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 */
606 u16 tun_flags
; /* tunnel flags */
611 * The information about the virtual service offered to the net and the
612 * forwarding entries.
614 struct ip_vs_service
{
615 struct hlist_node s_list
; /* for normal service table */
616 struct hlist_node f_list
; /* for fwmark-based service table */
617 atomic_t refcnt
; /* reference counter */
619 u16 af
; /* address family */
620 __u16 protocol
; /* which protocol (TCP/UDP) */
621 union nf_inet_addr addr
; /* IP address for virtual service */
622 __be16 port
; /* port number for the service */
623 __u32 fwmark
; /* firewall mark of the service */
624 unsigned int flags
; /* service status flags */
625 unsigned int timeout
; /* persistent timeout in ticks */
626 __be32 netmask
; /* grouping granularity, mask/plen */
627 struct netns_ipvs
*ipvs
;
629 struct list_head destinations
; /* real server d-linked list */
630 __u32 num_dests
; /* number of servers */
631 struct ip_vs_stats stats
; /* statistics for the service */
634 struct ip_vs_scheduler __rcu
*scheduler
; /* bound scheduler object */
635 spinlock_t sched_lock
; /* lock sched_data */
636 void *sched_data
; /* scheduler application data */
638 /* alternate persistence engine */
639 struct ip_vs_pe __rcu
*pe
;
640 int conntrack_afmask
;
642 struct rcu_head rcu_head
;
645 /* Information for cached dst */
646 struct ip_vs_dest_dst
{
647 struct dst_entry
*dst_cache
; /* destination cache entry */
649 union nf_inet_addr dst_saddr
;
650 struct rcu_head rcu_head
;
653 /* The real server destination forwarding entry with ip address, port number,
657 struct list_head n_list
; /* for the dests in the service */
658 struct hlist_node d_list
; /* for table with all the dests */
660 u16 af
; /* address family */
661 __be16 port
; /* port number of the server */
662 union nf_inet_addr addr
; /* IP address of the server */
663 volatile unsigned int flags
; /* dest status flags */
664 atomic_t conn_flags
; /* flags to copy to conn */
665 atomic_t weight
; /* server weight */
666 atomic_t last_weight
; /* server latest weight */
667 __u16 tun_type
; /* tunnel type */
668 __be16 tun_port
; /* tunnel port */
669 __u16 tun_flags
; /* tunnel flags */
671 refcount_t refcnt
; /* reference counter */
672 struct ip_vs_stats stats
; /* statistics */
673 unsigned long idle_start
; /* start time, jiffies */
675 /* connection counters and thresholds */
676 atomic_t activeconns
; /* active connections */
677 atomic_t inactconns
; /* inactive connections */
678 atomic_t persistconns
; /* persistent connections */
679 __u32 u_threshold
; /* upper threshold */
680 __u32 l_threshold
; /* lower threshold */
682 /* for destination cache */
683 spinlock_t dst_lock
; /* lock of dst_cache */
684 struct ip_vs_dest_dst __rcu
*dest_dst
; /* cached dst info */
686 /* for virtual service */
687 struct ip_vs_service __rcu
*svc
; /* service it belongs to */
688 __u16 protocol
; /* which protocol (TCP/UDP) */
689 __be16 vport
; /* virtual port number */
690 union nf_inet_addr vaddr
; /* virtual IP address */
691 __u32 vfwmark
; /* firewall mark of service */
693 struct list_head t_list
; /* in dest_trash */
694 unsigned int in_rs_table
:1; /* we are in rs_table */
697 /* The scheduler object */
698 struct ip_vs_scheduler
{
699 struct list_head n_list
; /* d-linked list head */
700 char *name
; /* scheduler name */
701 atomic_t refcnt
