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 /* Connections' size value needed by ip_vs_ctl.c */
45 extern int ip_vs_conn_tab_size;
48 int hdr_flags; /* ipvs flags */
49 __u32 off; /* Where IP or IPv4 header starts */
50 __u32 len; /* IPv4 simply where L4 starts
51 * IPv6 where L4 Transport Header starts */
52 __u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
55 union nf_inet_addr saddr;
56 union nf_inet_addr daddr;
59 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
60 int len, void *buffer)
62 return skb_header_pointer(skb, offset, len, buffer);
65 /* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
66 * IPv6 requires some extra work, as finding proper header position,
67 * depend on the IPv6 extension headers.
70 ip_vs_fill_iph_skb_off(int af, const struct sk_buff *skb, int offset,
71 int hdr_flags, struct ip_vs_iphdr *iphdr)
73 iphdr->hdr_flags = hdr_flags;
76 #ifdef CONFIG_IP_VS_IPV6
79 const struct ipv6hdr *iph = skb_header_pointer(
80 skb, offset, sizeof(_iph), &_iph);
84 iphdr->saddr.in6 = iph->saddr;
85 iphdr->daddr.in6 = iph->daddr;
86 /* ipv6_find_hdr() updates len, flags */
89 iphdr->protocol = ipv6_find_hdr(skb, &iphdr->len, -1,
92 if (iphdr->protocol < 0)
98 const struct iphdr *iph = skb_header_pointer(
99 skb, offset, sizeof(_iph), &_iph);
103 iphdr->len = offset + iph->ihl * 4;
105 iphdr->protocol = iph->protocol;
106 iphdr->saddr.ip = iph->saddr;
107 iphdr->daddr.ip = iph->daddr;
114 ip_vs_fill_iph_skb_icmp(int af, const struct sk_buff *skb, int offset,
115 bool inverse, struct ip_vs_iphdr *iphdr)
117 int hdr_flags = IP_VS_HDR_ICMP;
120 hdr_flags |= IP_VS_HDR_INVERSE;
122 return ip_vs_fill_iph_skb_off(af, skb, offset, hdr_flags, iphdr);
126 ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, bool inverse,
127 struct ip_vs_iphdr *iphdr)
132 hdr_flags |= IP_VS_HDR_INVERSE;
134 return ip_vs_fill_iph_skb_off(af, skb, skb_network_offset(skb),
139 ip_vs_iph_inverse(const struct ip_vs_iphdr *iph)
141 return !!(iph->hdr_flags & IP_VS_HDR_INVERSE);
145 ip_vs_iph_icmp(const struct ip_vs_iphdr *iph)
147 return !!(iph->hdr_flags & IP_VS_HDR_ICMP);
150 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
151 const union nf_inet_addr *src)
153 #ifdef CONFIG_IP_VS_IPV6
161 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
162 const union nf_inet_addr *src)
164 #ifdef CONFIG_IP_VS_IPV6
165 if (af == AF_INET6) {
176 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
177 const union nf_inet_addr *b)
179 #ifdef CONFIG_IP_VS_IPV6
181 return ipv6_addr_equal(&a->in6, &b->in6);
183 return a->ip == b->ip;
186 #ifdef CONFIG_IP_VS_DEBUG
187 #include <linux/net.h>
189 int ip_vs_get_debug_level(void);
191 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
192 const union nf_inet_addr *addr,
196 #ifdef CONFIG_IP_VS_IPV6
198 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
202 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
206 BUG_ON(*idx > buf_len + 1);
207 return &buf[*idx - len];
210 #define IP_VS_DBG_BUF(level, msg, ...) \
212 char ip_vs_dbg_buf[160]; \
213 int ip_vs_dbg_idx = 0; \
214 if (level <= ip_vs_get_debug_level()) \
215 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
217 #define IP_VS_ERR_BUF(msg...) \
219 char ip_vs_dbg_buf[160]; \
220 int ip_vs_dbg_idx = 0; \
224 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
225 #define IP_VS_DBG_ADDR(af, addr) \
226 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
227 sizeof(ip_vs_dbg_buf), addr, \
230 #define IP_VS_DBG(level, msg, ...) \
232 if (level <= ip_vs_get_debug_level()) \
233 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
235 #define IP_VS_DBG_RL(msg, ...) \
237 if (net_ratelimit()) \
238 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
240 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
242 if (level <= ip_vs_get_debug_level()) \
243 pp->debug_packet(af, pp, skb, ofs, msg); \
245 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
247 if (level <= ip_vs_get_debug_level() && \
249 pp->debug_packet(af, pp, skb, ofs, msg); \
251 #else /* NO DEBUGGING at ALL */
252 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
253 #define IP_VS_ERR_BUF(msg...) do {} while (0)
254 #define IP_VS_DBG(level, msg...) do {} while (0)
255 #define IP_VS_DBG_RL(msg...) do {} while (0)
256 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
257 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
260 #define IP_VS_BUG() BUG()
261 #define IP_VS_ERR_RL(msg, ...) \
263 if (net_ratelimit()) \
264 pr_err(msg, ##__VA_ARGS__); \
267 #ifdef CONFIG_IP_VS_DEBUG
268 #define EnterFunction(level) \
270 if (level <= ip_vs_get_debug_level()) \
272 pr_fmt("Enter: %s, %s line %i\n"), \
