Linux Audio

Check our new training course

Loading...
v4.17
   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
  18#include <linux/compiler.h>
  19#include <linux/timer.h>
  20#include <linux/bug.h>
  21
  22#include <net/checksum.h>
  23#include <linux/netfilter.h>		/* for union nf_inet_addr */
  24#include <linux/ip.h>
  25#include <linux/ipv6.h>			/* for struct ipv6hdr */
  26#include <net/ipv6.h>
  27#if IS_ENABLED(CONFIG_IP_VS_IPV6)
  28#include <linux/netfilter_ipv6/ip6_tables.h>
  29#endif
  30#if IS_ENABLED(CONFIG_NF_CONNTRACK)
  31#include <net/netfilter/nf_conntrack.h>
  32#endif
  33#include <net/net_namespace.h>		/* Netw namespace */
  34
  35#define IP_VS_HDR_INVERSE	1
  36#define IP_VS_HDR_ICMP		2
  37
  38/* Generic access of ipvs struct */
  39static inline struct netns_ipvs *net_ipvs(struct net* net)
  40{
  41	return net->ipvs;
  42}
  43
  44/* This one needed for single_open_net since net is stored directly in
  45 * private not as a struct i.e. seq_file_net can't be used.
  46 */
  47static inline struct net *seq_file_single_net(struct seq_file *seq)
  48{
  49#ifdef CONFIG_NET_NS
  50	return (struct net *)seq->private;
  51#else
  52	return &init_net;
  53#endif
  54}
  55
  56/* Connections' size value needed by ip_vs_ctl.c */
  57extern int ip_vs_conn_tab_size;
  58
  59struct ip_vs_iphdr {
  60	int hdr_flags;	/* ipvs flags */
  61	__u32 off;	/* Where IP or IPv4 header starts */
  62	__u32 len;	/* IPv4 simply where L4 starts
  63			 * IPv6 where L4 Transport Header starts */
  64	__u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
  65	__s16 protocol;
  66	__s32 flags;
  67	union nf_inet_addr saddr;
  68	union nf_inet_addr daddr;
  69};
  70
  71static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
  72				      int len, void *buffer)
  73{
  74	return skb_header_pointer(skb, offset, len, buffer);
  75}
  76
  77/* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
  78 * IPv6 requires some extra work, as finding proper header position,
  79 * depend on the IPv6 extension headers.
  80 */
  81static inline int
  82ip_vs_fill_iph_skb_off(int af, const struct sk_buff *skb, int offset,
  83		       int hdr_flags, struct ip_vs_iphdr *iphdr)
  84{
  85	iphdr->hdr_flags = hdr_flags;
  86	iphdr->off = offset;
  87
  88#ifdef CONFIG_IP_VS_IPV6
  89	if (af == AF_INET6) {
  90		struct ipv6hdr _iph;
  91		const struct ipv6hdr *iph = skb_header_pointer(
  92			skb, offset, sizeof(_iph), &_iph);
  93		if (!iph)
  94			return 0;
  95
  96		iphdr->saddr.in6 = iph->saddr;
  97		iphdr->daddr.in6 = iph->daddr;
  98		/* ipv6_find_hdr() updates len, flags */
  99		iphdr->len	 = offset;
 100		iphdr->flags	 = 0;
 101		iphdr->protocol  = ipv6_find_hdr(skb, &iphdr->len, -1,
 102						 &iphdr->fragoffs,
 103						 &iphdr->flags);
 104		if (iphdr->protocol < 0)
 105			return 0;
 106	} else
 107#endif
 108	{
 109		struct iphdr _iph;
 110		const struct iphdr *iph = skb_header_pointer(
 111			skb, offset, sizeof(_iph), &_iph);
 112		if (!iph)
 113			return 0;
 114
 115		iphdr->len	= offset + iph->ihl * 4;
 116		iphdr->fragoffs	= 0;
 117		iphdr->protocol	= iph->protocol;
 118		iphdr->saddr.ip	= iph->saddr;
 119		iphdr->daddr.ip	= iph->daddr;
 120	}
 121
 122	return 1;
 123}
 124
 125static inline int
 126ip_vs_fill_iph_skb_icmp(int af, const struct sk_buff *skb, int offset,
 127			bool inverse, struct ip_vs_iphdr *iphdr)
 128{
 129	int hdr_flags = IP_VS_HDR_ICMP;
 130
 131	if (inverse)
 132		hdr_flags |= IP_VS_HDR_INVERSE;
 133
 134	return ip_vs_fill_iph_skb_off(af, skb, offset, hdr_flags, iphdr);
 135}
 136
 137static inline int
 138ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, bool inverse,
 139		   struct ip_vs_iphdr *iphdr)
 140{
 141	int hdr_flags = 0;
 142
 143	if (inverse)
 144		hdr_flags |= IP_VS_HDR_INVERSE;
 145
 146	return ip_vs_fill_iph_skb_off(af, skb, skb_network_offset(skb),
 147				      hdr_flags, iphdr);
 148}
 149
 150static inline bool
 151ip_vs_iph_inverse(const struct ip_vs_iphdr *iph)
 152{
 153	return !!(iph->hdr_flags & IP_VS_HDR_INVERSE);
 154}
 155
 156static inline bool
 157ip_vs_iph_icmp(const struct ip_vs_iphdr *iph)
 158{
 159	return !!(iph->hdr_flags & IP_VS_HDR_ICMP);
 160}
 161
 162static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
 163				   const union nf_inet_addr *src)
 164{
 165#ifdef CONFIG_IP_VS_IPV6
 166	if (af == AF_INET6)
 167		dst->in6 = src->in6;
 168	else
 169#endif
 170	dst->ip = src->ip;
 171}
 172
 173static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
 174				  const union nf_inet_addr *src)
 175{
 176#ifdef CONFIG_IP_VS_IPV6
 177	if (af == AF_INET6) {
 178		dst->in6 = src->in6;
 179		return;
 180	}
 181#endif
 182	dst->ip = src->ip;
 183	dst->all[1] = 0;
 184	dst->all[2] = 0;
 185	dst->all[3] = 0;
 186}
 187
 188static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
 189				   const union nf_inet_addr *b)
 190{
 191#ifdef CONFIG_IP_VS_IPV6
 192	if (af == AF_INET6)
 193		return ipv6_addr_equal(&a->in6, &b->in6);
 194#endif
 195	return a->ip == b->ip;
 196}
 197
 198#ifdef CONFIG_IP_VS_DEBUG
 199#include <linux/net.h>
 200
 201int ip_vs_get_debug_level(void);
 202
 203static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
 204					 const union nf_inet_addr *addr,
 205					 int *idx)
 206{
 207	int len;
 208#ifdef CONFIG_IP_VS_IPV6
 209	if (af == AF_INET6)
 210		len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
 211			       &addr->in6) + 1;
 212	else
 213#endif
 214		len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
 215			       &addr->ip) + 1;
 216
 217	*idx += len;
 218	BUG_ON(*idx > buf_len + 1);
 219	return &buf[*idx - len];
 220}
 221
 222#define IP_VS_DBG_BUF(level, msg, ...)					\
 223	do {								\
 224		char ip_vs_dbg_buf[160];				\
 225		int ip_vs_dbg_idx = 0;					\
 226		if (level <= ip_vs_get_debug_level())			\
 227			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
 228	} while (0)
 229#define IP_VS_ERR_BUF(msg...)						\
 230	do {								\
 231		char ip_vs_dbg_buf[160];				\
 232		int ip_vs_dbg_idx = 0;					\
 233		pr_err(msg);						\
 234	} while (0)
 235
 236/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
 237#define IP_VS_DBG_ADDR(af, addr)					\
 238	ip_vs_dbg_addr(af, ip_vs_dbg_buf,				\
 239		       sizeof(ip_vs_dbg_buf), addr,			\
 240		       &ip_vs_dbg_idx)
 241
 242#define IP_VS_DBG(level, msg, ...)					\
 243	do {								\
 244		if (level <= ip_vs_get_debug_level())			\
 245			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
 246	} while (0)
 247#define IP_VS_DBG_RL(msg, ...)						\
 248	do {								\
 249		if (net_ratelimit())					\
 250			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
 251	} while (0)
 252#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)			\
 253	do {								\
 254		if (level <= ip_vs_get_debug_level())			\
 255			pp->debug_packet(af, pp, skb, ofs, msg);	\
 256	} while (0)
 257#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)			\
 258	do {								\
 259		if (level <= ip_vs_get_debug_level() &&			\
 260		    net_ratelimit())					\
 261			pp->debug_packet(af, pp, skb, ofs, msg);	\
 262	} while (0)
 263#else	/* NO DEBUGGING at ALL */
 264#define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
 265#define IP_VS_ERR_BUF(msg...)  do {} while (0)
 266#define IP_VS_DBG(level, msg...)  do {} while (0)
 267#define IP_VS_DBG_RL(msg...)  do {} while (0)
 268#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)	do {} while (0)
 269#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)	do {} while (0)
 270#endif
 271
 272#define IP_VS_BUG() BUG()
 273#define IP_VS_ERR_RL(msg, ...)						\
 274	do {								\
 275		if (net_ratelimit())					\
 276			pr_err(msg, ##__VA_ARGS__);			\
 277	} while (0)
 278
 279#ifdef CONFIG_IP_VS_DEBUG
 280#define EnterFunction(level)						\
 281	do {								\
 282		if (level <= ip_vs_get_debug_level())			\
 283			printk(KERN_DEBUG				\
 284			       pr_fmt("Enter: %s, %s line %i\n"),	\
 285			       __func__, __FILE__, __LINE__);		\
 286	} while (0)
 287#define LeaveFunction(level)						\
 288	do {								\
 289		if (level <= ip_vs_get_debug_level())			\
 290			printk(KERN_DEBUG				\
 291			       pr_fmt("Leave: %s, %s line %i\n"),	\
 292			       __func__, __FILE__, __LINE__);		\
 293	} while (0)
 294#else
 295#define EnterFunction(level)   do {} while (0)
 296#define LeaveFunction(level)   do {} while (0)
 297#endif
 298
 299/* The port number of FTP service (in network order). */
 300#define FTPPORT  cpu_to_be16(21)
 301#define FTPDATA  cpu_to_be16(20)
 302
 303/* TCP State Values */
 304enum {
 305	IP_VS_TCP_S_NONE = 0,
 306	IP_VS_TCP_S_ESTABLISHED,
 307	IP_VS_TCP_S_SYN_SENT,
 308	IP_VS_TCP_S_SYN_RECV,
 309	IP_VS_TCP_S_FIN_WAIT,
 310	IP_VS_TCP_S_TIME_WAIT,
 311	IP_VS_TCP_S_CLOSE,
 312	IP_VS_TCP_S_CLOSE_WAIT,
 313	IP_VS_TCP_S_LAST_ACK,
 314	IP_VS_TCP_S_LISTEN,
 315	IP_VS_TCP_S_SYNACK,
 316	IP_VS_TCP_S_LAST
 317};
 318
 319/* UDP State Values */
 320enum {
 321	IP_VS_UDP_S_NORMAL,
 322	IP_VS_UDP_S_LAST,
 323};
 324
 325/* ICMP State Values */
 326enum {
 327	IP_VS_ICMP_S_NORMAL,
 328	IP_VS_ICMP_S_LAST,
 329};
 330
 331/* SCTP State Values */
 332enum ip_vs_sctp_states {
 333	IP_VS_SCTP_S_NONE,
 334	IP_VS_SCTP_S_INIT1,
 335	IP_VS_SCTP_S_INIT,
 336	IP_VS_SCTP_S_COOKIE_SENT,
 337	IP_VS_SCTP_S_COOKIE_REPLIED,
 338	IP_VS_SCTP_S_COOKIE_WAIT,
 339	IP_VS_SCTP_S_COOKIE,
 340	IP_VS_SCTP_S_COOKIE_ECHOED,
 341	IP_VS_SCTP_S_ESTABLISHED,
 342	IP_VS_SCTP_S_SHUTDOWN_SENT,
 343	IP_VS_SCTP_S_SHUTDOWN_RECEIVED,
 344	IP_VS_SCTP_S_SHUTDOWN_ACK_SENT,
 345	IP_VS_SCTP_S_REJECTED,
 346	IP_VS_SCTP_S_CLOSED,
 347	IP_VS_SCTP_S_LAST
 348};
 349
 
 
 
 
 
