aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--include/linux/udp.h10
-rw-r--r--include/net/ipv6.h5
-rw-r--r--include/net/netns/ipv6.h3
-rw-r--r--include/net/tipc/tipc_bearer.h9
-rw-r--r--include/net/tipc/tipc_port.h7
-rw-r--r--include/net/transp_v6.h5
-rw-r--r--include/net/udplite.h9
-rw-r--r--net/ipv4/Kconfig10
-rw-r--r--net/ipv4/Makefile3
-rw-r--r--net/ipv4/af_inet.c7
-rw-r--r--net/ipv4/proc.c5
-rw-r--r--net/ipv4/udp.c1090
-rw-r--r--net/ipv4/udp_ipv4.c1134
-rw-r--r--net/ipv4/udplite.c (renamed from net/ipv4/udplite_ipv4.c)0
-rw-r--r--net/ipv6/Makefile3
-rw-r--r--net/ipv6/af_inet6.c14
-rw-r--r--net/ipv6/ip6_output.c10
-rw-r--r--net/ipv6/ipv6_sockglue.c9
-rw-r--r--net/ipv6/mcast.c116
-rw-r--r--net/ipv6/ndisc.c101
-rw-r--r--net/ipv6/proc.c6
-rw-r--r--net/ipv6/route.c2
-rw-r--r--net/ipv6/tcp_ipv6.c45
-rw-r--r--net/ipv6/udp.c (renamed from net/ipv6/udp_ipv6.c)19
-rw-r--r--net/ipv6/udplite.c (renamed from net/ipv6/udplite_ipv6.c)0
-rw-r--r--net/tipc/core.c11
-rw-r--r--net/tipc/core.h6
-rw-r--r--net/tipc/link.c16
-rw-r--r--net/tipc/msg.c16
-rw-r--r--net/tipc/msg.h50
-rw-r--r--net/tipc/port.c12
-rw-r--r--net/tipc/socket.c5
32 files changed, 1345 insertions, 1393 deletions
diff --git a/include/linux/udp.h b/include/linux/udp.h
index 4144664d69d9..1e7b7cb5703b 100644
--- a/include/linux/udp.h
+++ b/include/linux/udp.h
@@ -70,10 +70,8 @@ struct udp_sock {
70#define UDPLITE_BIT 0x1 /* set by udplite proto init function */ 70#define UDPLITE_BIT 0x1 /* set by udplite proto init function */
71#define UDPLITE_SEND_CC 0x2 /* set via udplite setsockopt */ 71#define UDPLITE_SEND_CC 0x2 /* set via udplite setsockopt */
72#define UDPLITE_RECV_CC 0x4 /* set via udplite setsocktopt */ 72#define UDPLITE_RECV_CC 0x4 /* set via udplite setsocktopt */
73#ifdef CONFIG_IP_UDPLITE
74 __u8 pcflag; /* marks socket as UDP-Lite if > 0 */ 73 __u8 pcflag; /* marks socket as UDP-Lite if > 0 */
75 __u8 unused[3]; 74 __u8 unused[3];
76#endif
77 /* 75 /*
78 * For encapsulation sockets. 76 * For encapsulation sockets.
79 */ 77 */
@@ -85,15 +83,7 @@ static inline struct udp_sock *udp_sk(const struct sock *sk)
85 return (struct udp_sock *)sk; 83 return (struct udp_sock *)sk;
86} 84}
87 85
88#ifdef CONFIG_IP_UDPLITE
89#define IS_UDPLITE(__sk) (udp_sk(__sk)->pcflag) 86#define IS_UDPLITE(__sk) (udp_sk(__sk)->pcflag)
90#define IS_PROTO_UDPLITE(__proto) ((__proto) == IPPROTO_UDPLITE)
91#define IS_SOL_UDPFAMILY(level) ((level) == SOL_UDP || (level) == SOL_UDPLITE)
92#else
93#define IS_UDPLITE(__sk) 0
94#define IS_PROTO_UDPLITE(__proto) 0
95#define IS_SOL_UDPFAMILY(level) ((level) == SOL_UDP)
96#endif
97 87
98#endif 88#endif
99 89
diff --git a/include/net/ipv6.h b/include/net/ipv6.h
index 5f6df50a33a9..8db06af1efbb 100644
--- a/include/net/ipv6.h
+++ b/include/net/ipv6.h
@@ -599,13 +599,8 @@ extern int tcp6_proc_init(void);
599extern void tcp6_proc_exit(void); 599extern void tcp6_proc_exit(void);
600extern int udp6_proc_init(void); 600extern int udp6_proc_init(void);
601extern void udp6_proc_exit(void); 601extern void udp6_proc_exit(void);
602#ifdef CONFIG_IP_UDPLITE
603extern int udplite6_proc_init(void); 602extern int udplite6_proc_init(void);
604extern void udplite6_proc_exit(void); 603extern void udplite6_proc_exit(void);
605#else
606static inline int udplite6_proc_init(void) { return 0; }
607static inline void udplite6_proc_exit(void) { }
608#endif
609extern int ipv6_misc_proc_init(void); 604extern int ipv6_misc_proc_init(void);
610extern void ipv6_misc_proc_exit(void); 605extern void ipv6_misc_proc_exit(void);
611extern int snmp6_register_dev(struct inet6_dev *idev); 606extern int snmp6_register_dev(struct inet6_dev *idev);
diff --git a/include/net/netns/ipv6.h b/include/net/netns/ipv6.h
index 90e6e24df858..ac053be6c256 100644
--- a/include/net/netns/ipv6.h
+++ b/include/net/netns/ipv6.h
@@ -51,5 +51,8 @@ struct netns_ipv6 {
51 struct fib_rules_ops *fib6_rules_ops; 51 struct fib_rules_ops *fib6_rules_ops;
52#endif 52#endif
53 struct sock **icmp_sk; 53 struct sock **icmp_sk;
54 struct sock *ndisc_sk;
55 struct sock *tcp_sk;
56 struct sock *igmp_sk;
54}; 57};
55#endif 58#endif
diff --git a/include/net/tipc/tipc_bearer.h b/include/net/tipc/tipc_bearer.h
index 2151a80cdf30..ee2f304e4919 100644
--- a/include/net/tipc/tipc_bearer.h
+++ b/include/net/tipc/tipc_bearer.h
@@ -99,6 +99,9 @@ struct tipc_bearer {
99 char name[TIPC_MAX_BEARER_NAME]; 99 char name[TIPC_MAX_BEARER_NAME];
100}; 100};
101 101
102/*
103 * TIPC routines available to supported media types
104 */
102 105
103int tipc_register_media(u32 media_type, 106int tipc_register_media(u32 media_type,
104 char *media_name, 107 char *media_name,
@@ -123,6 +126,12 @@ void tipc_continue(struct tipc_bearer *tb_ptr);
123int tipc_enable_bearer(const char *bearer_name, u32 bcast_scope, u32 priority); 126int tipc_enable_bearer(const char *bearer_name, u32 bcast_scope, u32 priority);
124int tipc_disable_bearer(const char *name); 127int tipc_disable_bearer(const char *name);
125 128
129/*
130 * Routines made available to TIPC by supported media types
131 */
132
133int tipc_eth_media_start(void);
134void tipc_eth_media_stop(void);
126 135
127#endif 136#endif
128 137
diff --git a/include/net/tipc/tipc_port.h b/include/net/tipc/tipc_port.h
index cfc4ba46de8f..c9b36b77a0b9 100644
--- a/include/net/tipc/tipc_port.h
+++ b/include/net/tipc/tipc_port.h
@@ -86,13 +86,6 @@ u32 tipc_createport_raw(void *usr_handle,
86 void (*wakeup)(struct tipc_port *), 86 void (*wakeup)(struct tipc_port *),
87 const u32 importance); 87 const u32 importance);
88 88
89/*
90 * tipc_set_msg_option(): port must be locked.
91 */
92int tipc_set_msg_option(struct tipc_port *tp_ptr,
93 const char *opt,
94 const u32 len);
95
96int tipc_reject_msg(struct sk_buff *buf, u32 err); 89int tipc_reject_msg(struct sk_buff *buf, u32 err);
97 90
98int tipc_send_buf_fast(struct sk_buff *buf, u32 destnode); 91int tipc_send_buf_fast(struct sk_buff *buf, u32 destnode);
diff --git a/include/net/transp_v6.h b/include/net/transp_v6.h
index 902e6c6bc793..27394e0447d8 100644
--- a/include/net/transp_v6.h
+++ b/include/net/transp_v6.h
@@ -27,13 +27,8 @@ extern int rawv6_init(void);
27extern void rawv6_exit(void); 27extern void rawv6_exit(void);
28extern int udpv6_init(void); 28extern int udpv6_init(void);
29extern void udpv6_exit(void); 29extern void udpv6_exit(void);
30#ifdef CONFIG_IP_UDPLITE
31extern int udplitev6_init(void); 30extern int udplitev6_init(void);
32extern void udplitev6_exit(void); 31extern void udplitev6_exit(void);
33#else
34static inline int udplitev6_init(void) { return 0; }
35static inline void udplitev6_exit(void) { }
36#endif
37extern int tcpv6_init(void); 32extern int tcpv6_init(void);
38extern void tcpv6_exit(void); 33extern void tcpv6_exit(void);
39 34
diff --git a/include/net/udplite.h b/include/net/udplite.h
index 01ddb2c20264..b76b2e377af4 100644
--- a/include/net/udplite.h
+++ b/include/net/udplite.h
@@ -25,9 +25,7 @@ static __inline__ int udplite_getfrag(void *from, char *to, int offset,
25/* Designate sk as UDP-Lite socket */ 25/* Designate sk as UDP-Lite socket */
26static inline int udplite_sk_init(struct sock *sk) 26static inline int udplite_sk_init(struct sock *sk)
27{ 27{
28#ifdef CONFIG_IP_UDPLITE
29 udp_sk(sk)->pcflag = UDPLITE_BIT; 28 udp_sk(sk)->pcflag = UDPLITE_BIT;
30#endif
31 return 0; 29 return 0;
32} 30}
33 31
@@ -71,7 +69,7 @@ static inline int udplite_checksum_init(struct sk_buff *skb, struct udphdr *uh)
71static inline int udplite_sender_cscov(struct udp_sock *up, struct udphdr *uh) 69static inline int udplite_sender_cscov(struct udp_sock *up, struct udphdr *uh)
72{ 70{
73 int cscov = up->len; 71 int cscov = up->len;
74#ifdef CONFIG_IP_UDPLITE 72
75 /* 73 /*
76 * Sender has set `partial coverage' option on UDP-Lite socket 74 * Sender has set `partial coverage' option on UDP-Lite socket
77 */ 75 */
@@ -95,15 +93,13 @@ static inline int udplite_sender_cscov(struct udp_sock *up, struct udphdr *uh)
95 * illegal, we fall back to the defaults here. 93 * illegal, we fall back to the defaults here.
96 */ 94 */
97 } 95 }
98#endif
99 return cscov; 96 return cscov;
100} 97}
101 98
102static inline __wsum udplite_csum_outgoing(struct sock *sk, struct sk_buff *skb) 99static inline __wsum udplite_csum_outgoing(struct sock *sk, struct sk_buff *skb)
103{ 100{
104 __wsum csum = 0;
105#ifdef CONFIG_IP_UDPLITE
106 int cscov = udplite_sender_cscov(udp_sk(sk), udp_hdr(skb)); 101 int cscov = udplite_sender_cscov(udp_sk(sk), udp_hdr(skb));
102 __wsum csum = 0;
107 103
108 skb->ip_summed = CHECKSUM_NONE; /* no HW support for checksumming */ 104 skb->ip_summed = CHECKSUM_NONE; /* no HW support for checksumming */
109 105
@@ -116,7 +112,6 @@ static inline __wsum udplite_csum_outgoing(struct sock *sk, struct sk_buff *skb)
116 if ((cscov -= len) <= 0) 112 if ((cscov -= len) <= 0)
117 break; 113 break;
118 } 114 }
119#endif
120 return csum; 115 return csum;
121} 116}
122 117
diff --git a/net/ipv4/Kconfig b/net/ipv4/Kconfig
index 5098fd2ff4d0..9c7e5ffb223d 100644
--- a/net/ipv4/Kconfig
+++ b/net/ipv4/Kconfig
@@ -632,15 +632,5 @@ config TCP_MD5SIG
632 632
633 If unsure, say N. 633 If unsure, say N.
634 634
635config IP_UDPLITE
636 bool "IP: UDP-Lite Protocol (RFC 3828)"
637 default n
638 ---help---
639 UDP-Lite (RFC 3828) is a UDP-like protocol with variable-length
640 checksum. Read <file:Documentation/networking/udplite.txt> for
641 details.
642
643 If unsure, say N.
644
645source "net/ipv4/ipvs/Kconfig" 635source "net/ipv4/ipvs/Kconfig"
646 636
diff --git a/net/ipv4/Makefile b/net/ipv4/Makefile
index d5226241d5ed..ad40ef3f9ebc 100644
--- a/net/ipv4/Makefile
+++ b/net/ipv4/Makefile
@@ -8,7 +8,7 @@ obj-y := route.o inetpeer.o protocol.o \
8 inet_timewait_sock.o inet_connection_sock.o \ 8 inet_timewait_sock.o inet_connection_sock.o \
9 tcp.o tcp_input.o tcp_output.o tcp_timer.o tcp_ipv4.o \ 9 tcp.o tcp_input.o tcp_output.o tcp_timer.o tcp_ipv4.o \
10 tcp_minisocks.o tcp_cong.o \ 10 tcp_minisocks.o tcp_cong.o \
11 datagram.o raw.o udp.o udp_ipv4.o \ 11 datagram.o raw.o udp.o udplite.o \
12 arp.o icmp.o devinet.o af_inet.o igmp.o \ 12 arp.o icmp.o devinet.o af_inet.o igmp.o \
13 fib_frontend.o fib_semantics.o \ 13 fib_frontend.o fib_semantics.o \
14 inet_fragment.o 14 inet_fragment.o
@@ -49,7 +49,6 @@ obj-$(CONFIG_TCP_CONG_SCALABLE) += tcp_scalable.o
49obj-$(CONFIG_TCP_CONG_LP) += tcp_lp.o 49obj-$(CONFIG_TCP_CONG_LP) += tcp_lp.o
50obj-$(CONFIG_TCP_CONG_YEAH) += tcp_yeah.o 50obj-$(CONFIG_TCP_CONG_YEAH) += tcp_yeah.o
51obj-$(CONFIG_TCP_CONG_ILLINOIS) += tcp_illinois.o 51obj-$(CONFIG_TCP_CONG_ILLINOIS) += tcp_illinois.o
52obj-$(CONFIG_IP_UDPLITE) += udplite_ipv4.o
53obj-$(CONFIG_NETLABEL) += cipso_ipv4.o 52obj-$(CONFIG_NETLABEL) += cipso_ipv4.o
54 53
55obj-$(CONFIG_XFRM) += xfrm4_policy.o xfrm4_state.o xfrm4_input.o \ 54obj-$(CONFIG_XFRM) += xfrm4_policy.o xfrm4_state.o xfrm4_input.o \
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 25871c6c7444..4cb8a1385539 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1317,18 +1317,15 @@ static int __init init_ipv4_mibs(void)
1317 if (snmp_mib_init((void **)udp_statistics, 1317 if (snmp_mib_init((void **)udp_statistics,
1318 sizeof(struct udp_mib)) < 0) 1318 sizeof(struct udp_mib)) < 0)
1319 goto err_udp_mib; 1319 goto err_udp_mib;
1320#ifdef CONFIG_IP_UDPLITE
1321 if (snmp_mib_init((void **)udplite_statistics, 1320 if (snmp_mib_init((void **)udplite_statistics,
1322 sizeof(struct udp_mib)) < 0) 1321 sizeof(struct udp_mib)) < 0)
1323 goto err_udplite_mib; 1322 goto err_udplite_mib;
1324#endif 1323
1325 tcp_mib_init(); 1324 tcp_mib_init();
1326 1325
1327 return 0; 1326 return 0;
1328 1327
1329#ifdef CONFIG_IP_UDPLITE
1330err_udplite_mib: 1328err_udplite_mib:
1331#endif
1332 snmp_mib_free((void **)udp_statistics); 1329 snmp_mib_free((void **)udp_statistics);
1333err_udp_mib: 1330err_udp_mib:
1334 snmp_mib_free((void **)tcp_statistics); 1331 snmp_mib_free((void **)tcp_statistics);
@@ -1426,10 +1423,8 @@ static int __init inet_init(void)
1426 /* Setup UDP memory threshold */ 1423 /* Setup UDP memory threshold */
1427 udp_init(); 1424 udp_init();
1428 1425
1429#ifdef CONFIG_IP_UDPLITE
1430 /* Add UDP-Lite (RFC 3828) */ 1426 /* Add UDP-Lite (RFC 3828) */
1431 udplite4_register(); 1427 udplite4_register();
1432#endif
1433 1428
1434 /* 1429 /*
1435 * Set the ICMP layer up 1430 * Set the ICMP layer up
diff --git a/net/ipv4/proc.c b/net/ipv4/proc.c
index d75ddb7fa4b8..d63474c6b400 100644
--- a/net/ipv4/proc.c
+++ b/net/ipv4/proc.c
@@ -59,9 +59,7 @@ static int sockstat_seq_show(struct seq_file *seq, void *v)
59 atomic_read(&tcp_memory_allocated)); 59 atomic_read(&tcp_memory_allocated));
60 seq_printf(seq, "UDP: inuse %d mem %d\n", sock_prot_inuse_get(&udp_prot), 60 seq_printf(seq, "UDP: inuse %d mem %d\n", sock_prot_inuse_get(&udp_prot),
61 atomic_read(&udp_memory_allocated)); 61 atomic_read(&udp_memory_allocated));
62#ifdef CONFIG_IP_UDPLITE
63 seq_printf(seq, "UDPLITE: inuse %d\n", sock_prot_inuse_get(&udplite_prot)); 62 seq_printf(seq, "UDPLITE: inuse %d\n", sock_prot_inuse_get(&udplite_prot));
64#endif
65 seq_printf(seq, "RAW: inuse %d\n", sock_prot_inuse_get(&raw_prot)); 63 seq_printf(seq, "RAW: inuse %d\n", sock_prot_inuse_get(&raw_prot));
66 seq_printf(seq, "FRAG: inuse %d memory %d\n", 64 seq_printf(seq, "FRAG: inuse %d memory %d\n",
67 ip_frag_nqueues(&init_net), ip_frag_mem(&init_net)); 65 ip_frag_nqueues(&init_net), ip_frag_mem(&init_net));
@@ -351,7 +349,6 @@ static int snmp_seq_show(struct seq_file *seq, void *v)
351 snmp_fold_field((void **)udp_statistics, 349 snmp_fold_field((void **)udp_statistics,
352 snmp4_udp_list[i].entry)); 350 snmp4_udp_list[i].entry));
353 351
354#ifdef CONFIG_IP_UDPLITE
355 /* the UDP and UDP-Lite MIBs are the same */ 352 /* the UDP and UDP-Lite MIBs are the same */
356 seq_puts(seq, "\nUdpLite:"); 353 seq_puts(seq, "\nUdpLite:");
357 for (i = 0; snmp4_udp_list[i].name != NULL; i++) 354 for (i = 0; snmp4_udp_list[i].name != NULL; i++)
@@ -362,7 +359,7 @@ static int snmp_seq_show(struct seq_file *seq, void *v)
362 seq_printf(seq, " %lu", 359 seq_printf(seq, " %lu",
363 snmp_fold_field((void **)udplite_statistics, 360 snmp_fold_field((void **)udplite_statistics,
364 snmp4_udp_list[i].entry)); 361 snmp4_udp_list[i].entry));
365#endif 362
366 seq_putc(seq, '\n'); 363 seq_putc(seq, '\n');
367 return 0; 364 return 0;
368} 365}
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index c53d7673b57d..7ea1b67b6de1 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -246,6 +246,553 @@ int udp_get_port(struct sock *sk, unsigned short snum,
246 return __udp_lib_get_port(sk, snum, udp_hash, scmp); 246 return __udp_lib_get_port(sk, snum, udp_hash, scmp);
247} 247}
248 248
249int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
250{
251 struct inet_sock *inet1 = inet_sk(sk1), *inet2 = inet_sk(sk2);
252
253 return ( !ipv6_only_sock(sk2) &&
254 (!inet1->rcv_saddr || !inet2->rcv_saddr ||
255 inet1->rcv_saddr == inet2->rcv_saddr ));
256}
257
258static inline int udp_v4_get_port(struct sock *sk, unsigned short snum)
259{
260 return udp_get_port(sk, snum, ipv4_rcv_saddr_equal);
261}
262
263/* UDP is nearly always wildcards out the wazoo, it makes no sense to try
264 * harder than this. -DaveM
265 */
266static struct sock *__udp4_lib_lookup(struct net *net, __be32 saddr,
267 __be16 sport, __be32 daddr, __be16 dport,
268 int dif, struct hlist_head udptable[])
269{
270 struct sock *sk, *result = NULL;
271 struct hlist_node *node;
272 unsigned short hnum = ntohs(dport);
273 int badness = -1;
274
275 read_lock(&udp_hash_lock);
276 sk_for_each(sk, node, &udptable[hnum & (UDP_HTABLE_SIZE - 1)]) {
277 struct inet_sock *inet = inet_sk(sk);
278
279 if (sk->sk_net == net && sk->sk_hash == hnum &&
280 !ipv6_only_sock(sk)) {
281 int score = (sk->sk_family == PF_INET ? 1 : 0);
282 if (inet->rcv_saddr) {
283 if (inet->rcv_saddr != daddr)
284 continue;
285 score+=2;
286 }
287 if (inet->daddr) {
288 if (inet->daddr != saddr)
289 continue;
290 score+=2;
291 }
292 if (inet->dport) {
293 if (inet->dport != sport)
294 continue;
295 score+=2;
296 }
297 if (sk->sk_bound_dev_if) {
298 if (sk->sk_bound_dev_if != dif)
299 continue;
300 score+=2;
301 }
302 if (score == 9) {
303 result = sk;
304 break;
305 } else if (score > badness) {
306 result = sk;
307 badness = score;
308 }
309 }
310 }
311 if (result)
312 sock_hold(result);
313 read_unlock(&udp_hash_lock);
314 return result;
315}
316
317static inline struct sock *udp_v4_mcast_next(struct sock *sk,
318 __be16 loc_port, __be32 loc_addr,
319 __be16 rmt_port, __be32 rmt_addr,
320 int dif)
321{
322 struct hlist_node *node;
323 struct sock *s = sk;
324 unsigned short hnum = ntohs(loc_port);
325
326 sk_for_each_from(s, node) {
327 struct inet_sock *inet = inet_sk(s);
328
329 if (s->sk_hash != hnum ||
330 (inet->daddr && inet->daddr != rmt_addr) ||
331 (inet->dport != rmt_port && inet->dport) ||
332 (inet->rcv_saddr && inet->rcv_saddr != loc_addr) ||
333 ipv6_only_sock(s) ||
334 (s->sk_bound_dev_if && s->sk_bound_dev_if != dif))
335 continue;
336 if (!ip_mc_sf_allow(s, loc_addr, rmt_addr, dif))
337 continue;
338 goto found;
339 }
340 s = NULL;
341found:
342 return s;
343}
344
345/*
346 * This routine is called by the ICMP module when it gets some
347 * sort of error condition. If err < 0 then the socket should
348 * be closed and the error returned to the user. If err > 0
349 * it's just the icmp type << 8 | icmp code.
