blob: ff68cf288f9bf3a3ac18d696e215980e4fee54f5 [file] [log] [blame]
Greg Kroah-Hartmanb2441312017-11-01 15:07:57 +01001/* SPDX-License-Identifier: GPL-2.0 */
Simon Horman07dcc682014-09-30 10:50:06 +09002/* IP Virtual Server
3 * data structure and functionality definitions
Linus Torvalds1da177e2005-04-16 15:20:36 -07004 */
5
Julius Volzbc4768e2008-07-31 20:45:24 -07006#ifndef _NET_IP_VS_H
7#define _NET_IP_VS_H
Linus Torvalds1da177e2005-04-16 15:20:36 -07008
Julius Volzbc4768e2008-07-31 20:45:24 -07009#include <linux/ip_vs.h> /* definitions shared with userland */
Linus Torvalds1da177e2005-04-16 15:20:36 -070010
Julius Volzbc4768e2008-07-31 20:45:24 -070011#include <asm/types.h> /* for __uXX types */
12
Linus Torvalds1da177e2005-04-16 15:20:36 -070013#include <linux/list.h> /* for struct list_head */
14#include <linux/spinlock.h> /* for struct rwlock_t */
Simon Horman07dcc682014-09-30 10:50:06 +090015#include <linux/atomic.h> /* for struct atomic_t */
Reshetova, Elenab54ab922017-03-16 10:03:34 +020016#include <linux/refcount.h> /* for struct refcount_t */
17
Linus Torvalds1da177e2005-04-16 15:20:36 -070018#include <linux/compiler.h>
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -020019#include <linux/timer.h>
Paul Gortmaker187f1882011-11-23 20:12:59 -050020#include <linux/bug.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070021
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -020022#include <net/checksum.h>
Julius Volze7ade462008-09-02 15:55:33 +020023#include <linux/netfilter.h> /* for union nf_inet_addr */
KOVACS Krisztian1668e012008-10-01 07:33:10 -070024#include <linux/ip.h>
Julius Volze7ade462008-09-02 15:55:33 +020025#include <linux/ipv6.h> /* for struct ipv6hdr */
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +000026#include <net/ipv6.h>
Jesper Dangaard Brouer54d83ef2012-10-22 21:22:45 +020027#if IS_ENABLED(CONFIG_IP_VS_IPV6)
Jesper Dangaard Brouer63dca2c2012-09-26 14:06:41 +020028#include <linux/netfilter_ipv6/ip6_tables.h>
29#endif
Jesper Dangaard Brouera638e512012-09-26 14:06:20 +020030#if IS_ENABLED(CONFIG_NF_CONNTRACK)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +020031#include <net/netfilter/nf_conntrack.h>
32#endif
Hans Schillstrom61b1ab42011-01-03 14:44:42 +010033#include <net/net_namespace.h> /* Netw namespace */
34
Alex Gartrell4fd9bee2015-08-26 09:40:29 -070035#define IP_VS_HDR_INVERSE 1
36#define IP_VS_HDR_ICMP 2
37
Simon Horman07dcc682014-09-30 10:50:06 +090038/* Generic access of ipvs struct */
Hans Schillstrom61b1ab42011-01-03 14:44:42 +010039static inline struct netns_ipvs *net_ipvs(struct net* net)
40{
41 return net->ipvs;
42}
Simon Horman07dcc682014-09-30 10:50:06 +090043
Simon Horman07dcc682014-09-30 10:50:06 +090044/* This one needed for single_open_net since net is stored directly in
Lucas De Marchi25985ed2011-03-30 22:57:33 -030045 * private not as a struct i.e. seq_file_net can't be used.
Hans Schillstromb17fc992011-01-03 14:44:56 +010046 */
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}
Catalin(ux) M. BOIE6f7edb42010-01-05 05:50:24 +010055
56/* Connections' size value needed by ip_vs_ctl.c */
57extern int ip_vs_conn_tab_size;
58
Julius Volz64aae3c2008-09-02 15:55:34 +020059struct ip_vs_iphdr {
Alex Gartrell4fd9bee2015-08-26 09:40:29 -070060 int hdr_flags; /* ipvs flags */
Alex Gartrellb0e010c2015-08-26 09:40:28 -070061 __u32 off; /* Where IP or IPv4 header starts */
Jesper Dangaard Brouer63dca2c2012-09-26 14:06:41 +020062 __u32 len; /* IPv4 simply where L4 starts
Simon Horman07dcc682014-09-30 10:50:06 +090063 * IPv6 where L4 Transport Header starts */
Jesper Dangaard Brouer63dca2c2012-09-26 14:06:41 +020064 __u16 fragoffs; /* IPv6 fragment offset, 0 if first frag (or not frag)*/
65 __s16 protocol;
66 __s32 flags;
Julius Volz64aae3c2008-09-02 15:55:34 +020067 union nf_inet_addr saddr;
68 union nf_inet_addr daddr;
69};
70
Jesper Dangaard Brouer2f747132012-09-26 14:06:59 +020071static inline void *frag_safe_skb_hp(const struct sk_buff *skb, int offset,
72 int len, void *buffer,
73 const struct ip_vs_iphdr *ipvsh)
74{
75 return skb_header_pointer(skb, offset, len, buffer);
76}
Jesper Dangaard Brouer2f747132012-09-26 14:06:59 +020077
Jesper Dangaard Brouer63dca2c2012-09-26 14:06:41 +020078/* This function handles filling *ip_vs_iphdr, both for IPv4 and IPv6.
79 * IPv6 requires some extra work, as finding proper header position,
80 * depend on the IPv6 extension headers.
81 */
Alex Gartrellb0e010c2015-08-26 09:40:28 -070082static inline int
83ip_vs_fill_iph_skb_off(int af, const struct sk_buff *skb, int offset,
Alex Gartrell4fd9bee2015-08-26 09:40:29 -070084 int hdr_flags, struct ip_vs_iphdr *iphdr)
Julius Volz64aae3c2008-09-02 15:55:34 +020085{
Alex Gartrell4fd9bee2015-08-26 09:40:29 -070086 iphdr->hdr_flags = hdr_flags;
Alex Gartrellb0e010c2015-08-26 09:40:28 -070087 iphdr->off = offset;
Alex Gartrell4fd9bee2015-08-26 09:40:29 -070088
Julius Volz64aae3c2008-09-02 15:55:34 +020089#ifdef CONFIG_IP_VS_IPV6
90 if (af == AF_INET6) {
Alex Gartrellb0e010c2015-08-26 09:40:28 -070091 struct ipv6hdr _iph;
92 const struct ipv6hdr *iph = skb_header_pointer(
93 skb, offset, sizeof(_iph), &_iph);
94 if (!iph)
95 return 0;
96
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +000097 iphdr->saddr.in6 = iph->saddr;
98 iphdr->daddr.in6 = iph->daddr;
Jiri Pirko6aafeef2013-11-06 17:52:20 +010099 /* ipv6_find_hdr() updates len, flags */
Alex Gartrellb0e010c2015-08-26 09:40:28 -0700100 iphdr->len = offset;
Jesper Dangaard Brouer63dca2c2012-09-26 14:06:41 +0200101 iphdr->flags = 0;
102 iphdr->protocol = ipv6_find_hdr(skb, &iphdr->len, -1,
103 &iphdr->fragoffs,
104 &iphdr->flags);
Alex Gartrellb0e010c2015-08-26 09:40:28 -0700105 if (iphdr->protocol < 0)
106 return 0;
Julius Volz64aae3c2008-09-02 15:55:34 +0200107 } else
108#endif
109 {
Alex Gartrellb0e010c2015-08-26 09:40:28 -0700110 struct iphdr _iph;
111 const struct iphdr *iph = skb_header_pointer(
112 skb, offset, sizeof(_iph), &_iph);
113 if (!iph)
114 return 0;
115
116 iphdr->len = offset + iph->ihl * 4;
Jesper Dangaard Brouer63dca2c2012-09-26 14:06:41 +0200117 iphdr->fragoffs = 0;
118 iphdr->protocol = iph->protocol;
119 iphdr->saddr.ip = iph->saddr;
120 iphdr->daddr.ip = iph->daddr;
121 }
Alex Gartrellb0e010c2015-08-26 09:40:28 -0700122
123 return 1;
124}
125
126static inline int
Alex Gartrell4fd9bee2015-08-26 09:40:29 -0700127ip_vs_fill_iph_skb_icmp(int af, const struct sk_buff *skb, int offset,
128 bool inverse, struct ip_vs_iphdr *iphdr)
Alex Gartrellb0e010c2015-08-26 09:40:28 -0700129{
Alex Gartrell4fd9bee2015-08-26 09:40:29 -0700130 int hdr_flags = IP_VS_HDR_ICMP;
131
132 if (inverse)
133 hdr_flags |= IP_VS_HDR_INVERSE;
134
135 return ip_vs_fill_iph_skb_off(af, skb, offset, hdr_flags, iphdr);
136}
137
138static inline int
139ip_vs_fill_iph_skb(int af, const struct sk_buff *skb, bool inverse,
140 struct ip_vs_iphdr *iphdr)
141{
142 int hdr_flags = 0;
143
144 if (inverse)
145 hdr_flags |= IP_VS_HDR_INVERSE;
146
147 return ip_vs_fill_iph_skb_off(af, skb, skb_network_offset(skb),
148 hdr_flags, iphdr);
149}
150
151static inline bool
152ip_vs_iph_inverse(const struct ip_vs_iphdr *iph)
153{
154 return !!(iph->hdr_flags & IP_VS_HDR_INVERSE);
155}
156
157static inline bool
158ip_vs_iph_icmp(const struct ip_vs_iphdr *iph)
159{
160 return !!(iph->hdr_flags & IP_VS_HDR_ICMP);
Jesper Dangaard Brouer63dca2c2012-09-26 14:06:41 +0200161}
162
Julius Volz64aae3c2008-09-02 15:55:34 +0200163static inline void ip_vs_addr_copy(int af, union nf_inet_addr *dst,
164 const union nf_inet_addr *src)
165{
166#ifdef CONFIG_IP_VS_IPV6
167 if (af == AF_INET6)
Alexey Dobriyan4e3fd7a2011-11-21 03:39:03 +0000168 dst->in6 = src->in6;
Julius Volz64aae3c2008-09-02 15:55:34 +0200169 else
170#endif
171 dst->ip = src->ip;
172}
173
Julian Anastasov9a05475c2013-03-21 11:58:12 +0200174static inline void ip_vs_addr_set(int af, union nf_inet_addr *dst,
175 const union nf_inet_addr *src)
176{
177#ifdef CONFIG_IP_VS_IPV6
178 if (af == AF_INET6) {
179 dst->in6 = src->in6;
180 return;
181 }
182#endif
183 dst->ip = src->ip;
184 dst->all[1] = 0;
185 dst->all[2] = 0;
186 dst->all[3] = 0;
187}
188
Julius Volz64aae3c2008-09-02 15:55:34 +0200189static inline int ip_vs_addr_equal(int af, const union nf_inet_addr *a,
190 const union nf_inet_addr *b)
191{
192#ifdef CONFIG_IP_VS_IPV6
193 if (af == AF_INET6)
194 return ipv6_addr_equal(&a->in6, &b->in6);
195#endif
196 return a->ip == b->ip;
197}
198
Linus Torvalds1da177e2005-04-16 15:20:36 -0700199#ifdef CONFIG_IP_VS_DEBUG
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -0200200#include <linux/net.h>
201
Joe Perches5c3a0fd2013-09-21 10:22:42 -0700202int ip_vs_get_debug_level(void);
Julius Volzc842a3a2008-09-02 15:55:35 +0200203
204static inline const char *ip_vs_dbg_addr(int af, char *buf, size_t buf_len,
205 const union nf_inet_addr *addr,
206 int *idx)
207{
208 int len;
209#ifdef CONFIG_IP_VS_IPV6
210 if (af == AF_INET6)
Jesper Dangaard Brouer120b9c12012-09-26 14:05:53 +0200211 len = snprintf(&buf[*idx], buf_len - *idx, "[%pI6c]",
Harvey Harrison0c6ce782008-10-28 16:09:23 -0700212 &addr->in6) + 1;
