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 netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
488 void (*exit_netns)(struct netns_ipvs *ipvs, 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)(struct netns_ipvs *ipvs,
498 const struct sk_buff *skb,
499 const struct ip_vs_iphdr *iph);
502 (*conn_out_get)(struct netns_ipvs *ipvs,
504 const struct sk_buff *skb,
505 const struct ip_vs_iphdr *iph);
507 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
508 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
510 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
511 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
513 int (*csum_check)(int af, struct sk_buff *skb,
514 struct ip_vs_protocol *pp);
516 const char *(*state_name)(int state);
518 void (*state_transition)(struct ip_vs_conn *cp, int direction,
519 const struct sk_buff *skb,
520 struct ip_vs_proto_data *pd);
522 int (*register_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
524 void (*unregister_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
526 int (*app_conn_bind)(struct ip_vs_conn *cp);
528 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
529 const struct sk_buff *skb,
533 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
536 /* protocol data per netns */
537 struct ip_vs_proto_data {
538 struct ip_vs_proto_data *next;
539 struct ip_vs_protocol *pp;
540 int *timeout_table; /* protocol timeout table */
541 atomic_t appcnt; /* counter of proto app incs. */
542 struct tcp_states_t *tcp_state_table;
545 struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
546 struct ip_vs_proto_data *ip_vs_proto_data_get(struct netns_ipvs *ipvs,
547 unsigned short proto);
549 struct ip_vs_conn_param {
550 struct netns_ipvs *ipvs;
551 const union nf_inet_addr *caddr;
552 const union nf_inet_addr *vaddr;
558 const struct ip_vs_pe *pe;
563 /* IP_VS structure allocated for each dynamically scheduled connection */
565 struct hlist_node c_list; /* hashed list heads */
566 /* Protocol, addresses and port numbers */
570 u16 af; /* address family */
571 union nf_inet_addr caddr; /* client address */
572 union nf_inet_addr vaddr; /* virtual address */
573 union nf_inet_addr daddr; /* destination address */
574 volatile __u32 flags; /* status flags */
575 __u16 protocol; /* Which protocol (TCP/UDP) */
576 __u16 daf; /* Address family of the dest */
577 struct netns_ipvs *ipvs;
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 /* Extended internal versions of struct ip_vs_service_user and ip_vs_dest_user
627 * We need these to conveniently pass around service and destination
628 * options, but unfortunately, we also need to keep the old definitions to
629 * maintain userspace backwards compatibility for the setsockopt interface.
631 struct ip_vs_service_user_kern {
632 /* virtual service addresses */
635 union nf_inet_addr addr; /* virtual ip address */
637 u32 fwmark; /* firwall mark of service */
639 /* virtual service options */
642 unsigned int flags; /* virtual service flags */
643 unsigned int timeout; /* persistent timeout in sec */
644 __be32 netmask; /* persistent netmask or plen */
648 struct ip_vs_dest_user_kern {
649 /* destination server address */
650 union nf_inet_addr addr;
653 /* real server options */
654 unsigned int conn_flags; /* connection flags */
655 int weight; /* destination weight */
657 /* thresholds for active connections */
658 u32 u_threshold; /* upper threshold */
659 u32 l_threshold; /* lower threshold */
661 /* Address family of addr */
667 * The information about the virtual service offered to the net and the
668 * forwarding entries.