273 __func__, __FILE__, __LINE__); \
275 #define LeaveFunction(level) \
277 if (level <= ip_vs_get_debug_level()) \
279 pr_fmt("Leave: %s, %s line %i\n"), \
280 __func__, __FILE__, __LINE__); \
283 #define EnterFunction(level) do {} while (0)
284 #define LeaveFunction(level) do {} while (0)
287 /* The port number of FTP service (in network order). */
288 #define FTPPORT cpu_to_be16(21)
289 #define FTPDATA cpu_to_be16(20)
291 /* TCP State Values */
293 IP_VS_TCP_S_NONE = 0,
294 IP_VS_TCP_S_ESTABLISHED,
295 IP_VS_TCP_S_SYN_SENT,
296 IP_VS_TCP_S_SYN_RECV,
297 IP_VS_TCP_S_FIN_WAIT,
298 IP_VS_TCP_S_TIME_WAIT,
300 IP_VS_TCP_S_CLOSE_WAIT,
301 IP_VS_TCP_S_LAST_ACK,
307 /* UDP State Values */
313 /* ICMP State Values */
319 /* SCTP State Values */
320 enum ip_vs_sctp_states {
324 IP_VS_SCTP_S_COOKIE_SENT,
325 IP_VS_SCTP_S_COOKIE_REPLIED,
326 IP_VS_SCTP_S_COOKIE_WAIT,
328 IP_VS_SCTP_S_COOKIE_ECHOED,
329 IP_VS_SCTP_S_ESTABLISHED,
330 IP_VS_SCTP_S_SHUTDOWN_SENT,
331 IP_VS_SCTP_S_SHUTDOWN_RECEIVED,
332 IP_VS_SCTP_S_SHUTDOWN_ACK_SENT,
333 IP_VS_SCTP_S_REJECTED,
338 /* Connection templates use bits from state */
339 #define IP_VS_CTPL_S_NONE 0x0000
340 #define IP_VS_CTPL_S_ASSURED 0x0001
341 #define IP_VS_CTPL_S_LAST 0x0002
343 /* Delta sequence info structure
344 * Each ip_vs_conn has 2 (output AND input seq. changes).
345 * Only used in the VS/NAT.
348 __u32 init_seq; /* Add delta from this seq */
349 __u32 delta; /* Delta in sequence numbers */
350 __u32 previous_delta; /* Delta in sequence numbers
351 * before last resized pkt */
354 /* counters per cpu */
355 struct ip_vs_counters {
356 __u64 conns; /* connections scheduled */
357 __u64 inpkts; /* incoming packets */
358 __u64 outpkts; /* outgoing packets */
359 __u64 inbytes; /* incoming bytes */
360 __u64 outbytes; /* outgoing bytes */
363 struct ip_vs_cpu_stats {
364 struct ip_vs_counters cnt;
365 struct u64_stats_sync syncp;
368 /* IPVS statistics objects */
369 struct ip_vs_estimator {
370 struct list_head list;
386 * IPVS statistics object, 64-bit kernel version of struct ip_vs_stats_user
388 struct ip_vs_kstats {
389 u64 conns; /* connections scheduled */
390 u64 inpkts; /* incoming packets */
391 u64 outpkts; /* outgoing packets */
392 u64 inbytes; /* incoming bytes */
393 u64 outbytes; /* outgoing bytes */
395 u64 cps; /* current connection rate */
396 u64 inpps; /* current in packet rate */
397 u64 outpps; /* current out packet rate */
398 u64 inbps; /* current in byte rate */
399 u64 outbps; /* current out byte rate */
403 struct ip_vs_kstats kstats; /* kernel statistics */
404 struct ip_vs_estimator est; /* estimator */
405 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
406 spinlock_t lock; /* spin lock */
407 struct ip_vs_kstats kstats0; /* reset values */
415 struct ip_vs_proto_data;
417 struct ip_vs_protocol {
418 struct ip_vs_protocol *next;
424 void (*init)(struct ip_vs_protocol *pp);
426 void (*exit)(struct ip_vs_protocol *pp);
428 int (*init_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
430 void (*exit_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
432 int (*conn_schedule)(struct netns_ipvs *ipvs,
433 int af, struct sk_buff *skb,
434 struct ip_vs_proto_data *pd,
435 int *verdict, struct ip_vs_conn **cpp,
436 struct ip_vs_iphdr *iph);
439 (*conn_in_get)(struct netns_ipvs *ipvs,
441 const struct sk_buff *skb,
442 const struct ip_vs_iphdr *iph);
445 (*conn_out_get)(struct netns_ipvs *ipvs,
447 const struct sk_buff *skb,
448 const struct ip_vs_iphdr *iph);
450 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
451 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
453 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
454 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
456 const char *(*state_name)(int state);
458 void (*state_transition)(struct ip_vs_conn *cp, int direction,
459 const struct sk_buff *skb,
460 struct ip_vs_proto_data *pd);
462 int (*register_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
464 void (*unregister_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
466 int (*app_conn_bind)(struct ip_vs_conn *cp);
468 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
469 const struct sk_buff *skb,
473 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
476 /* protocol data per netns */
477 struct ip_vs_proto_data {
478 struct ip_vs_proto_data *next;
479 struct ip_vs_protocol *pp;
480 int *timeout_table; /* protocol timeout table */
481 atomic_t appcnt; /* counter of proto app incs. */
482 struct tcp_states_t *tcp_state_table;
485 struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
486 struct ip_vs_proto_data *ip_vs_proto_data_get(struct netns_ipvs *ipvs,
