1 /* SPDX-License-Identifier: GPL-2.0 */
2 /* IP Virtual Server
3  * data structure and functionality definitions
4  */
5 
6 #ifndef _NET_IP_VS_H
7 #define _NET_IP_VS_H
8 
9 #include <linux/ip_vs.h>                /* definitions shared with userland */
10 
11 #include <asm/types.h>                  /* for __uXX types */
12 
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 */
17 #include <linux/workqueue.h>
18 
19 #include <linux/compiler.h>
20 #include <linux/timer.h>
21 #include <linux/bug.h>
22 
23 #include <net/checksum.h>
24 #include <linux/netfilter.h>		/* for union nf_inet_addr */
25 #include <linux/ip.h>
26 #include <linux/ipv6.h>			/* for struct ipv6hdr */
27 #include <net/ipv6.h>
28 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
29 #include <net/netfilter/nf_conntrack.h>
30 #endif
31 #include <net/net_namespace.h>		/* Netw namespace */
32 
33 #define IP_VS_HDR_INVERSE	1
34 #define IP_VS_HDR_ICMP		2
35 
36 /* Generic access of ipvs struct */
net_ipvs(struct net * net)37 static inline struct netns_ipvs *net_ipvs(struct net* net)
38 {
39 	return net->ipvs;
40 }
41 
42 /* Connections' size value needed by ip_vs_ctl.c */
43 extern int ip_vs_conn_tab_size;
44 
45 struct ip_vs_iphdr {
46 	int hdr_flags;	/* ipvs flags */
47 	__u32 off;	/* Where IP or IPv4 header starts */
48 	__u32 len;	/* IPv4 simply where L4 starts
49 			 * IPv6 where L4 Transport Header starts */
50 	__u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
51 	__s16 protocol;
52 	__s32 flags;
53 	union nf_inet_addr saddr;
54 	union nf_inet_addr daddr;
55 };
56 
frag_safe_skb_hp(const struct sk_buff * skb,int offset,int len,void * buffer)57 static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
58 				      int len, void *buffer)
59 {
60 	return skb_header_pointer(skb, offset, len, buffer);
61 }
62 
63 /* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
64  * IPv6 requires some extra work, as finding proper header position,
65  * depend on the IPv6 extension headers.
66  */
67 static inline int
ip_vs_fill_iph_skb_off(int af,const struct sk_buff * skb,int offset,int hdr_flags,struct ip_vs_iphdr * iphdr)68 ip_vs_fill_iph_skb_off(int af, const struct sk_buff *skb, int offset,
69 		       int hdr_flags, struct ip_vs_iphdr *iphdr)
70 {
71 	iphdr->hdr_flags = hdr_flags;
72 	iphdr->off = offset;
73 
74 #ifdef CONFIG_IP_VS_IPV6
75 	if (af == AF_INET6) {
76 		struct ipv6hdr _iph;
77 		const struct ipv6hdr *iph = skb_header_pointer(
78 			skb, offset, sizeof(_iph), &_iph);
79 		if (!iph)
80 			return 0;
81 
82 		iphdr->saddr.in6 = iph->saddr;
83 		iphdr->daddr.in6 = iph->daddr;
84 		/* ipv6_find_hdr() updates len, flags */
85 		iphdr->len	 = offset;
86 		iphdr->flags	 = 0;
87 		iphdr->protocol  = ipv6_find_hdr(skb, &iphdr->len, -1,
88 						 &iphdr->fragoffs,
89 						 &iphdr->flags);
90 		if (iphdr->protocol < 0)
91 			return 0;
92 	} else
93 #endif
94 	{
95 		struct iphdr _iph;
96 		const struct iphdr *iph = skb_header_pointer(
97 			skb, offset, sizeof(_iph), &_iph);
98 		if (!iph)
99 			return 0;
100 
101 		iphdr->len	= offset + iph->ihl * 4;
102 		iphdr->fragoffs	= 0;
103 		iphdr->protocol	= iph->protocol;
104 		iphdr->saddr.ip	= iph->saddr;
105 		iphdr->daddr.ip	= iph->daddr;
106 	}
107 
108 	return 1;
109 }
110 
111 static inline int
ip_vs_fill_iph_skb_icmp(int af,const struct sk_buff * skb,int offset,bool inverse,struct ip_vs_iphdr * iphdr)112 ip_vs_fill_iph_skb_icmp(int af, const struct sk_buff *skb, int offset,
113 			bool inverse, struct ip_vs_iphdr *iphdr)
114 {
115 	int hdr_flags = IP_VS_HDR_ICMP;
116 
117 	if (inverse)
118 		hdr_flags |= IP_VS_HDR_INVERSE;
119 
120 	return ip_vs_fill_iph_skb_off(af, skb, offset, hdr_flags, iphdr);
121 }
122 
123 static inline int
ip_vs_fill_iph_skb(int af,const struct sk_buff * skb,bool inverse,struct ip_vs_iphdr * iphdr)124 ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, bool inverse,
125 		   struct ip_vs_iphdr *iphdr)
126 {
127 	int hdr_flags = 0;
128 
129 	if (inverse)
130 		hdr_flags |= IP_VS_HDR_INVERSE;
131 
132 	return ip_vs_fill_iph_skb_off(af, skb, skb_network_offset(skb),
133 				      hdr_flags, iphdr);
134 }
135 
136 static inline bool
ip_vs_iph_inverse(const struct ip_vs_iphdr * iph)137 ip_vs_iph_inverse(const struct ip_vs_iphdr *iph)
138 {
139 	return !!(iph->hdr_flags & IP_VS_HDR_INVERSE);
140 }
141 
142 static inline bool
ip_vs_iph_icmp(const struct ip_vs_iphdr * iph)143 ip_vs_iph_icmp(const struct ip_vs_iphdr *iph)
144 {
145 	return !!(iph->hdr_flags & IP_VS_HDR_ICMP);
146 }
147 
ip_vs_addr_copy(int af,union nf_inet_addr * dst,const union nf_inet_addr * src)148 static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
149 				   const union nf_inet_addr *src)
150 {
151 #ifdef CONFIG_IP_VS_IPV6
152 	if (af == AF_INET6)
153 		dst->in6 = src->in6;
154 	else
155 #endif
156 	dst->ip = src->ip;
157 }
158 
ip_vs_addr_set(int af,union nf_inet_addr * dst,const union nf_inet_addr * src)159 static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
160 				  const union nf_inet_addr *src)
161 {
162 #ifdef CONFIG_IP_VS_IPV6
163 	if (af == AF_INET6) {
164 		dst->in6 = src->in6;
165 		return;
166 	}
167 #endif
168 	dst->ip = src->ip;
169 	dst->all[1] = 0;
170 	dst->all[2] = 0;
171 	dst->all[3] = 0;
172 }
173 