Julius Volzc842a3a2008-09-02 15:55:35 +0200213 else
214#endif
Harvey Harrison3685f25d2008-10-31 00:56:49 -0700215 len = snprintf(&buf[*idx], buf_len - *idx, "%pI4",
216 &addr->ip) + 1;
Julius Volzc842a3a2008-09-02 15:55:35 +0200217
218 *idx += len;
219 BUG_ON(*idx > buf_len + 1);
220 return &buf[*idx - len];
221}
222
Hannes Eder9aada7a2009-07-30 14:29:44 -0700223#define IP_VS_DBG_BUF(level, msg, ...) \
224 do { \
225 char ip_vs_dbg_buf[160]; \
226 int ip_vs_dbg_idx = 0; \
227 if (level <= ip_vs_get_debug_level()) \
228 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
229 } while (0)
230#define IP_VS_ERR_BUF(msg...) \
231 do { \
232 char ip_vs_dbg_buf[160]; \
233 int ip_vs_dbg_idx = 0; \
234 pr_err(msg); \
235 } while (0)
Julius Volzc842a3a2008-09-02 15:55:35 +0200236
237/* Only use from within IP_VS_DBG_BUF() or IP_VS_ERR_BUF macros */
Hannes Eder9aada7a2009-07-30 14:29:44 -0700238#define IP_VS_DBG_ADDR(af, addr) \
239 ip_vs_dbg_addr(af, ip_vs_dbg_buf, \
240 sizeof(ip_vs_dbg_buf), addr, \
241 &ip_vs_dbg_idx)
Julius Volzc842a3a2008-09-02 15:55:35 +0200242
Hannes Eder9aada7a2009-07-30 14:29:44 -0700243#define IP_VS_DBG(level, msg, ...) \
244 do { \
245 if (level <= ip_vs_get_debug_level()) \
246 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
247 } while (0)
248#define IP_VS_DBG_RL(msg, ...) \
249 do { \
250 if (net_ratelimit()) \
251 printk(KERN_DEBUG pr_fmt(msg), ##__VA_ARGS__); \
252 } while (0)
Julian Anastasov0d796412010-10-17 16:46:17 +0300253#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700254 do { \
255 if (level <= ip_vs_get_debug_level()) \
Julian Anastasov0d796412010-10-17 16:46:17 +0300256 pp->debug_packet(af, pp, skb, ofs, msg); \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700257 } while (0)
Julian Anastasov0d796412010-10-17 16:46:17 +0300258#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700259 do { \
260 if (level <= ip_vs_get_debug_level() && \
261 net_ratelimit()) \
Julian Anastasov0d796412010-10-17 16:46:17 +0300262 pp->debug_packet(af, pp, skb, ofs, msg); \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700263 } while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700264#else /* NO DEBUGGING at ALL */
Julius Volzc842a3a2008-09-02 15:55:35 +0200265#define IP_VS_DBG_BUF(level, msg...) do {} while (0)
266#define IP_VS_ERR_BUF(msg...) do {} while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700267#define IP_VS_DBG(level, msg...) do {} while (0)
268#define IP_VS_DBG_RL(msg...) do {} while (0)
Julian Anastasov0d796412010-10-17 16:46:17 +0300269#define IP_VS_DBG_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
270#define IP_VS_DBG_RL_PKT(level, af, pp, skb, ofs, msg) do {} while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700271#endif
272
273#define IP_VS_BUG() BUG()
Hannes Eder1e3e2382009-08-02 11:05:41 +0000274#define IP_VS_ERR_RL(msg, ...) \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700275 do { \
276 if (net_ratelimit()) \
Hannes Eder1e3e2382009-08-02 11:05:41 +0000277 pr_err(msg, ##__VA_ARGS__); \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700278 } while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700279
280#ifdef CONFIG_IP_VS_DEBUG
281#define EnterFunction(level) \
Hannes Eder9aada7a2009-07-30 14:29:44 -0700282 do { \
283 if (level <= ip_vs_get_debug_level()) \
284 printk(KERN_DEBUG \
285 pr_fmt("Enter: %s, %s line %i\n"), \
286 __func__, __FILE__, __LINE__); \
287 } while (0)
288#define LeaveFunction(level) \
289 do { \
290 if (level <= ip_vs_get_debug_level()) \
291 printk(KERN_DEBUG \
292 pr_fmt("Leave: %s, %s line %i\n"), \
293 __func__, __FILE__, __LINE__); \
294 } while (0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700295#else
296#define EnterFunction(level) do {} while (0)
297#define LeaveFunction(level) do {} while (0)
298#endif
299
Simon Horman07dcc682014-09-30 10:50:06 +0900300/* The port number of FTP service (in network order). */
Harvey Harrisonf3a7c662009-02-14 22:58:35 -0800301#define FTPPORT cpu_to_be16(21)
302#define FTPDATA cpu_to_be16(20)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700303
Simon Horman07dcc682014-09-30 10:50:06 +0900304/* TCP State Values */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305enum {
306 IP_VS_TCP_S_NONE = 0,
307 IP_VS_TCP_S_ESTABLISHED,
308 IP_VS_TCP_S_SYN_SENT,
309 IP_VS_TCP_S_SYN_RECV,
310 IP_VS_TCP_S_FIN_WAIT,
311 IP_VS_TCP_S_TIME_WAIT,
312 IP_VS_TCP_S_CLOSE,
313 IP_VS_TCP_S_CLOSE_WAIT,
314 IP_VS_TCP_S_LAST_ACK,
315 IP_VS_TCP_S_LISTEN,
316 IP_VS_TCP_S_SYNACK,
317 IP_VS_TCP_S_LAST
318};
319
Simon Horman07dcc682014-09-30 10:50:06 +0900320/* UDP State Values */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321enum {
322 IP_VS_UDP_S_NORMAL,
323 IP_VS_UDP_S_LAST,
324};
325
Simon Horman07dcc682014-09-30 10:50:06 +0900326/* ICMP State Values */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700327enum {
328 IP_VS_ICMP_S_NORMAL,
329 IP_VS_ICMP_S_LAST,
330};
331
Simon Horman07dcc682014-09-30 10:50:06 +0900332/* SCTP State Values */
Venkata Mohan Reddy2906f662010-02-18 12:31:05 +0100333enum ip_vs_sctp_states {
334 IP_VS_SCTP_S_NONE,
Julian Anastasov61e7c422013-06-18 10:08:07 +0300335 IP_VS_SCTP_S_INIT1,
336 IP_VS_SCTP_S_INIT,
337 IP_VS_SCTP_S_COOKIE_SENT,
338 IP_VS_SCTP_S_COOKIE_REPLIED,
339 IP_VS_SCTP_S_COOKIE_WAIT,
340 IP_VS_SCTP_S_COOKIE,
341 IP_VS_SCTP_S_COOKIE_ECHOED,
Venkata Mohan Reddy2906f662010-02-18 12:31:05 +0100342 IP_VS_SCTP_S_ESTABLISHED,
Julian Anastasov61e7c422013-06-18 10:08:07 +0300343 IP_VS_SCTP_S_SHUTDOWN_SENT,
344 IP_VS_SCTP_S_SHUTDOWN_RECEIVED,
345 IP_VS_SCTP_S_SHUTDOWN_ACK_SENT,
346 IP_VS_SCTP_S_REJECTED,
Venkata Mohan Reddy2906f662010-02-18 12:31:05 +0100347 IP_VS_SCTP_S_CLOSED,
348 IP_VS_SCTP_S_LAST
349};
350
Simon Horman07dcc682014-09-30 10:50:06 +0900351/* Delta sequence info structure
352 * Each ip_vs_conn has 2 (output AND input seq. changes).
353 * Only used in the VS/NAT.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354 */
355struct ip_vs_seq {
356 __u32 init_seq; /* Add delta from this seq */
357 __u32 delta; /* Delta in sequence numbers */
358 __u32 previous_delta; /* Delta in sequence numbers
Simon Horman07dcc682014-09-30 10:50:06 +0900359 * before last resized pkt */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700360};
361
Simon Horman07dcc682014-09-30 10:50:06 +0900362/* counters per cpu */
Hans Schillstromb17fc992011-01-03 14:44:56 +0100363struct ip_vs_counters {
Julian Anastasovcd67cd52015-02-06 09:44:44 +0200364 __u64 conns; /* connections scheduled */
365 __u64 inpkts; /* incoming packets */
366 __u64 outpkts; /* outgoing packets */
Hans Schillstromb17fc992011-01-03 14:44:56 +0100367 __u64 inbytes; /* incoming bytes */
368 __u64 outbytes; /* outgoing bytes */
369};
Simon Horman07dcc682014-09-30 10:50:06 +0900370/* Stats per cpu */
Hans Schillstromb17fc992011-01-03 14:44:56 +0100371struct ip_vs_cpu_stats {
Julian Anastasovcd67cd52015-02-06 09:44:44 +0200372 struct ip_vs_counters cnt;
Hans Schillstromb17fc992011-01-03 14:44:56 +0100373 struct u64_stats_sync syncp;
374};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700375
Simon Horman07dcc682014-09-30 10:50:06 +0900376/* IPVS statistics objects */
Sven Wegener3a14a312008-08-10 18:24:41 +0000377struct ip_vs_estimator {
378 struct list_head list;
379
380 u64 last_inbytes;
381 u64 last_outbytes;
Julian Anastasovcd67cd52015-02-06 09:44:44 +0200382 u64 last_conns;
383 u64 last_inpkts;
384 u64 last_outpkts;
Sven Wegener3a14a312008-08-10 18:24:41 +0000385
Julian Anastasovcd67cd52015-02-06 09:44:44 +0200386 u64 cps;
387 u64 inpps;
388 u64 outpps;
389 u64 inbps;
390 u64 outbps;
391};
392
393/*
394 * IPVS statistics object, 64-bit kernel version of struct ip_vs_stats_user
395 */
396struct ip_vs_kstats {
397 u64 conns; /* connections scheduled */
398 u64 inpkts; /* incoming packets */
399 u64 outpkts; /* outgoing packets */
400 u64 inbytes; /* incoming bytes */
401 u64 outbytes; /* outgoing bytes */
402
403 u64 cps; /* current connection rate */
404 u64 inpps; /* current in packet rate */
405 u64 outpps; /* current out packet rate */
406 u64 inbps; /* current in byte rate */
407 u64 outbps; /* current out byte rate */
Sven Wegener3a14a312008-08-10 18:24:41 +0000408};
409
Eric Dumazetfd2c3ef2009-11-03 03:26:03 +0000410struct ip_vs_stats {
Julian Anastasovcd67cd52015-02-06 09:44:44 +0200411 struct ip_vs_kstats kstats; /* kernel statistics */
Sven Wegenere9c0ce22008-09-08 13:39:04 +0200412 struct ip_vs_estimator est; /* estimator */
Julian Anastasovb962abd2013-03-09 23:25:08 +0200413 struct ip_vs_cpu_stats __percpu *cpustats; /* per cpu counters */
Hans Schillstromb17fc992011-01-03 14:44:56 +0100414 spinlock_t lock; /* spin lock */
Julian Anastasovcd67cd52015-02-06 09:44:44 +0200415 struct ip_vs_kstats kstats0; /* reset values */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700416};
417
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -0200418struct dst_entry;
419struct iphdr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700420struct ip_vs_conn;
421struct ip_vs_app;
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -0200422struct sk_buff;