487 unsigned short proto);
489 struct ip_vs_conn_param {
490 struct netns_ipvs *ipvs;
491 const union nf_inet_addr *caddr;
492 const union nf_inet_addr *vaddr;
498 const struct ip_vs_pe *pe;
503 /* IP_VS structure allocated for each dynamically scheduled connection */
505 struct hlist_node c_list; /* hashed list heads */
506 /* Protocol, addresses and port numbers */
510 u16 af; /* address family */
511 union nf_inet_addr caddr; /* client address */
512 union nf_inet_addr vaddr; /* virtual address */
513 union nf_inet_addr daddr; /* destination address */
514 volatile __u32 flags; /* status flags */
515 __u16 protocol; /* Which protocol (TCP/UDP) */
516 __u16 daf; /* Address family of the dest */
517 struct netns_ipvs *ipvs;
519 /* counter and timer */
520 refcount_t refcnt; /* reference count */
521 struct timer_list timer; /* Expiration timer */
522 volatile unsigned long timeout; /* timeout */
524 /* Flags and state transition */
525 spinlock_t lock; /* lock for state transition */
526 volatile __u16 state; /* state info */
527 volatile __u16 old_state; /* old state, to be used for
528 * state transition triggerd
531 __u32 fwmark; /* Fire wall mark from skb */
532 unsigned long sync_endtime; /* jiffies + sent_retries */
534 /* Control members */
535 struct ip_vs_conn *control; /* Master control connection */
536 atomic_t n_control; /* Number of controlled ones */
537 struct ip_vs_dest *dest; /* real server */
538 atomic_t in_pkts; /* incoming packet counter */
540 /* Packet transmitter for different forwarding methods. If it
541 * mangles the packet, it must return NF_DROP or better NF_STOLEN,
542 * otherwise this must be changed to a sk_buff **.
543 * NF_ACCEPT can be returned when destination is local.
545 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
546 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
548 /* Note: we can group the following members into a structure,
549 * in order to save more space, and the following members are
550 * only used in VS/NAT anyway
552 struct ip_vs_app *app; /* bound ip_vs_app object */
553 void *app_data; /* Application private data */
554 struct ip_vs_seq in_seq; /* incoming seq. struct */
555 struct ip_vs_seq out_seq; /* outgoing seq. struct */
557 const struct ip_vs_pe *pe;
561 struct rcu_head rcu_head;
564 /* Extended internal versions of struct ip_vs_service_user and ip_vs_dest_user
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 int flags; /* virtual service flags */
583 unsigned int timeout; /* persistent timeout in sec */
584 __be32 netmask; /* persistent netmask or plen */
588 struct ip_vs_dest_user_kern {
589 /* destination server address */
590 union nf_inet_addr addr;
593 /* real server options */
594 unsigned int 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 */
601 /* Address family of addr */
604 u16 tun_type; /* tunnel type */
605 __be16 tun_port; /* tunnel port */
610 * The information about the virtual service offered to the net and the
611 * forwarding entries.
613 struct ip_vs_service {
614 struct hlist_node s_list; /* for normal service table */
615 struct hlist_node f_list; /* for fwmark-based service table */
616 atomic_t refcnt; /* reference counter */
618 u16 af; /* address family */
619 __u16 protocol; /* which protocol (TCP/UDP) */
620 union nf_inet_addr addr; /* IP address for virtual service */
621 __be16 port; /* port number for the service */
622 __u32 fwmark; /* firewall mark of the service */
623 unsigned int flags; /* service status flags */
624 unsigned int timeout; /* persistent timeout in ticks */
625 __be32 netmask; /* grouping granularity, mask/plen */
626 struct netns_ipvs *ipvs;
628 struct list_head destinations; /* real server d-linked list */
629 __u32 num_dests; /* number of servers */
630 struct ip_vs_stats stats; /* statistics for the service */
633 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
634 spinlock_t sched_lock; /* lock sched_data */
635 void *sched_data; /* scheduler application data */
637 /* alternate persistence engine */
638 struct ip_vs_pe __rcu *pe;
639 int conntrack_afmask;
641 struct rcu_head rcu_head;
644 /* Information for cached dst */
645 struct ip_vs_dest_dst {
646 struct dst_entry *dst_cache; /* destination cache entry */
648 union nf_inet_addr dst_saddr;
649 struct rcu_head rcu_head;
652 /* The real server destination forwarding entry with ip address, port number,
656 struct list_head n_list; /* for the dests in the service */
657 struct hlist_node d_list; /* for table with all the dests */
659 u16 af; /* address family */
660 __be16 port; /* port number of the server */
661 union nf_inet_addr addr; /* IP address of the server */
