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/compiler.h>
17 #include <linux/timer.h>
18 #include <linux/bug.h>
20 #include <net/checksum.h>
21 #include <linux/netfilter.h> /* for union nf_inet_addr */
23 #include <linux/ipv6.h> /* for struct ipv6hdr */
25 #if IS_ENABLED(CONFIG_IP_VS_IPV6)
26 #include <linux/netfilter_ipv6/ip6_tables.h>
28 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
29 #include <net/netfilter/nf_conntrack.h>
31 #include <net/net_namespace.h> /* Netw namespace */
34 * Generic access of ipvs struct
36 static inline struct netns_ipvs
*net_ipvs(struct net
* net
)
41 * Get net ptr from skb in traffic cases
42 * use skb_sknet when call is from userland (ioctl or netlink)
44 static inline struct net
*skb_net(const struct sk_buff
*skb
)
47 #ifdef CONFIG_IP_VS_DEBUG
49 * This is used for debug only.
50 * Start with the most likely hit
53 if (likely(skb
->dev
&& skb
->dev
->nd_net
))
54 return dev_net(skb
->dev
);
55 if (skb_dst(skb
) && skb_dst(skb
)->dev
)
56 return dev_net(skb_dst(skb
)->dev
);
57 WARN(skb
->sk
, "Maybe skb_sknet should be used in %s() at line:%d\n",
59 if (likely(skb
->sk
&& skb
->sk
->sk_net
))
60 return sock_net(skb
->sk
);
61 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
65 return dev_net(skb
->dev
? : skb_dst(skb
)->dev
);
72 static inline struct net
*skb_sknet(const struct sk_buff
*skb
)
75 #ifdef CONFIG_IP_VS_DEBUG
76 /* Start with the most likely hit */
77 if (likely(skb
->sk
&& skb
->sk
->sk_net
))
78 return sock_net(skb
->sk
);
79 WARN(skb
->dev
, "Maybe skb_net should be used instead in %s() line:%d\n",
81 if (likely(skb
->dev
&& skb
->dev
->nd_net
))
82 return dev_net(skb
->dev
);
83 pr_err("There is no net ptr to find in the skb in %s() line:%d\n",
87 return sock_net(skb
->sk
);
94 * This one needed for single_open_net since net is stored directly in
95 * private not as a struct i.e. seq_file_net can't be used.
97 static inline struct net
*seq_file_single_net(struct seq_file
*seq
)
100 return (struct net
*)seq
->private;
106 /* Connections' size value needed by ip_vs_ctl.c */
107 extern int ip_vs_conn_tab_size
;
110 __u32 len
; /* IPv4 simply where L4 starts
111 IPv6 where L4 Transport Header starts */
112 __u32 thoff_reasm
; /* Transport Header Offset in nfct_reasm skb */
113 __u16 fragoffs
; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
116 union nf_inet_addr saddr
;
117 union nf_inet_addr daddr
;
120 /* Dependency to module: nf_defrag_ipv6 */
121 #if defined(CONFIG_NF_DEFRAG_IPV6) || defined(CONFIG_NF_DEFRAG_IPV6_MODULE)
122 static inline struct sk_buff
*skb_nfct_reasm(const struct sk_buff
*skb
)
124 return skb
->nfct_reasm
;
126 static inline void *frag_safe_skb_hp(const struct sk_buff
*skb
, int offset
,
127 int len
, void *buffer
,
128 const struct ip_vs_iphdr
*ipvsh
)
130 if (unlikely(ipvsh
->fragoffs
&& skb_nfct_reasm(skb
)))
131 return skb_header_pointer(skb_nfct_reasm(skb
),
132 ipvsh
->thoff_reasm
, len
, buffer
);
134 return skb_header_pointer(skb
, offset
, len
, buffer
);
137 static inline struct sk_buff
*skb_nfct_reasm(const struct sk_buff
*skb
)
141 static inline void *frag_safe_skb_hp(const struct sk_buff
*skb
, int offset
,
142 int len
, void *buffer
,
143 const struct ip_vs_iphdr
*ipvsh
)
145 return skb_header_pointer(skb
, offset
, len
, buffer
);
150 ip_vs_fill_ip4hdr(const void *nh
, struct ip_vs_iphdr
*iphdr
)
152 const struct iphdr
*iph
= nh
;
154 iphdr
->len
= iph
->ihl
* 4;
156 iphdr
->protocol
= iph
->protocol
;
157 iphdr
->saddr
.ip
= iph
->saddr
;
158 iphdr
->daddr
.ip
= iph
->daddr
;
161 /* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
162 * IPv6 requires some extra work, as finding proper header position,
163 * depend on the IPv6 extension headers.
166 ip_vs_fill_iph_skb(int af
, const struct sk_buff
*skb
, struct ip_vs_iphdr
*iphdr
)
168 #ifdef CONFIG_IP_VS_IPV6
169 if (af
== AF_INET6
) {
170 const struct ipv6hdr
*iph
=
171 (struct ipv6hdr
*)skb_network_header(skb
);
172 iphdr
->saddr
.in6
= iph
->saddr
;
173 iphdr
->daddr
.in6
= iph
->daddr
;
174 /* ipv6_find_hdr() updates len, flags, thoff_reasm */
175 iphdr
->thoff_reasm
= 0;
178 iphdr
->protocol
= ipv6_find_hdr(skb
, &iphdr
->len
, -1,
181 /* get proto from re-assembled packet and it's offset */
182 if (skb_nfct_reasm(skb
))
183 iphdr
->protocol
= ipv6_find_hdr(skb_nfct_reasm(skb
),
190 const struct iphdr
*iph
=
191 (struct iphdr
*)skb_network_header(skb
);
192 iphdr
->len
= iph
->ihl
* 4;
194 iphdr
->protocol
= iph
->protocol
;
195 iphdr
->saddr
.ip
= iph
->saddr
;
196 iphdr
->daddr
.ip
= iph
->daddr
;
200 /* This function is a faster version of ip_vs_fill_iph_skb().
201 * Where we only populate {s,d}addr (and avoid calling ipv6_find_hdr()).
202 * This is used by the some of the ip_vs_*_schedule() functions.