Hans Schillstrom252c6412011-01-03 14:44:46 +0100423struct ip_vs_proto_data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700424
425struct ip_vs_protocol {
426 struct ip_vs_protocol *next;
427 char *name;
Julian Anastasov2ad17de2008-04-29 03:21:23 -0700428 u16 protocol;
429 u16 num_states;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700430 int dont_defrag;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700431
432 void (*init)(struct ip_vs_protocol *pp);
433
434 void (*exit)(struct ip_vs_protocol *pp);
435
Eric W. Biederman1281a9c22015-09-21 13:02:36 -0500436 int (*init_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
Hans Schillstrom252c6412011-01-03 14:44:46 +0100437
Eric W. Biederman1281a9c22015-09-21 13:02:36 -0500438 void (*exit_netns)(struct netns_ipvs *ipvs, struct ip_vs_proto_data *pd);
Hans Schillstrom252c6412011-01-03 14:44:46 +0100439
Eric W. Biedermand8f44c32015-09-21 13:02:43 -0500440 int (*conn_schedule)(struct netns_ipvs *ipvs,
441 int af, struct sk_buff *skb,
Hans Schillstrom93304192011-01-03 14:44:51 +0100442 struct ip_vs_proto_data *pd,
Jesper Dangaard Brouerd4383f02012-09-26 14:07:17 +0200443 int *verdict, struct ip_vs_conn **cpp,
444 struct ip_vs_iphdr *iph);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700445
446 struct ip_vs_conn *
Eric W. Biedermanab161972015-09-21 13:02:38 -0500447 (*conn_in_get)(struct netns_ipvs *ipvs,
448 int af,
Julius Volz51ef3482008-09-02 15:55:40 +0200449 const struct sk_buff *skb,
Alex Gartrell802c41a2015-08-26 09:40:32 -0700450 const struct ip_vs_iphdr *iph);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700451
452 struct ip_vs_conn *
Eric W. Biederman0cf705c8c2015-09-21 13:02:39 -0500453 (*conn_out_get)(struct netns_ipvs *ipvs,
454 int af,
Julius Volz51ef3482008-09-02 15:55:40 +0200455 const struct sk_buff *skb,
Alex Gartrell802c41a2015-08-26 09:40:32 -0700456 const struct ip_vs_iphdr *iph);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700457
Jesper Dangaard Brouerd4383f02012-09-26 14:07:17 +0200458 int (*snat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
459 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700460
Jesper Dangaard Brouerd4383f02012-09-26 14:07:17 +0200461 int (*dnat_handler)(struct sk_buff *skb, struct ip_vs_protocol *pp,
462 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700463
Julius Volz51ef3482008-09-02 15:55:40 +0200464 int (*csum_check)(int af, struct sk_buff *skb,
465 struct ip_vs_protocol *pp);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700466
467 const char *(*state_name)(int state);
468
Simon Horman4a516f12011-09-16 14:11:49 +0900469 void (*state_transition)(struct ip_vs_conn *cp, int direction,
470 const struct sk_buff *skb,
471 struct ip_vs_proto_data *pd);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700472
Eric W. Biederman19648912015-09-21 13:02:29 -0500473 int (*register_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700474
Eric W. Biederman19648912015-09-21 13:02:29 -0500475 void (*unregister_app)(struct netns_ipvs *ipvs, struct ip_vs_app *inc);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700476
477 int (*app_conn_bind)(struct ip_vs_conn *cp);
478
Julian Anastasov0d796412010-10-17 16:46:17 +0300479 void (*debug_packet)(int af, struct ip_vs_protocol *pp,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700480 const struct sk_buff *skb,
481 int offset,
482 const char *msg);
483
Hans Schillstrom93304192011-01-03 14:44:51 +0100484 void (*timeout_change)(struct ip_vs_proto_data *pd, int flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700485};
486
Simon Horman07dcc682014-09-30 10:50:06 +0900487/* protocol data per netns */
Hans Schillstrom252c6412011-01-03 14:44:46 +0100488struct ip_vs_proto_data {
489 struct ip_vs_proto_data *next;
490 struct ip_vs_protocol *pp;
491 int *timeout_table; /* protocol timeout table */
492 atomic_t appcnt; /* counter of proto app incs. */
493 struct tcp_states_t *tcp_state_table;
494};
495
Joe Perches5c3a0fd2013-09-21 10:22:42 -0700496struct ip_vs_protocol *ip_vs_proto_get(unsigned short proto);
Eric W. Biederman18d6ade2015-09-21 13:02:01 -0500497struct ip_vs_proto_data *ip_vs_proto_data_get(struct netns_ipvs *ipvs,
Joe Perches5c3a0fd2013-09-21 10:22:42 -0700498 unsigned short proto);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700499
Simon Hormanf11017e2010-08-22 21:37:52 +0900500struct ip_vs_conn_param {
Eric W. Biedermane64e2b42015-09-21 13:01:42 -0500501 struct netns_ipvs *ipvs;
Simon Hormanf11017e2010-08-22 21:37:52 +0900502 const union nf_inet_addr *caddr;
503 const union nf_inet_addr *vaddr;
504 __be16 cport;
505 __be16 vport;
506 __u16 protocol;
507 u16 af;
Simon Horman85999282010-08-22 21:37:53 +0900508
509 const struct ip_vs_pe *pe;
510 char *pe_data;
511 __u8 pe_data_len;
Simon Hormanf11017e2010-08-22 21:37:52 +0900512};
513
Simon Horman07dcc682014-09-30 10:50:06 +0900514/* IP_VS structure allocated for each dynamically scheduled connection */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700515struct ip_vs_conn {
Changli Gao731109e2011-02-19 18:05:08 +0800516 struct hlist_node c_list; /* hashed list heads */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700517 /* Protocol, addresses and port numbers */
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100518 __be16 cport;
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100519 __be16 dport;
Julian Anastasov1845ed02013-03-21 11:58:11 +0200520 __be16 vport;
521 u16 af; /* address family */
Hans Schillstrom6e67e582011-01-03 14:44:57 +0100522 union nf_inet_addr caddr; /* client address */
523 union nf_inet_addr vaddr; /* virtual address */
524 union nf_inet_addr daddr; /* destination address */
525 volatile __u32 flags; /* status flags */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700526 __u16 protocol; /* Which protocol (TCP/UDP) */
Alex Gartrellba385282014-09-09 16:40:23 -0700527 __u16 daf; /* Address family of the dest */
Eric W. Biederman58dbc6f2015-09-21 13:01:41 -0500528 struct netns_ipvs *ipvs;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700529
530 /* counter and timer */
Reshetova, Elenab54ab922017-03-16 10:03:34 +0200531 refcount_t refcnt; /* reference count */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700532 struct timer_list timer; /* Expiration timer */
533 volatile unsigned long timeout; /* timeout */
534
535 /* Flags and state transition */
536 spinlock_t lock; /* lock for state transition */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700537 volatile __u16 state; /* state info */
Rumen G. Bogdanovskiefac5272007-11-07 02:36:55 -0800538 volatile __u16 old_state; /* old state, to be used for
539 * state transition triggerd
540 * synchronization
541 */
Julian Anastasov1845ed02013-03-21 11:58:11 +0200542 __u32 fwmark; /* Fire wall mark from skb */
Julian Anastasov749c42b2012-04-24 23:46:40 +0300543 unsigned long sync_endtime; /* jiffies + sent_retries */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700544
545 /* Control members */
546 struct ip_vs_conn *control; /* Master control connection */
547 atomic_t n_control; /* Number of controlled ones */
548 struct ip_vs_dest *dest; /* real server */
549 atomic_t in_pkts; /* incoming packet counter */
550
Simon Horman07dcc682014-09-30 10:50:06 +0900551 /* Packet transmitter for different forwarding methods. If it
552 * mangles the packet, it must return NF_DROP or better NF_STOLEN,
553 * otherwise this must be changed to a sk_buff **.
554 * NF_ACCEPT can be returned when destination is local.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700555 */
556 int (*packet_xmit)(struct sk_buff *skb, struct ip_vs_conn *cp,
Jesper Dangaard Brouerd4383f02012-09-26 14:07:17 +0200557 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700558
559 /* Note: we can group the following members into a structure,
Simon Horman07dcc682014-09-30 10:50:06 +0900560 * in order to save more space, and the following members are
561 * only used in VS/NAT anyway
562 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700563 struct ip_vs_app *app; /* bound ip_vs_app object */
564 void *app_data; /* Application private data */
565 struct ip_vs_seq in_seq; /* incoming seq. struct */
566 struct ip_vs_seq out_seq; /* outgoing seq. struct */
Simon Horman85999282010-08-22 21:37:53 +0900567
Simon Hormane9e5eee2010-11-08 20:05:57 +0900568 const struct ip_vs_pe *pe;
Simon Horman85999282010-08-22 21:37:53 +0900569 char *pe_data;
570 __u8 pe_data_len;
Julian Anastasov088339a2013-03-21 11:58:10 +0200571
572 struct rcu_head rcu_head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700573};
574
Simon Horman07dcc682014-09-30 10:50:06 +0900575/* Extended internal versions of struct ip_vs_service_user and ip_vs_dest_user
576 * for IPv6 support.
Julius Volzc860c6b2008-09-02 15:55:36 +0200577 *
Simon Horman07dcc682014-09-30 10:50:06 +0900578 * We need these to conveniently pass around service and destination
579 * options, but unfortunately, we also need to keep the old definitions to
580 * maintain userspace backwards compatibility for the setsockopt interface.