662 volatile unsigned int flags; /* dest status flags */
663 atomic_t conn_flags; /* flags to copy to conn */
664 atomic_t weight; /* server weight */
665 atomic_t last_weight; /* server latest weight */
666 __u16 tun_type; /* tunnel type */
667 __be16 tun_port; /* tunnel port */
669 refcount_t refcnt; /* reference counter */
670 struct ip_vs_stats stats; /* statistics */
671 unsigned long idle_start; /* start time, jiffies */
673 /* connection counters and thresholds */
674 atomic_t activeconns; /* active connections */
675 atomic_t inactconns; /* inactive connections */
676 atomic_t persistconns; /* persistent connections */
677 __u32 u_threshold; /* upper threshold */
678 __u32 l_threshold; /* lower threshold */
680 /* for destination cache */
681 spinlock_t dst_lock; /* lock of dst_cache */
682 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
684 /* for virtual service */
685 struct ip_vs_service __rcu *svc; /* service it belongs to */
686 __u16 protocol; /* which protocol (TCP/UDP) */
687 __be16 vport; /* virtual port number */
688 union nf_inet_addr vaddr; /* virtual IP address */
689 __u32 vfwmark; /* firewall mark of service */
691 struct list_head t_list; /* in dest_trash */
692 unsigned int in_rs_table:1; /* we are in rs_table */
695 /* The scheduler object */
696 struct ip_vs_scheduler {
697 struct list_head n_list; /* d-linked list head */
698 char *name; /* scheduler name */
699 atomic_t refcnt; /* reference counter */
700 struct module *module; /* THIS_MODULE/NULL */
702 /* scheduler initializing service */
703 int (*init_service)(struct ip_vs_service *svc);
704 /* scheduling service finish */
705 void (*done_service)(struct ip_vs_service *svc);
707 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
708 /* dest is unlinked */
709 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
710 /* dest is updated */
711 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
713 /* selecting a server from the given service */
714 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
715 const struct sk_buff *skb,
716 struct ip_vs_iphdr *iph);
719 /* The persistence engine object */
721 struct list_head n_list; /* d-linked list head */
722 char *name; /* scheduler name */
723 atomic_t refcnt; /* reference counter */
724 struct module *module; /* THIS_MODULE/NULL */
726 /* get the connection template, if any */
727 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
728 bool (*ct_match)(const struct ip_vs_conn_param *p,
729 struct ip_vs_conn *ct);
730 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
732 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
733 /* create connections for real-server outgoing packets */
734 struct ip_vs_conn* (*conn_out)(struct ip_vs_service *svc,
735 struct ip_vs_dest *dest,
737 const struct ip_vs_iphdr *iph,
738 __be16 dport, __be16 cport);
741 /* The application module object (a.k.a. app incarnation) */
743 struct list_head a_list; /* member in app list */
744 int type; /* IP_VS_APP_TYPE_xxx */
745 char *name; /* application module name */
747 struct module *module; /* THIS_MODULE/NULL */
748 struct list_head incs_list; /* list of incarnations */
750 /* members for application incarnations */
751 struct list_head p_list; /* member in proto app list */
752 struct ip_vs_app *app; /* its real application */
753 __be16 port; /* port number in net order */
754 atomic_t usecnt; /* usage counter */
755 struct rcu_head rcu_head;
757 /* output hook: Process packet in inout direction, diff set for TCP.
758 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
759 * 2=Mangled but checksum was not updated
761 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
762 struct sk_buff *, int *diff, struct ip_vs_iphdr *ipvsh);
764 /* input hook: Process packet in outin direction, diff set for TCP.
765 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
766 * 2=Mangled but checksum was not updated
768 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
769 struct sk_buff *, int *diff, struct ip_vs_iphdr *ipvsh);
771 /* ip_vs_app initializer */
772 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
774 /* ip_vs_app finish */
775 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
779 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
780 struct ip_vs_protocol *);
782 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
788 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
789 int *verdict, struct ip_vs_conn **cpp);
792 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
793 const struct iphdr *iph, int inverse);
796 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
797 const struct iphdr *iph, int inverse);