203 * (Mostly done to avoid ABI breakage of external schedulers)
206 ip_vs_fill_iph_addr_only(int af
, const struct sk_buff
*skb
,
207 struct ip_vs_iphdr
*iphdr
)
209 #ifdef CONFIG_IP_VS_IPV6
210 if (af
== AF_INET6
) {
211 const struct ipv6hdr
*iph
=
212 (struct ipv6hdr
*)skb_network_header(skb
);
213 iphdr
->saddr
.in6
= iph
->saddr
;
214 iphdr
->daddr
.in6
= iph
->daddr
;
218 const struct iphdr
*iph
=
219 (struct iphdr
*)skb_network_header(skb
);
220 iphdr
->saddr
.ip
= iph
->saddr
;
221 iphdr
->daddr
.ip
= iph
->daddr
;
225 static inline void ip_vs_addr_copy(int af
, union nf_inet_addr
*dst
,
226 const union nf_inet_addr
*src
)
228 #ifdef CONFIG_IP_VS_IPV6
236 static inline void ip_vs_addr_set(int af
, union nf_inet_addr
*dst
,
237 const union nf_inet_addr
*src
)
239 #ifdef CONFIG_IP_VS_IPV6
240 if (af
== AF_INET6
) {
251 static inline int ip_vs_addr_equal(int af
, const union nf_inet_addr
*a
,
252 const union nf_inet_addr
*b
)
254 #ifdef CONFIG_IP_VS_IPV6
256 return ipv6_addr_equal(&a
->in6
, &b
->in6
);
258 return a
->ip
== b
->ip
;
261 #ifdef CONFIG_IP_VS_DEBUG
262 #include <linux/net.h>
264 extern int ip_vs_get_debug_level(void);
266 static inline const char *ip_vs_dbg_addr(int af
, char *buf
, size_t buf_len
,
267 const union nf_inet_addr
*addr
,
271 #ifdef CONFIG_IP_VS_IPV6
273 len
= snprintf(&buf
[*idx
], buf_len
- *idx
, "[%pI6c]",
277 len
= snprintf(&buf
[*idx
], buf_len
- *idx
, "%pI4",
281 BUG_ON(*idx
> buf_len
+ 1);
282 return &buf
[*idx
- len
];
285 #define IP_VS_DBG_BUF(level, msg, ...) \
287 char ip_vs_dbg_buf[160]; \
288 int ip_vs_dbg_idx = 0; \
289 if (level <= ip_vs_get_debug_level()) \
290 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
292 #define IP_VS_ERR_BUF(msg...) \
294 char ip_vs_dbg_buf[160]; \
295 int ip_vs_dbg_idx = 0; \
299 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
300 #define IP_VS_DBG_ADDR(af, addr) \
301 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
302 sizeof(ip_vs_dbg_buf), addr, \
305 #define IP_VS_DBG(level, msg, ...) \
307 if (level <= ip_vs_get_debug_level()) \
308 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
310 #define IP_VS_DBG_RL(msg, ...) \
312 if (net_ratelimit()) \
313 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
315 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
317 if (level <= ip_vs_get_debug_level()) \
318 pp->debug_packet(af, pp, skb, ofs, msg); \
320 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
322 if (level <= ip_vs_get_debug_level() && \
324 pp->debug_packet(af, pp, skb, ofs, msg); \
326 #else /* NO DEBUGGING at ALL */
327 #define IP_VS_DBG_BUF(level, msg...) do {} while (0)
328 #define IP_VS_ERR_BUF(msg...) do {} while (0)
329 #define IP_VS_DBG(level, msg...) do {} while (0)
330 #define IP_VS_DBG_RL(msg...) do {} while (0)
331 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
332 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
335 #define IP_VS_BUG() BUG()
336 #define IP_VS_ERR_RL(msg, ...) \
338 if (net_ratelimit()) \
339 pr_err(msg, ##__VA_ARGS__); \
342 #ifdef CONFIG_IP_VS_DEBUG
343 #define EnterFunction(level) \
345 if (level <= ip_vs_get_debug_level()) \
347 pr_fmt("Enter: %s, %s line %i\n"), \
348 __func__, __FILE__, __LINE__); \
350 #define LeaveFunction(level) \
352 if (level <= ip_vs_get_debug_level()) \
354 pr_fmt("Leave: %s, %s line %i\n"), \
355 __func__, __FILE__, __LINE__); \
358 #define EnterFunction(level) do {} while (0)
359 #define LeaveFunction(level) do {} while (0)
362 #define IP_VS_WAIT_WHILE(expr) while (expr) { cpu_relax(); }
366 * The port number of FTP service (in network order).
368 #define FTPPORT cpu_to_be16(21)
369 #define FTPDATA cpu_to_be16(20)
375 IP_VS_TCP_S_NONE
= 0,
376 IP_VS_TCP_S_ESTABLISHED
,
377 IP_VS_TCP_S_SYN_SENT
,
378 IP_VS_TCP_S_SYN_RECV
,
379 IP_VS_TCP_S_FIN_WAIT
,
380 IP_VS_TCP_S_TIME_WAIT
,
382 IP_VS_TCP_S_CLOSE_WAIT
,
383 IP_VS_TCP_S_LAST_ACK
,
408 enum ip_vs_sctp_states
{
410 IP_VS_SCTP_S_INIT_CLI
,
411 IP_VS_SCTP_S_INIT_SER
,
412 IP_VS_SCTP_S_INIT_ACK_CLI
,
413 IP_VS_SCTP_S_INIT_ACK_SER
,
414 IP_VS_SCTP_S_ECHO_CLI
,
415 IP_VS_SCTP_S_ECHO_SER
,
416 IP_VS_SCTP_S_ESTABLISHED
,
417 IP_VS_SCTP_S_SHUT_CLI
,
418 IP_VS_SCTP_S_SHUT_SER
,
419 IP_VS_SCTP_S_SHUT_ACK_CLI
,
420 IP_VS_SCTP_S_SHUT_ACK_SER
,
426 * Delta sequence info structure
427 * Each ip_vs_conn has 2 (output AND input seq. changes).
428 * Only used in the VS/NAT.