ip_vs_addr_equal(int af,const union nf_inet_addr * a,const union nf_inet_addr * b)174 static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
175 				   const union nf_inet_addr *b)
176 {
177 #ifdef CONFIG_IP_VS_IPV6
178 	if (af == AF_INET6)
179 		return ipv6_addr_equal(&a->in6, &b->in6);
180 #endif
181 	return a->ip == b->ip;
182 }
183 
184 #ifdef CONFIG_IP_VS_DEBUG
185 #include <linux/net.h>
186 
187 int ip_vs_get_debug_level(void);
188 
ip_vs_dbg_addr(int af,char * buf,size_t buf_len,const union nf_inet_addr * addr,int * idx)189 static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
190 					 const union nf_inet_addr *addr,
191 					 int *idx)
192 {
193 	int len;
194 #ifdef CONFIG_IP_VS_IPV6
195 	if (af == AF_INET6)
196 		len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
197 			       &addr->in6) + 1;
198 	else
199 #endif
200 		len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
201 			       &addr->ip) + 1;
202 
203 	*idx += len;
204 	BUG_ON(*idx > buf_len + 1);
205 	return &buf[*idx - len];
206 }
207 
208 #define IP_VS_DBG_BUF(level, msg, ...)					\
209 	do {								\
210 		char ip_vs_dbg_buf[160];				\
211 		int ip_vs_dbg_idx = 0;					\
212 		if (level <= ip_vs_get_debug_level())			\
213 			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
214 	} while (0)
215 #define IP_VS_ERR_BUF(msg...)						\
216 	do {								\
217 		char ip_vs_dbg_buf[160];				\
218 		int ip_vs_dbg_idx = 0;					\
219 		pr_err(msg);						\
220 	} while (0)
221 
222 /* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
223 #define IP_VS_DBG_ADDR(af, addr)					\
224 	ip_vs_dbg_addr(af, ip_vs_dbg_buf,				\
225 		       sizeof(ip_vs_dbg_buf), addr,			\
226 		       &ip_vs_dbg_idx)
227 
228 #define IP_VS_DBG(level, msg, ...)					\
229 	do {								\
230 		if (level <= ip_vs_get_debug_level())			\
231 			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
232 	} while (0)
233 #define IP_VS_DBG_RL(msg, ...)						\
234 	do {								\
235 		if (net_ratelimit())					\
236 			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
237 	} while (0)
238 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)			\
239 	do {								\
240 		if (level <= ip_vs_get_debug_level())			\
241 			pp->debug_packet(af, pp, skb, ofs, msg);	\
242 	} while (0)
243 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)			\
244 	do {								\
245 		if (level <= ip_vs_get_debug_level() &&			\
246 		    net_ratelimit())					\
247 			pp->debug_packet(af, pp, skb, ofs, msg);	\
248 	} while (0)
249 #else	/* NO DEBUGGING at ALL */
250 #define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
251 #define IP_VS_ERR_BUF(msg...)  do {} while (0)
252 #define IP_VS_DBG(level, msg...)  do {} while (0)
253 #define IP_VS_DBG_RL(msg...)  do {} while (0)
254 #define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)	do {} while (0)
255 #define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)	do {} while (0)
256 #endif
257 
258 #define IP_VS_BUG() BUG()
259 #define IP_VS_ERR_RL(msg, ...)						\
260 	do {								\
261 		if (net_ratelimit())					\
262 			pr_err(msg, ##__VA_ARGS__);			\
263 	} while (0)
264 
265 #ifdef CONFIG_IP_VS_DEBUG
266 #define EnterFunction(level)						\
267 	do {								\
268 		if (level <= ip_vs_get_debug_level())			\
269 			printk(KERN_DEBUG				\
270 			       pr_fmt("Enter: %s, %s line %i\n"),	\
271 			       __func__, __FILE__, __LINE__);		\
272 	} while (0)
273 #define LeaveFunction(level)						\
274 	do {								\
275 		if (level <= ip_vs_get_debug_level())			\
276 			printk(KERN_DEBUG				\
277 			       pr_fmt("Leave: %s, %s line %i\n"),	\
278 			       __func__, __FILE__, __LINE__);		\
279 	} while (0)
280 #else
281 #define EnterFunction(level)   do {} while (0)
282 #define LeaveFunction(level)   do {} while (0)
283 #endif
284 
285 /* The port number of FTP service (in network order). */
286 #define FTPPORT  cpu_to_be16(21)
287 #define FTPDATA  cpu_to_be16(20)
288 
289 /* TCP State Values */
290 enum {
291 	IP_VS_TCP_S_NONE = 0,
292 	IP_VS_TCP_S_ESTABLISHED,
293 	IP_VS_TCP_S_SYN_SENT,
294 	IP_VS_TCP_S_SYN_RECV,
295 	IP_VS_TCP_S_FIN_WAIT,
296 	IP_VS_TCP_S_TIME_WAIT,
297 	IP_VS_TCP_S_CLOSE,
298 	IP_VS_TCP_S_CLOSE_WAIT,
299 	IP_VS_TCP_S_LAST_ACK,
300 	IP_VS_TCP_S_LISTEN,
301 	IP_VS_TCP_S_SYNACK,
302 	IP_VS_TCP_S_LAST
303 };
304 
305 /* UDP State Values */
306 enum {
307 	IP_VS_UDP_S_NORMAL,
308 	IP_VS_UDP_S_LAST,
309 };
310 
311 /* ICMP State Values */
312 enum {
313 	IP_VS_ICMP_S_NORMAL,
314 	IP_VS_ICMP_S_LAST,
315 };
316 
317 /* SCTP State Values */
318 enum ip_vs_sctp_states {
319 	IP_VS_SCTP_S_NONE,
320 	IP_VS_SCTP_S_INIT1,
321 	IP_VS_SCTP_S_INIT,
322 	IP_VS_SCTP_S_COOKIE_SENT,
323 	IP_VS_SCTP_S_COOKIE_REPLIED,
324 	IP_VS_SCTP_S_COOKIE_WAIT,
325 	IP_VS_SCTP_S_COOKIE,
326 	IP_VS_SCTP_S_COOKIE_ECHOED,
327 	IP_VS_SCTP_S_ESTABLISHED,
328 	IP_VS_SCTP_S_SHUTDOWN_SENT,
329 	IP_VS_SCTP_S_SHUTDOWN_RECEIVED,
330 	IP_VS_SCTP_S_SHUTDOWN_ACK_SENT,
331 	IP_VS_SCTP_S_REJECTED,
332 	IP_VS_SCTP_S_CLOSED,
333 	IP_VS_SCTP_S_LAST
334 };
335 
336 /* Connection templates use bits from state */
337 #define IP_VS_CTPL_S_NONE		0x0000
338 #define IP_VS_CTPL_S_ASSURED		0x0001
339 #define IP_VS_CTPL_S_LAST		0x0002
340 
341 /* Delta sequence info structure
342  * Each ip_vs_conn has 2 (output AND input seq. changes).
343  * Only used in the VS/NAT.