670 struct ip_vs_service {
671 struct hlist_node s_list; /* for normal service table */
672 struct hlist_node f_list; /* for fwmark-based service table */
673 atomic_t refcnt; /* reference counter */
675 u16 af; /* address family */
676 __u16 protocol; /* which protocol (TCP/UDP) */
677 union nf_inet_addr addr; /* IP address for virtual service */
678 __be16 port; /* port number for the service */
679 __u32 fwmark; /* firewall mark of the service */
680 unsigned int flags; /* service status flags */
681 unsigned int timeout; /* persistent timeout in ticks */
682 __be32 netmask; /* grouping granularity, mask/plen */
683 struct netns_ipvs *ipvs;
685 struct list_head destinations; /* real server d-linked list */
686 __u32 num_dests; /* number of servers */
687 struct ip_vs_stats stats; /* statistics for the service */
690 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
691 spinlock_t sched_lock; /* lock sched_data */
692 void *sched_data; /* scheduler application data */
694 /* alternate persistence engine */
695 struct ip_vs_pe __rcu *pe;
697 struct rcu_head rcu_head;
700 /* Information for cached dst */
701 struct ip_vs_dest_dst {
702 struct dst_entry *dst_cache; /* destination cache entry */
704 union nf_inet_addr dst_saddr;
705 struct rcu_head rcu_head;
708 /* The real server destination forwarding entry with ip address, port number,
712 struct list_head n_list; /* for the dests in the service */
713 struct hlist_node d_list; /* for table with all the dests */
715 u16 af; /* address family */
716 __be16 port; /* port number of the server */
717 union nf_inet_addr addr; /* IP address of the server */
718 volatile unsigned int flags; /* dest status flags */
719 atomic_t conn_flags; /* flags to copy to conn */
720 atomic_t weight; /* server weight */
722 atomic_t refcnt; /* reference counter */
723 struct ip_vs_stats stats; /* statistics */
724 unsigned long idle_start; /* start time, jiffies */
726 /* connection counters and thresholds */
727 atomic_t activeconns; /* active connections */
728 atomic_t inactconns; /* inactive connections */
729 atomic_t persistconns; /* persistent connections */
730 __u32 u_threshold; /* upper threshold */
731 __u32 l_threshold; /* lower threshold */
733 /* for destination cache */
734 spinlock_t dst_lock; /* lock of dst_cache */
735 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
737 /* for virtual service */
738 struct ip_vs_service __rcu *svc; /* service it belongs to */
739 __u16 protocol; /* which protocol (TCP/UDP) */
740 __be16 vport; /* virtual port number */
741 union nf_inet_addr vaddr; /* virtual IP address */
742 __u32 vfwmark; /* firewall mark of service */
744 struct list_head t_list; /* in dest_trash */
745 unsigned int in_rs_table:1; /* we are in rs_table */
748 /* The scheduler object */
749 struct ip_vs_scheduler {
750 struct list_head n_list; /* d-linked list head */
751 char *name; /* scheduler name */
752 atomic_t refcnt; /* reference counter */
753 struct module *module; /* THIS_MODULE/NULL */
755 /* scheduler initializing service */
756 int (*init_service)(struct ip_vs_service *svc);
757 /* scheduling service finish */
758 void (*done_service)(struct ip_vs_service *svc);
760 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
761 /* dest is unlinked */
762 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
763 /* dest is updated */
764 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
766 /* selecting a server from the given service */
767 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
768 const struct sk_buff *skb,
769 struct ip_vs_iphdr *iph);
772 /* The persistence engine object */
774 struct list_head n_list; /* d-linked list head */
775 char *name; /* scheduler name */
776 atomic_t refcnt; /* reference counter */
777 struct module *module; /* THIS_MODULE/NULL */
779 /* get the connection template, if any */
780 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
781 bool (*ct_match)(const struct ip_vs_conn_param *p,
782 struct ip_vs_conn *ct);
783 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
785 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
788 /* The application module object (a.k.a. app incarnation) */
790 struct list_head a_list; /* member in app list */
791 int type; /* IP_VS_APP_TYPE_xxx */
792 char *name; /* application module name */
794 struct module *module; /* THIS_MODULE/NULL */
795 struct list_head incs_list; /* list of incarnations */
797 /* members for application incarnations */
798 struct list_head p_list; /* member in proto app list */
799 struct ip_vs_app *app; /* its real application */
800 __be16 port; /* port number in net order */
801 atomic_t usecnt; /* usage counter */
802 struct rcu_head rcu_head;
804 /* output hook: Process packet in inout direction, diff set for TCP.