431 __u32 init_seq
; /* Add delta from this seq */
432 __u32 delta
; /* Delta in sequence numbers */
433 __u32 previous_delta
; /* Delta in sequence numbers
434 before last resized pkt */
440 struct ip_vs_counters
{
441 __u32 conns
; /* connections scheduled */
442 __u32 inpkts
; /* incoming packets */
443 __u32 outpkts
; /* outgoing packets */
444 __u64 inbytes
; /* incoming bytes */
445 __u64 outbytes
; /* outgoing bytes */
450 struct ip_vs_cpu_stats
{
451 struct ip_vs_counters ustats
;
452 struct u64_stats_sync syncp
;
456 * IPVS statistics objects
458 struct ip_vs_estimator
{
459 struct list_head list
;
475 struct ip_vs_stats_user ustats
; /* statistics */
476 struct ip_vs_estimator est
; /* estimator */
477 struct ip_vs_cpu_stats __percpu
*cpustats
; /* per cpu counters */
478 spinlock_t lock
; /* spin lock */
479 struct ip_vs_stats_user ustats0
; /* reset values */
487 struct ip_vs_proto_data
;
489 struct ip_vs_protocol
{
490 struct ip_vs_protocol
*next
;
496 void (*init
)(struct ip_vs_protocol
*pp
);
498 void (*exit
)(struct ip_vs_protocol
*pp
);
500 int (*init_netns
)(struct net
*net
, struct ip_vs_proto_data
*pd
);
502 void (*exit_netns
)(struct net
*net
, struct ip_vs_proto_data
*pd
);
504 int (*conn_schedule
)(int af
, struct sk_buff
*skb
,
505 struct ip_vs_proto_data
*pd
,
506 int *verdict
, struct ip_vs_conn
**cpp
,
507 struct ip_vs_iphdr
*iph
);
510 (*conn_in_get
)(int af
,
511 const struct sk_buff
*skb
,
512 const struct ip_vs_iphdr
*iph
,
516 (*conn_out_get
)(int af
,
517 const struct sk_buff
*skb
,
518 const struct ip_vs_iphdr
*iph
,
521 int (*snat_handler
)(struct sk_buff
*skb
, struct ip_vs_protocol
*pp
,
522 struct ip_vs_conn
*cp
, struct ip_vs_iphdr
*iph
);
524 int (*dnat_handler
)(struct sk_buff
*skb
, struct ip_vs_protocol
*pp
,
525 struct ip_vs_conn
*cp
, struct ip_vs_iphdr
*iph
);
527 int (*csum_check
)(int af
, struct sk_buff
*skb
,
528 struct ip_vs_protocol
*pp
);
530 const char *(*state_name
)(int state
);
532 void (*state_transition
)(struct ip_vs_conn
*cp
, int direction
,
533 const struct sk_buff
*skb
,
534 struct ip_vs_proto_data
*pd
);
536 int (*register_app
)(struct net
*net
, struct ip_vs_app
*inc
);
538 void (*unregister_app
)(struct net
*net
, struct ip_vs_app
*inc
);
540 int (*app_conn_bind
)(struct ip_vs_conn
*cp
);
542 void (*debug_packet
)(int af
, struct ip_vs_protocol
*pp
,
543 const struct sk_buff
*skb
,
547 void (*timeout_change
)(struct ip_vs_proto_data
*pd
, int flags
);
551 * protocol data per netns
553 struct ip_vs_proto_data
{
554 struct ip_vs_proto_data
*next
;
555 struct ip_vs_protocol
*pp
;
556 int *timeout_table
; /* protocol timeout table */
557 atomic_t appcnt
; /* counter of proto app incs. */
558 struct tcp_states_t
*tcp_state_table
;
561 extern struct ip_vs_protocol
*ip_vs_proto_get(unsigned short proto
);
562 extern struct ip_vs_proto_data
*ip_vs_proto_data_get(struct net
*net
,
563 unsigned short proto
);
565 struct ip_vs_conn_param
{
567 const union nf_inet_addr
*caddr
;
568 const union nf_inet_addr
*vaddr
;
574 const struct ip_vs_pe
*pe
;
580 * IP_VS structure allocated for each dynamically scheduled connection
583 struct hlist_node c_list
; /* hashed list heads */
584 /* Protocol, addresses and port numbers */
588 u16 af
; /* address family */
589 union nf_inet_addr caddr
; /* client address */
590 union nf_inet_addr vaddr
; /* virtual address */
591 union nf_inet_addr daddr
; /* destination address */
592 volatile __u32 flags
; /* status flags */
593 __u16 protocol
; /* Which protocol (TCP/UDP) */
595 struct net
*net
; /* Name space */
598 /* counter and timer */
599 atomic_t refcnt
; /* reference count */
600 struct timer_list timer
; /* Expiration timer */
601 volatile unsigned long timeout
; /* timeout */
603 /* Flags and state transition */
604 spinlock_t lock
; /* lock for state transition */
605 volatile __u16 state
; /* state info */
606 volatile __u16 old_state
; /* old state, to be used for
607 * state transition triggerd
610 __u32 fwmark
; /* Fire wall mark from skb */
611 unsigned long sync_endtime
; /* jiffies + sent_retries */
613 /* Control members */
614 struct ip_vs_conn
*control
; /* Master control connection */
615 atomic_t n_control
; /* Number of controlled ones */
616 struct ip_vs_dest
*dest
; /* real server */
617 atomic_t in_pkts
; /* incoming packet counter */
619 /* packet transmitter for different forwarding methods. If it
620 mangles the packet, it must return NF_DROP or better NF_STOLEN,
621 otherwise this must be changed to a sk_buff **.
622 NF_ACCEPT can be returned when destination is local.
624 int (*packet_xmit
)(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
625 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
627 /* Note: we can group the following members into a structure,
628 in order to save more space, and the following members are
629 only used in VS/NAT anyway */
630 struct ip_vs_app
*app
; /* bound ip_vs_app object */
631 void *app_data
; /* Application private data */
632 struct ip_vs_seq in_seq
; /* incoming seq. struct */
633 struct ip_vs_seq out_seq
; /* outgoing seq. struct */
635 const struct ip_vs_pe
*pe
;
639 struct rcu_head rcu_head
;
643 * To save some memory in conn table when name space is disabled.
645 static inline struct net
*ip_vs_conn_net(const struct ip_vs_conn
*cp
)
653 static inline void ip_vs_conn_net_set(struct ip_vs_conn
*cp
, struct net
*net
)
660 static inline int ip_vs_conn_net_eq(const struct ip_vs_conn
*cp
,
664 return cp
->net
== net
;
671 * Extended internal versions of struct ip_vs_service_user and
672 * ip_vs_dest_user for IPv6 support.
674 * We need these to conveniently pass around service and destination
675 * options, but unfortunately, we also need to keep the old definitions to
676 * maintain userspace backwards compatibility for the setsockopt interface.
678 struct ip_vs_service_user_kern
{
679 /* virtual service addresses */
682 union nf_inet_addr addr
; /* virtual ip address */
684 u32 fwmark
; /* firwall mark of service */
686 /* virtual service options */
689 unsigned int flags
; /* virtual service flags */
690 unsigned int timeout
; /* persistent timeout in sec */
691 u32 netmask
; /* persistent netmask */
695 struct ip_vs_dest_user_kern
{
696 /* destination server address */
697 union nf_inet_addr addr
;
700 /* real server options */
701 unsigned int conn_flags
; /* connection flags */
702 int weight
; /* destination weight */
704 /* thresholds for active connections */
705 u32 u_threshold
; /* upper threshold */
706 u32 l_threshold
; /* lower threshold */
711 * The information about the virtual service offered to the net
712 * and the forwarding entries
714 struct ip_vs_service
{
715 struct list_head s_list
; /* for normal service table */
716 struct list_head f_list
; /* for fwmark-based service table */
717 atomic_t refcnt
; /* reference counter */
718 atomic_t usecnt
; /* use counter */
720 u16 af
; /* address family */
721 __u16 protocol
; /* which protocol (TCP/UDP) */
722 union nf_inet_addr addr
; /* IP address for virtual service */
723 __be16 port
; /* port number for the service */
724 __u32 fwmark
; /* firewall mark of the service */
725 unsigned int flags
; /* service status flags */
726 unsigned int timeout
; /* persistent timeout in ticks */
727 __be32 netmask
; /* grouping granularity */
730 struct list_head destinations
; /* real server d-linked list */
731 __u32 num_dests
; /* number of servers */
732 struct ip_vs_stats stats
; /* statistics for the service */
733 struct ip_vs_app
*inc
; /* bind conns to this app inc */
736 struct ip_vs_scheduler
*scheduler
; /* bound scheduler object */
737 rwlock_t sched_lock
; /* lock sched_data */
738 void *sched_data
; /* scheduler application data */
740 /* alternate persistence engine */
744 /* Information for cached dst */
745 struct ip_vs_dest_dst
{
746 struct dst_entry
*dst_cache
; /* destination cache entry */
748 union nf_inet_addr dst_saddr
;
749 struct rcu_head rcu_head
;
752 /* In grace period after removing */
753 #define IP_VS_DEST_STATE_REMOVING 0x01
755 * The real server destination forwarding entry
756 * with ip address, port number, and so on.
