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/sysctl.h> /* for ctl_path */
14 #include <linux/list.h> /* for struct list_head */
15 #include <linux/spinlock.h> /* for struct rwlock_t */
16 #include <asm/atomic.h> /* for struct atomic_t */
17 #include <linux/compiler.h>
18 #include <linux/timer.h>
20 #include <net/checksum.h>
21 #include <linux/netfilter.h> /* for union nf_inet_addr */
23 #include <linux/ipv6.h> /* for struct ipv6hdr */
24 #include <net/ipv6.h> /* for ipv6_addr_copy */
25 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
26 #include <net/netfilter/nf_conntrack.h>
28 #include <net/net_namespace.h> /* Netw namespace */
31 * Generic access of ipvs struct
33 static inline struct netns_ipvs *net_ipvs(struct net* net)
38 * Get net ptr from skb in traffic cases
39 * use skb_sknet when call is from userland (ioctl or netlink)
41 static inline struct net *skb_net(const struct sk_buff *skb)
44 #ifdef CONFIG_IP_VS_DEBUG
46 * This is used for debug only.
47 * Start with the most likely hit
50 if (likely(skb->dev && skb->dev->nd_net))
51 return dev_net(skb->dev);
52 if (skb_dst(skb) && skb_dst(skb)->dev)
53 return dev_net(skb_dst(skb)->dev);
54 WARN(skb->sk, "Maybe skb_sknet should be used in %s() at line:%d\n",
56 if (likely(skb->sk && skb->sk->sk_net))
57 return sock_net(skb->sk);
58 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
62 return dev_net(skb->dev ? : skb_dst(skb)->dev);
69 static inline struct net *skb_sknet(const struct sk_buff *skb)
72 #ifdef CONFIG_IP_VS_DEBUG
73 /* Start with the most likely hit */
74 if (likely(skb->sk && skb->sk->sk_net))
75 return sock_net(skb->sk);
76 WARN(skb->dev, "Maybe skb_net should be used instead in %s() line:%d\n",
78 if (likely(skb->dev && skb->dev->nd_net))
79 return dev_net(skb->dev);
80 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
84 return sock_net(skb->sk);
91 * This one needed for single_open_net since net is stored directly in
92 * private not as a struct i.e. seq_file_net can't be used.
94 static inline struct net *seq_file_single_net(struct seq_file *seq)
97 return (struct net *)seq->private;
103 /* Connections' size value needed by ip_vs_ctl.c */
104 extern int ip_vs_conn_tab_size;
110 union nf_inet_addr saddr;
111 union nf_inet_addr daddr;
115 ip_vs_fill_iphdr(int af, const void *nh, struct ip_vs_iphdr *iphdr)
117 #ifdef CONFIG_IP_VS_IPV6
118 if (af == AF_INET6) {
119 const struct ipv6hdr *iph = nh;
120 iphdr->len = sizeof(struct ipv6hdr);
121 iphdr->protocol = iph->nexthdr;
122 ipv6_addr_copy(&iphdr->saddr.in6, &iph->saddr);
123 ipv6_addr_copy(&iphdr->daddr.in6, &iph->daddr);
127 const struct iphdr *iph = nh;
128 iphdr->len = iph->ihl * 4;
129 iphdr->protocol = iph->protocol;
130 iphdr->saddr.ip = iph->saddr;
131 iphdr->daddr.ip = iph->daddr;
135 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
136 const union nf_inet_addr *src)
138 #ifdef CONFIG_IP_VS_IPV6
140 ipv6_addr_copy(&dst->in6, &src->in6);
146 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
147 const union nf_inet_addr *b)
149 #ifdef CONFIG_IP_VS_IPV6
151 return ipv6_addr_equal(&a->in6, &b->in6);
153 return a->ip == b->ip;
156 #ifdef CONFIG_IP_VS_DEBUG
157 #include <linux/net.h>
159 extern int ip_vs_get_debug_level(void);
161 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
162 const union nf_inet_addr *addr,
166 #ifdef CONFIG_IP_VS_IPV6
168 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6]",
172 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
176 BUG_ON(*idx > buf_len + 1);
177 return &buf[*idx - len];
180 #define IP_VS_DBG_BUF(level, msg, ...) \
182 char ip_vs_dbg_buf[160]; \
183 int ip_vs_dbg_idx = 0; \
184 if (level <= ip_vs_get_debug_level()) \
185 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
187 #define IP_VS_ERR_BUF(msg...) \
189 char ip_vs_dbg_buf[160]; \
190 int ip_vs_dbg_idx = 0; \
194 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
195 #define IP_VS_DBG_ADDR(af, addr) \
196 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