 350/* Delta sequence info structure
 351 * Each ip_vs_conn has 2 (output AND input seq. changes).
 352 * Only used in the VS/NAT.
 353 */
 354struct ip_vs_seq {
 355	__u32			init_seq;	/* Add delta from this seq */
 356	__u32			delta;		/* Delta in sequence numbers */
 357	__u32			previous_delta;	/* Delta in sequence numbers
 358						 * before last resized pkt */
 359};
 360
 361/* counters per cpu */
 362struct ip_vs_counters {
 363	__u64		conns;		/* connections scheduled */
 364	__u64		inpkts;		/* incoming packets */
 365	__u64		outpkts;	/* outgoing packets */
 366	__u64		inbytes;	/* incoming bytes */
 367	__u64		outbytes;	/* outgoing bytes */
 368};
 369/* Stats per cpu */
 370struct ip_vs_cpu_stats {
 371	struct ip_vs_counters   cnt;
 372	struct u64_stats_sync   syncp;
 373};
 374
 375/* IPVS statistics objects */
 376struct ip_vs_estimator {
 377	struct list_head	list;
 378
 379	u64			last_inbytes;
 380	u64			last_outbytes;
 381	u64			last_conns;
 382	u64			last_inpkts;
 383	u64			last_outpkts;
 384
 385	u64			cps;
 386	u64			inpps;
 387	u64			outpps;
 388	u64			inbps;
 389	u64			outbps;
 390};
 391
 392/*
 393 * IPVS statistics object, 64-bit kernel version of struct ip_vs_stats_user
 394 */
 395struct ip_vs_kstats {
 396	u64			conns;		/* connections scheduled */
 397	u64			inpkts;		/* incoming packets */
 398	u64			outpkts;	/* outgoing packets */
 399	u64			inbytes;	/* incoming bytes */
 400	u64			outbytes;	/* outgoing bytes */
 401
 402	u64			cps;		/* current connection rate */
 403	u64			inpps;		/* current in packet rate */
 404	u64			outpps;		/* current out packet rate */
 405	u64			inbps;		/* current in byte rate */
 406	u64			outbps;		/* current out byte rate */
 407};
 408
 409struct ip_vs_stats {
 410	struct ip_vs_kstats	kstats;		/* kernel statistics */
 411	struct ip_vs_estimator	est;		/* estimator */
 412	struct ip_vs_cpu_stats __percpu	*cpustats;	/* per cpu counters */
 413	spinlock_t		lock;		/* spin lock */
 414	struct ip_vs_kstats	kstats0;	/* reset values */
 415};
 416
 417struct dst_entry;
 418struct iphdr;
 419struct ip_vs_conn;
 420struct ip_vs_app;
 421struct sk_buff;
 422struct ip_vs_proto_data;
 423
 424struct ip_vs_protocol {
 425	struct ip_vs_protocol	*next;
 426	char			*name;
 427	u16			protocol;
 428	u16			num_states;
 429	int			dont_defrag;
 430
 431	void (*init)(struct ip_vs_protocol *pp);
 432
 433	void (*exit)(struct ip_vs_protocol *pp);
 434
 435	int (*init_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
 436
 437	void (*exit_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
 438
 439	int (*conn_schedule)(struct netns_ipvs *ipvs,
 440			     int af, struct sk_buff *skb,
 441			     struct ip_vs_proto_data *pd,
 442			     int *verdict, struct ip_vs_conn **cpp,
 443			     struct ip_vs_iphdr *iph);
 444
 445	struct ip_vs_conn *
 446	(*conn_in_get)(struct netns_ipvs *ipvs,
 447		       int af,
 448		       const struct sk_buff *skb,
 449		       const struct ip_vs_iphdr *iph);
 450
 451	struct ip_vs_conn *
 452	(*conn_out_get)(struct netns_ipvs *ipvs,
 453			int af,
 454			const struct sk_buff *skb,
 455			const struct ip_vs_iphdr *iph);
 456
 457	int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
 458			    struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
 459
 460	int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
 461			    struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
 462
 463	int (*csum_check)(int af, struct sk_buff *skb,
 464			  struct ip_vs_protocol *pp);
 465
 466	const char *(*state_name)(int state);
 467
 468	void (*state_transition)(struct ip_vs_conn *cp, int direction,
 469				 const struct sk_buff *skb,
 470				 struct ip_vs_proto_data *pd);
 471
 472	int (*register_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
 473
 474	void (*unregister_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
 475
 476	int (*app_conn_bind)(struct ip_vs_conn *cp);
 477
 478	void (*debug_packet)(int af, struct ip_vs_protocol *pp,
 479			     const struct sk_buff *skb,
 480			     int offset,
 481			     const char *msg);
 482
 483	void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
 484};
 485
 486/* protocol data per netns */
 487struct ip_vs_proto_data {
 488	struct ip_vs_proto_data	*next;
 489	struct ip_vs_protocol	*pp;
 490	int			*timeout_table;	/* protocol timeout table */
 491	atomic_t		appcnt;		/* counter of proto app incs. */
 492	struct tcp_states_t	*tcp_state_table;
 493};
 494
 495struct ip_vs_protocol   *ip_vs_proto_get(unsigned short proto);
 496struct ip_vs_proto_data *ip_vs_proto_data_get(struct netns_ipvs *ipvs,
 497					      unsigned short proto);
 498
 499struct ip_vs_conn_param {
 500	struct netns_ipvs		*ipvs;
 501	const union nf_inet_addr	*caddr;
 502	const union nf_inet_addr	*vaddr;
 503	__be16				cport;
 504	__be16				vport;
 505	__u16				protocol;
 506	u16				af;
 507
 508	const struct ip_vs_pe		*pe;
 509	char				*pe_data;
 510	__u8				pe_data_len;
 511};
 512
 513/* IP_VS structure allocated for each dynamically scheduled connection */
 514struct ip_vs_conn {
 515	struct hlist_node	c_list;         /* hashed list heads */
 516	/* Protocol, addresses and port numbers */
 517	__be16                  cport;
 518	__be16                  dport;
 519	__be16                  vport;
 520	u16			af;		/* address family */
 521	union nf_inet_addr      caddr;          /* client address */
 522	union nf_inet_addr      vaddr;          /* virtual address */
 523	union nf_inet_addr      daddr;          /* destination address */
 524	volatile __u32          flags;          /* status flags */
 525	__u16                   protocol;       /* Which protocol (TCP/UDP) */
 526	__u16			daf;		/* Address family of the dest */
 527	struct netns_ipvs	*ipvs;
 528
 529	/* counter and timer */
 530	refcount_t		refcnt;		/* reference count */
 531	struct timer_list	timer;		/* Expiration timer */
 532	volatile unsigned long	timeout;	/* timeout */
 533
 534	/* Flags and state transition */
 535	spinlock_t              lock;           /* lock for state transition */
 536	volatile __u16          state;          /* state info */
 537	volatile __u16          old_state;      /* old state, to be used for
 538						 * state transition triggerd
 539						 * synchronization
 540						 */
 541	__u32			fwmark;		/* Fire wall mark from skb */
 542	unsigned long		sync_endtime;	/* jiffies + sent_retries */
 543
 544	/* Control members */
 545	struct ip_vs_conn       *control;       /* Master control connection */
 546	atomic_t                n_control;      /* Number of controlled ones */
 547	struct ip_vs_dest       *dest;          /* real server */
 548	atomic_t                in_pkts;        /* incoming packet counter */
 549
 550	/* Packet transmitter for different forwarding methods.  If it
 551	 * mangles the packet, it must return NF_DROP or better NF_STOLEN,
 552	 * otherwise this must be changed to a sk_buff **.
 553	 * NF_ACCEPT can be returned when destination is local.
 554	 */
 555	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
 556			   struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
 557
 558	/* Note: we can group the following members into a structure,
 559	 * in order to save more space, and the following members are
 560	 * only used in VS/NAT anyway
 561	 */
 562	struct ip_vs_app        *app;           /* bound ip_vs_app object */
 563	void                    *app_data;      /* Application private data */
 564	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
 565	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
 566
 567	const struct ip_vs_pe	*pe;
 568	char			*pe_data;
 569	__u8			pe_data_len;
 570
 571	struct rcu_head		rcu_head;
 572};
 573
 574/* Extended internal versions of struct ip_vs_service_user and ip_vs_dest_user
 575 * for IPv6 support.
 576 *
 577 * We need these to conveniently pass around service and destination
 578 * options, but unfortunately, we also need to keep the old definitions to
 579 * maintain userspace backwards compatibility for the setsockopt interface.
 580 */
 581struct ip_vs_service_user_kern {
 582	/* virtual service addresses */
 583	u16			af;
 584	u16			protocol;
 585	union nf_inet_addr	addr;		/* virtual ip address */
 586	__be16			port;
 587	u32			fwmark;		/* firwall mark of service */
 588
 589	/* virtual service options */
 590	char			*sched_name;
 591	char			*pe_name;
 592	unsigned int		flags;		/* virtual service flags */
 593	unsigned int		timeout;	/* persistent timeout in sec */
 594	__be32			netmask;	/* persistent netmask or plen */
 595};
 596
 597
 598struct ip_vs_dest_user_kern {
 599	/* destination server address */
 600	union nf_inet_addr	addr;
 601	__be16			port;
 602
 603	/* real server options */
 604	unsigned int		conn_flags;	/* connection flags */
 605	int			weight;		/* destination weight */
 606
 607	/* thresholds for active connections */
 608	u32			u_threshold;	/* upper threshold */
 609	u32			l_threshold;	/* lower threshold */
 610
 611	/* Address family of addr */
 612	u16			af;
 
 
 
 
 613};
 614
 615
 616/*
 617 * The information about the virtual service offered to the net and the
 618 * forwarding entries.
 619 */
 620struct ip_vs_service {
 621	struct hlist_node	s_list;   /* for normal service table */
 622	struct hlist_node	f_list;   /* for fwmark-based service table */
 623	atomic_t		refcnt;   /* reference counter */
 624
 625	u16			af;       /* address family */
 626	__u16			protocol; /* which protocol (TCP/UDP) */
 627	union nf_inet_addr	addr;	  /* IP address for virtual service */
 628	__be16			port;	  /* port number for the service */
 629	__u32                   fwmark;   /* firewall mark of the service */
 630	unsigned int		flags;	  /* service status flags */
 631	unsigned int		timeout;  /* persistent timeout in ticks */
 632	__be32			netmask;  /* grouping granularity, mask/plen */
 633	struct netns_ipvs	*ipvs;
 634
 635	struct list_head	destinations;  /* real server d-linked list */
 636	__u32			num_dests;     /* number of servers */
 637	struct ip_vs_stats      stats;         /* statistics for the service */
 638
 639	/* for scheduling */
 640	struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
 641	spinlock_t		sched_lock;    /* lock sched_data */
 642	void			*sched_data;   /* scheduler application data */
 643
 644	/* alternate persistence engine */
 645	struct ip_vs_pe __rcu	*pe;
 
 646
 647	struct rcu_head		rcu_head;
 648};
 649
 650/* Information for cached dst */
 651struct ip_vs_dest_dst {
 652	struct dst_entry	*dst_cache;	/* destination cache entry */
 653	u32			dst_cookie;
 654	union nf_inet_addr	dst_saddr;
 655	struct rcu_head		rcu_head;
 656};
 657
 658/* The real server destination forwarding entry with ip address, port number,
 659 * and so on.
 660 */
 661struct ip_vs_dest {
 662	struct list_head	n_list;   /* for the dests in the service */
 663	struct hlist_node	d_list;   /* for table with all the dests */
 664
 665	u16			af;		/* address family */
 666	__be16			port;		/* port number of the server */
 667	union nf_inet_addr	addr;		/* IP address of the server */
 668	volatile unsigned int	flags;		/* dest status flags */
 669	atomic_t		conn_flags;	/* flags to copy to conn */
 670	atomic_t		weight;		/* server weight */
 
 
 
 
 671
 672	refcount_t		refcnt;		/* reference counter */
 673	struct ip_vs_stats      stats;          /* statistics */
 674	unsigned long		idle_start;	/* start time, jiffies */
 675
 676	/* connection counters and thresholds */
 677	atomic_t		activeconns;	/* active connections */
 678	atomic_t		inactconns;	/* inactive connections */
 679	atomic_t		persistconns;	/* persistent connections */
 680	__u32			u_threshold;	/* upper threshold */
 681	__u32			l_threshold;	/* lower threshold */
 682
 683	/* for destination cache */
 684	spinlock_t		dst_lock;	/* lock of dst_cache */
 685	struct ip_vs_dest_dst __rcu *dest_dst;	/* cached dst info */
 686
 687	/* for virtual service */
 688	struct ip_vs_service __rcu *svc;	/* service it belongs to */
 689	__u16			protocol;	/* which protocol (TCP/UDP) */
 690	__be16			vport;		/* virtual port number */
 691	union nf_inet_addr	vaddr;		/* virtual IP address */
 692	__u32			vfwmark;	/* firewall mark of service */
 693
 694	struct list_head	t_list;		/* in dest_trash */
 695	unsigned int		in_rs_table:1;	/* we are in rs_table */
 696};
 697
 698/* The scheduler object */
 699struct ip_vs_scheduler {
 700	struct list_head	n_list;		/* d-linked list head */
 701	char			*name;		/* scheduler name */
 702	atomic_t		refcnt;		/* reference counter */
 703	struct module		*module;	/* THIS_MODULE/NULL */
 704
 705	/* scheduler initializing service */
 706	int (*init_service)(struct ip_vs_service *svc);
 707	/* scheduling service finish */
 708	void (*done_service)(struct ip_vs_service *svc);
 709	/* dest is linked */
 710	int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
 711	/* dest is unlinked */
 712	int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
 713	/* dest is updated */
 714	int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
 715
 716	/* selecting a server from the given service */
 717	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 718				       const struct sk_buff *skb,
 719				       struct ip_vs_iphdr *iph);
 720};
 721
 722/* The persistence engine object */
 723struct ip_vs_pe {
 724	struct list_head	n_list;		/* d-linked list head */
 725	char			*name;		/* scheduler name */
 726	atomic_t		refcnt;		/* reference counter */
 727	struct module		*module;	/* THIS_MODULE/NULL */
 728
 729	/* get the connection template, if any */
 730	int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
 731	bool (*ct_match)(const struct ip_vs_conn_param *p,
 732			 struct ip_vs_conn *ct);
 733	u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
 734			   bool inverse);
 735	int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
 736	/* create connections for real-server outgoing packets */
 737	struct ip_vs_conn* (*conn_out)(struct ip_vs_service *svc,
 738				       struct ip_vs_dest *dest,
 739				       struct sk_buff *skb,
 740				       const struct ip_vs_iphdr *iph,
 741				       __be16 dport, __be16 cport);
 742};
 743
 744/* The application module object (a.k.a. app incarnation) */
 745struct ip_vs_app {
 746	struct list_head	a_list;		/* member in app list */
 747	int			type;		/* IP_VS_APP_TYPE_xxx */
 748	char			*name;		/* application module name */
 749	__u16			protocol;
 750	struct module		*module;	/* THIS_MODULE/NULL */
 751	struct list_head	incs_list;	/* list of incarnations */
 752
 753	/* members for application incarnations */
 754	struct list_head	p_list;		/* member in proto app list */
 755	struct ip_vs_app	*app;		/* its real application */
 756	__be16			port;		/* port number in net order */
 757	atomic_t		usecnt;		/* usage counter */
 758	struct rcu_head		rcu_head;
 759
 760	/* output hook: Process packet in inout direction, diff set for TCP.
 761	 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 762	 *	   2=Mangled but checksum was not updated
 763	 */
 764	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 765		       struct sk_buff *, int *diff);
 766
 767	/* input hook: Process packet in outin direction, diff set for TCP.
 768	 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 769	 *	   2=Mangled but checksum was not updated
 770	 */
 771	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 772		      struct sk_buff *, int *diff);
 773
 774	/* ip_vs_app initializer */
 775	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 776
 777	/* ip_vs_app finish */
 778	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 779
 780
 781	/* not used now */
 782	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 783			 struct ip_vs_protocol *);
 784
 785	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 786
 787	int *			timeout_table;
 788	int *			timeouts;
 789	int			timeouts_size;
 790
 791	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 792			     int *verdict, struct ip_vs_conn **cpp);
 793
 794	struct ip_vs_conn *
 795	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 796		       const struct iphdr *iph, int inverse);
 797
 798	struct ip_vs_conn *
 799	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 800			const struct iphdr *iph, int inverse);
 801
 802	int (*state_transition)(struct ip_vs_conn *cp, int direction,
 803				const struct sk_buff *skb,
 804				struct ip_vs_app *app);
 805
 806	void (*timeout_change)(struct ip_vs_app *app, int flags);
 807};
 808
 809struct ipvs_master_sync_state {
 810	struct list_head	sync_queue;
 811	struct ip_vs_sync_buff	*sync_buff;
 812	unsigned long		sync_queue_len;
 813	unsigned int		sync_queue_delay;
 814	struct task_struct	*master_thread;
 815	struct delayed_work	master_wakeup_work;
 816	struct netns_ipvs	*ipvs;
 817};
 818
 