805 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
806 * 2=Mangled but checksum was not updated
808 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
809 struct sk_buff *, int *diff);
811 /* input hook: Process packet in outin direction, diff set for TCP.
812 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
813 * 2=Mangled but checksum was not updated
815 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
816 struct sk_buff *, int *diff);
818 /* ip_vs_app initializer */
819 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
821 /* ip_vs_app finish */
822 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
826 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
827 struct ip_vs_protocol *);
829 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
835 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
836 int *verdict, struct ip_vs_conn **cpp);
839 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
840 const struct iphdr *iph, int inverse);
843 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
844 const struct iphdr *iph, int inverse);
846 int (*state_transition)(struct ip_vs_conn *cp, int direction,
847 const struct sk_buff *skb,
848 struct ip_vs_app *app);
850 void (*timeout_change)(struct ip_vs_app *app, int flags);
853 struct ipvs_master_sync_state {
854 struct list_head sync_queue;
855 struct ip_vs_sync_buff *sync_buff;
856 unsigned long sync_queue_len;
857 unsigned int sync_queue_delay;
858 struct task_struct *master_thread;
859 struct delayed_work master_wakeup_work;
860 struct netns_ipvs *ipvs;
863 /* How much time to keep dests in trash */
864 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
866 struct ipvs_sync_daemon_cfg {
867 union nf_inet_addr mcast_group;
873 /* multicast interface name */
874 char mcast_ifn[IP_VS_IFNAME_MAXLEN];
877 /* IPVS in network namespace */
879 int gen; /* Generation */
880 int enable; /* enable like nf_hooks do */
881 /* Hash table: for real service lookups */
882 #define IP_VS_RTAB_BITS 4
883 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
884 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
886 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
888 struct list_head app_list;
890 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
891 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
892 /* ip_vs_proto_tcp */
893 #ifdef CONFIG_IP_VS_PROTO_TCP
894 #define TCP_APP_TAB_BITS 4
895 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
896 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
897 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
899 /* ip_vs_proto_udp */
900 #ifdef CONFIG_IP_VS_PROTO_UDP
901 #define UDP_APP_TAB_BITS 4
902 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
903 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
904 struct list_head udp_apps[UDP_APP_TAB_SIZE];
906 /* ip_vs_proto_sctp */
907 #ifdef CONFIG_IP_VS_PROTO_SCTP
908 #define SCTP_APP_TAB_BITS 4
909 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
910 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
911 /* Hash table for SCTP application incarnations */
912 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
915 atomic_t conn_count; /* connection counter */
918 struct ip_vs_stats tot_stats; /* Statistics & est. */
920 int num_services; /* no of virtual services */
922 /* Trash for destinations */
923 struct list_head dest_trash;
924 spinlock_t dest_trash_lock;
925 struct timer_list dest_trash_timer; /* expiration timer */
926 /* Service counters */
927 atomic_t ftpsvc_counter;
928 atomic_t nullsvc_counter;
931 /* 1/rate drop and drop-entry variables */
932 struct delayed_work defense_work; /* Work handler */
937 spinlock_t dropentry_lock; /* drop entry handling */
938 spinlock_t droppacket_lock; /* drop packet handling */
939 spinlock_t securetcp_lock; /* state and timeout tables */
942 struct ctl_table_header *sysctl_hdr;
943 struct ctl_table *sysctl_tbl;
946 /* sysctl variables */
947 int sysctl_amemthresh;
948 int sysctl_am_droprate;
949 int sysctl_drop_entry;
950 int sysctl_drop_packet;
951 int sysctl_secure_tcp;
952 #ifdef CONFIG_IP_VS_NFCT
953 int sysctl_conntrack;
955 int sysctl_snat_reroute;
957 int sysctl_sync_ports;
958 int sysctl_sync_persist_mode;
959 unsigned long sysctl_sync_qlen_max;
960 int sysctl_sync_sock_size;
961 int sysctl_cache_bypass;
962 int sysctl_expire_nodest_conn;
963 int sysctl_sloppy_tcp;
964 int sysctl_sloppy_sctp;
965 int sysctl_expire_quiescent_template;
966 int sysctl_sync_threshold[2];
967 unsigned int sysctl_sync_refresh_period;
968 int sysctl_sync_retries;
969 int sysctl_nat_icmp_send;
970 int sysctl_pmtu_disc;
971 int sysctl_backup_only;