; /* reference counter */
702 struct module
*module
; /* THIS_MODULE/NULL */
704 /* scheduler initializing service */
705 int (*init_service
)(struct ip_vs_service
*svc
);
706 /* scheduling service finish */
707 void (*done_service
)(struct ip_vs_service
*svc
);
709 int (*add_dest
)(struct ip_vs_service
*svc
, struct ip_vs_dest
*dest
);
710 /* dest is unlinked */
711 int (*del_dest
)(struct ip_vs_service
*svc
, struct ip_vs_dest
*dest
);
712 /* dest is updated */
713 int (*upd_dest
)(struct ip_vs_service
*svc
, struct ip_vs_dest
*dest
);
715 /* selecting a server from the given service */
716 struct ip_vs_dest
* (*schedule
)(struct ip_vs_service
*svc
,
717 const struct sk_buff
*skb
,
718 struct ip_vs_iphdr
*iph
);
721 /* The persistence engine object */
723 struct list_head n_list
; /* d-linked list head */
724 char *name
; /* scheduler name */
725 atomic_t refcnt
; /* reference counter */
726 struct module
*module
; /* THIS_MODULE/NULL */
728 /* get the connection template, if any */
729 int (*fill_param
)(struct ip_vs_conn_param
*p
, struct sk_buff
*skb
);
730 bool (*ct_match
)(const struct ip_vs_conn_param
*p
,
731 struct ip_vs_conn
*ct
);
732 u32 (*hashkey_raw
)(const struct ip_vs_conn_param
*p
, u32 initval
,
734 int (*show_pe_data
)(const struct ip_vs_conn
*cp
, char *buf
);
735 /* create connections for real-server outgoing packets */
736 struct ip_vs_conn
* (*conn_out
)(struct ip_vs_service
*svc
,
737 struct ip_vs_dest
*dest
,
739 const struct ip_vs_iphdr
*iph
,
740 __be16 dport
, __be16 cport
);
743 /* The application module object (a.k.a. app incarnation) */
745 struct list_head a_list
; /* member in app list */
746 int type
; /* IP_VS_APP_TYPE_xxx */
747 char *name
; /* application module name */
749 struct module
*module
; /* THIS_MODULE/NULL */
750 struct list_head incs_list
; /* list of incarnations */
752 /* members for application incarnations */
753 struct list_head p_list
; /* member in proto app list */
754 struct ip_vs_app
*app
; /* its real application */
755 __be16 port
; /* port number in net order */
756 atomic_t usecnt
; /* usage counter */
757 struct rcu_head rcu_head
;
759 /* output hook: Process packet in inout direction, diff set for TCP.
760 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
761 * 2=Mangled but checksum was not updated
763 int (*pkt_out
)(struct ip_vs_app
*, struct ip_vs_conn
*,
764 struct sk_buff
*, int *diff
, struct ip_vs_iphdr
*ipvsh
);
766 /* input hook: Process packet in outin direction, diff set for TCP.
767 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
768 * 2=Mangled but checksum was not updated
770 int (*pkt_in
)(struct ip_vs_app
*, struct ip_vs_conn
*,
771 struct sk_buff
*, int *diff
, struct ip_vs_iphdr
*ipvsh
);
773 /* ip_vs_app initializer */
774 int (*init_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
776 /* ip_vs_app finish */
777 int (*done_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
781 int (*bind_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*,
782 struct ip_vs_protocol
*);
784 void (*unbind_conn
)(struct ip_vs_app
*, struct ip_vs_conn
*);
790 int (*conn_schedule
)(struct sk_buff
*skb
, struct ip_vs_app
*app
,
791 int *verdict
, struct ip_vs_conn
**cpp
);
794 (*conn_in_get
)(const struct sk_buff
*skb
, struct ip_vs_app
*app
,
795 const struct iphdr
*iph
, int inverse
);
798 (*conn_out_get
)(const struct sk_buff
*skb
, struct ip_vs_app
*app
,
799 const struct iphdr
