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