* scheduler name */
803 atomic_t refcnt; /* reference counter */
804 struct module *module; /* THIS_MODULE/NULL */
806 /* get the connection template, if any */
807 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
808 bool (*ct_match)(const struct ip_vs_conn_param *p,
809 struct ip_vs_conn *ct);
810 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
812 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
815 /* The application module object (a.k.a. app incarnation) */
817 struct list_head a_list; /* member in app list */
818 int type; /* IP_VS_APP_TYPE_xxx */
819 char *name; /* application module name */
821 struct module *module; /* THIS_MODULE/NULL */
822 struct list_head incs_list; /* list of incarnations */
824 /* members for application incarnations */
825 struct list_head p_list; /* member in proto app list */
826 struct ip_vs_app *app; /* its real application */
827 __be16 port; /* port number in net order */
828 atomic_t usecnt; /* usage counter */
829 struct rcu_head rcu_head;
831 /* output hook: Process packet in inout direction, diff set for TCP.
832 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
833 * 2=Mangled but checksum was not updated
835 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
836 struct sk_buff *, int *diff);
838 /* input hook: Process packet in outin direction, diff set for TCP.
839 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
840 * 2=Mangled but checksum was not updated
842 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
843 struct sk_buff *, int *diff);
845 /* ip_vs_app initializer */
846 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
848 /* ip_vs_app finish */
849 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
853 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
854 struct ip_vs_protocol *);
856 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
862 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
863 int *verdict, struct ip_vs_conn **cpp);
866 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
867 const struct iphdr *iph, int inverse);
870 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
871 const struct iphdr *iph, int inverse);
873 int (*state_transition)(struct ip_vs_conn *cp, int direction,
874 const struct sk_buff *skb,
875 struct ip_vs_app *app);
877 void (*timeout_change)(struct ip_vs_app *app, int flags);
880 struct ipvs_master_sync_state {
881 struct list_head sync_queue;
882 struct ip_vs_sync_buff *sync_buff;
883 unsigned long sync_queue_len;
884 unsigned int sync_queue_delay;
885 struct task_struct *master_thread;
886 struct delayed_work master_wakeup_work;
887 struct netns_ipvs *ipvs;
890 /* How much time to keep dests in trash */
891 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
893 struct ipvs_sync_daemon_cfg {
894 union nf_inet_addr mcast_group;
900 /* multicast interface name */
901 char mcast_ifn[IP_VS_IFNAME_MAXLEN];
904 /* IPVS in network namespace */
906 int gen; /* Generation */
907 int enable; /* enable like nf_hooks do */
908 /* Hash table: for real service lookups */
909 #define IP_VS_RTAB_BITS 4
910 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
911 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
913 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
915 struct list_head app_list;
917 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
918 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
919 /* ip_vs_proto_tcp */
920 #ifdef CONFIG_IP_VS_PROTO_TCP
921 #define TCP_APP_TAB_BITS 4
922 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
923 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
924 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
926 /* ip_vs_proto_udp */
927 #ifdef CONFIG_IP_VS_PROTO_UDP
928 #define UDP_APP_TAB_BITS 4
929 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
930 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
931 struct list_head udp_apps[UDP_APP_TAB_SIZE];
933 /* ip_vs_proto_sctp */
934 #ifdef CONFIG_IP_VS_PROTO_SCTP
935 #define SCTP_APP_TAB_BITS 4
936 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
937 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
938 /* Hash table for SCTP application incarnations */
939 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
942 atomic_t conn_count; /* connection counter */
945 struct ip_vs_stats tot_stats; /* Statistics & est. */
947 int num_services; /* no of virtual services */
949 /* Trash for destinations */
950 struct list_head dest_trash;
951 spinlock_t dest_trash_lock;
952 struct timer_list dest_trash_timer; /* expiration timer */
953 /* Service counters */
954 atomic_t ftpsvc_counter;
955 atomic_t nullsvc_counter;
958 /* 1/rate drop and drop-entry variables */