*iph
, int inverse
);
801 int (*state_transition
)(struct ip_vs_conn
*cp
, int direction
,
802 const struct sk_buff
*skb
,
803 struct ip_vs_app
*app
);
805 void (*timeout_change
)(struct ip_vs_app
*app
, int flags
);
808 struct ipvs_master_sync_state
{
809 struct list_head sync_queue
;
810 struct ip_vs_sync_buff
*sync_buff
;
811 unsigned long sync_queue_len
;
812 unsigned int sync_queue_delay
;
813 struct delayed_work master_wakeup_work
;
814 struct netns_ipvs
*ipvs
;
817 struct ip_vs_sync_thread_data
;
819 /* How much time to keep dests in trash */
820 #define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
822 struct ipvs_sync_daemon_cfg
{
823 union nf_inet_addr mcast_group
;
829 /* multicast interface name */
830 char mcast_ifn
[IP_VS_IFNAME_MAXLEN
];
833 /* IPVS in network namespace */
835 int gen
; /* Generation */
836 int enable
; /* enable like nf_hooks do */
837 /* Hash table: for real service lookups */
838 #define IP_VS_RTAB_BITS 4
839 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
840 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
842 struct hlist_head rs_table
[IP_VS_RTAB_SIZE
];
844 struct list_head app_list
;
846 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
847 struct ip_vs_proto_data
*proto_data_table
[IP_VS_PROTO_TAB_SIZE
];
848 /* ip_vs_proto_tcp */
849 #ifdef CONFIG_IP_VS_PROTO_TCP
850 #define TCP_APP_TAB_BITS 4
851 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
852 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
853 struct list_head tcp_apps
[TCP_APP_TAB_SIZE
];
855 /* ip_vs_proto_udp */
856 #ifdef CONFIG_IP_VS_PROTO_UDP
857 #define UDP_APP_TAB_BITS 4
858 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
859 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
860 struct list_head udp_apps
[UDP_APP_TAB_SIZE
];
862 /* ip_vs_proto_sctp */
863 #ifdef CONFIG_IP_VS_PROTO_SCTP
864 #define SCTP_APP_TAB_BITS 4
865 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
866 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
867 /* Hash table for SCTP application incarnations */
868 struct list_head sctp_apps
[SCTP_APP_TAB_SIZE
];
871 atomic_t conn_count
; /* connection counter */
874 struct ip_vs_stats tot_stats
; /* Statistics & est. */
876 int num_services
; /* no of virtual services */
878 /* Trash for destinations */
879 struct list_head dest_trash
;
880 spinlock_t dest_trash_lock
;
881 struct timer_list dest_trash_timer
; /* expiration timer */
882 /* Service counters */
883 atomic_t ftpsvc_counter
;
884 atomic_t nullsvc_counter
;
885 atomic_t conn_out_counter
;
888 /* 1/rate drop and drop-entry variables */
889 struct delayed_work defense_work
; /* Work handler */
894 spinlock_t dropentry_lock
; /* drop entry handling */
895 spinlock_t droppacket_lock
; /* drop packet handling */
896 spinlock_t securetcp_lock
; /* state and timeout tables */
899 struct ctl_table_header
*sysctl_hdr
;
900 struct ctl_table
*sysctl_tbl
;
903 /* sysctl variables */
904 int sysctl_amemthresh
;
905 int sysctl_am_droprate
;
906 int sysctl_drop_entry
;
907 int sysctl_drop_packet
;
908 int sysctl_secure_tcp
;
909 #ifdef CONFIG_IP_VS_NFCT
910 int sysctl_conntrack
;
912 int sysctl_snat_reroute
;
914 int sysctl_sync_ports
;
915 int sysctl_sync_persist_mode
;
916 unsigned long sysctl_sync_qlen_max
;
917 int sysctl_sync_sock_size
;
918 int sysctl_cache_bypass
;
919 int sysctl_expire_nodest_conn
;
920 int sysctl_sloppy_tcp
;