799 int (*state_transition)(struct ip_vs_conn *cp, int direction,
800 const struct sk_buff *skb,
801 struct ip_vs_app *app);
803 void (*timeout_change)(struct ip_vs_app *app, int flags);
806 struct ipvs_master_sync_state {
807 struct list_head sync_queue;
808 struct ip_vs_sync_buff *sync_buff;
809 unsigned long sync_queue_len;
810 unsigned int sync_queue_delay;
811 struct delayed_work master_wakeup_work;
812 struct netns_ipvs *ipvs;
815 struct ip_vs_sync_thread_data;
817 /* How much time to keep dests in trash */
818 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
820 struct ipvs_sync_daemon_cfg {
821 union nf_inet_addr mcast_group;
827 /* multicast interface name */
828 char mcast_ifn[IP_VS_IFNAME_MAXLEN];
831 /* IPVS in network namespace */
833 int gen; /* Generation */
834 int enable; /* enable like nf_hooks do */
835 /* Hash table: for real service lookups */
836 #define IP_VS_RTAB_BITS 4
837 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
838 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
840 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
842 struct list_head app_list;
844 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
845 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
846 /* ip_vs_proto_tcp */
847 #ifdef CONFIG_IP_VS_PROTO_TCP
848 #define TCP_APP_TAB_BITS 4
849 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
850 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
851 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
853 /* ip_vs_proto_udp */
854 #ifdef CONFIG_IP_VS_PROTO_UDP
855 #define UDP_APP_TAB_BITS 4
856 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
857 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
858 struct list_head udp_apps[UDP_APP_TAB_SIZE];
860 /* ip_vs_proto_sctp */
861 #ifdef CONFIG_IP_VS_PROTO_SCTP
862 #define SCTP_APP_TAB_BITS 4
863 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
864 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
865 /* Hash table for SCTP application incarnations */
866 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
869 atomic_t conn_count; /* connection counter */
872 struct ip_vs_stats tot_stats; /* Statistics & est. */
874 int num_services; /* no of virtual services */
876 /* Trash for destinations */
877 struct list_head dest_trash;
878 spinlock_t dest_trash_lock;
879 struct timer_list dest_trash_timer; /* expiration timer */
880 /* Service counters */
881 atomic_t ftpsvc_counter;
882 atomic_t nullsvc_counter;
883 atomic_t conn_out_counter;
886 /* 1/rate drop and drop-entry variables */
887 struct delayed_work defense_work; /* Work handler */
892 spinlock_t dropentry_lock; /* drop entry handling */
893 spinlock_t droppacket_lock; /* drop packet handling */
894 spinlock_t securetcp_lock; /* state and timeout tables */
897 struct ctl_table_header *sysctl_hdr;
898 struct ctl_table *sysctl_tbl;
901 /* sysctl variables */
902 int sysctl_amemthresh;
903 int sysctl_am_droprate;
904 int sysctl_drop_entry;
905 int sysctl_drop_packet;
906 int sysctl_secure_tcp;
907 #ifdef CONFIG_IP_VS_NFCT
908 int sysctl_conntrack;
910 int sysctl_snat_reroute;
912 int sysctl_sync_ports;
913 int sysctl_sync_persist_mode;
914 unsigned long sysctl_sync_qlen_max;
915 int sysctl_sync_sock_size;
916 int sysctl_cache_bypass;
917 int sysctl_expire_nodest_conn;
918 int sysctl_sloppy_tcp;
919 int sysctl_sloppy_sctp;
920 int sysctl_expire_quiescent_template;
921 int sysctl_sync_threshold[2];
922 unsigned int sysctl_sync_refresh_period;
923 int sysctl_sync_retries;
924 int sysctl_nat_icmp_send;
925 int sysctl_pmtu_disc;
926 int sysctl_backup_only;
927 int sysctl_conn_reuse_mode;
928 int sysctl_schedule_icmp;
929 int sysctl_ignore_tunneled;
932 int sysctl_lblc_expiration;
933 struct ctl_table_header *lblc_ctl_header;
934 struct ctl_table *lblc_ctl_table;
936 int sysctl_lblcr_expiration;
937 struct ctl_table_header *lblcr_ctl_header;
938 struct ctl_table *lblcr_ctl_table;
940 struct list_head est_list; /* estimator list */
942 struct timer_list est_timer; /* Estimation timer */
944 spinlock_t sync_lock;
945 struct ipvs_master_sync_state *ms;
946 spinlock_t sync_buff_lock;
947 struct ip_vs_sync_thread_data *master_tinfo;
948 struct ip_vs_sync_thread_data *backup_tinfo;
950 volatile int sync_state;
951 struct mutex sync_mutex;
952 struct ipvs_sync_daemon_cfg mcfg; /* Master Configuration */
953 struct ipvs_sync_daemon_cfg bcfg; /* Backup Configuration */
954 /* net name space ptr */
955 struct net *net; /* Needed by timer routines */
956 /* Number of heterogeneous destinations, needed becaus heterogeneous
957 * are not supported when synchronization is enabled.