959 struct delayed_work defense_work; /* Work handler */
964 spinlock_t dropentry_lock; /* drop entry handling */
965 spinlock_t droppacket_lock; /* drop packet handling */
966 spinlock_t securetcp_lock; /* state and timeout tables */
969 struct ctl_table_header *sysctl_hdr;
970 struct ctl_table *sysctl_tbl;
973 /* sysctl variables */
974 int sysctl_amemthresh;
975 int sysctl_am_droprate;
976 int sysctl_drop_entry;
977 int sysctl_drop_packet;
978 int sysctl_secure_tcp;
979 #ifdef CONFIG_IP_VS_NFCT
980 int sysctl_conntrack;
982 int sysctl_snat_reroute;
984 int sysctl_sync_ports;
985 int sysctl_sync_persist_mode;
986 unsigned long sysctl_sync_qlen_max;
987 int sysctl_sync_sock_size;
988 int sysctl_cache_bypass;
989 int sysctl_expire_nodest_conn;
990 int sysctl_sloppy_tcp;
991 int sysctl_sloppy_sctp;
992 int sysctl_expire_quiescent_template;
993 int sysctl_sync_threshold[2];
994 unsigned int sysctl_sync_refresh_period;
995 int sysctl_sync_retries;
996 int sysctl_nat_icmp_send;
997 int sysctl_pmtu_disc;
998 int sysctl_backup_only;
999 int sysctl_conn_reuse_mode;
1000 int sysctl_schedule_icmp;
1003 int sysctl_lblc_expiration;
1004 struct ctl_table_header *lblc_ctl_header;
1005 struct ctl_table *lblc_ctl_table;
1007 int sysctl_lblcr_expiration;
1008 struct ctl_table_header *lblcr_ctl_header;
1009 struct ctl_table *lblcr_ctl_table;
1011 struct list_head est_list; /* estimator list */
1012 spinlock_t est_lock;
1013 struct timer_list est_timer; /* Estimation timer */
1015 spinlock_t sync_lock;
1016 struct ipvs_master_sync_state *ms;
1017 spinlock_t sync_buff_lock;
1018 struct task_struct **backup_threads;
1020 volatile int sync_state;
1021 struct mutex sync_mutex;
1022 struct ipvs_sync_daemon_cfg mcfg; /* Master Configuration */
1023 struct ipvs_sync_daemon_cfg bcfg; /* Backup Configuration */
1024 /* net name space ptr */
1025 struct net *net; /* Needed by timer routines */
1026 /* Number of heterogeneous destinations, needed becaus heterogeneous
1027 * are not supported when synchronization is enabled.
1029 unsigned int mixed_address_family_dests;
1032 #define DEFAULT_SYNC_THRESHOLD 3
1033 #define DEFAULT_SYNC_PERIOD 50
1034 #define DEFAULT_SYNC_VER 1
1035 #define DEFAULT_SLOPPY_TCP 0
1036 #define DEFAULT_SLOPPY_SCTP 0
1037 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1038 #define DEFAULT_SYNC_RETRIES 0
1039 #define IPVS_SYNC_WAKEUP_RATE 8
1040 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1041 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1042 #define IPVS_SYNC_CHECK_PERIOD HZ
1043 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1044 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1046 #ifdef CONFIG_SYSCTL
1048 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1050 return ipvs->sysctl_sync_threshold[0];
1053 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1055 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
1058 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1060 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
1063 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1065 return ipvs->sysctl_sync_retries;
1068 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1070 return ipvs->sysctl_sync_ver;
1073 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1075 return ipvs->sysctl_sloppy_tcp;
1078 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1080 return ipvs->sysctl_sloppy_sctp;
1083 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1085 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
1088 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1090 return ipvs->sysctl_sync_persist_mode;
1093 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1095 return ipvs->sysctl_sync_qlen_max;
1098 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1100 return ipvs->sysctl_sync_sock_size;
1103 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1105 return ipvs->sysctl_pmtu_disc;
1108 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1110 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1111 ipvs->sysctl_backup_only;
1114 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1116 return ipvs->sysctl_conn_reuse_mode;
1119 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1121 return ipvs->sysctl_schedule_icmp;
1126 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1128 return DEFAULT_SYNC_THRESHOLD;
1131 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1133 return DEFAULT_SYNC_PERIOD;
1136 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1138 return DEFAULT_SYNC_REFRESH_PERIOD;
1141 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1143 return DEFAULT_SYNC_RETRIES & 3;
1146 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1148 return DEFAULT_SYNC_VER;