 
 819/* How much time to keep dests in trash */
 820#define IP_VS_DEST_TRASH_PERIOD		(120 * HZ)
 821
 822struct ipvs_sync_daemon_cfg {
 823	union nf_inet_addr	mcast_group;
 824	int			syncid;
 825	u16			sync_maxlen;
 826	u16			mcast_port;
 827	u8			mcast_af;
 828	u8			mcast_ttl;
 829	/* multicast interface name */
 830	char			mcast_ifn[IP_VS_IFNAME_MAXLEN];
 831};
 832
 833/* IPVS in network namespace */
 834struct netns_ipvs {
 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)
 841
 842	struct hlist_head	rs_table[IP_VS_RTAB_SIZE];
 843	/* ip_vs_app */
 844	struct list_head	app_list;
 845	/* ip_vs_proto */
 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];
 854#endif
 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];
 861#endif
 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];
 869#endif
 870	/* ip_vs_conn */
 871	atomic_t		conn_count;      /* connection counter */
 872
 873	/* ip_vs_ctl */
 874	struct ip_vs_stats		tot_stats;  /* Statistics & est. */
 875
 876	int			num_services;    /* no of virtual services */
 
 877
 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;
 886
 887#ifdef CONFIG_SYSCTL
 
 
 888	/* 1/rate drop and drop-entry variables */
 889	struct delayed_work	defense_work;   /* Work handler */
 890	int			drop_rate;
 891	int			drop_counter;
 
 892	atomic_t		dropentry;
 893	/* locks in ctl.c */
 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 */
 897
 898	/* sys-ctl struct */
 899	struct ctl_table_header	*sysctl_hdr;
 900	struct ctl_table	*sysctl_tbl;
 901#endif
 902
 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;
 911#endif
 912	int			sysctl_snat_reroute;
 913	int			sysctl_sync_ver;
 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;
 932
 933	/* ip_vs_lblc */
 934	int			sysctl_lblc_expiration;
 935	struct ctl_table_header	*lblc_ctl_header;
 936	struct ctl_table	*lblc_ctl_table;
 937	/* ip_vs_lblcr */
 938	int			sysctl_lblcr_expiration;
 939	struct ctl_table_header	*lblcr_ctl_header;
 940	struct ctl_table	*lblcr_ctl_table;
 941	/* ip_vs_est */
 942	struct list_head	est_list;	/* estimator list */
 943	spinlock_t		est_lock;
 944	struct timer_list	est_timer;	/* Estimation timer */
 945	/* ip_vs_sync */
 946	spinlock_t		sync_lock;
 947	struct ipvs_master_sync_state *ms;
 948	spinlock_t		sync_buff_lock;
 949	struct task_struct	**backup_threads;
 
 950	int			threads_mask;
 951	volatile int		sync_state;
 952	struct mutex		sync_mutex;
 953	struct ipvs_sync_daemon_cfg	mcfg;	/* Master Configuration */
 954	struct ipvs_sync_daemon_cfg	bcfg;	/* Backup Configuration */
 955	/* net name space ptr */
 956	struct net		*net;            /* Needed by timer routines */
 957	/* Number of heterogeneous destinations, needed becaus heterogeneous
 958	 * are not supported when synchronization is enabled.
 959	 */
 960	unsigned int		mixed_address_family_dests;
 
 961};
 962
 963#define DEFAULT_SYNC_THRESHOLD	3
 964#define DEFAULT_SYNC_PERIOD	50
 965#define DEFAULT_SYNC_VER	1
 966#define DEFAULT_SLOPPY_TCP	0
 967#define DEFAULT_SLOPPY_SCTP	0
 968#define DEFAULT_SYNC_REFRESH_PERIOD	(0U * HZ)
 969#define DEFAULT_SYNC_RETRIES		0
 970#define IPVS_SYNC_WAKEUP_RATE	8
 971#define IPVS_SYNC_QLEN_MAX	(IPVS_SYNC_WAKEUP_RATE * 4)
 972#define IPVS_SYNC_SEND_DELAY	(HZ / 50)
 973#define IPVS_SYNC_CHECK_PERIOD	HZ
 974#define IPVS_SYNC_FLUSH_TIME	(HZ * 2)
 975#define IPVS_SYNC_PORTS_MAX	(1 << 6)
 976
 977#ifdef CONFIG_SYSCTL
 978
 979static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
 980{
 981	return ipvs->sysctl_sync_threshold[0];
 982}
 983
 984static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
 985{
 986	return READ_ONCE(ipvs->sysctl_sync_threshold[1]);
 987}
 988
 989static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
 990{
 991	return READ_ONCE(ipvs->sysctl_sync_refresh_period);
 992}
 993
 994static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
 995{
 996	return ipvs->sysctl_sync_retries;
 997}
 998
 999static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1000{
1001	return ipvs->sysctl_sync_ver;
1002}
1003
1004static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1005{
1006	return ipvs->sysctl_sloppy_tcp;
1007}
1008
1009static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1010{
1011	return ipvs->sysctl_sloppy_sctp;
1012}
1013
1014static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1015{
1016	return READ_ONCE(ipvs->sysctl_sync_ports);
1017}
1018
1019static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1020{
1021	return ipvs->sysctl_sync_persist_mode;
1022}
1023
1024static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1025{
1026	return ipvs->sysctl_sync_qlen_max;
1027}
1028
1029static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1030{
1031	return ipvs->sysctl_sync_sock_size;
1032}
1033
1034static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1035{
1036	return ipvs->sysctl_pmtu_disc;
1037}
1038
1039static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1040{
1041	return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1042	       ipvs->sysctl_backup_only;
1043}
1044
1045static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1046{
1047	return ipvs->sysctl_conn_reuse_mode;
1048}
1049
 
 
 
 
 
1050static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1051{
1052	return ipvs->sysctl_schedule_icmp;
1053}
1054
1055static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1056{
1057	return ipvs->sysctl_ignore_tunneled;
1058}
1059
1060static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1061{
1062	return ipvs->sysctl_cache_bypass;
1063}
1064
1065#else
1066
1067static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1068{
1069	return DEFAULT_SYNC_THRESHOLD;
1070}
1071
1072static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1073{
1074	return DEFAULT_SYNC_PERIOD;
1075}
1076
1077static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1078{
1079	return DEFAULT_SYNC_REFRESH_PERIOD;
1080}
1081
1082static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1083{
1084	return DEFAULT_SYNC_RETRIES & 3;
1085}
1086
1087static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1088{
1089	return DEFAULT_SYNC_VER;
1090}
1091
1092static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1093{
1094	return DEFAULT_SLOPPY_TCP;
1095}
1096
1097static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1098{
1099	return DEFAULT_SLOPPY_SCTP;
1100}
1101
1102static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1103{
1104	return 1;
1105}
1106
1107static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1108{
1109	return 0;
1110}
1111
1112static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1113{
1114	return IPVS_SYNC_QLEN_MAX;
1115}
1116
1117static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1118{
1119	return 0;
1120}
1121
1122static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1123{
1124	return 1;
1125}
1126
1127static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1128{
1129	return 0;
1130}
1131
1132static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1133{
1134	return 1;
1135}
1136
 
 
 
 
 
1137static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1138{
1139	return 0;
1140}
1141
1142static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1143{
1144	return 0;
1145}
1146
1147static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1148{
1149	return 0;
1150}
1151
1152#endif
1153
1154/* IPVS core functions
1155 * (from ip_vs_core.c)
1156 */
1157const char *ip_vs_proto_name(unsigned int proto);
1158void ip_vs_init_hash_table(struct list_head *table, int rows);
1159struct ip_vs_conn *ip_vs_new_conn_out(struct ip_vs_service *svc,
1160				      struct ip_vs_dest *dest,
1161				      struct sk_buff *skb,
1162				      const struct ip_vs_iphdr *iph,
1163				      __be16 dport,
1164				      __be16 cport);
1165#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1166
1167#define IP_VS_APP_TYPE_FTP	1
1168
1169/* ip_vs_conn handling functions
1170 * (from ip_vs_conn.c)
1171 */
1172enum {
1173	IP_VS_DIR_INPUT = 0,
1174	IP_VS_DIR_OUTPUT,
1175	IP_VS_DIR_INPUT_ONLY,
1176	IP_VS_DIR_LAST,
1177};
1178
1179static inline void ip_vs_conn_fill_param(struct netns_ipvs *ipvs, int af, int protocol,
1180					 const union nf_inet_addr *caddr,
1181					 __be16 cport,
1182					 const union nf_inet_addr *vaddr,
1183					 __be16 vport,
1184					 struct ip_vs_conn_param *p)
1185{
1186	p->ipvs = ipvs;
1187	p->af = af;
1188	p->protocol = protocol;
1189	p->caddr = caddr;
1190	p->cport = cport;
1191	p->vaddr = vaddr;
1192	p->vport = vport;
1193	p->pe = NULL;
1194	p->pe_data = NULL;
1195}
1196
1197struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1198struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1199
1200struct ip_vs_conn * ip_vs_conn_in_get_proto(struct netns_ipvs *ipvs, int af,
1201					    const struct sk_buff *skb,
1202					    const struct ip_vs_iphdr *iph);
1203
1204struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1205
1206struct ip_vs_conn * ip_vs_conn_out_get_proto(struct netns_ipvs *ipvs, int af,
1207					     const struct sk_buff *skb,
1208					     const struct ip_vs_iphdr *iph);
1209
1210/* Get reference to gain full access to conn.
1211 * By default, RCU read-side critical sections have access only to
1212 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1213 */
1214static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1215{
1216	return refcount_inc_not_zero(&cp->refcnt);
1217}
1218
1219/* put back the conn without restarting its timer */
1220static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1221{
1222	smp_mb__before_atomic();
1223	refcount_dec(&cp->refcnt);
1224}
1225void ip_vs_conn_put(struct ip_vs_conn *cp);
1226void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1227
1228struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, int dest_af,
1229				  const union nf_inet_addr *daddr,
1230				  __be16 dport, unsigned int flags,
1231				  struct ip_vs_dest *dest, __u32 fwmark);
1232void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1233
1234const char *ip_vs_state_name(__u16 proto, int state);
1235
1236void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
1237int ip_vs_check_template(struct ip_vs_conn *ct, struct ip_vs_dest *cdest);
1238void ip_vs_random_dropentry(struct netns_ipvs *ipvs);
1239int ip_vs_conn_init(void);
1240void ip_vs_conn_cleanup(void);
1241
1242static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1243{
1244	struct ip_vs_conn *ctl_cp = cp->control;
1245	if (!ctl_cp) {
1246		IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1247			      "%s:%d to %s:%d\n",
1248			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1249			      ntohs(cp->cport),
1250			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1251			      ntohs(cp->vport));
1252
1253		return;
1254	}
1255
1256	IP_VS_DBG_BUF(7, "DELeting control for: "
1257		      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1258		      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1259		      ntohs(cp->cport),
1260		      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1261		      ntohs(ctl_cp->cport));
1262
1263	cp->control = NULL;
1264	if (atomic_read(&ctl_cp->n_control) == 0) {
1265		IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1266			      "%s:%d to %s:%d\n",
1267			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1268			      ntohs(cp->cport),
1269			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1270			      ntohs(cp->vport));
1271
1272		return;
1273	}
1274	atomic_dec(&ctl_cp->n_control);
1275}
1276
1277static inline void
1278ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1279{
1280	if (cp->control) {
1281		IP_VS_ERR_BUF("request control ADD for already controlled: "
1282			      "%s:%d to %s:%d\n",
1283			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1284			      ntohs(cp->cport),
1285			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1286			      ntohs(cp->vport));
1287
1288		ip_vs_control_del(cp);
1289	}
1290
1291	IP_VS_DBG_BUF(7, "ADDing control for: "
1292		      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1293		      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1294		      ntohs(cp->cport),
1295		      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1296		      ntohs(ctl_cp->cport));
1297
1298	cp->control = ctl_cp;
1299	atomic_inc(&ctl_cp->n_control);
1300}
1301
 
 
 
 
 
 
 
 
 
 
 