759 struct list_head n_list
; /* for the dests in the service */
760 struct hlist_node d_list
; /* for table with all the dests */
762 u16 af
; /* address family */
763 __be16 port
; /* port number of the server */
764 union nf_inet_addr addr
; /* IP address of the server */
765 volatile unsigned int flags
; /* dest status flags */
766 atomic_t conn_flags
; /* flags to copy to conn */
767 atomic_t weight
; /* server weight */
769 atomic_t refcnt
; /* reference counter */
770 struct ip_vs_stats stats
; /* statistics */
771 unsigned long state
; /* state flags */
773 /* connection counters and thresholds */
774 atomic_t activeconns
; /* active connections */
775 atomic_t inactconns
; /* inactive connections */
776 atomic_t persistconns
; /* persistent connections */
777 __u32 u_threshold
; /* upper threshold */
778 __u32 l_threshold
; /* lower threshold */
780 /* for destination cache */
781 spinlock_t dst_lock
; /* lock of dst_cache */
782 struct ip_vs_dest_dst __rcu
*dest_dst
; /* cached dst info */
784 /* for virtual service */
785 struct ip_vs_service
*svc
; /* service it belongs to */
786 __u16 protocol
; /* which protocol (TCP/UDP) */
787 __be16 vport
; /* virtual port number */
788 union nf_inet_addr vaddr
; /* virtual IP address */
789 __u32 vfwmark
; /* firewall mark of service */
791 struct list_head t_list
; /* in dest_trash */
792 struct rcu_head rcu_head
;
793 unsigned int in_rs_table
:1; /* we are in rs_table */
798 * The scheduler object
800 struct ip_vs_scheduler
{
801 struct list_head n_list
; /* d-linked list head */
802 char *name
; /* scheduler name */
803 atomic_t refcnt
; /* reference counter */
804 struct module
*module
; /* THIS_MODULE/NULL */
806 /* scheduler initializing service */
807 int (*init_service
)(struct ip_vs_service
*svc
);
808 /* scheduling service finish */
809 void (*done_service
)(struct ip_vs_service
*svc
);
810 /* scheduler updating service */
811 int (*update_service
)(struct ip_vs_service
*svc
);
813 int (*add_dest
)(struct ip_vs_service
*svc
, struct ip_vs_dest
*dest
);
814 /* dest is unlinked */
815 int (*del_dest
)(struct ip_vs_service
*svc
, struct ip_vs_dest
*dest
);
816 /* dest is updated */
817 int (*upd_dest
)(struct ip_vs_service
*svc
, struct ip_vs_dest
*dest
);
819 /* selecting a server from the given service */
820 struct ip_vs_dest
* (*schedule
)(struct ip_vs_service
*svc
,
821 const struct sk_buff
*skb
);
824 /* The persistence engine object */
826 struct list_head n_list
; /* d-linked list head */
827 char *name
; /* scheduler name */
828 atomic_t refcnt
; /* reference counter */
829 struct module
*module
; /* THIS_MODULE/NULL */
831 /* get the connection template, if any */
832 int (*fill_param
)(struct ip_vs_conn_param
*p
, struct sk_buff
*skb
);
833 bool (*ct_match
)(const struct ip_vs_conn_param
*p
,
834 struct ip_vs_conn
*ct
);
835 u32 (*hashkey_raw
)(const struct ip_vs_conn_param
*p
, u32 initval
,
837 int (*show_pe_data
)(const struct ip_vs_conn
*cp
, char *buf
);
841 * The application module object (a.k.a. app incarnation)
844 struct list_head a_list
; /* member in app list */
845 int type
; /* IP_VS_APP_TYPE_xxx */
846 char *name
; /* application module name */
848 struct module
*module
; /* THIS_MODULE/NULL */
849 struct list_head incs_list
; /* list of incarnations */
851 /* members for application incarnations */
852 struct list_head p_list
; /* member in proto app list */
853 struct ip_vs_app
*app
; /* its real application */
854 __be16 port
; /* port number in net order */
855 atomic_t usecnt
; /* usage counter */
856 struct rcu_head rcu_head
;
859 * output hook: Process packet in inout direction, diff set for TCP.
860 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
861 * 2=Mangled but checksum was not updated
863 int (*pkt_out
)(struct ip_vs_app
*, struct ip_vs_conn
*,
864 struct sk_buff
*, int *diff
);
867 * input hook: Process packet in outin direction, diff set for TCP.