959 unsigned int mixed_address_family_dests;
962 #define DEFAULT_SYNC_THRESHOLD 3
963 #define DEFAULT_SYNC_PERIOD 50
964 #define DEFAULT_SYNC_VER 1
965 #define DEFAULT_SLOPPY_TCP 0
966 #define DEFAULT_SLOPPY_SCTP 0
967 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
968 #define DEFAULT_SYNC_RETRIES 0
969 #define IPVS_SYNC_WAKEUP_RATE 8
970 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
971 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
972 #define IPVS_SYNC_CHECK_PERIOD HZ
973 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
974 #define IPVS_SYNC_PORTS_MAX (1 << 6)
978 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
980 return ipvs->sysctl_sync_threshold[0];
983 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
985 return READ_ONCE(ipvs->sysctl_sync_threshold[1]);
988 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
990 return READ_ONCE(ipvs->sysctl_sync_refresh_period);
993 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
995 return ipvs->sysctl_sync_retries;
998 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1000 return ipvs->sysctl_sync_ver;
1003 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1005 return ipvs->sysctl_sloppy_tcp;
1008 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1010 return ipvs->sysctl_sloppy_sctp;
1013 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1015 return READ_ONCE(ipvs->sysctl_sync_ports);
1018 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1020 return ipvs->sysctl_sync_persist_mode;
1023 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1025 return ipvs->sysctl_sync_qlen_max;
1028 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1030 return ipvs->sysctl_sync_sock_size;
1033 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1035 return ipvs->sysctl_pmtu_disc;
1038 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1040 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1041 ipvs->sysctl_backup_only;
1044 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1046 return ipvs->sysctl_conn_reuse_mode;
1049 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1051 return ipvs->sysctl_schedule_icmp;
1054 static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1056 return ipvs->sysctl_ignore_tunneled;
1059 static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1061 return ipvs->sysctl_cache_bypass;
1066 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1068 return DEFAULT_SYNC_THRESHOLD;
1071 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1073 return DEFAULT_SYNC_PERIOD;
1076 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1078 return DEFAULT_SYNC_REFRESH_PERIOD;
1081 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1083 return DEFAULT_SYNC_RETRIES & 3;
1086 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1088 return DEFAULT_SYNC_VER;
1091 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1093 return DEFAULT_SLOPPY_TCP;
1096 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1098 return DEFAULT_SLOPPY_SCTP;
1101 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1106 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1111 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1113 return IPVS_SYNC_QLEN_MAX;
1116 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1121 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1126 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1131 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1136 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1141 static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1146 static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1153 /* IPVS core functions
1154 * (from ip_vs_core.c)
1156 const char *ip_vs_proto_name(unsigned int proto);
1157 void ip_vs_init_hash_table(struct list_head *table, int rows);
1158 struct ip_vs_conn *ip_vs_new_conn_out(struct ip_vs_service *svc,
1159 struct ip_vs_dest *dest,
1160 struct sk_buff *skb,
1161 const struct ip_vs_iphdr *iph,
1164 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1166 #define IP_VS_APP_TYPE_FTP 1
1168 /* ip_vs_conn handling functions
1169 * (from ip_vs_conn.c)
1172 IP_VS_DIR_INPUT = 0,
1174 IP_VS_DIR_INPUT_ONLY,
1178 static inline void ip_vs_conn_fill_param(struct netns_ipvs *ipvs, int af, int protocol,
1179 const union nf_inet_addr *caddr,
1181 const union nf_inet_addr *vaddr,
1183 struct ip_vs_conn_param *p)
1187 p->protocol = protocol;
1196 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1197 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1199 struct ip_vs_conn * ip_vs_conn_in_get_proto(struct netns_ipvs *ipvs, int af,
1200 const struct sk_buff *skb,
1201 const struct ip_vs_iphdr *iph);
1203 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1205 struct ip_vs_conn * ip_vs_conn_out_get_proto(struct netns_ipvs *ipvs, int af,
1206 const struct sk_buff *skb,
1207 const struct ip_vs_iphdr *iph);
1209 /* Get reference to gain full access to conn.
1210 * By default, RCU read-side critical sections have access only to
1211 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1213 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1215 return refcount_inc_not_zero(&cp->refcnt);
1218 /* put back the conn without restarting its timer */
1219 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1221 smp_mb__before_atomic();
1222 refcount_dec(&cp->refcnt);
1224 void ip_vs_conn_put(struct ip_vs_conn *cp);
1225 void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1227 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, int dest_af,
1228 const union nf_inet_addr *daddr,
1229 __be16 dport, unsigned int flags,
1230 struct ip_vs_dest *dest, __u32 fwmark);
1231 void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1233 const char *ip_vs_state_name(const struct ip_vs_conn *cp);
1235 void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
1236 int ip_vs_check_template(struct ip_vs_conn *ct, struct ip_vs_dest *cdest);
1237 void ip_vs_random_dropentry(struct netns_ipvs *ipvs);
1238 int ip_vs_conn_init(void);
1239 void ip_vs_conn_cleanup(void);
1241 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1243 struct ip_vs_conn *ctl_cp = cp->control;
1245 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1247 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1249 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1255 IP_VS_DBG_BUF(7, "DELeting control for: "