Julius Volzc860c6b2008-09-02 15:55:36 +0200581 */
582struct ip_vs_service_user_kern {
583 /* virtual service addresses */
584 u16 af;
585 u16 protocol;
586 union nf_inet_addr addr; /* virtual ip address */
Julian Anastasov0a925862013-04-17 23:50:49 +0300587 __be16 port;
Julius Volzc860c6b2008-09-02 15:55:36 +0200588 u32 fwmark; /* firwall mark of service */
589
590 /* virtual service options */
591 char *sched_name;
Simon Horman0d1e71b2010-08-22 21:37:54 +0900592 char *pe_name;
Eric Dumazet95c96172012-04-15 05:58:06 +0000593 unsigned int flags; /* virtual service flags */
594 unsigned int timeout; /* persistent timeout in sec */
Julian Anastasov0a925862013-04-17 23:50:49 +0300595 __be32 netmask; /* persistent netmask or plen */
Julius Volzc860c6b2008-09-02 15:55:36 +0200596};
597
598
599struct ip_vs_dest_user_kern {
600 /* destination server address */
601 union nf_inet_addr addr;
Julian Anastasov0a925862013-04-17 23:50:49 +0300602 __be16 port;
Julius Volzc860c6b2008-09-02 15:55:36 +0200603
604 /* real server options */
Eric Dumazet95c96172012-04-15 05:58:06 +0000605 unsigned int conn_flags; /* connection flags */
Julius Volzc860c6b2008-09-02 15:55:36 +0200606 int weight; /* destination weight */
607
608 /* thresholds for active connections */
609 u32 u_threshold; /* upper threshold */
610 u32 l_threshold; /* lower threshold */
Alex Gartrell6cff3392014-09-09 16:40:20 -0700611
612 /* Address family of addr */
613 u16 af;
Julius Volzc860c6b2008-09-02 15:55:36 +0200614};
615
616
617/*
Simon Horman07dcc682014-09-30 10:50:06 +0900618 * The information about the virtual service offered to the net and the
619 * forwarding entries.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700620 */
621struct ip_vs_service {
Julian Anastasovceec4c32013-03-22 11:46:53 +0200622 struct hlist_node s_list; /* for normal service table */
623 struct hlist_node f_list; /* for fwmark-based service table */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700624 atomic_t refcnt; /* reference counter */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700625
Julius Volze7ade462008-09-02 15:55:33 +0200626 u16 af; /* address family */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700627 __u16 protocol; /* which protocol (TCP/UDP) */
Julius Volze7ade462008-09-02 15:55:33 +0200628 union nf_inet_addr addr; /* IP address for virtual service */
Al Viro014d7302006-09-28 14:29:52 -0700629 __be16 port; /* port number for the service */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700630 __u32 fwmark; /* firewall mark of the service */
Eric Dumazet95c96172012-04-15 05:58:06 +0000631 unsigned int flags; /* service status flags */
632 unsigned int timeout; /* persistent timeout in ticks */
Julian Anastasov0a925862013-04-17 23:50:49 +0300633 __be32 netmask; /* grouping granularity, mask/plen */
Eric W. Biederman3109d2f2015-09-21 13:01:44 -0500634 struct netns_ipvs *ipvs;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700635
636 struct list_head destinations; /* real server d-linked list */
637 __u32 num_dests; /* number of servers */
638 struct ip_vs_stats stats; /* statistics for the service */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639
640 /* for scheduling */
Julian Anastasovceec4c32013-03-22 11:46:53 +0200641 struct ip_vs_scheduler __rcu *scheduler; /* bound scheduler object */
Julian Anastasovba3a3ce2013-03-22 11:46:51 +0200642 spinlock_t sched_lock; /* lock sched_data */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700643 void *sched_data; /* scheduler application data */
Simon Horman85999282010-08-22 21:37:53 +0900644
645 /* alternate persistence engine */
Julian Anastasovceec4c32013-03-22 11:46:53 +0200646 struct ip_vs_pe __rcu *pe;
647
648 struct rcu_head rcu_head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700649};
650
Julian Anastasov026ace02013-03-21 11:58:06 +0200651/* Information for cached dst */
652struct ip_vs_dest_dst {
653 struct dst_entry *dst_cache; /* destination cache entry */
654 u32 dst_cookie;
655 union nf_inet_addr dst_saddr;
656 struct rcu_head rcu_head;
657};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700658
Simon Horman07dcc682014-09-30 10:50:06 +0900659/* The real server destination forwarding entry with ip address, port number,
660 * and so on.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700661 */
662struct ip_vs_dest {
663 struct list_head n_list; /* for the dests in the service */
Julian Anastasov276472e2013-03-21 11:58:08 +0200664 struct hlist_node d_list; /* for table with all the dests */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700665
Julius Volze7ade462008-09-02 15:55:33 +0200666 u16 af; /* address family */
Al Viro014d7302006-09-28 14:29:52 -0700667 __be16 port; /* port number of the server */
Hans Schillstromf2431e62011-01-03 14:45:00 +0100668 union nf_inet_addr addr; /* IP address of the server */
Eric Dumazet95c96172012-04-15 05:58:06 +0000669 volatile unsigned int flags; /* dest status flags */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700670 atomic_t conn_flags; /* flags to copy to conn */
671 atomic_t weight; /* server weight */
672
Reshetova, Elenab54ab922017-03-16 10:03:34 +0200673 refcount_t refcnt; /* reference counter */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700674 struct ip_vs_stats stats; /* statistics */
Julian Anastasovbcbde4c2013-09-12 11:21:07 +0300675 unsigned long idle_start; /* start time, jiffies */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700676
677 /* connection counters and thresholds */
678 atomic_t activeconns; /* active connections */
679 atomic_t inactconns; /* inactive connections */
680 atomic_t persistconns; /* persistent connections */
681 __u32 u_threshold; /* upper threshold */
682 __u32 l_threshold; /* lower threshold */
683
684 /* for destination cache */
685 spinlock_t dst_lock; /* lock of dst_cache */
Julian Anastasov026ace02013-03-21 11:58:06 +0200686 struct ip_vs_dest_dst __rcu *dest_dst; /* cached dst info */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700687
688 /* for virtual service */
Julian Anastasovbcbde4c2013-09-12 11:21:07 +0300689 struct ip_vs_service __rcu *svc; /* service it belongs to */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700690 __u16 protocol; /* which protocol (TCP/UDP) */
Al Viro014d7302006-09-28 14:29:52 -0700691 __be16 vport; /* virtual port number */
Hans Schillstromf2431e62011-01-03 14:45:00 +0100692 union nf_inet_addr vaddr; /* virtual IP address */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700693 __u32 vfwmark; /* firewall mark of service */
Julian Anastasov276472e2013-03-21 11:58:08 +0200694
Julian Anastasov578bc3e2013-03-22 11:46:49 +0200695 struct list_head t_list; /* in dest_trash */
Julian Anastasov276472e2013-03-21 11:58:08 +0200696 unsigned int in_rs_table:1; /* we are in rs_table */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700697};
698
Simon Horman07dcc682014-09-30 10:50:06 +0900699/* The scheduler object */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700700struct ip_vs_scheduler {
701 struct list_head n_list; /* d-linked list head */
702 char *name; /* scheduler name */
703 atomic_t refcnt; /* reference counter */
704 struct module *module; /* THIS_MODULE/NULL */
705
706 /* scheduler initializing service */
707 int (*init_service)(struct ip_vs_service *svc);
708 /* scheduling service finish */
Julian Anastasoved3ffc42013-03-22 11:46:50 +0200709 void (*done_service)(struct ip_vs_service *svc);
Julian Anastasov6b6df462013-03-22 11:46:37 +0200710 /* dest is linked */
711 int (*add_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
712 /* dest is unlinked */
713 int (*del_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
714 /* dest is updated */
715 int (*upd_dest)(struct ip_vs_service *svc, struct ip_vs_dest *dest);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700716
717 /* selecting a server from the given service */
718 struct ip_vs_dest* (*schedule)(struct ip_vs_service *svc,
Julian Anastasovbba54de2013-06-16 09:09:36 +0300719 const struct sk_buff *skb,
720 struct ip_vs_iphdr *iph);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700721};
722
Simon Horman85999282010-08-22 21:37:53 +0900723/* The persistence engine object */
724struct ip_vs_pe {
725 struct list_head n_list; /* d-linked list head */
726 char *name; /* scheduler name */
727 atomic_t refcnt; /* reference counter */
728 struct module *module; /* THIS_MODULE/NULL */
729
730 /* get the connection template, if any */
731 int (*fill_param)(struct ip_vs_conn_param *p, struct sk_buff *skb);
732 bool (*ct_match)(const struct ip_vs_conn_param *p,
733 struct ip_vs_conn *ct);
734 u32 (*hashkey_raw)(const struct ip_vs_conn_param *p, u32 initval,
735 bool inverse);
Simon Hormana3c918a2010-08-22 21:37:53 +0900736 int (*show_pe_data)(const struct ip_vs_conn *cp, char *buf);
Marco Angaroni39b97222016-04-05 18:26:29 +0200737 /* create connections for real-server outgoing packets */
738 struct ip_vs_conn* (*conn_out)(struct ip_vs_service *svc,
739 struct ip_vs_dest *dest,
740 struct sk_buff *skb,
741 const struct ip_vs_iphdr *iph,
742 __be16 dport, __be16 cport);
Simon Horman85999282010-08-22 21:37:53 +0900743};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700744
Simon Horman07dcc682014-09-30 10:50:06 +0900745/* The application module object (a.k.a. app incarnation) */
Eric Dumazetfd2c3ef2009-11-03 03:26:03 +0000746struct ip_vs_app {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700747 struct list_head a_list; /* member in app list */
748 int type; /* IP_VS_APP_TYPE_xxx */
749 char *name; /* application module name */
750 __u16 protocol;
751 struct module *module; /* THIS_MODULE/NULL */
752 struct list_head incs_list; /* list of incarnations */
753
754 /* members for application incarnations */
755 struct list_head p_list; /* member in proto app list */
756 struct ip_vs_app *app; /* its real application */
Al Viro014d7302006-09-28 14:29:52 -0700757 __be16 port; /* port number in net order */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700758 atomic_t usecnt; /* usage counter */
Julian Anastasov363c97d2013-03-21 11:58:07 +0200759 struct rcu_head rcu_head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700760
Simon Horman07dcc682014-09-30 10:50:06 +0900761 /* output hook: Process packet in inout direction, diff set for TCP.
Julian Anastasov8b27b102010-10-17 16:17:20 +0300762 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
763 * 2=Mangled but checksum was not updated
764 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700765 int (*pkt_out)(struct ip_vs_app *, struct ip_vs_conn *,
Herbert Xu3db05fea2007-10-15 00:53:15 -0700766 struct sk_buff *, int *diff);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700767
Simon Horman07dcc682014-09-30 10:50:06 +0900768 /* input hook: Process packet in outin direction, diff set for TCP.