350 * Header points to the ip header of the error packet. We move
351 * on past this. Then (as it used to claim before adjustment)
352 * header points to the first 8 bytes of the udp header. We need
353 * to find the appropriate port.
354 */
355
356void __udp4_lib_err(struct sk_buff *skb, u32 info, struct hlist_head udptable[])
357{
358 struct inet_sock *inet;
359 struct iphdr *iph = (struct iphdr*)skb->data;
360 struct udphdr *uh = (struct udphdr*)(skb->data+(iph->ihl<<2));
361 const int type = icmp_hdr(skb)->type;
362 const int code = icmp_hdr(skb)->code;
363 struct sock *sk;
364 int harderr;
365 int err;
366
367 sk = __udp4_lib_lookup(skb->dev->nd_net, iph->daddr, uh->dest,
368 iph->saddr, uh->source, skb->dev->ifindex, udptable);
369 if (sk == NULL) {
370 ICMP_INC_STATS_BH(ICMP_MIB_INERRORS);
371 return; /* No socket for error */
372 }
373
374 err = 0;
375 harderr = 0;
376 inet = inet_sk(sk);
377
378 switch (type) {
379 default:
380 case ICMP_TIME_EXCEEDED:
381 err = EHOSTUNREACH;
382 break;
383 case ICMP_SOURCE_QUENCH:
384 goto out;
385 case ICMP_PARAMETERPROB:
386 err = EPROTO;
387 harderr = 1;
388 break;
389 case ICMP_DEST_UNREACH:
390 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
391 if (inet->pmtudisc != IP_PMTUDISC_DONT) {
392 err = EMSGSIZE;
393 harderr = 1;
394 break;
395 }
396 goto out;
397 }
398 err = EHOSTUNREACH;
399 if (code <= NR_ICMP_UNREACH) {
400 harderr = icmp_err_convert[code].fatal;
401 err = icmp_err_convert[code].errno;
402 }
403 break;
404 }
405
406 /*
407 * RFC1122: OK. Passes ICMP errors back to application, as per
408 * 4.1.3.3.
409 */
410 if (!inet->recverr) {
411 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
412 goto out;
413 } else {
414 ip_icmp_error(sk, skb, err, uh->dest, info, (u8*)(uh+1));
415 }
416 sk->sk_err = err;
417 sk->sk_error_report(sk);
418out:
419 sock_put(sk);
420}
421
422void udp_err(struct sk_buff *skb, u32 info)
423{
424 __udp4_lib_err(skb, info, udp_hash);
425}
426
427/*
428 * Throw away all pending data and cancel the corking. Socket is locked.
429 */
430static void udp_flush_pending_frames(struct sock *sk)
431{
432 struct udp_sock *up = udp_sk(sk);
433
434 if (up->pending) {
435 up->len = 0;
436 up->pending = 0;
437 ip_flush_pending_frames(sk);
438 }
439}
440
441/**
442 * udp4_hwcsum_outgoing - handle outgoing HW checksumming
443 * @sk: socket we are sending on
444 * @skb: sk_buff containing the filled-in UDP header
445 * (checksum field must be zeroed out)
446 */
447static void udp4_hwcsum_outgoing(struct sock *sk, struct sk_buff *skb,
448 __be32 src, __be32 dst, int len )
449{
450 unsigned int offset;
451 struct udphdr *uh = udp_hdr(skb);
452 __wsum csum = 0;
453
454 if (skb_queue_len(&sk->sk_write_queue) == 1) {
455 /*
456 * Only one fragment on the socket.
457 */
458 skb->csum_start = skb_transport_header(skb) - skb->head;
459 skb->csum_offset = offsetof(struct udphdr, check);
460 uh->check = ~csum_tcpudp_magic(src, dst, len, IPPROTO_UDP, 0);
461 } else {
462 /*
463 * HW-checksum won't work as there are two or more
464 * fragments on the socket so that all csums of sk_buffs
465 * should be together
466 */
467 offset = skb_transport_offset(skb);
468 skb->csum = skb_checksum(skb, offset, skb->len - offset, 0);
469
470 skb->ip_summed = CHECKSUM_NONE;
471
472 skb_queue_walk(&sk->sk_write_queue, skb) {
473 csum = csum_add(csum, skb->csum);
474 }
475
476 uh->check = csum_tcpudp_magic(src, dst, len, IPPROTO_UDP, csum);
477 if (uh->check == 0)
478 uh->check = CSUM_MANGLED_0;
479 }
480}
481
482/*
483 * Push out all pending data as one UDP datagram. Socket is locked.
484 */
485static int udp_push_pending_frames(struct sock *sk)
486{
487 struct udp_sock *up = udp_sk(sk);
488 struct inet_sock *inet = inet_sk(sk);
489 struct flowi *fl = &inet->cork.fl;
490 struct sk_buff *skb;
491 struct udphdr *uh;
492 int err = 0;
493 int is_udplite = IS_UDPLITE(sk);
494 __wsum csum = 0;
495
496 /* Grab the skbuff where UDP header space exists. */
497 if ((skb = skb_peek(&sk->sk_write_queue)) == NULL)
498 goto out;
499
500 /*
501 * Create a UDP header
502 */
503 uh = udp_hdr(skb);
504 uh->source = fl->fl_ip_sport;
505 uh->dest = fl->fl_ip_dport;
506 uh->len = htons(up->len);
507 uh->check = 0;
508
509 if (is_udplite) /* UDP-Lite */
510 csum = udplite_csum_outgoing(sk, skb);
511
512 else if (sk->sk_no_check == UDP_CSUM_NOXMIT) { /* UDP csum disabled */
513
514 skb->ip_summed = CHECKSUM_NONE;
515 goto send;
516
517 } else if (skb->ip_summed == CHECKSUM_PARTIAL) { /* UDP hardware csum */
518
519 udp4_hwcsum_outgoing(sk, skb, fl->fl4_src,fl->fl4_dst, up->len);
520 goto send;
521
522 } else /* `normal' UDP */
523 csum = udp_csum_outgoing(sk, skb);
524
525 /* add protocol-dependent pseudo-header */
526 uh->check = csum_tcpudp_magic(fl->fl4_src, fl->fl4_dst, up->len,
527 sk->sk_protocol, csum );
528 if (uh->check == 0)
529 uh->check = CSUM_MANGLED_0;
530
531send:
532 err = ip_push_pending_frames(sk);
533out:
534 up->len = 0;
535 up->pending = 0;
536 if (!err)
537 UDP_INC_STATS_USER(UDP_MIB_OUTDATAGRAMS, is_udplite);
538 return err;
539}
540
541int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
542 size_t len)
543{
544 struct inet_sock *inet = inet_sk(sk);
545 struct udp_sock *up = udp_sk(sk);
546 int ulen = len;
547 struct ipcm_cookie ipc;
548 struct rtable *rt = NULL;
549 int free = 0;
550 int connected = 0;
551 __be32 daddr, faddr, saddr;
552 __be16 dport;
553 u8 tos;
554 int err, is_udplite = IS_UDPLITE(sk);
555 int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
556 int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
557
558 if (len > 0xFFFF)
559 return -EMSGSIZE;
560
561 /*
562 * Check the flags.
563 */
564
565 if (msg->msg_flags&MSG_OOB) /* Mirror BSD error message compatibility */
566 return -EOPNOTSUPP;
567
568 ipc.opt = NULL;
569
570 if (up->pending) {
571 /*
572 * There are pending frames.
573 * The socket lock must be held while it's corked.
574 */
575 lock_sock(sk);
576 if (likely(up->pending)) {
577 if (unlikely(up->pending != AF_INET)) {
578 release_sock(sk);
579 return -EINVAL;
580 }
581 goto do_append_data;
582 }
583 release_sock(sk);
584 }
585 ulen += sizeof(struct udphdr);
586
587 /*
588 * Get and verify the address.
589 */
590 if (msg->msg_name) {
591 struct sockaddr_in * usin = (struct sockaddr_in*)msg->msg_name;
592 if (msg->msg_namelen < sizeof(*usin))
593 return -EINVAL;
594 if (usin->sin_family != AF_INET) {
595 if (usin->sin_family != AF_UNSPEC)
596 return -EAFNOSUPPORT;
597 }
598
599 daddr = usin->sin_addr.s_addr;
600 dport = usin->sin_port;
601 if (dport == 0)
602 return -EINVAL;
603 } else {
604 if (sk->sk_state != TCP_ESTABLISHED)
605 return -EDESTADDRREQ;
606 daddr = inet->daddr;
607 dport = inet->dport;
608 /* Open fast path for connected socket.
609 Route will not be used, if at least one option is set.
610 */
611 connected = 1;
612 }
613 ipc.addr = inet->saddr;
614
615 ipc.oif = sk->sk_bound_dev_if;
616 if (msg->msg_controllen) {
617 err = ip_cmsg_send(msg, &ipc);
618 if (err)
619 return err;
620 if (ipc.opt)
621 free = 1;
622 connected = 0;
623 }
624 if (!ipc.opt)
625 ipc.opt = inet->opt;
626
627 saddr = ipc.addr;
628 ipc.addr = faddr = daddr;
629
630 if (ipc.opt && ipc.opt->srr) {
631 if (!daddr)
632 return -EINVAL;
633 faddr = ipc.opt->faddr;
634 connected = 0;
635 }
636 tos = RT_TOS(inet->tos);
637 if (sock_flag(sk, SOCK_LOCALROUTE) ||
638 (msg->msg_flags & MSG_DONTROUTE) ||
639 (ipc.opt && ipc.opt->is_strictroute)) {
640 tos |= RTO_ONLINK;
641 connected = 0;
642 }
643
644 if (ipv4_is_multicast(daddr)) {
645 if (!ipc.oif)
646 ipc.oif = inet->mc_index;
647 if (!saddr)
648 saddr = inet->mc_addr;
649 connected = 0;
650 }
651
652 if (connected)
653 rt = (struct rtable*)sk_dst_check(sk, 0);
654
655 if (rt == NULL) {
656 struct flowi fl = { .oif = ipc.oif,
657 .nl_u = { .ip4_u =
658 { .daddr = faddr,
659 .saddr = saddr,
660 .tos = tos } },
661 .proto = sk->sk_protocol,
662 .uli_u = { .ports =
663 { .sport = inet->sport,
664 .dport = dport } } };
665 security_sk_classify_flow(sk, &fl);
666 err = ip_route_output_flow(&init_net, &rt, &fl, sk, 1);
667 if (err) {
668 if (err == -ENETUNREACH)
669 IP_INC_STATS_BH(IPSTATS_MIB_OUTNOROUTES);
670 goto out;
671 }
672
673 err = -EACCES;
674 if ((rt->rt_flags & RTCF_BROADCAST) &&
675 !sock_flag(sk, SOCK_BROADCAST))
676 goto out;
677 if (connected)
678 sk_dst_set(sk, dst_clone(&rt->u.dst));
679 }
680
681 if (msg->msg_flags&MSG_CONFIRM)
682 goto do_confirm;
683back_from_confirm:
684
685 saddr = rt->rt_src;
686 if (!ipc.addr)
687 daddr = ipc.addr = rt->rt_dst;
688
689 lock_sock(sk);
690 if (unlikely(up->pending)) {
691 /* The socket is already corked while preparing it. */
692 /* ... which is an evident application bug. --ANK */
693 release_sock(sk);
694
695 LIMIT_NETDEBUG(KERN_DEBUG "udp cork app bug 2\n");
696 err = -EINVAL;
697 goto out;
698 }
699 /*
700 * Now cork the socket to pend data.
701 */
702 inet->cork.fl.fl4_dst = daddr;
703 inet->cork.fl.fl_ip_dport = dport;
704 inet->cork.fl.fl4_src = saddr;
705 inet->cork.fl.fl_ip_sport = inet->sport;
706 up->pending = AF_INET;
707
708do_append_data:
709 up->len += ulen;
710 getfrag = is_udplite ? udplite_getfrag : ip_generic_getfrag;
711 err = ip_append_data(sk, getfrag, msg->msg_iov, ulen,
712 sizeof(struct udphdr), &ipc, rt,
713 corkreq ? msg->msg_flags|MSG_MORE : msg->msg_flags);
714 if (err)
715 udp_flush_pending_frames(sk);
716 else if (!corkreq)
717 err = udp_push_pending_frames(sk);
718 else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
719 up->pending = 0;
720 release_sock(sk);
721
722out:
723 ip_rt_put(rt);
724 if (free)
725 kfree(ipc.opt);
726 if (!err)
727 return len;
728 /*
729 * ENOBUFS = no kernel mem, SOCK_NOSPACE = no sndbuf space. Reporting
730 * ENOBUFS might not be good (it's not tunable per se), but otherwise
731 * we don't have a good statistic (IpOutDiscards but it can be too many
732 * things). We could add another new stat but at least for now that
733 * seems like overkill.
734 */
735 if (err == -ENOBUFS || test_bit(SOCK_NOSPACE, &sk->sk_socket->flags)) {
736 UDP_INC_STATS_USER(UDP_MIB_SNDBUFERRORS, is_udplite);
737 }
738 return err;
739
740do_confirm:
741 dst_confirm(&rt->u.dst);
742 if (!(msg->msg_flags&MSG_PROBE) || len)
743 goto back_from_confirm;
744 err = 0;
745 goto out;
746}
747
748int udp_sendpage(struct sock *sk, struct page *page, int offset,
749 size_t size, int flags)
750{
751 struct udp_sock *up = udp_sk(sk);
752 int ret;
753
754 if (!up->pending) {
755 struct msghdr msg = { .msg_flags = flags|MSG_MORE };
756
757 /* Call udp_sendmsg to specify destination address which
758 * sendpage interface can't pass.
759 * This will succeed only when the socket is connected.
760 */
761 ret = udp_sendmsg(NULL, sk, &msg, 0);
762 if (ret < 0)
763 return ret;
764 }
765
766 lock_sock(sk);
767
768 if (unlikely(!up->pending)) {
769 release_sock(sk);
770
771 LIMIT_NETDEBUG(KERN_DEBUG "udp cork app bug 3\n");
772 return -EINVAL;
773 }
774
775 ret = ip_append_page(sk, page, offset, size, flags);
776 if (ret == -EOPNOTSUPP) {
777 release_sock(sk);
778 return sock_no_sendpage(sk->sk_socket, page, offset,
779 size, flags);
780 }
781 if (ret < 0) {
782 udp_flush_pending_frames(sk);
783 goto out;
784 }
785
786 up->len += size;
787 if (!(up->corkflag || (flags&MSG_MORE)))
788 ret = udp_push_pending_frames(sk);
789 if (!ret)
790 ret = size;
791out:
792 release_sock(sk);
793 return ret;
794}
795
249/* 796/*
250 * IOCTL requests applicable to the UDP protocol 797 * IOCTL requests applicable to the UDP protocol
251 */ 798 */
@@ -286,6 +833,107 @@ int udp_ioctl(struct sock *sk, int cmd, unsigned long arg)
286 return 0; 833 return 0;
287} 834}
288 835
836/*
837 * This should be easy, if there is something there we
838 * return it, otherwise we block.
839 */
840
841int udp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
842 size_t len, int noblock, int flags, int *addr_len)
843{
844 struct inet_sock *inet = inet_sk(sk);
845 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
846 struct sk_buff *skb;
847 unsigned int ulen, copied;
848 int peeked;
849 int err;
850 int is_udplite = IS_UDPLITE(sk);
851
852 /*
853 * Check any passed addresses
854 */
855 if (addr_len)
856 *addr_len=sizeof(*sin);
857
858 if (flags & MSG_ERRQUEUE)
859 return ip_recv_error(sk, msg, len);
860
861try_again:
862 skb = __skb_recv_datagram(sk, flags | (noblock ? MSG_DONTWAIT : 0),
863 &peeked, &err);
864 if (!skb)
865 goto out;
866
867 ulen = skb->len - sizeof(struct udphdr);
868 copied = len;
869 if (copied > ulen)
870 copied = ulen;
871 else if (copied < ulen)
872 msg->msg_flags |= MSG_TRUNC;
873
874 /*
875 * If checksum is needed at all, try to do it while copying the
876 * data. If the data is truncated, or if we only want a partial
877 * coverage checksum (UDP-Lite), do it before the copy.
878 */
879
880 if (copied < ulen || UDP_SKB_CB(skb)->partial_cov) {
881 if (udp_lib_checksum_complete(skb))
882 goto csum_copy_err;
883 }
884
885 if (skb_csum_unnecessary(skb))
886 err = skb_copy_datagram_iovec(skb, sizeof(struct udphdr),
887 msg->msg_iov, copied );
888 else {
889 err = skb_copy_and_csum_datagram_iovec(skb, sizeof(struct udphdr), msg->msg_iov);
890
891 if (err == -EINVAL)
892 goto csum_copy_err;
893 }
894
895 if (err)
896 goto out_free;
897
898 if (!peeked)
899 UDP_INC_STATS_USER(UDP_MIB_INDATAGRAMS, is_udplite);
900
901 sock_recv_timestamp(msg, sk, skb);
902
903 /* Copy the address. */
904 if (sin)
905 {
906 sin->sin_family = AF_INET;
907 sin->sin_port = udp_hdr(skb)->source;
908 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
909 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
910 }
911 if (inet->cmsg_flags)
912 ip_cmsg_recv(msg, skb);
913
914 err = copied;
915 if (flags & MSG_TRUNC)
916 err = ulen;
917
918out_free:
919 lock_sock(sk);
920 skb_free_datagram(sk, skb);
921 release_sock(sk);
922out:
923 return err;
924
925csum_copy_err:
926 lock_sock(sk);
927 if (!skb_kill_datagram(sk, skb, flags))
928 UDP_INC_STATS_USER(UDP_MIB_INERRORS, is_udplite);
929 release_sock(sk);
930
931 if (noblock)
932 return -EAGAIN;
933 goto try_again;
934}
935
936
289int udp_disconnect(struct sock *sk, int flags) 937int udp_disconnect(struct sock *sk, int flags)
290{ 938{
291 struct inet_sock *inet = inet_sk(sk); 939 struct inet_sock *inet = inet_sk(sk);
@@ -308,6 +956,319 @@ int udp_disconnect(struct sock *sk, int flags)
308 return 0; 956 return 0;
309} 957}
310 958
959/* returns:
960 * -1: error
961 * 0: success
962 * >0: "udp encap" protocol resubmission
963 *
964 * Note that in the success and error cases, the skb is assumed to
965 * have either been requeued or freed.
966 */
967int udp_queue_rcv_skb(struct sock * sk, struct sk_buff *skb)
968{
969 struct udp_sock *up = udp_sk(sk);
970 int rc;
971 int is_udplite = IS_UDPLITE(sk);
972
973 /*
974 * Charge it to the socket, dropping if the queue is full.
975 */
976 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
977 goto drop;
978 nf_reset(skb);
979
980 if (up->encap_type) {
981 /*
982 * This is an encapsulation socket so pass the skb to
983 * the socket's udp_encap_rcv() hook. Otherwise, just
984 * fall through and pass this up the UDP socket.
985 * up->encap_rcv() returns the following value:
986 * =0 if skb was successfully passed to the encap
987 * handler or was discarded by it.
988 * >0 if skb should be passed on to UDP.
989 * <0 if skb should be resubmitted as proto -N
990 */
991
992 /* if we're overly short, let UDP handle it */
993 if (skb->len > sizeof(struct udphdr) &&
994 up->encap_rcv != NULL) {
995 int ret;
996
997 ret = (*up->encap_rcv)(sk, skb);
998 if (ret <= 0) {
999 UDP_INC_STATS_BH(UDP_MIB_INDATAGRAMS,
1000 is_udplite);
1001 return -ret;
1002 }
1003 }
1004
1005 /* FALLTHROUGH -- it's a UDP Packet */
1006 }
1007
1008 /*
1009 * UDP-Lite specific tests, ignored on UDP sockets
1010 */
1011 if ((is_udplite & UDPLITE_RECV_CC) && UDP_SKB_CB(skb)->partial_cov) {
1012
1013 /*
1014 * MIB statistics other than incrementing the error count are
1015 * disabled for the following two types of errors: these depend
1016 * on the application settings, not on the functioning of the
1017 * protocol stack as such.
1018 *
1019 * RFC 3828 here recommends (sec 3.3): "There should also be a
1020 * way ... to ... at least let the receiving application block
1021 * delivery of packets with coverage values less than a value
1022 * provided by the application."
1023 */
1024 if (up->pcrlen == 0) { /* full coverage was set */
1025 LIMIT_NETDEBUG(KERN_WARNING "UDPLITE: partial coverage "
1026 "%d while full coverage %d requested\n",
1027 UDP_SKB_CB(skb)->cscov, skb->len);
1028 goto drop;
1029 }
1030 /* The next case involves violating the min. coverage requested
1031 * by the receiver. This is subtle: if receiver wants x and x is
1032 * greater than the buffersize/MTU then receiver will complain
1033 * that it wants x while sender emits packets of smaller size y.
1034 * Therefore the above ...()->partial_cov statement is essential.
1035 */
1036 if (UDP_SKB_CB(skb)->cscov < up->pcrlen) {
1037 LIMIT_NETDEBUG(KERN_WARNING
1038 "UDPLITE: coverage %d too small, need min %d\n",
1039 UDP_SKB_CB(skb)->cscov, up->pcrlen);
1040 goto drop;
1041 }
1042 }
1043
1044 if (sk->sk_filter) {
1045 if (udp_lib_checksum_complete(skb))
1046 goto drop;
1047 }
1048
1049 if ((rc = sock_queue_rcv_skb(sk,skb)) < 0) {
1050 /* Note that an ENOMEM error is charged twice */
1051 if (rc == -ENOMEM)
1052 UDP_INC_STATS_BH(UDP_MIB_RCVBUFERRORS, is_udplite);
1053 goto drop;
1054 }
1055
1056 return 0;
1057
1058drop:
1059 UDP_INC_STATS_BH(UDP_MIB_INERRORS, is_udplite);
1060 kfree_skb(skb);
1061 return -1;
1062}
1063
1064/*
1065 * Multicasts and broadcasts go to each listener.
1066 *
1067 * Note: called only from the BH handler context,
1068 * so we don't need to lock the hashes.