344  */
345 struct ip_vs_seq {
346 	__u32			init_seq;	/* Add delta from this seq */
347 	__u32			delta;		/* Delta in sequence numbers */
348 	__u32			previous_delta;	/* Delta in sequence numbers
349 						 * before last resized pkt */
350 };
351 
352 /* counters per cpu */
353 struct ip_vs_counters {
354 	u64_stats_t	conns;		/* connections scheduled */
355 	u64_stats_t	inpkts;		/* incoming packets */
356 	u64_stats_t	outpkts;	/* outgoing packets */
357 	u64_stats_t	inbytes;	/* incoming bytes */
358 	u64_stats_t	outbytes;	/* outgoing bytes */
359 };
360 /* Stats per cpu */
361 struct ip_vs_cpu_stats {
362 	struct ip_vs_counters   cnt;
363 	struct u64_stats_sync   syncp;
364 };
365 
366 /* IPVS statistics objects */
367 struct ip_vs_estimator {
368 	struct list_head	list;
369 
370 	u64			last_inbytes;
371 	u64			last_outbytes;
372 	u64			last_conns;
373 	u64			last_inpkts;
374 	u64			last_outpkts;
375 
376 	u64			cps;
377 	u64			inpps;
378 	u64			outpps;
379 	u64			inbps;
380 	u64			outbps;
381 };
382 
383 /*
384  * IPVS statistics object, 64-bit kernel version of struct ip_vs_stats_user
385  */
386 struct ip_vs_kstats {
387 	u64			conns;		/* connections scheduled */
388 	u64			inpkts;		/* incoming packets */
389 	u64			outpkts;	/* outgoing packets */
390 	u64			inbytes;	/* incoming bytes */
391 	u64			outbytes;	/* outgoing bytes */
392 
393 	u64			cps;		/* current connection rate */
394 	u64			inpps;		/* current in packet rate */
395 	u64			outpps;		/* current out packet rate */
396 	u64			inbps;		/* current in byte rate */
397 	u64			outbps;		/* current out byte rate */
398 };
399 
400 struct ip_vs_stats {
401 	struct ip_vs_kstats	kstats;		/* kernel statistics */
402 	struct ip_vs_estimator	est;		/* estimator */
403 	struct ip_vs_cpu_stats __percpu	*cpustats;	/* per cpu counters */
404 	spinlock_t		lock;		/* spin lock */
405 	struct ip_vs_kstats	kstats0;	/* reset values */
406 };
407 
408 struct dst_entry;
409 struct iphdr;
410 struct ip_vs_conn;
411 struct ip_vs_app;
412 struct sk_buff;
413 struct ip_vs_proto_data;
414 
415 struct ip_vs_protocol {
416 	struct ip_vs_protocol	*next;
417 	char			*name;
418 	u16			protocol;
419 	u16			num_states;
420 	int			dont_defrag;
421 
422 	void (*init)(struct ip_vs_protocol *pp);
423 
424 	void (*exit)(struct ip_vs_protocol *pp);
425 
426 	int (*init_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
427 
428 	void (*exit_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
429 
430 	int (*conn_schedule)(struct netns_ipvs *ipvs,
431 			     int af, struct sk_buff *skb,
432 			     struct ip_vs_proto_data *pd,
433 			     int *verdict, struct ip_vs_conn **cpp,
434 			     struct ip_vs_iphdr *iph);
435 
436 	struct ip_vs_conn *
437 	(*conn_in_get)(struct netns_ipvs *ipvs,
438 		       int af,
439 		       const struct sk_buff *skb,
440 		       const struct ip_vs_iphdr *iph);
441 
442 	struct ip_vs_conn *
443 	(*conn_out_get)(struct netns_ipvs *ipvs,
444 			int af,
445 			const struct sk_buff *skb,
446 			const struct ip_vs_iphdr *iph);
447 
448 	int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
449 			    struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
450 
451 	int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
452 			    struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
453 
454 	const char *(*state_name)(int state);
455 
456 	void (*state_transition)(struct ip_vs_conn *cp, int direction,
457 				 const struct sk_buff *skb,
458 				 struct ip_vs_proto_data *pd);
459 
460 	int (*register_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
461 
462 	void (*unregister_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
463 
464 	int (*app_conn_bind)(struct ip_vs_conn *cp);
465 
466 	void (*debug_packet)(int af, struct ip_vs_protocol *pp,
467 			     const struct sk_buff *skb,
468 			     int offset,
469 			     const char *msg);
470 
471 	void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
472 };
473 
474 /* protocol data per netns */
475 struct ip_vs_proto_data {
476 	struct ip_vs_proto_data	*next;
477 	struct ip_vs_protocol	*pp;
478 	int			*timeout_table;	/* protocol timeout table */
479 	atomic_t		appcnt;		/* counter of proto app incs. */
480 	struct tcp_states_t	*tcp_state_table;
481 };
482 
483 struct ip_vs_protocol   *ip_vs_proto_get(unsigned short proto);
484 struct ip_vs_proto_data *ip_vs_proto_data_get(struct netns_ipvs *ipvs,
485 					      unsigned short proto);
486 
487 struct ip_vs_conn_param {
488 	struct netns_ipvs		*ipvs;
489 	const union nf_inet_addr	*caddr;
490 	const union nf_inet_addr	*vaddr;
491 	__be16				cport;
492 	__be16				vport;
493 	__u16				protocol;
494 	u16				af;
495 
496 	const struct ip_vs_pe		*pe;
497 	char				*pe_data;
498 	__u8				pe_data_len;
499 };
500 
501 /* IP_VS structure allocated for each dynamically scheduled connection */
502 struct ip_vs_conn {
503 	struct hlist_node	c_list;         /* hashed list heads */
504 	/* Protocol, addresses and port numbers */
505 	__be16                  cport;
506 	__be16                  dport;
507 	__be16                  vport;
508 	u16			af;		/* address family */
509 	union nf_inet_addr      caddr;          /* client address */
510 	union nf_inet_addr      vaddr;          /* virtual address */
511 	union nf_inet_addr      daddr;          /* destination address */
512 	volatile __u32          flags;          /* status flags */
513 	__u16                   protocol;       /* Which protocol (TCP/UDP) */
514 	__u16			daf;		/* Address family of the dest */
515 	struct netns_ipvs	*ipvs;
516 
517 	/* counter and timer */
518 	refcount_t		refcnt;		/* reference count */
519 	struct timer_list	timer;		/* Expiration timer */
520 	volatile unsigned long	timeout;	/* timeout */
521 
522 	/* Flags and state transition */
523 	spinlock_t              lock;           /* lock for state transition */
524 	volatile __u16          state;          /* state info */
525 	volatile __u16          old_state;      /* old state, to be used for
526 						 * state transition triggerd
527 						 * synchronization
528 						 */
529 	__u32			fwmark;		/* Fire wall mark from skb */
530 	unsigned long		sync_endtime;	/* jiffies + sent_retries */
531 
532 	/* Control members */
533 	struct ip_vs_conn       *control;       /* Master control connection */
534 	atomic_t                n_control;      /* Number of controlled ones */
535 	struct ip_vs_dest       *dest;          /* real server */
536 	atomic_t                in_pkts;        /* incoming packet counter */
537 
538 	/* Packet transmitter for different forwarding methods.  If it
539 	 * mangles the packet, it must return NF_DROP or better NF_STOLEN,
540 	 * otherwise this must be changed to a sk_buff **.
541 	 * NF_ACCEPT can be returned when destination is local.
542 	 */
543 	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
544 			   struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
545 
546 	/* Note: we can group the following members into a structure,
547 	 * in order to save more space, and the following members are
548 	 * only used in VS/NAT anyway
549 	 */
550 	struct ip_vs_app        *app;           /* bound ip_vs_app object */
551 	void                    *app_data;      /* Application private data */
552 	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
553 	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
554 
555 	const struct ip_vs_pe	*pe;
556 	char			*pe_data;
557 	__u8			pe_data_len;
558 
559 	struct rcu_head		rcu_head;
560 };
561 
562 /* Extended internal versions of struct ip_vs_service_user and ip_vs_dest_user
563  * for IPv6 support.
564  *
565  * We need these to conveniently pass around service and destination
566  * options, but unfortunately, we also need to keep the old definitions to
567  * maintain userspace backwards compatibility for the setsockopt interface.
568  */
569 struct ip_vs_service_user_kern {
570 	/* virtual service addresses */
571 	u16			af;
572 	u16			protocol;
573 	union nf_inet_addr	addr;		/* virtual ip address */
574 	__be16			port;
575 	u32			fwmark;		/* firwall mark of service */
576 
577 	/* virtual service options */
578 	char			*sched_name;
579 	char			*pe_name;
580 	unsigned int		flags;		/* virtual service flags */
581 	unsigned int		timeout;	/* persistent timeout in sec */
582 	__be32			netmask;	/* persistent netmask or plen */
583 };
584 
585 
586 struct ip_vs_dest_user_kern {
587 	/* destination server address */
588 	union nf_inet_addr	addr;
589 	__be16			port;
590 
591 	/* real server options */
592 	unsigned int		conn_flags;	/* connection flags */
593 	int			weight;		/* destination weight */
594 
595 	/* thresholds for active connections */
596 	u32			u_threshold;	/* upper threshold */
597 	u32			l_threshold;	/* lower threshold */
598 
599 	/* Address family of addr */
600 	u16			af;
601 
602 	u16			tun_type;	/* tunnel type */
603 	__be16			tun_port;	/* tunnel port */
604 	u16			tun_flags;	/* tunnel flags */
605 };
606 
607 
608 /*
609  * The information about the virtual service offered to the net and the
610  * forwarding entries.