Julian Anastasov8b27b102010-10-17 16:17:20 +0300769 * Return: 0=Error, 1=Payload Not Mangled/Mangled but checksum is ok,
770 * 2=Mangled but checksum was not updated
771 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700772 int (*pkt_in)(struct ip_vs_app *, struct ip_vs_conn *,
Herbert Xu3db05fea2007-10-15 00:53:15 -0700773 struct sk_buff *, int *diff);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700774
775 /* ip_vs_app initializer */
776 int (*init_conn)(struct ip_vs_app *, struct ip_vs_conn *);
777
778 /* ip_vs_app finish */
779 int (*done_conn)(struct ip_vs_app *, struct ip_vs_conn *);
780
781
782 /* not used now */
783 int (*bind_conn)(struct ip_vs_app *, struct ip_vs_conn *,
784 struct ip_vs_protocol *);
785
786 void (*unbind_conn)(struct ip_vs_app *, struct ip_vs_conn *);
787
788 int * timeout_table;
789 int * timeouts;
790 int timeouts_size;
791
792 int (*conn_schedule)(struct sk_buff *skb, struct ip_vs_app *app,
793 int *verdict, struct ip_vs_conn **cpp);
794
795 struct ip_vs_conn *
796 (*conn_in_get)(const struct sk_buff *skb, struct ip_vs_app *app,
Jesper Dangaard Brouerd4383f02012-09-26 14:07:17 +0200797 const struct iphdr *iph, int inverse);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700798
799 struct ip_vs_conn *
800 (*conn_out_get)(const struct sk_buff *skb, struct ip_vs_app *app,
Jesper Dangaard Brouerd4383f02012-09-26 14:07:17 +0200801 const struct iphdr *iph, int inverse);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700802
803 int (*state_transition)(struct ip_vs_conn *cp, int direction,
804 const struct sk_buff *skb,
805 struct ip_vs_app *app);
806
807 void (*timeout_change)(struct ip_vs_app *app, int flags);
808};
809
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +0200810struct ipvs_master_sync_state {
811 struct list_head sync_queue;
812 struct ip_vs_sync_buff *sync_buff;
Zhang Yanfei07995672013-04-29 11:55:10 -0700813 unsigned long sync_queue_len;
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +0200814 unsigned int sync_queue_delay;
815 struct task_struct *master_thread;
816 struct delayed_work master_wakeup_work;
817 struct netns_ipvs *ipvs;
818};
819
Julian Anastasov578bc3e2013-03-22 11:46:49 +0200820/* How much time to keep dests in trash */
821#define IP_VS_DEST_TRASH_PERIOD (120 * HZ)
822
Julian Anastasove4ff6752015-07-26 15:03:27 +0300823struct ipvs_sync_daemon_cfg {
Julian Anastasovd3328812015-07-26 15:03:28 +0300824 union nf_inet_addr mcast_group;
Julian Anastasove4ff6752015-07-26 15:03:27 +0300825 int syncid;
826 u16 sync_maxlen;
Julian Anastasovd3328812015-07-26 15:03:28 +0300827 u16 mcast_port;
828 u8 mcast_af;
829 u8 mcast_ttl;
Julian Anastasove4ff6752015-07-26 15:03:27 +0300830 /* multicast interface name */
831 char mcast_ifn[IP_VS_IFNAME_MAXLEN];
832};
833
Julian Anastasov2553d062011-03-04 12:18:07 +0200834/* IPVS in network namespace */
835struct netns_ipvs {
836 int gen; /* Generation */
Hans Schillstrom7a4f0762011-05-03 22:09:31 +0200837 int enable; /* enable like nf_hooks do */
Simon Horman07dcc682014-09-30 10:50:06 +0900838 /* Hash table: for real service lookups */
Julian Anastasov2553d062011-03-04 12:18:07 +0200839 #define IP_VS_RTAB_BITS 4
840 #define IP_VS_RTAB_SIZE (1 << IP_VS_RTAB_BITS)
841 #define IP_VS_RTAB_MASK (IP_VS_RTAB_SIZE - 1)
842
Julian Anastasov276472e2013-03-21 11:58:08 +0200843 struct hlist_head rs_table[IP_VS_RTAB_SIZE];
Julian Anastasov2553d062011-03-04 12:18:07 +0200844 /* ip_vs_app */
845 struct list_head app_list;
Julian Anastasov2553d062011-03-04 12:18:07 +0200846 /* ip_vs_proto */
847 #define IP_VS_PROTO_TAB_SIZE 32 /* must be power of 2 */
848 struct ip_vs_proto_data *proto_data_table[IP_VS_PROTO_TAB_SIZE];
849 /* ip_vs_proto_tcp */
850#ifdef CONFIG_IP_VS_PROTO_TCP
851 #define TCP_APP_TAB_BITS 4
852 #define TCP_APP_TAB_SIZE (1 << TCP_APP_TAB_BITS)
853 #define TCP_APP_TAB_MASK (TCP_APP_TAB_SIZE - 1)
854 struct list_head tcp_apps[TCP_APP_TAB_SIZE];
Julian Anastasov2553d062011-03-04 12:18:07 +0200855#endif
856 /* ip_vs_proto_udp */
857#ifdef CONFIG_IP_VS_PROTO_UDP
858 #define UDP_APP_TAB_BITS 4
859 #define UDP_APP_TAB_SIZE (1 << UDP_APP_TAB_BITS)
860 #define UDP_APP_TAB_MASK (UDP_APP_TAB_SIZE - 1)
861 struct list_head udp_apps[UDP_APP_TAB_SIZE];
Julian Anastasov2553d062011-03-04 12:18:07 +0200862#endif
863 /* ip_vs_proto_sctp */
864#ifdef CONFIG_IP_VS_PROTO_SCTP
865 #define SCTP_APP_TAB_BITS 4
866 #define SCTP_APP_TAB_SIZE (1 << SCTP_APP_TAB_BITS)
867 #define SCTP_APP_TAB_MASK (SCTP_APP_TAB_SIZE - 1)
868 /* Hash table for SCTP application incarnations */
869 struct list_head sctp_apps[SCTP_APP_TAB_SIZE];
Julian Anastasov2553d062011-03-04 12:18:07 +0200870#endif
871 /* ip_vs_conn */
Simon Horman07dcc682014-09-30 10:50:06 +0900872 atomic_t conn_count; /* connection counter */
Julian Anastasov2553d062011-03-04 12:18:07 +0200873
874 /* ip_vs_ctl */
Julian Anastasov2a0751a2011-03-04 12:20:35 +0200875 struct ip_vs_stats tot_stats; /* Statistics & est. */
Julian Anastasov2553d062011-03-04 12:18:07 +0200876
877 int num_services; /* no of virtual services */
Simon Hormanf2247fb2011-02-04 18:33:02 +0900878
Julian Anastasov2553d062011-03-04 12:18:07 +0200879 /* Trash for destinations */
880 struct list_head dest_trash;
Julian Anastasov578bc3e2013-03-22 11:46:49 +0200881 spinlock_t dest_trash_lock;
882 struct timer_list dest_trash_timer; /* expiration timer */
Julian Anastasov2553d062011-03-04 12:18:07 +0200883 /* Service counters */
884 atomic_t ftpsvc_counter;
885 atomic_t nullsvc_counter;
Marco Angaroni39b97222016-04-05 18:26:29 +0200886 atomic_t conn_out_counter;
Julian Anastasov2553d062011-03-04 12:18:07 +0200887
Simon Hormanf2247fb2011-02-04 18:33:02 +0900888#ifdef CONFIG_SYSCTL
889 /* 1/rate drop and drop-entry variables */
890 struct delayed_work defense_work; /* Work handler */
891 int drop_rate;
892 int drop_counter;
893 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
Julian Anastasov2553d062011-03-04 12:18:07 +0200899 /* sys-ctl struct */
900 struct ctl_table_header *sysctl_hdr;
901 struct ctl_table *sysctl_tbl;
Simon Hormanf2247fb2011-02-04 18:33:02 +0900902#endif
903
Julian Anastasov2553d062011-03-04 12:18:07 +0200904 /* 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;
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +0200915 int sysctl_sync_ports;
Julian Anastasov4d0c8752013-06-24 22:44:41 +0300916 int sysctl_sync_persist_mode;
Zhang Yanfei07995672013-04-29 11:55:10 -0700917 unsigned long sysctl_sync_qlen_max;
Pablo Neira Ayuso1c003b12012-05-08 19:39:49 +0200918 int sysctl_sync_sock_size;
Julian Anastasov2553d062011-03-04 12:18:07 +0200919 int sysctl_cache_bypass;
920 int sysctl_expire_nodest_conn;
Alexander Frolkinc6c96c12013-06-13 08:56:15 +0100921 int sysctl_sloppy_tcp;
922 int sysctl_sloppy_sctp;
Julian Anastasov2553d062011-03-04 12:18:07 +0200923 int sysctl_expire_quiescent_template;
924 int sysctl_sync_threshold[2];
Julian Anastasov749c42b2012-04-24 23:46:40 +0300925 unsigned int sysctl_sync_refresh_period;
926 int sysctl_sync_retries;
Julian Anastasov2553d062011-03-04 12:18:07 +0200927 int sysctl_nat_icmp_send;
Julian Anastasov3654e612012-07-20 11:59:53 +0300928 int sysctl_pmtu_disc;
Julian Anastasov0c125822013-03-09 23:25:04 +0200929 int sysctl_backup_only;
Marcelo Ricardo Leitnerd752c362015-02-23 15:02:34 -0300930 int sysctl_conn_reuse_mode;
Alex Gartrell94485fe2015-08-26 09:40:34 -0700931 int sysctl_schedule_icmp;
Alex Gartrell4e478092015-09-14 23:23:05 -0700932 int sysctl_ignore_tunneled;
Julian Anastasov2553d062011-03-04 12:18:07 +0200933
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 */
Julian Anastasov2553d062011-03-04 12:18:07 +0200947 spinlock_t sync_lock;
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +0200948 struct ipvs_master_sync_state *ms;
Julian Anastasov2553d062011-03-04 12:18:07 +0200949 spinlock_t sync_buff_lock;
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +0200950 struct task_struct **backup_threads;
951 int threads_mask;
Julian Anastasov2553d062011-03-04 12:18:07 +0200952 volatile int sync_state;
Hans Schillstromae1d48b2011-10-11 10:54:35 +0900953 struct mutex sync_mutex;
Julian Anastasove4ff6752015-07-26 15:03:27 +0300954 struct ipvs_sync_daemon_cfg mcfg; /* Master Configuration */
955 struct ipvs_sync_daemon_cfg bcfg; /* Backup Configuration */
Julian Anastasov2553d062011-03-04 12:18:07 +0200956 /* net name space ptr */
957 struct net *net; /* Needed by timer routines */
Simon Horman07dcc682014-09-30 10:50:06 +0900958 /* Number of heterogeneous destinations, needed becaus heterogeneous
959 * are not supported when synchronization is enabled.
960 */
Alex Gartrell391f503d2014-09-09 16:40:24 -0700961 unsigned int mixed_address_family_dests;
Julian Anastasov2553d062011-03-04 12:18:07 +0200962};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700963
Simon Horman59e03502011-02-04 18:33:01 +0900964#define DEFAULT_SYNC_THRESHOLD 3
965#define DEFAULT_SYNC_PERIOD 50
Simon Horman7532e8d2011-02-04 18:33:01 +0900966#define DEFAULT_SYNC_VER 1
Alexander Frolkinc6c96c12013-06-13 08:56:15 +0100967#define DEFAULT_SLOPPY_TCP 0
968#define DEFAULT_SLOPPY_SCTP 0
Julian Anastasov749c42b2012-04-24 23:46:40 +0300969#define DEFAULT_SYNC_REFRESH_PERIOD (0U * HZ)