197 sizeof(ip_vs_dbg_buf), addr, \
200 #define IP_VS_DBG(level, msg, ...) \
202 if (level <= ip_vs_get_debug_level()) \
203 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
205 #define IP_VS_DBG_RL(msg, ...) \
207 if (net_ratelimit()) \
208 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
210 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
212 if (level <= ip_vs_get_debug_level()) \
213 pp->debug_packet(af, pp, skb, ofs, msg); \
215 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
217 if (level <= ip_vs_get_debug_level() && \
219 pp->debug_packet(af, pp, skb, ofs, msg); \
221 #else /* NO DEBUGGING at ALL */
222 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
223 #define IP_VS_ERR_BUF(msg...) do {} while (0)
224 #define IP_VS_DBG(level, msg...) do {} while (0)
225 #define IP_VS_DBG_RL(msg...) do {} while (0)
226 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
227 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
230 #define IP_VS_BUG() BUG()
231 #define IP_VS_ERR_RL(msg, ...) \
233 if (net_ratelimit()) \
234 pr_err(msg, ##__VA_ARGS__); \
237 #ifdef CONFIG_IP_VS_DEBUG
238 #define EnterFunction(level) \
240 if (level <= ip_vs_get_debug_level()) \
242 pr_fmt("Enter: %s, %s line %i\n"), \
243 __func__, __FILE__, __LINE__); \
245 #define LeaveFunction(level) \
247 if (level <= ip_vs_get_debug_level()) \
249 pr_fmt("Leave: %s, %s line %i\n"), \
250 __func__, __FILE__, __LINE__); \
253 #define EnterFunction(level) do {} while (0)
254 #define LeaveFunction(level) do {} while (0)
257 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
261 * The port number of FTP service (in network order).
263 #define FTPPORT cpu_to_be16(21)
264 #define FTPDATA cpu_to_be16(20)
270 IP_VS_TCP_S_NONE = 0,
271 IP_VS_TCP_S_ESTABLISHED,
272 IP_VS_TCP_S_SYN_SENT,
273 IP_VS_TCP_S_SYN_RECV,
274 IP_VS_TCP_S_FIN_WAIT,
275 IP_VS_TCP_S_TIME_WAIT,
277 IP_VS_TCP_S_CLOSE_WAIT,
278 IP_VS_TCP_S_LAST_ACK,
303 enum ip_vs_sctp_states {
305 IP_VS_SCTP_S_INIT_CLI,
306 IP_VS_SCTP_S_INIT_SER,
307 IP_VS_SCTP_S_INIT_ACK_CLI,
308 IP_VS_SCTP_S_INIT_ACK_SER,
309 IP_VS_SCTP_S_ECHO_CLI,
310 IP_VS_SCTP_S_ECHO_SER,
311 IP_VS_SCTP_S_ESTABLISHED,
312 IP_VS_SCTP_S_SHUT_CLI,
313 IP_VS_SCTP_S_SHUT_SER,
314 IP_VS_SCTP_S_SHUT_ACK_CLI,
315 IP_VS_SCTP_S_SHUT_ACK_SER,
321 * Delta sequence info structure
322 * Each ip_vs_conn has 2 (output AND input seq. changes).
323 * Only used in the VS/NAT.
326 __u32 init_seq; /* Add delta from this seq */
327 __u32 delta; /* Delta in sequence numbers */
328 __u32 previous_delta; /* Delta in sequence numbers
329 before last resized pkt */
335 struct ip_vs_counters {
336 __u32 conns; /* connections scheduled */
337 __u32 inpkts; /* incoming packets */
338 __u32 outpkts; /* outgoing packets */
339 __u64 inbytes; /* incoming bytes */
340 __u64 outbytes; /* outgoing bytes */
345 struct ip_vs_cpu_stats {
346 struct ip_vs_counters ustats;
347 struct u64_stats_sync syncp;
351 * IPVS statistics objects
353 struct ip_vs_estimator {
354 struct list_head list;
370 struct ip_vs_stats_user ustats; /* statistics */
371 struct ip_vs_estimator est; /* estimator */
372 struct ip_vs_cpu_stats *cpustats; /* per cpu counters */
373 spinlock_t lock; /* spin lock */
374 struct ip_vs_stats_user ustats0; /* reset values */
382 struct ip_vs_proto_data;
384 struct ip_vs_protocol {
385 struct ip_vs_protocol *next;
391 void (*init)(struct ip_vs_protocol *pp);
393 void (*exit)(struct ip_vs_protocol *pp);
395 void (*init_netns)(struct net *net, struct ip_vs_proto_data *pd);
397 void (*exit_netns)(struct net *net, struct ip_vs_proto_data *pd);
399 int (*conn_schedule)(int af, struct sk_buff *skb,
400 struct ip_vs_proto_data *pd,
401 int *verdict, struct ip_vs_conn **cpp);
404 (*conn_in_get)(int af,
405 const struct sk_buff *skb,
406 const struct ip_vs_iphdr *iph,
407 unsigned int proto_off,
411 (*conn_out_get)(int af,
412 const struct sk_buff *skb,
413 const struct ip_vs_iphdr *iph,