921 int sysctl_sloppy_sctp
;
922 int sysctl_expire_quiescent_template
;
923 int sysctl_sync_threshold
[2];
924 unsigned int sysctl_sync_refresh_period
;
925 int sysctl_sync_retries
;
926 int sysctl_nat_icmp_send
;
927 int sysctl_pmtu_disc
;
928 int sysctl_backup_only
;
929 int sysctl_conn_reuse_mode
;
930 int sysctl_schedule_icmp
;
931 int sysctl_ignore_tunneled
;
934 int sysctl_lblc_expiration
;
935 struct ctl_table_header
*lblc_ctl_header
;
936 struct ctl_table
*lblc_ctl_table
;
938 int sysctl_lblcr_expiration
;
939 struct ctl_table_header
*lblcr_ctl_header
;
940 struct ctl_table
*lblcr_ctl_table
;
942 struct list_head est_list
; /* estimator list */
944 struct timer_list est_timer
; /* Estimation timer */
946 spinlock_t sync_lock
;
947 struct ipvs_master_sync_state
*ms
;
948 spinlock_t sync_buff_lock
;
949 struct ip_vs_sync_thread_data
*master_tinfo
;
950 struct ip_vs_sync_thread_data
*backup_tinfo
;
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(const struct ip_vs_conn
*cp
);
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 /* Mark our template as assured */
1305 ip_vs_control_assure_ct(struct ip_vs_conn
*cp
)
1307 struct ip_vs_conn
*ct
= cp
->control
;
1309 if (ct
&& !(ct
->state
& IP_VS_CTPL_S_ASSURED
) &&
1310 (ct
->flags
& IP_VS_CONN_F_TEMPLATE
))
1311 ct
->state
|= IP_VS_CTPL_S_ASSURED
;
1314 /* IPVS netns init & cleanup functions */
1315 int ip_vs_estimator_net_init(struct netns_ipvs
*ipvs
);
1316 int ip_vs_control_net_init(struct netns_ipvs
*ipvs
);
1317 int ip_vs_protocol_net_init(struct netns_ipvs
*ipvs
);
1318 int ip_vs_app_net_init(struct netns_ipvs
*ipvs
);
1319 int ip_vs_conn_net_init(struct netns_ipvs
*ipvs
);
1320 int ip_vs_sync_net_init(struct netns_ipvs
*ipvs
);
1321 void ip_vs_conn_net_cleanup(struct netns_ipvs
*ipvs
);
1322 void ip_vs_app_net_cleanup(struct netns_ipvs
*ipvs
);
1323 void ip_vs_protocol_net_cleanup(struct netns_ipvs
*ipvs
);
1324 void ip_vs_control_net_cleanup(struct netns_ipvs
*ipvs
);
1325 void ip_vs_estimator_net_cleanup(struct netns_ipvs
*ipvs
);
1326 void ip_vs_sync_net_cleanup(struct netns_ipvs
*ipvs
);
1327 void ip_vs_service_net_cleanup(struct netns_ipvs
*ipvs
);
1329 /* IPVS application functions
1330 * (from ip_vs_app.c)
1332 #define IP_VS_APP_MAX_PORTS 8
1333 struct ip_vs_app
*register_ip_vs_app(struct netns_ipvs
*ipvs
, struct ip_vs_app
*app
);
1334 void unregister_ip_vs_app(struct netns_ipvs
*ipvs
, struct ip_vs_app
*app
);
1335 int ip_vs_bind_app(struct ip_vs_conn
*cp
, struct ip_vs_protocol
*pp
);
1336 void ip_vs_unbind_app(struct ip_vs_conn
*cp
);
1337 int register_ip_vs_app_inc(struct netns_ipvs
*ipvs
, struct ip_vs_app
*app
, __u16 proto
,
1339 int ip_vs_app_inc_get(struct ip_vs_app
*inc
);
1340 void ip_vs_app_inc_put(struct ip_vs_app
*inc
);
1342 int ip_vs_app_pkt_out(struct ip_vs_conn
*, struct sk_buff
*skb
,
1343 struct ip_vs_iphdr
*ipvsh
);
1344 int ip_vs_app_pkt_in(struct ip_vs_conn
*, struct sk_buff
*skb
,
1345 struct ip_vs_iphdr
*ipvsh
);
1347 int register_ip_vs_pe(struct ip_vs_pe
*pe
);
1348 int unregister_ip_vs_pe(struct ip_vs_pe
*pe
);
1349 struct ip_vs_pe
*ip_vs_pe_getbyname(const char *name
);
1350 struct ip_vs_pe
*__ip_vs_pe_getbyname(const char *pe_name
);
1352 /* Use a #define to avoid all of module.h just for these trivial ops */
1353 #define ip_vs_pe_get(pe) \
1354 if (pe && pe->module) \
1355 __module_get(pe->module);
1357 #define ip_vs_pe_put(pe) \
1358 if (pe && pe->module) \