1069 */
1070static int __udp4_lib_mcast_deliver(struct sk_buff *skb,
1071 struct udphdr *uh,
1072 __be32 saddr, __be32 daddr,
1073 struct hlist_head udptable[])
1074{
1075 struct sock *sk;
1076 int dif;
1077
1078 read_lock(&udp_hash_lock);
1079 sk = sk_head(&udptable[ntohs(uh->dest) & (UDP_HTABLE_SIZE - 1)]);
1080 dif = skb->dev->ifindex;
1081 sk = udp_v4_mcast_next(sk, uh->dest, daddr, uh->source, saddr, dif);
1082 if (sk) {
1083 struct sock *sknext = NULL;
1084
1085 do {
1086 struct sk_buff *skb1 = skb;
1087
1088 sknext = udp_v4_mcast_next(sk_next(sk), uh->dest, daddr,
1089 uh->source, saddr, dif);
1090 if (sknext)
1091 skb1 = skb_clone(skb, GFP_ATOMIC);
1092
1093 if (skb1) {
1094 int ret = 0;
1095
1096 bh_lock_sock_nested(sk);
1097 if (!sock_owned_by_user(sk))
1098 ret = udp_queue_rcv_skb(sk, skb1);
1099 else
1100 sk_add_backlog(sk, skb1);
1101 bh_unlock_sock(sk);
1102
1103 if (ret > 0)
1104 /* we should probably re-process instead
1105 * of dropping packets here. */
1106 kfree_skb(skb1);
1107 }
1108 sk = sknext;
1109 } while (sknext);
1110 } else
1111 kfree_skb(skb);
1112 read_unlock(&udp_hash_lock);
1113 return 0;
1114}
1115
1116/* Initialize UDP checksum. If exited with zero value (success),
1117 * CHECKSUM_UNNECESSARY means, that no more checks are required.
1118 * Otherwise, csum completion requires chacksumming packet body,
1119 * including udp header and folding it to skb->csum.
1120 */
1121static inline int udp4_csum_init(struct sk_buff *skb, struct udphdr *uh,
1122 int proto)
1123{
1124 const struct iphdr *iph;
1125 int err;
1126
1127 UDP_SKB_CB(skb)->partial_cov = 0;
1128 UDP_SKB_CB(skb)->cscov = skb->len;
1129
1130 if (proto == IPPROTO_UDPLITE) {
1131 err = udplite_checksum_init(skb, uh);
1132 if (err)
1133 return err;
1134 }
1135
1136 iph = ip_hdr(skb);
1137 if (uh->check == 0) {
1138 skb->ip_summed = CHECKSUM_UNNECESSARY;
1139 } else if (skb->ip_summed == CHECKSUM_COMPLETE) {
1140 if (!csum_tcpudp_magic(iph->saddr, iph->daddr, skb->len,
1141 proto, skb->csum))
1142 skb->ip_summed = CHECKSUM_UNNECESSARY;
1143 }
1144 if (!skb_csum_unnecessary(skb))
1145 skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
1146 skb->len, proto, 0);
1147 /* Probably, we should checksum udp header (it should be in cache
1148 * in any case) and data in tiny packets (< rx copybreak).
1149 */
1150
1151 return 0;
1152}
1153
1154/*
1155 * All we need to do is get the socket, and then do a checksum.
1156 */
1157
1158int __udp4_lib_rcv(struct sk_buff *skb, struct hlist_head udptable[],
1159 int proto)
1160{
1161 struct sock *sk;
1162 struct udphdr *uh = udp_hdr(skb);
1163 unsigned short ulen;
1164 struct rtable *rt = (struct rtable*)skb->dst;
1165 __be32 saddr = ip_hdr(skb)->saddr;
1166 __be32 daddr = ip_hdr(skb)->daddr;
1167
1168 /*
1169 * Validate the packet.
1170 */
1171 if (!pskb_may_pull(skb, sizeof(struct udphdr)))
1172 goto drop; /* No space for header. */
1173
1174 ulen = ntohs(uh->len);
1175 if (ulen > skb->len)
1176 goto short_packet;
1177
1178 if (proto == IPPROTO_UDP) {
1179 /* UDP validates ulen. */
1180 if (ulen < sizeof(*uh) || pskb_trim_rcsum(skb, ulen))
1181 goto short_packet;
1182 uh = udp_hdr(skb);
1183 }
1184
1185 if (udp4_csum_init(skb, uh, proto))
1186 goto csum_error;
1187
1188 if (rt->rt_flags & (RTCF_BROADCAST|RTCF_MULTICAST))
1189 return __udp4_lib_mcast_deliver(skb, uh, saddr, daddr, udptable);
1190
1191 sk = __udp4_lib_lookup(skb->dev->nd_net, saddr, uh->source, daddr,
1192 uh->dest, inet_iif(skb), udptable);
1193
1194 if (sk != NULL) {
1195 int ret = 0;
1196 bh_lock_sock_nested(sk);
1197 if (!sock_owned_by_user(sk))
1198 ret = udp_queue_rcv_skb(sk, skb);
1199 else
1200 sk_add_backlog(sk, skb);
1201 bh_unlock_sock(sk);
1202 sock_put(sk);
1203
1204 /* a return value > 0 means to resubmit the input, but
1205 * it wants the return to be -protocol, or 0
1206 */
1207 if (ret > 0)
1208 return -ret;
1209 return 0;
1210 }
1211
1212 if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
1213 goto drop;
1214 nf_reset(skb);
1215
1216 /* No socket. Drop packet silently, if checksum is wrong */
1217 if (udp_lib_checksum_complete(skb))
1218 goto csum_error;
1219
1220 UDP_INC_STATS_BH(UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE);
1221 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PORT_UNREACH, 0);
1222
1223 /*
1224 * Hmm. We got an UDP packet to a port to which we
1225 * don't wanna listen. Ignore it.
1226 */
1227 kfree_skb(skb);
1228 return 0;
1229
1230short_packet:
1231 LIMIT_NETDEBUG(KERN_DEBUG "UDP%s: short packet: From %u.%u.%u.%u:%u %d/%d to %u.%u.%u.%u:%u\n",
1232 proto == IPPROTO_UDPLITE ? "-Lite" : "",
1233 NIPQUAD(saddr),
1234 ntohs(uh->source),
1235 ulen,
1236 skb->len,
1237 NIPQUAD(daddr),
1238 ntohs(uh->dest));
1239 goto drop;
1240
1241csum_error:
1242 /*
1243 * RFC1122: OK. Discards the bad packet silently (as far as
1244 * the network is concerned, anyway) as per 4.1.3.4 (MUST).
1245 */
1246 LIMIT_NETDEBUG(KERN_DEBUG "UDP%s: bad checksum. From %d.%d.%d.%d:%d to %d.%d.%d.%d:%d ulen %d\n",
1247 proto == IPPROTO_UDPLITE ? "-Lite" : "",
1248 NIPQUAD(saddr),
1249 ntohs(uh->source),
1250 NIPQUAD(daddr),
1251 ntohs(uh->dest),
1252 ulen);
1253drop:
1254 UDP_INC_STATS_BH(UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE);
1255 kfree_skb(skb);
1256 return 0;
1257}
1258
1259int udp_rcv(struct sk_buff *skb)
1260{
1261 return __udp4_lib_rcv(skb, udp_hash, IPPROTO_UDP);
1262}
1263
1264int udp_destroy_sock(struct sock *sk)
1265{
1266 lock_sock(sk);
1267 udp_flush_pending_frames(sk);
1268 release_sock(sk);
1269 return 0;
1270}
1271
311/* 1272/*
312 * Socket option code for UDP 1273 * Socket option code for UDP
313 */ 1274 */
@@ -318,9 +1279,7 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
318 struct udp_sock *up = udp_sk(sk); 1279 struct udp_sock *up = udp_sk(sk);
319 int val; 1280 int val;
320 int err = 0; 1281 int err = 0;
321#ifdef CONFIG_IP_UDPLITE
322 int is_udplite = IS_UDPLITE(sk); 1282 int is_udplite = IS_UDPLITE(sk);
323#endif
324 1283
325 if (optlen<sizeof(int)) 1284 if (optlen<sizeof(int))
326 return -EINVAL; 1285 return -EINVAL;
@@ -356,7 +1315,6 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
356 } 1315 }
357 break; 1316 break;
358 1317
359#ifdef CONFIG_IP_UDPLITE
360 /* 1318 /*
361 * UDP-Lite's partial checksum coverage (RFC 3828). 1319 * UDP-Lite's partial checksum coverage (RFC 3828).
362 */ 1320 */
@@ -382,7 +1340,6 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
382 up->pcrlen = val; 1340 up->pcrlen = val;
383 up->pcflag |= UDPLITE_RECV_CC; 1341 up->pcflag |= UDPLITE_RECV_CC;
384 break; 1342 break;
385#endif
386 1343
387 default: 1344 default:
388 err = -ENOPROTOOPT; 1345 err = -ENOPROTOOPT;
@@ -392,6 +1349,26 @@ int udp_lib_setsockopt(struct sock *sk, int level, int optname,
392 return err; 1349 return err;
393} 1350}
394 1351
1352int udp_setsockopt(struct sock *sk, int level, int optname,
1353 char __user *optval, int optlen)
1354{
1355 if (level == SOL_UDP || level == SOL_UDPLITE)
1356 return udp_lib_setsockopt(sk, level, optname, optval, optlen,
1357 udp_push_pending_frames);
1358 return ip_setsockopt(sk, level, optname, optval, optlen);
1359}
1360
1361#ifdef CONFIG_COMPAT
1362int compat_udp_setsockopt(struct sock *sk, int level, int optname,
1363 char __user *optval, int optlen)
1364{
1365 if (level == SOL_UDP || level == SOL_UDPLITE)
1366 return udp_lib_setsockopt(sk, level, optname, optval, optlen,
1367 udp_push_pending_frames);
1368 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
1369}
1370#endif
1371
395int udp_lib_getsockopt(struct sock *sk, int level, int optname, 1372int udp_lib_getsockopt(struct sock *sk, int level, int optname,
396 char __user *optval, int __user *optlen) 1373 char __user *optval, int __user *optlen)
397{ 1374{
@@ -436,6 +1413,23 @@ int udp_lib_getsockopt(struct sock *sk, int level, int optname,
436 return 0; 1413 return 0;
437} 1414}
438 1415
1416int udp_getsockopt(struct sock *sk, int level, int optname,
1417 char __user *optval, int __user *optlen)
1418{
1419 if (level == SOL_UDP || level == SOL_UDPLITE)
1420 return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1421 return ip_getsockopt(sk, level, optname, optval, optlen);
1422}
1423
1424#ifdef CONFIG_COMPAT
1425int compat_udp_getsockopt(struct sock *sk, int level, int optname,
1426 char __user *optval, int __user *optlen)
1427{
1428 if (level == SOL_UDP || level == SOL_UDPLITE)
1429 return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1430 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
1431}
1432#endif
439/** 1433/**
440 * udp_poll - wait for a UDP event. 1434 * udp_poll - wait for a UDP event.
441 * @file - file struct 1435 * @file - file struct
@@ -480,6 +1474,36 @@ unsigned int udp_poll(struct file *file, struct socket *sock, poll_table *wait)
480 1474
481} 1475}
482 1476
1477DEFINE_PROTO_INUSE(udp)
1478
1479struct proto udp_prot = {
1480 .name = "UDP",
1481 .owner = THIS_MODULE,
1482 .close = udp_lib_close,
1483 .connect = ip4_datagram_connect,
1484 .disconnect = udp_disconnect,
1485 .ioctl = udp_ioctl,
1486 .destroy = udp_destroy_sock,
1487 .setsockopt = udp_setsockopt,
1488 .getsockopt = udp_getsockopt,
1489 .sendmsg = udp_sendmsg,
1490 .recvmsg = udp_recvmsg,
1491 .sendpage = udp_sendpage,
1492 .backlog_rcv = udp_queue_rcv_skb,
1493 .hash = udp_lib_hash,
1494 .unhash = udp_lib_unhash,
1495 .get_port = udp_v4_get_port,
1496 .memory_allocated = &udp_memory_allocated,
1497 .sysctl_mem = sysctl_udp_mem,
1498 .sysctl_wmem = &sysctl_udp_wmem_min,
1499 .sysctl_rmem = &sysctl_udp_rmem_min,
1500 .obj_size = sizeof(struct udp_sock),
1501#ifdef CONFIG_COMPAT
1502 .compat_setsockopt = compat_udp_setsockopt,
1503 .compat_getsockopt = compat_udp_getsockopt,
1504#endif
1505 REF_PROTO_INUSE(udp)
1506};
483 1507
484/* ------------------------------------------------------------------------ */ 1508/* ------------------------------------------------------------------------ */
485#ifdef CONFIG_PROC_FS 1509#ifdef CONFIG_PROC_FS
@@ -612,6 +1636,62 @@ void udp_proc_unregister(struct udp_seq_afinfo *afinfo)
612 proc_net_remove(&init_net, afinfo->name); 1636 proc_net_remove(&init_net, afinfo->name);
613 memset(afinfo->seq_fops, 0, sizeof(*afinfo->seq_fops)); 1637 memset(afinfo->seq_fops, 0, sizeof(*afinfo->seq_fops));
614} 1638}
1639
1640/* ------------------------------------------------------------------------ */
1641static void udp4_format_sock(struct sock *sp, char *tmpbuf, int bucket)
1642{
1643 struct inet_sock *inet = inet_sk(sp);
1644 __be32 dest = inet->daddr;
1645 __be32 src = inet->rcv_saddr;
1646 __u16 destp = ntohs(inet->dport);
1647 __u16 srcp = ntohs(inet->sport);
1648
1649 sprintf(tmpbuf, "%4d: %08X:%04X %08X:%04X"
1650 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p",
1651 bucket, src, srcp, dest, destp, sp->sk_state,
1652 atomic_read(&sp->sk_wmem_alloc),
1653 atomic_read(&sp->sk_rmem_alloc),
1654 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
1655 atomic_read(&sp->sk_refcnt), sp);
1656}
1657
1658int udp4_seq_show(struct seq_file *seq, void *v)
1659{
1660 if (v == SEQ_START_TOKEN)
1661 seq_printf(seq, "%-127s\n",
1662 " sl local_address rem_address st tx_queue "
1663 "rx_queue tr tm->when retrnsmt uid timeout "
1664 "inode");
1665 else {
1666 char tmpbuf[129];
1667 struct udp_iter_state *state = seq->private;
1668
1669 udp4_format_sock(v, tmpbuf, state->bucket);
1670 seq_printf(seq, "%-127s\n", tmpbuf);
1671 }
1672 return 0;
1673}
1674
1675/* ------------------------------------------------------------------------ */
1676static struct file_operations udp4_seq_fops;
1677static struct udp_seq_afinfo udp4_seq_afinfo = {
1678 .owner = THIS_MODULE,
1679 .name = "udp",
1680 .family = AF_INET,
1681 .hashtable = udp_hash,
1682 .seq_show = udp4_seq_show,
1683 .seq_fops = &udp4_seq_fops,
1684};
1685
1686int __init udp4_proc_init(void)
1687{
1688 return udp_proc_register(&udp4_seq_afinfo);
1689}
1690
1691void udp4_proc_exit(void)
1692{
1693 udp_proc_unregister(&udp4_seq_afinfo);
1694}
615#endif /* CONFIG_PROC_FS */ 1695#endif /* CONFIG_PROC_FS */
616 1696
617void __init udp_init(void) 1697void __init udp_init(void)
@@ -638,6 +1718,8 @@ EXPORT_SYMBOL(udp_hash);
638EXPORT_SYMBOL(udp_hash_lock); 1718EXPORT_SYMBOL(udp_hash_lock);
639EXPORT_SYMBOL(udp_ioctl); 1719EXPORT_SYMBOL(udp_ioctl);
640EXPORT_SYMBOL(udp_get_port); 1720EXPORT_SYMBOL(udp_get_port);
1721EXPORT_SYMBOL(udp_prot);
1722EXPORT_SYMBOL(udp_sendmsg);
641EXPORT_SYMBOL(udp_lib_getsockopt); 1723EXPORT_SYMBOL(udp_lib_getsockopt);
642EXPORT_SYMBOL(udp_lib_setsockopt); 1724EXPORT_SYMBOL(udp_lib_setsockopt);
643EXPORT_SYMBOL(udp_poll); 1725EXPORT_SYMBOL(udp_poll);
diff --git a/net/ipv4/udp_ipv4.c b/net/ipv4/udp_ipv4.c
deleted file mode 100644
index fd14c2c50ed4..000000000000
--- a/net/ipv4/udp_ipv4.c
+++ /dev/null
@@ -1,1134 +0,0 @@
1/*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
5 *
6 * UDP for IPv4.
7 *
8 * For full credits, see net/ipv4/udp.c.
9 *
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public License
12 * as published by the Free Software Foundation; either version
13 * 2 of the License, or (at your option) any later version.
14 */
15
16#include <asm/system.h>
17#include <asm/uaccess.h>
18#include <asm/ioctls.h>
19#include <linux/bootmem.h>
20#include <linux/types.h>
21#include <linux/fcntl.h>
22#include <linux/module.h>
23#include <linux/socket.h>
24#include <linux/sockios.h>
25#include <linux/igmp.h>
26#include <linux/in.h>
27#include <linux/errno.h>
28#include <linux/timer.h>
29#include <linux/mm.h>
30#include <linux/inet.h>
31#include <linux/netdevice.h>
32#include <net/tcp_states.h>
33#include <linux/skbuff.h>
34#include <linux/proc_fs.h>
35#include <linux/seq_file.h>
36#include <net/net_namespace.h>
37#include <net/icmp.h>
38#include <net/route.h>
39#include <net/checksum.h>
40#include <net/xfrm.h>
41#include "udp_impl.h"
42
43int ipv4_rcv_saddr_equal(const struct sock *sk1, const struct sock *sk2)
44{
45 struct inet_sock *inet1 = inet_sk(sk1), *inet2 = inet_sk(sk2);
46
47 return ( !ipv6_only_sock(sk2) &&
48 (!inet1->rcv_saddr || !inet2->rcv_saddr ||
49 inet1->rcv_saddr == inet2->rcv_saddr ));
50}
51
52static inline int udp_v4_get_port(struct sock *sk, unsigned short snum)
53{
54 return udp_get_port(sk, snum, ipv4_rcv_saddr_equal);
55}
56
57/* UDP is nearly always wildcards out the wazoo, it makes no sense to try
58 * harder than this. -DaveM
59 */
60static struct sock *__udp4_lib_lookup(struct net *net, __be32 saddr,
61 __be16 sport, __be32 daddr, __be16 dport,
62 int dif, struct hlist_head udptable[])
63{
64 struct sock *sk, *result = NULL;
65 struct hlist_node *node;
66 unsigned short hnum = ntohs(dport);
67 int badness = -1;
68
69 read_lock(&udp_hash_lock);
70 sk_for_each(sk, node, &udptable[hnum & (UDP_HTABLE_SIZE - 1)]) {
71 struct inet_sock *inet = inet_sk(sk);
72
73 if (sk->sk_net == net && sk->sk_hash == hnum &&
74 !ipv6_only_sock(sk)) {
75 int score = (sk->sk_family == PF_INET ? 1 : 0);
76 if (inet->rcv_saddr) {
77 if (inet->rcv_saddr != daddr)
78 continue;
79 score+=2;
80 }
81 if (inet->daddr) {
82 if (inet->daddr != saddr)
83 continue;
84 score+=2;
85 }
86 if (inet->dport) {
87 if (inet->dport != sport)
88 continue;
89 score+=2;
90 }
91 if (sk->sk_bound_dev_if) {
92 if (sk->sk_bound_dev_if != dif)
93 continue;
94 score+=2;
95 }
96 if (score == 9) {
97 result = sk;
98 break;
99 } else if (score > badness) {
100 result = sk;
101 badness = score;
102 }
103 }
104 }
105 if (result)
106 sock_hold(result);
107 read_unlock(&udp_hash_lock);
108 return result;
109}
110
111static inline struct sock *udp_v4_mcast_next(struct sock *sk,
112 __be16 loc_port, __be32 loc_addr,
113 __be16 rmt_port, __be32 rmt_addr,
114 int dif)
115{
116 struct hlist_node *node;
117 struct sock *s = sk;
118 unsigned short hnum = ntohs(loc_port);
119
120 sk_for_each_from(s, node) {
121 struct inet_sock *inet = inet_sk(s);
122
123 if (s->sk_hash != hnum ||
124 (inet->daddr && inet->daddr != rmt_addr) ||
125 (inet->dport != rmt_port && inet->dport) ||
126 (inet->rcv_saddr && inet->rcv_saddr != loc_addr) ||
127 ipv6_only_sock(s) ||
128 (s->sk_bound_dev_if && s->sk_bound_dev_if != dif))
129 continue;
130 if (!ip_mc_sf_allow(s, loc_addr, rmt_addr, dif))
131 continue;
132 goto found;
133 }
134 s = NULL;
135found:
136 return s;
137}
138
139/*
140 * This routine is called by the ICMP module when it gets some
141 * sort of error condition. If err < 0 then the socket should
142 * be closed and the error returned to the user. If err > 0
143 * it's just the icmp type << 8 | icmp code.
144 * Header points to the ip header of the error packet. We move
145 * on past this. Then (as it used to claim before adjustment)
146 * header points to the first 8 bytes of the udp header. We need
147 * to find the appropriate port.
148 */
149
150void __udp4_lib_err(struct sk_buff *skb, u32 info, struct hlist_head udptable[])
151{
152 struct inet_sock *inet;
153 struct iphdr *iph = (struct iphdr*)skb->data;
154 struct udphdr *uh = (struct udphdr*)(skb->data+(iph->ihl<<2));
155 const int type = icmp_hdr(skb)->type;
156 const int code = icmp_hdr(skb)->code;
157 struct sock *sk;
158 int harderr;
159 int err;
160
161 sk = __udp4_lib_lookup(skb->dev->nd_net, iph->daddr, uh->dest,
162 iph->saddr, uh->source, skb->dev->ifindex, udptable);
163 if (sk == NULL) {
164 ICMP_INC_STATS_BH(ICMP_MIB_INERRORS);
165 return; /* No socket for error */
166 }
167
168 err = 0;
169 harderr = 0;
170 inet = inet_sk(sk);
171
172 switch (type) {
173 default:
174 case ICMP_TIME_EXCEEDED:
175 err = EHOSTUNREACH;
176 break;
177 case ICMP_SOURCE_QUENCH:
178 goto out;
179 case ICMP_PARAMETERPROB:
180 err = EPROTO;
181 harderr = 1;
182 break;
183 case ICMP_DEST_UNREACH:
184 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
185 if (inet->pmtudisc != IP_PMTUDISC_DONT) {
186 err = EMSGSIZE;
187 harderr = 1;
188 break;
189 }
190 goto out;
191 }
192 err = EHOSTUNREACH;
193 if (code <= NR_ICMP_UNREACH) {
194 harderr = icmp_err_convert[code].fatal;
195 err = icmp_err_convert[code].errno;
196 }
197 break;
198 }
199
200 /*
201 * RFC1122: OK. Passes ICMP errors back to application, as per
202 * 4.1.3.3.