1151 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1153 return DEFAULT_SLOPPY_TCP;
1156 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1158 return DEFAULT_SLOPPY_SCTP;
1161 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1166 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1171 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1173 return IPVS_SYNC_QLEN_MAX;
1176 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1181 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1186 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1191 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1196 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1203 /* IPVS core functions
1204 * (from ip_vs_core.c)
1206 const char *ip_vs_proto_name(unsigned int proto);
1207 void ip_vs_init_hash_table(struct list_head *table, int rows);
1208 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1210 #define IP_VS_APP_TYPE_FTP 1
1212 /* ip_vs_conn handling functions
1213 * (from ip_vs_conn.c)
1216 IP_VS_DIR_INPUT = 0,
1218 IP_VS_DIR_INPUT_ONLY,
1222 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
1223 const union nf_inet_addr *caddr,
1225 const union nf_inet_addr *vaddr,
1227 struct ip_vs_conn_param *p)
1231 p->protocol = protocol;
1240 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1241 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1243 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
1244 const struct ip_vs_iphdr *iph);
1246 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1248 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1249 const struct ip_vs_iphdr *iph);
1251 /* Get reference to gain full access to conn.
1252 * By default, RCU read-side critical sections have access only to
1253 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1255 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1257 return atomic_inc_not_zero(&cp->refcnt);
1260 /* put back the conn without restarting its timer */
1261 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1263 smp_mb__before_atomic();
1264 atomic_dec(&cp->refcnt);
1266 void ip_vs_conn_put(struct ip_vs_conn *cp);
1267 void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1269 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, int dest_af,
1270 const union nf_inet_addr *daddr,
1271 __be16 dport, unsigned int flags,
1272 struct ip_vs_dest *dest, __u32 fwmark);
1273 void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1275 const char *ip_vs_state_name(__u16 proto, int state);
1277 void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1278 int ip_vs_check_template(struct ip_vs_conn *ct);
1279 void ip_vs_random_dropentry(struct net *net);
1280 int ip_vs_conn_init(void);
1281 void ip_vs_conn_cleanup(void);
1283 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1285 struct ip_vs_conn *ctl_cp = cp->control;
1287 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1289 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1291 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1297 IP_VS_DBG_BUF(7, "DELeting control for: "
1298 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1299 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1301 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1302 ntohs(ctl_cp->cport));
1305 if (atomic_read(&ctl_cp->n_control) == 0) {
1306 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1308 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1310 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1315 atomic_dec(&ctl_cp->n_control);
1319 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1322 IP_VS_ERR_BUF("request control ADD for already controlled: "
1324 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1326 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1329 ip_vs_control_del(cp);
1332 IP_VS_DBG_BUF(7, "ADDing control for: "
1333 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1334 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1336 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1337 ntohs(ctl_cp->cport));
1339 cp->control = ctl_cp;
1340 atomic_inc(&ctl_cp->n_control);
1343 /* IPVS netns init & cleanup functions */
1344 int ip_vs_estimator_net_init(struct net *net);
1345 int ip_vs_control_net_init(struct net *net);
1346 int ip_vs_protocol_net_init(struct net *net);
1347 int ip_vs_app_net_init(struct net *net);
1348 int ip_vs_conn_net_init(struct net *net);
1349 int ip_vs_sync_net_init(struct net *net);
1350 void ip_vs_conn_net_cleanup(struct net *net);
1351 void ip_vs_app_net_cleanup(struct net *net);
1352 void ip_vs_protocol_net_cleanup(struct net *net);
1353 void ip_vs_control_net_cleanup(struct net *net);