1359 module_put(pe->module);
1361 /* IPVS protocol functions (from ip_vs_proto.c) */
1362 int ip_vs_protocol_init(void);
1363 void ip_vs_protocol_cleanup(void);
1364 void ip_vs_protocol_timeout_change(struct netns_ipvs
*ipvs
, int flags
);
1365 int *ip_vs_create_timeout_table(int *table
, int size
);
1366 void ip_vs_tcpudp_debug_packet(int af
, struct ip_vs_protocol
*pp
,
1367 const struct sk_buff
*skb
, int offset
,
1370 extern struct ip_vs_protocol ip_vs_protocol_tcp
;
1371 extern struct ip_vs_protocol ip_vs_protocol_udp
;
1372 extern struct ip_vs_protocol ip_vs_protocol_icmp
;
1373 extern struct ip_vs_protocol ip_vs_protocol_esp
;
1374 extern struct ip_vs_protocol ip_vs_protocol_ah
;
1375 extern struct ip_vs_protocol ip_vs_protocol_sctp
;
1377 /* Registering/unregistering scheduler functions
1378 * (from ip_vs_sched.c)
1380 int register_ip_vs_scheduler(struct ip_vs_scheduler
*scheduler
);
1381 int unregister_ip_vs_scheduler(struct ip_vs_scheduler
*scheduler
);
1382 int ip_vs_bind_scheduler(struct ip_vs_service
*svc
,
1383 struct ip_vs_scheduler
*scheduler
);
1384 void ip_vs_unbind_scheduler(struct ip_vs_service
*svc
,
1385 struct ip_vs_scheduler
*sched
);
1386 struct ip_vs_scheduler
*ip_vs_scheduler_get(const char *sched_name
);
1387 void ip_vs_scheduler_put(struct ip_vs_scheduler
*scheduler
);
1389 ip_vs_schedule(struct ip_vs_service
*svc
, struct sk_buff
*skb
,
1390 struct ip_vs_proto_data
*pd
, int *ignored
,
1391 struct ip_vs_iphdr
*iph
);
1392 int ip_vs_leave(struct ip_vs_service
*svc
, struct sk_buff
*skb
,
1393 struct ip_vs_proto_data
*pd
, struct ip_vs_iphdr
*iph
);
1395 void ip_vs_scheduler_err(struct ip_vs_service
*svc
, const char *msg
);
1397 /* IPVS control data and functions (from ip_vs_ctl.c) */
1398 extern struct ip_vs_stats ip_vs_stats
;
1399 extern int sysctl_ip_vs_sync_ver
;
1401 struct ip_vs_service
*
1402 ip_vs_service_find(struct netns_ipvs
*ipvs
, int af
, __u32 fwmark
, __u16 protocol
,
1403 const union nf_inet_addr
*vaddr
, __be16 vport
);
1405 bool ip_vs_has_real_service(struct netns_ipvs
*ipvs
, int af
, __u16 protocol
,
1406 const union nf_inet_addr
*daddr
, __be16 dport
);
1409 ip_vs_find_real_service(struct netns_ipvs
*ipvs
, int af
, __u16 protocol
,
1410 const union nf_inet_addr
*daddr
, __be16 dport
);
1411 struct ip_vs_dest
*ip_vs_find_tunnel(struct netns_ipvs
*ipvs
, int af
,
1412 const union nf_inet_addr
*daddr
,
1415 int ip_vs_use_count_inc(void);
1416 void ip_vs_use_count_dec(void);
1417 int ip_vs_register_nl_ioctl(void);
1418 void ip_vs_unregister_nl_ioctl(void);
1419 int ip_vs_control_init(void);
1420 void ip_vs_control_cleanup(void);
1422 ip_vs_find_dest(struct netns_ipvs
*ipvs
, int svc_af
, int dest_af
,
1423 const union nf_inet_addr
*daddr
, __be16 dport
,
1424 const union nf_inet_addr
*vaddr
, __be16 vport
,
1425 __u16 protocol
, __u32 fwmark
, __u32 flags
);
1426 void ip_vs_try_bind_dest(struct ip_vs_conn
*cp
);
1428 static inline void ip_vs_dest_hold(struct ip_vs_dest
*dest
)
1430 refcount_inc(&dest
->refcnt
);
1433 static inline void ip_vs_dest_put(struct ip_vs_dest
*dest
)
1435 smp_mb__before_atomic();
1436 refcount_dec(&dest
->refcnt
);
1439 static inline void ip_vs_dest_put_and_free(struct ip_vs_dest
*dest
)
1441 if (refcount_dec_and_test(&dest
->refcnt
))
1445 /* IPVS sync daemon data and function prototypes
1446 * (from ip_vs_sync.c)
1448 int start_sync_thread(struct netns_ipvs
*ipvs