972 int sysctl_conn_reuse_mode;
973 int sysctl_schedule_icmp;
974 int sysctl_ignore_tunneled;
977 int sysctl_lblc_expiration;
978 struct ctl_table_header *lblc_ctl_header;
979 struct ctl_table *lblc_ctl_table;
981 int sysctl_lblcr_expiration;
982 struct ctl_table_header *lblcr_ctl_header;
983 struct ctl_table *lblcr_ctl_table;
985 struct list_head est_list; /* estimator list */
987 struct timer_list est_timer; /* Estimation timer */
989 spinlock_t sync_lock;
990 struct ipvs_master_sync_state *ms;
991 spinlock_t sync_buff_lock;
992 struct task_struct **backup_threads;
994 volatile int sync_state;
995 struct mutex sync_mutex;
996 struct ipvs_sync_daemon_cfg mcfg; /* Master Configuration */
997 struct ipvs_sync_daemon_cfg bcfg; /* Backup Configuration */
998 /* net name space ptr */
999 struct net *net; /* Needed by timer routines */
1000 /* Number of heterogeneous destinations, needed becaus heterogeneous
1001 * are not supported when synchronization is enabled.
1003 unsigned int mixed_address_family_dests;
1006 #define DEFAULT_SYNC_THRESHOLD 3
1007 #define DEFAULT_SYNC_PERIOD 50
1008 #define DEFAULT_SYNC_VER 1
1009 #define DEFAULT_SLOPPY_TCP 0
1010 #define DEFAULT_SLOPPY_SCTP 0
1011 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1012 #define DEFAULT_SYNC_RETRIES 0
1013 #define IPVS_SYNC_WAKEUP_RATE 8
1014 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1015 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1016 #define IPVS_SYNC_CHECK_PERIOD HZ
1017 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1018 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1020 #ifdef CONFIG_SYSCTL
1022 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1024 return ipvs->sysctl_sync_threshold[0];
1027 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1029 return ACCESS_ONCE(ipvs->sysctl_sync_threshold[1]);
1032 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1034 return ACCESS_ONCE(ipvs->sysctl_sync_refresh_period);
1037 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1039 return ipvs->sysctl_sync_retries;
1042 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1044 return ipvs->sysctl_sync_ver;
1047 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1049 return ipvs->sysctl_sloppy_tcp;
1052 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1054 return ipvs->sysctl_sloppy_sctp;
1057 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1059 return ACCESS_ONCE(ipvs->sysctl_sync_ports);
1062 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1064 return ipvs->sysctl_sync_persist_mode;
1067 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1069 return ipvs->sysctl_sync_qlen_max;
1072 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1074 return ipvs->sysctl_sync_sock_size;
1077 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1079 return ipvs->sysctl_pmtu_disc;
1082 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1084 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1085 ipvs->sysctl_backup_only;
1088 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1090 return ipvs->sysctl_conn_reuse_mode;
1093 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1095 return ipvs->sysctl_schedule_icmp;
1098 static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1100 return ipvs->sysctl_ignore_tunneled;
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)
1170 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1175 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1180 static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1187 /* IPVS core functions
1188 * (from ip_vs_core.c)
1190 const char *ip_vs_proto_name(unsigned int proto);
1191 void ip_vs_init_hash_table(struct list_head *table, int rows);
1192 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1194 #define IP_VS_APP_TYPE_FTP 1
1196 /* ip_vs_conn handling functions
1197 * (from ip_vs_conn.c)
1200 IP_VS_DIR_INPUT = 0,
1202 IP_VS_DIR_INPUT_ONLY,
1206 static inline void ip_vs_conn_fill_param(struct netns_ipvs *ipvs, int af, int protocol,
1207 const union nf_inet_addr *caddr,
1209 const union nf_inet_addr *vaddr,
1211 struct ip_vs_conn_param *p)
1215 p->protocol = protocol;
1224 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1225 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1227 struct ip_vs_conn * ip_vs_conn_in_get_proto(struct netns_ipvs *ipvs, int af,
1228 const struct sk_buff *skb,
1229 const struct ip_vs_iphdr *iph);
1231 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1233 struct ip_vs_conn * ip_vs_conn_out_get_proto(struct netns_ipvs *ipvs, int af,
1234 const struct sk_buff *skb,
1235 const struct ip_vs_iphdr *iph);
1237 /* Get reference to gain full access to conn.