970#define DEFAULT_SYNC_RETRIES 0
Pablo Neira Ayuso1c003b12012-05-08 19:39:49 +0200971#define IPVS_SYNC_WAKEUP_RATE 8
972#define IPVS_SYNC_QLEN_MAX (IPVS_SYNC_WAKEUP_RATE * 4)
973#define IPVS_SYNC_SEND_DELAY (HZ / 50)
974#define IPVS_SYNC_CHECK_PERIOD HZ
Julian Anastasov749c42b2012-04-24 23:46:40 +0300975#define IPVS_SYNC_FLUSH_TIME (HZ * 2)
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +0200976#define IPVS_SYNC_PORTS_MAX (1 << 6)
Simon Horman59e03502011-02-04 18:33:01 +0900977
978#ifdef CONFIG_SYSCTL
979
980static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
981{
982 return ipvs->sysctl_sync_threshold[0];
983}
984
985static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
986{
Mark Rutland6aa7de02017-10-23 14:07:29 -0700987 return READ_ONCE(ipvs->sysctl_sync_threshold[1]);
Julian Anastasov749c42b2012-04-24 23:46:40 +0300988}
989
990static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
991{
Mark Rutland6aa7de02017-10-23 14:07:29 -0700992 return READ_ONCE(ipvs->sysctl_sync_refresh_period);
Julian Anastasov749c42b2012-04-24 23:46:40 +0300993}
994
995static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
996{
997 return ipvs->sysctl_sync_retries;
Simon Horman59e03502011-02-04 18:33:01 +0900998}
999
Simon Horman7532e8d2011-02-04 18:33:01 +09001000static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1001{
1002 return ipvs->sysctl_sync_ver;
1003}
1004
Alexander Frolkinc6c96c12013-06-13 08:56:15 +01001005static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1006{
1007 return ipvs->sysctl_sloppy_tcp;
1008}
1009
1010static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1011{
1012 return ipvs->sysctl_sloppy_sctp;
1013}
1014
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +02001015static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1016{
Mark Rutland6aa7de02017-10-23 14:07:29 -07001017 return READ_ONCE(ipvs->sysctl_sync_ports);
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +02001018}
1019
Julian Anastasov4d0c8752013-06-24 22:44:41 +03001020static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1021{
1022 return ipvs->sysctl_sync_persist_mode;
1023}
1024
Zhang Yanfei07995672013-04-29 11:55:10 -07001025static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
Pablo Neira Ayuso1c003b12012-05-08 19:39:49 +02001026{
1027 return ipvs->sysctl_sync_qlen_max;
1028}
1029
1030static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1031{
1032 return ipvs->sysctl_sync_sock_size;
1033}
1034
Julian Anastasov3654e612012-07-20 11:59:53 +03001035static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1036{
1037 return ipvs->sysctl_pmtu_disc;
1038}
1039
Julian Anastasov0c125822013-03-09 23:25:04 +02001040static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1041{
1042 return ipvs->sync_state & IP_VS_STATE_BACKUP &&
1043 ipvs->sysctl_backup_only;
1044}
1045
Marcelo Ricardo Leitnerd752c362015-02-23 15:02:34 -03001046static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1047{
1048 return ipvs->sysctl_conn_reuse_mode;
1049}
1050
Alex Gartrell94485fe2015-08-26 09:40:34 -07001051static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1052{
1053 return ipvs->sysctl_schedule_icmp;
1054}
1055
Alex Gartrell4e478092015-09-14 23:23:05 -07001056static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1057{
1058 return ipvs->sysctl_ignore_tunneled;
1059}
1060
Eric W. Biederman57032942015-09-21 13:02:49 -05001061static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1062{
1063 return ipvs->sysctl_cache_bypass;
1064}
1065
Simon Horman59e03502011-02-04 18:33:01 +09001066#else
1067
1068static inline int sysctl_sync_threshold(struct netns_ipvs *ipvs)
1069{
1070 return DEFAULT_SYNC_THRESHOLD;
1071}
1072
1073static inline int sysctl_sync_period(struct netns_ipvs *ipvs)
1074{
1075 return DEFAULT_SYNC_PERIOD;
1076}
1077
Julian Anastasov749c42b2012-04-24 23:46:40 +03001078static inline unsigned int sysctl_sync_refresh_period(struct netns_ipvs *ipvs)
1079{
1080 return DEFAULT_SYNC_REFRESH_PERIOD;
1081}
1082
1083static inline int sysctl_sync_retries(struct netns_ipvs *ipvs)
1084{
1085 return DEFAULT_SYNC_RETRIES & 3;
1086}
1087
Simon Horman7532e8d2011-02-04 18:33:01 +09001088static inline int sysctl_sync_ver(struct netns_ipvs *ipvs)
1089{
1090 return DEFAULT_SYNC_VER;
1091}
1092
Alexander Frolkinc6c96c12013-06-13 08:56:15 +01001093static inline int sysctl_sloppy_tcp(struct netns_ipvs *ipvs)
1094{
1095 return DEFAULT_SLOPPY_TCP;
1096}
1097
1098static inline int sysctl_sloppy_sctp(struct netns_ipvs *ipvs)
1099{
1100 return DEFAULT_SLOPPY_SCTP;
1101}
1102
Pablo Neira Ayusof73181c2012-05-08 19:40:30 +02001103static inline int sysctl_sync_ports(struct netns_ipvs *ipvs)
1104{
1105 return 1;
1106}
1107
Julian Anastasov4d0c8752013-06-24 22:44:41 +03001108static inline int sysctl_sync_persist_mode(struct netns_ipvs *ipvs)
1109{
1110 return 0;
1111}
1112
Zhang Yanfei07995672013-04-29 11:55:10 -07001113static inline unsigned long sysctl_sync_qlen_max(struct netns_ipvs *ipvs)
Pablo Neira Ayuso1c003b12012-05-08 19:39:49 +02001114{
1115 return IPVS_SYNC_QLEN_MAX;
1116}
1117
1118static inline int sysctl_sync_sock_size(struct netns_ipvs *ipvs)
1119{
1120 return 0;
1121}
1122
Julian Anastasov3654e612012-07-20 11:59:53 +03001123static inline int sysctl_pmtu_disc(struct netns_ipvs *ipvs)
1124{
1125 return 1;
1126}
1127
Julian Anastasov0c125822013-03-09 23:25:04 +02001128static inline int sysctl_backup_only(struct netns_ipvs *ipvs)
1129{
1130 return 0;
1131}
1132
Marcelo Ricardo Leitnerd752c362015-02-23 15:02:34 -03001133static inline int sysctl_conn_reuse_mode(struct netns_ipvs *ipvs)
1134{
1135 return 1;
1136}
1137
Alex Gartrell94485fe2015-08-26 09:40:34 -07001138static inline int sysctl_schedule_icmp(struct netns_ipvs *ipvs)
1139{
1140 return 0;
1141}
1142
Alex Gartrell4e478092015-09-14 23:23:05 -07001143static inline int sysctl_ignore_tunneled(struct netns_ipvs *ipvs)
1144{
1145 return 0;
1146}
1147
Eric W. Biederman57032942015-09-21 13:02:49 -05001148static inline int sysctl_cache_bypass(struct netns_ipvs *ipvs)
1149{
1150 return 0;
1151}
1152
Simon Horman59e03502011-02-04 18:33:01 +09001153#endif
1154
Simon Horman07dcc682014-09-30 10:50:06 +09001155/* IPVS core functions
1156 * (from ip_vs_core.c)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001157 */
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001158const char *ip_vs_proto_name(unsigned int proto);
1159void ip_vs_init_hash_table(struct list_head *table, int rows);
Marco Angaroni39b97222016-04-05 18:26:29 +02001160struct ip_vs_conn *ip_vs_new_conn_out(struct ip_vs_service *svc,
1161 struct ip_vs_dest *dest,
1162 struct sk_buff *skb,
1163 const struct ip_vs_iphdr *iph,
1164 __be16 dport,
1165 __be16 cport);
Sven Wegenerafdd6142008-08-10 09:18:01 +00001166#define IP_VS_INIT_HASH_TABLE(t) ip_vs_init_hash_table((t), ARRAY_SIZE((t)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001167
Linus Torvalds1da177e2005-04-16 15:20:36 -07001168#define IP_VS_APP_TYPE_FTP 1
1169
Simon Horman07dcc682014-09-30 10:50:06 +09001170/* ip_vs_conn handling functions
1171 * (from ip_vs_conn.c)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001172 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001173enum {
1174 IP_VS_DIR_INPUT = 0,
1175 IP_VS_DIR_OUTPUT,
1176 IP_VS_DIR_INPUT_ONLY,
1177 IP_VS_DIR_LAST,
1178};
1179
Eric W. Biederman19913de2015-09-21 13:01:43 -05001180static inline void ip_vs_conn_fill_param(struct netns_ipvs *ipvs, int af, int protocol,
Simon Hormanf11017e2010-08-22 21:37:52 +09001181 const union nf_inet_addr *caddr,
1182 __be16 cport,
1183 const union nf_inet_addr *vaddr,
1184 __be16 vport,
1185 struct ip_vs_conn_param *p)
1186{
Eric W. Biederman19913de2015-09-21 13:01:43 -05001187 p->ipvs = ipvs;
Simon Hormanf11017e2010-08-22 21:37:52 +09001188 p->af = af;
1189 p->protocol = protocol;
1190 p->caddr = caddr;
1191 p->cport = cport;
1192 p->vaddr = vaddr;
1193 p->vport = vport;
Simon Horman85999282010-08-22 21:37:53 +09001194 p->pe = NULL;
1195 p->pe_data = NULL;
Simon Hormanf11017e2010-08-22 21:37:52 +09001196}
Julius Volz28364a52008-09-02 15:55:43 +02001197
Simon Hormanf11017e2010-08-22 21:37:52 +09001198struct ip_vs_conn *ip_vs_conn_in_get(const struct ip_vs_conn_param *p);
1199struct ip_vs_conn *ip_vs_ct_in_get(const struct ip_vs_conn_param *p);
Julius Volz28364a52008-09-02 15:55:43 +02001200
Eric W. Biedermanab161972015-09-21 13:02:38 -05001201struct ip_vs_conn * ip_vs_conn_in_get_proto(struct netns_ipvs *ipvs, int af,
1202 const struct sk_buff *skb,
Alex Gartrell802c41a2015-08-26 09:40:32 -07001203 const struct ip_vs_iphdr *iph);
Simon Horman5c0d2372010-08-02 17:12:44 +02001204
Simon Hormanf11017e2010-08-22 21:37:52 +09001205struct ip_vs_conn *ip_vs_conn_out_get(const struct ip_vs_conn_param *p);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001206
Eric W. Biederman0cf705c8c2015-09-21 13:02:39 -05001207struct ip_vs_conn * ip_vs_conn_out_get_proto(struct netns_ipvs *ipvs, int af,
1208 const struct sk_buff *skb,
Alex Gartrell802c41a2015-08-26 09:40:32 -07001209 const struct ip_vs_iphdr *iph);
Simon Horman5c0d2372010-08-02 17:12:44 +02001210
Julian Anastasov088339a2013-03-21 11:58:10 +02001211/* Get reference to gain full access to conn.