, struct ipvs_sync_daemon_cfg
*cfg
,
1450 int stop_sync_thread(struct netns_ipvs
*ipvs
, int state
);
1451 void ip_vs_sync_conn(struct netns_ipvs
*ipvs
, struct ip_vs_conn
*cp
, int pkts
);
1453 /* IPVS rate estimator prototypes (from ip_vs_est.c) */
1454 void ip_vs_start_estimator(struct netns_ipvs
*ipvs
, struct ip_vs_stats
*stats
);
1455 void ip_vs_stop_estimator(struct netns_ipvs
*ipvs
, struct ip_vs_stats
*stats
);
1456 void ip_vs_zero_estimator(struct ip_vs_stats
*stats
);
1457 void ip_vs_read_estimator(struct ip_vs_kstats
*dst
, struct ip_vs_stats
*stats
);
1459 /* Various IPVS packet transmitters (from ip_vs_xmit.c) */
1460 int ip_vs_null_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1461 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1462 int ip_vs_bypass_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1463 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1464 int ip_vs_nat_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1465 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1466 int ip_vs_tunnel_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1467 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1468 int ip_vs_dr_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1469 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1470 int ip_vs_icmp_xmit(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1471 struct ip_vs_protocol
*pp
, int offset
,
1472 unsigned int hooknum
, struct ip_vs_iphdr
*iph
);
1473 void ip_vs_dest_dst_rcu_free(struct rcu_head
*head
);
1475 #ifdef CONFIG_IP_VS_IPV6
1476 int ip_vs_bypass_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1477 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1478 int ip_vs_nat_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1479 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1480 int ip_vs_tunnel_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1481 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1482 int ip_vs_dr_xmit_v6(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1483 struct ip_vs_protocol
*pp
, struct ip_vs_iphdr
*iph
);
1484 int ip_vs_icmp_xmit_v6(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
);
1489 #ifdef CONFIG_SYSCTL
1490 /* This is a simple mechanism to ignore packets when
1491 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1492 * we start to drop 1/rate of the packets
1494 static inline int ip_vs_todrop(struct netns_ipvs
*ipvs
)
1496 if (!ipvs
->drop_rate
)
1498 if (--ipvs
->drop_counter
> 0)
1500 ipvs
->drop_counter
= ipvs
->drop_rate
;
1504 static inline int ip_vs_todrop(struct netns_ipvs
*ipvs
) { return 0; }
1507 #define IP_VS_DFWD_METHOD(dest) (atomic_read(&(dest)->conn_flags) & \
1508 IP_VS_CONN_F_FWD_MASK)
1510 /* ip_vs_fwd_tag returns the forwarding tag of the connection */
1511 #define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1513 static inline char ip_vs_fwd_tag(struct ip_vs_conn
*cp
)
1517 switch (IP_VS_FWD_METHOD(cp
)) {
1518 case IP_VS_CONN_F_MASQ
:
1520 case IP_VS_CONN_F_LOCALNODE
:
1522 case IP_VS_CONN_F_TUNNEL
:
1524 case IP_VS_CONN_F_DROUTE
:
1526 case IP_VS_CONN_F_BYPASS
:
1534 void ip_vs_nat_icmp(struct sk_buff
*skb
, struct ip_vs_protocol
*pp
,
1535 struct ip_vs_conn
*cp
, int dir
);
1537 #ifdef CONFIG_IP_VS_IPV6
1538 void ip_vs_nat_icmp_v6(struct sk_buff
*skb
, struct ip_vs_protocol
*pp
,
1539 struct ip_vs_conn
*cp
, int dir
);
1542 __sum16
ip_vs_checksum_complete(struct sk_buff
*skb
