1 /* SPDX-License-Identifier: GPL-2.0 */
3 * data structure and functionality definitions
9 #include <linux/ip_vs.h> /* definitions shared with userland */
11 #include <asm/types.h> /* for __uXX types */
13 #include <linux/list.h> /* for struct list_head */
14 #include <linux/spinlock.h> /* for struct rwlock_t */
15 #include <linux/atomic.h> /* for struct atomic_t */
16 #include <linux/refcount.h> /* for struct refcount_t */
18 #include <linux/compiler.h>
19 #include <linux/timer.h>
20 #include <linux/bug.h>
22 #include <net/checksum.h>
23 #include <linux/netfilter.h> /* for union nf_inet_addr */
25 #include <linux/ipv6.h> /* for struct ipv6hdr */
27 #if IS_ENABLED(CONFIG_IP_VS_IPV6)
28 #include <linux/netfilter_ipv6/ip6_tables.h>
30 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
31 #include <net/netfilter/nf_conntrack.h>
33 #include <net/net_namespace.h> /* Netw namespace */
35 #define IP_VS_HDR_INVERSE 1
36 #define IP_VS_HDR_ICMP 2
38 /* Generic access of ipvs struct */
39 static inline struct netns_ipvs *net_ipvs(struct net* net)
44 /* This one needed for single_open_net since net is stored directly in
45 * private not as a struct i.e. seq_file_net can't be used.
47 static inline struct net *seq_file_single_net(struct seq_file *seq)
50 return (struct net *)seq->private;
56 /* Connections' size value needed by ip_vs_ctl.c */
57 extern int ip_vs_conn_tab_size;
60 int hdr_flags; /* ipvs flags */
61 __u32 off; /* Where IP or IPv4 header starts */
62 __u32 len; /* IPv4 simply where L4 starts
63 * IPv6 where L4 Transport Header starts */
64 __u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
67 union nf_inet_addr saddr;
68 union nf_inet_addr daddr;
71 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
72 int len, void *buffer)
74 return skb_header_pointer(skb, offset, len, buffer);
77 /* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
78 * IPv6 requires some extra work, as finding proper header position,
79 * depend on the IPv6 extension headers.
82 ip_vs_fill_iph_skb_off(int af, const struct sk_buff *skb, int offset,
83 int hdr_flags, struct ip_vs_iphdr *iphdr)
85 iphdr->hdr_flags = hdr_flags;
88 #ifdef CONFIG_IP_VS_IPV6
91 const struct ipv6hdr *iph = skb_header_pointer(
92 skb, offset, sizeof(_iph), &_iph);
96 iphdr->saddr.in6 = iph->saddr;
97 iphdr->daddr.in6 = iph->daddr;
98 /* ipv6_find_hdr() updates len, flags */
101 iphdr->protocol = ipv6_find_hdr(skb, &iphdr->len, -1,
104 if (iphdr->protocol < 0)
110 const struct iphdr *iph = skb_header_pointer(
111 skb, offset, sizeof(_iph), &_iph);
115 iphdr->len = offset + iph->ihl * 4;
117 iphdr->protocol = iph->protocol;
118 iphdr->saddr.ip = iph->saddr;
119 iphdr->daddr.ip = iph->daddr;
126 ip_vs_fill_iph_skb_icmp(int af, const struct sk_buff *skb, int offset,
127 bool inverse, struct ip_vs_iphdr *iphdr)
129 int hdr_flags = IP_VS_HDR_ICMP;
132 hdr_flags |= IP_VS_HDR_INVERSE;
134 return ip_vs_fill_iph_skb_off(af, skb, offset, hdr_flags, iphdr);
138 ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, bool inverse,
139 struct ip_vs_iphdr *iphdr)
144 hdr_flags |= IP_VS_HDR_INVERSE;
146 return ip_vs_fill_iph_skb_off(af, skb, skb_network_offset(skb),
151 ip_vs_iph_inverse(const struct ip_vs_iphdr *iph)
153 return !!(iph->hdr_flags & IP_VS_HDR_INVERSE);
157 ip_vs_iph_icmp(const struct ip_vs_iphdr *iph)
159 return !!(iph->hdr_flags & IP_VS_HDR_ICMP);
162 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
163 const union nf_inet_addr *src)
165 #ifdef CONFIG_IP_VS_IPV6
173 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
174 const union nf_inet_addr *src)
176 #ifdef CONFIG_IP_VS_IPV6
177 if (af == AF_INET6) {
188 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
189 const union nf_inet_addr *b)
191 #ifdef CONFIG_IP_VS_IPV6
193 return ipv6_addr_equal(&a->in6, &b->in6);
195 return a->ip == b->ip;
198 #ifdef CONFIG_IP_VS_DEBUG
199 #include <linux/net.h>
201 int ip_vs_get_debug_level(void);
203 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
204 const union nf_inet_addr *addr,
208 #ifdef CONFIG_IP_VS_IPV6
210 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
214 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
218 BUG_ON(*idx > buf_len + 1);
219 return &buf[*idx - len];
222 #define IP_VS_DBG_BUF(level, msg, ...) \
224 char ip_vs_dbg_buf[160]; \
225 int ip_vs_dbg_idx = 0; \
226 if (level <= ip_vs_get_debug_level()) \
227 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
229 #define IP_VS_ERR_BUF(msg...) \
231 char ip_vs_dbg_buf[160]; \
232 int ip_vs_dbg_idx = 0; \
236 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
237 #define IP_VS_DBG_ADDR(af, addr) \
238 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