1302/* IPVS netns init & cleanup functions */
1303int ip_vs_estimator_net_init(struct netns_ipvs *ipvs);
1304int ip_vs_control_net_init(struct netns_ipvs *ipvs);
1305int ip_vs_protocol_net_init(struct netns_ipvs *ipvs);
1306int ip_vs_app_net_init(struct netns_ipvs *ipvs);
1307int ip_vs_conn_net_init(struct netns_ipvs *ipvs);
1308int ip_vs_sync_net_init(struct netns_ipvs *ipvs);
1309void ip_vs_conn_net_cleanup(struct netns_ipvs *ipvs);
1310void ip_vs_app_net_cleanup(struct netns_ipvs *ipvs);
1311void ip_vs_protocol_net_cleanup(struct netns_ipvs *ipvs);
1312void ip_vs_control_net_cleanup(struct netns_ipvs *ipvs);
1313void ip_vs_estimator_net_cleanup(struct netns_ipvs *ipvs);
1314void ip_vs_sync_net_cleanup(struct netns_ipvs *ipvs);
1315void ip_vs_service_net_cleanup(struct netns_ipvs *ipvs);
1316
1317/* IPVS application functions
1318 * (from ip_vs_app.c)
1319 */
1320#define IP_VS_APP_MAX_PORTS  8
1321struct ip_vs_app *register_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1322void unregister_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1323int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1324void ip_vs_unbind_app(struct ip_vs_conn *cp);
1325int register_ip_vs_app_inc(struct netns_ipvs *ipvs, struct ip_vs_app *app, __u16 proto,
1326			   __u16 port);
1327int ip_vs_app_inc_get(struct ip_vs_app *inc);
1328void ip_vs_app_inc_put(struct ip_vs_app *inc);
1329
1330int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1331int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
 
 
1332
1333int register_ip_vs_pe(struct ip_vs_pe *pe);
1334int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1335struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1336struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1337
1338/* Use a #define to avoid all of module.h just for these trivial ops */
1339#define ip_vs_pe_get(pe)			\
1340	if (pe && pe->module)			\
1341		__module_get(pe->module);
1342
1343#define ip_vs_pe_put(pe)			\
1344	if (pe && pe->module)			\
1345		module_put(pe->module);
1346
1347/* IPVS protocol functions (from ip_vs_proto.c) */
1348int ip_vs_protocol_init(void);
1349void ip_vs_protocol_cleanup(void);
1350void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1351int *ip_vs_create_timeout_table(int *table, int size);
1352void ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1353			       const struct sk_buff *skb, int offset,
1354			       const char *msg);
1355
1356extern struct ip_vs_protocol ip_vs_protocol_tcp;
1357extern struct ip_vs_protocol ip_vs_protocol_udp;
1358extern struct ip_vs_protocol ip_vs_protocol_icmp;
1359extern struct ip_vs_protocol ip_vs_protocol_esp;
1360extern struct ip_vs_protocol ip_vs_protocol_ah;
1361extern struct ip_vs_protocol ip_vs_protocol_sctp;
1362
1363/* Registering/unregistering scheduler functions
1364 * (from ip_vs_sched.c)
1365 */
1366int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1367int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1368int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1369			 struct ip_vs_scheduler *scheduler);
1370void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1371			    struct ip_vs_scheduler *sched);
1372struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1373void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1374struct ip_vs_conn *
1375ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1376	       struct ip_vs_proto_data *pd, int *ignored,
1377	       struct ip_vs_iphdr *iph);
1378int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1379		struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1380
1381void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1382
1383/* IPVS control data and functions (from ip_vs_ctl.c) */
1384extern struct ip_vs_stats ip_vs_stats;
1385extern int sysctl_ip_vs_sync_ver;
1386
1387struct ip_vs_service *
1388ip_vs_service_find(struct netns_ipvs *ipvs, int af, __u32 fwmark, __u16 protocol,
1389		  const union nf_inet_addr *vaddr, __be16 vport);
1390
1391bool ip_vs_has_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1392			    const union nf_inet_addr *daddr, __be16 dport);
1393
1394struct ip_vs_dest *
1395ip_vs_find_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1396			const union nf_inet_addr *daddr, __be16 dport);
 
 
 
1397
1398int ip_vs_use_count_inc(void);
1399void ip_vs_use_count_dec(void);
1400int ip_vs_register_nl_ioctl(void);
1401void ip_vs_unregister_nl_ioctl(void);
1402int ip_vs_control_init(void);
1403void ip_vs_control_cleanup(void);
1404struct ip_vs_dest *
1405ip_vs_find_dest(struct netns_ipvs *ipvs, int svc_af, int dest_af,
1406		const union nf_inet_addr *daddr, __be16 dport,
1407		const union nf_inet_addr *vaddr, __be16 vport,
1408		__u16 protocol, __u32 fwmark, __u32 flags);
1409void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1410
1411static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1412{
1413	refcount_inc(&dest->refcnt);
1414}
1415
1416static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1417{
1418	smp_mb__before_atomic();
1419	refcount_dec(&dest->refcnt);
1420}
1421
1422static inline void ip_vs_dest_put_and_free(struct ip_vs_dest *dest)
1423{
1424	if (refcount_dec_and_test(&dest->refcnt))
1425		kfree(dest);
1426}
1427
1428/* IPVS sync daemon data and function prototypes
1429 * (from ip_vs_sync.c)
1430 */
1431int start_sync_thread(struct netns_ipvs *ipvs, struct ipvs_sync_daemon_cfg *cfg,
1432		      int state);
1433int stop_sync_thread(struct netns_ipvs *ipvs, int state);
1434void ip_vs_sync_conn(struct netns_ipvs *ipvs, struct ip_vs_conn *cp, int pkts);
1435
1436/* IPVS rate estimator prototypes (from ip_vs_est.c) */
1437void ip_vs_start_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1438void ip_vs_stop_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1439void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1440void ip_vs_read_estimator(struct ip_vs_kstats *dst, struct ip_vs_stats *stats);
1441
1442/* Various IPVS packet transmitters (from ip_vs_xmit.c) */
1443int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1444		    struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1445int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1446		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1447int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1448		   struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1449int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1450		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1451int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1452		  struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1453int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1454		    struct ip_vs_protocol *pp, int offset,
1455		    unsigned int hooknum, struct ip_vs_iphdr *iph);
1456void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1457
1458#ifdef CONFIG_IP_VS_IPV6
1459int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1460			 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1461int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1462		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1463int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1464			 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1465int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1466		     struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1467int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1468		       struct ip_vs_protocol *pp, int offset,
1469		       unsigned int hooknum, struct ip_vs_iphdr *iph);
1470#endif
1471
1472#ifdef CONFIG_SYSCTL
1473/* This is a simple mechanism to ignore packets when
1474 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1475 * we start to drop 1/rate of the packets
1476 */
1477static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1478{
1479	if (!ipvs->drop_rate)
1480		return 0;
1481	if (--ipvs->drop_counter > 0)
1482		return 0;
1483	ipvs->drop_counter = ipvs->drop_rate;
1484	return 1;
1485}
1486#else
1487static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1488#endif
1489
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1490/* ip_vs_fwd_tag returns the forwarding tag of the connection */
1491#define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
1492
1493static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1494{
1495	char fwd;
1496
1497	switch (IP_VS_FWD_METHOD(cp)) {
1498	case IP_VS_CONN_F_MASQ:
1499		fwd = 'M'; break;
1500	case IP_VS_CONN_F_LOCALNODE:
1501		fwd = 'L'; break;
1502	case IP_VS_CONN_F_TUNNEL:
1503		fwd = 'T'; break;
1504	case IP_VS_CONN_F_DROUTE:
1505		fwd = 'R'; break;
1506	case IP_VS_CONN_F_BYPASS:
1507		fwd = 'B'; break;
1508	default:
1509		fwd = '?'; break;
1510	}
1511	return fwd;
1512}
1513
1514void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1515		    struct ip_vs_conn *cp, int dir);
1516
1517#ifdef CONFIG_IP_VS_IPV6
1518void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1519		       struct ip_vs_conn *cp, int dir);
1520#endif
1521
1522__sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1523
1524static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1525{
1526	__be32 diff[2] = { ~old, new };
1527
1528	return csum_partial(diff, sizeof(diff), oldsum);
1529}
1530
1531#ifdef CONFIG_IP_VS_IPV6
1532static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1533					__wsum oldsum)
1534{
1535	__be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1536			    new[3],  new[2],  new[1],  new[0] };
1537
1538	return csum_partial(diff, sizeof(diff), oldsum);
1539}
1540#endif
1541
1542static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1543{
1544	__be16 diff[2] = { ~old, new };
1545
1546	return csum_partial(diff, sizeof(diff), oldsum);
1547}
1548
1549/* Forget current conntrack (unconfirmed) and attach notrack entry */
1550static inline void ip_vs_notrack(struct sk_buff *skb)
1551{
1552#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1553	enum ip_conntrack_info ctinfo;
1554	struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1555
1556	if (ct) {
1557		nf_conntrack_put(&ct->ct_general);
1558		nf_ct_set(skb, NULL, IP_CT_UNTRACKED);
1559	}
1560#endif
1561}
1562
1563#ifdef CONFIG_IP_VS_NFCT
1564/* Netfilter connection tracking
1565 * (from ip_vs_nfct.c)
1566 */
1567static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1568{
1569#ifdef CONFIG_SYSCTL
1570	return ipvs->sysctl_conntrack;
1571#else
1572	return 0;
1573#endif
1574}
1575
1576void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1577			    int outin);
1578int ip_vs_confirm_conntrack(struct sk_buff *skb);
1579void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1580			       struct ip_vs_conn *cp, u_int8_t proto,
1581			       const __be16 port, int from_rs);
1582void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1583
1584#else
1585
1586static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1587{
1588	return 0;
1589}
1590
1591static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1592					  struct ip_vs_conn *cp, int outin)
1593{
1594}
1595
1596static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1597{
1598	return NF_ACCEPT;
1599}
1600
1601static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1602{
1603}
1604#endif /* CONFIG_IP_VS_NFCT */
1605
1606/* Really using conntrack? */
1607static inline bool ip_vs_conn_uses_conntrack(struct ip_vs_conn *cp,
1608					     struct sk_buff *skb)
1609{
1610#ifdef CONFIG_IP_VS_NFCT
1611	enum ip_conntrack_info ctinfo;
1612	struct nf_conn *ct;
1613
1614	if (!(cp->flags & IP_VS_CONN_F_NFCT))
1615		return false;
1616	ct = nf_ct_get(skb, &ctinfo);
1617	if (ct)
1618		return true;
1619#endif
1620	return false;
1621}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1622
1623static inline int
1624ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1625{
1626	/* We think the overhead of processing active connections is 256
1627	 * times higher than that of inactive connections in average. (This
1628	 * 256 times might not be accurate, we will change it later) We
1629	 * use the following formula to estimate the overhead now:
1630	 *		  dest->activeconns*256 + dest->inactconns
1631	 */
1632	return (atomic_read(&dest->activeconns) << 8) +
1633		atomic_read(&dest->inactconns);
1634}
1635
1636#endif	/* _NET_IP_VS_H */
v5.9
   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_IP_VS_IPV6)
  29#include <linux/netfilter_ipv6/ip6_tables.h>
  30#endif
  31#if IS_ENABLED(CONFIG_NF_CONNTRACK)
  32#include <net/netfilter/nf_conntrack.h>
  33#endif
  34#include <net/net_namespace.h>		/* Netw namespace */
  35
  36#define IP_VS_HDR_INVERSE	1
  37#define IP_VS_HDR_ICMP		2
  38
  39/* Generic access of ipvs struct */
  40static inline struct netns_ipvs *net_ipvs(struct net* net)
  41{
  42	return net->ipvs;
  43}
  44
 
 
 
 
 
 
 
 
 
 
 