414 unsigned int proto_off,
417 int (*snat_handler)(struct sk_buff *skb,
418 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
420 int (*dnat_handler)(struct sk_buff *skb,
421 struct ip_vs_protocol *pp, struct ip_vs_conn *cp);
423 int (*csum_check)(int af, struct sk_buff *skb,
424 struct ip_vs_protocol *pp);
426 const char *(*state_name)(int state);
428 int (*state_transition)(struct ip_vs_conn *cp, int direction,
429 const struct sk_buff *skb,
430 struct ip_vs_proto_data *pd);
432 int (*register_app)(struct net *net, struct ip_vs_app *inc);
434 void (*unregister_app)(struct net *net, struct ip_vs_app *inc);
436 int (*app_conn_bind)(struct ip_vs_conn *cp);
438 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
439 const struct sk_buff *skb,
443 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
447 * protocol data per netns
449 struct ip_vs_proto_data {
450 struct ip_vs_proto_data *next;
451 struct ip_vs_protocol *pp;
452 int *timeout_table; /* protocol timeout table */
453 atomic_t appcnt; /* counter of proto app incs. */
454 struct tcp_states_t *tcp_state_table;
457 extern struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
458 extern struct ip_vs_proto_data *ip_vs_proto_data_get(struct net *net,
459 unsigned short proto);
461 struct ip_vs_conn_param {
463 const union nf_inet_addr *caddr;
464 const union nf_inet_addr *vaddr;
470 const struct ip_vs_pe *pe;
476 * IP_VS structure allocated for each dynamically scheduled connection
479 struct hlist_node c_list; /* hashed list heads */
481 struct net *net; /* Name space */
483 /* Protocol, addresses and port numbers */
484 u16 af; /* address family */
488 __u32 fwmark; /* Fire wall mark from skb */
489 union nf_inet_addr caddr; /* client address */
490 union nf_inet_addr vaddr; /* virtual address */
491 union nf_inet_addr daddr; /* destination address */
492 volatile __u32 flags; /* status flags */
493 __u16 protocol; /* Which protocol (TCP/UDP) */
495 /* counter and timer */
496 atomic_t refcnt; /* reference count */
497 struct timer_list timer; /* Expiration timer */
498 volatile unsigned long timeout; /* timeout */
500 /* Flags and state transition */
501 spinlock_t lock; /* lock for state transition */
502 volatile __u16 state; /* state info */
503 volatile __u16 old_state; /* old state, to be used for
504 * state transition triggerd
508 /* Control members */
509 struct ip_vs_conn *control; /* Master control connection */
510 atomic_t n_control; /* Number of controlled ones */
511 struct ip_vs_dest *dest; /* real server */
512 atomic_t in_pkts; /* incoming packet counter */
514 /* packet transmitter for different forwarding methods. If it
515 mangles the packet, it must return NF_DROP or better NF_STOLEN,
516 otherwise this must be changed to a sk_buff **.
517 NF_ACCEPT can be returned when destination is local.
519 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
520 struct ip_vs_protocol *pp);
522 /* Note: we can group the following members into a structure,
523 in order to save more space, and the following members are
524 only used in VS/NAT anyway */
525 struct ip_vs_app *app; /* bound ip_vs_app object */
526 void *app_data; /* Application private data */
527 struct ip_vs_seq in_seq; /* incoming seq. struct */
528 struct ip_vs_seq out_seq; /* outgoing seq. struct */
530 const struct ip_vs_pe *pe;
536 * To save some memory in conn table when name space is disabled.
538 static inline struct net *ip_vs_conn_net(const struct ip_vs_conn *cp)
546 static inline void ip_vs_conn_net_set(struct ip_vs_conn *cp, struct net *net)
553 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn *cp,
557 return cp->net == net;
564 * Extended internal versions of struct ip_vs_service_user and
565 * ip_vs_dest_user for IPv6 support.
567 * We need these to conveniently pass around service and destination
568 * options, but unfortunately, we also need to keep the old definitions to
569 * maintain userspace backwards compatibility for the setsockopt interface.