868 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
869 * 2=Mangled but checksum was not updated
871 int (*pkt_in
)(struct ip_vs_app
*, struct ip_vs_conn
*,
872 struct sk_buff
*, int *diff
);
874 /* ip_vs_app initializer */
875 int (*init_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
877 /* ip_vs_app finish */
878 int (*done_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
882 int (*bind_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*,
883 struct ip_vs_protocol
*);
885 void (*unbind_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
891 int (*conn_schedule
)(struct sk_buff
*skb
, struct ip_vs_app
*app
,
892 int *verdict
, struct ip_vs_conn
**cpp
);
895 (*conn_in_get
)(const struct sk_buff
*skb
, struct ip_vs_app
*app
,
896 const struct iphdr
*iph
, int inverse
);
899 (*conn_out_get
)(const struct sk_buff
*skb
, struct ip_vs_app
*app
,
900 const struct iphdr
*iph
, int inverse
);
902 int (*state_transition
)(struct ip_vs_conn
*cp
, int direction
,
903 const struct sk_buff
*skb
,
904 struct ip_vs_app
*app
);
906 void (*timeout_change
)(struct ip_vs_app
*app
, int flags
);
909 struct ipvs_master_sync_state
{
910 struct list_head sync_queue
;
911 struct ip_vs_sync_buff
*sync_buff
;
913 unsigned int sync_queue_delay
;
914 struct task_struct
*master_thread
;
915 struct delayed_work master_wakeup_work
;
916 struct netns_ipvs
*ipvs
;
919 /* How much time to keep dests in trash */
920 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
922 /* IPVS in network namespace */
924 int gen
; /* Generation */
925 int enable
; /* enable like nf_hooks do */
927 * Hash table: for real service lookups
929 #define IP_VS_RTAB_BITS 4
930 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
931 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
933 struct hlist_head rs_table
[IP_VS_RTAB_SIZE
];
935 struct list_head app_list
;
937 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
938 struct ip_vs_proto_data
*proto_data_table
[IP_VS_PROTO_TAB_SIZE
];
939 /* ip_vs_proto_tcp */
940 #ifdef CONFIG_IP_VS_PROTO_TCP
941 #define TCP_APP_TAB_BITS 4
942 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
943 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
944 struct list_head tcp_apps
[TCP_APP_TAB_SIZE
];
946 /* ip_vs_proto_udp */
947 #ifdef CONFIG_IP_VS_PROTO_UDP
948 #define UDP_APP_TAB_BITS 4
949 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
950 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
951 struct list_head udp_apps
[UDP_APP_TAB_SIZE
];
953 /* ip_vs_proto_sctp */
954 #ifdef CONFIG_IP_VS_PROTO_SCTP
955 #define SCTP_APP_TAB_BITS 4
956 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
957 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
958 /* Hash table for SCTP application incarnations */
959 struct list_head sctp_apps
[SCTP_APP_TAB_SIZE
];
962 atomic_t conn_count
; /* connection counter */
965 struct ip_vs_stats tot_stats
; /* Statistics & est. */
967 int num_services
; /* no of virtual services */
969 /* Trash for destinations */
970 struct list_head dest_trash
;
971 spinlock_t dest_trash_lock
;
972 struct timer_list dest_trash_timer
; /* expiration timer */
973 /* Service counters */
974 atomic_t ftpsvc_counter
;
975 atomic_t nullsvc_counter
;
978 /* 1/rate drop and drop-entry variables */
979 struct delayed_work defense_work
; /* Work handler */
984 spinlock_t dropentry_lock
; /* drop entry handling */
985 spinlock_t droppacket_lock
; /* drop packet handling */
986 spinlock_t securetcp_lock
; /* state and timeout tables */
989 struct ctl_table_header
*sysctl_hdr
;
990 struct ctl_table
*sysctl_tbl
;
993 /* sysctl variables */
994 int sysctl_amemthresh
;
995 int sysctl_am_droprate
;
996 int sysctl_drop_entry
;
997 int sysctl_drop_packet
;
998 int sysctl_secure_tcp
;
999 #ifdef CONFIG_IP_VS_NFCT
1000 int sysctl_conntrack
;
1002 int sysctl_snat_reroute
;
1003 int sysctl_sync_ver
;
1004 int sysctl_sync_ports
;
1005 int sysctl_sync_qlen_max
;
1006 int sysctl_sync_sock_size
;
1007 int sysctl_cache_bypass
;
1008 int sysctl_expire_nodest_conn
;
1009 int sysctl_expire_quiescent_template
;
1010 int sysctl_sync_threshold
[2];
1011 unsigned int sysctl_sync_refresh_period
;
1012 int sysctl_sync_retries
;
1013 int sysctl_nat_icmp_send
;
1014 int sysctl_pmtu_disc
;
1015 int sysctl_backup_only
;
1018 int sysctl_lblc_expiration
;
1019 struct ctl_table_header
*lblc_ctl_header
;
1020 struct ctl_table
*lblc_ctl_table
;
1022 int sysctl_lblcr_expiration
;
1023 struct ctl_table_header
*lblcr_ctl_header
;
1024 struct ctl_table
*lblcr_ctl_table
;
1026 struct list_head est_list
; /* estimator list */
1027 spinlock_t est_lock
;
1028 struct timer_list est_timer
; /* Estimation timer */
1030 spinlock_t sync_lock
;
1031 struct ipvs_master_sync_state
*ms
;
1032 spinlock_t sync_buff_lock
;
1033 struct task_struct
**backup_threads
;
1035 int send_mesg_maxlen
;
1036 int recv_mesg_maxlen
;
1037 volatile int sync_state
;
1038 volatile int master_syncid
;
1039 volatile int backup_syncid
;
1040 struct mutex sync_mutex
;
1041 /* multicast interface name */
1042 char master_mcast_ifn
[IP_VS_IFNAME_MAXLEN
];
1043 char backup_mcast_ifn
[IP_VS_IFNAME_MAXLEN
];
1044 /* net name space ptr */
1045 struct net
*net
; /* Needed by timer routines */
1048 #define DEFAULT_SYNC_THRESHOLD 3
1049 #define DEFAULT_SYNC_PERIOD 50
1050 #define DEFAULT_SYNC_VER 1
1051 #define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
1052 #define DEFAULT_SYNC_RETRIES 0
1053 #define IPVS_SYNC_WAKEUP_RATE 8
1054 #define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
1055 #define IPVS_SYNC_SEND_DELAY (HZ / 50)
1056 #define IPVS_SYNC_CHECK_PERIOD HZ
1057 #define IPVS_SYNC_FLUSH_TIME (HZ * 2)
1058 #define IPVS_SYNC_PORTS_MAX (1 << 6)
1060 #ifdef CONFIG_SYSCTL
1062 static inline int sysctl_sync_threshold(struct netns_ipvs
*ipvs
)