203 */
204 if (!inet->recverr) {
205 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
206 goto out;
207 } else {
208 ip_icmp_error(sk, skb, err, uh->dest, info, (u8*)(uh+1));
209 }
210 sk->sk_err = err;
211 sk->sk_error_report(sk);
212out:
213 sock_put(sk);
214}
215
216void udp_err(struct sk_buff *skb, u32 info)
217{
218 __udp4_lib_err(skb, info, udp_hash);
219}
220
221/*
222 * Throw away all pending data and cancel the corking. Socket is locked.
223 */
224static void udp_flush_pending_frames(struct sock *sk)
225{
226 struct udp_sock *up = udp_sk(sk);
227
228 if (up->pending) {
229 up->len = 0;
230 up->pending = 0;
231 ip_flush_pending_frames(sk);
232 }
233}
234
235/**
236 * udp4_hwcsum_outgoing - handle outgoing HW checksumming
237 * @sk: socket we are sending on
238 * @skb: sk_buff containing the filled-in UDP header
239 * (checksum field must be zeroed out)
240 */
241static void udp4_hwcsum_outgoing(struct sock *sk, struct sk_buff *skb,
242 __be32 src, __be32 dst, int len )
243{
244 unsigned int offset;
245 struct udphdr *uh = udp_hdr(skb);
246 __wsum csum = 0;
247
248 if (skb_queue_len(&sk->sk_write_queue) == 1) {
249 /*
250 * Only one fragment on the socket.
251 */
252 skb->csum_start = skb_transport_header(skb) - skb->head;
253 skb->csum_offset = offsetof(struct udphdr, check);
254 uh->check = ~csum_tcpudp_magic(src, dst, len, IPPROTO_UDP, 0);
255 } else {
256 /*
257 * HW-checksum won't work as there are two or more
258 * fragments on the socket so that all csums of sk_buffs
259 * should be together
260 */
261 offset = skb_transport_offset(skb);
262 skb->csum = skb_checksum(skb, offset, skb->len - offset, 0);
263
264 skb->ip_summed = CHECKSUM_NONE;
265
266 skb_queue_walk(&sk->sk_write_queue, skb) {
267 csum = csum_add(csum, skb->csum);
268 }
269
270 uh->check = csum_tcpudp_magic(src, dst, len, IPPROTO_UDP, csum);
271 if (uh->check == 0)
272 uh->check = CSUM_MANGLED_0;
273 }
274}
275
276/*
277 * Push out all pending data as one UDP datagram. Socket is locked.
278 */
279static int udp_push_pending_frames(struct sock *sk)
280{
281 struct udp_sock *up = udp_sk(sk);
282 struct inet_sock *inet = inet_sk(sk);
283 struct flowi *fl = &inet->cork.fl;
284 struct sk_buff *skb;
285 struct udphdr *uh;
286 int err = 0;
287 int is_udplite = IS_UDPLITE(sk);
288 __wsum csum = 0;
289
290 /* Grab the skbuff where UDP header space exists. */
291 if ((skb = skb_peek(&sk->sk_write_queue)) == NULL)
292 goto out;
293
294 /*
295 * Create a UDP header
296 */
297 uh = udp_hdr(skb);
298 uh->source = fl->fl_ip_sport;
299 uh->dest = fl->fl_ip_dport;
300 uh->len = htons(up->len);
301 uh->check = 0;
302
303 if (is_udplite) /* UDP-Lite */
304 csum = udplite_csum_outgoing(sk, skb);
305
306 else if (sk->sk_no_check == UDP_CSUM_NOXMIT) { /* UDP csum disabled */
307
308 skb->ip_summed = CHECKSUM_NONE;
309 goto send;
310
311 } else if (skb->ip_summed == CHECKSUM_PARTIAL) { /* UDP hardware csum */
312
313 udp4_hwcsum_outgoing(sk, skb, fl->fl4_src,fl->fl4_dst, up->len);
314 goto send;
315
316 } else /* `normal' UDP */
317 csum = udp_csum_outgoing(sk, skb);
318
319 /* add protocol-dependent pseudo-header */
320 uh->check = csum_tcpudp_magic(fl->fl4_src, fl->fl4_dst, up->len,
321 sk->sk_protocol, csum );
322 if (uh->check == 0)
323 uh->check = CSUM_MANGLED_0;
324
325send:
326 err = ip_push_pending_frames(sk);
327out:
328 up->len = 0;
329 up->pending = 0;
330 if (!err)
331 UDP_INC_STATS_USER(UDP_MIB_OUTDATAGRAMS, is_udplite);
332 return err;
333}
334
335int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
336 size_t len)
337{
338 struct inet_sock *inet = inet_sk(sk);
339 struct udp_sock *up = udp_sk(sk);
340 int ulen = len;
341 struct ipcm_cookie ipc;
342 struct rtable *rt = NULL;
343 int free = 0;
344 int connected = 0;
345 __be32 daddr, faddr, saddr;
346 __be16 dport;
347 u8 tos;
348 int err, is_udplite = IS_UDPLITE(sk);
349 int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
350 int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
351
352 if (len > 0xFFFF)
353 return -EMSGSIZE;
354
355 /*
356 * Check the flags.
357 */
358
359 if (msg->msg_flags&MSG_OOB) /* Mirror BSD error message compatibility */
360 return -EOPNOTSUPP;
361
362 ipc.opt = NULL;
363
364 if (up->pending) {
365 /*
366 * There are pending frames.
367 * The socket lock must be held while it's corked.
368 */
369 lock_sock(sk);
370 if (likely(up->pending)) {
371 if (unlikely(up->pending != AF_INET)) {
372 release_sock(sk);
373 return -EINVAL;
374 }
375 goto do_append_data;
376 }
377 release_sock(sk);
378 }
379 ulen += sizeof(struct udphdr);
380
381 /*
382 * Get and verify the address.
383 */
384 if (msg->msg_name) {
385 struct sockaddr_in * usin = (struct sockaddr_in*)msg->msg_name;
386 if (msg->msg_namelen < sizeof(*usin))
387 return -EINVAL;
388 if (usin->sin_family != AF_INET) {
389 if (usin->sin_family != AF_UNSPEC)
390 return -EAFNOSUPPORT;
391 }
392
393 daddr = usin->sin_addr.s_addr;
394 dport = usin->sin_port;
395 if (dport == 0)
396 return -EINVAL;
397 } else {
398 if (sk->sk_state != TCP_ESTABLISHED)
399 return -EDESTADDRREQ;
400 daddr = inet->daddr;
401 dport = inet->dport;
402 /* Open fast path for connected socket.
403 Route will not be used, if at least one option is set.
404 */
405 connected = 1;
406 }
407 ipc.addr = inet->saddr;
408
409 ipc.oif = sk->sk_bound_dev_if;
410 if (msg->msg_controllen) {
411 err = ip_cmsg_send(msg, &ipc);
412 if (err)
413 return err;
414 if (ipc.opt)
415 free = 1;
416 connected = 0;
417 }
418 if (!ipc.opt)
419 ipc.opt = inet->opt;
420
421 saddr = ipc.addr;
422 ipc.addr = faddr = daddr;
423
424 if (ipc.opt && ipc.opt->srr) {
425 if (!daddr)
426 return -EINVAL;
427 faddr = ipc.opt->faddr;
428 connected = 0;
429 }
430 tos = RT_TOS(inet->tos);
431 if (sock_flag(sk, SOCK_LOCALROUTE) ||
432 (msg->msg_flags & MSG_DONTROUTE) ||
433 (ipc.opt && ipc.opt->is_strictroute)) {
434 tos |= RTO_ONLINK;
435 connected = 0;
436 }
437
438 if (ipv4_is_multicast(daddr)) {
439 if (!ipc.oif)
440 ipc.oif = inet->mc_index;
441 if (!saddr)
442 saddr = inet->mc_addr;
443 connected = 0;
444 }
445
446 if (connected)
447 rt = (struct rtable*)sk_dst_check(sk, 0);
448
449 if (rt == NULL) {
450 struct flowi fl = { .oif = ipc.oif,
451 .nl_u = { .ip4_u =
452 { .daddr = faddr,
453 .saddr = saddr,
454 .tos = tos } },
455 .proto = sk->sk_protocol,
456 .uli_u = { .ports =
457 { .sport = inet->sport,
458 .dport = dport } } };
459 security_sk_classify_flow(sk, &fl);
460 err = ip_route_output_flow(&init_net, &rt, &fl, sk, 1);
461 if (err) {
462 if (err == -ENETUNREACH)
463 IP_INC_STATS_BH(IPSTATS_MIB_OUTNOROUTES);
464 goto out;
465 }
466
467 err = -EACCES;
468 if ((rt->rt_flags & RTCF_BROADCAST) &&
469 !sock_flag(sk, SOCK_BROADCAST))
470 goto out;
471 if (connected)
472 sk_dst_set(sk, dst_clone(&rt->u.dst));
473 }
474
475 if (msg->msg_flags&MSG_CONFIRM)
476 goto do_confirm;
477back_from_confirm:
478
479 saddr = rt->rt_src;
480 if (!ipc.addr)
481 daddr = ipc.addr = rt->rt_dst;
482
483 lock_sock(sk);
484 if (unlikely(up->pending)) {
485 /* The socket is already corked while preparing it. */
486 /* ... which is an evident application bug. --ANK */
487 release_sock(sk);
488
489 LIMIT_NETDEBUG(KERN_DEBUG "udp cork app bug 2\n");
490 err = -EINVAL;
491 goto out;
492 }
493 /*
494 * Now cork the socket to pend data.
495 */
496 inet->cork.fl.fl4_dst = daddr;
497 inet->cork.fl.fl_ip_dport = dport;
498 inet->cork.fl.fl4_src = saddr;
499 inet->cork.fl.fl_ip_sport = inet->sport;
500 up->pending = AF_INET;
501
502do_append_data:
503 up->len += ulen;
504 getfrag = is_udplite ? udplite_getfrag : ip_generic_getfrag;
505 err = ip_append_data(sk, getfrag, msg->msg_iov, ulen,
506 sizeof(struct udphdr), &ipc, rt,
507 corkreq ? msg->msg_flags|MSG_MORE : msg->msg_flags);
508 if (err)
509 udp_flush_pending_frames(sk);
510 else if (!corkreq)
511 err = udp_push_pending_frames(sk);
512 else if (unlikely(skb_queue_empty(&sk->sk_write_queue)))
513 up->pending = 0;
514 release_sock(sk);
515
516out:
517 ip_rt_put(rt);
518 if (free)
519 kfree(ipc.opt);
520 if (!err)
521 return len;
522 /*
523 * ENOBUFS = no kernel mem, SOCK_NOSPACE = no sndbuf space. Reporting
524 * ENOBUFS might not be good (it's not tunable per se), but otherwise
525 * we don't have a good statistic (IpOutDiscards but it can be too many
526 * things). We could add another new stat but at least for now that
527 * seems like overkill.
528 */
529 if (err == -ENOBUFS || test_bit(SOCK_NOSPACE, &sk->sk_socket->flags)) {
530 UDP_INC_STATS_USER(UDP_MIB_SNDBUFERRORS, is_udplite);
531 }
532 return err;
533
534do_confirm:
535 dst_confirm(&rt->u.dst);
536 if (!(msg->msg_flags&MSG_PROBE) || len)
537 goto back_from_confirm;
538 err = 0;
539 goto out;
540}
541
542int udp_sendpage(struct sock *sk, struct page *page, int offset,
543 size_t size, int flags)
544{
545 struct udp_sock *up = udp_sk(sk);
546 int ret;
547
548 if (!up->pending) {
549 struct msghdr msg = { .msg_flags = flags|MSG_MORE };
550
551 /* Call udp_sendmsg to specify destination address which
552 * sendpage interface can't pass.
553 * This will succeed only when the socket is connected.
554 */
555 ret = udp_sendmsg(NULL, sk, &msg, 0);
556 if (ret < 0)
557 return ret;
558 }
559
560 lock_sock(sk);
561
562 if (unlikely(!up->pending)) {
563 release_sock(sk);
564
565 LIMIT_NETDEBUG(KERN_DEBUG "udp cork app bug 3\n");
566 return -EINVAL;
567 }
568
569 ret = ip_append_page(sk, page, offset, size, flags);
570 if (ret == -EOPNOTSUPP) {
571 release_sock(sk);
572 return sock_no_sendpage(sk->sk_socket, page, offset,
573 size, flags);
574 }
575 if (ret < 0) {
576 udp_flush_pending_frames(sk);
577 goto out;
578 }
579
580 up->len += size;
581 if (!(up->corkflag || (flags&MSG_MORE)))
582 ret = udp_push_pending_frames(sk);
583 if (!ret)
584 ret = size;
585out:
586 release_sock(sk);
587 return ret;
588}
589
590/*
591 * This should be easy, if there is something there we
592 * return it, otherwise we block.
593 */
594
595int udp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
596 size_t len, int noblock, int flags, int *addr_len)
597{
598 struct inet_sock *inet = inet_sk(sk);
599 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
600 struct sk_buff *skb;
601 unsigned int ulen, copied;
602 int peeked;
603 int err;
604 int is_udplite = IS_UDPLITE(sk);
605
606 /*
607 * Check any passed addresses
608 */
609 if (addr_len)
610 *addr_len=sizeof(*sin);
611
612 if (flags & MSG_ERRQUEUE)
613 return ip_recv_error(sk, msg, len);
614
615try_again:
616 skb = __skb_recv_datagram(sk, flags | (noblock ? MSG_DONTWAIT : 0),
617 &peeked, &err);
618 if (!skb)
619 goto out;
620
621 ulen = skb->len - sizeof(struct udphdr);
622 copied = len;
623 if (copied > ulen)
624 copied = ulen;
625 else if (copied < ulen)
626 msg->msg_flags |= MSG_TRUNC;
627
628 /*
629 * If checksum is needed at all, try to do it while copying the
630 * data. If the data is truncated, or if we only want a partial
631 * coverage checksum (UDP-Lite), do it before the copy.
632 */
633
634 if (copied < ulen || UDP_SKB_CB(skb)->partial_cov) {
635 if (udp_lib_checksum_complete(skb))
636 goto csum_copy_err;
637 }
638
639 if (skb_csum_unnecessary(skb))
640 err = skb_copy_datagram_iovec(skb, sizeof(struct udphdr),
641 msg->msg_iov, copied );
642 else {
643 err = skb_copy_and_csum_datagram_iovec(skb, sizeof(struct udphdr), msg->msg_iov);
644
645 if (err == -EINVAL)
646 goto csum_copy_err;
647 }
648
649 if (err)
650 goto out_free;
651
652 if (!peeked)
653 UDP_INC_STATS_USER(UDP_MIB_INDATAGRAMS, is_udplite);
654
655 sock_recv_timestamp(msg, sk, skb);
656
657 /* Copy the address. */
658 if (sin)
659 {
660 sin->sin_family = AF_INET;
661 sin->sin_port = udp_hdr(skb)->source;
662 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
663 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
664 }
665 if (inet->cmsg_flags)
666 ip_cmsg_recv(msg, skb);
667
668 err = copied;
669 if (flags & MSG_TRUNC)
670 err = ulen;
671
672out_free:
673 lock_sock(sk);
674 skb_free_datagram(sk, skb);
675 release_sock(sk);
676out:
677 return err;
678
679csum_copy_err:
680 lock_sock(sk);
681 if (!skb_kill_datagram(sk, skb, flags))
682 UDP_INC_STATS_USER(UDP_MIB_INERRORS, is_udplite);
683 release_sock(sk);
684
685 if (noblock)
686 return -EAGAIN;
687 goto try_again;
688}
689
690
691/* returns:
692 * -1: error
693 * 0: success
694 * >0: "udp encap" protocol resubmission
695 *
696 * Note that in the success and error cases, the skb is assumed to
697 * have either been requeued or freed.
698 */
699int udp_queue_rcv_skb(struct sock * sk, struct sk_buff *skb)
700{
701 struct udp_sock *up = udp_sk(sk);
702 int rc;
703 int is_udplite = IS_UDPLITE(sk);
704
705 /*
706 * Charge it to the socket, dropping if the queue is full.
707 */
708 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb))
709 goto drop;
710 nf_reset(skb);
711
712 if (up->encap_type) {
713 /*
714 * This is an encapsulation socket so pass the skb to
715 * the socket's udp_encap_rcv() hook. Otherwise, just
716 * fall through and pass this up the UDP socket.
717 * up->encap_rcv() returns the following value:
718 * =0 if skb was successfully passed to the encap
719 * handler or was discarded by it.
720 * >0 if skb should be passed on to UDP.
721 * <0 if skb should be resubmitted as proto -N
722 */
723
724 /* if we're overly short, let UDP handle it */
725 if (skb->len > sizeof(struct udphdr) &&
726 up->encap_rcv != NULL) {
727 int ret;
728
729 ret = (*up->encap_rcv)(sk, skb);
730 if (ret <= 0) {
731 UDP_INC_STATS_BH(UDP_MIB_INDATAGRAMS,
732 is_udplite);
733 return -ret;
734 }
735 }
736
737 /* FALLTHROUGH -- it's a UDP Packet */
738 }
739
740 /*
741 * UDP-Lite specific tests, ignored on UDP sockets
742 */
743 if ((is_udplite & UDPLITE_RECV_CC) && UDP_SKB_CB(skb)->partial_cov) {
744
745 /*
746 * MIB statistics other than incrementing the error count are
747 * disabled for the following two types of errors: these depend
748 * on the application settings, not on the functioning of the
749 * protocol stack as such.
750 *
751 * RFC 3828 here recommends (sec 3.3): "There should also be a
752 * way ... to ... at least let the receiving application block
753 * delivery of packets with coverage values less than a value
754 * provided by the application."
755 */
756 if (up->pcrlen == 0) { /* full coverage was set */
757 LIMIT_NETDEBUG(KERN_WARNING "UDPLITE: partial coverage "
758 "%d while full coverage %d requested\n",
759 UDP_SKB_CB(skb)->cscov, skb->len);
760 goto drop;
761 }
762 /* The next case involves violating the min. coverage requested
763 * by the receiver. This is subtle: if receiver wants x and x is
764 * greater than the buffersize/MTU then receiver will complain
765 * that it wants x while sender emits packets of smaller size y.
766 * Therefore the above ...()->partial_cov statement is essential.
767 */
768 if (UDP_SKB_CB(skb)->cscov < up->pcrlen) {
769 LIMIT_NETDEBUG(KERN_WARNING
770 "UDPLITE: coverage %d too small, need min %d\n",
771 UDP_SKB_CB(skb)->cscov, up->pcrlen);
772 goto drop;
773 }
774 }
775
776 if (sk->sk_filter) {
777 if (udp_lib_checksum_complete(skb))
778 goto drop;
779 }
780
781 if ((rc = sock_queue_rcv_skb(sk,skb)) < 0) {
782 /* Note that an ENOMEM error is charged twice */
783 if (rc == -ENOMEM)
784 UDP_INC_STATS_BH(UDP_MIB_RCVBUFERRORS, is_udplite);
785 goto drop;
786 }
787
788 return 0;
789
790drop:
791 UDP_INC_STATS_BH(UDP_MIB_INERRORS, is_udplite);
792 kfree_skb(skb);
793 return -1;
794}
795
796/*
797 * Multicasts and broadcasts go to each listener.
798 *
799 * Note: called only from the BH handler context,
800 * so we don't need to lock the hashes.
801 */
802static int __udp4_lib_mcast_deliver(struct sk_buff *skb,
803 struct udphdr *uh,
804 __be32 saddr, __be32 daddr,
805 struct hlist_head udptable[])
806{
807 struct sock *sk;
808 int dif;
809
810 read_lock(&udp_hash_lock);
811 sk = sk_head(&udptable[ntohs(uh->dest) & (UDP_HTABLE_SIZE - 1)]);
812 dif = skb->dev->ifindex;
813 sk = udp_v4_mcast_next(sk, uh->dest, daddr, uh->source, saddr, dif);
814 if (sk) {
815 struct sock *sknext = NULL;
816
817 do {
818 struct sk_buff *skb1 = skb;
819
820 sknext = udp_v4_mcast_next(sk_next(sk), uh->dest, daddr,
821 uh->source, saddr, dif);
822 if (sknext)
823 skb1 = skb_clone(skb, GFP_ATOMIC);
824
825 if (skb1) {
826 int ret = 0;
827
828 bh_lock_sock_nested(sk);
829 if (!sock_owned_by_user(sk))
830 ret = udp_queue_rcv_skb(sk, skb1);
831 else
832 sk_add_backlog(sk, skb1);
833 bh_unlock_sock(sk);
834
835 if (ret > 0)
836 /* we should probably re-process instead
837 * of dropping packets here. */
838 kfree_skb(skb1);
839 }
840 sk = sknext;
841 } while (sknext);
842 } else
843 kfree_skb(skb);
844 read_unlock(&udp_hash_lock);
845 return 0;
846}
847
848/* Initialize UDP checksum. If exited with zero value (success),
849 * CHECKSUM_UNNECESSARY means, that no more checks are required.
850 * Otherwise, csum completion requires chacksumming packet body,
851 * including udp header and folding it to skb->csum.
852 */
853static inline int udp4_csum_init(struct sk_buff *skb, struct udphdr *uh,
854 int proto)
855{
856 const struct iphdr *iph;
857 int err;
858
859 UDP_SKB_CB(skb)->partial_cov = 0;
860 UDP_SKB_CB(skb)->cscov = skb->len;
861
862 if (IS_PROTO_UDPLITE(proto)) {
863 err = udplite_checksum_init(skb, uh);
864 if (err)
865 return err;
866 }
867
868 iph = ip_hdr(skb);
869 if (uh->check == 0) {
870 skb->ip_summed = CHECKSUM_UNNECESSARY;
871 } else if (skb->ip_summed == CHECKSUM_COMPLETE) {
872 if (!csum_tcpudp_magic(iph->saddr, iph->daddr, skb->len,
873 proto, skb->csum))
874 skb->ip_summed = CHECKSUM_UNNECESSARY;
875 }
876 if (!skb_csum_unnecessary(skb))
877 skb->csum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
878 skb->len, proto, 0);
879 /* Probably, we should checksum udp header (it should be in cache
880 * in any case) and data in tiny packets (< rx copybreak).
881 */
882
883 return 0;
884}
885
886/*
887 * All we need to do is get the socket, and then do a checksum.
888 */
889
890int __udp4_lib_rcv(struct sk_buff *skb, struct hlist_head udptable[],
891 int proto)
892{
893 struct sock *sk;
894 struct udphdr *uh = udp_hdr(skb);
895 unsigned short ulen;
896 struct rtable *rt = skb->rtable;
897 __be32 saddr = ip_hdr(skb)->saddr;
898 __be32 daddr = ip_hdr(skb)->daddr;
899
900 /*
901 * Validate the packet.