571 struct ip_vs_service_user_kern {
572 /* virtual service addresses */
575 union nf_inet_addr addr; /* virtual ip address */
577 u32 fwmark; /* firwall mark of service */
579 /* virtual service options */
582 unsigned flags; /* virtual service flags */
583 unsigned timeout; /* persistent timeout in sec */
584 u32 netmask; /* persistent netmask */
588 struct ip_vs_dest_user_kern {
589 /* destination server address */
590 union nf_inet_addr addr;
593 /* real server options */
594 unsigned conn_flags; /* connection flags */
595 int weight; /* destination weight */
597 /* thresholds for active connections */
598 u32 u_threshold; /* upper threshold */
599 u32 l_threshold; /* lower threshold */
604 * The information about the virtual service offered to the net
605 * and the forwarding entries
607 struct ip_vs_service {
608 struct list_head s_list; /* for normal service table */
609 struct list_head f_list; /* for fwmark-based service table */
610 atomic_t refcnt; /* reference counter */
611 atomic_t usecnt; /* use counter */
613 u16 af; /* address family */
614 __u16 protocol; /* which protocol (TCP/UDP) */
615 union nf_inet_addr addr; /* IP address for virtual service */
616 __be16 port; /* port number for the service */
617 __u32 fwmark; /* firewall mark of the service */
618 unsigned flags; /* service status flags */
619 unsigned timeout; /* persistent timeout in ticks */
620 __be32 netmask; /* grouping granularity */
623 struct list_head destinations; /* real server d-linked list */
624 __u32 num_dests; /* number of servers */
625 struct ip_vs_stats stats; /* statistics for the service */
626 struct ip_vs_app *inc; /* bind conns to this app inc */
629 struct ip_vs_scheduler *scheduler; /* bound scheduler object */
630 rwlock_t sched_lock; /* lock sched_data */
631 void *sched_data; /* scheduler application data */
633 /* alternate persistence engine */
639 * The real server destination forwarding entry
640 * with ip address, port number, and so on.
643 struct list_head n_list; /* for the dests in the service */
644 struct list_head d_list; /* for table with all the dests */
646 u16 af; /* address family */
647 __be16 port; /* port number of the server */
648 union nf_inet_addr addr; /* IP address of the server */
649 volatile unsigned flags; /* dest status flags */
650 atomic_t conn_flags; /* flags to copy to conn */
651 atomic_t weight; /* server weight */
653 atomic_t refcnt; /* reference counter */
654 struct ip_vs_stats stats; /* statistics */
656 /* connection counters and thresholds */
657 atomic_t activeconns; /* active connections */
658 atomic_t inactconns; /* inactive connections */
659 atomic_t persistconns; /* persistent connections */
660 __u32 u_threshold; /* upper threshold */
661 __u32 l_threshold; /* lower threshold */
663 /* for destination cache */
664 spinlock_t dst_lock; /* lock of dst_cache */
665 struct dst_entry *dst_cache; /* destination cache entry */
666 u32 dst_rtos; /* RT_TOS(tos) for dst */
668 union nf_inet_addr dst_saddr;
670 /* for virtual service */
671 struct ip_vs_service *svc; /* service it belongs to */
672 __u16 protocol; /* which protocol (TCP/UDP) */
673 __be16 vport; /* virtual port number */
674 union nf_inet_addr vaddr; /* virtual IP address */
675 __u32 vfwmark; /* firewall mark of service */
680 * The scheduler object
682 struct ip_vs_scheduler {
683 struct list_head n_list; /* d-linked list head */
684 char *name; /* scheduler name */
685 atomic_t refcnt; /* reference counter */
686 struct module *module; /* THIS_MODULE/NULL */
688 /* scheduler initializing service */
689 int (*init_service)(struct ip_vs_service *svc);
690 /* scheduling service finish */
691 int (*done_service)(struct ip_vs_service *svc);
692 /* scheduler updating service */
693 int (*update_service)(struct ip_vs_service *svc);
695 /* selecting a server from the given service */
696 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
697 const struct sk_buff *skb);
700 /* The persistence engine object */
702 struct list_head n_list; /* d-linked list head */
703 char *name; /* scheduler name */
704 atomic_t refcnt; /* reference counter */
705 struct module *module; /* THIS_MODULE/NULL */
707 /* get the connection template, if any */
708 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
709 bool (*ct_match)(const struct ip_vs_conn_param *p,
710 struct ip_vs_conn *ct);
711 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
713 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
717 * The application module object (a.k.a. app incarnation)
720 struct list_head a_list; /* member in app list */
721 int type; /* IP_VS_APP_TYPE_xxx */
722 char *name; /* application module name */
724 struct module *module; /* THIS_MODULE/NULL */
725 struct list_head incs_list; /* list of incarnations */
727 /* members for application incarnations */
728 struct list_head p_list; /* member in proto app list */
729 struct ip_vs_app *app; /* its real application */
730 __be16 port; /* port number in net order */
731 atomic_t usecnt; /* usage counter */
734 * output hook: Process packet in inout direction, diff set for TCP.