611  */
612 struct ip_vs_service {
613 	struct hlist_node	s_list;   /* for normal service table */
614 	struct hlist_node	f_list;   /* for fwmark-based service table */
615 	atomic_t		refcnt;   /* reference counter */
616 
617 	u16			af;       /* address family */
618 	__u16			protocol; /* which protocol (TCP/UDP) */
619 	union nf_inet_addr	addr;	  /* IP address for virtual service */
620 	__be16			port;	  /* port number for the service */
621 	__u32                   fwmark;   /* firewall mark of the service */
622 	unsigned int		flags;	  /* service status flags */
623 	unsigned int		timeout;  /* persistent timeout in ticks */
624 	__be32			netmask;  /* grouping granularity, mask/plen */
625 	struct netns_ipvs	*ipvs;
626 
627 	struct list_head	destinations;  /* real server d-linked list */
628 	__u32			num_dests;     /* number of servers */
629 	struct ip_vs_stats      stats;         /* statistics for the service */
630 
631 	/* for scheduling */
632 	struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
633 	spinlock_t		sched_lock;    /* lock sched_data */
634 	void			*sched_data;   /* scheduler application data */
635 
636 	/* alternate persistence engine */
637 	struct ip_vs_pe __rcu	*pe;
638 	int			conntrack_afmask;
639 
640 	struct rcu_head		rcu_head;
641 };
642 
643 /* Information for cached dst */
644 struct ip_vs_dest_dst {
645 	struct dst_entry	*dst_cache;	/* destination cache entry */
646 	u32			dst_cookie;
647 	union nf_inet_addr	dst_saddr;
648 	struct rcu_head		rcu_head;
649 };
650 
651 /* The real server destination forwarding entry with ip address, port number,
652  * and so on.
653  */
654 struct ip_vs_dest {
655 	struct list_head	n_list;   /* for the dests in the service */
656 	struct hlist_node	d_list;   /* for table with all the dests */
657 
658 	u16			af;		/* address family */
659 	__be16			port;		/* port number of the server */
660 	union nf_inet_addr	addr;		/* IP address of the server */
661 	volatile unsigned int	flags;		/* dest status flags */
662 	atomic_t		conn_flags;	/* flags to copy to conn */
663 	atomic_t		weight;		/* server weight */
664 	atomic_t		last_weight;	/* server latest weight */
665 	__u16			tun_type;	/* tunnel type */
666 	__be16			tun_port;	/* tunnel port */
667 	__u16			tun_flags;	/* tunnel flags */
668 
669 	refcount_t		refcnt;		/* reference counter */
670 	struct ip_vs_stats      stats;          /* statistics */
671 	unsigned long		idle_start;	/* start time, jiffies */
672 
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 */
679 
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 */
683 
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 */
690 
691 	struct list_head	t_list;		/* in dest_trash */
692 	unsigned int		in_rs_table:1;	/* we are in rs_table */
693 };
694 
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 */
701 
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);
706 	/* dest is linked */
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);
712 
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);
717 };
718 
719 /* The persistence engine object */
720 struct ip_vs_pe {
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 */
725 
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,
731 			   bool inverse);
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,
736 				       struct sk_buff *skb,
737 				       const struct ip_vs_iphdr *iph,
738 				       __be16 dport, __be16 cport);
739 };
740 
741 /* The application module object (a.k.a. app incarnation) */
742 struct ip_vs_app {
743 	struct list_head	a_list;		/* member in app list */
744 	int			type;		/* IP_VS_APP_TYPE_xxx */
745 	char			*name;		/* application module name */
746 	__u16			protocol;
747 	struct module		*module;	/* THIS_MODULE/NULL */
748 	struct list_head	incs_list;	/* list of incarnations */
749 
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;
756 
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
760 	 */
761 	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
762 		       struct sk_buff *, int *diff, struct ip_vs_iphdr *ipvsh);
763 
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
767 	 */
768 	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
769 		      struct sk_buff *, int *diff, struct ip_vs_iphdr *ipvsh);
770 
771 	/* ip_vs_app initializer */
772 	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
773 
774 	/* ip_vs_app finish */
775 	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
776 
777 
778 	/* not used now */
779 	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
780 			 struct ip_vs_protocol *);
781 
782 	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
783 
784 	int *			timeout_table;
785 	int *			timeouts;
786 	int			timeouts_size;
787 
788 	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
789 			     int *verdict, struct ip_vs_conn **cpp);
790 
791 	struct ip_vs_conn *
792 	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
793 		       const struct iphdr *iph, int inverse);
794 
795 	struct ip_vs_conn *
796 	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
797 			const struct iphdr *iph, int inverse);
798 
799 	int (*state_transition)(struct ip_vs_conn *cp, int direction,
800 				const struct sk_buff *skb,
801 				struct ip_vs_app *app);
802 
803 	void (*timeout_change)(struct ip_vs_app *app, int flags);
804 };
805 
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;
813 };
814 
815 struct ip_vs_sync_thread_data;
816 
817 /* How much time to keep dests in trash */
818 #define IP_VS_DEST_TRASH_PERIOD		(120 * HZ)
819 
820 struct ipvs_sync_daemon_cfg {
821 	union nf_inet_addr	mcast_group;
822 	int			syncid;
823 	u16			sync_maxlen;
824 	u16			mcast_port;
825 	u8			mcast_af;
826 	u8			mcast_ttl;
827 	/* multicast interface name */
828 	char			mcast_ifn[IP_VS_IFNAME_MAXLEN];
829 };
830 
831 /* IPVS in network namespace */
832 struct netns_ipvs {
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)
839 
840 	struct hlist_head	rs_table[IP_VS_RTAB_SIZE];
841 	/* ip_vs_app */
842 	struct list_head	app_list;
843 	/* ip_vs_proto */
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];
852 #endif
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];
859 #endif
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];
867 #endif
868 	/* ip_vs_conn */
869 	atomic_t		conn_count;      /* connection counter */
870 
871 	/* ip_vs_ctl */
872 	struct ip_vs_stats		tot_stats;  /* Statistics & est. */
873 
874 	int			num_services;    /* no of virtual services */
875 	int			num_services6;   /* IPv6 virtual services */
876 
877 	/* Trash for destinations */
878 	struct list_head	dest_trash;
879 	spinlock_t		dest_trash_lock;
880 	struct timer_list	dest_trash_timer; /* expiration timer */
881 	/* Service counters */
882 	atomic_t		ftpsvc_counter;
883 	atomic_t		nullsvc_counter;
884 	atomic_t		conn_out_counter;
885 
886 #ifdef CONFIG_SYSCTL
887 	/* delayed work for expiring no dest connections */
888 	struct delayed_work	expire_nodest_conn_work;
889 	/* 1/rate drop and drop-entry variables */
890 	struct delayed_work	defense_work;   /* Work handler */
891 	int			drop_rate;
892 	int			drop_counter;
893 	int			old_secure_tcp;
894 	atomic_t		dropentry;
895 	/* locks in ctl.c */
896 	spinlock_t		dropentry_lock;  /* drop entry handling */
897 	spinlock_t		droppacket_lock; /* drop packet handling */
898 	spinlock_t		securetcp_lock;  /* state and timeout tables */
899 
900 	/* sys-ctl struct */
901 	struct ctl_table_header	*sysctl_hdr;
902 	struct ctl_table	*sysctl_tbl;
903 #endif
904 
905 	/* sysctl variables */
906 	int			sysctl_amemthresh;
907 	int			sysctl_am_droprate;
908 	int			sysctl_drop_entry;
909 	int			sysctl_drop_packet;
910 	int			sysctl_secure_tcp;
911 #ifdef CONFIG_IP_VS_NFCT
912 	int			sysctl_conntrack;
913 #endif
914 	int			sysctl_snat_reroute;
915 	int			sysctl_sync_ver;
916 	int			sysctl_sync_ports;
917 	int			sysctl_sync_persist_mode;
918 	unsigned long		sysctl_sync_qlen_max;
919 	int			sysctl_sync_sock_size;
920 	int			sysctl_cache_bypass;
921 	int			sysctl_expire_nodest_conn;
922 	int			sysctl_sloppy_tcp;
923 	int			sysctl_sloppy_sctp;
924 	int			sysctl_expire_quiescent_template;
925 	int			sysctl_sync_threshold[2];
926 	unsigned int		sysctl_sync_refresh_period;
927 	int			sysctl_sync_retries;
928 	int			sysctl_nat_icmp_send;
929 	int			sysctl_pmtu_disc;
930 	int			sysctl_backup_only;
931 	int			sysctl_conn_reuse_mode;
932 	int			sysctl_schedule_icmp;
933 	int			sysctl_ignore_tunneled;
934 	int			sysctl_run_estimation;
935 
936 	/* ip_vs_lblc */
937 	int			sysctl_lblc_expiration;
938 	struct ctl_table_header	*lblc_ctl_header;
939 	struct ctl_table	*lblc_ctl_table;
940 	/* ip_vs_lblcr */
941 	int			sysctl_lblcr_expiration;
942 	struct ctl_table_header	*lblcr_ctl_header;
943 	struct ctl_table	*lblcr_ctl_table;
944 	/* ip_vs_est */
945 	struct list_head	est_list;	/* estimator list */
946 	spinlock_t		est_lock;
947 	struct timer_list	est_timer;	/* Estimation timer */
948 	/* ip_vs_sync */
949 	spinlock_t		sync_lock;
950 	struct ipvs_master_sync_state *ms;
951 	spinlock_t		sync_buff_lock;
952 	struct ip_vs_sync_thread_data *master_tinfo;
953 	struct ip_vs_sync_thread_data *backup_tinfo;
954 	int			threads_mask;
955 	volatile int		sync_state;
956 	struct mutex		sync_mutex;
957 	struct ipvs_sync_daemon_cfg	mcfg;	/* Master Configuration */
958 	struct ipvs_sync_daemon_cfg	bcfg;	/* Backup Configuration */
959 	/* net name space ptr */
960 	struct net		*net;            /* Needed by timer routines */
961 	/* Number of heterogeneous destinations, needed becaus heterogeneous
962 	 * are not supported when synchronization is enabled.