 
  45/* Connections' size value needed by ip_vs_ctl.c */
  46extern int ip_vs_conn_tab_size;
  47
  48struct ip_vs_iphdr {
  49	int hdr_flags;	/* ipvs flags */
  50	__u32 off;	/* Where IP or IPv4 header starts */
  51	__u32 len;	/* IPv4 simply where L4 starts
  52			 * IPv6 where L4 Transport Header starts */
  53	__u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
  54	__s16 protocol;
  55	__s32 flags;
  56	union nf_inet_addr saddr;
  57	union nf_inet_addr daddr;
  58};
  59
  60static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
  61				      int len, void *buffer)
  62{
  63	return skb_header_pointer(skb, offset, len, buffer);
  64}
  65
  66/* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
  67 * IPv6 requires some extra work, as finding proper header position,
  68 * depend on the IPv6 extension headers.
  69 */
  70static inline int
  71ip_vs_fill_iph_skb_off(int af, const struct sk_buff *skb, int offset,
  72		       int hdr_flags, struct ip_vs_iphdr *iphdr)
  73{
  74	iphdr->hdr_flags = hdr_flags;
  75	iphdr->off = offset;
  76
  77#ifdef CONFIG_IP_VS_IPV6
  78	if (af == AF_INET6) {
  79		struct ipv6hdr _iph;
  80		const struct ipv6hdr *iph = skb_header_pointer(
  81			skb, offset, sizeof(_iph), &_iph);
  82		if (!iph)
  83			return 0;
  84
  85		iphdr->saddr.in6 = iph->saddr;
  86		iphdr->daddr.in6 = iph->daddr;
  87		/* ipv6_find_hdr() updates len, flags */
  88		iphdr->len	 = offset;
  89		iphdr->flags	 = 0;
  90		iphdr->protocol  = ipv6_find_hdr(skb, &iphdr->len, -1,
  91						 &iphdr->fragoffs,
  92						 &iphdr->flags);
  93		if (iphdr->protocol < 0)
  94			return 0;
  95	} else
  96#endif
  97	{
  98		struct iphdr _iph;
  99		const struct iphdr *iph = skb_header_pointer(
 100			skb, offset, sizeof(_iph), &_iph);
 101		if (!iph)
 102			return 0;
 103
 104		iphdr->len	= offset + iph->ihl * 4;
 105		iphdr->fragoffs	= 0;
 106		iphdr->protocol	= iph->protocol;
 107		iphdr->saddr.ip	= iph->saddr;
 108		iphdr->daddr.ip	= iph->daddr;
 109	}
 110
 111	return 1;
 112}
 113
 114static inline int
 115ip_vs_fill_iph_skb_icmp(int af, const struct sk_buff *skb, int offset,
 116			bool inverse, struct ip_vs_iphdr *iphdr)
 117{
 118	int hdr_flags = IP_VS_HDR_ICMP;
 119
 120	if (inverse)
 121		hdr_flags |= IP_VS_HDR_INVERSE;
 122
 123	return ip_vs_fill_iph_skb_off(af, skb, offset, hdr_flags, iphdr);
 124}
 125
 126static inline int
 127ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, bool inverse,
 128		   struct ip_vs_iphdr *iphdr)
 129{
 130	int hdr_flags = 0;
 131
 132	if (inverse)
 133		hdr_flags |= IP_VS_HDR_INVERSE;
 134
 135	return ip_vs_fill_iph_skb_off(af, skb, skb_network_offset(skb),
 136				      hdr_flags, iphdr);
 137}
 138
 139static inline bool
 140ip_vs_iph_inverse(const struct ip_vs_iphdr *iph)
 141{
 142	return !!(iph->hdr_flags & IP_VS_HDR_INVERSE);
 143}
 144
 145static inline bool
 146ip_vs_iph_icmp(const struct ip_vs_iphdr *iph)
 147{
 148	return !!(iph->hdr_flags & IP_VS_HDR_ICMP);
 149}
 150
 151static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
 152				   const union nf_inet_addr *src)
 153{
 154#ifdef CONFIG_IP_VS_IPV6
 155	if (af == AF_INET6)
 156		dst->in6 = src->in6;
 157	else
 158#endif
 159	dst->ip = src->ip;
 160}
 161
 162static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
 163				  const union nf_inet_addr *src)
 164{
 165#ifdef CONFIG_IP_VS_IPV6
 166	if (af == AF_INET6) {
 167		dst->in6 = src->in6;
 168		return;
 169	}
 170#endif
 171	dst->ip = src->ip;
 172	dst->all[1] = 0;
 173	dst->all[2] = 0;
 174	dst->all[3] = 0;
 175}
 176
 177static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
 178				   const union nf_inet_addr *b)
 179{
 180#ifdef CONFIG_IP_VS_IPV6
 181	if (af == AF_INET6)
 182		return ipv6_addr_equal(&a->in6, &b->in6);
 183#endif
 184	return a->ip == b->ip;
 185}
 186
 187#ifdef CONFIG_IP_VS_DEBUG
 188#include <linux/net.h>
 189
 190int ip_vs_get_debug_level(void);
 191
 192static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
 193					 const union nf_inet_addr *addr,
 194					 int *idx)
 195{
 196	int len;
 197#ifdef CONFIG_IP_VS_IPV6
 198	if (af == AF_INET6)
 199		len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
 200			       &addr->in6) + 1;
 201	else
 202#endif
 203		len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
 204			       &addr->ip) + 1;
 205
 206	*idx += len;
 207	BUG_ON(*idx > buf_len + 1);
 208	return &buf[*idx - len];
 209}
 210
 211#define IP_VS_DBG_BUF(level, msg, ...)					\
 212	do {								\
 213		char ip_vs_dbg_buf[160];				\
 214		int ip_vs_dbg_idx = 0;					\
 215		if (level <= ip_vs_get_debug_level())			\
 216			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
 217	} while (0)
 218#define IP_VS_ERR_BUF(msg...)						\
 219	do {								\
 220		char ip_vs_dbg_buf[160];				\
 221		int ip_vs_dbg_idx = 0;					\
 222		pr_err(msg);						\
 223	} while (0)
 224
 225/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
 226#define IP_VS_DBG_ADDR(af, addr)					\
 227	ip_vs_dbg_addr(af, ip_vs_dbg_buf,				\
 228		       sizeof(ip_vs_dbg_buf), addr,			\
 229		       &ip_vs_dbg_idx)
 230
 231#define IP_VS_DBG(level, msg, ...)					\
 232	do {								\
 233		if (level <= ip_vs_get_debug_level())			\
 234			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
 235	} while (0)
 236#define IP_VS_DBG_RL(msg, ...)						\
 237	do {								\
 238		if (net_ratelimit())					\
 239			printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__);	\
 240	} while (0)
 241#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)			\
 242	do {								\
 243		if (level <= ip_vs_get_debug_level())			\
 244			pp->debug_packet(af, pp, skb, ofs, msg);	\
 245	} while (0)
 246#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)			\
 247	do {								\
 248		if (level <= ip_vs_get_debug_level() &&			\
 249		    net_ratelimit())					\
 250			pp->debug_packet(af, pp, skb, ofs, msg);	\
 251	} while (0)
 252#else	/* NO DEBUGGING at ALL */
 253#define IP_VS_DBG_BUF(level, msg...)  do {} while (0)
 254#define IP_VS_ERR_BUF(msg...)  do {} while (0)
 255#define IP_VS_DBG(level, msg...)  do {} while (0)
 256#define IP_VS_DBG_RL(msg...)  do {} while (0)
 257#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg)	do {} while (0)
 258#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg)	do {} while (0)
 259#endif
 260
 261#define IP_VS_BUG() BUG()
 262#define IP_VS_ERR_RL(msg, ...)						\
 263	do {								\
 264		if (net_ratelimit())					\
 265			pr_err(msg, ##__VA_ARGS__);			\
 266	} while (0)
 267
 268#ifdef CONFIG_IP_VS_DEBUG
 269#define EnterFunction(level)						\
 270	do {								\
 271		if (level <= ip_vs_get_debug_level())			\
 272			printk(KERN_DEBUG				\
 273			       pr_fmt("Enter: %s, %s line %i\n"),	\
 274			       __func__, __FILE__, __LINE__);		\
 275	} while (0)
 276#define LeaveFunction(level)						\
 277	do {								\
 278		if (level <= ip_vs_get_debug_level())			\
 279			printk(KERN_DEBUG				\
 280			       pr_fmt("Leave: %s, %s line %i\n"),	\
 281			       __func__, __FILE__, __LINE__);		\
 282	} while (0)
 283#else
 284#define EnterFunction(level)   do {} while (0)
 285#define LeaveFunction(level)   do {} while (0)
 286#endif
 287
 288/* The port number of FTP service (in network order). */
 289#define FTPPORT  cpu_to_be16(21)
 290#define FTPDATA  cpu_to_be16(20)
 291
 292/* TCP State Values */
 293enum {
 294	IP_VS_TCP_S_NONE = 0,
 295	IP_VS_TCP_S_ESTABLISHED,
 296	IP_VS_TCP_S_SYN_SENT,
 297	IP_VS_TCP_S_SYN_RECV,
 298	IP_VS_TCP_S_FIN_WAIT,
 299	IP_VS_TCP_S_TIME_WAIT,
 300	IP_VS_TCP_S_CLOSE,
 301	IP_VS_TCP_S_CLOSE_WAIT,
 302	IP_VS_TCP_S_LAST_ACK,
 303	IP_VS_TCP_S_LISTEN,
 304	IP_VS_TCP_S_SYNACK,
 305	IP_VS_TCP_S_LAST
 306};
 307
 308/* UDP State Values */
 309enum {
 310	IP_VS_UDP_S_NORMAL,
 311	IP_VS_UDP_S_LAST,
 312};
 313
 314/* ICMP State Values */
 315enum {
 316	IP_VS_ICMP_S_NORMAL,
 317	IP_VS_ICMP_S_LAST,
 318};
 319
 320/* SCTP State Values */
 321enum ip_vs_sctp_states {
 322	IP_VS_SCTP_S_NONE,
 323	IP_VS_SCTP_S_INIT1,
 324	IP_VS_SCTP_S_INIT,
 325	IP_VS_SCTP_S_COOKIE_SENT,
 326	IP_VS_SCTP_S_COOKIE_REPLIED,
 327	IP_VS_SCTP_S_COOKIE_WAIT,
 328	IP_VS_SCTP_S_COOKIE,
 329	IP_VS_SCTP_S_COOKIE_ECHOED,
 330	IP_VS_SCTP_S_ESTABLISHED,
 331	IP_VS_SCTP_S_SHUTDOWN_SENT,
 332	IP_VS_SCTP_S_SHUTDOWN_RECEIVED,
 333	IP_VS_SCTP_S_SHUTDOWN_ACK_SENT,
 334	IP_VS_SCTP_S_REJECTED,
 335	IP_VS_SCTP_S_CLOSED,
 336	IP_VS_SCTP_S_LAST
 337};
 338
 339/* Connection templates use bits from state */
 340#define IP_VS_CTPL_S_NONE		0x0000
 341#define IP_VS_CTPL_S_ASSURED		0x0001
 342#define IP_VS_CTPL_S_LAST		0x0002
 343
 344/* Delta sequence info structure
 345 * Each ip_vs_conn has 2 (output AND input seq. changes).
 346 * Only used in the VS/NAT.
 347 */
 348struct ip_vs_seq {
 349	__u32			init_seq;	/* Add delta from this seq */
 350	__u32			delta;		/* Delta in sequence numbers */
 351	__u32			previous_delta;	/* Delta in sequence numbers
 352						 * before last resized pkt */
 353};
 354
 355/* counters per cpu */
 356struct ip_vs_counters {
 357	__u64		conns;		/* connections scheduled */
 358	__u64		inpkts;		/* incoming packets */
 359	__u64		outpkts;	/* outgoing packets */
 360	__u64		inbytes;	/* incoming bytes */
 361	__u64		outbytes;	/* outgoing bytes */
 362};
 363/* Stats per cpu */
 364struct ip_vs_cpu_stats {
 365	struct ip_vs_counters   cnt;
 366	struct u64_stats_sync   syncp;
 367};
 368
 369/* IPVS statistics objects */
 370struct ip_vs_estimator {
 371	struct list_head	list;
 372
 373	u64			last_inbytes;
 374	u64			last_outbytes;
 375	u64			last_conns;
 376	u64			last_inpkts;
 377	u64			last_outpkts;
 378
 379	u64			cps;
 380	u64			inpps;
 381	u64			outpps;
 382	u64			inbps;
 383	u64			outbps;
 384};
 385
 386/*
 387 * IPVS statistics object, 64-bit kernel version of struct ip_vs_stats_user
 388 */
 389struct ip_vs_kstats {
 390	u64			conns;		/* connections scheduled */
 391	u64			inpkts;		/* incoming packets */
 392	u64			outpkts;	/* outgoing packets */
 393	u64			inbytes;	/* incoming bytes */
 394	u64			outbytes;	/* outgoing bytes */
 395
 396	u64			cps;		/* current connection rate */
 397	u64			inpps;		/* current in packet rate */
 398	u64			outpps;		/* current out packet rate */
 399	u64			inbps;		/* current in byte rate */
 400	u64			outbps;		/* current out byte rate */
 401};
 402
 403struct ip_vs_stats {
 404	struct ip_vs_kstats	kstats;		/* kernel statistics */
 405	struct ip_vs_estimator	est;		/* estimator */
 406	struct ip_vs_cpu_stats __percpu	*cpustats;	/* per cpu counters */
 407	spinlock_t		lock;		/* spin lock */
 408	struct ip_vs_kstats	kstats0;	/* reset values */
 409};
 410
 411struct dst_entry;
 412struct iphdr;
 413struct ip_vs_conn;
 414struct ip_vs_app;
 415struct sk_buff;
 416struct ip_vs_proto_data;
 417
 418struct ip_vs_protocol {
 419	struct ip_vs_protocol	*next;
 420	char			*name;
 421	u16			protocol;
 422	u16			num_states;
 423	int			dont_defrag;
 424
 425	void (*init)(struct ip_vs_protocol *pp);
 426
 427	void (*exit)(struct ip_vs_protocol *pp);
 428
 429	int (*init_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
 430
 431	void (*exit_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
 432
 433	int (*conn_schedule)(struct netns_ipvs *ipvs,
 434			     int af, struct sk_buff *skb,
 435			     struct ip_vs_proto_data *pd,
 436			     int *verdict, struct ip_vs_conn **cpp,
 437			     struct ip_vs_iphdr *iph);
 438
 439	struct ip_vs_conn *
 440	(*conn_in_get)(struct netns_ipvs *ipvs,
 441		       int af,
 442		       const struct sk_buff *skb,
 443		       const struct ip_vs_iphdr *iph);
 444
 445	struct ip_vs_conn *
 446	(*conn_out_get)(struct netns_ipvs *ipvs,
 447			int af,
 448			const struct sk_buff *skb,
 449			const struct ip_vs_iphdr *iph);
 450
 451	int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
 452			    struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
 453
 454	int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
 455			    struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
 456
 
 
 