902 */
903 if (!pskb_may_pull(skb, sizeof(struct udphdr)))
904 goto drop; /* No space for header. */
905
906 ulen = ntohs(uh->len);
907 if (ulen > skb->len)
908 goto short_packet;
909
910 if (IS_PROTO_UDPLITE(proto)) {
911 /* UDP validates ulen. */
912 if (ulen < sizeof(*uh) || pskb_trim_rcsum(skb, ulen))
913 goto short_packet;
914 uh = udp_hdr(skb);
915 }
916
917 if (udp4_csum_init(skb, uh, proto))
918 goto csum_error;
919
920 if (rt->rt_flags & (RTCF_BROADCAST|RTCF_MULTICAST))
921 return __udp4_lib_mcast_deliver(skb, uh, saddr, daddr, udptable);
922
923 sk = __udp4_lib_lookup(skb->dev->nd_net, saddr, uh->source, daddr,
924 uh->dest, inet_iif(skb), udptable);
925
926 if (sk != NULL) {
927 int ret = 0;
928 bh_lock_sock_nested(sk);
929 if (!sock_owned_by_user(sk))
930 ret = udp_queue_rcv_skb(sk, skb);
931 else
932 sk_add_backlog(sk, skb);
933 bh_unlock_sock(sk);
934 sock_put(sk);
935
936 /* a return value > 0 means to resubmit the input, but
937 * it wants the return to be -protocol, or 0
938 */
939 if (ret > 0)
940 return -ret;
941 return 0;
942 }
943
944 if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
945 goto drop;
946 nf_reset(skb);
947
948 /* No socket. Drop packet silently, if checksum is wrong */
949 if (udp_lib_checksum_complete(skb))
950 goto csum_error;
951
952 UDP_INC_STATS_BH(UDP_MIB_NOPORTS, IS_PROTO_UDPLITE(proto));
953 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PORT_UNREACH, 0);
954
955 /*
956 * Hmm. We got an UDP packet to a port to which we
957 * don't wanna listen. Ignore it.
958 */
959 kfree_skb(skb);
960 return 0;
961
962short_packet:
963 LIMIT_NETDEBUG(KERN_DEBUG "UDP%s: short packet: From %u.%u.%u.%u:%u %d/%d to %u.%u.%u.%u:%u\n",
964 IS_PROTO_UDPLITE(proto) ? "-Lite" : "",
965 NIPQUAD(saddr),
966 ntohs(uh->source),
967 ulen,
968 skb->len,
969 NIPQUAD(daddr),
970 ntohs(uh->dest));
971 goto drop;
972
973csum_error:
974 /*
975 * RFC1122: OK. Discards the bad packet silently (as far as
976 * the network is concerned, anyway) as per 4.1.3.4 (MUST).
977 */
978 LIMIT_NETDEBUG(KERN_DEBUG "UDP%s: bad checksum. From %d.%d.%d.%d:%d to %d.%d.%d.%d:%d ulen %d\n",
979 IS_PROTO_UDPLITE(proto) ? "-Lite" : "",
980 NIPQUAD(saddr),
981 ntohs(uh->source),
982 NIPQUAD(daddr),
983 ntohs(uh->dest),
984 ulen);
985drop:
986 UDP_INC_STATS_BH(UDP_MIB_INERRORS, IS_PROTO_UDPLITE(proto));
987 kfree_skb(skb);
988 return 0;
989}
990
991int udp_rcv(struct sk_buff *skb)
992{
993 return __udp4_lib_rcv(skb, udp_hash, IPPROTO_UDP);
994}
995
996int udp_destroy_sock(struct sock *sk)
997{
998 lock_sock(sk);
999 udp_flush_pending_frames(sk);
1000 release_sock(sk);
1001 return 0;
1002}
1003
1004int udp_setsockopt(struct sock *sk, int level, int optname,
1005 char __user *optval, int optlen)
1006{
1007 if (IS_SOL_UDPFAMILY(level))
1008 return udp_lib_setsockopt(sk, level, optname, optval, optlen,
1009 udp_push_pending_frames);
1010 return ip_setsockopt(sk, level, optname, optval, optlen);
1011}
1012
1013#ifdef CONFIG_COMPAT
1014int compat_udp_setsockopt(struct sock *sk, int level, int optname,
1015 char __user *optval, int optlen)
1016{
1017 if (IS_SOL_UDPFAMILY(level))
1018 return udp_lib_setsockopt(sk, level, optname, optval, optlen,
1019 udp_push_pending_frames);
1020 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
1021}
1022#endif
1023
1024int udp_getsockopt(struct sock *sk, int level, int optname,
1025 char __user *optval, int __user *optlen)
1026{
1027 if (IS_SOL_UDPFAMILY(level))
1028 return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1029 return ip_getsockopt(sk, level, optname, optval, optlen);
1030}
1031
1032#ifdef CONFIG_COMPAT
1033int compat_udp_getsockopt(struct sock *sk, int level, int optname,
1034 char __user *optval, int __user *optlen)
1035{
1036 if (IS_SOL_UDPFAMILY(level))
1037 return udp_lib_getsockopt(sk, level, optname, optval, optlen);
1038 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
1039}
1040#endif
1041
1042/* ------------------------------------------------------------------------ */
1043DEFINE_PROTO_INUSE(udp)
1044
1045struct proto udp_prot = {
1046 .name = "UDP",
1047 .owner = THIS_MODULE,
1048 .close = udp_lib_close,
1049 .connect = ip4_datagram_connect,
1050 .disconnect = udp_disconnect,
1051 .ioctl = udp_ioctl,
1052 .destroy = udp_destroy_sock,
1053 .setsockopt = udp_setsockopt,
1054 .getsockopt = udp_getsockopt,
1055 .sendmsg = udp_sendmsg,
1056 .recvmsg = udp_recvmsg,
1057 .sendpage = udp_sendpage,
1058 .backlog_rcv = udp_queue_rcv_skb,
1059 .hash = udp_lib_hash,
1060 .unhash = udp_lib_unhash,
1061 .get_port = udp_v4_get_port,
1062 .memory_allocated = &udp_memory_allocated,
1063 .sysctl_mem = sysctl_udp_mem,
1064 .sysctl_wmem = &sysctl_udp_wmem_min,
1065 .sysctl_rmem = &sysctl_udp_rmem_min,
1066 .obj_size = sizeof(struct udp_sock),
1067#ifdef CONFIG_COMPAT
1068 .compat_setsockopt = compat_udp_setsockopt,
1069 .compat_getsockopt = compat_udp_getsockopt,
1070#endif
1071 REF_PROTO_INUSE(udp)
1072};
1073
1074/* ------------------------------------------------------------------------ */
1075static void udp4_format_sock(struct sock *sp, char *tmpbuf, int bucket)
1076{
1077 struct inet_sock *inet = inet_sk(sp);
1078 __be32 dest = inet->daddr;
1079 __be32 src = inet->rcv_saddr;
1080 __u16 destp = ntohs(inet->dport);
1081 __u16 srcp = ntohs(inet->sport);
1082
1083 sprintf(tmpbuf, "%4d: %08X:%04X %08X:%04X"
1084 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p",
1085 bucket, src, srcp, dest, destp, sp->sk_state,
1086 atomic_read(&sp->sk_wmem_alloc),
1087 atomic_read(&sp->sk_rmem_alloc),
1088 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
1089 atomic_read(&sp->sk_refcnt), sp);
1090}
1091
1092int udp4_seq_show(struct seq_file *seq, void *v)
1093{
1094 if (v == SEQ_START_TOKEN)
1095 seq_printf(seq, "%-127s\n",
1096 " sl local_address rem_address st tx_queue "
1097 "rx_queue tr tm->when retrnsmt uid timeout "
1098 "inode");
1099 else {
1100 char tmpbuf[129];
1101 struct udp_iter_state *state = seq->private;
1102
1103 udp4_format_sock(v, tmpbuf, state->bucket);
1104 seq_printf(seq, "%-127s\n", tmpbuf);
1105 }
1106 return 0;
1107}
1108
1109/* ------------------------------------------------------------------------ */
1110#ifdef CONFIG_PROC_FS
1111static struct file_operations udp4_seq_fops;
1112static struct udp_seq_afinfo udp4_seq_afinfo = {
1113 .owner = THIS_MODULE,
1114 .name = "udp",
1115 .family = AF_INET,
1116 .hashtable = udp_hash,
1117 .seq_show = udp4_seq_show,
1118 .seq_fops = &udp4_seq_fops,
1119};
1120
1121int __init udp4_proc_init(void)
1122{
1123 return udp_proc_register(&udp4_seq_afinfo);
1124}
1125
1126void udp4_proc_exit(void)
1127{
1128 udp_proc_unregister(&udp4_seq_afinfo);
1129}
1130#endif /* CONFIG_PROC_FS */
1131
1132EXPORT_SYMBOL(udp_prot);
1133EXPORT_SYMBOL(udp_sendmsg);
1134
diff --git a/net/ipv4/udplite_ipv4.c b/net/ipv4/udplite.c
index d49c6d68c8a9..d49c6d68c8a9 100644
--- a/net/ipv4/udplite_ipv4.c
+++ b/net/ipv4/udplite.c
diff --git a/net/ipv6/Makefile b/net/ipv6/Makefile
index 107051f7c227..ae14617e607f 100644
--- a/net/ipv6/Makefile
+++ b/net/ipv6/Makefile
@@ -6,7 +6,7 @@ obj-$(CONFIG_IPV6) += ipv6.o
6 6
7ipv6-objs := af_inet6.o anycast.o ip6_output.o ip6_input.o addrconf.o \ 7ipv6-objs := af_inet6.o anycast.o ip6_output.o ip6_input.o addrconf.o \
8 addrlabel.o \ 8 addrlabel.o \
9 route.o ip6_fib.o ipv6_sockglue.o ndisc.o udp_ipv6.o \ 9 route.o ip6_fib.o ipv6_sockglue.o ndisc.o udp.o udplite.o \
10 raw.o protocol.o icmp.o mcast.o reassembly.o tcp_ipv6.o \ 10 raw.o protocol.o icmp.o mcast.o reassembly.o tcp_ipv6.o \
11 exthdrs.o datagram.o ip6_flowlabel.o inet6_connection_sock.o 11 exthdrs.o datagram.o ip6_flowlabel.o inet6_connection_sock.o
12 12
@@ -17,7 +17,6 @@ ipv6-$(CONFIG_NETFILTER) += netfilter.o
17ipv6-$(CONFIG_IPV6_MULTIPLE_TABLES) += fib6_rules.o 17ipv6-$(CONFIG_IPV6_MULTIPLE_TABLES) += fib6_rules.o
18ipv6-$(CONFIG_PROC_FS) += proc.o 18ipv6-$(CONFIG_PROC_FS) += proc.o
19ipv6-$(CONFIG_SYN_COOKIES) += syncookies.o 19ipv6-$(CONFIG_SYN_COOKIES) += syncookies.o
20ipv6-$(CONFIG_IP_UDPLITE) += udplite_ipv6.o
21 20
22ipv6-objs += $(ipv6-y) 21ipv6-objs += $(ipv6-y)
23 22
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index afe9276d0420..730a861b8f41 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -813,16 +813,12 @@ static int __init init_ipv6_mibs(void)
813 goto err_icmpmsg_mib; 813 goto err_icmpmsg_mib;
814 if (snmp_mib_init((void **)udp_stats_in6, sizeof (struct udp_mib)) < 0) 814 if (snmp_mib_init((void **)udp_stats_in6, sizeof (struct udp_mib)) < 0)
815 goto err_udp_mib; 815 goto err_udp_mib;
816#ifdef CONFIG_IP_UDPLITE
817 if (snmp_mib_init((void **)udplite_stats_in6, 816 if (snmp_mib_init((void **)udplite_stats_in6,
818 sizeof (struct udp_mib)) < 0) 817 sizeof (struct udp_mib)) < 0)
819 goto err_udplite_mib; 818 goto err_udplite_mib;
820#endif
821 return 0; 819 return 0;
822 820
823#ifdef CONFIG_IP_UDPLITE
824err_udplite_mib: 821err_udplite_mib:
825#endif
826 snmp_mib_free((void **)udp_stats_in6); 822 snmp_mib_free((void **)udp_stats_in6);
827err_udp_mib: 823err_udp_mib:
828 snmp_mib_free((void **)icmpv6msg_statistics); 824 snmp_mib_free((void **)icmpv6msg_statistics);
@@ -841,9 +837,7 @@ static void cleanup_ipv6_mibs(void)
841 snmp_mib_free((void **)icmpv6_statistics); 837 snmp_mib_free((void **)icmpv6_statistics);
842 snmp_mib_free((void **)icmpv6msg_statistics); 838 snmp_mib_free((void **)icmpv6msg_statistics);
843 snmp_mib_free((void **)udp_stats_in6); 839 snmp_mib_free((void **)udp_stats_in6);
844#ifdef CONFIG_IP_UDPLITE
845 snmp_mib_free((void **)udplite_stats_in6); 840 snmp_mib_free((void **)udplite_stats_in6);
846#endif
847} 841}
848 842
849static int inet6_net_init(struct net *net) 843static int inet6_net_init(struct net *net)
@@ -888,11 +882,9 @@ static int __init inet6_init(void)
888 if (err) 882 if (err)
889 goto out_unregister_tcp_proto; 883 goto out_unregister_tcp_proto;
890 884
891#ifdef CONFIG_IP_UDPLITE
892 err = proto_register(&udplitev6_prot, 1); 885 err = proto_register(&udplitev6_prot, 1);
893 if (err) 886 if (err)
894 goto out_unregister_udp_proto; 887 goto out_unregister_udp_proto;
895#endif
896 888
897 err = proto_register(&rawv6_prot, 1); 889 err = proto_register(&rawv6_prot, 1);
898 if (err) 890 if (err)
@@ -1063,10 +1055,8 @@ out_sock_register_fail:
1063out_unregister_raw_proto: 1055out_unregister_raw_proto:
1064 proto_unregister(&rawv6_prot); 1056 proto_unregister(&rawv6_prot);
1065out_unregister_udplite_proto: 1057out_unregister_udplite_proto:
1066#ifdef CONFIG_IP_UDPLITE
1067 proto_unregister(&udplitev6_prot); 1058 proto_unregister(&udplitev6_prot);
1068out_unregister_udp_proto: 1059out_unregister_udp_proto:
1069#endif
1070 proto_unregister(&udpv6_prot); 1060 proto_unregister(&udpv6_prot);
1071out_unregister_tcp_proto: 1061out_unregister_tcp_proto:
1072 proto_unregister(&tcpv6_prot); 1062 proto_unregister(&tcpv6_prot);
@@ -1085,9 +1075,7 @@ static void __exit inet6_exit(void)
1085 ipv6_sysctl_unregister(); 1075 ipv6_sysctl_unregister();
1086#endif 1076#endif
1087 udpv6_exit(); 1077 udpv6_exit();
1088#ifdef CONFIG_IP_UDPLITE
1089 udplitev6_exit(); 1078 udplitev6_exit();
1090#endif
1091 tcpv6_exit(); 1079 tcpv6_exit();
1092 1080
1093 /* Cleanup code parts. */ 1081 /* Cleanup code parts. */
@@ -1117,9 +1105,7 @@ static void __exit inet6_exit(void)
1117 unregister_pernet_subsys(&inet6_net_ops); 1105 unregister_pernet_subsys(&inet6_net_ops);
1118 cleanup_ipv6_mibs(); 1106 cleanup_ipv6_mibs();
1119 proto_unregister(&rawv6_prot); 1107 proto_unregister(&rawv6_prot);
1120#ifdef CONFIG_IP_UDPLITE
1121 proto_unregister(&udplitev6_prot); 1108 proto_unregister(&udplitev6_prot);
1122#endif
1123 proto_unregister(&udpv6_prot); 1109 proto_unregister(&udpv6_prot);
1124 proto_unregister(&tcpv6_prot); 1110 proto_unregister(&tcpv6_prot);
1125} 1111}
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index 937018529d18..98762fde2b65 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -404,6 +404,7 @@ int ip6_forward(struct sk_buff *skb)
404 struct dst_entry *dst = skb->dst; 404 struct dst_entry *dst = skb->dst;
405 struct ipv6hdr *hdr = ipv6_hdr(skb); 405 struct ipv6hdr *hdr = ipv6_hdr(skb);
406 struct inet6_skb_parm *opt = IP6CB(skb); 406 struct inet6_skb_parm *opt = IP6CB(skb);
407 struct net *net = dst->dev->nd_net;
407 408
408 if (ipv6_devconf.forwarding == 0) 409 if (ipv6_devconf.forwarding == 0)
409 goto error; 410 goto error;
@@ -450,7 +451,7 @@ int ip6_forward(struct sk_buff *skb)
450 451
451 /* XXX: idev->cnf.proxy_ndp? */ 452 /* XXX: idev->cnf.proxy_ndp? */
452 if (ipv6_devconf.proxy_ndp && 453 if (ipv6_devconf.proxy_ndp &&
453 pneigh_lookup(&nd_tbl, &init_net, &hdr->daddr, skb->dev, 0)) { 454 pneigh_lookup(&nd_tbl, net, &hdr->daddr, skb->dev, 0)) {
454 int proxied = ip6_forward_proxy_check(skb); 455 int proxied = ip6_forward_proxy_check(skb);
455 if (proxied > 0) 456 if (proxied > 0)
456 return ip6_input(skb); 457 return ip6_input(skb);
@@ -911,9 +912,10 @@ static int ip6_dst_lookup_tail(struct sock *sk,
911 struct dst_entry **dst, struct flowi *fl) 912 struct dst_entry **dst, struct flowi *fl)
912{ 913{
913 int err; 914 int err;
915 struct net *net = sk->sk_net;
914 916
915 if (*dst == NULL) 917 if (*dst == NULL)
916 *dst = ip6_route_output(sk->sk_net, sk, fl); 918 *dst = ip6_route_output(net, sk, fl);
917 919
918 if ((err = (*dst)->error)) 920 if ((err = (*dst)->error))
919 goto out_err_release; 921 goto out_err_release;
@@ -939,7 +941,7 @@ static int ip6_dst_lookup_tail(struct sock *sk,
939 struct flowi fl_gw; 941 struct flowi fl_gw;
940 int redirect; 942 int redirect;
941 943
942 ifp = ipv6_get_ifaddr(&init_net, &fl->fl6_src, 944 ifp = ipv6_get_ifaddr(net, &fl->fl6_src,
943 (*dst)->dev, 1); 945 (*dst)->dev, 1);
944 946
945 redirect = (ifp && ifp->flags & IFA_F_OPTIMISTIC); 947 redirect = (ifp && ifp->flags & IFA_F_OPTIMISTIC);
@@ -954,7 +956,7 @@ static int ip6_dst_lookup_tail(struct sock *sk,
954 dst_release(*dst); 956 dst_release(*dst);
955 memcpy(&fl_gw, fl, sizeof(struct flowi)); 957 memcpy(&fl_gw, fl, sizeof(struct flowi));
956 memset(&fl_gw.fl6_dst, 0, sizeof(struct in6_addr)); 958 memset(&fl_gw.fl6_dst, 0, sizeof(struct in6_addr));
957 *dst = ip6_route_output(sk->sk_net, sk, &fl_gw); 959 *dst = ip6_route_output(net, sk, &fl_gw);
958 if ((err = (*dst)->error)) 960 if ((err = (*dst)->error))
959 goto out_err_release; 961 goto out_err_release;
960 } 962 }
diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index 3bbfdff698d2..c11c76cab371 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -107,6 +107,7 @@ static int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
107 char __user *optval, int optlen) 107 char __user *optval, int optlen)
108{ 108{
109 struct ipv6_pinfo *np = inet6_sk(sk); 109 struct ipv6_pinfo *np = inet6_sk(sk);
110 struct net *net = sk->sk_net;
110 int val, valbool; 111 int val, valbool;
111 int retv = -ENOPROTOOPT; 112 int retv = -ENOPROTOOPT;
112 113
@@ -127,9 +128,7 @@ static int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
127 struct sk_buff *pktopt; 128 struct sk_buff *pktopt;
128 129
129 if (sk->sk_protocol != IPPROTO_UDP && 130 if (sk->sk_protocol != IPPROTO_UDP &&
130#ifdef CONFIG_IP_UDPLITE
131 sk->sk_protocol != IPPROTO_UDPLITE && 131 sk->sk_protocol != IPPROTO_UDPLITE &&
132#endif
133 sk->sk_protocol != IPPROTO_TCP) 132 sk->sk_protocol != IPPROTO_TCP)
134 break; 133 break;
135 134
@@ -169,7 +168,7 @@ static int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
169 } else { 168 } else {
170 struct proto *prot = &udp_prot; 169 struct proto *prot = &udp_prot;
171 170
172 if (IS_PROTO_UDPLITE(sk->sk_protocol)) 171 if (sk->sk_protocol == IPPROTO_UDPLITE)
173 prot = &udplite_prot; 172 prot = &udplite_prot;
174 local_bh_disable(); 173 local_bh_disable();
175 sock_prot_inuse_add(sk->sk_prot, -1); 174 sock_prot_inuse_add(sk->sk_prot, -1);
@@ -434,7 +433,7 @@ done:
434 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != val) 433 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != val)
435 goto e_inval; 434 goto e_inval;
436 435
437 if (__dev_get_by_index(&init_net, val) == NULL) { 436 if (__dev_get_by_index(net, val) == NULL) {
438 retv = -ENODEV; 437 retv = -ENODEV;
439 break; 438 break;
440 } 439 }
@@ -734,9 +733,7 @@ static int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
734 switch (optname) { 733 switch (optname) {
735 case IPV6_ADDRFORM: 734 case IPV6_ADDRFORM:
736 if (sk->sk_protocol != IPPROTO_UDP && 735 if (sk->sk_protocol != IPPROTO_UDP &&
737#ifdef CONFIG_IP_UDPLITE
738 sk->sk_protocol != IPPROTO_UDPLITE && 736 sk->sk_protocol != IPPROTO_UDPLITE &&
739#endif
740 sk->sk_protocol != IPPROTO_TCP) 737 sk->sk_protocol != IPPROTO_TCP)
741 return -EINVAL; 738 return -EINVAL;
742 if (sk->sk_state != TCP_ESTABLISHED) 739 if (sk->sk_state != TCP_ESTABLISHED)
diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
index 197ca390a15d..f2879056fab0 100644
--- a/net/ipv6/mcast.c
+++ b/net/ipv6/mcast.c
@@ -126,8 +126,6 @@ static struct in6_addr mld2_all_mcr = MLD2_ALL_MCR_INIT;
126/* Big mc list lock for all the sockets */ 126/* Big mc list lock for all the sockets */
127static DEFINE_RWLOCK(ipv6_sk_mc_lock); 127static DEFINE_RWLOCK(ipv6_sk_mc_lock);
128 128
129static struct socket *igmp6_socket;
130
131int __ipv6_dev_mc_dec(struct inet6_dev *idev, struct in6_addr *addr); 129int __ipv6_dev_mc_dec(struct inet6_dev *idev, struct in6_addr *addr);
132 130
133static void igmp6_join_group(struct ifmcaddr6 *ma); 131static void igmp6_join_group(struct ifmcaddr6 *ma);
@@ -183,6 +181,7 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, struct in6_addr *addr)
183 struct net_device *dev = NULL; 181 struct net_device *dev = NULL;
184 struct ipv6_mc_socklist *mc_lst; 182 struct ipv6_mc_socklist *mc_lst;
185 struct ipv6_pinfo *np = inet6_sk(sk); 183 struct ipv6_pinfo *np = inet6_sk(sk);
184 struct net *net = sk->sk_net;