1212 * By default, RCU read-side critical sections have access only to
1213 * conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
1214 */
1215static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
1216{
Reshetova, Elenab54ab922017-03-16 10:03:34 +02001217 return refcount_inc_not_zero(&cp->refcnt);
Julian Anastasov088339a2013-03-21 11:58:10 +02001218}
1219
Linus Torvalds1da177e2005-04-16 15:20:36 -07001220/* put back the conn without restarting its timer */
1221static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
1222{
Peter Zijlstra4e857c52014-03-17 18:06:10 +01001223 smp_mb__before_atomic();
Reshetova, Elenab54ab922017-03-16 10:03:34 +02001224 refcount_dec(&cp->refcnt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001225}
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001226void ip_vs_conn_put(struct ip_vs_conn *cp);
1227void ip_vs_conn_fill_cport(struct ip_vs_conn *cp, __be16 cport);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001228
Alex Gartrellba385282014-09-09 16:40:23 -07001229struct ip_vs_conn *ip_vs_conn_new(const struct ip_vs_conn_param *p, int dest_af,
Simon Hormanf11017e2010-08-22 21:37:52 +09001230 const union nf_inet_addr *daddr,
Eric Dumazet95c96172012-04-15 05:58:06 +00001231 __be16 dport, unsigned int flags,
Hans Schillstrom0e051e62010-11-19 14:25:07 +01001232 struct ip_vs_dest *dest, __u32 fwmark);
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001233void ip_vs_conn_expire_now(struct ip_vs_conn *cp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001234
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001235const char *ip_vs_state_name(__u16 proto, int state);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001236
Eric W. Biederman69f39092015-09-21 13:02:57 -05001237void ip_vs_tcp_conn_listen(struct ip_vs_conn *cp);
Marco Angaroni3ec10d32016-05-16 19:18:09 +02001238int ip_vs_check_template(struct ip_vs_conn *ct, struct ip_vs_dest *cdest);
Eric W. Biederman423b5592015-09-21 13:02:24 -05001239void ip_vs_random_dropentry(struct netns_ipvs *ipvs);
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001240int ip_vs_conn_init(void);
1241void ip_vs_conn_cleanup(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001242
1243static inline void ip_vs_control_del(struct ip_vs_conn *cp)
1244{
1245 struct ip_vs_conn *ctl_cp = cp->control;
1246 if (!ctl_cp) {
Julius Volzcfc78c52008-09-02 15:55:53 +02001247 IP_VS_ERR_BUF("request control DEL for uncontrolled: "
1248 "%s:%d to %s:%d\n",
1249 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1250 ntohs(cp->cport),
1251 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1252 ntohs(cp->vport));
1253
Linus Torvalds1da177e2005-04-16 15:20:36 -07001254 return;
1255 }
1256
Julius Volzcfc78c52008-09-02 15:55:53 +02001257 IP_VS_DBG_BUF(7, "DELeting control for: "
1258 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1259 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1260 ntohs(cp->cport),
1261 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1262 ntohs(ctl_cp->cport));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001263
1264 cp->control = NULL;
1265 if (atomic_read(&ctl_cp->n_control) == 0) {
Julius Volzcfc78c52008-09-02 15:55:53 +02001266 IP_VS_ERR_BUF("BUG control DEL with n=0 : "
1267 "%s:%d to %s:%d\n",
1268 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1269 ntohs(cp->cport),
1270 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1271 ntohs(cp->vport));
1272
Linus Torvalds1da177e2005-04-16 15:20:36 -07001273 return;
1274 }
1275 atomic_dec(&ctl_cp->n_control);
1276}
1277
1278static inline void
1279ip_vs_control_add(struct ip_vs_conn *cp, struct ip_vs_conn *ctl_cp)
1280{
1281 if (cp->control) {
Julius Volzcfc78c52008-09-02 15:55:53 +02001282 IP_VS_ERR_BUF("request control ADD for already controlled: "
1283 "%s:%d to %s:%d\n",
1284 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1285 ntohs(cp->cport),
1286 IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
1287 ntohs(cp->vport));
1288
Linus Torvalds1da177e2005-04-16 15:20:36 -07001289 ip_vs_control_del(cp);
1290 }
1291
Julius Volzcfc78c52008-09-02 15:55:53 +02001292 IP_VS_DBG_BUF(7, "ADDing control for: "
1293 "cp.dst=%s:%d ctl_cp.dst=%s:%d\n",
1294 IP_VS_DBG_ADDR(cp->af, &cp->caddr),
1295 ntohs(cp->cport),
1296 IP_VS_DBG_ADDR(cp->af, &ctl_cp->caddr),
1297 ntohs(ctl_cp->cport));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001298
1299 cp->control = ctl_cp;
1300 atomic_inc(&ctl_cp->n_control);
1301}
1302
Simon Horman07dcc682014-09-30 10:50:06 +09001303/* IPVS netns init & cleanup functions */
Eric W. Biedermana4dd0362015-09-21 13:02:28 -05001304int ip_vs_estimator_net_init(struct netns_ipvs *ipvs);
Eric W. Biederman3d993762015-09-21 13:02:26 -05001305int ip_vs_control_net_init(struct netns_ipvs *ipvs);
Eric W. Biederman7d1f88e2015-09-21 13:02:58 -05001306int ip_vs_protocol_net_init(struct netns_ipvs *ipvs);
Eric W. Biedermanb5dd2122015-09-21 13:02:34 -05001307int ip_vs_app_net_init(struct netns_ipvs *ipvs);
Eric W. Biederman2f3edc6a2015-09-21 13:02:42 -05001308int ip_vs_conn_net_init(struct netns_ipvs *ipvs);
Eric W. Biederman802cb432015-09-21 13:02:20 -05001309int ip_vs_sync_net_init(struct netns_ipvs *ipvs);
Eric W. Biederman2f3edc6a2015-09-21 13:02:42 -05001310void ip_vs_conn_net_cleanup(struct netns_ipvs *ipvs);
Eric W. Biedermanb5dd2122015-09-21 13:02:34 -05001311void ip_vs_app_net_cleanup(struct netns_ipvs *ipvs);
Eric W. Biederman7d1f88e2015-09-21 13:02:58 -05001312void ip_vs_protocol_net_cleanup(struct netns_ipvs *ipvs);
Eric W. Biederman3d993762015-09-21 13:02:26 -05001313void ip_vs_control_net_cleanup(struct netns_ipvs *ipvs);
Eric W. Biedermana4dd0362015-09-21 13:02:28 -05001314void ip_vs_estimator_net_cleanup(struct netns_ipvs *ipvs);
Eric W. Biedermanebea1f72015-09-21 13:02:21 -05001315void ip_vs_sync_net_cleanup(struct netns_ipvs *ipvs);
Eric W. Biederman56d21692015-09-21 13:01:58 -05001316void ip_vs_service_net_cleanup(struct netns_ipvs *ipvs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001317
Simon Horman07dcc682014-09-30 10:50:06 +09001318/* IPVS application functions
1319 * (from ip_vs_app.c)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001320 */
1321#define IP_VS_APP_MAX_PORTS 8
Eric W. Biederman9f8128a2015-09-21 13:02:32 -05001322struct ip_vs_app *register_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
1323void unregister_ip_vs_app(struct netns_ipvs *ipvs, struct ip_vs_app *app);
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001324int ip_vs_bind_app(struct ip_vs_conn *cp, struct ip_vs_protocol *pp);
1325void ip_vs_unbind_app(struct ip_vs_conn *cp);
Eric W. Biederman3250dc92015-09-21 13:02:31 -05001326int register_ip_vs_app_inc(struct netns_ipvs *ipvs, struct ip_vs_app *app, __u16 proto,
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001327 __u16 port);
1328int ip_vs_app_inc_get(struct ip_vs_app *inc);
1329void ip_vs_app_inc_put(struct ip_vs_app *inc);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001330
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001331int ip_vs_app_pkt_out(struct ip_vs_conn *, struct sk_buff *skb);
1332int ip_vs_app_pkt_in(struct ip_vs_conn *, struct sk_buff *skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001333
Simon Horman8be67a62010-08-22 21:37:54 +09001334int register_ip_vs_pe(struct ip_vs_pe *pe);
1335int unregister_ip_vs_pe(struct ip_vs_pe *pe);
Simon Hormane9e5eee2010-11-08 20:05:57 +09001336struct ip_vs_pe *ip_vs_pe_getbyname(const char *name);
Hans Schillstromfe5e7a12010-11-19 14:25:12 +01001337struct ip_vs_pe *__ip_vs_pe_getbyname(const char *pe_name);
Simon Hormane9e5eee2010-11-08 20:05:57 +09001338
Simon Horman07dcc682014-09-30 10:50:06 +09001339/* Use a #define to avoid all of module.h just for these trivial ops */
Paul Gortmaker69e7dae2011-08-29 12:40:15 -04001340#define ip_vs_pe_get(pe) \
1341 if (pe && pe->module) \
Simon Hormane9e5eee2010-11-08 20:05:57 +09001342 __module_get(pe->module);
Simon Hormane9e5eee2010-11-08 20:05:57 +09001343
Paul Gortmaker69e7dae2011-08-29 12:40:15 -04001344#define ip_vs_pe_put(pe) \
1345 if (pe && pe->module) \
Simon Hormane9e5eee2010-11-08 20:05:57 +09001346 module_put(pe->module);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001347
Simon Horman07dcc682014-09-30 10:50:06 +09001348/* IPVS protocol functions (from ip_vs_proto.c) */
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001349int ip_vs_protocol_init(void);
1350void ip_vs_protocol_cleanup(void);
1351void ip_vs_protocol_timeout_change(struct netns_ipvs *ipvs, int flags);
1352int *ip_vs_create_timeout_table(int *table, int size);
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001353void ip_vs_tcpudp_debug_packet(int af, struct ip_vs_protocol *pp,
1354 const struct sk_buff *skb, int offset,
1355 const char *msg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001356
1357extern struct ip_vs_protocol ip_vs_protocol_tcp;
1358extern struct ip_vs_protocol ip_vs_protocol_udp;
1359extern struct ip_vs_protocol ip_vs_protocol_icmp;
1360extern struct ip_vs_protocol ip_vs_protocol_esp;
1361extern struct ip_vs_protocol ip_vs_protocol_ah;
Venkata Mohan Reddy2906f662010-02-18 12:31:05 +01001362extern struct ip_vs_protocol ip_vs_protocol_sctp;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001363
Simon Horman07dcc682014-09-30 10:50:06 +09001364/* Registering/unregistering scheduler functions
1365 * (from ip_vs_sched.c)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001366 */
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001367int register_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1368int unregister_ip_vs_scheduler(struct ip_vs_scheduler *scheduler);
1369int ip_vs_bind_scheduler(struct ip_vs_service *svc,
1370 struct ip_vs_scheduler *scheduler);
1371void ip_vs_unbind_scheduler(struct ip_vs_service *svc,
1372 struct ip_vs_scheduler *sched);
1373struct ip_vs_scheduler *ip_vs_scheduler_get(const char *sched_name);
1374void ip_vs_scheduler_put(struct ip_vs_scheduler *scheduler);
1375struct ip_vs_conn *
Julian Anastasov190ecd22010-10-17 16:24:37 +03001376ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
Jesper Dangaard Brouerd4383f02012-09-26 14:07:17 +02001377 struct ip_vs_proto_data *pd, int *ignored,
1378 struct ip_vs_iphdr *iph);
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001379int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
1380 struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001381
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001382void ip_vs_scheduler_err(struct ip_vs_service *svc, const char *msg);
Patrick Schaaf41ac51e2011-02-11 14:01:12 +01001383
Simon Horman07dcc682014-09-30 10:50:06 +09001384/* IPVS control data and functions (from ip_vs_ctl.c) */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001385extern struct ip_vs_stats ip_vs_stats;
Hans Schillstromb880c1f2010-11-19 14:25:14 +01001386extern int sysctl_ip_vs_sync_ver;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001387
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001388struct ip_vs_service *
Eric W. Biederman0a4fd6c2015-09-21 13:01:49 -05001389ip_vs_service_find(struct netns_ipvs *ipvs, int af, __u32 fwmark, __u16 protocol,
Julius Volz3c2e0502008-09-02 15:55:38 +02001390 const union nf_inet_addr *vaddr, __be16 vport);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001391
Eric W. Biederman48aed1b2015-09-21 13:01:50 -05001392bool ip_vs_has_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001393 const union nf_inet_addr *daddr, __be16 dport);
Julius Volz7937df12008-09-02 15:55:48 +02001394
Marco Angaroni39b97222016-04-05 18:26:29 +02001395struct ip_vs_dest *
1396ip_vs_find_real_service(struct netns_ipvs *ipvs, int af, __u16 protocol,
1397 const union nf_inet_addr *daddr, __be16 dport);
1398
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001399int ip_vs_use_count_inc(void);
1400void ip_vs_use_count_dec(void);
1401int ip_vs_register_nl_ioctl(void);
1402void ip_vs_unregister_nl_ioctl(void);
1403int ip_vs_control_init(void);
1404void ip_vs_control_cleanup(void);
1405struct ip_vs_dest *
Eric W. Biedermandc2add62015-09-21 13:01:51 -05001406ip_vs_find_dest(struct netns_ipvs *ipvs, int svc_af, int dest_af,