1064 return ipvs
->sysctl_sync_threshold
[0];
1067 static inline int sysctl_sync_period(struct netns_ipvs
*ipvs
)
1069 return ACCESS_ONCE(ipvs
->sysctl_sync_threshold
[1]);
1072 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs
*ipvs
)
1074 return ACCESS_ONCE(ipvs
->sysctl_sync_refresh_period
);
1077 static inline int sysctl_sync_retries(struct netns_ipvs
*ipvs
)
1079 return ipvs
->sysctl_sync_retries
;
1082 static inline int sysctl_sync_ver(struct netns_ipvs
*ipvs
)
1084 return ipvs
->sysctl_sync_ver
;
1087 static inline int sysctl_sync_ports(struct netns_ipvs
*ipvs
)
1089 return ACCESS_ONCE(ipvs
->sysctl_sync_ports
);
1092 static inline int sysctl_sync_qlen_max(struct netns_ipvs
*ipvs
)
1094 return ipvs
->sysctl_sync_qlen_max
;
1097 static inline int sysctl_sync_sock_size(struct netns_ipvs
*ipvs
)
1099 return ipvs
->sysctl_sync_sock_size
;
1102 static inline int sysctl_pmtu_disc(struct netns_ipvs
*ipvs
)
1104 return ipvs
->sysctl_pmtu_disc
;
1107 static inline int sysctl_backup_only(struct netns_ipvs
*ipvs
)
1109 return ipvs
->sync_state
& IP_VS_STATE_BACKUP
&&
1110 ipvs
->sysctl_backup_only
;
1115 static inline int sysctl_sync_threshold(struct netns_ipvs
*ipvs
)
1117 return DEFAULT_SYNC_THRESHOLD
;
1120 static inline int sysctl_sync_period(struct netns_ipvs
*ipvs
)
1122 return DEFAULT_SYNC_PERIOD
;
1125 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs
*ipvs
)
1127 return DEFAULT_SYNC_REFRESH_PERIOD
;
1130 static inline int sysctl_sync_retries(struct netns_ipvs
*ipvs
)
1132 return DEFAULT_SYNC_RETRIES
& 3;
1135 static inline int sysctl_sync_ver(struct netns_ipvs
*ipvs
)
1137 return DEFAULT_SYNC_VER
;
1140 static inline int sysctl_sync_ports(struct netns_ipvs
*ipvs
)
1145 static inline int sysctl_sync_qlen_max(struct netns_ipvs
*ipvs
)
1147 return IPVS_SYNC_QLEN_MAX
;
1150 static inline int sysctl_sync_sock_size(struct netns_ipvs
*ipvs
)
1155 static inline int sysctl_pmtu_disc(struct netns_ipvs
*ipvs
)
1160 static inline int sysctl_backup_only(struct netns_ipvs
*ipvs
)
1168 * IPVS core functions
1169 * (from ip_vs_core.c)
1171 extern const char *ip_vs_proto_name(unsigned int proto
);
1172 extern void ip_vs_init_hash_table(struct list_head
*table
, int rows
);
1173 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1175 #define IP_VS_APP_TYPE_FTP 1
1178 * ip_vs_conn handling functions
1179 * (from ip_vs_conn.c)
1183 IP_VS_DIR_INPUT
= 0,
1185 IP_VS_DIR_INPUT_ONLY
,
1189 static inline void ip_vs_conn_fill_param(struct net
*net
, int af
, int protocol
,
1190 const union nf_inet_addr
*caddr
,
1192 const union nf_inet_addr
*vaddr
,
1194 struct ip_vs_conn_param
*p
)
1198 p
->protocol
= protocol
;
1207 struct ip_vs_conn
*ip_vs_conn_in_get(const struct ip_vs_conn_param
*p
);
1208 struct ip_vs_conn
*ip_vs_ct_in_get(const struct ip_vs_conn_param
*p
);
1210 struct ip_vs_conn
* ip_vs_conn_in_get_proto(int af
, const struct sk_buff
*skb
,
1211 const struct ip_vs_iphdr
*iph
,
1214 struct ip_vs_conn
*ip_vs_conn_out_get(const struct ip_vs_conn_param
*p
);
1216 struct ip_vs_conn
* ip_vs_conn_out_get_proto(int af
, const struct sk_buff
*skb
,
1217 const struct ip_vs_iphdr
*iph
,
1220 /* Get reference to gain full access to conn.
1221 * By default, RCU read-side critical sections have access only to
1222 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1224 static inline bool __ip_vs_conn_get(struct ip_vs_conn
*cp
)
1226 return atomic_inc_not_zero(&cp
->refcnt
);
1229 /* put back the conn without restarting its timer */
1230 static inline void __ip_vs_conn_put(struct ip_vs_conn
*cp
)
1232 smp_mb__before_atomic_dec();
1233 atomic_dec(&cp
->refcnt
);
1235 extern void ip_vs_conn_put(struct ip_vs_conn
*cp
);
1236 extern void ip_vs_conn_fill_cport(struct ip_vs_conn
*cp
, __be16 cport
);
1238 struct ip_vs_conn
*ip_vs_conn_new(const struct ip_vs_conn_param
*p
,
1239 const union nf_inet_addr
*daddr
,
1240 __be16 dport
, unsigned int flags
,
1241 struct ip_vs_dest
*dest
, __u32 fwmark
);
1242 extern void ip_vs_conn_expire_now(struct ip_vs_conn
*cp
);
1244 extern const char * ip_vs_state_name(__u16 proto
, int state
);
1246 extern void ip_vs_tcp_conn_listen(struct net
*net
, struct ip_vs_conn
*cp
);
1247 extern int ip_vs_check_template(struct ip_vs_conn
*ct
);
1248 extern void ip_vs_random_dropentry(struct net
*net
);
1249 extern int ip_vs_conn_init(void);
1250 extern void ip_vs_conn_cleanup(void);
1252 static inline void ip_vs_control_del(struct ip_vs_conn
*cp
)
1254 struct ip_vs_conn
*ctl_cp
= cp
->control
;
1256 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1258 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1260 IP_VS_DBG_ADDR(cp
->af
, &cp
->vaddr
),
1266 IP_VS_DBG_BUF(7, "DELeting control for: "
1267 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1268 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1270 IP_VS_DBG_ADDR(cp
->af
, &ctl_cp
->caddr
),
1271 ntohs(ctl_cp
->cport
));
1274 if (atomic_read(&ctl_cp
->n_control
) == 0) {
1275 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1277 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1279 IP_VS_DBG_ADDR(cp
->af
, &cp
->vaddr
),
1284 atomic_dec(&ctl_cp
->n_control
);
1288 ip_vs_control_add(struct ip_vs_conn
*cp
, struct ip_vs_conn
*ctl_cp
)
1291 IP_VS_ERR_BUF("request control ADD for already controlled: "
1293 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1295 IP_VS_DBG_ADDR(cp
->af
, &cp
->vaddr
),
1298 ip_vs_control_del(cp
);
1301 IP_VS_DBG_BUF(7, "ADDing control for: "
1302 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1303 IP_VS_DBG_ADDR(cp
->af
, &cp
->caddr
),
1305 IP_VS_DBG_ADDR(cp
->af
, &ctl_cp
->caddr
),
1306 ntohs(ctl_cp
->cport
));
1308 cp
->control
= ctl_cp
;
1309 atomic_inc(&ctl_cp
->n_control
);
1313 * IPVS netns init & cleanup functions
1315 extern int ip_vs_estimator_net_init(struct net
*net
);