 457	const char *(*state_name)(int state);
 458
 459	void (*state_transition)(struct ip_vs_conn *cp, int direction,
 460				 const struct sk_buff *skb,
 461				 struct ip_vs_proto_data *pd);
 462
 463	int (*register_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
 464
 465	void (*unregister_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
 466
 467	int (*app_conn_bind)(struct ip_vs_conn *cp);
 468
 469	void (*debug_packet)(int af, struct ip_vs_protocol *pp,
 470			     const struct sk_buff *skb,
 471			     int offset,
 472			     const char *msg);
 473
 474	void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
 475};
 476
 477/* protocol data per netns */
 478struct ip_vs_proto_data {
 479	struct ip_vs_proto_data	*next;
 480	struct ip_vs_protocol	*pp;
 481	int			*timeout_table;	/* protocol timeout table */
 482	atomic_t		appcnt;		/* counter of proto app incs. */
 483	struct tcp_states_t	*tcp_state_table;
 484};
 485
 486struct ip_vs_protocol   *ip_vs_proto_get(unsigned short proto);
 487struct ip_vs_proto_data *ip_vs_proto_data_get(struct netns_ipvs *ipvs,
 488					      unsigned short proto);
 489
 490struct ip_vs_conn_param {
 491	struct netns_ipvs		*ipvs;
 492	const union nf_inet_addr	*caddr;
 493	const union nf_inet_addr	*vaddr;
 494	__be16				cport;
 495	__be16				vport;
 496	__u16				protocol;
 497	u16				af;
 498
 499	const struct ip_vs_pe		*pe;
 500	char				*pe_data;
 501	__u8				pe_data_len;
 502};
 503
 504/* IP_VS structure allocated for each dynamically scheduled connection */
 505struct ip_vs_conn {
 506	struct hlist_node	c_list;         /* hashed list heads */
 507	/* Protocol, addresses and port numbers */
 508	__be16                  cport;
 509	__be16                  dport;
 510	__be16                  vport;
 511	u16			af;		/* address family */
 512	union nf_inet_addr      caddr;          /* client address */
 513	union nf_inet_addr      vaddr;          /* virtual address */
 514	union nf_inet_addr      daddr;          /* destination address */
 515	volatile __u32          flags;          /* status flags */
 516	__u16                   protocol;       /* Which protocol (TCP/UDP) */
 517	__u16			daf;		/* Address family of the dest */
 518	struct netns_ipvs	*ipvs;
 519
 520	/* counter and timer */
 521	refcount_t		refcnt;		/* reference count */
 522	struct timer_list	timer;		/* Expiration timer */
 523	volatile unsigned long	timeout;	/* timeout */
 524
 525	/* Flags and state transition */
 526	spinlock_t              lock;           /* lock for state transition */
 527	volatile __u16          state;          /* state info */
 528	volatile __u16          old_state;      /* old state, to be used for
 529						 * state transition triggerd
 530						 * synchronization
 531						 */
 532	__u32			fwmark;		/* Fire wall mark from skb */
 533	unsigned long		sync_endtime;	/* jiffies + sent_retries */
 534
 535	/* Control members */
 536	struct ip_vs_conn       *control;       /* Master control connection */
 537	atomic_t                n_control;      /* Number of controlled ones */
 538	struct ip_vs_dest       *dest;          /* real server */
 539	atomic_t                in_pkts;        /* incoming packet counter */
 540
 541	/* Packet transmitter for different forwarding methods.  If it
 542	 * mangles the packet, it must return NF_DROP or better NF_STOLEN,
 543	 * otherwise this must be changed to a sk_buff **.
 544	 * NF_ACCEPT can be returned when destination is local.
 545	 */
 546	int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
 547			   struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
 548
 549	/* Note: we can group the following members into a structure,
 550	 * in order to save more space, and the following members are
 551	 * only used in VS/NAT anyway
 552	 */
 553	struct ip_vs_app        *app;           /* bound ip_vs_app object */
 554	void                    *app_data;      /* Application private data */
 555	struct ip_vs_seq        in_seq;         /* incoming seq. struct */
 556	struct ip_vs_seq        out_seq;        /* outgoing seq. struct */
 557
 558	const struct ip_vs_pe	*pe;
 559	char			*pe_data;
 560	__u8			pe_data_len;
 561
 562	struct rcu_head		rcu_head;
 563};
 564
 565/* Extended internal versions of struct ip_vs_service_user and ip_vs_dest_user
 566 * for IPv6 support.
 567 *
 568 * We need these to conveniently pass around service and destination
 569 * options, but unfortunately, we also need to keep the old definitions to
 570 * maintain userspace backwards compatibility for the setsockopt interface.
 571 */
 572struct ip_vs_service_user_kern {
 573	/* virtual service addresses */
 574	u16			af;
 575	u16			protocol;
 576	union nf_inet_addr	addr;		/* virtual ip address */
 577	__be16			port;
 578	u32			fwmark;		/* firwall mark of service */
 579
 580	/* virtual service options */
 581	char			*sched_name;
 582	char			*pe_name;
 583	unsigned int		flags;		/* virtual service flags */
 584	unsigned int		timeout;	/* persistent timeout in sec */
 585	__be32			netmask;	/* persistent netmask or plen */
 586};
 587
 588
 589struct ip_vs_dest_user_kern {
 590	/* destination server address */
 591	union nf_inet_addr	addr;
 592	__be16			port;
 593
 594	/* real server options */
 595	unsigned int		conn_flags;	/* connection flags */
 596	int			weight;		/* destination weight */
 597
 598	/* thresholds for active connections */
 599	u32			u_threshold;	/* upper threshold */
 600	u32			l_threshold;	/* lower threshold */
 601
 602	/* Address family of addr */
 603	u16			af;
 604
 605	u16			tun_type;	/* tunnel type */
 606	__be16			tun_port;	/* tunnel port */
 607	u16			tun_flags;	/* tunnel flags */
 608};
 609
 610
 611/*
 612 * The information about the virtual service offered to the net and the
 613 * forwarding entries.
 614 */
 615struct ip_vs_service {
 616	struct hlist_node	s_list;   /* for normal service table */
 617	struct hlist_node	f_list;   /* for fwmark-based service table */
 618	atomic_t		refcnt;   /* reference counter */
 619
 620	u16			af;       /* address family */
 621	__u16			protocol; /* which protocol (TCP/UDP) */
 622	union nf_inet_addr	addr;	  /* IP address for virtual service */
 623	__be16			port;	  /* port number for the service */
 624	__u32                   fwmark;   /* firewall mark of the service */
 625	unsigned int		flags;	  /* service status flags */
 626	unsigned int		timeout;  /* persistent timeout in ticks */
 627	__be32			netmask;  /* grouping granularity, mask/plen */
 628	struct netns_ipvs	*ipvs;
 629
 630	struct list_head	destinations;  /* real server d-linked list */
 631	__u32			num_dests;     /* number of servers */
 632	struct ip_vs_stats      stats;         /* statistics for the service */
 633
 634	/* for scheduling */
 635	struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
 636	spinlock_t		sched_lock;    /* lock sched_data */
 637	void			*sched_data;   /* scheduler application data */
 638
 639	/* alternate persistence engine */
 640	struct ip_vs_pe __rcu	*pe;
 641	int			conntrack_afmask;
 642
 643	struct rcu_head		rcu_head;
 644};
 645
 646/* Information for cached dst */
 647struct ip_vs_dest_dst {
 648	struct dst_entry	*dst_cache;	/* destination cache entry */
 649	u32			dst_cookie;
 650	union nf_inet_addr	dst_saddr;
 651	struct rcu_head		rcu_head;
 652};
 653
 654/* The real server destination forwarding entry with ip address, port number,
 655 * and so on.
 656 */
 657struct ip_vs_dest {
 658	struct list_head	n_list;   /* for the dests in the service */
 659	struct hlist_node	d_list;   /* for table with all the dests */
 660
 661	u16			af;		/* address family */
 662	__be16			port;		/* port number of the server */
 663	union nf_inet_addr	addr;		/* IP address of the server */
 664	volatile unsigned int	flags;		/* dest status flags */
 665	atomic_t		conn_flags;	/* flags to copy to conn */
 666	atomic_t		weight;		/* server weight */
 667	atomic_t		last_weight;	/* server latest weight */
 668	__u16			tun_type;	/* tunnel type */
 669	__be16			tun_port;	/* tunnel port */
 670	__u16			tun_flags;	/* tunnel flags */
 671
 672	refcount_t		refcnt;		/* reference counter */
 673	struct ip_vs_stats      stats;          /* statistics */
 674	unsigned long		idle_start;	/* start time, jiffies */
 675
 676	/* connection counters and thresholds */
 677	atomic_t		activeconns;	/* active connections */
 678	atomic_t		inactconns;	/* inactive connections */
 679	atomic_t		persistconns;	/* persistent connections */
 680	__u32			u_threshold;	/* upper threshold */
 681	__u32			l_threshold;	/* lower threshold */
 682
 683	/* for destination cache */
 684	spinlock_t		dst_lock;	/* lock of dst_cache */
 685	struct ip_vs_dest_dst __rcu *dest_dst;	/* cached dst info */
 686
 687	/* for virtual service */
 688	struct ip_vs_service __rcu *svc;	/* service it belongs to */
 689	__u16			protocol;	/* which protocol (TCP/UDP) */
 690	__be16			vport;		/* virtual port number */
 691	union nf_inet_addr	vaddr;		/* virtual IP address */
 692	__u32			vfwmark;	/* firewall mark of service */
 693
 694	struct list_head	t_list;		/* in dest_trash */
 695	unsigned int		in_rs_table:1;	/* we are in rs_table */
 696};
 697
 698/* The scheduler object */
 699struct ip_vs_scheduler {
 700	struct list_head	n_list;		/* d-linked list head */
 701	char			*name;		/* scheduler name */
 702	atomic_t		refcnt;		/* reference counter */
 703	struct module		*module;	/* THIS_MODULE/NULL */
 704
 705	/* scheduler initializing service */
 706	int (*init_service)(struct ip_vs_service *svc);
 707	/* scheduling service finish */
 708	void (*done_service)(struct ip_vs_service *svc);
 709	/* dest is linked */
 710	int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
 711	/* dest is unlinked */
 712	int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
 713	/* dest is updated */
 714	int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
 715
 716	/* selecting a server from the given service */
 717	struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
 718				       const struct sk_buff *skb,
 719				       struct ip_vs_iphdr *iph);
 720};
 721
 722/* The persistence engine object */
 723struct ip_vs_pe {
 724	struct list_head	n_list;		/* d-linked list head */
 725	char			*name;		/* scheduler name */
 726	atomic_t		refcnt;		/* reference counter */
 727	struct module		*module;	/* THIS_MODULE/NULL */
 728
 729	/* get the connection template, if any */
 730	int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
 731	bool (*ct_match)(const struct ip_vs_conn_param *p,
 732			 struct ip_vs_conn *ct);
 733	u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
 734			   bool inverse);
 735	int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
 736	/* create connections for real-server outgoing packets */
 737	struct ip_vs_conn* (*conn_out)(struct ip_vs_service *svc,
 738				       struct ip_vs_dest *dest,
 739				       struct sk_buff *skb,
 740				       const struct ip_vs_iphdr *iph,
 741				       __be16 dport, __be16 cport);
 742};
 743
 744/* The application module object (a.k.a. app incarnation) */
 745struct ip_vs_app {
 746	struct list_head	a_list;		/* member in app list */
 747	int			type;		/* IP_VS_APP_TYPE_xxx */
 748	char			*name;		/* application module name */
 749	__u16			protocol;
 750	struct module		*module;	/* THIS_MODULE/NULL */
 751	struct list_head	incs_list;	/* list of incarnations */
 752
 753	/* members for application incarnations */
 754	struct list_head	p_list;		/* member in proto app list */
 755	struct ip_vs_app	*app;		/* its real application */
 756	__be16			port;		/* port number in net order */
 757	atomic_t		usecnt;		/* usage counter */
 758	struct rcu_head		rcu_head;
 759
 760	/* output hook: Process packet in inout direction, diff set for TCP.
 761	 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 762	 *	   2=Mangled but checksum was not updated
 763	 */
 764	int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
 765		       struct sk_buff *, int *diff, struct ip_vs_iphdr *ipvsh);
 766
 767	/* input hook: Process packet in outin direction, diff set for TCP.
 768	 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
 769	 *	   2=Mangled but checksum was not updated
 770	 */
 771	int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
 772		      struct sk_buff *, int *diff, struct ip_vs_iphdr *ipvsh);
 773
 774	/* ip_vs_app initializer */
 775	int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 776
 777	/* ip_vs_app finish */
 778	int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 779
 780
 781	/* not used now */
 782	int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
 783			 struct ip_vs_protocol *);
 784
 785	void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
 786
 787	int *			timeout_table;
 788	int *			timeouts;
 789	int			timeouts_size;
 790
 791	int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
 792			     int *verdict, struct ip_vs_conn **cpp);
 793
 794	struct ip_vs_conn *
 795	(*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 796		       const struct iphdr *iph, int inverse);
 797
 798	struct ip_vs_conn *
 799	(*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
 800			const struct iphdr *iph, int inverse);
 801
 802	int (*state_transition)(struct ip_vs_conn *cp, int direction,
 803				const struct sk_buff *skb,
 804				struct ip_vs_app *app);
 805
 806	void (*timeout_change)(struct ip_vs_app *app, int flags);
 807};
 808
 809struct ipvs_master_sync_state {
 810	struct list_head	sync_queue;
 811	struct ip_vs_sync_buff	*sync_buff;
 812	unsigned long		sync_queue_len;
 813	unsigned int		sync_queue_delay;
 