1256 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1257 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1259 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1260 ntohs(ctl_cp->cport));
1263 if (atomic_read(&ctl_cp->n_control) == 0) {
1264 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1266 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1268 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1273 atomic_dec(&ctl_cp->n_control);
1277 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1280 IP_VS_ERR_BUF("request control ADD for already controlled: "
1282 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1284 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1287 ip_vs_control_del(cp);
1290 IP_VS_DBG_BUF(7, "ADDing control for: "
1291 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1292 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1294 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1295 ntohs(ctl_cp->cport));
1297 cp->control = ctl_cp;
1298 atomic_inc(&ctl_cp->n_control);
1301 /* Mark our template as assured */
1303 ip_vs_control_assure_ct(struct ip_vs_conn *cp)
1305 struct ip_vs_conn *ct = cp->control;
1307 if (ct && !(ct->state & IP_VS_CTPL_S_ASSURED) &&
1308 (ct->flags & IP_VS_CONN_F_TEMPLATE))
1309 ct->state |= IP_VS_CTPL_S_ASSURED;
1312 /* IPVS netns init & cleanup functions */
1313 int ip_vs_estimator_net_init(struct netns_ipvs *ipvs);
1314 int ip_vs_control_net_init(struct netns_ipvs *ipvs);
1315 int ip_vs_protocol_net_init(struct netns_ipvs *ipvs);
1316 int ip_vs_app_net_init(struct netns_ipvs *ipvs);
1317 int ip_vs_conn_net_init(struct netns_ipvs *ipvs);
1318 int ip_vs_sync_net_init(struct netns_ipvs *ipvs);
1319 void ip_vs_conn_net_cleanup(struct netns_ipvs *ipvs);
1320 void ip_vs_app_net_cleanup(struct netns_ipvs *ipvs);
1321 void ip_vs_protocol_net_cleanup(struct netns_ipvs *ipvs);
1322 void ip_vs_control_net_cleanup(struct netns_ipvs *ipvs);
1323 void ip_vs_estimator_net_cleanup(struct netns_ipvs *ipvs);
1324 void ip_vs_sync_net_cleanup(struct netns_ipvs *ipvs);
1325 void ip_vs_service_net_cleanup(struct netns_ipvs *ipvs);
1327 /* IPVS application functions
1328 * (from ip_vs_app.c)
1330 #define IP_VS_APP_MAX_PORTS 8
1331 struct ip_vs_app *register_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1332 void unregister_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1333 int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1334 void ip_vs_unbind_app(struct ip_vs_conn *cp);
1335 int register_ip_vs_app_inc(struct netns_ipvs *ipvs, struct ip_vs_app *app, __u16 proto,
1337 int ip_vs_app_inc_get(struct ip_vs_app *inc);
1338 void ip_vs_app_inc_put(struct ip_vs_app *inc);
1340 int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb,
1341 struct ip_vs_iphdr *ipvsh);
1342 int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb,
1343 struct ip_vs_iphdr *ipvsh);
1345 int register_ip_vs_pe(struct ip_vs_pe *pe);
1346 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1347 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1348 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1350 /* Use a #define to avoid all of module.h just for these trivial ops */
1351 #define ip_vs_pe_get(pe) \
1352 if (pe && pe->module) \
1353 __module_get(pe->module);
1355 #define ip_vs_pe_put(pe) \
1356 if (pe && pe->module) \
1357 module_put(pe->module);
1359 /* IPVS protocol functions (from ip_vs_proto.c) */
1360 int ip_vs_protocol_init(void);
1361 void ip_vs_protocol_cleanup(void);
1362 void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1363 int *ip_vs_create_timeout_table(int *table, int size);
1364 void ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1365 const struct sk_buff *skb, int offset,
1368 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1369 extern struct ip_vs_protocol ip_vs_protocol_udp;
1370 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1371 extern struct ip_vs_protocol ip_vs_protocol_esp;
1372 extern struct ip_vs_protocol ip_vs_protocol_ah;
1373 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1375 /* Registering/unregistering scheduler functions
1376 * (from ip_vs_sched.c)
1378 int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1379 int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1380 int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1381 struct ip_vs_scheduler *scheduler);
1382 void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1383 struct ip_vs_scheduler *sched);
1384 struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1385 void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1387 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1388 struct ip_vs_proto_data *pd, int *ignored,
1389 struct ip_vs_iphdr *iph);
1390 int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1391 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1393 void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1395 /* IPVS control data and functions (from ip_vs_ctl.c) */
1396 extern struct ip_vs_stats ip_vs_stats;
1397 extern int sysctl_ip_vs_sync_ver;
1399 struct ip_vs_service *
1400 ip_vs_service_find(struct netns_ipvs *ipvs, int af, __u32 fwmark, __u16 protocol,
1401 const union nf_inet_addr *vaddr, __be16 vport);
1403 bool ip_vs_has_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1404 const union nf_inet_addr *daddr, __be16 dport);
1407 ip_vs_find_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1408 const union nf_inet_addr *daddr, __be16 dport);
1410 int ip_vs_use_count_inc(void);
1411 void ip_vs_use_count_dec(void);
1412 int ip_vs_register_nl_ioctl(void);
1413 void ip_vs_unregister_nl_ioctl(void);
1414 int ip_vs_control_init(void);
1415 void ip_vs_control_cleanup(void);
1417 ip_vs_find_dest(struct netns_ipvs *ipvs, int svc_af, int dest_af,
1418 const union nf_inet_addr *daddr, __be16 dport,
1419 const union nf_inet_addr *vaddr, __be16 vport,
1420 __u16 protocol, __u32 fwmark, __u32 flags);
1421 void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1423 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1425 refcount_inc(&dest->refcnt);
1428 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1430 smp_mb__before_atomic();
1431 refcount_dec(&dest->refcnt);
1434 static inline void ip_vs_dest_put_and_free(struct ip_vs_dest *dest)
1436 if (refcount_dec_and_test(&dest->refcnt))
1440 /* IPVS sync daemon data and function prototypes
1441 * (from ip_vs_sync.c)