1238 * By default, RCU read-side critical sections have access only to
1239 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1241 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1243 return atomic_inc_not_zero(&cp->refcnt);
1246 /* put back the conn without restarting its timer */
1247 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1249 smp_mb__before_atomic();
1250 atomic_dec(&cp->refcnt);
1252 void ip_vs_conn_put(struct ip_vs_conn *cp);
1253 void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1255 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, int dest_af,
1256 const union nf_inet_addr *daddr,
1257 __be16 dport, unsigned int flags,
1258 struct ip_vs_dest *dest, __u32 fwmark);
1259 void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1261 const char *ip_vs_state_name(__u16 proto, int state);
1263 void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1264 int ip_vs_check_template(struct ip_vs_conn *ct);
1265 void ip_vs_random_dropentry(struct netns_ipvs *ipvs);
1266 int ip_vs_conn_init(void);
1267 void ip_vs_conn_cleanup(void);
1269 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1271 struct ip_vs_conn *ctl_cp = cp->control;
1273 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1275 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1277 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1283 IP_VS_DBG_BUF(7, "DELeting control for: "
1284 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1285 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1287 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1288 ntohs(ctl_cp->cport));
1291 if (atomic_read(&ctl_cp->n_control) == 0) {
1292 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1294 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1296 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1301 atomic_dec(&ctl_cp->n_control);
1305 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1308 IP_VS_ERR_BUF("request control ADD for already controlled: "
1310 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1312 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1315 ip_vs_control_del(cp);
1318 IP_VS_DBG_BUF(7, "ADDing control for: "
1319 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1320 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1322 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1323 ntohs(ctl_cp->cport));
1325 cp->control = ctl_cp;
1326 atomic_inc(&ctl_cp->n_control);
1329 /* IPVS netns init & cleanup functions */
1330 int ip_vs_estimator_net_init(struct netns_ipvs *ipvs);
1331 int ip_vs_control_net_init(struct netns_ipvs *ipvs);
1332 int ip_vs_protocol_net_init(struct net *net);
1333 int ip_vs_app_net_init(struct netns_ipvs *ipvs);
1334 int ip_vs_conn_net_init(struct netns_ipvs *ipvs);
1335 int ip_vs_sync_net_init(struct netns_ipvs *ipvs);
1336 void ip_vs_conn_net_cleanup(struct netns_ipvs *ipvs);
1337 void ip_vs_app_net_cleanup(struct netns_ipvs *ipvs);
1338 void ip_vs_protocol_net_cleanup(struct net *net);
1339 void ip_vs_control_net_cleanup(struct netns_ipvs *ipvs);
1340 void ip_vs_estimator_net_cleanup(struct netns_ipvs *ipvs);
1341 void ip_vs_sync_net_cleanup(struct netns_ipvs *ipvs);
1342 void ip_vs_service_net_cleanup(struct netns_ipvs *ipvs);
1344 /* IPVS application functions
1345 * (from ip_vs_app.c)
1347 #define IP_VS_APP_MAX_PORTS 8
1348 struct ip_vs_app *register_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1349 void unregister_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1350 int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1351 void ip_vs_unbind_app(struct ip_vs_conn *cp);
1352 int register_ip_vs_app_inc(struct netns_ipvs *ipvs, struct ip_vs_app *app, __u16 proto,
1354 int ip_vs_app_inc_get(struct ip_vs_app *inc);
1355 void ip_vs_app_inc_put(struct ip_vs_app *inc);
1357 int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1358 int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1360 int register_ip_vs_pe(struct ip_vs_pe *pe);
1361 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1362 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1363 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1365 /* Use a #define to avoid all of module.h just for these trivial ops */
1366 #define ip_vs_pe_get(pe) \