, int offset
);
1544 static inline __wsum
ip_vs_check_diff4(__be32 old
, __be32
new, __wsum oldsum
)
1546 __be32 diff
[2] = { ~old
, new };
1548 return csum_partial(diff
, sizeof(diff
), oldsum
);
1551 #ifdef CONFIG_IP_VS_IPV6
1552 static inline __wsum
ip_vs_check_diff16(const __be32
*old
, const __be32
*new,
1555 __be32 diff
[8] = { ~old
[3], ~old
[2], ~old
[1], ~old
[0],
1556 new[3], new[2], new[1], new[0] };
1558 return csum_partial(diff
, sizeof(diff
), oldsum
);
1562 static inline __wsum
ip_vs_check_diff2(__be16 old
, __be16
new, __wsum oldsum
)
1564 __be16 diff
[2] = { ~old
, new };
1566 return csum_partial(diff
, sizeof(diff
), oldsum
);
1569 /* Forget current conntrack (unconfirmed) and attach notrack entry */
1570 static inline void ip_vs_notrack(struct sk_buff
*skb
)
1572 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1573 enum ip_conntrack_info ctinfo
;
1574 struct nf_conn
*ct
= nf_ct_get(skb
, &ctinfo
);
1577 nf_conntrack_put(&ct
->ct_general
);
1578 nf_ct_set(skb
, NULL
, IP_CT_UNTRACKED
);
1583 #ifdef CONFIG_IP_VS_NFCT
1584 /* Netfilter connection tracking
1585 * (from ip_vs_nfct.c)
1587 static inline int ip_vs_conntrack_enabled(struct netns_ipvs
*ipvs
)
1589 #ifdef CONFIG_SYSCTL
1590 return ipvs
->sysctl_conntrack
;
1596 void ip_vs_update_conntrack(struct sk_buff
*skb
, struct ip_vs_conn
*cp
,
1598 int ip_vs_confirm_conntrack(struct sk_buff
*skb
);
1599 void ip_vs_nfct_expect_related(struct sk_buff
*skb
, struct nf_conn
*ct
,
1600 struct ip_vs_conn
*cp
, u_int8_t proto
,
1601 const __be16 port
, int from_rs
);
1602 void ip_vs_conn_drop_conntrack(struct ip_vs_conn
*cp
);
1606 static inline int ip_vs_conntrack_enabled(struct netns_ipvs
*ipvs
)
1611 static inline void ip_vs_update_conntrack(struct sk_buff
*skb
,
1612 struct ip_vs_conn
*cp
, int outin
)
1616 static inline int ip_vs_confirm_conntrack(struct sk_buff
*skb
)
1621 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn
*cp
)
1624 #endif /* CONFIG_IP_VS_NFCT */
1626 /* Really using conntrack? */
1627 static inline bool ip_vs_conn_uses_conntrack(struct ip_vs_conn
*cp
,
1628 struct sk_buff
*skb
)
1630 #ifdef CONFIG_IP_VS_NFCT
1631 enum ip_conntrack_info ctinfo
;
1634 if (!(cp
->flags
& IP_VS_CONN_F_NFCT
))
1636 ct
= nf_ct_get(skb
, &ctinfo
);
1643 static inline int ip_vs_register_conntrack(struct ip_vs_service
*svc
)
1645 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
1646 int afmask
= (svc
->af
== AF_INET6
) ? 2 : 1;
1649 if (!(svc
->conntrack_afmask
& afmask
)) {
1650 ret
= nf_ct_netns_get(svc
->ipvs
->net
, svc
->af
);
1652 svc
->conntrack_afmask
|= afmask
;
1660 static inline void ip_vs_unregister_conntrack(struct ip_vs_service
*svc
)
1662 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
1663 int afmask
= (svc
->af
== AF_INET6
) ? 2 : 1;
1665 if (svc
->conntrack_afmask
& afmask
) {
1666 nf_ct_netns_put(svc
->ipvs
->net
, svc
->af
);
1667 svc
->conntrack_afmask
&= ~afmask
;
1673 ip_vs_dest_conn_overhead(struct ip_vs_dest
*dest
)
1675 /* We think the overhead of processing active connections is 256
1676 * times higher than that of inactive connections in average. (This
1677 * 256 times might not be accurate, we will change it later) We
1678 * use the following formula to estimate the overhead now:
1679 * dest->activeconns*256 + dest->inactconns
1681 return (atomic_read(&dest
->activeconns
) << 8) +
1682 atomic_read(&dest
->inactconns
);
1685 #endif /* _NET_IP_VS_H */