963 	 */
964 	unsigned int		mixed_address_family_dests;
965 	unsigned int		hooks_afmask;	/* &1=AF_INET, &2=AF_INET6 */
966 };
967 
968 #define DEFAULT_SYNC_THRESHOLD	3
969 #define DEFAULT_SYNC_PERIOD	50
970 #define DEFAULT_SYNC_VER	1
971 #define DEFAULT_SLOPPY_TCP	0
972 #define DEFAULT_SLOPPY_SCTP	0
973 #define DEFAULT_SYNC_REFRESH_PERIOD	(0U * HZ)
974 #define DEFAULT_SYNC_RETRIES		0
975 #define IPVS_SYNC_WAKEUP_RATE	8
976 #define IPVS_SYNC_QLEN_MAX	(IPVS_SYNC_WAKEUP_RATE * 4)
977 #define IPVS_SYNC_SEND_DELAY	(HZ / 50)
978 #define IPVS_SYNC_CHECK_PERIOD	HZ
979 #define IPVS_SYNC_FLUSH_TIME	(HZ * 2)
980 #define IPVS_SYNC_PORTS_MAX	(1 << 6)
981 
982 #ifdef CONFIG_SYSCTL
983 
sysctl_sync_threshold(struct netns_ipvs * ipvs)984 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
985 {
986 	return ipvs->sysctl_sync_threshold[0];
987 }
988 
sysctl_sync_period(struct netns_ipvs * ipvs)989 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
990 {
991 	return READ_ONCE(ipvs->sysctl_sync_threshold[1]);
992 }
993 
sysctl_sync_refresh_period(struct netns_ipvs * ipvs)994 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
995 {
996 	return READ_ONCE(ipvs->sysctl_sync_refresh_period);
997 }
998 
sysctl_sync_retries(struct netns_ipvs * ipvs)999 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1000 {
1001 	return ipvs->sysctl_sync_retries;
1002 }
1003 
sysctl_sync_ver(struct netns_ipvs * ipvs)1004 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1005 {
1006 	return ipvs->sysctl_sync_ver;
1007 }
1008 
sysctl_sloppy_tcp(struct netns_ipvs * ipvs)1009 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1010 {
1011 	return ipvs->sysctl_sloppy_tcp;
1012 }
1013 
sysctl_sloppy_sctp(struct netns_ipvs * ipvs)1014 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1015 {
1016 	return ipvs->sysctl_sloppy_sctp;
1017 }
1018 
sysctl_sync_ports(struct netns_ipvs * ipvs)1019 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1020 {
1021 	return READ_ONCE(ipvs->sysctl_sync_ports);
1022 }
1023 
sysctl_sync_persist_mode(struct netns_ipvs * ipvs)1024 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1025 {
1026 	return ipvs->sysctl_sync_persist_mode;
1027 }
1028 
sysctl_sync_qlen_max(struct netns_ipvs * ipvs)1029 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1030 {
1031 	return ipvs->sysctl_sync_qlen_max;
1032 }
1033 
sysctl_sync_sock_size(struct netns_ipvs * ipvs)1034 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1035 {
1036 	return ipvs->sysctl_sync_sock_size;
1037 }
1038 
sysctl_pmtu_disc(struct netns_ipvs * ipvs)1039 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1040 {
1041 	return ipvs->sysctl_pmtu_disc;
1042 }
1043 
sysctl_backup_only(struct netns_ipvs * ipvs)1044 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1045 {
1046 	return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1047 	       ipvs->sysctl_backup_only;
1048 }
1049 
sysctl_conn_reuse_mode(struct netns_ipvs * ipvs)1050 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1051 {
1052 	return ipvs->sysctl_conn_reuse_mode;
1053 }
1054 
sysctl_expire_nodest_conn(struct netns_ipvs * ipvs)1055 static inline int sysctl_expire_nodest_conn(struct netns_ipvs *ipvs)
1056 {
1057 	return ipvs->sysctl_expire_nodest_conn;
1058 }
1059 
sysctl_schedule_icmp(struct netns_ipvs * ipvs)1060 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1061 {
1062 	return ipvs->sysctl_schedule_icmp;
1063 }
1064 
sysctl_ignore_tunneled(struct netns_ipvs * ipvs)1065 static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1066 {
1067 	return ipvs->sysctl_ignore_tunneled;
1068 }
1069 
sysctl_cache_bypass(struct netns_ipvs * ipvs)1070 static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1071 {
1072 	return ipvs->sysctl_cache_bypass;
1073 }
1074 
sysctl_run_estimation(struct netns_ipvs * ipvs)1075 static inline int sysctl_run_estimation(struct netns_ipvs *ipvs)
1076 {
1077 	return ipvs->sysctl_run_estimation;
1078 }
1079 
1080 #else
1081 
sysctl_sync_threshold(struct netns_ipvs * ipvs)1082 static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1083 {
1084 	return DEFAULT_SYNC_THRESHOLD;
1085 }
1086 
sysctl_sync_period(struct netns_ipvs * ipvs)1087 static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1088 {
1089 	return DEFAULT_SYNC_PERIOD;
1090 }
1091 
sysctl_sync_refresh_period(struct netns_ipvs * ipvs)1092 static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1093 {
1094 	return DEFAULT_SYNC_REFRESH_PERIOD;
1095 }
1096 
sysctl_sync_retries(struct netns_ipvs * ipvs)1097 static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1098 {
1099 	return DEFAULT_SYNC_RETRIES & 3;
1100 }
1101 
sysctl_sync_ver(struct netns_ipvs * ipvs)1102 static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1103 {
1104 	return DEFAULT_SYNC_VER;
1105 }
1106 
sysctl_sloppy_tcp(struct netns_ipvs * ipvs)1107 static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1108 {
1109 	return DEFAULT_SLOPPY_TCP;
1110 }
1111 
sysctl_sloppy_sctp(struct netns_ipvs * ipvs)1112 static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1113 {
1114 	return DEFAULT_SLOPPY_SCTP;
1115 }
1116 
sysctl_sync_ports(struct netns_ipvs * ipvs)1117 static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1118 {
1119 	return 1;
1120 }
1121 
sysctl_sync_persist_mode(struct netns_ipvs * ipvs)1122 static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1123 {
1124 	return 0;
1125 }
1126 
sysctl_sync_qlen_max(struct netns_ipvs * ipvs)1127 static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1128 {
1129 	return IPVS_SYNC_QLEN_MAX;
1130 }
1131 
sysctl_sync_sock_size(struct netns_ipvs * ipvs)1132 static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1133 {
1134 	return 0;
1135 }
1136 
sysctl_pmtu_disc(struct netns_ipvs * ipvs)1137 static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1138 {
1139 	return 1;
1140 }
1141 
sysctl_backup_only(struct netns_ipvs * ipvs)1142 static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1143 {
1144 	return 0;
1145 }
1146 
sysctl_conn_reuse_mode(struct netns_ipvs * ipvs)1147 static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1148 {
1149 	return 1;
1150 }
1151 
sysctl_expire_nodest_conn(struct netns_ipvs * ipvs)1152 static inline int sysctl_expire_nodest_conn(struct netns_ipvs *ipvs)
1153 {
1154 	return 0;
1155 }
1156 
sysctl_schedule_icmp(struct netns_ipvs * ipvs)1157 static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1158 {
1159 	return 0;
1160 }
1161 
sysctl_ignore_tunneled(struct netns_ipvs * ipvs)1162 static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1163 {
1164 	return 0;
1165 }
1166 
sysctl_cache_bypass(struct netns_ipvs * ipvs)1167 static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1168 {
1169 	return 0;
1170 }
1171 
sysctl_run_estimation(struct netns_ipvs * ipvs)1172 static inline int sysctl_run_estimation(struct netns_ipvs *ipvs)
1173 {
1174 	return 1;
1175 }
1176 
1177 #endif
1178 
1179 /* IPVS core functions
1180  * (from ip_vs_core.c)
1181  */
1182 const char *ip_vs_proto_name(unsigned int proto);
1183 void ip_vs_init_hash_table(struct list_head *table, int rows);
1184 struct ip_vs_conn *ip_vs_new_conn_out(struct ip_vs_service *svc,