186 int err; 185 int err;
187 186
188 if (!ipv6_addr_is_multicast(addr)) 187 if (!ipv6_addr_is_multicast(addr))
@@ -208,14 +207,14 @@ int ipv6_sock_mc_join(struct sock *sk, int ifindex, struct in6_addr *addr)
208 207
209 if (ifindex == 0) { 208 if (ifindex == 0) {
210 struct rt6_info *rt; 209 struct rt6_info *rt;
211 rt = rt6_lookup(&init_net, addr, NULL, 0, 0); 210 rt = rt6_lookup(net, addr, NULL, 0, 0);
212 if (rt) { 211 if (rt) {
213 dev = rt->rt6i_dev; 212 dev = rt->rt6i_dev;
214 dev_hold(dev); 213 dev_hold(dev);
215 dst_release(&rt->u.dst); 214 dst_release(&rt->u.dst);
216 } 215 }
217 } else 216 } else
218 dev = dev_get_by_index(&init_net, ifindex); 217 dev = dev_get_by_index(net, ifindex);
219 218
220 if (dev == NULL) { 219 if (dev == NULL) {
221 sock_kfree_s(sk, mc_lst, sizeof(*mc_lst)); 220 sock_kfree_s(sk, mc_lst, sizeof(*mc_lst));
@@ -256,6 +255,7 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, struct in6_addr *addr)
256{ 255{
257 struct ipv6_pinfo *np = inet6_sk(sk); 256 struct ipv6_pinfo *np = inet6_sk(sk);
258 struct ipv6_mc_socklist *mc_lst, **lnk; 257 struct ipv6_mc_socklist *mc_lst, **lnk;
258 struct net *net = sk->sk_net;
259 259
260 write_lock_bh(&ipv6_sk_mc_lock); 260 write_lock_bh(&ipv6_sk_mc_lock);
261 for (lnk = &np->ipv6_mc_list; (mc_lst = *lnk) !=NULL ; lnk = &mc_lst->next) { 261 for (lnk = &np->ipv6_mc_list; (mc_lst = *lnk) !=NULL ; lnk = &mc_lst->next) {
@@ -266,7 +266,8 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, struct in6_addr *addr)
266 *lnk = mc_lst->next; 266 *lnk = mc_lst->next;
267 write_unlock_bh(&ipv6_sk_mc_lock); 267 write_unlock_bh(&ipv6_sk_mc_lock);
268 268
269 if ((dev = dev_get_by_index(&init_net, mc_lst->ifindex)) != NULL) { 269 dev = dev_get_by_index(net, mc_lst->ifindex);
270 if (dev != NULL) {
270 struct inet6_dev *idev = in6_dev_get(dev); 271 struct inet6_dev *idev = in6_dev_get(dev);
271 272
272 (void) ip6_mc_leave_src(sk, mc_lst, idev); 273 (void) ip6_mc_leave_src(sk, mc_lst, idev);
@@ -286,7 +287,9 @@ int ipv6_sock_mc_drop(struct sock *sk, int ifindex, struct in6_addr *addr)
286 return -EADDRNOTAVAIL; 287 return -EADDRNOTAVAIL;
287} 288}
288 289
289static struct inet6_dev *ip6_mc_find_dev(struct in6_addr *group, int ifindex) 290static struct inet6_dev *ip6_mc_find_dev(struct net *net,
291 struct in6_addr *group,
292 int ifindex)
290{ 293{
291 struct net_device *dev = NULL; 294 struct net_device *dev = NULL;
292 struct inet6_dev *idev = NULL; 295 struct inet6_dev *idev = NULL;
@@ -294,14 +297,14 @@ static struct inet6_dev *ip6_mc_find_dev(struct in6_addr *group, int ifindex)
294 if (ifindex == 0) { 297 if (ifindex == 0) {
295 struct rt6_info *rt; 298 struct rt6_info *rt;
296 299
297 rt = rt6_lookup(&init_net, group, NULL, 0, 0); 300 rt = rt6_lookup(net, group, NULL, 0, 0);
298 if (rt) { 301 if (rt) {
299 dev = rt->rt6i_dev; 302 dev = rt->rt6i_dev;
300 dev_hold(dev); 303 dev_hold(dev);
301 dst_release(&rt->u.dst); 304 dst_release(&rt->u.dst);
302 } 305 }
303 } else 306 } else
304 dev = dev_get_by_index(&init_net, ifindex); 307 dev = dev_get_by_index(net, ifindex);
305 308
306 if (!dev) 309 if (!dev)
307 return NULL; 310 return NULL;
@@ -324,6 +327,7 @@ void ipv6_sock_mc_close(struct sock *sk)
324{ 327{
325 struct ipv6_pinfo *np = inet6_sk(sk); 328 struct ipv6_pinfo *np = inet6_sk(sk);
326 struct ipv6_mc_socklist *mc_lst; 329 struct ipv6_mc_socklist *mc_lst;
330 struct net *net = sk->sk_net;
327 331
328 write_lock_bh(&ipv6_sk_mc_lock); 332 write_lock_bh(&ipv6_sk_mc_lock);
329 while ((mc_lst = np->ipv6_mc_list) != NULL) { 333 while ((mc_lst = np->ipv6_mc_list) != NULL) {
@@ -332,7 +336,7 @@ void ipv6_sock_mc_close(struct sock *sk)
332 np->ipv6_mc_list = mc_lst->next; 336 np->ipv6_mc_list = mc_lst->next;
333 write_unlock_bh(&ipv6_sk_mc_lock); 337 write_unlock_bh(&ipv6_sk_mc_lock);
334 338
335 dev = dev_get_by_index(&init_net, mc_lst->ifindex); 339 dev = dev_get_by_index(net, mc_lst->ifindex);
336 if (dev) { 340 if (dev) {
337 struct inet6_dev *idev = in6_dev_get(dev); 341 struct inet6_dev *idev = in6_dev_get(dev);
338 342
@@ -361,6 +365,7 @@ int ip6_mc_source(int add, int omode, struct sock *sk,
361 struct inet6_dev *idev; 365 struct inet6_dev *idev;
362 struct ipv6_pinfo *inet6 = inet6_sk(sk); 366 struct ipv6_pinfo *inet6 = inet6_sk(sk);
363 struct ip6_sf_socklist *psl; 367 struct ip6_sf_socklist *psl;
368 struct net *net = sk->sk_net;
364 int i, j, rv; 369 int i, j, rv;
365 int leavegroup = 0; 370 int leavegroup = 0;
366 int pmclocked = 0; 371 int pmclocked = 0;
@@ -376,7 +381,7 @@ int ip6_mc_source(int add, int omode, struct sock *sk,
376 if (!ipv6_addr_is_multicast(group)) 381 if (!ipv6_addr_is_multicast(group))
377 return -EINVAL; 382 return -EINVAL;
378 383
379 idev = ip6_mc_find_dev(group, pgsr->gsr_interface); 384 idev = ip6_mc_find_dev(net, group, pgsr->gsr_interface);
380 if (!idev) 385 if (!idev)
381 return -ENODEV; 386 return -ENODEV;
382 dev = idev->dev; 387 dev = idev->dev;
@@ -500,6 +505,7 @@ int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf)
500 struct inet6_dev *idev; 505 struct inet6_dev *idev;
501 struct ipv6_pinfo *inet6 = inet6_sk(sk); 506 struct ipv6_pinfo *inet6 = inet6_sk(sk);
502 struct ip6_sf_socklist *newpsl, *psl; 507 struct ip6_sf_socklist *newpsl, *psl;
508 struct net *net = sk->sk_net;
503 int leavegroup = 0; 509 int leavegroup = 0;
504 int i, err; 510 int i, err;
505 511
@@ -511,7 +517,7 @@ int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf)
511 gsf->gf_fmode != MCAST_EXCLUDE) 517 gsf->gf_fmode != MCAST_EXCLUDE)
512 return -EINVAL; 518 return -EINVAL;
513 519
514 idev = ip6_mc_find_dev(group, gsf->gf_interface); 520 idev = ip6_mc_find_dev(net, group, gsf->gf_interface);
515 521
516 if (!idev) 522 if (!idev)
517 return -ENODEV; 523 return -ENODEV;
@@ -592,13 +598,14 @@ int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
592 struct net_device *dev; 598 struct net_device *dev;
593 struct ipv6_pinfo *inet6 = inet6_sk(sk); 599 struct ipv6_pinfo *inet6 = inet6_sk(sk);
594 struct ip6_sf_socklist *psl; 600 struct ip6_sf_socklist *psl;
601 struct net *net = sk->sk_net;
595 602
596 group = &((struct sockaddr_in6 *)&gsf->gf_group)->sin6_addr; 603 group = &((struct sockaddr_in6 *)&gsf->gf_group)->sin6_addr;
597 604
598 if (!ipv6_addr_is_multicast(group)) 605 if (!ipv6_addr_is_multicast(group))
599 return -EINVAL; 606 return -EINVAL;
600 607
601 idev = ip6_mc_find_dev(group, gsf->gf_interface); 608 idev = ip6_mc_find_dev(net, group, gsf->gf_interface);
602 609
603 if (!idev) 610 if (!idev)
604 return -ENODEV; 611 return -ENODEV;
@@ -1393,7 +1400,8 @@ mld_scount(struct ifmcaddr6 *pmc, int type, int gdeleted, int sdeleted)
1393 1400
1394static struct sk_buff *mld_newpack(struct net_device *dev, int size) 1401static struct sk_buff *mld_newpack(struct net_device *dev, int size)
1395{ 1402{
1396 struct sock *sk = igmp6_socket->sk; 1403 struct net *net = dev->nd_net;
1404 struct sock *sk = net->ipv6.igmp_sk;
1397 struct sk_buff *skb; 1405 struct sk_buff *skb;
1398 struct mld2_report *pmr; 1406 struct mld2_report *pmr;
1399 struct in6_addr addr_buf; 1407 struct in6_addr addr_buf;
@@ -1440,6 +1448,7 @@ static void mld_sendpack(struct sk_buff *skb)
1440 (struct mld2_report *)skb_transport_header(skb); 1448 (struct mld2_report *)skb_transport_header(skb);
1441 int payload_len, mldlen; 1449 int payload_len, mldlen;
1442 struct inet6_dev *idev = in6_dev_get(skb->dev); 1450 struct inet6_dev *idev = in6_dev_get(skb->dev);
1451 struct net *net = skb->dev->nd_net;
1443 int err; 1452 int err;
1444 struct flowi fl; 1453 struct flowi fl;
1445 1454
@@ -1459,7 +1468,7 @@ static void mld_sendpack(struct sk_buff *skb)
1459 goto err_out; 1468 goto err_out;
1460 } 1469 }
1461 1470
1462 icmpv6_flow_init(igmp6_socket->sk, &fl, ICMPV6_MLD2_REPORT, 1471 icmpv6_flow_init(net->ipv6.igmp_sk, &fl, ICMPV6_MLD2_REPORT,
1463 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr, 1472 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr,
1464 skb->dev->ifindex); 1473 skb->dev->ifindex);
1465 1474
@@ -1753,7 +1762,8 @@ static void mld_send_cr(struct inet6_dev *idev)
1753 1762
1754static void igmp6_send(struct in6_addr *addr, struct net_device *dev, int type) 1763static void igmp6_send(struct in6_addr *addr, struct net_device *dev, int type)
1755{ 1764{
1756 struct sock *sk = igmp6_socket->sk; 1765 struct net *net = dev->nd_net;
1766 struct sock *sk = net->ipv6.igmp_sk;
1757 struct inet6_dev *idev; 1767 struct inet6_dev *idev;
1758 struct sk_buff *skb; 1768 struct sk_buff *skb;
1759 struct icmp6hdr *hdr; 1769 struct icmp6hdr *hdr;
@@ -1824,7 +1834,7 @@ static void igmp6_send(struct in6_addr *addr, struct net_device *dev, int type)
1824 goto err_out; 1834 goto err_out;
1825 } 1835 }
1826 1836
1827 icmpv6_flow_init(igmp6_socket->sk, &fl, type, 1837 icmpv6_flow_init(sk, &fl, type,
1828 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr, 1838 &ipv6_hdr(skb)->saddr, &ipv6_hdr(skb)->daddr,
1829 skb->dev->ifindex); 1839 skb->dev->ifindex);
1830 1840
@@ -2334,6 +2344,7 @@ void ipv6_mc_destroy_dev(struct inet6_dev *idev)
2334 2344
2335#ifdef CONFIG_PROC_FS 2345#ifdef CONFIG_PROC_FS
2336struct igmp6_mc_iter_state { 2346struct igmp6_mc_iter_state {
2347 struct seq_net_private p;
2337 struct net_device *dev; 2348 struct net_device *dev;
2338 struct inet6_dev *idev; 2349 struct inet6_dev *idev;
2339}; 2350};
@@ -2344,9 +2355,10 @@ static inline struct ifmcaddr6 *igmp6_mc_get_first(struct seq_file *seq)
2344{ 2355{
2345 struct ifmcaddr6 *im = NULL; 2356 struct ifmcaddr6 *im = NULL;
2346 struct igmp6_mc_iter_state *state = igmp6_mc_seq_private(seq); 2357 struct igmp6_mc_iter_state *state = igmp6_mc_seq_private(seq);
2358 struct net *net = state->p.net;
2347 2359
2348 state->idev = NULL; 2360 state->idev = NULL;
2349 for_each_netdev(&init_net, state->dev) { 2361 for_each_netdev(net, state->dev) {
2350 struct inet6_dev *idev; 2362 struct inet6_dev *idev;
2351 idev = in6_dev_get(state->dev); 2363 idev = in6_dev_get(state->dev);
2352 if (!idev) 2364 if (!idev)
@@ -2448,8 +2460,8 @@ static const struct seq_operations igmp6_mc_seq_ops = {
2448 2460
2449static int igmp6_mc_seq_open(struct inode *inode, struct file *file) 2461static int igmp6_mc_seq_open(struct inode *inode, struct file *file)
2450{ 2462{
2451 return seq_open_private(file, &igmp6_mc_seq_ops, 2463 return seq_open_net(inode, file, &igmp6_mc_seq_ops,
2452 sizeof(struct igmp6_mc_iter_state)); 2464 sizeof(struct igmp6_mc_iter_state));
2453} 2465}
2454 2466
2455static const struct file_operations igmp6_mc_seq_fops = { 2467static const struct file_operations igmp6_mc_seq_fops = {
@@ -2457,10 +2469,11 @@ static const struct file_operations igmp6_mc_seq_fops = {
2457 .open = igmp6_mc_seq_open, 2469 .open = igmp6_mc_seq_open,
2458 .read = seq_read, 2470 .read = seq_read,
2459 .llseek = seq_lseek, 2471 .llseek = seq_lseek,
2460 .release = seq_release_private, 2472 .release = seq_release_net,
2461}; 2473};
2462 2474
2463struct igmp6_mcf_iter_state { 2475struct igmp6_mcf_iter_state {
2476 struct seq_net_private p;
2464 struct net_device *dev; 2477 struct net_device *dev;
2465 struct inet6_dev *idev; 2478 struct inet6_dev *idev;
2466 struct ifmcaddr6 *im; 2479 struct ifmcaddr6 *im;
@@ -2473,10 +2486,11 @@ static inline struct ip6_sf_list *igmp6_mcf_get_first(struct seq_file *seq)
2473 struct ip6_sf_list *psf = NULL; 2486 struct ip6_sf_list *psf = NULL;
2474 struct ifmcaddr6 *im = NULL; 2487 struct ifmcaddr6 *im = NULL;
2475 struct igmp6_mcf_iter_state *state = igmp6_mcf_seq_private(seq); 2488 struct igmp6_mcf_iter_state *state = igmp6_mcf_seq_private(seq);
2489 struct net *net = state->p.net;
2476 2490
2477 state->idev = NULL; 2491 state->idev = NULL;
2478 state->im = NULL; 2492 state->im = NULL;
2479 for_each_netdev(&init_net, state->dev) { 2493 for_each_netdev(net, state->dev) {
2480 struct inet6_dev *idev; 2494 struct inet6_dev *idev;
2481 idev = in6_dev_get(state->dev); 2495 idev = in6_dev_get(state->dev);
2482 if (unlikely(idev == NULL)) 2496 if (unlikely(idev == NULL))
@@ -2608,8 +2622,8 @@ static const struct seq_operations igmp6_mcf_seq_ops = {
2608 2622
2609static int igmp6_mcf_seq_open(struct inode *inode, struct file *file) 2623static int igmp6_mcf_seq_open(struct inode *inode, struct file *file)
2610{ 2624{
2611 return seq_open_private(file, &igmp6_mcf_seq_ops, 2625 return seq_open_net(inode, file, &igmp6_mcf_seq_ops,
2612 sizeof(struct igmp6_mcf_iter_state)); 2626 sizeof(struct igmp6_mcf_iter_state));
2613} 2627}
2614 2628
2615static const struct file_operations igmp6_mcf_seq_fops = { 2629static const struct file_operations igmp6_mcf_seq_fops = {
@@ -2617,26 +2631,27 @@ static const struct file_operations igmp6_mcf_seq_fops = {
2617 .open = igmp6_mcf_seq_open, 2631 .open = igmp6_mcf_seq_open,
2618 .read = seq_read, 2632 .read = seq_read,
2619 .llseek = seq_lseek, 2633 .llseek = seq_lseek,
2620 .release = seq_release_private, 2634 .release = seq_release_net,
2621}; 2635};
2622#endif 2636#endif
2623 2637
2624int __init igmp6_init(void) 2638static int igmp6_net_init(struct net *net)
2625{ 2639{
2626 struct ipv6_pinfo *np; 2640 struct ipv6_pinfo *np;
2641 struct socket *sock;
2627 struct sock *sk; 2642 struct sock *sk;
2628 int err; 2643 int err;
2629 2644
2630 err = sock_create_kern(PF_INET6, SOCK_RAW, IPPROTO_ICMPV6, &igmp6_socket); 2645 err = sock_create_kern(PF_INET6, SOCK_RAW, IPPROTO_ICMPV6, &sock);
2631 if (err < 0) { 2646 if (err < 0) {
2632 printk(KERN_ERR 2647 printk(KERN_ERR
2633 "Failed to initialize the IGMP6 control socket (err %d).\n", 2648 "Failed to initialize the IGMP6 control socket (err %d).\n",
2634 err); 2649 err);
2635 igmp6_socket = NULL; /* For safety. */ 2650 goto out;
2636 return err;
2637 } 2651 }
2638 2652
2639 sk = igmp6_socket->sk; 2653 net->ipv6.igmp_sk = sk = sock->sk;
2654 sk_change_net(sk, net);
2640 sk->sk_allocation = GFP_ATOMIC; 2655 sk->sk_allocation = GFP_ATOMIC;
2641 sk->sk_prot->unhash(sk); 2656 sk->sk_prot->unhash(sk);
2642 2657
@@ -2644,20 +2659,45 @@ int __init igmp6_init(void)
2644 np->hop_limit = 1; 2659 np->hop_limit = 1;
2645 2660
2646#ifdef CONFIG_PROC_FS 2661#ifdef CONFIG_PROC_FS
2647 proc_net_fops_create(&init_net, "igmp6", S_IRUGO, &igmp6_mc_seq_fops); 2662 err = -ENOMEM;
2648 proc_net_fops_create(&init_net, "mcfilter6", S_IRUGO, &igmp6_mcf_seq_fops); 2663 if (!proc_net_fops_create(net, "igmp6", S_IRUGO, &igmp6_mc_seq_fops))
2664 goto out_sock_create;
2665 if (!proc_net_fops_create(net, "mcfilter6", S_IRUGO,
2666 &igmp6_mcf_seq_fops)) {
2667 proc_net_remove(net, "igmp6");
2668 goto out_sock_create;
2669 }
2649#endif 2670#endif
2650 2671
2651 return 0; 2672 err = 0;
2673out:
2674 return err;
2675
2676out_sock_create:
2677 sk_release_kernel(net->ipv6.igmp_sk);
2678 goto out;
2652} 2679}
2653 2680
2654void igmp6_cleanup(void) 2681static void igmp6_net_exit(struct net *net)
2655{ 2682{
2656 sock_release(igmp6_socket); 2683 sk_release_kernel(net->ipv6.igmp_sk);
2657 igmp6_socket = NULL; /* for safety */
2658
2659#ifdef CONFIG_PROC_FS 2684#ifdef CONFIG_PROC_FS
2660 proc_net_remove(&init_net, "mcfilter6"); 2685 proc_net_remove(net, "mcfilter6");
2661 proc_net_remove(&init_net, "igmp6"); 2686 proc_net_remove(net, "igmp6");
2662#endif 2687#endif
2663} 2688}
2689
2690static struct pernet_operations igmp6_net_ops = {
2691 .init = igmp6_net_init,
2692 .exit = igmp6_net_exit,
2693};
2694
2695int __init igmp6_init(void)
2696{
2697 return register_pernet_subsys(&igmp6_net_ops);
2698}
2699
2700void igmp6_cleanup(void)
2701{
2702 unregister_pernet_subsys(&igmp6_net_ops);
2703}
diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c
index e0d023360a88..24e76ed98884 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -89,8 +89,6 @@
89#include <linux/netfilter.h> 89#include <linux/netfilter.h>
90#include <linux/netfilter_ipv6.h> 90#include <linux/netfilter_ipv6.h>
91 91
92static struct socket *ndisc_socket;
93
94static u32 ndisc_hash(const void *pkey, const struct net_device *dev); 92static u32 ndisc_hash(const void *pkey, const struct net_device *dev);
95static int ndisc_constructor(struct neighbour *neigh); 93static int ndisc_constructor(struct neighbour *neigh);
96static void ndisc_solicit(struct neighbour *neigh, struct sk_buff *skb); 94static void ndisc_solicit(struct neighbour *neigh, struct sk_buff *skb);
@@ -449,7 +447,8 @@ static void __ndisc_send(struct net_device *dev,
449{ 447{
450 struct flowi fl; 448 struct flowi fl;
451 struct dst_entry *dst; 449 struct dst_entry *dst;
452 struct sock *sk = ndisc_socket->sk; 450 struct net *net = dev->nd_net;
451 struct sock *sk = net->ipv6.ndisc_sk;
453 struct sk_buff *skb; 452 struct sk_buff *skb;
454 struct icmp6hdr *hdr; 453 struct icmp6hdr *hdr;
455 struct inet6_dev *idev; 454 struct inet6_dev *idev;
@@ -459,8 +458,7 @@ static void __ndisc_send(struct net_device *dev,
459 458
460 type = icmp6h->icmp6_type; 459 type = icmp6h->icmp6_type;
461 460
462 icmpv6_flow_init(ndisc_socket->sk, &fl, type, 461 icmpv6_flow_init(sk, &fl, type, saddr, daddr, dev->ifindex);
463 saddr, daddr, dev->ifindex);
464 462
465 dst = icmp6_dst_alloc(dev, neigh, daddr); 463 dst = icmp6_dst_alloc(dev, neigh, daddr);
466 if (!dst) 464 if (!dst)
@@ -541,7 +539,7 @@ static void ndisc_send_na(struct net_device *dev, struct neighbour *neigh,
541 }; 539 };
542 540
543 /* for anycast or proxy, solicited_addr != src_addr */ 541 /* for anycast or proxy, solicited_addr != src_addr */
544 ifp = ipv6_get_ifaddr(&init_net, solicited_addr, dev, 1); 542 ifp = ipv6_get_ifaddr(dev->nd_net, solicited_addr, dev, 1);
545 if (ifp) { 543 if (ifp) {
546 src_addr = solicited_addr; 544 src_addr = solicited_addr;
547 if (ifp->flags & IFA_F_OPTIMISTIC) 545 if (ifp->flags & IFA_F_OPTIMISTIC)
@@ -601,7 +599,7 @@ void ndisc_send_rs(struct net_device *dev, struct in6_addr *saddr,
601 * suppress the inclusion of the sllao. 599 * suppress the inclusion of the sllao.