239 sizeof(ip_vs_dbg_buf), addr, \
242 #define IP_VS_DBG(level, msg, ...) \
244 if (level <= ip_vs_get_debug_level()) \
245 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
247 #define IP_VS_DBG_RL(msg, ...) \
249 if (net_ratelimit()) \
250 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
252 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
254 if (level <= ip_vs_get_debug_level()) \
255 pp->debug_packet(af, pp, skb, ofs, msg); \
257 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
259 if (level <= ip_vs_get_debug_level() && \
261 pp->debug_packet(af, pp, skb, ofs, msg); \
263 #else /* NO DEBUGGING at ALL */
264 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
265 #define IP_VS_ERR_BUF(msg...) do {} while (0)
266 #define IP_VS_DBG(level, msg...) do {} while (0)
267 #define IP_VS_DBG_RL(msg...) do {} while (0)
268 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
269 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
272 #define IP_VS_BUG() BUG()
273 #define IP_VS_ERR_RL(msg, ...) \
275 if (net_ratelimit()) \
276 pr_err(msg, ##__VA_ARGS__); \
279 #ifdef CONFIG_IP_VS_DEBUG
280 #define EnterFunction(level) \
282 if (level <= ip_vs_get_debug_level()) \
284 pr_fmt("Enter: %s, %s line %i\n"), \
285 __func__, __FILE__, __LINE__); \
287 #define LeaveFunction(level) \
289 if (level <= ip_vs_get_debug_level()) \
291 pr_fmt("Leave: %s, %s line %i\n"), \
292 __func__, __FILE__, __LINE__); \
295 #define EnterFunction(level) do {} while (0)
296 #define LeaveFunction(level) do {} while (0)
299 /* The port number of FTP service (in network order). */
300 #define FTPPORT cpu_to_be16(21)
301 #define FTPDATA cpu_to_be16(20)
303 /* TCP State Values */
305 IP_VS_TCP_S_NONE = 0,
306 IP_VS_TCP_S_ESTABLISHED,
307 IP_VS_TCP_S_SYN_SENT,
308 IP_VS_TCP_S_SYN_RECV,
309 IP_VS_TCP_S_FIN_WAIT,
310 IP_VS_TCP_S_TIME_WAIT,
312 IP_VS_TCP_S_CLOSE_WAIT,
313 IP_VS_TCP_S_LAST_ACK,
319 /* UDP State Values */
325 /* ICMP State Values */
331 /* SCTP State Values */
332 enum ip_vs_sctp_states {
336 IP_VS_SCTP_S_COOKIE_SENT,
337 IP_VS_SCTP_S_COOKIE_REPLIED,
338 IP_VS_SCTP_S_COOKIE_WAIT,
340 IP_VS_SCTP_S_COOKIE_ECHOED,
341 IP_VS_SCTP_S_ESTABLISHED,
342 IP_VS_SCTP_S_SHUTDOWN_SENT,
343 IP_VS_SCTP_S_SHUTDOWN_RECEIVED,
344 IP_VS_SCTP_S_SHUTDOWN_ACK_SENT,
345 IP_VS_SCTP_S_REJECTED,
350 /* Delta sequence info structure
351 * Each ip_vs_conn has 2 (output AND input seq. changes).
352 * Only used in the VS/NAT.
355 __u32 init_seq; /* Add delta from this seq */
356 __u32 delta; /* Delta in sequence numbers */
357 __u32 previous_delta; /* Delta in sequence numbers
358 * before last resized pkt */
361 /* counters per cpu */
362 struct ip_vs_counters {
363 __u64 conns; /* connections scheduled */
364 __u64 inpkts; /* incoming packets */
365 __u64 outpkts; /* outgoing packets */
366 __u64 inbytes; /* incoming bytes */
367 __u64 outbytes; /* outgoing bytes */
370 struct ip_vs_cpu_stats {
371 struct ip_vs_counters cnt;
372 struct u64_stats_sync syncp;
375 /* IPVS statistics objects */
376 struct ip_vs_estimator {
377 struct list_head list;
393 * IPVS statistics object, 64-bit kernel version of struct ip_vs_stats_user
395 struct ip_vs_kstats {
396 u64 conns; /* connections scheduled */
397 u64 inpkts; /* incoming packets */
398 u64 outpkts; /* outgoing packets */
399 u64 inbytes; /* incoming bytes */
400 u64 outbytes; /* outgoing bytes */
402 u64 cps; /* current connection rate */
403 u64 inpps; /* current in packet rate */
404 u64 outpps; /* current out packet rate */
405 u64 inbps; /* current in byte rate */
406 u64 outbps; /* current out byte rate */
410 struct ip_vs_kstats kstats; /* kernel statistics */
411 struct ip_vs_estimator est; /* estimator */
412 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
413 spinlock_t lock; /* spin lock */
414 struct ip_vs_kstats kstats0; /* reset values */
422 struct ip_vs_proto_data;
424 struct ip_vs_protocol {
425 struct ip_vs_protocol *next;
431 void (*init)(struct ip_vs_protocol *pp);
433 void (*exit)(struct ip_vs_protocol *pp);
435 int (*init_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
437 void (*exit_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
439 int (*conn_schedule)(struct netns_ipvs *ipvs,
440 int af, struct sk_buff *skb,
441 struct ip_vs_proto_data *pd,
442 int *verdict, struct ip_vs_conn **cpp,
443 struct ip_vs_iphdr *iph);
446 (*conn_in_get)(struct netns_ipvs *ipvs,
448 const struct sk_buff *skb,
449 const struct ip_vs_iphdr *iph);
452 (*conn_out_get)(struct netns_ipvs *ipvs,
454 const struct sk_buff *skb,
455 const struct ip_vs_iphdr *iph);
457 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