1367 if (pe && pe->module) \
1368 __module_get(pe->module);
1370 #define ip_vs_pe_put(pe) \
1371 if (pe && pe->module) \
1372 module_put(pe->module);
1374 /* IPVS protocol functions (from ip_vs_proto.c) */
1375 int ip_vs_protocol_init(void);
1376 void ip_vs_protocol_cleanup(void);
1377 void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1378 int *ip_vs_create_timeout_table(int *table, int size);
1379 int ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1380 const char *name, int to);
1381 void ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1382 const struct sk_buff *skb, int offset,
1385 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1386 extern struct ip_vs_protocol ip_vs_protocol_udp;
1387 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1388 extern struct ip_vs_protocol ip_vs_protocol_esp;
1389 extern struct ip_vs_protocol ip_vs_protocol_ah;
1390 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1392 /* Registering/unregistering scheduler functions
1393 * (from ip_vs_sched.c)
1395 int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1396 int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1397 int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1398 struct ip_vs_scheduler *scheduler);
1399 void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1400 struct ip_vs_scheduler *sched);
1401 struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1402 void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1404 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1405 struct ip_vs_proto_data *pd, int *ignored,
1406 struct ip_vs_iphdr *iph);
1407 int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1408 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1410 void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1412 /* IPVS control data and functions (from ip_vs_ctl.c) */
1413 extern struct ip_vs_stats ip_vs_stats;
1414 extern int sysctl_ip_vs_sync_ver;
1416 struct ip_vs_service *
1417 ip_vs_service_find(struct netns_ipvs *ipvs, int af, __u32 fwmark, __u16 protocol,
1418 const union nf_inet_addr *vaddr, __be16 vport);
1420 bool ip_vs_has_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1421 const union nf_inet_addr *daddr, __be16 dport);
1423 int ip_vs_use_count_inc(void);
1424 void ip_vs_use_count_dec(void);
1425 int ip_vs_register_nl_ioctl(void);
1426 void ip_vs_unregister_nl_ioctl(void);
1427 int ip_vs_control_init(void);
1428 void ip_vs_control_cleanup(void);
1430 ip_vs_find_dest(struct netns_ipvs *ipvs, int svc_af, int dest_af,
1431 const union nf_inet_addr *daddr, __be16 dport,
1432 const union nf_inet_addr *vaddr, __be16 vport,
1433 __u16 protocol, __u32 fwmark, __u32 flags);
1434 void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1436 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1438 atomic_inc(&dest->refcnt);
1441 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1443 smp_mb__before_atomic();
1444 atomic_dec(&dest->refcnt);
1447 static inline void ip_vs_dest_put_and_free(struct ip_vs_dest *dest)
1449 if (atomic_dec_return(&dest->refcnt) < 0)
1453 /* IPVS sync daemon data and function prototypes
1454 * (from ip_vs_sync.c)
1456 int start_sync_thread(struct netns_ipvs *ipvs, struct ipvs_sync_daemon_cfg *cfg,
1458 int stop_sync_thread(struct netns_ipvs *ipvs, int state);
1459 void ip_vs_sync_conn(struct netns_ipvs *ipvs, struct ip_vs_conn *cp, int pkts);
1461 /* IPVS rate estimator prototypes (from ip_vs_est.c) */
1462 void ip_vs_start_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1463 void ip_vs_stop_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1464 void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1465 void ip_vs_read_estimator(struct ip_vs_kstats *dst, struct ip_vs_stats *stats);
1467 /* Various IPVS packet transmitters (from ip_vs_xmit.c) */
1468 int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1469 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1470 int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1471 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1472 int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1473 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1474 int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1475 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1476 int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1477 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1478 int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1479 struct ip_vs_protocol *pp, int offset,