1354 void ip_vs_estimator_net_cleanup(struct net *net);
1355 void ip_vs_sync_net_cleanup(struct net *net);
1356 void ip_vs_service_net_cleanup(struct net *net);
1358 /* IPVS application functions
1359 * (from ip_vs_app.c)
1361 #define IP_VS_APP_MAX_PORTS 8
1362 struct ip_vs_app *register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1363 void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1364 int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1365 void ip_vs_unbind_app(struct ip_vs_conn *cp);
1366 int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app, __u16 proto,
1368 int ip_vs_app_inc_get(struct ip_vs_app *inc);
1369 void ip_vs_app_inc_put(struct ip_vs_app *inc);
1371 int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1372 int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1374 int register_ip_vs_pe(struct ip_vs_pe *pe);
1375 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1376 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1377 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1379 /* Use a #define to avoid all of module.h just for these trivial ops */
1380 #define ip_vs_pe_get(pe) \
1381 if (pe && pe->module) \
1382 __module_get(pe->module);
1384 #define ip_vs_pe_put(pe) \
1385 if (pe && pe->module) \
1386 module_put(pe->module);
1388 /* IPVS protocol functions (from ip_vs_proto.c) */
1389 int ip_vs_protocol_init(void);
1390 void ip_vs_protocol_cleanup(void);
1391 void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1392 int *ip_vs_create_timeout_table(int *table, int size);
1393 int ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1394 const char *name, int to);
1395 void ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1396 const struct sk_buff *skb, int offset,
1399 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1400 extern struct ip_vs_protocol ip_vs_protocol_udp;
1401 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1402 extern struct ip_vs_protocol ip_vs_protocol_esp;
1403 extern struct ip_vs_protocol ip_vs_protocol_ah;
1404 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1406 /* Registering/unregistering scheduler functions
1407 * (from ip_vs_sched.c)
1409 int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1410 int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1411 int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1412 struct ip_vs_scheduler *scheduler);
1413 void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1414 struct ip_vs_scheduler *sched);
1415 struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1416 void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1418 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1419 struct ip_vs_proto_data *pd, int *ignored,
1420 struct ip_vs_iphdr *iph);
1421 int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1422 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1424 void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1426 /* IPVS control data and functions (from ip_vs_ctl.c) */
1427 extern struct ip_vs_stats ip_vs_stats;
1428 extern int sysctl_ip_vs_sync_ver;
1430 struct ip_vs_service *
1431 ip_vs_service_find(struct net *net, int af, __u32 fwmark, __u16 protocol,
1432 const union nf_inet_addr *vaddr, __be16 vport);
1434 bool ip_vs_has_real_service(struct net *net, int af, __u16 protocol,
1435 const union nf_inet_addr *daddr, __be16 dport);
1437 int ip_vs_use_count_inc(void);
1438 void ip_vs_use_count_dec(void);
1439 int ip_vs_register_nl_ioctl(void);
1440 void ip_vs_unregister_nl_ioctl(void);
1441 int ip_vs_control_init(void);
1442 void ip_vs_control_cleanup(void);
1444 ip_vs_find_dest(struct net *net, int svc_af, int dest_af,
1445 const union nf_inet_addr *daddr, __be16 dport,
1446 const union nf_inet_addr *vaddr, __be16 vport,
1447 __u16 protocol, __u32 fwmark, __u32 flags);
1448 void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1450 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1452 atomic_inc(&dest->refcnt);
1455 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1457 smp_mb__before_atomic();
1458 atomic_dec(&dest->refcnt);
1461 static inline void ip_vs_dest_put_and_free(struct ip_vs_dest *dest)
1463 if (atomic_dec_return(&dest->refcnt) < 0)
1467 /* IPVS sync daemon data and function prototypes
1468 * (from ip_vs_sync.c)
1470 int start_sync_thread(struct net *net, struct ipvs_sync_daemon_cfg *cfg,
1472 int stop_sync_thread(struct net *net, int state);
1473 void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp, int pkts);
1475 /* IPVS rate estimator prototypes (from ip_vs_est.c) */