1316 extern int ip_vs_control_net_init(struct net
*net
);
1317 extern int ip_vs_protocol_net_init(struct net
*net
);
1318 extern int ip_vs_app_net_init(struct net
*net
);
1319 extern int ip_vs_conn_net_init(struct net
*net
);
1320 extern int ip_vs_sync_net_init(struct net
*net
);
1321 extern void ip_vs_conn_net_cleanup(struct net
*net
);
1322 extern void ip_vs_app_net_cleanup(struct net
*net
);
1323 extern void ip_vs_protocol_net_cleanup(struct net
*net
);
1324 extern void ip_vs_control_net_cleanup(struct net
*net
);
1325 extern void ip_vs_estimator_net_cleanup(struct net
*net
);
1326 extern void ip_vs_sync_net_cleanup(struct net
*net
);
1327 extern void ip_vs_service_net_cleanup(struct net
*net
);
1330 * IPVS application functions
1331 * (from ip_vs_app.c)
1333 #define IP_VS_APP_MAX_PORTS 8
1334 extern struct ip_vs_app
*register_ip_vs_app(struct net
*net
,
1335 struct ip_vs_app
*app
);
1336 extern void unregister_ip_vs_app(struct net
*net
, struct ip_vs_app
*app
);
1337 extern int ip_vs_bind_app(struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1338 extern void ip_vs_unbind_app(struct ip_vs_conn
*cp
);
1339 extern int register_ip_vs_app_inc(struct net
*net
, struct ip_vs_app
*app
,
1340 __u16 proto
, __u16 port
);
1341 extern int ip_vs_app_inc_get(struct ip_vs_app
*inc
);
1342 extern void ip_vs_app_inc_put(struct ip_vs_app
*inc
);
1344 extern int ip_vs_app_pkt_out(struct ip_vs_conn
*, struct sk_buff
*skb
);
1345 extern int ip_vs_app_pkt_in(struct ip_vs_conn
*, struct sk_buff
*skb
);
1347 void ip_vs_bind_pe(struct ip_vs_service
*svc
, struct ip_vs_pe
*pe
);
1348 void ip_vs_unbind_pe(struct ip_vs_service
*svc
);
1349 int register_ip_vs_pe(struct ip_vs_pe
*pe
);
1350 int unregister_ip_vs_pe(struct ip_vs_pe
*pe
);
1351 struct ip_vs_pe
*ip_vs_pe_getbyname(const char *name
);
1352 struct ip_vs_pe
*__ip_vs_pe_getbyname(const char *pe_name
);
1355 * Use a #define to avoid all of module.h just for these trivial ops
1357 #define ip_vs_pe_get(pe) \
1358 if (pe && pe->module) \
1359 __module_get(pe->module);
1361 #define ip_vs_pe_put(pe) \
1362 if (pe && pe->module) \
1363 module_put(pe->module);
1366 * IPVS protocol functions (from ip_vs_proto.c)
1368 extern int ip_vs_protocol_init(void);
1369 extern void ip_vs_protocol_cleanup(void);
1370 extern void ip_vs_protocol_timeout_change(struct netns_ipvs
*ipvs
, int flags
);
1371 extern int *ip_vs_create_timeout_table(int *table
, int size
);
1373 ip_vs_set_state_timeout(int *table
, int num
, const char *const *names
,
1374 const char *name
, int to
);
1376 ip_vs_tcpudp_debug_packet(int af
, struct ip_vs_protocol
*pp
,
1377 const struct sk_buff
*skb
,
1378 int offset
, const char *msg
);
1380 extern struct ip_vs_protocol ip_vs_protocol_tcp
;
1381 extern struct ip_vs_protocol ip_vs_protocol_udp
;
1382 extern struct ip_vs_protocol ip_vs_protocol_icmp
;
1383 extern struct ip_vs_protocol ip_vs_protocol_esp
;
1384 extern struct ip_vs_protocol ip_vs_protocol_ah
;
1385 extern struct ip_vs_protocol ip_vs_protocol_sctp
;
1388 * Registering/unregistering scheduler functions
1389 * (from ip_vs_sched.c)
1391 extern int register_ip_vs_scheduler(struct ip_vs_scheduler
*scheduler
);
1392 extern int unregister_ip_vs_scheduler(struct ip_vs_scheduler
*scheduler
);
1393 extern int ip_vs_bind_scheduler(struct ip_vs_service
*svc
,
1394 struct ip_vs_scheduler
*scheduler
);
1395 extern void ip_vs_unbind_scheduler(struct ip_vs_service
*svc
);
1396 extern struct ip_vs_scheduler
*ip_vs_scheduler_get(const char *sched_name
);
1397 extern void ip_vs_scheduler_put(struct ip_vs_scheduler
*scheduler
);
1398 extern struct ip_vs_conn
*
1399 ip_vs_schedule(struct ip_vs_service
*svc
, struct sk_buff
*skb
,
1400 struct ip_vs_proto_data
*pd
, int *ignored
,
1401 struct ip_vs_iphdr
*iph
);
1402 extern int ip_vs_leave(struct ip_vs_service
*svc
, struct sk_buff
*skb
,
1403 struct ip_vs_proto_data
*pd
, struct ip_vs_iphdr
*iph
);
1405 extern void ip_vs_scheduler_err(struct ip_vs_service
*svc
, const char *msg
);
1409 * IPVS control data and functions (from ip_vs_ctl.c)
1411 extern struct ip_vs_stats ip_vs_stats
;
1412 extern int sysctl_ip_vs_sync_ver
;
1414 extern struct ip_vs_service
*
1415 ip_vs_service_get(struct net
*net
, int af
, __u32 fwmark
, __u16 protocol
,
1416 const union nf_inet_addr
*vaddr
, __be16 vport
);
1418 static inline void ip_vs_service_put(struct ip_vs_service
*svc
)
1420 atomic_dec(&svc
->usecnt
);
1424 ip_vs_has_real_service(struct net
*net
, int af
, __u16 protocol
,
1425 const union nf_inet_addr
*daddr
, __be16 dport
);
1427 extern int ip_vs_use_count_inc(void);
1428 extern void ip_vs_use_count_dec(void);
1429 extern int ip_vs_register_nl_ioctl(void);
1430 extern void ip_vs_unregister_nl_ioctl(void);
1431 extern int ip_vs_control_init(void);
1432 extern void ip_vs_control_cleanup(void);
1433 extern struct ip_vs_dest
*
1434 ip_vs_find_dest(struct net
*net
, int af
, const union nf_inet_addr
*daddr
,
1435 __be16 dport
, const union nf_inet_addr
*vaddr
, __be16 vport
,
1436 __u16 protocol
, __u32 fwmark
, __u32 flags
);
1437 extern struct ip_vs_dest
*ip_vs_try_bind_dest(struct ip_vs_conn
*cp
);
1439 static inline void ip_vs_dest_hold(struct ip_vs_dest
*dest
)
1441 atomic_inc(&dest
->refcnt
);
1444 static inline void ip_vs_dest_put(struct ip_vs_dest
*dest
)
1446 smp_mb__before_atomic_dec();
1447 atomic_dec(&dest
->refcnt
);
1451 * IPVS sync daemon data and function prototypes
1452 * (from ip_vs_sync.c)
1454 extern int start_sync_thread(struct net
*net
, int state
, char *mcast_ifn
,
1456 extern int stop_sync_thread(struct net
*net
, int state
);
1457 extern void ip_vs_sync_conn(struct net
*net
, struct ip_vs_conn
*cp
, int pkts
);
1461 * IPVS rate estimator prototypes (from ip_vs_est.c)
1463 extern void ip_vs_start_estimator(struct net
*net
, struct ip_vs_stats
*stats
);
1464 extern void ip_vs_stop_estimator(struct net