458 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
460 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
461 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
463 int (*csum_check)(int af, struct sk_buff *skb,
464 struct ip_vs_protocol *pp);
466 const char *(*state_name)(int state);
468 void (*state_transition)(struct ip_vs_conn *cp, int direction,
469 const struct sk_buff *skb,
470 struct ip_vs_proto_data *pd);
472 int (*register_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
474 void (*unregister_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
476 int (*app_conn_bind)(struct ip_vs_conn *cp);
478 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
479 const struct sk_buff *skb,
483 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
486 /* protocol data per netns */
487 struct ip_vs_proto_data {
488 struct ip_vs_proto_data *next;
489 struct ip_vs_protocol *pp;
490 int *timeout_table; /* protocol timeout table */
491 atomic_t appcnt; /* counter of proto app incs. */
492 struct tcp_states_t *tcp_state_table;
495 struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
496 struct ip_vs_proto_data *ip_vs_proto_data_get(struct netns_ipvs *ipvs,
497 unsigned short proto);
499 struct ip_vs_conn_param {
500 struct netns_ipvs *ipvs;
501 const union nf_inet_addr *caddr;
502 const union nf_inet_addr *vaddr;
508 const struct ip_vs_pe *pe;
513 /* IP_VS structure allocated for each dynamically scheduled connection */
515 struct hlist_node c_list; /* hashed list heads */
516 /* Protocol, addresses and port numbers */
520 u16 af; /* address family */
521 union nf_inet_addr caddr; /* client address */
522 union nf_inet_addr vaddr; /* virtual address */
523 union nf_inet_addr daddr; /* destination address */
524 volatile __u32 flags; /* status flags */
525 __u16 protocol; /* Which protocol (TCP/UDP) */
526 __u16 daf; /* Address family of the dest */
527 struct netns_ipvs *ipvs;
529 /* counter and timer */
530 refcount_t refcnt; /* reference count */
531 struct timer_list timer; /* Expiration timer */
532 volatile unsigned long timeout; /* timeout */
534 /* Flags and state transition */
535 spinlock_t lock; /* lock for state transition */
536 volatile __u16 state; /* state info */
537 volatile __u16 old_state; /* old state, to be used for
538 * state transition triggerd
541 __u32 fwmark; /* Fire wall mark from skb */
542 unsigned long sync_endtime; /* jiffies + sent_retries */
544 /* Control members */
545 struct ip_vs_conn *control; /* Master control connection */
546 atomic_t n_control; /* Number of controlled ones */
547 struct ip_vs_dest *dest; /* real server */
548 atomic_t in_pkts; /* incoming packet counter */
550 /* Packet transmitter for different forwarding methods. If it
551 * mangles the packet, it must return NF_DROP or better NF_STOLEN,
552 * otherwise this must be changed to a sk_buff **.
553 * NF_ACCEPT can be returned when destination is local.
555 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
556 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
558 /* Note: we can group the following members into a structure,
559 * in order to save more space, and the following members are
560 * only used in VS/NAT anyway
562 struct ip_vs_app *app; /* bound ip_vs_app object */
563 void *app_data; /* Application private data */
564 struct ip_vs_seq in_seq; /* incoming seq. struct */
565 struct ip_vs_seq out_seq; /* outgoing seq. struct */
567 const struct ip_vs_pe *pe;
571 struct rcu_head rcu_head;
574 /* Extended internal versions of struct ip_vs_service_user and ip_vs_dest_user
577 * We need these to conveniently pass around service and destination
578 * options, but unfortunately, we also need to keep the old definitions to
579 * maintain userspace backwards compatibility for the setsockopt interface.
581 struct ip_vs_service_user_kern {
582 /* virtual service addresses */
585 union nf_inet_addr addr; /* virtual ip address */
587 u32 fwmark; /* firwall mark of service */
589 /* virtual service options */
592 unsigned int flags; /* virtual service flags */
593 unsigned int timeout; /* persistent timeout in sec */
594 __be32 netmask; /* persistent netmask or plen */
598 struct ip_vs_dest_user_kern {
599 /* destination server address */
600 union nf_inet_addr addr;
603 /* real server options */
604 unsigned int conn_flags; /* connection flags */
605 int weight; /* destination weight */
607 /* thresholds for active connections */
608 u32 u_threshold; /* upper threshold */
609 u32 l_threshold; /* lower threshold */
611 /* Address family of addr */
617 * The information about the virtual service offered to the net and the
618 * forwarding entries.