735 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
736 * 2=Mangled but checksum was not updated
738 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
739 struct sk_buff *, int *diff);
742 * input hook: Process packet in outin direction, diff set for TCP.
743 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
744 * 2=Mangled but checksum was not updated
746 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
747 struct sk_buff *, int *diff);
749 /* ip_vs_app initializer */
750 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
752 /* ip_vs_app finish */
753 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
757 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
758 struct ip_vs_protocol *);
760 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
766 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
767 int *verdict, struct ip_vs_conn **cpp);
770 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
771 const struct iphdr *iph, unsigned int proto_off,
775 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
776 const struct iphdr *iph, unsigned int proto_off,
779 int (*state_transition)(struct ip_vs_conn *cp, int direction,
780 const struct sk_buff *skb,
781 struct ip_vs_app *app);
783 void (*timeout_change)(struct ip_vs_app *app, int flags);
786 /* IPVS in network namespace */
788 int gen; /* Generation */
789 int enable; /* enable like nf_hooks do */
791 * Hash table: for real service lookups
793 #define IP_VS_RTAB_BITS 4
794 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
795 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
797 struct list_head rs_table[IP_VS_RTAB_SIZE];
799 struct list_head app_list;
801 struct ip_vs_app *ftp_app;
803 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
804 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
805 /* ip_vs_proto_tcp */
806 #ifdef CONFIG_IP_VS_PROTO_TCP
807 #define TCP_APP_TAB_BITS 4
808 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
809 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
810 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
811 spinlock_t tcp_app_lock;
813 /* ip_vs_proto_udp */
814 #ifdef CONFIG_IP_VS_PROTO_UDP
815 #define UDP_APP_TAB_BITS 4
816 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
817 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
818 struct list_head udp_apps[UDP_APP_TAB_SIZE];
819 spinlock_t udp_app_lock;
821 /* ip_vs_proto_sctp */
822 #ifdef CONFIG_IP_VS_PROTO_SCTP
823 #define SCTP_APP_TAB_BITS 4
824 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
825 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
826 /* Hash table for SCTP application incarnations */
827 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
828 spinlock_t sctp_app_lock;
831 atomic_t conn_count; /* connection counter */
834 struct ip_vs_stats tot_stats; /* Statistics & est. */
836 int num_services; /* no of virtual services */
838 rwlock_t rs_lock; /* real services table */
839 /* semaphore for IPVS sockopts. And, [gs]etsockopt may sleep. */
840 struct lock_class_key ctl_key; /* ctl_mutex debuging */
841 /* Trash for destinations */
842 struct list_head dest_trash;
843 /* Service counters */
844 atomic_t ftpsvc_counter;
845 atomic_t nullsvc_counter;
848 /* 1/rate drop and drop-entry variables */
849 struct delayed_work defense_work; /* Work handler */
854 spinlock_t dropentry_lock; /* drop entry handling */
855 spinlock_t droppacket_lock; /* drop packet handling */
856 spinlock_t securetcp_lock; /* state and timeout tables */
859 struct ctl_table_header *sysctl_hdr;
860 struct ctl_table *sysctl_tbl;
863 /* sysctl variables */
864 int sysctl_amemthresh;
865 int sysctl_am_droprate;
866 int sysctl_drop_entry;
867 int sysctl_drop_packet;
868 int sysctl_secure_tcp;
869 #ifdef CONFIG_IP_VS_NFCT
870 int sysctl_conntrack;
872 int sysctl_snat_reroute;
874 int sysctl_cache_bypass;
875 int sysctl_expire_nodest_conn;
876 int sysctl_expire_quiescent_template;
877 int sysctl_sync_threshold[2];
878 int sysctl_nat_icmp_send;
881 int sysctl_lblc_expiration;
882 struct ctl_table_header *lblc_ctl_header;
883 struct ctl_table *lblc_ctl_table;
885 int sysctl_lblcr_expiration;
886 struct ctl_table_header *lblcr_ctl_header;
887 struct ctl_table *lblcr_ctl_table;
889 struct list_head est_list; /* estimator list */
891 struct timer_list est_timer; /* Estimation timer */
893 struct list_head sync_queue;
894 spinlock_t sync_lock;
895 struct ip_vs_sync_buff *sync_buff;
896 spinlock_t sync_buff_lock;
897 struct sockaddr_in sync_mcast_addr;
898 struct task_struct *master_thread;
899 struct task_struct *backup_thread;
900 int send_mesg_maxlen;
901 int recv_mesg_maxlen;
902 volatile int sync_state;
903 volatile int master_syncid;
904 volatile int backup_syncid;
905 /* multicast interface name */
906 char master_mcast_ifn[IP_VS_IFNAME_MAXLEN];
907 char backup_mcast_ifn[IP_VS_IFNAME_MAXLEN];