602 */ 600 */
603 if (send_sllao) { 601 if (send_sllao) {
604 struct inet6_ifaddr *ifp = ipv6_get_ifaddr(&init_net, saddr, 602 struct inet6_ifaddr *ifp = ipv6_get_ifaddr(dev->nd_net, saddr,
605 dev, 1); 603 dev, 1);
606 if (ifp) { 604 if (ifp) {
607 if (ifp->flags & IFA_F_OPTIMISTIC) { 605 if (ifp->flags & IFA_F_OPTIMISTIC) {
@@ -639,7 +637,7 @@ static void ndisc_solicit(struct neighbour *neigh, struct sk_buff *skb)
639 struct in6_addr *target = (struct in6_addr *)&neigh->primary_key; 637 struct in6_addr *target = (struct in6_addr *)&neigh->primary_key;
640 int probes = atomic_read(&neigh->probes); 638 int probes = atomic_read(&neigh->probes);
641 639
642 if (skb && ipv6_chk_addr(&init_net, &ipv6_hdr(skb)->saddr, dev, 1)) 640 if (skb && ipv6_chk_addr(dev->nd_net, &ipv6_hdr(skb)->saddr, dev, 1))
643 saddr = &ipv6_hdr(skb)->saddr; 641 saddr = &ipv6_hdr(skb)->saddr;
644 642
645 if ((probes -= neigh->parms->ucast_probes) < 0) { 643 if ((probes -= neigh->parms->ucast_probes) < 0) {
@@ -727,7 +725,8 @@ static void ndisc_recv_ns(struct sk_buff *skb)
727 725
728 inc = ipv6_addr_is_multicast(daddr); 726 inc = ipv6_addr_is_multicast(daddr);
729 727
730 if ((ifp = ipv6_get_ifaddr(&init_net, &msg->target, dev, 1)) != NULL) { 728 ifp = ipv6_get_ifaddr(dev->nd_net, &msg->target, dev, 1);
729 if (ifp) {
731 730
732 if (ifp->flags & (IFA_F_TENTATIVE|IFA_F_OPTIMISTIC)) { 731 if (ifp->flags & (IFA_F_TENTATIVE|IFA_F_OPTIMISTIC)) {
733 if (dad) { 732 if (dad) {
@@ -775,7 +774,7 @@ static void ndisc_recv_ns(struct sk_buff *skb)
775 if (ipv6_chk_acast_addr(dev, &msg->target) || 774 if (ipv6_chk_acast_addr(dev, &msg->target) ||
776 (idev->cnf.forwarding && 775 (idev->cnf.forwarding &&
777 (ipv6_devconf.proxy_ndp || idev->cnf.proxy_ndp) && 776 (ipv6_devconf.proxy_ndp || idev->cnf.proxy_ndp) &&
778 (pneigh = pneigh_lookup(&nd_tbl, &init_net, 777 (pneigh = pneigh_lookup(&nd_tbl, dev->nd_net,
779 &msg->target, dev, 0)) != NULL)) { 778 &msg->target, dev, 0)) != NULL)) {
780 if (!(NEIGH_CB(skb)->flags & LOCALLY_ENQUEUED) && 779 if (!(NEIGH_CB(skb)->flags & LOCALLY_ENQUEUED) &&
781 skb->pkt_type != PACKET_HOST && 780 skb->pkt_type != PACKET_HOST &&
@@ -885,7 +884,8 @@ static void ndisc_recv_na(struct sk_buff *skb)
885 return; 884 return;
886 } 885 }
887 } 886 }
888 if ((ifp = ipv6_get_ifaddr(&init_net, &msg->target, dev, 1))) { 887 ifp = ipv6_get_ifaddr(dev->nd_net, &msg->target, dev, 1);
888 if (ifp) {
889 if (ifp->flags & IFA_F_TENTATIVE) { 889 if (ifp->flags & IFA_F_TENTATIVE) {
890 addrconf_dad_failure(ifp); 890 addrconf_dad_failure(ifp);
891 return; 891 return;
@@ -916,7 +916,7 @@ static void ndisc_recv_na(struct sk_buff *skb)
916 */ 916 */
917 if (lladdr && !memcmp(lladdr, dev->dev_addr, dev->addr_len) && 917 if (lladdr && !memcmp(lladdr, dev->dev_addr, dev->addr_len) &&
918 ipv6_devconf.forwarding && ipv6_devconf.proxy_ndp && 918 ipv6_devconf.forwarding && ipv6_devconf.proxy_ndp &&
919 pneigh_lookup(&nd_tbl, &init_net, &msg->target, dev, 0)) { 919 pneigh_lookup(&nd_tbl, dev->nd_net, &msg->target, dev, 0)) {
920 /* XXX: idev->cnf.prixy_ndp */ 920 /* XXX: idev->cnf.prixy_ndp */
921 goto out; 921 goto out;
922 } 922 }
@@ -1006,6 +1006,7 @@ static void ndisc_ra_useropt(struct sk_buff *ra, struct nd_opt_hdr *opt)
1006 struct sk_buff *skb; 1006 struct sk_buff *skb;
1007 struct nlmsghdr *nlh; 1007 struct nlmsghdr *nlh;
1008 struct nduseroptmsg *ndmsg; 1008 struct nduseroptmsg *ndmsg;
1009 struct net *net = ra->dev->nd_net;
1009 int err; 1010 int err;
1010 int base_size = NLMSG_ALIGN(sizeof(struct nduseroptmsg) 1011 int base_size = NLMSG_ALIGN(sizeof(struct nduseroptmsg)
1011 + (opt->nd_opt_len << 3)); 1012 + (opt->nd_opt_len << 3));
@@ -1035,7 +1036,7 @@ static void ndisc_ra_useropt(struct sk_buff *ra, struct nd_opt_hdr *opt)
1035 &ipv6_hdr(ra)->saddr); 1036 &ipv6_hdr(ra)->saddr);
1036 nlmsg_end(skb, nlh); 1037 nlmsg_end(skb, nlh);
1037 1038
1038 err = rtnl_notify(skb, &init_net, 0, RTNLGRP_ND_USEROPT, NULL, 1039 err = rtnl_notify(skb, net, 0, RTNLGRP_ND_USEROPT, NULL,
1039 GFP_ATOMIC); 1040 GFP_ATOMIC);
1040 if (err < 0) 1041 if (err < 0)
1041 goto errout; 1042 goto errout;
@@ -1046,7 +1047,7 @@ nla_put_failure:
1046 nlmsg_free(skb); 1047 nlmsg_free(skb);
1047 err = -EMSGSIZE; 1048 err = -EMSGSIZE;
1048errout: 1049errout:
1049 rtnl_set_sk_err(&init_net, RTNLGRP_ND_USEROPT, err); 1050 rtnl_set_sk_err(net, RTNLGRP_ND_USEROPT, err);
1050} 1051}
1051 1052
1052static void ndisc_router_discovery(struct sk_buff *skb) 1053static void ndisc_router_discovery(struct sk_buff *skb)
@@ -1391,13 +1392,14 @@ static void ndisc_redirect_rcv(struct sk_buff *skb)
1391void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh, 1392void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
1392 struct in6_addr *target) 1393 struct in6_addr *target)
1393{ 1394{
1394 struct sock *sk = ndisc_socket->sk; 1395 struct net_device *dev = skb->dev;
1396 struct net *net = dev->nd_net;
1397 struct sock *sk = net->ipv6.ndisc_sk;
1395 int len = sizeof(struct icmp6hdr) + 2 * sizeof(struct in6_addr); 1398 int len = sizeof(struct icmp6hdr) + 2 * sizeof(struct in6_addr);
1396 struct sk_buff *buff; 1399 struct sk_buff *buff;
1397 struct icmp6hdr *icmph; 1400 struct icmp6hdr *icmph;
1398 struct in6_addr saddr_buf; 1401 struct in6_addr saddr_buf;
1399 struct in6_addr *addrp; 1402 struct in6_addr *addrp;
1400 struct net_device *dev;
1401 struct rt6_info *rt; 1403 struct rt6_info *rt;
1402 struct dst_entry *dst; 1404 struct dst_entry *dst;
1403 struct inet6_dev *idev; 1405 struct inet6_dev *idev;
@@ -1408,8 +1410,6 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
1408 int hlen; 1410 int hlen;
1409 u8 ha_buf[MAX_ADDR_LEN], *ha = NULL; 1411 u8 ha_buf[MAX_ADDR_LEN], *ha = NULL;
1410 1412
1411 dev = skb->dev;
1412
1413 if (ipv6_get_lladdr(dev, &saddr_buf, IFA_F_TENTATIVE)) { 1413 if (ipv6_get_lladdr(dev, &saddr_buf, IFA_F_TENTATIVE)) {
1414 ND_PRINTK2(KERN_WARNING 1414 ND_PRINTK2(KERN_WARNING
1415 "ICMPv6 Redirect: no link-local address on %s\n", 1415 "ICMPv6 Redirect: no link-local address on %s\n",
@@ -1424,10 +1424,10 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
1424 return; 1424 return;
1425 } 1425 }
1426 1426
1427 icmpv6_flow_init(ndisc_socket->sk, &fl, NDISC_REDIRECT, 1427 icmpv6_flow_init(sk, &fl, NDISC_REDIRECT,
1428 &saddr_buf, &ipv6_hdr(skb)->saddr, dev->ifindex); 1428 &saddr_buf, &ipv6_hdr(skb)->saddr, dev->ifindex);
1429 1429
1430 dst = ip6_route_output(&init_net, NULL, &fl); 1430 dst = ip6_route_output(net, NULL, &fl);
1431 if (dst == NULL) 1431 if (dst == NULL)
1432 return; 1432 return;
1433 1433
@@ -1600,9 +1600,6 @@ static int ndisc_netdev_event(struct notifier_block *this, unsigned long event,
1600 struct net_device *dev = ptr; 1600 struct net_device *dev = ptr;
1601 struct net *net = dev->nd_net; 1601 struct net *net = dev->nd_net;
1602 1602
1603 if (dev->nd_net != &init_net)
1604 return NOTIFY_DONE;
1605
1606 switch (event) { 1603 switch (event) {
1607 case NETDEV_CHANGEADDR: 1604 case NETDEV_CHANGEADDR:
1608 neigh_changeaddr(&nd_tbl, dev); 1605 neigh_changeaddr(&nd_tbl, dev);
@@ -1719,22 +1716,24 @@ static int ndisc_ifinfo_sysctl_strategy(ctl_table *ctl, int __user *name,
1719 1716
1720#endif 1717#endif
1721 1718
1722int __init ndisc_init(void) 1719static int ndisc_net_init(struct net *net)
1723{ 1720{
1721 struct socket *sock;
1724 struct ipv6_pinfo *np; 1722 struct ipv6_pinfo *np;
1725 struct sock *sk; 1723 struct sock *sk;
1726 int err; 1724 int err;
1727 1725
1728 err = sock_create_kern(PF_INET6, SOCK_RAW, IPPROTO_ICMPV6, &ndisc_socket); 1726 err = sock_create_kern(PF_INET6, SOCK_RAW, IPPROTO_ICMPV6, &sock);
1729 if (err < 0) { 1727 if (err < 0) {
1730 ND_PRINTK0(KERN_ERR 1728 ND_PRINTK0(KERN_ERR
1731 "ICMPv6 NDISC: Failed to initialize the control socket (err %d).\n", 1729 "ICMPv6 NDISC: Failed to initialize the control socket (err %d).\n",
1732 err); 1730 err);
1733 ndisc_socket = NULL; /* For safety. */
1734 return err; 1731 return err;
1735 } 1732 }
1736 1733
1737 sk = ndisc_socket->sk; 1734 net->ipv6.ndisc_sk = sk = sock->sk;
1735 sk_change_net(sk, net);
1736
1738 np = inet6_sk(sk); 1737 np = inet6_sk(sk);
1739 sk->sk_allocation = GFP_ATOMIC; 1738 sk->sk_allocation = GFP_ATOMIC;
1740 np->hop_limit = 255; 1739 np->hop_limit = 255;
@@ -1742,21 +1741,52 @@ int __init ndisc_init(void)
1742 np->mc_loop = 0; 1741 np->mc_loop = 0;
1743 sk->sk_prot->unhash(sk); 1742 sk->sk_prot->unhash(sk);
1744 1743
1744 return 0;
1745}
1746
1747static void ndisc_net_exit(struct net *net)
1748{
1749 sk_release_kernel(net->ipv6.ndisc_sk);
1750}
1751
1752static struct pernet_operations ndisc_net_ops = {
1753 .init = ndisc_net_init,
1754 .exit = ndisc_net_exit,
1755};
1756
1757int __init ndisc_init(void)
1758{
1759 int err;
1760
1761 err = register_pernet_subsys(&ndisc_net_ops);
1762 if (err)
1763 return err;
1745 /* 1764 /*
1746 * Initialize the neighbour table 1765 * Initialize the neighbour table
1747 */ 1766 */
1748
1749 neigh_table_init(&nd_tbl); 1767 neigh_table_init(&nd_tbl);
1750 1768
1751#ifdef CONFIG_SYSCTL 1769#ifdef CONFIG_SYSCTL
1752 neigh_sysctl_register(NULL, &nd_tbl.parms, NET_IPV6, NET_IPV6_NEIGH, 1770 err = neigh_sysctl_register(NULL, &nd_tbl.parms, NET_IPV6,
1753 "ipv6", 1771 NET_IPV6_NEIGH, "ipv6",
1754 &ndisc_ifinfo_sysctl_change, 1772 &ndisc_ifinfo_sysctl_change,
1755 &ndisc_ifinfo_sysctl_strategy); 1773 &ndisc_ifinfo_sysctl_strategy);
1774 if (err)
1775 goto out_unregister_pernet;
1756#endif 1776#endif
1777 err = register_netdevice_notifier(&ndisc_netdev_notifier);
1778 if (err)
1779 goto out_unregister_sysctl;
1780out:
1781 return err;
1757 1782
1758 register_netdevice_notifier(&ndisc_netdev_notifier); 1783out_unregister_sysctl:
1759 return 0; 1784#ifdef CONFIG_SYSCTL
1785 neigh_sysctl_unregister(&nd_tbl.parms);
1786out_unregister_pernet:
1787#endif
1788 unregister_pernet_subsys(&ndisc_net_ops);
1789 goto out;
1760} 1790}
1761 1791
1762void ndisc_cleanup(void) 1792void ndisc_cleanup(void)
@@ -1766,6 +1796,5 @@ void ndisc_cleanup(void)
1766 neigh_sysctl_unregister(&nd_tbl.parms); 1796 neigh_sysctl_unregister(&nd_tbl.parms);
1767#endif 1797#endif
1768 neigh_table_clear(&nd_tbl); 1798 neigh_table_clear(&nd_tbl);
1769 sock_release(ndisc_socket); 1799 unregister_pernet_subsys(&ndisc_net_ops);
1770 ndisc_socket = NULL; /* For safety. */
1771} 1800}
diff --git a/net/ipv6/proc.c b/net/ipv6/proc.c
index 2453f2229ef7..8a5be290c710 100644
--- a/net/ipv6/proc.c
+++ b/net/ipv6/proc.c
@@ -39,10 +39,8 @@ static int sockstat6_seq_show(struct seq_file *seq, void *v)
39 sock_prot_inuse_get(&tcpv6_prot)); 39 sock_prot_inuse_get(&tcpv6_prot));
40 seq_printf(seq, "UDP6: inuse %d\n", 40 seq_printf(seq, "UDP6: inuse %d\n",
41 sock_prot_inuse_get(&udpv6_prot)); 41 sock_prot_inuse_get(&udpv6_prot));
42#ifdef CONFIG_IP_UDPLITE
43 seq_printf(seq, "UDPLITE6: inuse %d\n", 42 seq_printf(seq, "UDPLITE6: inuse %d\n",
44 sock_prot_inuse_get(&udplitev6_prot)); 43 sock_prot_inuse_get(&udplitev6_prot));
45#endif
46 seq_printf(seq, "RAW6: inuse %d\n", 44 seq_printf(seq, "RAW6: inuse %d\n",
47 sock_prot_inuse_get(&rawv6_prot)); 45 sock_prot_inuse_get(&rawv6_prot));
48 seq_printf(seq, "FRAG6: inuse %d memory %d\n", 46 seq_printf(seq, "FRAG6: inuse %d memory %d\n",
@@ -113,7 +111,6 @@ static struct snmp_mib snmp6_udp6_list[] = {
113 SNMP_MIB_SENTINEL 111 SNMP_MIB_SENTINEL
114}; 112};
115 113
116#ifdef CONFIG_IP_UDPLITE
117static struct snmp_mib snmp6_udplite6_list[] = { 114static struct snmp_mib snmp6_udplite6_list[] = {
118 SNMP_MIB_ITEM("UdpLite6InDatagrams", UDP_MIB_INDATAGRAMS), 115 SNMP_MIB_ITEM("UdpLite6InDatagrams", UDP_MIB_INDATAGRAMS),
119 SNMP_MIB_ITEM("UdpLite6NoPorts", UDP_MIB_NOPORTS), 116 SNMP_MIB_ITEM("UdpLite6NoPorts", UDP_MIB_NOPORTS),
@@ -121,7 +118,6 @@ static struct snmp_mib snmp6_udplite6_list[] = {
121 SNMP_MIB_ITEM("UdpLite6OutDatagrams", UDP_MIB_OUTDATAGRAMS), 118 SNMP_MIB_ITEM("UdpLite6OutDatagrams", UDP_MIB_OUTDATAGRAMS),
122 SNMP_MIB_SENTINEL 119 SNMP_MIB_SENTINEL
123}; 120};
124#endif
125 121
126static void snmp6_seq_show_icmpv6msg(struct seq_file *seq, void **mib) 122static void snmp6_seq_show_icmpv6msg(struct seq_file *seq, void **mib)
127{ 123{
@@ -180,9 +176,7 @@ static int snmp6_seq_show(struct seq_file *seq, void *v)
180 snmp6_seq_show_item(seq, (void **)icmpv6_statistics, snmp6_icmp6_list); 176 snmp6_seq_show_item(seq, (void **)icmpv6_statistics, snmp6_icmp6_list);
181 snmp6_seq_show_icmpv6msg(seq, (void **)icmpv6msg_statistics); 177 snmp6_seq_show_icmpv6msg(seq, (void **)icmpv6msg_statistics);
182 snmp6_seq_show_item(seq, (void **)udp_stats_in6, snmp6_udp6_list); 178 snmp6_seq_show_item(seq, (void **)udp_stats_in6, snmp6_udp6_list);
183#ifdef CONFIG_IP_UDPLITE
184 snmp6_seq_show_item(seq, (void **)udplite_stats_in6, snmp6_udplite6_list); 179 snmp6_seq_show_item(seq, (void **)udplite_stats_in6, snmp6_udplite6_list);
185#endif
186 } 180 }
187 return 0; 181 return 0;
188} 182}
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 15e9a86f28c8..a4b5aee0f68a 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -2261,7 +2261,7 @@ static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void
2261 skb_reset_mac_header(skb); 2261 skb_reset_mac_header(skb);
2262 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr)); 2262 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
2263 2263
2264 rt = (struct rt6_info*) ip6_route_output(&init_net, NULL, &fl); 2264 rt = (struct rt6_info*) ip6_route_output(net, NULL, &fl);
2265 skb->dst = &rt->u.dst; 2265 skb->dst = &rt->u.dst;
2266 2266
2267 err = rt6_fill_node(skb, rt, &fl.fl6_dst, &fl.fl6_src, iif, 2267 err = rt6_fill_node(skb, rt, &fl.fl6_dst, &fl.fl6_src, iif,
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index b47cce8fea44..caf0cc1c00e1 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -69,9 +69,6 @@
69#include <linux/crypto.h> 69#include <linux/crypto.h>
70#include <linux/scatterlist.h> 70#include <linux/scatterlist.h>
71 71
72/* Socket used for sending RSTs and ACKs */
73static struct socket *tcp6_socket;
74
75static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb); 72static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb);
76static void tcp_v6_reqsk_send_ack(struct sk_buff *skb, struct request_sock *req); 73static void tcp_v6_reqsk_send_ack(struct sk_buff *skb, struct request_sock *req);
77static void tcp_v6_send_check(struct sock *sk, int len, 74static void tcp_v6_send_check(struct sock *sk, int len,
@@ -991,6 +988,8 @@ static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb)
991 struct tcphdr *th = tcp_hdr(skb), *t1; 988 struct tcphdr *th = tcp_hdr(skb), *t1;
992 struct sk_buff *buff; 989 struct sk_buff *buff;
993 struct flowi fl; 990 struct flowi fl;
991 struct net *net = skb->dst->dev->nd_net;
992 struct sock *ctl_sk = net->ipv6.tcp_sk;
994 unsigned int tot_len = sizeof(*th); 993 unsigned int tot_len = sizeof(*th);
995#ifdef CONFIG_TCP_MD5SIG 994#ifdef CONFIG_TCP_MD5SIG
996 struct tcp_md5sig_key *key; 995 struct tcp_md5sig_key *key;
@@ -1075,10 +1074,10 @@ static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb)
1075 * Underlying function will use this to retrieve the network 1074 * Underlying function will use this to retrieve the network
1076 * namespace 1075 * namespace
1077 */ 1076 */
1078 if (!ip6_dst_lookup(tcp6_socket->sk, &buff->dst, &fl)) { 1077 if (!ip6_dst_lookup(ctl_sk, &buff->dst, &fl)) {
1079 1078
1080 if (xfrm_lookup(&buff->dst, &fl, NULL, 0) >= 0) { 1079 if (xfrm_lookup(&buff->dst, &fl, NULL, 0) >= 0) {
1081 ip6_xmit(tcp6_socket->sk, buff, &fl, NULL, 0); 1080 ip6_xmit(ctl_sk, buff, &fl, NULL, 0);
1082 TCP_INC_STATS_BH(TCP_MIB_OUTSEGS); 1081 TCP_INC_STATS_BH(TCP_MIB_OUTSEGS);
1083 TCP_INC_STATS_BH(TCP_MIB_OUTRSTS); 1082 TCP_INC_STATS_BH(TCP_MIB_OUTRSTS);
1084 return; 1083 return;
@@ -1094,6 +1093,8 @@ static void tcp_v6_send_ack(struct tcp_timewait_sock *tw,
1094 struct tcphdr *th = tcp_hdr(skb), *t1; 1093 struct tcphdr *th = tcp_hdr(skb), *t1;
1095 struct sk_buff *buff; 1094 struct sk_buff *buff;
1096 struct flowi fl; 1095 struct flowi fl;
1096 struct net *net = skb->dev->nd_net;
1097 struct sock *ctl_sk = net->ipv6.tcp_sk;
1097 unsigned int tot_len = sizeof(struct tcphdr); 1098 unsigned int tot_len = sizeof(struct tcphdr);