620 struct ip_vs_service {
621 struct hlist_node s_list; /* for normal service table */
622 struct hlist_node f_list; /* for fwmark-based service table */
623 atomic_t refcnt; /* reference counter */
625 u16 af; /* address family */
626 __u16 protocol; /* which protocol (TCP/UDP) */
627 union nf_inet_addr addr; /* IP address for virtual service */
628 __be16 port; /* port number for the service */
629 __u32 fwmark; /* firewall mark of the service */
630 unsigned int flags; /* service status flags */
631 unsigned int timeout; /* persistent timeout in ticks */
632 __be32 netmask; /* grouping granularity, mask/plen */
633 struct netns_ipvs *ipvs;
635 struct list_head destinations; /* real server d-linked list */
636 __u32 num_dests; /* number of servers */
637 struct ip_vs_stats stats; /* statistics for the service */
640 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
641 spinlock_t sched_lock; /* lock sched_data */
642 void *sched_data; /* scheduler application data */
644 /* alternate persistence engine */
645 struct ip_vs_pe __rcu *pe;
647 struct rcu_head rcu_head;
650 /* Information for cached dst */
651 struct ip_vs_dest_dst {
652 struct dst_entry *dst_cache; /* destination cache entry */
654 union nf_inet_addr dst_saddr;
655 struct rcu_head rcu_head;
658 /* The real server destination forwarding entry with ip address, port number,
662 struct list_head n_list; /* for the dests in the service */
663 struct hlist_node d_list; /* for table with all the dests */
665 u16 af; /* address family */
666 __be16 port; /* port number of the server */
667 union nf_inet_addr addr; /* IP address of the server */
668 volatile unsigned int flags; /* dest status flags */
669 atomic_t conn_flags; /* flags to copy to conn */
670 atomic_t weight; /* server weight */
671 atomic_t last_weight; /* server latest weight */
673 refcount_t refcnt; /* reference counter */
674 struct ip_vs_stats stats; /* statistics */
675 unsigned long idle_start; /* start time, jiffies */
677 /* connection counters and thresholds */
678 atomic_t activeconns; /* active connections */
679 atomic_t inactconns; /* inactive connections */
680 atomic_t persistconns; /* persistent connections */
681 __u32 u_threshold; /* upper threshold */
682 __u32 l_threshold; /* lower threshold */
684 /* for destination cache */
685 spinlock_t dst_lock; /* lock of dst_cache */
686 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
688 /* for virtual service */
689 struct ip_vs_service __rcu *svc; /* service it belongs to */
690 __u16 protocol; /* which protocol (TCP/UDP) */
691 __be16 vport; /* virtual port number */
692 union nf_inet_addr vaddr; /* virtual IP address */
693 __u32 vfwmark; /* firewall mark of service */
695 struct list_head t_list; /* in dest_trash */
696 unsigned int in_rs_table:1; /* we are in rs_table */
699 /* The scheduler object */
700 struct ip_vs_scheduler {
701 struct list_head n_list; /* d-linked list head */
702 char *name; /* scheduler name */
703 atomic_t refcnt; /* reference counter */
704 struct module *module; /* THIS_MODULE/NULL */
706 /* scheduler initializing service */
707 int (*init_service)(struct ip_vs_service *svc);
708 /* scheduling service finish */
709 void (*done_service)(struct ip_vs_service *svc);
711 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
712 /* dest is unlinked */
713 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
714 /* dest is updated */
715 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
717 /* selecting a server from the given service */
718 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
719 const struct sk_buff *skb,
720 struct ip_vs_iphdr *iph);
723 /* The persistence engine object */
725 struct list_head n_list; /* d-linked list head */
726 char *name; /* scheduler name */
727 atomic_t refcnt; /* reference counter */
728 struct module *module; /* THIS_MODULE/NULL */
730 /* get the connection template, if any */
731 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
732 bool (*ct_match)(const struct ip_vs_conn_param *p,
733 struct ip_vs_conn *ct);
734 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
736 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
737 /* create connections for real-server outgoing packets */
738 struct ip_vs_conn* (*conn_out)(struct ip_vs_service *svc,
739 struct ip_vs_dest *dest,
741 const struct ip_vs_iphdr *iph,
742 __be16 dport, __be16 cport);
745 /* The application module object (a.k.a. app incarnation) */
747 struct list_head a_list; /* member in app list */
748 int type; /* IP_VS_APP_TYPE_xxx */
749 char *name; /* application module name */
751 struct module *module; /* THIS_MODULE/NULL */
752 struct list_head incs_list; /* list of incarnations */
754 /* members for application incarnations */
755 struct list_head p_list; /* member in proto app list */
756 struct ip_vs_app *app; /* its real application */
757 __be16 port; /* port number in net order */
758 atomic_t usecnt; /* usage counter */
759 struct rcu_head rcu_head;
761 /* output hook: Process packet in inout direction, diff set for TCP.
762 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
763 * 2=Mangled but checksum was not updated
765 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
766 struct sk_buff *, int *diff);
768 /* input hook: Process packet in outin direction, diff set for TCP.