908 /* net name space ptr */
909 struct net *net; /* Needed by timer routines */
912 #define DEFAULT_SYNC_THRESHOLD 3
913 #define DEFAULT_SYNC_PERIOD 50
914 #define DEFAULT_SYNC_VER 1
918 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
920 return ipvs->sysctl_sync_threshold[0];
923 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
925 return ipvs->sysctl_sync_threshold[1];
928 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
930 return ipvs->sysctl_sync_ver;
935 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
937 return DEFAULT_SYNC_THRESHOLD;
940 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
942 return DEFAULT_SYNC_PERIOD;
945 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
947 return DEFAULT_SYNC_VER;
953 * IPVS core functions
954 * (from ip_vs_core.c)
956 extern const char *ip_vs_proto_name(unsigned proto);
957 extern void ip_vs_init_hash_table(struct list_head *table, int rows);
958 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
960 #define IP_VS_APP_TYPE_FTP 1
963 * ip_vs_conn handling functions
964 * (from ip_vs_conn.c)
970 IP_VS_DIR_INPUT_ONLY,
974 static inline void ip_vs_conn_fill_param(struct net *net, int af, int protocol,
975 const union nf_inet_addr *caddr,
977 const union nf_inet_addr *vaddr,
979 struct ip_vs_conn_param *p)
983 p->protocol = protocol;
992 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
993 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
995 struct ip_vs_conn * ip_vs_conn_in_get_proto(int af, const struct sk_buff *skb,
996 const struct ip_vs_iphdr *iph,
997 unsigned int proto_off,
1000 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1002 struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
1003 const struct ip_vs_iphdr *iph,
1004 unsigned int proto_off,
1007 /* put back the conn without restarting its timer */
1008 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1010 atomic_dec(&cp->refcnt);
1012 extern void ip_vs_conn_put(struct ip_vs_conn *cp);
1013 extern void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1015 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p,
1016 const union nf_inet_addr *daddr,
1017 __be16 dport, unsigned flags,
1018 struct ip_vs_dest *dest, __u32 fwmark);
1019 extern void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1021 extern const char * ip_vs_state_name(__u16 proto, int state);
1023 extern void ip_vs_tcp_conn_listen(struct net *net, struct ip_vs_conn *cp);
1024 extern int ip_vs_check_template(struct ip_vs_conn *ct);
1025 extern void ip_vs_random_dropentry(struct net *net);
1026 extern int ip_vs_conn_init(void);
1027 extern void ip_vs_conn_cleanup(void);
1029 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1031 struct ip_vs_conn *ctl_cp = cp->control;
1033 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1035 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1037 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1043 IP_VS_DBG_BUF(7, "DELeting control for: "
1044 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1045 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1047 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1048 ntohs(ctl_cp->cport));
1051 if (atomic_read(&ctl_cp->n_control) == 0) {
1052 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1054 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1056 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1061 atomic_dec(&ctl_cp->n_control);
1065 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1068 IP_VS_ERR_BUF("request control ADD for already controlled: "
1070 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1072 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1075 ip_vs_control_del(cp);
1078 IP_VS_DBG_BUF(7, "ADDing control for: "
1079 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1080 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1082 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1083 ntohs(ctl_cp->cport));
1085 cp->control = ctl_cp;
1086 atomic_inc(&ctl_cp->n_control);
1090 * IPVS netns init & cleanup functions
1092 extern int __ip_vs_estimator_init(struct net *net);
1093 extern int __ip_vs_control_init(struct net *net);
1094 extern int __ip_vs_protocol_init(struct net *net);
1095 extern int __ip_vs_app_init(struct net *net);
1096 extern int __ip_vs_conn_init(struct net *net);
1097 extern int __ip_vs_sync_init(struct net *net);
1098 extern void __ip_vs_conn_cleanup(struct net *net);
1099 extern void __ip_vs_app_cleanup(struct net *net);
1100 extern void __ip_vs_protocol_cleanup(struct net *net);
1101 extern void __ip_vs_control_cleanup(struct net *net);
1102 extern void __ip_vs_estimator_cleanup(struct net *net);
1103 extern void __ip_vs_sync_cleanup(struct net *net);
1104 extern void __ip_vs_service_cleanup(struct net *net);
1107 * IPVS application functions
1108 * (from ip_vs_app.c)
1110 #define IP_VS_APP_MAX_PORTS 8