 814	struct delayed_work	master_wakeup_work;
 815	struct netns_ipvs	*ipvs;
 816};
 817
 818struct ip_vs_sync_thread_data;
 819
 820/* How much time to keep dests in trash */
 821#define IP_VS_DEST_TRASH_PERIOD		(120 * HZ)
 822
 823struct ipvs_sync_daemon_cfg {
 824	union nf_inet_addr	mcast_group;
 825	int			syncid;
 826	u16			sync_maxlen;
 827	u16			mcast_port;
 828	u8			mcast_af;
 829	u8			mcast_ttl;
 830	/* multicast interface name */
 831	char			mcast_ifn[IP_VS_IFNAME_MAXLEN];
 832};
 833
 834/* IPVS in network namespace */
 835struct netns_ipvs {
 836	int			gen;		/* Generation */
 837	int			enable;		/* enable like nf_hooks do */
 838	/* Hash table: for real service lookups */
 839	#define IP_VS_RTAB_BITS 4
 840	#define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
 841	#define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
 842
 843	struct hlist_head	rs_table[IP_VS_RTAB_SIZE];
 844	/* ip_vs_app */
 845	struct list_head	app_list;
 846	/* ip_vs_proto */
 847	#define IP_VS_PROTO_TAB_SIZE	32	/* must be power of 2 */
 848	struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
 849	/* ip_vs_proto_tcp */
 850#ifdef CONFIG_IP_VS_PROTO_TCP
 851	#define	TCP_APP_TAB_BITS	4
 852	#define	TCP_APP_TAB_SIZE	(1 << TCP_APP_TAB_BITS)
 853	#define	TCP_APP_TAB_MASK	(TCP_APP_TAB_SIZE - 1)
 854	struct list_head	tcp_apps[TCP_APP_TAB_SIZE];
 855#endif
 856	/* ip_vs_proto_udp */
 857#ifdef CONFIG_IP_VS_PROTO_UDP
 858	#define	UDP_APP_TAB_BITS	4
 859	#define	UDP_APP_TAB_SIZE	(1 << UDP_APP_TAB_BITS)
 860	#define	UDP_APP_TAB_MASK	(UDP_APP_TAB_SIZE - 1)
 861	struct list_head	udp_apps[UDP_APP_TAB_SIZE];
 862#endif
 863	/* ip_vs_proto_sctp */
 864#ifdef CONFIG_IP_VS_PROTO_SCTP
 865	#define SCTP_APP_TAB_BITS	4
 866	#define SCTP_APP_TAB_SIZE	(1 << SCTP_APP_TAB_BITS)
 867	#define SCTP_APP_TAB_MASK	(SCTP_APP_TAB_SIZE - 1)
 868	/* Hash table for SCTP application incarnations	 */
 869	struct list_head	sctp_apps[SCTP_APP_TAB_SIZE];
 870#endif
 871	/* ip_vs_conn */
 872	atomic_t		conn_count;      /* connection counter */
 873
 874	/* ip_vs_ctl */
 875	struct ip_vs_stats		tot_stats;  /* Statistics & est. */
 876
 877	int			num_services;    /* no of virtual services */
 878	int			num_services6;   /* IPv6 virtual services */
 879
 880	/* Trash for destinations */
 881	struct list_head	dest_trash;
 882	spinlock_t		dest_trash_lock;
 883	struct timer_list	dest_trash_timer; /* expiration timer */
 884	/* Service counters */
 885	atomic_t		ftpsvc_counter;
 886	atomic_t		nullsvc_counter;
 887	atomic_t		conn_out_counter;
 888
 889#ifdef CONFIG_SYSCTL
 890	/* delayed work for expiring no dest connections */
 891	struct delayed_work	expire_nodest_conn_work;
 892	/* 1/rate drop and drop-entry variables */
 893	struct delayed_work	defense_work;   /* Work handler */
 894	int			drop_rate;
 895	int			drop_counter;
 896	int			old_secure_tcp;
 897	atomic_t		dropentry;
 898	/* locks in ctl.c */
 899	spinlock_t		dropentry_lock;  /* drop entry handling */
 900	spinlock_t		droppacket_lock; /* drop packet handling */
 901	spinlock_t		securetcp_lock;  /* state and timeout tables */
 902
 903	/* sys-ctl struct */
 904	struct ctl_table_header	*sysctl_hdr;
 905	struct ctl_table	*sysctl_tbl;
 906#endif
 907
 908	/* sysctl variables */
 909	int			sysctl_amemthresh;
 910	int			sysctl_am_droprate;
 911	int			sysctl_drop_entry;
 912	int			sysctl_drop_packet;
 913	int			sysctl_secure_tcp;
 914#ifdef CONFIG_IP_VS_NFCT
 915	int			sysctl_conntrack;
 916#endif
 917	int			sysctl_snat_reroute;
 918	int			sysctl_sync_ver;
 919	int			sysctl_sync_ports;
 920	int			sysctl_sync_persist_mode;
 921	unsigned long		sysctl_sync_qlen_max;
 922	int			sysctl_sync_sock_size;
 923	int			sysctl_cache_bypass;
 924	int			sysctl_expire_nodest_conn;
 925	int			sysctl_sloppy_tcp;
 926	int			sysctl_sloppy_sctp;
 927	int			sysctl_expire_quiescent_template;
 928	int			sysctl_sync_threshold[2];
 929	unsigned int		sysctl_sync_refresh_period;
 930	int			sysctl_sync_retries;
 931	int			sysctl_nat_icmp_send;
 932	int			sysctl_pmtu_disc;
 933	int			sysctl_backup_only;
 934	int			sysctl_conn_reuse_mode;
 935	int			sysctl_schedule_icmp;
 936	int			sysctl_ignore_tunneled;
 937
 938	/* ip_vs_lblc */
 939	int			sysctl_lblc_expiration;
 940	struct ctl_table_header	*lblc_ctl_header;
 941	struct ctl_table	*lblc_ctl_table;
 942	/* ip_vs_lblcr */
 943	int			sysctl_lblcr_expiration;
 944	struct ctl_table_header	*lblcr_ctl_header;
 945	struct ctl_table	*lblcr_ctl_table;
 946	/* ip_vs_est */
 947	struct list_head	est_list;	/* estimator list */
 948	spinlock_t		est_lock;
 949	struct timer_list	est_timer;	/* Estimation timer */
 950	/* ip_vs_sync */
 951	spinlock_t		sync_lock;
 952	struct ipvs_master_sync_state *ms;
 953	spinlock_t		sync_buff_lock;
 954	struct ip_vs_sync_thread_data *master_tinfo;
 955	struct ip_vs_sync_thread_data *backup_tinfo;
 956	int			threads_mask;
 957	volatile int		sync_state;
 958	struct mutex		sync_mutex;
 959	struct ipvs_sync_daemon_cfg	mcfg;	/* Master Configuration */
 960	struct ipvs_sync_daemon_cfg	bcfg;	/* Backup Configuration */
 961	/* net name space ptr */
 962	struct net		*net;            /* Needed by timer routines */
 963	/* Number of heterogeneous destinations, needed becaus heterogeneous
 964	 * are not supported when synchronization is enabled.
 965	 */
 966	unsigned int		mixed_address_family_dests;
 967	unsigned int		hooks_afmask;	/* &1=AF_INET, &2=AF_INET6 */
 968};
 969
 970#define DEFAULT_SYNC_THRESHOLD	3
 971#define DEFAULT_SYNC_PERIOD	50
 972#define DEFAULT_SYNC_VER	1
 973#define DEFAULT_SLOPPY_TCP	0
 974#define DEFAULT_SLOPPY_SCTP	0
 975#define DEFAULT_SYNC_REFRESH_PERIOD	(0U * HZ)
 976#define DEFAULT_SYNC_RETRIES		0
 977#define IPVS_SYNC_WAKEUP_RATE	8
 978#define IPVS_SYNC_QLEN_MAX	(IPVS_SYNC_WAKEUP_RATE * 4)
 979#define IPVS_SYNC_SEND_DELAY	(HZ / 50)
 980#define IPVS_SYNC_CHECK_PERIOD	HZ
 981#define IPVS_SYNC_FLUSH_TIME	(HZ * 2)
 982#define IPVS_SYNC_PORTS_MAX	(1 << 6)
 983
 984#ifdef CONFIG_SYSCTL
 985
 986static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
 987{
 988	return ipvs->sysctl_sync_threshold[0];
 989}
 990
 991static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
 992{
 993	return READ_ONCE(ipvs->sysctl_sync_threshold[1]);
 994}
 995
 996static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
 997{
 998	return READ_ONCE(ipvs->sysctl_sync_refresh_period);
 999}
1000
1001static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1002{
1003	return ipvs->sysctl_sync_retries;
1004}
1005
1006static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1007{
1008	return ipvs->sysctl_sync_ver;
1009}
1010
1011static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1012{
1013	return ipvs->sysctl_sloppy_tcp;
1014}
1015
1016static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1017{
1018	return ipvs->sysctl_sloppy_sctp;
1019}
1020
1021static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1022{
1023	return READ_ONCE(ipvs->sysctl_sync_ports);
1024}
1025
1026static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1027{
1028	return ipvs->sysctl_sync_persist_mode;
1029}
1030
1031static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1032{
1033	return ipvs->sysctl_sync_qlen_max;
1034}
1035
1036static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1037{
1038	return ipvs->sysctl_sync_sock_size;
1039}
1040
1041static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1042{
1043	return ipvs->sysctl_pmtu_disc;
1044}
1045
1046static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1047{
1048	return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1049	       ipvs->sysctl_backup_only;
1050}
1051
1052static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1053{
1054	return ipvs->sysctl_conn_reuse_mode;
1055}
1056
1057static inline int sysctl_expire_nodest_conn(struct netns_ipvs *ipvs)
1058{
1059	return ipvs->sysctl_expire_nodest_conn;
1060}
1061
1062static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1063{
1064	return ipvs->sysctl_schedule_icmp;
1065}
1066
1067static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1068{
1069	return ipvs->sysctl_ignore_tunneled;
1070}
1071
1072static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1073{
1074	return ipvs->sysctl_cache_bypass;
1075}
1076
1077#else
1078
1079static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1080{
1081	return DEFAULT_SYNC_THRESHOLD;
1082}
1083
1084static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1085{
1086	return DEFAULT_SYNC_PERIOD;
1087}
1088
1089static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1090{
1091	return DEFAULT_SYNC_REFRESH_PERIOD;
1092}
1093
1094static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1095{
1096	return DEFAULT_SYNC_RETRIES & 3;
1097}
1098
1099static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1100{
1101	return DEFAULT_SYNC_VER;
1102}
1103
1104static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1105{
1106	return DEFAULT_SLOPPY_TCP;
1107}
1108
1109static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1110{
1111	return DEFAULT_SLOPPY_SCTP;
1112}
1113
1114static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1115{
1116	return 1;
1117}
1118
1119static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1120{
1121	return 0;
1122}
1123
1124static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
1125{
1126	return IPVS_SYNC_QLEN_MAX;
1127}
1128
1129static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1130{
1131	return 0;
1132}
1133
1134static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1135{
1136	return 1;
1137}
1138
1139static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1140{
1141	return 0;
1142}
1143
1144static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1145{
1146	return 1;
1147}
1148
1149static inline int sysctl_expire_nodest_conn(struct netns_ipvs *ipvs)
1150{
1151	return 0;
1152}
1153
1154static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1155{
1156	return 0;
1157}
1158
1159static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1160{
1161	return 0;
1162}
1163
1164static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1165{
1166	return 0;
1167}
1168
1169#endif
1170
1171/* IPVS core functions
1172 * (from ip_vs_core.c)
1173 */
1174const char *ip_vs_proto_name(unsigned int proto);
1175void ip_vs_init_hash_table(struct list_head *table, int rows);
1176struct ip_vs_conn *ip_vs_new_conn_out(struct ip_vs_service *svc,
1177				      struct ip_vs_dest *dest,
1178				      struct sk_buff *skb,
1179				      const struct ip_vs_iphdr *iph,
1180				      __be16 dport,
1181				      __be16 cport);
1182#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
1183
1184#define IP_VS_APP_TYPE_FTP	1
1185
1186/* ip_vs_conn handling functions
1187 * (from ip_vs_conn.c)
1188 */
1189enum {
1190	IP_VS_DIR_INPUT = 0,
1191	IP_VS_DIR_OUTPUT,
1192	IP_VS_DIR_INPUT_ONLY,
1193	IP_VS_DIR_LAST,
1194};
1195
1196static inline void ip_vs_conn_fill_param(struct netns_ipvs *ipvs, int af, int protocol,
1197					 const union nf_inet_addr *caddr,
1198					 __be16 cport,
1199					 const union nf_inet_addr *vaddr,
1200					 __be16 vport,
1201					 struct ip_vs_conn_param *p)
1202{
1203	p->ipvs = ipvs;
1204	p->af = af;
1205	p->protocol = protocol;
1206	p->caddr = caddr;
1207	p->cport = cport;
1208	p->vaddr = vaddr;
1209	p->vport = vport;
1210	p->pe = NULL;
1211	p->pe_data = NULL;
1212}
1213
1214struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1215struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
1216
1217struct ip_vs_conn * ip_vs_conn_in_get_proto(struct netns_ipvs *ipvs, int af,
1218					    const struct sk_buff *skb,
1219					    const struct ip_vs_iphdr *iph);
1220
1221struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
1222
1223struct ip_vs_conn * ip_vs_conn_out_get_proto(struct netns_ipvs *ipvs, int af,
1224					     const struct sk_buff *skb,
1225					     const struct ip_vs_iphdr *iph);
1226
1227/* Get reference to gain full access to conn.
1228 * By default, RCU read-side critical sections have access only to
1229 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1230 */
1231static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1232{
1233	return refcount_inc_not_zero(&cp->refcnt);
1234}
1235
1236/* put back the conn without restarting its timer */
1237static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1238{
1239	smp_mb__before_atomic();
1240	refcount_dec(&cp->refcnt);
1241}
1242void ip_vs_conn_put(struct ip_vs_conn *cp);
1243void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
1244
1245struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, int dest_af,
1246				  const union nf_inet_addr *daddr,
1247				  __be16 dport, unsigned int flags,
1248				  struct ip_vs_dest *dest, __u32 fwmark);
1249void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
1250
1251const char *ip_vs_state_name(const struct ip_vs_conn *cp);
1252
1253void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
1254int ip_vs_check_template(struct ip_vs_conn *ct, struct ip_vs_dest *cdest);
1255void ip_vs_random_dropentry(struct netns_ipvs *ipvs);
1256int ip_vs_conn_init(void);
1257void ip_vs_conn_cleanup(void);
1258
1259static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1260{
1261	struct ip_vs_conn *ctl_cp = cp->control;
1262	if (!ctl_cp) {
1263		IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1264			      "%s:%d to %s:%d\n",
1265			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1266			      ntohs(cp->cport),
1267			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1268			      ntohs(cp->vport));
1269
1270		return;
1271	}
1272
1273	IP_VS_DBG_BUF(7, "DELeting control for: "
1274		      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1275		      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1276		      ntohs(cp->cport),
1277		      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1278		      ntohs(ctl_cp->cport));
1279
1280	cp->control = NULL;
1281	if (atomic_read(&ctl_cp->n_control) == 0) {
1282		IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1283			      "%s:%d to %s:%d\n",
1284			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1285			      ntohs(cp->cport),
1286			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1287			      ntohs(cp->vport));
1288
1289		return;
1290	}
1291	atomic_dec(&ctl_cp->n_control);
1292}
1293
1294static inline void
1295ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1296{
1297	if (cp->control) {
1298		IP_VS_ERR_BUF("request control ADD for already controlled: "
1299			      "%s:%d to %s:%d\n",
1300			      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1301			      ntohs(cp->cport),
1302			      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1303			      ntohs(cp->vport));
1304
1305		ip_vs_control_del(cp);
1306	}
1307
1308	IP_VS_DBG_BUF(7, "ADDing control for: "
1309		      "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1310		      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1311		      ntohs(cp->cport),
1312		      IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1313		      ntohs(ctl_cp->cport));
1314
1315	cp->control = ctl_cp;
1316	atomic_inc(&ctl_cp->n_control);
1317}
1318
1319/* Mark our template as assured */
1320static inline void
1321ip_vs_control_assure_ct(struct ip_vs_conn *cp)
1322{
1323	struct ip_vs_conn *ct = cp->control;
1324
1325	if (ct && !(ct->state & IP_VS_CTPL_S_ASSURED) &&
1326	    (ct->flags & IP_VS_CONN_F_TEMPLATE))
1327		ct->state |= IP_VS_CTPL_S_ASSURED;
1328}
1329
1330/* IPVS netns init & cleanup functions */
1331int ip_vs_estimator_net_init(struct netns_ipvs *ipvs);
1332int ip_vs_control_net_init(struct netns_ipvs *ipvs);
1333int ip_vs_protocol_net_init(struct netns_ipvs *ipvs);
1334int ip_vs_app_net_init(struct netns_ipvs *ipvs);
1335int ip_vs_conn_net_init(struct netns_ipvs *ipvs);
1336int ip_vs_sync_net_init(struct netns_ipvs *ipvs);
1337void ip_vs_conn_net_cleanup(struct netns_ipvs *ipvs);
1338void ip_vs_app_net_cleanup(struct netns_ipvs *ipvs);
1339void ip_vs_protocol_net_cleanup(struct netns_ipvs *ipvs);
1340void ip_vs_control_net_cleanup(struct netns_ipvs *ipvs);
1341void ip_vs_estimator_net_cleanup(struct netns_ipvs *ipvs);
1342void ip_vs_sync_net_cleanup(struct netns_ipvs *ipvs);
1343void ip_vs_service_nets_cleanup(struct list_head *net_list);
1344
1345/* IPVS application functions
1346 * (from ip_vs_app.c)
1347 */
1348#define IP_VS_APP_MAX_PORTS  8
1349struct ip_vs_app *register_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1350void unregister_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1351int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1352void ip_vs_unbind_app(struct ip_vs_conn *cp);
1353int register_ip_vs_app_inc(struct netns_ipvs *ipvs, struct ip_vs_app *app, __u16 proto,
1354			   __u16 port);
1355int ip_vs_app_inc_get(struct ip_vs_app *inc);
1356void ip_vs_app_inc_put(struct ip_vs_app *inc);
1357
1358int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb,
1359		      struct ip_vs_iphdr *ipvsh);
1360int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb,
1361		     struct ip_vs_iphdr *ipvsh);
1362
1363int register_ip_vs_pe(struct ip_vs_pe *pe);
1364int unregister_ip_vs_pe(struct ip_vs_pe *pe);
1365struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
1366struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
1367
1368/* Use a #define to avoid all of module.h just for these trivial ops */
1369#define ip_vs_pe_get(pe)			\
1370	if (pe && pe->module)			\
1371		__module_get(pe->module);
1372
1373#define ip_vs_pe_put(pe)			\
1374	if (pe && pe->module)			\
1375		module_put(pe->module);
1376
1377/* IPVS protocol functions (from ip_vs_proto.c) */
1378int ip_vs_protocol_init(void);
1379void ip_vs_protocol_cleanup(void);
1380void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1381int *ip_vs_create_timeout_table(int *table, int size);
1382void ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1383			       const struct sk_buff *skb, int offset,
1384			       const char *msg);
1385
1386extern struct ip_vs_protocol ip_vs_protocol_tcp;
1387extern struct ip_vs_protocol ip_vs_protocol_udp;
1388extern struct ip_vs_protocol ip_vs_protocol_icmp;
1389extern struct ip_vs_protocol ip_vs_protocol_esp;
1390extern struct ip_vs_protocol ip_vs_protocol_ah;
1391extern struct ip_vs_protocol ip_vs_protocol_sctp;
1392
1393/* Registering/unregistering scheduler functions
1394 * (from ip_vs_sched.c)
1395 */
1396int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1397int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1398int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1399			 struct ip_vs_scheduler *scheduler);
1400void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1401			    struct ip_vs_scheduler *sched);
1402struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1403void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1404struct ip_vs_conn *
1405ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
1406	       struct ip_vs_proto_data *pd, int *ignored,
1407	       struct ip_vs_iphdr *iph);
1408int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1409		struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
1410
1411void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
1412
1413/* IPVS control data and functions (from ip_vs_ctl.c) */
1414extern struct ip_vs_stats ip_vs_stats;
1415extern int sysctl_ip_vs_sync_ver;
1416
1417struct ip_vs_service *
1418ip_vs_service_find(struct netns_ipvs *ipvs, int af, __u32 fwmark, __u16 protocol,
1419		  const union nf_inet_addr *vaddr, __be16 vport);
1420
1421bool ip_vs_has_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1422			    const union nf_inet_addr *daddr, __be16 dport);
1423
1424struct ip_vs_dest *
1425ip_vs_find_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1426			const union nf_inet_addr *daddr, __be16 dport);
1427struct ip_vs_dest *ip_vs_find_tunnel(struct netns_ipvs *ipvs, int af,
1428				     const union nf_inet_addr *daddr,
1429				     __be16 tun_port);
1430
1431int ip_vs_use_count_inc(void);
1432void ip_vs_use_count_dec(void);
1433int ip_vs_register_nl_ioctl(void);
1434void ip_vs_unregister_nl_ioctl(void);
1435int ip_vs_control_init(void);
1436void ip_vs_control_cleanup(void);
1437struct ip_vs_dest *
1438ip_vs_find_dest(struct netns_ipvs *ipvs, int svc_af, int dest_af,
1439		const union nf_inet_addr *daddr, __be16 dport,
1440		const union nf_inet_addr *vaddr, __be16 vport,
1441		__u16 protocol, __u32 fwmark, __u32 flags);
1442void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
1443
1444static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1445{
1446	refcount_inc(&dest->refcnt);
1447}
1448
1449static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1450{
1451	smp_mb__before_atomic();
1452	refcount_dec(&dest->refcnt);
1453}
1454
1455static inline void ip_vs_dest_put_and_free(struct ip_vs_dest *dest)
1456{
1457	if (refcount_dec_and_test(&dest->refcnt))
1458		kfree(dest);
1459}
1460
1461/* IPVS sync daemon data and function prototypes
1462 * (from ip_vs_sync.c)
1463 */
1464int start_sync_thread(struct netns_ipvs *ipvs, struct ipvs_sync_daemon_cfg *cfg,
1465		      int state);
1466int stop_sync_thread(struct netns_ipvs *ipvs, int state);
1467void ip_vs_sync_conn(struct netns_ipvs *ipvs, struct ip_vs_conn *cp, int pkts);
1468
1469/* IPVS rate estimator prototypes (from ip_vs_est.c) */
1470void ip_vs_start_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1471void ip_vs_stop_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1472void ip_vs_zero_estimator(struct ip_vs_stats *stats);
1473void ip_vs_read_estimator(struct ip_vs_kstats *dst, struct ip_vs_stats *stats);
1474
1475/* Various IPVS packet transmitters (from ip_vs_xmit.c) */
1476int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1477		    struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1478int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1479		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1480int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1481		   struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1482int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1483		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1484int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1485		  struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1486int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1487		    struct ip_vs_protocol *pp, int offset,
1488		    unsigned int hooknum, struct ip_vs_iphdr *iph);
1489void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
1490
1491#ifdef CONFIG_IP_VS_IPV6
1492int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1493			 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1494int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1495		      struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1496int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1497			 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1498int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1499		     struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1500int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1501		       struct ip_vs_protocol *pp, int offset,
1502		       unsigned int hooknum, struct ip_vs_iphdr *iph);
1503#endif
1504
1505#ifdef CONFIG_SYSCTL
1506/* This is a simple mechanism to ignore packets when
1507 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1508 * we start to drop 1/rate of the packets
1509 */
1510static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
1511{
1512	if (!ipvs->drop_rate)
1513		return 0;
1514	if (--ipvs->drop_counter > 0)
1515		return 0;
1516	ipvs->drop_counter = ipvs->drop_rate;
1517	return 1;
1518}
1519#else
1520static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1521#endif
1522
1523#ifdef CONFIG_SYSCTL
1524/* Enqueue delayed work for expiring no dest connections
1525 * Only run when sysctl_expire_nodest=1
1526 */
1527static inline void ip_vs_enqueue_expire_nodest_conns(struct netns_ipvs *ipvs)
1528{
1529	if (sysctl_expire_nodest_conn(ipvs))
1530		queue_delayed_work(system_long_wq,
1531				   &ipvs->expire_nodest_conn_work, 1);
1532}
1533
1534void ip_vs_expire_nodest_conn_flush(struct netns_ipvs *ipvs);
1535#else
1536static inline void ip_vs_enqueue_expire_nodest_conns(struct netns_ipvs *ipvs) {}
1537#endif
1538
1539#define IP_VS_DFWD_METHOD(dest) (atomic_read(&(dest)->conn_flags) & \
1540				 IP_VS_CONN_F_FWD_MASK)
1541
1542/* ip_vs_fwd_tag returns the forwarding tag of the connection */
1543#define IP_VS_FWD_METHOD(cp)  (cp->flags & IP_VS_CONN_F_FWD_MASK)
1544
1545static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
1546{
1547	char fwd;
1548
1549	switch (IP_VS_FWD_METHOD(cp)) {
1550	case IP_VS_CONN_F_MASQ:
1551		fwd = 'M'; break;
1552	case IP_VS_CONN_F_LOCALNODE:
1553		fwd = 'L'; break;
1554	case IP_VS_CONN_F_TUNNEL:
1555		fwd = 'T'; break;
1556	case IP_VS_CONN_F_DROUTE:
1557		fwd = 'R'; break;
1558	case IP_VS_CONN_F_BYPASS:
1559		fwd = 'B'; break;
1560	default:
1561		fwd = '?'; break;
1562	}
1563	return fwd;
1564}
1565
1566void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1567		    struct ip_vs_conn *cp, int dir);
1568
1569#ifdef CONFIG_IP_VS_IPV6
1570void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1571		       struct ip_vs_conn *cp, int dir);
1572#endif
1573
1574__sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
1575
1576static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
1577{
1578	__be32 diff[2] = { ~old, new };
1579
1580	return csum_partial(diff, sizeof(diff), oldsum);
1581}
1582
1583#ifdef CONFIG_IP_VS_IPV6
1584static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1585					__wsum oldsum)
1586{
1587	__be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1588			    new[3],  new[2],  new[1],  new[0] };
1589
1590	return csum_partial(diff, sizeof(diff), oldsum);
1591}
1592#endif
1593
1594static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1595{
1596	__be16 diff[2] = { ~old, new };
1597
1598	return csum_partial(diff, sizeof(diff), oldsum);
1599}
1600
1601/* Forget current conntrack (unconfirmed) and attach notrack entry */
1602static inline void ip_vs_notrack(struct sk_buff *skb)
1603{
1604#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1605	enum ip_conntrack_info ctinfo;
1606	struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
1607
1608	if (ct) {
1609		nf_conntrack_put(&ct->ct_general);
1610		nf_ct_set(skb, NULL, IP_CT_UNTRACKED);
1611	}
1612#endif
1613}
1614
1615#ifdef CONFIG_IP_VS_NFCT
1616/* Netfilter connection tracking
1617 * (from ip_vs_nfct.c)
1618 */
1619static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1620{
1621#ifdef CONFIG_SYSCTL
1622	return ipvs->sysctl_conntrack;
1623#else
1624	return 0;
1625#endif
1626}
1627
1628void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1629			    int outin);
1630int ip_vs_confirm_conntrack(struct sk_buff *skb);
1631void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1632			       struct ip_vs_conn *cp, u_int8_t proto,
1633			       const __be16 port, int from_rs);
1634void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
1635
1636#else
1637
1638static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
1639{
1640	return 0;
1641}
1642
1643static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1644					  struct ip_vs_conn *cp, int outin)
1645{
1646}
1647
1648static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
1649{
1650	return NF_ACCEPT;
1651}
1652
1653static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1654{
1655}
1656#endif /* CONFIG_IP_VS_NFCT */
1657
1658/* Using old conntrack that can not be redirected to another real server? */
1659static inline bool ip_vs_conn_uses_old_conntrack(struct ip_vs_conn *cp,
1660						 struct sk_buff *skb)
1661{
1662#ifdef CONFIG_IP_VS_NFCT
1663	enum ip_conntrack_info ctinfo;
1664	struct nf_conn *ct;
1665
 