769 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
770 * 2=Mangled but checksum was not updated
772 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
773 struct sk_buff *, int *diff);
775 /* ip_vs_app initializer */
776 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
778 /* ip_vs_app finish */
779 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
783 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
784 struct ip_vs_protocol *);
786 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
792 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
793 int *verdict, struct ip_vs_conn **cpp);
796 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
797 const struct iphdr *iph, int inverse);
800 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
801 const struct iphdr *iph, int inverse);
803 int (*state_transition)(struct ip_vs_conn *cp, int direction,
804 const struct sk_buff *skb,
805 struct ip_vs_app *app);
807 void (*timeout_change)(struct ip_vs_app *app, int flags);
810 struct ipvs_master_sync_state {
811 struct list_head sync_queue;
812 struct ip_vs_sync_buff *sync_buff;
813 unsigned long sync_queue_len;
814 unsigned int sync_queue_delay;
815 struct task_struct *master_thread;
816 struct delayed_work master_wakeup_work;
817 struct netns_ipvs *ipvs;
820 /* How much time to keep dests in trash */
821 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
823 struct ipvs_sync_daemon_cfg {
824 union nf_inet_addr mcast_group;
830 /* multicast interface name */
831 char mcast_ifn[IP_VS_IFNAME_MAXLEN];
834 /* IPVS in network namespace */
836 int gen; /* Generation */
837 int enable; /* enable like nf_hooks do */
838 /* Hash table: for real service lookups */
839 #define IP_VS_RTAB_BITS 4
840 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
841 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
843 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
845 struct list_head app_list;
847 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
848 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
849 /* ip_vs_proto_tcp */
850 #ifdef CONFIG_IP_VS_PROTO_TCP
851 #define TCP_APP_TAB_BITS 4
852 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
853 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
854 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
856 /* ip_vs_proto_udp */
857 #ifdef CONFIG_IP_VS_PROTO_UDP
858 #define UDP_APP_TAB_BITS 4
859 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
860 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
861 struct list_head udp_apps[UDP_APP_TAB_SIZE];
863 /* ip_vs_proto_sctp */
864 #ifdef CONFIG_IP_VS_PROTO_SCTP
865 #define SCTP_APP_TAB_BITS 4
866 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
867 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
868 /* Hash table for SCTP application incarnations */
869 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
872 atomic_t conn_count; /* connection counter */
875 struct ip_vs_stats tot_stats; /* Statistics & est. */
877 int num_services; /* no of virtual services */
879 /* Trash for destinations */
880 struct list_head dest_trash;
881 spinlock_t dest_trash_lock;
882 struct timer_list dest_trash_timer; /* expiration timer */
883 /* Service counters */
884 atomic_t ftpsvc_counter;
885 atomic_t nullsvc_counter;
886 atomic_t conn_out_counter;
889 /* 1/rate drop and drop-entry variables */
890 struct delayed_work defense_work; /* Work handler */
895 spinlock_t dropentry_lock; /* drop entry handling */
896 spinlock_t droppacket_lock; /* drop packet handling */
897 spinlock_t securetcp_lock; /* state and timeout tables */
900 struct ctl_table_header *sysctl_hdr;
901 struct ctl_table *sysctl_tbl;
904 /* sysctl variables */
905 int sysctl_amemthresh;
906 int sysctl_am_droprate;
907 int sysctl_drop_entry;
908 int sysctl_drop_packet;
909 int sysctl_secure_tcp;
910 #ifdef CONFIG_IP_VS_NFCT
911 int sysctl_conntrack;
913 int sysctl_snat_reroute;
915 int sysctl_sync_ports;
916 int sysctl_sync_persist_mode;
917 unsigned long sysctl_sync_qlen_max;
918 int sysctl_sync_sock_size;
919 int sysctl_cache_bypass;
920 int sysctl_expire_nodest_conn;
921 int sysctl_sloppy_tcp;
922 int sysctl_sloppy_sctp;
923 int sysctl_expire_quiescent_template;
924 int sysctl_sync_threshold[2];
925 unsigned int sysctl_sync_refresh_period;
926 int sysctl_sync_retries;
927 int sysctl_nat_icmp_send;
928 int sysctl_pmtu_disc;
929 int sysctl_backup_only;
930 int sysctl_conn_reuse_mode;
931 int sysctl_schedule_icmp;
932 int sysctl_ignore_tunneled;
935 int sysctl_lblc_expiration;
936 struct ctl_table_header *lblc_ctl_header;
937 struct ctl_table *lblc_ctl_table;
939 int sysctl_lblcr_expiration;
940 struct ctl_table_header *lblcr_ctl_header;
941 struct ctl_table *lblcr_ctl_table;
943 struct list_head est_list; /* estimator list */
945 struct timer_list est_timer; /* Estimation timer */
947 spinlock_t sync_lock;
948 struct ipvs_master_sync_state *ms;
949 spinlock_t sync_buff_lock;
950 struct task_struct **backup_threads;
952 volatile int sync_state;
953 struct mutex sync_mutex;
954 struct ipvs_sync_daemon_cfg mcfg; /* Master Configuration */
955 struct ipvs_sync_daemon_cfg bcfg; /* Backup Configuration */
956 /* net name space ptr */
957 struct net *net; /* Needed by timer routines */
958 /* Number of heterogeneous destinations, needed becaus heterogeneous
959 * are not supported when synchronization is enabled.