1185 				      struct ip_vs_dest *dest,
1186 				      struct sk_buff *skb,
1187 				      const struct ip_vs_iphdr *iph,
1188 				      __be16 dport,
1189 				      __be16 cport);
1190 #define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1191 
1192 #define IP_VS_APP_TYPE_FTP	1
1193 
1194 /* ip_vs_conn handling functions
1195  * (from ip_vs_conn.c)
1196  */
1197 enum {
1198 	IP_VS_DIR_INPUT = 0,
1199 	IP_VS_DIR_OUTPUT,
1200 	IP_VS_DIR_INPUT_ONLY,
1201 	IP_VS_DIR_LAST,
1202 };
1203 
ip_vs_conn_fill_param(struct netns_ipvs * ipvs,int af,int protocol,const union nf_inet_addr * caddr,__be16 cport,const union nf_inet_addr * vaddr,__be16 vport,struct ip_vs_conn_param * p)1204 static inline void ip_vs_conn_fill_param(struct netns_ipvs *ipvs, int af, int protocol,
1205 					 const union nf_inet_addr *caddr,
1206 					 __be16 cport,
1207 					 const union nf_inet_addr *vaddr,
1208 					 __be16 vport,
1209 					 struct ip_vs_conn_param *p)
1210 {
1211 	p->ipvs = ipvs;
1212 	p->af = af;
1213 	p->protocol = protocol;
1214 	p->caddr = caddr;
1215 	p->cport = cport;
1216 	p->vaddr = vaddr;
1217 	p->vport = vport;
1218 	p->pe = NULL;
1219 	p->pe_data = NULL;
1220 }
1221 
1222 struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1223 struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1224 
1225 struct ip_vs_conn * ip_vs_conn_in_get_proto(struct netns_ipvs *ipvs, int af,
1226 					    const struct sk_buff *skb,
1227 					    const struct ip_vs_iphdr *iph);
1228 
1229 struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1230 
1231 struct ip_vs_conn * ip_vs_conn_out_get_proto(struct netns_ipvs *ipvs, int af,
1232 					     const struct sk_buff *skb,
1233 					     const struct ip_vs_iphdr *iph);
1234 
1235 /* Get reference to gain full access to conn.
1236  * By default, RCU read-side critical sections have access only to
1237  * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1238  */
__ip_vs_conn_get(struct ip_vs_conn * cp)1239 static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1240 {
1241 	return refcount_inc_not_zero(&cp->refcnt);
1242 }
1243 
1244 /* put back the conn without restarting its timer */
__ip_vs_conn_put(struct ip_vs_conn * cp)1245 static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1246 {
1247 	smp_mb__before_atomic();
1248 	refcount_dec(&cp->refcnt);
1249 }
1250 void ip_vs_conn_put(struct ip_vs_conn *cp);
1251 void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1252 
1253 struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, int dest_af,
1254 				  const union nf_inet_addr *daddr,
1255 				  __be16 dport, unsigned int flags,
1256 				  struct ip_vs_dest *dest, __u32 fwmark);
1257 void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1258 
1259 const char *ip_vs_state_name(const struct ip_vs_conn *cp);
1260 
1261 void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
1262 int ip_vs_check_template(struct ip_vs_conn *ct, struct ip_vs_dest *cdest);
1263 void ip_vs_random_dropentry(struct netns_ipvs *ipvs);
1264 int ip_vs_conn_init(void);
1265 void ip_vs_conn_cleanup(void);
1266 
ip_vs_control_del(struct ip_vs_conn * cp)1267 static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1268 {
1269 	struct ip_vs_conn *ctl_cp = cp->control;
1270 	if (!ctl_cp) {
1271 		IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1272 			      "%s:%d to %s:%d\n",
1273 			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1274 			      ntohs(cp->cport),
1275 			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1276 			      ntohs(cp->vport));
1277 
1278 		return;
1279 	}
1280 
1281 	IP_VS_DBG_BUF(7, "DELeting control for: "
1282 		      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1283 		      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1284 		      ntohs(cp->cport),
1285 		      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1286 		      ntohs(ctl_cp->cport));
1287 
1288 	cp->control = NULL;
1289 	if (atomic_read(&ctl_cp->n_control) == 0) {
1290 		IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1291 			      "%s:%d to %s:%d\n",
1292 			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1293 			      ntohs(cp->cport),
1294 			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1295 			      ntohs(cp->vport));
1296 
1297 		return;
1298 	}
1299 	atomic_dec(&ctl_cp->n_control);
1300 }
1301 
1302 static inline void
ip_vs_control_add(struct ip_vs_conn * cp,struct ip_vs_conn * ctl_cp)1303 ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1304 {
1305 	if (cp->control) {
1306 		IP_VS_ERR_BUF("request control ADD for already controlled: "
1307 			      "%s:%d to %s:%d\n",
1308 			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1309 			      ntohs(cp->cport),
1310 			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1311 			      ntohs(cp->vport));
1312 
1313 		ip_vs_control_del(cp);
1314 	}
1315 
1316 	IP_VS_DBG_BUF(7, "ADDing control for: "
1317 		      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1318 		      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1319 		      ntohs(cp->cport),
1320 		      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1321 		      ntohs(ctl_cp->cport));
1322 
1323 	cp->control = ctl_cp;
1324 	atomic_inc(&ctl_cp->n_control);
1325 }
1326 
1327 /* Mark our template as assured */
1328 static inline void
ip_vs_control_assure_ct(struct ip_vs_conn * cp)1329 ip_vs_control_assure_ct(struct ip_vs_conn *cp)
1330 {
1331 	struct ip_vs_conn *ct = cp->control;
1332 
1333 	if (ct && !(ct->state & IP_VS_CTPL_S_ASSURED) &&
1334 	    (ct->flags & IP_VS_CONN_F_TEMPLATE))
1335 		ct->state |= IP_VS_CTPL_S_ASSURED;
1336 }
1337 
1338 /* IPVS netns init & cleanup functions */
1339 int ip_vs_estimator_net_init(struct netns_ipvs *ipvs);
1340 int ip_vs_control_net_init(struct netns_ipvs *ipvs);
1341 int ip_vs_protocol_net_init(struct netns_ipvs *ipvs);
1342 int ip_vs_app_net_init(struct netns_ipvs *ipvs);
1343 int ip_vs_conn_net_init(struct netns_ipvs *ipvs);
1344 int ip_vs_sync_net_init(struct netns_ipvs *ipvs);
1345 void ip_vs_conn_net_cleanup(struct netns_ipvs *ipvs);
1346 void ip_vs_app_net_cleanup(struct netns_ipvs *ipvs);
1347 void ip_vs_protocol_net_cleanup(struct netns_ipvs *ipvs);
1348 void ip_vs_control_net_cleanup(struct netns_ipvs *ipvs);
1349 void ip_vs_estimator_net_cleanup(struct netns_ipvs *ipvs);
1350 void ip_vs_sync_net_cleanup(struct netns_ipvs *ipvs);
1351 void ip_vs_service_nets_cleanup(struct list_head *net_list);
1352 
1353 /* IPVS application functions
1354  * (from ip_vs_app.c)
1355  */
1356 #define IP_VS_APP_MAX_PORTS  8
1357 struct ip_vs_app *register_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1358 void unregister_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1359 int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1360 void ip_vs_unbind_app(struct ip_vs_conn *cp);
1361 int register_ip_vs_app_inc(struct netns_ipvs *ipvs, struct ip_vs_app *app, __u16 proto,
1362 			   __u16 port);
1363 int ip_vs_app_inc_get(struct ip_vs_app *inc);
1364 void ip_vs_app_inc_put(struct ip_vs_app *inc);
1365 
1366 int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb,
1367 		      struct ip_vs_iphdr *ipvsh);