1443 int start_sync_thread(struct netns_ipvs *ipvs, struct ipvs_sync_daemon_cfg *cfg,
1445 int stop_sync_thread(struct netns_ipvs *ipvs, int state);
1446 void ip_vs_sync_conn(struct netns_ipvs *ipvs, struct ip_vs_conn *cp, int pkts);
1448 /* IPVS rate estimator prototypes (from ip_vs_est.c) */
1449 void ip_vs_start_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1450 void ip_vs_stop_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1451 void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1452 void ip_vs_read_estimator(struct ip_vs_kstats *dst, struct ip_vs_stats *stats);
1454 /* Various IPVS packet transmitters (from ip_vs_xmit.c) */
1455 int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1456 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1457 int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1458 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1459 int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1460 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1461 int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1462 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1463 int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1464 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1465 int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1466 struct ip_vs_protocol *pp, int offset,
1467 unsigned int hooknum, struct ip_vs_iphdr *iph);
1468 void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1470 #ifdef CONFIG_IP_VS_IPV6
1471 int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1472 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1473 int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1474 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1475 int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1476 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1477 int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1478 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1479 int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1480 struct ip_vs_protocol *pp, int offset,
1481 unsigned int hooknum, struct ip_vs_iphdr *iph);
1484 #ifdef CONFIG_SYSCTL
1485 /* This is a simple mechanism to ignore packets when
1486 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1487 * we start to drop 1/rate of the packets
1489 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1491 if (!ipvs->drop_rate)
1493 if (--ipvs->drop_counter > 0)
1495 ipvs->drop_counter = ipvs->drop_rate;
1499 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1502 /* ip_vs_fwd_tag returns the forwarding tag of the connection */
1503 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1505 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1509 switch (IP_VS_FWD_METHOD(cp)) {
1510 case IP_VS_CONN_F_MASQ:
1512 case IP_VS_CONN_F_LOCALNODE:
1514 case IP_VS_CONN_F_TUNNEL:
1516 case IP_VS_CONN_F_DROUTE:
1518 case IP_VS_CONN_F_BYPASS:
1526 void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1527 struct ip_vs_conn *cp, int dir);
1529 #ifdef CONFIG_IP_VS_IPV6
1530 void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1531 struct ip_vs_conn *cp, int dir);
1534 __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1536 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1538 __be32 diff[2] = { ~old, new };
1540 return csum_partial(diff, sizeof(diff), oldsum);
1543 #ifdef CONFIG_IP_VS_IPV6
1544 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1547 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1548 new[3], new[2], new[1], new[0] };
1550 return csum_partial(diff, sizeof(diff), oldsum);
1554 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1556 __be16 diff[2] = { ~old, new };
1558 return csum_partial(diff, sizeof(diff), oldsum);
1561 /* Forget current conntrack (unconfirmed) and attach notrack entry */
1562 static inline void ip_vs_notrack(struct sk_buff *skb)
1564 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1565 enum ip_conntrack_info ctinfo;
1566 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1569 nf_conntrack_put(&ct->ct_general);
1570 nf_ct_set(skb, NULL, IP_CT_UNTRACKED);
1575 #ifdef CONFIG_IP_VS_NFCT
1576 /* Netfilter connection tracking
1577 * (from ip_vs_nfct.c)
1579 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1581 #ifdef CONFIG_SYSCTL
1582 return ipvs->sysctl_conntrack;
1588 void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1590 int ip_vs_confirm_conntrack(struct sk_buff *skb);
1591 void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1592 struct ip_vs_conn *cp, u_int8_t proto,
1593 const __be16 port, int from_rs);
1594 void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1598 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1603 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1604 struct ip_vs_conn *cp, int outin)
1608 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1613 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1616 #endif /* CONFIG_IP_VS_NFCT */
1618 /* Really using conntrack? */
1619 static inline bool ip_vs_conn_uses_conntrack(struct ip_vs_conn *cp,
1620 struct sk_buff *skb)
1622 #ifdef CONFIG_IP_VS_NFCT
1623 enum ip_conntrack_info ctinfo;
1626 if (!(cp->flags & IP_VS_CONN_F_NFCT))
1628 ct = nf_ct_get(skb, &ctinfo);
1635 static inline int ip_vs_register_conntrack(struct ip_vs_service *svc)
1637 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
1638 int afmask = (svc->af == AF_INET6) ? 2 : 1;
1641 if (!(svc->conntrack_afmask & afmask)) {
1642 ret = nf_ct_netns_get(svc->ipvs->net, svc->af);
1644 svc->conntrack_afmask |= afmask;
1652 static inline void ip_vs_unregister_conntrack(struct ip_vs_service *svc)
1654 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
1655 int afmask = (svc->af == AF_INET6) ? 2 : 1;
1657 if (svc->conntrack_afmask & afmask) {
1658 nf_ct_netns_put(svc->ipvs->net, svc->af);
1659 svc->conntrack_afmask &= ~afmask;
1665 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1667 /* We think the overhead of processing active connections is 256
1668 * times higher than that of inactive connections in average. (This
1669 * 256 times might not be accurate, we will change it later) We
1670 * use the following formula to estimate the overhead now:
1671 * dest->activeconns*256 + dest->inactconns
1673 return (atomic_read(&dest->activeconns) << 8) +
1674 atomic_read(&dest->inactconns);
1677 #endif /* _NET_IP_VS_H */