961 unsigned int mixed_address_family_dests;
964 #define DEFAULT_SYNC_THRESHOLD 3
965 #define DEFAULT_SYNC_PERIOD 50
966 #define DEFAULT_SYNC_VER 1
967 #define DEFAULT_SLOPPY_TCP 0
968 #define DEFAULT_SLOPPY_SCTP 0
969 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
970 #define DEFAULT_SYNC_RETRIES 0
971 #define IPVS_SYNC_WAKEUP_RATE 8
972 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
973 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
974 #define IPVS_SYNC_CHECK_PERIOD HZ
975 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
976 #define IPVS_SYNC_PORTS_MAX (1 << 6)
980 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
982 return ipvs->sysctl_sync_threshold[0];
985 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
987 return READ_ONCE(ipvs->sysctl_sync_threshold[1]);
990 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
992 return READ_ONCE(ipvs->sysctl_sync_refresh_period);
995 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
997 return ipvs->sysctl_sync_retries;
1000 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1002 return ipvs->sysctl_sync_ver;
1005 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1007 return ipvs->sysctl_sloppy_tcp;
1010 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1012 return ipvs->sysctl_sloppy_sctp;
1015 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1017 return READ_ONCE(ipvs->sysctl_sync_ports);
1020 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1022 return ipvs->sysctl_sync_persist_mode;
1025 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1027 return ipvs->sysctl_sync_qlen_max;
1030 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1032 return ipvs->sysctl_sync_sock_size;
1035 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1037 return ipvs->sysctl_pmtu_disc;
1040 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1042 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1043 ipvs->sysctl_backup_only;
1046 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1048 return ipvs->sysctl_conn_reuse_mode;
1051 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1053 return ipvs->sysctl_schedule_icmp;
1056 static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1058 return ipvs->sysctl_ignore_tunneled;
1061 static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1063 return ipvs->sysctl_cache_bypass;
1068 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1070 return DEFAULT_SYNC_THRESHOLD;
1073 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1075 return DEFAULT_SYNC_PERIOD;
1078 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1080 return DEFAULT_SYNC_REFRESH_PERIOD;
1083 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1085 return DEFAULT_SYNC_RETRIES & 3;
1088 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1090 return DEFAULT_SYNC_VER;
1093 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1095 return DEFAULT_SLOPPY_TCP;
1098 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1100 return DEFAULT_SLOPPY_SCTP;
1103 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1108 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1113 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1115 return IPVS_SYNC_QLEN_MAX;
1118 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1123 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1128 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1133 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1138 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1143 static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1148 static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1155 /* IPVS core functions
1156 * (from ip_vs_core.c)
1158 const char *ip_vs_proto_name(unsigned int proto);
1159 void ip_vs_init_hash_table(struct list_head *table, int rows);
1160 struct ip_vs_conn *ip_vs_new_conn_out(struct ip_vs_service *svc,
1161 struct ip_vs_dest *dest,
1162 struct sk_buff *skb,
1163 const struct ip_vs_iphdr *iph,
1166 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1168 #define IP_VS_APP_TYPE_FTP 1
1170 /* ip_vs_conn handling functions
1171 * (from ip_vs_conn.c)
1174 IP_VS_DIR_INPUT = 0,
1176 IP_VS_DIR_INPUT_ONLY,
1180 static inline void ip_vs_conn_fill_param(struct netns_ipvs *ipvs, int af, int protocol,
1181 const union nf_inet_addr *caddr,
1183 const union nf_inet_addr *vaddr,
1185 struct ip_vs_conn_param *p)
1189 p->protocol = protocol;
1198 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1199 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1201 struct ip_vs_conn * ip_vs_conn_in_get_proto(struct netns_ipvs *ipvs, int af,
1202 const struct sk_buff *skb,
1203 const struct ip_vs_iphdr *iph);
1205 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1207 struct ip_vs_conn * ip_vs_conn_out_get_proto(struct netns_ipvs *ipvs, int af,
1208 const struct sk_buff *skb,
1209 const struct ip_vs_iphdr *iph);
1211 /* Get reference to gain full access to conn.