1476 void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1477 void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1478 void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1479 void ip_vs_read_estimator(struct ip_vs_kstats *dst, struct ip_vs_stats *stats);
1481 /* Various IPVS packet transmitters (from ip_vs_xmit.c) */
1482 int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1483 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1484 int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1485 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1486 int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1487 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1488 int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1489 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1490 int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1491 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1492 int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1493 struct ip_vs_protocol *pp, int offset,
1494 unsigned int hooknum, struct ip_vs_iphdr *iph);
1495 void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1497 #ifdef CONFIG_IP_VS_IPV6
1498 int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1499 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1500 int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1501 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1502 int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1503 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1504 int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1505 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1506 int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1507 struct ip_vs_protocol *pp, int offset,
1508 unsigned int hooknum, struct ip_vs_iphdr *iph);
1511 #ifdef CONFIG_SYSCTL
1512 /* This is a simple mechanism to ignore packets when
1513 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1514 * we start to drop 1/rate of the packets
1516 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1518 if (!ipvs->drop_rate)
1520 if (--ipvs->drop_counter > 0)
1522 ipvs->drop_counter = ipvs->drop_rate;
1526 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1529 /* ip_vs_fwd_tag returns the forwarding tag of the connection */
1530 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1532 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1536 switch (IP_VS_FWD_METHOD(cp)) {
1537 case IP_VS_CONN_F_MASQ:
1539 case IP_VS_CONN_F_LOCALNODE:
1541 case IP_VS_CONN_F_TUNNEL:
1543 case IP_VS_CONN_F_DROUTE:
1545 case IP_VS_CONN_F_BYPASS:
1553 void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1554 struct ip_vs_conn *cp, int dir);
1556 #ifdef CONFIG_IP_VS_IPV6
1557 void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1558 struct ip_vs_conn *cp, int dir);
1561 __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1563 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1565 __be32 diff[2] = { ~old, new };
1567 return csum_partial(diff, sizeof(diff), oldsum);
1570 #ifdef CONFIG_IP_VS_IPV6
1571 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1574 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1575 new[3], new[2], new[1], new[0] };
1577 return csum_partial(diff, sizeof(diff), oldsum);
1581 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1583 __be16 diff[2] = { ~old, new };
1585 return csum_partial(diff, sizeof(diff), oldsum);
1588 /* 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
1605 /* Netfilter connection tracking
1606 * (from ip_vs_nfct.c)
1608 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1610 #ifdef CONFIG_SYSCTL
1611 return ipvs->sysctl_conntrack;
1617 void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1619 int ip_vs_confirm_conntrack(struct sk_buff *skb);
1620 void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1621 struct ip_vs_conn *cp, u_int8_t proto,
1622 const __be16 port, int from_rs);
1623 void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1627 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1632 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1633 struct ip_vs_conn *cp, int outin)
1637 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1642 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1645 #endif /* CONFIG_IP_VS_NFCT */
1648 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1650 /* We think the overhead of processing active connections is 256
1651 * times higher than that of inactive connections in average. (This
1652 * 256 times might not be accurate, we will change it later) We
1653 * use the following formula to estimate the overhead now:
1654 * dest->activeconns*256 + dest->inactconns
1656 return (atomic_read(&dest->activeconns) << 8) +
1657 atomic_read(&dest->inactconns);
1660 #endif /* _NET_IP_VS_H */