 
1666	ct = nf_ct_get(skb, &ctinfo);
1667	if (ct && nf_ct_is_confirmed(ct))
1668		return true;
1669#endif
1670	return false;
1671}
1672
1673static inline int ip_vs_register_conntrack(struct ip_vs_service *svc)
1674{
1675#if IS_ENABLED(CONFIG_NF_CONNTRACK)
1676	int afmask = (svc->af == AF_INET6) ? 2 : 1;
1677	int ret = 0;
1678
1679	if (!(svc->conntrack_afmask & afmask)) {
1680		ret = nf_ct_netns_get(svc->ipvs->net, svc->af);
1681		if (ret >= 0)
1682			svc->conntrack_afmask |= afmask;
1683	}
1684	return ret;
1685#else
1686	return 0;
1687#endif
1688}
1689
1690static inline void ip_vs_unregister_conntrack(struct ip_vs_service *svc)
1691{
1692#if IS_ENABLED(CONFIG_NF_CONNTRACK)
1693	int afmask = (svc->af == AF_INET6) ? 2 : 1;
1694
1695	if (svc->conntrack_afmask & afmask) {
1696		nf_ct_netns_put(svc->ipvs->net, svc->af);
1697		svc->conntrack_afmask &= ~afmask;
1698	}
1699#endif
1700}
1701
1702int ip_vs_register_hooks(struct netns_ipvs *ipvs, unsigned int af);
1703void ip_vs_unregister_hooks(struct netns_ipvs *ipvs, unsigned int af);
1704
1705static inline int
1706ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1707{
1708	/* We think the overhead of processing active connections is 256
1709	 * times higher than that of inactive connections in average. (This
1710	 * 256 times might not be accurate, we will change it later) We
1711	 * use the following formula to estimate the overhead now:
1712	 *		  dest->activeconns*256 + dest->inactconns
1713	 */
1714	return (atomic_read(&dest->activeconns) << 8) +
1715		atomic_read(&dest->inactconns);
1716}
1717
1718#endif	/* _NET_IP_VS_H */