1212 * By default, RCU read-side critical sections have access only to
1213 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1215 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1217 return refcount_inc_not_zero(&cp->refcnt);
1220 /* put back the conn without restarting its timer */
1221 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1223 smp_mb__before_atomic();
1224 refcount_dec(&cp->refcnt);
1226 void ip_vs_conn_put(struct ip_vs_conn *cp);
1227 void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1229 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, int dest_af,
1230 const union nf_inet_addr *daddr,
1231 __be16 dport, unsigned int flags,
1232 struct ip_vs_dest *dest, __u32 fwmark);
1233 void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1235 const char *ip_vs_state_name(__u16 proto, int state);
1237 void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
1238 int ip_vs_check_template(struct ip_vs_conn *ct, struct ip_vs_dest *cdest);
1239 void ip_vs_random_dropentry(struct netns_ipvs *ipvs);
1240 int ip_vs_conn_init(void);
1241 void ip_vs_conn_cleanup(void);
1243 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1245 struct ip_vs_conn *ctl_cp = cp->control;
1247 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1249 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1251 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1257 IP_VS_DBG_BUF(7, "DELeting control for: "
1258 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1259 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1261 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1262 ntohs(ctl_cp->cport));
1265 if (atomic_read(&ctl_cp->n_control) == 0) {
1266 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1268 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1270 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1275 atomic_dec(&ctl_cp->n_control);
1279 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1282 IP_VS_ERR_BUF("request control ADD for already controlled: "
1284 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1286 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1289 ip_vs_control_del(cp);
1292 IP_VS_DBG_BUF(7, "ADDing control for: "
1293 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1294 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1296 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1297 ntohs(ctl_cp->cport));
1299 cp->control = ctl_cp;
1300 atomic_inc(&ctl_cp->n_control);
1303 /* IPVS netns init & cleanup functions */
1304 int ip_vs_estimator_net_init(struct netns_ipvs *ipvs);
1305 int ip_vs_control_net_init(struct netns_ipvs *ipvs);
1306 int ip_vs_protocol_net_init(struct netns_ipvs *ipvs);
1307 int ip_vs_app_net_init(struct netns_ipvs *ipvs);
1308 int ip_vs_conn_net_init(struct netns_ipvs *ipvs);
1309 int ip_vs_sync_net_init(struct netns_ipvs *ipvs);
1310 void ip_vs_conn_net_cleanup(struct netns_ipvs *ipvs);
1311 void ip_vs_app_net_cleanup(struct netns_ipvs *ipvs);
1312 void ip_vs_protocol_net_cleanup(struct netns_ipvs *ipvs);
1313 void ip_vs_control_net_cleanup(struct netns_ipvs *ipvs);
1314 void ip_vs_estimator_net_cleanup(struct netns_ipvs *ipvs);
1315 void ip_vs_sync_net_cleanup(struct netns_ipvs *ipvs);
1316 void ip_vs_service_net_cleanup(struct netns_ipvs *ipvs);
1318 /* IPVS application functions
1319 * (from ip_vs_app.c)
1321 #define IP_VS_APP_MAX_PORTS 8
1322 struct ip_vs_app *register_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1323 void unregister_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1324 int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1325 void ip_vs_unbind_app(struct ip_vs_conn *cp);
1326 int register_ip_vs_app_inc(struct netns_ipvs *ipvs, struct ip_vs_app *app, __u16 proto,
1328 int ip_vs_app_inc_get(struct ip_vs_app *inc);
1329 void ip_vs_app_inc_put(struct ip_vs_app *inc);
1331 int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1332 int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
1334 int register_ip_vs_pe(struct ip_vs_pe *pe);
1335 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1336 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1337 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1339 /* Use a #define to avoid all of module.h just for these trivial ops */
1340 #define ip_vs_pe_get(pe) \
1341 if (pe && pe->module) \
1342 __module_get(pe->module);
1344 #define ip_vs_pe_put(pe) \
1345 if (pe && pe->module) \
1346 module_put(pe->module);
1348 /* IPVS protocol functions (from ip_vs_proto.c) */
1349 int ip_vs_protocol_init(void);
1350 void ip_vs_protocol_cleanup(void);
1351 void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1352 int *ip_vs_create_timeout_table(int *table, int size);
1353 void ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1354 const struct sk_buff *skb, int offset,
1357 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1358 extern struct ip_vs_protocol ip_vs_protocol_udp;
1359 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1360 extern struct ip_vs_protocol ip_vs_protocol_esp;
1361 extern struct ip_vs_protocol ip_vs_protocol_ah;
1362 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1364 /* Registering/unregistering scheduler functions
1365 * (from ip_vs_sched.c)
1367 int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1368 int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1369 int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1370 struct ip_vs_scheduler *scheduler);
1371 void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1372 struct ip_vs_scheduler *sched);
1373 struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1374 void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1376 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1377 struct ip_vs_proto_data *pd, int *ignored,
1378 struct ip_vs_iphdr *iph);
1379 int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1380 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1382 void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1384 /* IPVS control data and functions (from ip_vs_ctl.c) */
1385 extern struct ip_vs_stats ip_vs_stats;
1386 extern int sysctl_ip_vs_sync_ver;
1388 struct ip_vs_service *
1389 ip_vs_service_find(struct netns_ipvs *ipvs, int af, __u32 fwmark, __u16 protocol,
1390 const union nf_inet_addr *vaddr, __be16 vport);
1392 bool ip_vs_has_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1393 const union nf_inet_addr *daddr, __be16 dport);
1396 ip_vs_find_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1397 const union nf_inet_addr *daddr, __be16 dport);
1399 int ip_vs_use_count_inc(void);