Alex Gartrell655eef12014-09-09 16:40:21 -07001407 const union nf_inet_addr *daddr, __be16 dport,
1408 const union nf_inet_addr *vaddr, __be16 vport,
Julian Anastasov52793db2011-12-30 14:19:02 +09001409 __u16 protocol, __u32 fwmark, __u32 flags);
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001410void ip_vs_try_bind_dest(struct ip_vs_conn *cp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001411
Julian Anastasovfca9c202013-03-22 11:46:38 +02001412static inline void ip_vs_dest_hold(struct ip_vs_dest *dest)
1413{
Reshetova, Elenab54ab922017-03-16 10:03:34 +02001414 refcount_inc(&dest->refcnt);
Julian Anastasovfca9c202013-03-22 11:46:38 +02001415}
1416
1417static inline void ip_vs_dest_put(struct ip_vs_dest *dest)
1418{
Peter Zijlstra4e857c52014-03-17 18:06:10 +01001419 smp_mb__before_atomic();
Reshetova, Elenab54ab922017-03-16 10:03:34 +02001420 refcount_dec(&dest->refcnt);
Julian Anastasovfca9c202013-03-22 11:46:38 +02001421}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001422
Julian Anastasov9e4e9482013-10-09 09:24:27 +03001423static inline void ip_vs_dest_put_and_free(struct ip_vs_dest *dest)
1424{
Reshetova, Elenab54ab922017-03-16 10:03:34 +02001425 if (refcount_dec_and_test(&dest->refcnt))
Julian Anastasov9e4e9482013-10-09 09:24:27 +03001426 kfree(dest);
1427}
1428
Simon Horman07dcc682014-09-30 10:50:06 +09001429/* IPVS sync daemon data and function prototypes
1430 * (from ip_vs_sync.c)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001431 */
Eric W. Biederman6ac121d2015-09-21 13:02:10 -05001432int start_sync_thread(struct netns_ipvs *ipvs, struct ipvs_sync_daemon_cfg *cfg,
Julian Anastasove4ff6752015-07-26 15:03:27 +03001433 int state);
Eric W. Biedermanb3cf3cb2015-09-21 13:02:11 -05001434int stop_sync_thread(struct netns_ipvs *ipvs, int state);
Eric W. Biedermanb61a8c12015-09-21 13:02:17 -05001435void ip_vs_sync_conn(struct netns_ipvs *ipvs, struct ip_vs_conn *cp, int pkts);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001436
Simon Horman07dcc682014-09-30 10:50:06 +09001437/* IPVS rate estimator prototypes (from ip_vs_est.c) */
Eric W. Biederman0f34d542015-09-21 13:02:23 -05001438void ip_vs_start_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
1439void ip_vs_stop_estimator(struct netns_ipvs *ipvs, struct ip_vs_stats *stats);
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001440void ip_vs_zero_estimator(struct ip_vs_stats *stats);
Julian Anastasovcd67cd52015-02-06 09:44:44 +02001441void ip_vs_read_estimator(struct ip_vs_kstats *dst, struct ip_vs_stats *stats);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001442
Simon Horman07dcc682014-09-30 10:50:06 +09001443/* Various IPVS packet transmitters (from ip_vs_xmit.c) */
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001444int ip_vs_null_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1445 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1446int ip_vs_bypass_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1447 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1448int ip_vs_nat_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1449 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1450int ip_vs_tunnel_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1451 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1452int ip_vs_dr_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1453 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1454int ip_vs_icmp_xmit(struct sk_buff *skb, struct ip_vs_conn *cp,
1455 struct ip_vs_protocol *pp, int offset,
1456 unsigned int hooknum, struct ip_vs_iphdr *iph);
1457void ip_vs_dest_dst_rcu_free(struct rcu_head *head);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001458
Julius Volzb3cdd2a2008-09-02 15:55:45 +02001459#ifdef CONFIG_IP_VS_IPV6
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001460int ip_vs_bypass_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1461 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1462int ip_vs_nat_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1463 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1464int ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1465 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1466int ip_vs_dr_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1467 struct ip_vs_protocol *pp, struct ip_vs_iphdr *iph);
1468int ip_vs_icmp_xmit_v6(struct sk_buff *skb, struct ip_vs_conn *cp,
1469 struct ip_vs_protocol *pp, int offset,
1470 unsigned int hooknum, struct ip_vs_iphdr *iph);
Julius Volzb3cdd2a2008-09-02 15:55:45 +02001471#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001472
Simon Horman3a1bbf12011-02-04 18:33:02 +09001473#ifdef CONFIG_SYSCTL
Simon Horman07dcc682014-09-30 10:50:06 +09001474/* This is a simple mechanism to ignore packets when
1475 * we are loaded. Just set ip_vs_drop_rate to 'n' and
1476 * we start to drop 1/rate of the packets
Linus Torvalds1da177e2005-04-16 15:20:36 -07001477 */
Hans Schillstroma0840e22011-01-03 14:44:58 +01001478static inline int ip_vs_todrop(struct netns_ipvs *ipvs)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001479{
Hans Schillstroma0840e22011-01-03 14:44:58 +01001480 if (!ipvs->drop_rate)
1481 return 0;
1482 if (--ipvs->drop_counter > 0)
1483 return 0;
1484 ipvs->drop_counter = ipvs->drop_rate;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001485 return 1;
1486}
Simon Horman3a1bbf12011-02-04 18:33:02 +09001487#else
1488static inline int ip_vs_todrop(struct netns_ipvs *ipvs) { return 0; }
1489#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001490
Simon Horman07dcc682014-09-30 10:50:06 +09001491/* ip_vs_fwd_tag returns the forwarding tag of the connection */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001492#define IP_VS_FWD_METHOD(cp) (cp->flags & IP_VS_CONN_F_FWD_MASK)
1493
Adrian Bunk732db652005-09-01 17:40:26 -07001494static inline char ip_vs_fwd_tag(struct ip_vs_conn *cp)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001495{
1496 char fwd;
1497
1498 switch (IP_VS_FWD_METHOD(cp)) {
1499 case IP_VS_CONN_F_MASQ:
1500 fwd = 'M'; break;
1501 case IP_VS_CONN_F_LOCALNODE:
1502 fwd = 'L'; break;
1503 case IP_VS_CONN_F_TUNNEL:
1504 fwd = 'T'; break;
1505 case IP_VS_CONN_F_DROUTE:
1506 fwd = 'R'; break;
1507 case IP_VS_CONN_F_BYPASS:
1508 fwd = 'B'; break;
1509 default:
1510 fwd = '?'; break;
1511 }
1512 return fwd;
1513}
1514
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001515void ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
1516 struct ip_vs_conn *cp, int dir);
Julius Volzb3cdd2a2008-09-02 15:55:45 +02001517
1518#ifdef CONFIG_IP_VS_IPV6
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001519void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
1520 struct ip_vs_conn *cp, int dir);
Julius Volzb3cdd2a2008-09-02 15:55:45 +02001521#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001522
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001523__sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001524
Al Virof9214b22006-11-16 02:41:18 -08001525static inline __wsum ip_vs_check_diff4(__be32 old, __be32 new, __wsum oldsum)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001526{
Al Virof9214b22006-11-16 02:41:18 -08001527 __be32 diff[2] = { ~old, new };
Linus Torvalds1da177e2005-04-16 15:20:36 -07001528
Joe Perches07f07572008-11-19 15:44:53 -08001529 return csum_partial(diff, sizeof(diff), oldsum);
Al Virof9214b22006-11-16 02:41:18 -08001530}
1531
Julius Volz0bbdd422008-09-02 15:55:42 +02001532#ifdef CONFIG_IP_VS_IPV6
1533static inline __wsum ip_vs_check_diff16(const __be32 *old, const __be32 *new,
1534 __wsum oldsum)
1535{
1536 __be32 diff[8] = { ~old[3], ~old[2], ~old[1], ~old[0],
1537 new[3], new[2], new[1], new[0] };
1538
Joe Perches07f07572008-11-19 15:44:53 -08001539 return csum_partial(diff, sizeof(diff), oldsum);
Julius Volz0bbdd422008-09-02 15:55:42 +02001540}
1541#endif
1542
Al Virof9214b22006-11-16 02:41:18 -08001543static inline __wsum ip_vs_check_diff2(__be16 old, __be16 new, __wsum oldsum)
1544{
1545 __be16 diff[2] = { ~old, new };
1546
Joe Perches07f07572008-11-19 15:44:53 -08001547 return csum_partial(diff, sizeof(diff), oldsum);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001548}
1549
Simon Horman07dcc682014-09-30 10:50:06 +09001550/* Forget current conntrack (unconfirmed) and attach notrack entry */
Julian Anastasovcf356d62010-10-17 16:21:07 +03001551static inline void ip_vs_notrack(struct sk_buff *skb)
1552{
1553#if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)
1554 enum ip_conntrack_info ctinfo;
Jesper Juhl06b69392011-03-09 22:55:05 +01001555 struct nf_conn *ct = nf_ct_get(skb, &ctinfo);
Julian Anastasovcf356d62010-10-17 16:21:07 +03001556
Florian Westphalab8bc7e2017-04-14 20:31:09 +02001557 if (ct) {
Florian Westphal97a6ad12017-01-23 18:21:55 +01001558 nf_conntrack_put(&ct->ct_general);
Florian Westphalcc41c842017-04-14 20:31:08 +02001559 nf_ct_set(skb, NULL, IP_CT_UNTRACKED);
Julian Anastasovcf356d62010-10-17 16:21:07 +03001560 }
1561#endif
1562}
1563
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001564#ifdef CONFIG_IP_VS_NFCT
Simon Horman07dcc682014-09-30 10:50:06 +09001565/* Netfilter connection tracking
1566 * (from ip_vs_nfct.c)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001567 */
Hans Schillstroma0840e22011-01-03 14:44:58 +01001568static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001569{
Simon Hormana4e2f5a2011-02-04 18:33:02 +09001570#ifdef CONFIG_SYSCTL
Hans Schillstroma0840e22011-01-03 14:44:58 +01001571 return ipvs->sysctl_conntrack;
Simon Hormana4e2f5a2011-02-04 18:33:02 +09001572#else
1573 return 0;
1574#endif
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001575}
1576
Joe Perches5c3a0fd2013-09-21 10:22:42 -07001577void ip_vs_update_conntrack(struct sk_buff *skb, struct ip_vs_conn *cp,
1578 int outin);
1579int ip_vs_confirm_conntrack(struct sk_buff *skb);
1580void ip_vs_nfct_expect_related(struct sk_buff *skb, struct nf_conn *ct,
1581 struct ip_vs_conn *cp, u_int8_t proto,
1582 const __be16 port, int from_rs);
1583void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp);
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001584
1585#else
1586
Hans Schillstroma0840e22011-01-03 14:44:58 +01001587static inline int ip_vs_conntrack_enabled(struct netns_ipvs *ipvs)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001588{
1589 return 0;
1590}
1591
1592static inline void ip_vs_update_conntrack(struct sk_buff *skb,
1593 struct ip_vs_conn *cp, int outin)
1594{
1595}
1596
Krzysztof Wilczynskie23ebf02011-10-20 13:18:04 +01001597static inline int ip_vs_confirm_conntrack(struct sk_buff *skb)
Julian Anastasovf4bc17c2010-09-21 17:35:41 +02001598{
1599 return NF_ACCEPT;
1600}
1601
1602static inline void ip_vs_conn_drop_conntrack(struct ip_vs_conn *cp)
1603{
1604}
Simon Horman07dcc682014-09-30 10:50:06 +09001605#endif /* CONFIG_IP_VS_NFCT */
Julian Anastasov6523ce12010-09-05 18:02:29 +00001606
Julian Anastasovf719e372016-03-05 15:03:22 +02001607/* Really using conntrack? */
1608static inline bool ip_vs_conn_uses_conntrack(struct ip_vs_conn *cp,
1609 struct sk_buff *skb)
1610{
1611#ifdef CONFIG_IP_VS_NFCT
1612 enum ip_conntrack_info ctinfo;
1613 struct nf_conn *ct;
1614
1615 if (!(cp->flags & IP_VS_CONN_F_NFCT))
1616 return false;
1617 ct = nf_ct_get(skb, &ctinfo);
Florian Westphalab8bc7e2017-04-14 20:31:09 +02001618 if (ct)
Julian Anastasovf719e372016-03-05 15:03:22 +02001619 return true;
1620#endif
1621 return false;
1622}
1623
Simon Kirbyc16526a2013-08-10 01:26:18 -07001624static inline int
Changli Gaob552f7e2011-02-19 17:32:28 +08001625ip_vs_dest_conn_overhead(struct ip_vs_dest *dest)
1626{
Simon Horman07dcc682014-09-30 10:50:06 +09001627 /* We think the overhead of processing active connections is 256
Changli Gaob552f7e2011-02-19 17:32:28 +08001628 * times higher than that of inactive connections in average. (This
1629 * 256 times might not be accurate, we will change it later) We
1630 * use the following formula to estimate the overhead now:
1631 * dest->activeconns*256 + dest->inactconns
1632 */
1633 return (atomic_read(&dest->activeconns) << 8) +
1634 atomic_read(&dest->inactconns);
1635}
1636
Julius Volzbc4768e2008-07-31 20:45:24 -07001637#endif /* _NET_IP_VS_H */