1368 int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb,
1369 		     struct ip_vs_iphdr *ipvsh);
1370 
1371 int register_ip_vs_pe(struct ip_vs_pe *pe);
1372 int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1373 struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1374 struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1375 
1376 /* Use a #define to avoid all of module.h just for these trivial ops */
1377 #define ip_vs_pe_get(pe)			\
1378 	if (pe && pe->module)			\
1379 		__module_get(pe->module);
1380 
1381 #define ip_vs_pe_put(pe)			\
1382 	if (pe && pe->module)			\
1383 		module_put(pe->module);
1384 
1385 /* IPVS protocol functions (from ip_vs_proto.c) */
1386 int ip_vs_protocol_init(void);
1387 void ip_vs_protocol_cleanup(void);
1388 void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1389 int *ip_vs_create_timeout_table(int *table, int size);
1390 void ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1391 			       const struct sk_buff *skb, int offset,
1392 			       const char *msg);
1393 
1394 extern struct ip_vs_protocol ip_vs_protocol_tcp;
1395 extern struct ip_vs_protocol ip_vs_protocol_udp;
1396 extern struct ip_vs_protocol ip_vs_protocol_icmp;
1397 extern struct ip_vs_protocol ip_vs_protocol_esp;
1398 extern struct ip_vs_protocol ip_vs_protocol_ah;
1399 extern struct ip_vs_protocol ip_vs_protocol_sctp;
1400 
1401 /* Registering/unregistering scheduler functions
1402  * (from ip_vs_sched.c)
1403  */
1404 int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1405 int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1406 int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1407 			 struct ip_vs_scheduler *scheduler);
1408 void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1409 			    struct ip_vs_scheduler *sched);
1410 struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1411 void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1412 struct ip_vs_conn *
1413 ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1414 	       struct ip_vs_proto_data *pd, int *ignored,
1415 	       struct ip_vs_iphdr *iph);
1416 int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1417 		struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1418 
1419 void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1420 
1421 /* IPVS control data and functions (from ip_vs_ctl.c) */
1422 extern struct ip_vs_stats ip_vs_stats;
1423 extern int sysctl_ip_vs_sync_ver;
1424 
1425 struct ip_vs_service *
1426 ip_vs_service_find(struct netns_ipvs *ipvs, int af, __u32 fwmark, __u16 protocol,
1427 		  const union nf_inet_addr *vaddr, __be16 vport);
1428 
1429 bool ip_vs_has_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1430 			    const union nf_inet_addr *daddr, __be16 dport);
1431 
1432 struct ip_vs_dest *
1433 ip_vs_find_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1434 			const union nf_inet_addr *daddr, __be16 dport);
1435 struct ip_vs_dest *ip_vs_find_tunnel(struct netns_ipvs *ipvs, int af,
1436 				     const union nf_inet_addr *daddr,
1437 				     __be16 tun_port);
1438 
1439 int ip_vs_use_count_inc(void);
1440 void ip_vs_use_count_dec(void);
1441 int ip_vs_register_nl_ioctl(void);
1442 void ip_vs_unregister_nl_ioctl(void);
1443 int ip_vs_control_init(void);
1444 void ip_vs_control_cleanup(void);
1445 struct ip_vs_dest *
1446 ip_vs_find_dest(struct netns_ipvs *ipvs, int svc_af, int dest_af,
1447 		const union nf_inet_addr *daddr, __be16 dport,
1448 		const union nf_inet_addr *vaddr, __be16 vport,
1449 		__u16 protocol, __u32 fwmark, __u32 flags);
1450 void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1451 
ip_vs_dest_hold(struct ip_vs_dest * dest)1452 static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1453 {
1454 	refcount_inc(&dest->refcnt);
1455 }
1456 
ip_vs_dest_put(struct ip_vs_dest * dest)1457 static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1458 {
1459 	smp_mb__before_atomic();
1460 	refcount_dec(&dest->refcnt);
1461 }
1462 
ip_vs_dest_put_and_free(struct ip_vs_dest * dest)1463 static inline void ip_vs_dest_put_and_free(struct ip_vs_dest *dest)
1464 {
1465 	if (refcount_dec_and_test(&dest->refcnt))
1466 		kfree(dest);
1467 }
1468 
1469 /* IPVS sync daemon data and function prototypes
1470  * (from ip_vs_sync.c)
1471  */
1472 int start_sync_thread(struct netns_ipvs *ipvs, struct ipvs_sync_daemon_cfg *cfg,
1473 		      int state);
1474 int stop_sync_thread(struct netns_ipvs *ipvs, int state);
1475 void ip_vs_sync_conn(struct netns_ipvs *ipvs, struct ip_vs_conn *cp, int pkts);
1476 
1477 /* IPVS rate estimator prototypes (from ip_vs_est.c) */
1478 void ip_vs_start_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1479 void ip_vs_stop_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1480 void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1481 void ip_vs_read_estimator(struct ip_vs_kstats *dst, struct ip_vs_stats *stats);
1482 
1483 /* Various IPVS packet transmitters (from ip_vs_xmit.c) */
1484 int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1485 		    struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1486 int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1487 		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1488 int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1489 		   struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1490 int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1491 		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1492 int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1493 		  struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1494 int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1495 		    struct ip_vs_protocol *pp, int offset,
1496 		    unsigned int hooknum, struct ip_vs_iphdr *iph);
1497 void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1498 
1499 #ifdef CONFIG_IP_VS_IPV6
1500 int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1501 			 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1502 int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1503 		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1504 int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1505 			 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1506 int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1507 		     struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1508 int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1509 		       struct ip_vs_protocol *pp, int offset,
1510 		       unsigned int hooknum, struct ip_vs_iphdr *iph);
1511 #endif
1512 
1513 #ifdef CONFIG_SYSCTL
1514 /* This is a simple mechanism to ignore packets when
1515  * we are loaded. Just set ip_vs_drop_rate to 'n' and
1516  * we start to drop 1/rate of the packets
1517  */
ip_vs_todrop(struct netns_ipvs * ipvs)1518 static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1519 {
1520 	if (!ipvs->drop_rate)
1521 		return 0;
1522 	if (--ipvs->drop_counter > 0)
1523 		return 0;
1524 	ipvs->drop_counter = ipvs->drop_rate;
1525 	return 1;
1526 }
1527 #else
ip_vs_todrop(struct netns_ipvs * ipvs)1528 static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1529 #endif
1530 
1531 #ifdef CONFIG_SYSCTL