1111 extern int register_ip_vs_app(struct net *net, struct ip_vs_app *app);
1112 extern void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app);
1113 extern int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1114 extern void ip_vs_unbind_app(struct ip_vs_conn *cp);
1115 extern int register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app,
1116 __u16 proto, __u16 port);
1117 extern int ip_vs_app_inc_get(struct ip_vs_app *inc);
1118 extern void ip_vs_app_inc_put(struct ip_vs_app *inc);
1120 extern int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1121 extern int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1122 extern int ip_vs_app_init(void);
1123 extern void ip_vs_app_cleanup(void);
1125 void ip_vs_bind_pe(struct ip_vs_service *svc, struct ip_vs_pe *pe);
1126 void ip_vs_unbind_pe(struct ip_vs_service *svc);
1127 int register_ip_vs_pe(struct ip_vs_pe *pe);
1128 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1129 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1130 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1132 static inline void ip_vs_pe_get(const struct ip_vs_pe *pe)
1134 if (pe && pe->module)
1135 __module_get(pe->module);
1138 static inline void ip_vs_pe_put(const struct ip_vs_pe *pe)
1140 if (pe && pe->module)
1141 module_put(pe->module);
1145 * IPVS protocol functions (from ip_vs_proto.c)
1147 extern int ip_vs_protocol_init(void);
1148 extern void ip_vs_protocol_cleanup(void);
1149 extern void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1150 extern int *ip_vs_create_timeout_table(int *table, int size);
1152 ip_vs_set_state_timeout(int *table, int num, const char *const *names,
1153 const char *name, int to);
1155 ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1156 const struct sk_buff *skb,
1157 int offset, const char *msg);
1159 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1160 extern struct ip_vs_protocol ip_vs_protocol_udp;
1161 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1162 extern struct ip_vs_protocol ip_vs_protocol_esp;
1163 extern struct ip_vs_protocol ip_vs_protocol_ah;
1164 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1167 * Registering/unregistering scheduler functions
1168 * (from ip_vs_sched.c)
1170 extern int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1171 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1172 extern int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1173 struct ip_vs_scheduler *scheduler);
1174 extern int ip_vs_unbind_scheduler(struct ip_vs_service *svc);
1175 extern struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1176 extern void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1177 extern struct ip_vs_conn *
1178 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1179 struct ip_vs_proto_data *pd, int *ignored);
1180 extern int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1181 struct ip_vs_proto_data *pd);
1183 extern void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1187 * IPVS control data and functions (from ip_vs_ctl.c)
1189 extern struct ip_vs_stats ip_vs_stats;
1190 extern const struct ctl_path net_vs_ctl_path[];
1191 extern int sysctl_ip_vs_sync_ver;
1193 extern void ip_vs_sync_switch_mode(struct net *net, int mode);
1194 extern struct ip_vs_service *
1195 ip_vs_service_get(struct net *net, int af, __u32 fwmark, __u16 protocol,
1196 const union nf_inet_addr *vaddr, __be16 vport);
1198 static inline void ip_vs_service_put(struct ip_vs_service *svc)
1200 atomic_dec(&svc->usecnt);
1203 extern struct ip_vs_dest *
1204 ip_vs_lookup_real_service(struct net *net, int af, __u16 protocol,
1205 const union nf_inet_addr *daddr, __be16 dport);
1207 extern int ip_vs_use_count_inc(void);
1208 extern void ip_vs_use_count_dec(void);
1209 extern int ip_vs_control_init(void);
1210 extern void ip_vs_control_cleanup(void);
1211 extern struct ip_vs_dest *
1212 ip_vs_find_dest(struct net *net, int af, const union nf_inet_addr *daddr,
1213 __be16 dport, const union nf_inet_addr *vaddr, __be16 vport,
1214 __u16 protocol, __u32 fwmark);
1215 extern struct ip_vs_dest *ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1219 * IPVS sync daemon data and function prototypes
1220 * (from ip_vs_sync.c)
1222 extern int start_sync_thread(struct net *net, int state, char *mcast_ifn,
1224 extern int stop_sync_thread(struct net *net, int state);
1225 extern void ip_vs_sync_conn(struct net *net, struct ip_vs_conn *cp);
1226 extern int ip_vs_sync_init(void);
1227 extern void ip_vs_sync_cleanup(void);
1231 * IPVS rate estimator prototypes (from ip_vs_est.c)
1233 extern int ip_vs_estimator_init(void);
1234 extern void ip_vs_estimator_cleanup(void);
1235 extern void ip_vs_start_estimator(struct net *net, struct ip_vs_stats *stats);
1236 extern void ip_vs_stop_estimator(struct net *net, struct ip_vs_stats *stats);