1400 void ip_vs_use_count_dec(void);
1401 int ip_vs_register_nl_ioctl(void);
1402 void ip_vs_unregister_nl_ioctl(void);
1403 int ip_vs_control_init(void);
1404 void ip_vs_control_cleanup(void);
1406 ip_vs_find_dest(struct netns_ipvs *ipvs, int svc_af, int dest_af,
1407 const union nf_inet_addr *daddr, __be16 dport,
1408 const union nf_inet_addr *vaddr, __be16 vport,
1409 __u16 protocol, __u32 fwmark, __u32 flags);
1410 void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1412 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1414 refcount_inc(&dest->refcnt);
1417 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1419 smp_mb__before_atomic();
1420 refcount_dec(&dest->refcnt);
1423 static inline void ip_vs_dest_put_and_free(struct ip_vs_dest *dest)
1425 if (refcount_dec_and_test(&dest->refcnt))
1429 /* IPVS sync daemon data and function prototypes
1430 * (from ip_vs_sync.c)
1432 int start_sync_thread(struct netns_ipvs *ipvs, struct ipvs_sync_daemon_cfg *cfg,
1434 int stop_sync_thread(struct netns_ipvs *ipvs, int state);
1435 void ip_vs_sync_conn(struct netns_ipvs *ipvs, struct ip_vs_conn *cp, int pkts);
1437 /* IPVS rate estimator prototypes (from ip_vs_est.c) */
1438 void ip_vs_start_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1439 void ip_vs_stop_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1440 void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1441 void ip_vs_read_estimator(struct ip_vs_kstats *dst, struct ip_vs_stats *stats);
1443 /* Various IPVS packet transmitters (from ip_vs_xmit.c) */
1444 int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1445 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1446 int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1447 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1448 int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1449 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1450 int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1451 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1452 int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1453 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1454 int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1455 struct ip_vs_protocol *pp, int offset,
1456 unsigned int hooknum, struct ip_vs_iphdr *iph);
1457 void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1459 #ifdef CONFIG_IP_VS_IPV6
1460 int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1461 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1462 int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1463 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1464 int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1465 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1466 int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1467 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1468 int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1469 struct ip_vs_protocol *pp, int offset,
1470 unsigned int hooknum, struct ip_vs_iphdr *iph);
1473 #ifdef CONFIG_SYSCTL
1474 /* This is a simple mechanism to ignore packets when
1475 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1476 * we start to drop 1/rate of the packets
1478 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1480 if (!ipvs->drop_rate)
1482 if (--ipvs->drop_counter > 0)
1484 ipvs->drop_counter = ipvs->drop_rate;
1488 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1491 /* ip_vs_fwd_tag returns the forwarding tag of the connection */
1492 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1494 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1498 switch (IP_VS_FWD_METHOD(cp)) {
1499 case IP_VS_CONN_F_MASQ:
1501 case IP_VS_CONN_F_LOCALNODE:
1503 case IP_VS_CONN_F_TUNNEL:
1505 case IP_VS_CONN_F_DROUTE:
1507 case IP_VS_CONN_F_BYPASS:
1515 void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1516 struct ip_vs_conn *cp, int dir);
1518 #ifdef CONFIG_IP_VS_IPV6
1519 void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1520 struct ip_vs_conn *cp, int dir);
1523 __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1525 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1527 __be32 diff[2] = { ~old, new };
1529 return csum_partial(diff, sizeof(diff), oldsum);
1532 #ifdef CONFIG_IP_VS_IPV6
1533 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1536 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1537 new[3], new[2], new[1], new[0] };
1539 return csum_partial(diff, sizeof(diff), oldsum);
1543 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1545 __be16 diff[2] = { ~old, new };
1547 return csum_partial(diff, sizeof(diff), oldsum);
1550 /* Forget current conntrack (unconfirmed) and attach notrack entry */
1551 static inline void ip_vs_notrack(struct sk_buff *skb)
1553 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1554 enum ip_conntrack_info ctinfo;
1555 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1558 nf_conntrack_put(&ct->ct_general);
1559 nf_ct_set(skb, NULL, IP_CT_UNTRACKED);
1564 #ifdef CONFIG_IP_VS_NFCT
1565 /* Netfilter connection tracking
1566 * (from ip_vs_nfct.c)
1568 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1570 #ifdef CONFIG_SYSCTL
1571 return ipvs->sysctl_conntrack;
1577 void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1579 int ip_vs_confirm_conntrack(struct sk_buff *skb);
1580 void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1581 struct ip_vs_conn *cp, u_int8_t proto,
1582 const __be16 port, int from_rs);
1583 void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1587 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1592 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1593 struct ip_vs_conn *cp, int outin)
1597 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1602 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1605 #endif /* CONFIG_IP_VS_NFCT */
1607 /* Really using conntrack? */
1608 static inline bool ip_vs_conn_uses_conntrack(struct ip_vs_conn *cp,
1609 struct sk_buff *skb)
1611 #ifdef CONFIG_IP_VS_NFCT
1612 enum ip_conntrack_info ctinfo;
1615 if (!(cp->flags & IP_VS_CONN_F_NFCT))
1617 ct = nf_ct_get(skb, &ctinfo);
1625 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1627 /* We think the overhead of processing active connections is 256
1628 * times higher than that of inactive connections in average. (This
1629 * 256 times might not be accurate, we will change it later) We
1630 * use the following formula to estimate the overhead now:
1631 * dest->activeconns*256 + dest->inactconns
1633 return (atomic_read(&dest->activeconns) << 8) +
1634 atomic_read(&dest->inactconns);
1637 #endif /* _NET_IP_VS_H */