1532 /* Enqueue delayed work for expiring no dest connections
1533  * Only run when sysctl_expire_nodest=1
1534  */
ip_vs_enqueue_expire_nodest_conns(struct netns_ipvs * ipvs)1535 static inline void ip_vs_enqueue_expire_nodest_conns(struct netns_ipvs *ipvs)
1536 {
1537 	if (sysctl_expire_nodest_conn(ipvs))
1538 		queue_delayed_work(system_long_wq,
1539 				   &ipvs->expire_nodest_conn_work, 1);
1540 }
1541 
1542 void ip_vs_expire_nodest_conn_flush(struct netns_ipvs *ipvs);
1543 #else
ip_vs_enqueue_expire_nodest_conns(struct netns_ipvs * ipvs)1544 static inline void ip_vs_enqueue_expire_nodest_conns(struct netns_ipvs *ipvs) {}
1545 #endif
1546 
1547 #define IP_VS_DFWD_METHOD(dest) (atomic_read(&(dest)->conn_flags) & \
1548 				 IP_VS_CONN_F_FWD_MASK)
1549 
1550 /* ip_vs_fwd_tag returns the forwarding tag of the connection */
1551 #define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
1552 
ip_vs_fwd_tag(struct ip_vs_conn * cp)1553 static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1554 {
1555 	char fwd;
1556 
1557 	switch (IP_VS_FWD_METHOD(cp)) {
1558 	case IP_VS_CONN_F_MASQ:
1559 		fwd = 'M'; break;
1560 	case IP_VS_CONN_F_LOCALNODE:
1561 		fwd = 'L'; break;
1562 	case IP_VS_CONN_F_TUNNEL:
1563 		fwd = 'T'; break;
1564 	case IP_VS_CONN_F_DROUTE:
1565 		fwd = 'R'; break;
1566 	case IP_VS_CONN_F_BYPASS:
1567 		fwd = 'B'; break;
1568 	default:
1569 		fwd = '?'; break;
1570 	}
1571 	return fwd;
1572 }
1573 
1574 void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1575 		    struct ip_vs_conn *cp, int dir);
1576 
1577 #ifdef CONFIG_IP_VS_IPV6
1578 void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1579 		       struct ip_vs_conn *cp, int dir);
1580 #endif
1581 
1582 __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1583 
ip_vs_check_diff4(__be32 old,__be32 new,__wsum oldsum)1584 static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1585 {
1586 	__be32 diff[2] = { ~old, new };
1587 
1588 	return csum_partial(diff, sizeof(diff), oldsum);
1589 }
1590 
1591 #ifdef CONFIG_IP_VS_IPV6
ip_vs_check_diff16(const __be32 * old,const __be32 * new,__wsum oldsum)1592 static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1593 					__wsum oldsum)
1594 {
1595 	__be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1596 			    new[3],  new[2],  new[1],  new[0] };
1597 
1598 	return csum_partial(diff, sizeof(diff), oldsum);
1599 }
1600 #endif
1601 
ip_vs_check_diff2(__be16 old,__be16 new,__wsum oldsum)1602 static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1603 {
1604 	__be16 diff[2] = { ~old, new };
1605 
1606 	return csum_partial(diff, sizeof(diff), oldsum);
1607 }
1608 
1609 /* Forget current conntrack (unconfirmed) and attach notrack entry */
ip_vs_notrack(struct sk_buff * skb)1610 static inline void ip_vs_notrack(struct sk_buff *skb)
1611 {
1612 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1613 	enum ip_conntrack_info ctinfo;
1614 	struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1615 
1616 	if (ct) {
1617 		nf_conntrack_put(&ct->ct_general);
1618 		nf_ct_set(skb, NULL, IP_CT_UNTRACKED);
1619 	}
1620 #endif
1621 }
1622 
1623 #ifdef CONFIG_IP_VS_NFCT
1624 /* Netfilter connection tracking
1625  * (from ip_vs_nfct.c)
1626  */
ip_vs_conntrack_enabled(struct netns_ipvs * ipvs)1627 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1628 {
1629 #ifdef CONFIG_SYSCTL
1630 	return ipvs->sysctl_conntrack;
1631 #else
1632 	return 0;
1633 #endif
1634 }
1635 
1636 void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1637 			    int outin);
1638 int ip_vs_confirm_conntrack(struct sk_buff *skb);
1639 void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1640 			       struct ip_vs_conn *cp, u_int8_t proto,
1641 			       const __be16 port, int from_rs);
1642 void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1643 
1644 #else
1645 
ip_vs_conntrack_enabled(struct netns_ipvs * ipvs)1646 static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1647 {
1648 	return 0;
1649 }
1650 
ip_vs_update_conntrack(struct sk_buff * skb,struct ip_vs_conn * cp,int outin)1651 static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1652 					  struct ip_vs_conn *cp, int outin)
1653 {
1654 }
1655 
ip_vs_confirm_conntrack(struct sk_buff * skb)1656 static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1657 {
1658 	return NF_ACCEPT;
1659 }
1660 
ip_vs_conn_drop_conntrack(struct ip_vs_conn * cp)1661 static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1662 {
1663 }
1664 #endif /* CONFIG_IP_VS_NFCT */
1665 
1666 /* Using old conntrack that can not be redirected to another real server? */
ip_vs_conn_uses_old_conntrack(struct ip_vs_conn * cp,struct sk_buff * skb)1667 static inline bool ip_vs_conn_uses_old_conntrack(struct ip_vs_conn *cp,
1668 						 struct sk_buff *skb)
1669 {
1670 #ifdef CONFIG_IP_VS_NFCT
1671 	enum ip_conntrack_info ctinfo;
1672 	struct nf_conn *ct;
1673 
1674 	ct = nf_ct_get(skb, &ctinfo);
1675 	if (ct && nf_ct_is_confirmed(ct))
1676 		return true;
1677 #endif
1678 	return false;
1679 }
1680 
ip_vs_register_conntrack(struct ip_vs_service * svc)1681 static inline int ip_vs_register_conntrack(struct ip_vs_service *svc)
1682 {
1683 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
1684 	int afmask = (svc->af == AF_INET6) ? 2 : 1;
1685 	int ret = 0;
1686 
1687 	if (!(svc->conntrack_afmask & afmask)) {
1688 		ret = nf_ct_netns_get(svc->ipvs->net, svc->af);
1689 		if (ret >= 0)
1690 			svc->conntrack_afmask |= afmask;
1691 	}
1692 	return ret;
1693 #else
1694 	return 0;
1695 #endif
1696 }
1697 
ip_vs_unregister_conntrack(struct ip_vs_service * svc)1698 static inline void ip_vs_unregister_conntrack(struct ip_vs_service *svc)
1699 {
1700 #if IS_ENABLED(CONFIG_NF_CONNTRACK)
1701 	int afmask = (svc->af == AF_INET6) ? 2 : 1;
1702 
1703 	if (svc->conntrack_afmask & afmask) {
1704 		nf_ct_netns_put(svc->ipvs->net, svc->af);
1705 		svc->conntrack_afmask &= ~afmask;
1706 	}
1707 #endif
1708 }
1709 
1710 int ip_vs_register_hooks(struct netns_ipvs *ipvs, unsigned int af);
1711 void ip_vs_unregister_hooks(struct netns_ipvs *ipvs, unsigned int af);
1712 
1713 static inline int
ip_vs_dest_conn_overhead(struct ip_vs_dest * dest)1714 ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1715 {
1716 	/* We think the overhead of processing active connections is 256
1717 	 * times higher than that of inactive connections in average. (This
1718 	 * 256 times might not be accurate, we will change it later) We
1719 	 * use the following formula to estimate the overhead now:
1720 	 *		  dest->activeconns*256 + dest->inactconns
1721 	 */
1722 	return (atomic_read(&dest->activeconns) << 8) +
1723 		atomic_read(&dest->inactconns);
1724 }
1725 
1726 #ifdef CONFIG_IP_VS_PROTO_TCP
1727 INDIRECT_CALLABLE_DECLARE(int
1728 	tcp_snat_handler(struct sk_buff *skb, struct ip_vs_protocol *pp,
1729 			 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph));
1730 #endif
1731 
1732 #ifdef CONFIG_IP_VS_PROTO_UDP
1733 INDIRECT_CALLABLE_DECLARE(int
1734 	udp_snat_handler(struct sk_buff *skb, struct ip_vs_protocol *pp,
1735 			 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph));
1736 #endif
1737 #endif	/* _NET_IP_VS_H */
1738