1480 unsigned int hooknum, struct ip_vs_iphdr *iph);
1481 void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1483 #ifdef CONFIG_IP_VS_IPV6
1484 int ip_vs_bypass_xmit_v6(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_v6(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_v6(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_v6(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_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1493 struct ip_vs_protocol *pp, int offset,
1494 unsigned int hooknum, struct ip_vs_iphdr *iph);
1497 #ifdef CONFIG_SYSCTL
1498 /* This is a simple mechanism to ignore packets when
1499 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1500 * we start to drop 1/rate of the packets
1502 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1504 if (!ipvs->drop_rate)
1506 if (--ipvs->drop_counter > 0)
1508 ipvs->drop_counter = ipvs->drop_rate;
1512 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1515 /* ip_vs_fwd_tag returns the forwarding tag of the connection */
1516 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1518 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1522 switch (IP_VS_FWD_METHOD(cp)) {
1523 case IP_VS_CONN_F_MASQ:
1525 case IP_VS_CONN_F_LOCALNODE:
1527 case IP_VS_CONN_F_TUNNEL:
1529 case IP_VS_CONN_F_DROUTE:
1531 case IP_VS_CONN_F_BYPASS:
1539 void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1540 struct ip_vs_conn *cp, int dir);
1542 #ifdef CONFIG_IP_VS_IPV6
1543 void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1544 struct ip_vs_conn *cp, int dir);
1547 __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1549 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1551 __be32 diff[2] = { ~old, new };
1553 return csum_partial(diff, sizeof(diff), oldsum);
1556 #ifdef CONFIG_IP_VS_IPV6
1557 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1560 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1561 new[3], new[2], new[1], new[0] };
1563 return csum_partial(diff, sizeof(diff), oldsum);
1567 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1569 __be16 diff[2] = { ~old, new };
1571 return csum_partial(diff, sizeof(diff), oldsum);
1574 /* Forget current conntrack (unconfirmed) and attach notrack entry */
1575 static inline void ip_vs_notrack(struct sk_buff *skb)
1577 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1578 enum ip_conntrack_info ctinfo;
1579 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1581 if (!ct || !nf_ct_is_untracked(ct)) {
1582 nf_conntrack_put(skb->nfct);
1583 skb->nfct = &nf_ct_untracked_get()->ct_general;
1584 skb->nfctinfo = IP_CT_NEW;
1585 nf_conntrack_get(skb->nfct);
1590 #ifdef CONFIG_IP_VS_NFCT
1591 /* Netfilter connection tracking
1592 * (from ip_vs_nfct.c)
1594 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1596 #ifdef CONFIG_SYSCTL
1597 return ipvs->sysctl_conntrack;
1603 void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1605 int ip_vs_confirm_conntrack(struct sk_buff *skb);
1606 void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1607 struct ip_vs_conn *cp, u_int8_t proto,
1608 const __be16 port, int from_rs);
1609 void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1613 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1618 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1619 struct ip_vs_conn *cp, int outin)
1623 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1628 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1631 #endif /* CONFIG_IP_VS_NFCT */
1634 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1636 /* We think the overhead of processing active connections is 256
1637 * times higher than that of inactive connections in average. (This
1638 * 256 times might not be accurate, we will change it later) We
1639 * use the following formula to estimate the overhead now:
1640 * dest->activeconns*256 + dest->inactconns
1642 return (atomic_read(&dest->activeconns) << 8) +
1643 atomic_read(&dest->inactconns);
1646 #endif /* _NET_IP_VS_H */