1237 extern void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1238 extern void ip_vs_read_estimator(struct ip_vs_stats_user *dst,
1239 struct ip_vs_stats *stats);
1242 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1244 extern int ip_vs_null_xmit
1245 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1246 extern int ip_vs_bypass_xmit
1247 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1248 extern int ip_vs_nat_xmit
1249 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1250 extern int ip_vs_tunnel_xmit
1251 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1252 extern int ip_vs_dr_xmit
1253 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1254 extern int ip_vs_icmp_xmit
1255 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1256 int offset, unsigned int hooknum);
1257 extern void ip_vs_dst_reset(struct ip_vs_dest *dest);
1259 #ifdef CONFIG_IP_VS_IPV6
1260 extern int ip_vs_bypass_xmit_v6
1261 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1262 extern int ip_vs_nat_xmit_v6
1263 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1264 extern int ip_vs_tunnel_xmit_v6
1265 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1266 extern int ip_vs_dr_xmit_v6
1267 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1268 extern int ip_vs_icmp_xmit_v6
1269 (struct sk_buff *skb, struct ip_vs_conn *cp, struct ip_vs_protocol *pp,
1270 int offset, unsigned int hooknum);
1273 #ifdef CONFIG_SYSCTL
1275 * This is a simple mechanism to ignore packets when
1276 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1277 * we start to drop 1/rate of the packets
1280 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1282 if (!ipvs->drop_rate)
1284 if (--ipvs->drop_counter > 0)
1286 ipvs->drop_counter = ipvs->drop_rate;
1290 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1294 * ip_vs_fwd_tag returns the forwarding tag of the connection
1296 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1298 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1302 switch (IP_VS_FWD_METHOD(cp)) {
1303 case IP_VS_CONN_F_MASQ:
1305 case IP_VS_CONN_F_LOCALNODE:
1307 case IP_VS_CONN_F_TUNNEL:
1309 case IP_VS_CONN_F_DROUTE:
1311 case IP_VS_CONN_F_BYPASS:
1319 extern void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1320 struct ip_vs_conn *cp, int dir);
1322 #ifdef CONFIG_IP_VS_IPV6
1323 extern void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1324 struct ip_vs_conn *cp, int dir);
1327 extern __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1329 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1331 __be32 diff[2] = { ~old, new };
1333 return csum_partial(diff, sizeof(diff), oldsum);
1336 #ifdef CONFIG_IP_VS_IPV6
1337 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1340 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1341 new[3], new[2], new[1], new[0] };
1343 return csum_partial(diff, sizeof(diff), oldsum);
1347 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1349 __be16 diff[2] = { ~old, new };
1351 return csum_partial(diff, sizeof(diff), oldsum);
1355 * Forget current conntrack (unconfirmed) and attach notrack entry
1357 static inline void ip_vs_notrack(struct sk_buff *skb)
1359 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1360 enum ip_conntrack_info ctinfo;
1361 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1363 if (!ct || !nf_ct_is_untracked(ct)) {
1365 skb->nfct = &nf_ct_untracked_get()->ct_general;
1366 skb->nfctinfo = IP_CT_NEW;
1367 nf_conntrack_get(skb->nfct);
1372 #ifdef CONFIG_IP_VS_NFCT
1374 * Netfilter connection tracking
1375 * (from ip_vs_nfct.c)
1377 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1379 #ifdef CONFIG_SYSCTL
1380 return ipvs->sysctl_conntrack;
1386 extern void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1388 extern int ip_vs_confirm_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp);
1389 extern void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1390 struct ip_vs_conn *cp, u_int8_t proto,
1391 const __be16 port, int from_rs);
1392 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1396 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1401 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1402 struct ip_vs_conn *cp, int outin)
1406 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb,
1407 struct ip_vs_conn *cp)
1412 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1415 /* CONFIG_IP_VS_NFCT */
1418 static inline unsigned int
1419 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1422 * We think the overhead of processing active connections is 256
1423 * times higher than that of inactive connections in average. (This
1424 * 256 times might not be accurate, we will change it later) We
1425 * use the following formula to estimate the overhead now:
1426 * dest->activeconns*256 + dest->inactconns
1428 return (atomic_read(&dest->activeconns) << 8) +
1429 atomic_read(&dest->inactconns);
1432 #endif /* _NET_IP_VS_H */