*net
, struct ip_vs_stats
*stats
);
1465 extern void ip_vs_zero_estimator(struct ip_vs_stats
*stats
);
1466 extern void ip_vs_read_estimator(struct ip_vs_stats_user
*dst
,
1467 struct ip_vs_stats
*stats
);
1470 * Various IPVS packet transmitters (from ip_vs_xmit.c)
1472 extern int ip_vs_null_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1473 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1474 extern int ip_vs_bypass_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1475 struct ip_vs_protocol
*pp
,
1476 struct ip_vs_iphdr
*iph
);
1477 extern int ip_vs_nat_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1478 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1479 extern int ip_vs_tunnel_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1480 struct ip_vs_protocol
*pp
,
1481 struct ip_vs_iphdr
*iph
);
1482 extern int ip_vs_dr_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1483 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1484 extern int ip_vs_icmp_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1485 struct ip_vs_protocol
*pp
, int offset
,
1486 unsigned int hooknum
, struct ip_vs_iphdr
*iph
);
1487 extern void ip_vs_dest_dst_rcu_free(struct rcu_head
*head
);
1489 #ifdef CONFIG_IP_VS_IPV6
1490 extern int ip_vs_bypass_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1491 struct ip_vs_protocol
*pp
,
1492 struct ip_vs_iphdr
*iph
);
1493 extern int ip_vs_nat_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1494 struct ip_vs_protocol
*pp
,
1495 struct ip_vs_iphdr
*iph
);
1496 extern int ip_vs_tunnel_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1497 struct ip_vs_protocol
*pp
,
1498 struct ip_vs_iphdr
*iph
);
1499 extern int ip_vs_dr_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1500 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1501 extern int ip_vs_icmp_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1502 struct ip_vs_protocol
*pp
, int offset
,
1503 unsigned int hooknum
, struct ip_vs_iphdr
*iph
);
1506 #ifdef CONFIG_SYSCTL
1508 * This is a simple mechanism to ignore packets when
1509 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1510 * we start to drop 1/rate of the packets
1513 static inline int ip_vs_todrop(struct netns_ipvs
*ipvs
)
1515 if (!ipvs
->drop_rate
)
1517 if (--ipvs
->drop_counter
> 0)
1519 ipvs
->drop_counter
= ipvs
->drop_rate
;
1523 static inline int ip_vs_todrop(struct netns_ipvs
*ipvs
) { return 0; }
1527 * ip_vs_fwd_tag returns the forwarding tag of the connection
1529 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1531 static inline char ip_vs_fwd_tag(struct ip_vs_conn
*cp
)
1535 switch (IP_VS_FWD_METHOD(cp
)) {
1536 case IP_VS_CONN_F_MASQ
:
1538 case IP_VS_CONN_F_LOCALNODE
:
1540 case IP_VS_CONN_F_TUNNEL
:
1542 case IP_VS_CONN_F_DROUTE
:
1544 case IP_VS_CONN_F_BYPASS
:
1552 extern void ip_vs_nat_icmp(struct sk_buff
*skb
, struct ip_vs_protocol
*pp
,
1553 struct ip_vs_conn
*cp
, int dir
);
1555 #ifdef CONFIG_IP_VS_IPV6
1556 extern void ip_vs_nat_icmp_v6(struct sk_buff
*skb
, struct ip_vs_protocol
*pp
,
1557 struct ip_vs_conn
*cp
, int dir
);
1560 extern __sum16
ip_vs_checksum_complete(struct sk_buff
*skb
, int offset
);
1562 static inline __wsum
ip_vs_check_diff4(__be32 old
, __be32
new, __wsum oldsum
)
1564 __be32 diff
[2] = { ~old
, new };
1566 return csum_partial(diff
, sizeof(diff
), oldsum
);
1569 #ifdef CONFIG_IP_VS_IPV6
1570 static inline __wsum
ip_vs_check_diff16(const __be32
*old
, const __be32
*new,
1573 __be32 diff
[8] = { ~old
[3], ~old
[2], ~old
[1], ~old
[0],
1574 new[3], new[2], new[1], new[0] };
1576 return csum_partial(diff
, sizeof(diff
), oldsum
);
1580 static inline __wsum
ip_vs_check_diff2(__be16 old
, __be16
new, __wsum oldsum
)
1582 __be16 diff
[2] = { ~old
, new };
1584 return csum_partial(diff
, sizeof(diff
), oldsum
);
1588 * Forget current conntrack (unconfirmed) and attach notrack entry
1590 static inline void ip_vs_notrack(struct sk_buff
*skb
)
1592 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1593 enum ip_conntrack_info ctinfo
;
1594 struct nf_conn
*ct
= nf_ct_get(skb
, &ctinfo
);
1596 if (!ct
|| !nf_ct_is_untracked(ct
)) {
1597 nf_conntrack_put(skb
->nfct
);
1598 skb
->nfct
= &nf_ct_untracked_get()->ct_general
;
1599 skb
->nfctinfo
= IP_CT_NEW
;
1600 nf_conntrack_get(skb
->nfct
);
1605 #ifdef CONFIG_IP_VS_NFCT
1607 * Netfilter connection tracking
1608 * (from ip_vs_nfct.c)
1610 static inline int ip_vs_conntrack_enabled(struct netns_ipvs
*ipvs
)
1612 #ifdef CONFIG_SYSCTL
1613 return ipvs
->sysctl_conntrack
;
1619 extern void ip_vs_update_conntrack(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1621 extern int ip_vs_confirm_conntrack(struct sk_buff
*skb
);
1622 extern void ip_vs_nfct_expect_related(struct sk_buff
*skb
, struct nf_conn
*ct
,
1623 struct ip_vs_conn
*cp
, u_int8_t proto
,
1624 const __be16 port
, int from_rs
);
1625 extern void ip_vs_conn_drop_conntrack(struct ip_vs_conn
*cp
);
1629 static inline int ip_vs_conntrack_enabled(struct netns_ipvs
*ipvs
)
1634 static inline void ip_vs_update_conntrack(struct sk_buff
*skb
,
1635 struct ip_vs_conn
*cp
, int outin
)
1639 static inline int ip_vs_confirm_conntrack(struct sk_buff
*skb
)
1644 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn
*cp
)
1647 /* CONFIG_IP_VS_NFCT */
1650 static inline unsigned int
1651 ip_vs_dest_conn_overhead(struct ip_vs_dest
*dest
)
1654 * We think the overhead of processing active connections is 256
1655 * times higher than that of inactive connections in average. (This
1656 * 256 times might not be accurate, we will change it later) We
1657 * use the following formula to estimate the overhead now:
1658 * dest->activeconns*256 + dest->inactconns
1660 return (atomic_read(&dest
->activeconns
) << 8) +
1661 atomic_read(&dest
->inactconns
);
1664 #endif /* _NET_IP_VS_H */