1098 __be32 *topt; 1099 __be32 *topt;
1099#ifdef CONFIG_TCP_MD5SIG 1100#ifdef CONFIG_TCP_MD5SIG
@@ -1175,9 +1176,9 @@ static void tcp_v6_send_ack(struct tcp_timewait_sock *tw,
1175 fl.fl_ip_sport = t1->source; 1176 fl.fl_ip_sport = t1->source;
1176 security_skb_classify_flow(skb, &fl); 1177 security_skb_classify_flow(skb, &fl);
1177 1178
1178 if (!ip6_dst_lookup(tcp6_socket->sk, &buff->dst, &fl)) { 1179 if (!ip6_dst_lookup(ctl_sk, &buff->dst, &fl)) {
1179 if (xfrm_lookup(&buff->dst, &fl, NULL, 0) >= 0) { 1180 if (xfrm_lookup(&buff->dst, &fl, NULL, 0) >= 0) {
1180 ip6_xmit(tcp6_socket->sk, buff, &fl, NULL, 0); 1181 ip6_xmit(ctl_sk, buff, &fl, NULL, 0);
1181 TCP_INC_STATS_BH(TCP_MIB_OUTSEGS); 1182 TCP_INC_STATS_BH(TCP_MIB_OUTSEGS);
1182 return; 1183 return;
1183 } 1184 }
@@ -2198,6 +2199,31 @@ static struct inet_protosw tcpv6_protosw = {
2198 INET_PROTOSW_ICSK, 2199 INET_PROTOSW_ICSK,
2199}; 2200};
2200 2201
2202static int tcpv6_net_init(struct net *net)
2203{
2204 int err;
2205 struct socket *sock;
2206 struct sock *sk;
2207
2208 err = inet_csk_ctl_sock_create(&sock, PF_INET6, SOCK_RAW, IPPROTO_TCP);
2209 if (err)
2210 return err;
2211
2212 net->ipv6.tcp_sk = sk = sock->sk;
2213 sk_change_net(sk, net);
2214 return err;
2215}
2216
2217static void tcpv6_net_exit(struct net *net)
2218{
2219 sk_release_kernel(net->ipv6.tcp_sk);
2220}
2221
2222static struct pernet_operations tcpv6_net_ops = {
2223 .init = tcpv6_net_init,
2224 .exit = tcpv6_net_exit,
2225};
2226
2201int __init tcpv6_init(void) 2227int __init tcpv6_init(void)
2202{ 2228{
2203 int ret; 2229 int ret;
@@ -2211,8 +2237,7 @@ int __init tcpv6_init(void)
2211 if (ret) 2237 if (ret)
2212 goto out_tcpv6_protocol; 2238 goto out_tcpv6_protocol;
2213 2239
2214 ret = inet_csk_ctl_sock_create(&tcp6_socket, PF_INET6, 2240 ret = register_pernet_subsys(&tcpv6_net_ops);
2215 SOCK_RAW, IPPROTO_TCP);
2216 if (ret) 2241 if (ret)
2217 goto out_tcpv6_protosw; 2242 goto out_tcpv6_protosw;
2218out: 2243out:
@@ -2227,7 +2252,7 @@ out_tcpv6_protosw:
2227 2252
2228void tcpv6_exit(void) 2253void tcpv6_exit(void)
2229{ 2254{
2230 sock_release(tcp6_socket); 2255 unregister_pernet_subsys(&tcpv6_net_ops);
2231 inet6_unregister_protosw(&tcpv6_protosw); 2256 inet6_unregister_protosw(&tcpv6_protosw);
2232 inet6_del_protocol(&tcpv6_protocol, IPPROTO_TCP); 2257 inet6_del_protocol(&tcpv6_protocol, IPPROTO_TCP);
2233} 2258}
diff --git a/net/ipv6/udp_ipv6.c b/net/ipv6/udp.c
index 55feac7ba717..d6e311f6c8eb 100644
--- a/net/ipv6/udp_ipv6.c
+++ b/net/ipv6/udp.c
@@ -323,6 +323,9 @@ static struct sock *udp_v6_mcast_next(struct sock *sk,
323 sk_for_each_from(s, node) { 323 sk_for_each_from(s, node) {
324 struct inet_sock *inet = inet_sk(s); 324 struct inet_sock *inet = inet_sk(s);
325 325
326 if (s->sk_net != sk->sk_net)
327 continue;
328
326 if (s->sk_hash == num && s->sk_family == PF_INET6) { 329 if (s->sk_hash == num && s->sk_family == PF_INET6) {
327 struct ipv6_pinfo *np = inet6_sk(s); 330 struct ipv6_pinfo *np = inet6_sk(s);
328 if (inet->dport) { 331 if (inet->dport) {
@@ -400,7 +403,7 @@ static inline int udp6_csum_init(struct sk_buff *skb, struct udphdr *uh,
400 UDP_SKB_CB(skb)->partial_cov = 0; 403 UDP_SKB_CB(skb)->partial_cov = 0;
401 UDP_SKB_CB(skb)->cscov = skb->len; 404 UDP_SKB_CB(skb)->cscov = skb->len;
402 405
403 if (IS_PROTO_UDPLITE(proto)) { 406 if (proto == IPPROTO_UDPLITE) {
404 err = udplite_checksum_init(skb, uh); 407 err = udplite_checksum_init(skb, uh);
405 if (err) 408 if (err)
406 return err; 409 return err;
@@ -489,7 +492,7 @@ int __udp6_lib_rcv(struct sk_buff *skb, struct hlist_head udptable[],
489 492
490 if (udp_lib_checksum_complete(skb)) 493 if (udp_lib_checksum_complete(skb))
491 goto discard; 494 goto discard;
492 UDP6_INC_STATS_BH(UDP_MIB_NOPORTS, IS_PROTO_UDPLITE(proto)); 495 UDP6_INC_STATS_BH(UDP_MIB_NOPORTS, proto == IPPROTO_UDPLITE);
493 496
494 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_PORT_UNREACH, 0, dev); 497 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_PORT_UNREACH, 0, dev);
495 498
@@ -510,11 +513,11 @@ int __udp6_lib_rcv(struct sk_buff *skb, struct hlist_head udptable[],
510 513
511short_packet: 514short_packet:
512 LIMIT_NETDEBUG(KERN_DEBUG "UDP%sv6: short packet: %d/%u\n", 515 LIMIT_NETDEBUG(KERN_DEBUG "UDP%sv6: short packet: %d/%u\n",
513 IS_PROTO_UDPLITE(proto) ? "-Lite" : "", 516 proto == IPPROTO_UDPLITE ? "-Lite" : "",
514 ulen, skb->len); 517 ulen, skb->len);
515 518
516discard: 519discard:
517 UDP6_INC_STATS_BH(UDP_MIB_INERRORS, IS_PROTO_UDPLITE(proto)); 520 UDP6_INC_STATS_BH(UDP_MIB_INERRORS, proto == IPPROTO_UDPLITE);
518 kfree_skb(skb); 521 kfree_skb(skb);
519 return 0; 522 return 0;
520} 523}
@@ -890,7 +893,7 @@ int udpv6_destroy_sock(struct sock *sk)
890int udpv6_setsockopt(struct sock *sk, int level, int optname, 893int udpv6_setsockopt(struct sock *sk, int level, int optname,
891 char __user *optval, int optlen) 894 char __user *optval, int optlen)
892{ 895{
893 if (IS_SOL_UDPFAMILY(level)) 896 if (level == SOL_UDP || level == SOL_UDPLITE)
894 return udp_lib_setsockopt(sk, level, optname, optval, optlen, 897 return udp_lib_setsockopt(sk, level, optname, optval, optlen,
895 udp_v6_push_pending_frames); 898 udp_v6_push_pending_frames);
896 return ipv6_setsockopt(sk, level, optname, optval, optlen); 899 return ipv6_setsockopt(sk, level, optname, optval, optlen);
@@ -900,7 +903,7 @@ int udpv6_setsockopt(struct sock *sk, int level, int optname,
900int compat_udpv6_setsockopt(struct sock *sk, int level, int optname, 903int compat_udpv6_setsockopt(struct sock *sk, int level, int optname,
901 char __user *optval, int optlen) 904 char __user *optval, int optlen)
902{ 905{
903 if (IS_SOL_UDPFAMILY(level)) 906 if (level == SOL_UDP || level == SOL_UDPLITE)
904 return udp_lib_setsockopt(sk, level, optname, optval, optlen, 907 return udp_lib_setsockopt(sk, level, optname, optval, optlen,
905 udp_v6_push_pending_frames); 908 udp_v6_push_pending_frames);
906 return compat_ipv6_setsockopt(sk, level, optname, optval, optlen); 909 return compat_ipv6_setsockopt(sk, level, optname, optval, optlen);
@@ -910,7 +913,7 @@ int compat_udpv6_setsockopt(struct sock *sk, int level, int optname,
910int udpv6_getsockopt(struct sock *sk, int level, int optname, 913int udpv6_getsockopt(struct sock *sk, int level, int optname,
911 char __user *optval, int __user *optlen) 914 char __user *optval, int __user *optlen)
912{ 915{
913 if (IS_SOL_UDPFAMILY(level)) 916 if (level == SOL_UDP || level == SOL_UDPLITE)
914 return udp_lib_getsockopt(sk, level, optname, optval, optlen); 917 return udp_lib_getsockopt(sk, level, optname, optval, optlen);
915 return ipv6_getsockopt(sk, level, optname, optval, optlen); 918 return ipv6_getsockopt(sk, level, optname, optval, optlen);
916} 919}
@@ -919,7 +922,7 @@ int udpv6_getsockopt(struct sock *sk, int level, int optname,
919int compat_udpv6_getsockopt(struct sock *sk, int level, int optname, 922int compat_udpv6_getsockopt(struct sock *sk, int level, int optname,
920 char __user *optval, int __user *optlen) 923 char __user *optval, int __user *optlen)
921{ 924{
922 if (IS_SOL_UDPFAMILY(level)) 925 if (level == SOL_UDP || level == SOL_UDPLITE)
923 return udp_lib_getsockopt(sk, level, optname, optval, optlen); 926 return udp_lib_getsockopt(sk, level, optname, optval, optlen);
924 return compat_ipv6_getsockopt(sk, level, optname, optval, optlen); 927 return compat_ipv6_getsockopt(sk, level, optname, optval, optlen);
925} 928}
diff --git a/net/ipv6/udplite_ipv6.c b/net/ipv6/udplite.c
index 87d4202522ee..87d4202522ee 100644
--- a/net/ipv6/udplite_ipv6.c
+++ b/net/ipv6/udplite.c
diff --git a/net/tipc/core.c b/net/tipc/core.c
index d2d7d32c02c7..740aac5cdfb6 100644
--- a/net/tipc/core.c
+++ b/net/tipc/core.c
@@ -48,16 +48,8 @@
48#include "subscr.h" 48#include "subscr.h"
49#include "config.h" 49#include "config.h"
50 50
51int tipc_eth_media_start(void);
52void tipc_eth_media_stop(void);
53int tipc_handler_start(void);
54void tipc_handler_stop(void);
55int tipc_socket_init(void);
56void tipc_socket_stop(void);
57int tipc_netlink_start(void);
58void tipc_netlink_stop(void);
59 51
60#define TIPC_MOD_VER "1.6.2" 52#define TIPC_MOD_VER "1.6.3"
61 53
62#ifndef CONFIG_TIPC_ZONES 54#ifndef CONFIG_TIPC_ZONES
63#define CONFIG_TIPC_ZONES 3 55#define CONFIG_TIPC_ZONES 3
@@ -277,7 +269,6 @@ EXPORT_SYMBOL(tipc_register_media);
277/* TIPC API for external APIs (see tipc_port.h) */ 269/* TIPC API for external APIs (see tipc_port.h) */
278 270
279EXPORT_SYMBOL(tipc_createport_raw); 271EXPORT_SYMBOL(tipc_createport_raw);
280EXPORT_SYMBOL(tipc_set_msg_option);
281EXPORT_SYMBOL(tipc_reject_msg); 272EXPORT_SYMBOL(tipc_reject_msg);
282EXPORT_SYMBOL(tipc_send_buf_fast); 273EXPORT_SYMBOL(tipc_send_buf_fast);
283EXPORT_SYMBOL(tipc_acknowledge); 274EXPORT_SYMBOL(tipc_acknowledge);
diff --git a/net/tipc/core.h b/net/tipc/core.h
index feabca580820..3fe9b70331d9 100644
--- a/net/tipc/core.h
+++ b/net/tipc/core.h
@@ -180,6 +180,12 @@ extern int tipc_core_start(void);
180extern void tipc_core_stop(void); 180extern void tipc_core_stop(void);
181extern int tipc_core_start_net(void); 181extern int tipc_core_start_net(void);
182extern void tipc_core_stop_net(void); 182extern void tipc_core_stop_net(void);
183extern int tipc_handler_start(void);
184extern void tipc_handler_stop(void);
185extern int tipc_netlink_start(void);
186extern void tipc_netlink_stop(void);
187extern int tipc_socket_init(void);
188extern void tipc_socket_stop(void);
183 189
184static inline int delimit(int val, int min, int max) 190static inline int delimit(int val, int min, int max)
185{ 191{
diff --git a/net/tipc/link.c b/net/tipc/link.c
index cefa99824c58..a42f43430101 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -2832,15 +2832,15 @@ static void link_set_supervision_props(struct link *l_ptr, u32 tolerance)
2832void tipc_link_set_queue_limits(struct link *l_ptr, u32 window) 2832void tipc_link_set_queue_limits(struct link *l_ptr, u32 window)
2833{ 2833{
2834 /* Data messages from this node, inclusive FIRST_FRAGM */ 2834 /* Data messages from this node, inclusive FIRST_FRAGM */
2835 l_ptr->queue_limit[DATA_LOW] = window; 2835 l_ptr->queue_limit[TIPC_LOW_IMPORTANCE] = window;
2836 l_ptr->queue_limit[DATA_MEDIUM] = (window / 3) * 4; 2836 l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE] = (window / 3) * 4;
2837 l_ptr->queue_limit[DATA_HIGH] = (window / 3) * 5; 2837 l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE] = (window / 3) * 5;
2838 l_ptr->queue_limit[DATA_CRITICAL] = (window / 3) * 6; 2838 l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE] = (window / 3) * 6;
2839 /* Transiting data messages,inclusive FIRST_FRAGM */ 2839 /* Transiting data messages,inclusive FIRST_FRAGM */
2840 l_ptr->queue_limit[DATA_LOW + 4] = 300; 2840 l_ptr->queue_limit[TIPC_LOW_IMPORTANCE + 4] = 300;
2841 l_ptr->queue_limit[DATA_MEDIUM + 4] = 600; 2841 l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE + 4] = 600;
2842 l_ptr->queue_limit[DATA_HIGH + 4] = 900; 2842 l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE + 4] = 900;
2843 l_ptr->queue_limit[DATA_CRITICAL + 4] = 1200; 2843 l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE + 4] = 1200;
2844 l_ptr->queue_limit[CONN_MANAGER] = 1200; 2844 l_ptr->queue_limit[CONN_MANAGER] = 1200;
2845 l_ptr->queue_limit[ROUTE_DISTRIBUTOR] = 1200; 2845 l_ptr->queue_limit[ROUTE_DISTRIBUTOR] = 1200;
2846 l_ptr->queue_limit[CHANGEOVER_PROTOCOL] = 2500; 2846 l_ptr->queue_limit[CHANGEOVER_PROTOCOL] = 2500;
diff --git a/net/tipc/msg.c b/net/tipc/msg.c
index 782485468fb2..696a8633df75 100644
--- a/net/tipc/msg.c
+++ b/net/tipc/msg.c
@@ -73,10 +73,10 @@ void tipc_msg_print(struct print_buf *buf, struct tipc_msg *msg, const char *str
73 tipc_printf(buf, "NO(%u/%u):",msg_long_msgno(msg), 73 tipc_printf(buf, "NO(%u/%u):",msg_long_msgno(msg),
74 msg_fragm_no(msg)); 74 msg_fragm_no(msg));
75 break; 75 break;
76 case DATA_LOW: 76 case TIPC_LOW_IMPORTANCE:
77 case DATA_MEDIUM: 77 case TIPC_MEDIUM_IMPORTANCE:
78 case DATA_HIGH: 78 case TIPC_HIGH_IMPORTANCE:
79 case DATA_CRITICAL: 79 case TIPC_CRITICAL_IMPORTANCE:
80 tipc_printf(buf, "DAT%u:", msg_user(msg)); 80 tipc_printf(buf, "DAT%u:", msg_user(msg));
81 if (msg_short(msg)) { 81 if (msg_short(msg)) {
82 tipc_printf(buf, "CON:"); 82 tipc_printf(buf, "CON:");
@@ -229,10 +229,10 @@ void tipc_msg_print(struct print_buf *buf, struct tipc_msg *msg, const char *str
229 switch (usr) { 229 switch (usr) {
230 case CONN_MANAGER: 230 case CONN_MANAGER:
231 case NAME_DISTRIBUTOR: 231 case NAME_DISTRIBUTOR:
232 case DATA_LOW: 232 case TIPC_LOW_IMPORTANCE:
233 case DATA_MEDIUM: 233 case TIPC_MEDIUM_IMPORTANCE:
234 case DATA_HIGH: 234 case TIPC_HIGH_IMPORTANCE:
235 case DATA_CRITICAL: 235 case TIPC_CRITICAL_IMPORTANCE:
236 if (msg_short(msg)) 236 if (msg_short(msg))
237 break; /* No error */ 237 break; /* No error */
238 switch (msg_errcode(msg)) { 238 switch (msg_errcode(msg)) {
diff --git a/net/tipc/msg.h b/net/tipc/msg.h
index e9ef6df26562..6ad070d87702 100644
--- a/net/tipc/msg.h
+++ b/net/tipc/msg.h
@@ -40,18 +40,16 @@
40#include "core.h" 40#include "core.h"
41 41
42#define TIPC_VERSION 2 42#define TIPC_VERSION 2
43#define DATA_LOW TIPC_LOW_IMPORTANCE 43
44#define DATA_MEDIUM TIPC_MEDIUM_IMPORTANCE 44#define SHORT_H_SIZE 24 /* Connected, in-cluster messages */
45#define DATA_HIGH TIPC_HIGH_IMPORTANCE
46#define DATA_CRITICAL TIPC_CRITICAL_IMPORTANCE
47#define SHORT_H_SIZE 24 /* Connected,in cluster */
48#define DIR_MSG_H_SIZE 32 /* Directly addressed messages */ 45#define DIR_MSG_H_SIZE 32 /* Directly addressed messages */
49#define CONN_MSG_H_SIZE 36 /* Routed connected msgs*/ 46#define LONG_H_SIZE 40 /* Named messages */
50#define LONG_H_SIZE 40 /* Named Messages */
51#define MCAST_H_SIZE 44 /* Multicast messages */ 47#define MCAST_H_SIZE 44 /* Multicast messages */
52#define MAX_H_SIZE 60 /* Inclusive full options */ 48#define INT_H_SIZE 40 /* Internal messages */
49#define MIN_H_SIZE 24 /* Smallest legal TIPC header size */
50#define MAX_H_SIZE 60 /* Largest possible TIPC header size */
51
53#define MAX_MSG_SIZE (MAX_H_SIZE + TIPC_MAX_USER_MSG_SIZE) 52#define MAX_MSG_SIZE (MAX_H_SIZE + TIPC_MAX_USER_MSG_SIZE)
54#define LINK_CONFIG 13
55 53
56 54
57/* 55/*
@@ -72,8 +70,10 @@ static inline void msg_set_bits(struct tipc_msg *m, u32 w,
72 u32 pos, u32 mask, u32 val) 70 u32 pos, u32 mask, u32 val)
73{ 71{
74 val = (val & mask) << pos; 72 val = (val & mask) << pos;
75 m->hdr[w] &= ~htonl(mask << pos); 73 val = htonl(val);
76 m->hdr[w] |= htonl(val); 74 mask = htonl(mask << pos);
75 m->hdr[w] &= ~mask;
76 m->hdr[w] |= val;
77} 77}
78 78
79/* 79/*
@@ -87,7 +87,7 @@ static inline u32 msg_version(struct tipc_msg *m)
87 87
88static inline void msg_set_version(struct tipc_msg *m) 88static inline void msg_set_version(struct tipc_msg *m)
89{ 89{
90 msg_set_bits(m, 0, 29, 0xf, TIPC_VERSION); 90 msg_set_bits(m, 0, 29, 7, TIPC_VERSION);
91} 91}
92 92
93static inline u32 msg_user(struct tipc_msg *m) 93static inline u32 msg_user(struct tipc_msg *m)
@@ -97,7 +97,7 @@ static inline u32 msg_user(struct tipc_msg *m)
97 97
98static inline u32 msg_isdata(struct tipc_msg *m) 98static inline u32 msg_isdata(struct tipc_msg *m)
99{ 99{
100 return (msg_user(m) <= DATA_CRITICAL); 100 return (msg_user(m) <= TIPC_CRITICAL_IMPORTANCE);
101} 101}
102 102
103static inline void msg_set_user(struct tipc_msg *m, u32 n) 103static inline void msg_set_user(struct tipc_msg *m, u32 n)
@@ -190,18 +190,6 @@ static inline void msg_set_lookup_scope(struct tipc_msg *m, u32 n)
190 msg_set_bits(m, 1, 19, 0x3, n); 190 msg_set_bits(m, 1, 19, 0x3, n);
191} 191}
192 192
193static inline void msg_set_options(struct tipc_msg *m, const char *opt, u32 sz)
194{
195 u32 hsz = msg_hdr_sz(m);
196 char *to = (char *)&m->hdr[hsz/4];
197
198 if ((hsz < DIR_MSG_H_SIZE) || ((hsz + sz) > MAX_H_SIZE))
199 return;
200 msg_set_bits(m, 1, 16, 0x7, (hsz - 28)/4);
201 msg_set_hdr_sz(m, hsz + sz);
202 memcpy(to, opt, sz);
203}
204
205static inline u32 msg_bcast_ack(struct tipc_msg *m) 193static inline u32 msg_bcast_ack(struct tipc_msg *m)
206{ 194{
207 return msg_bits(m, 1, 0, 0xffff); 195 return msg_bits(m, 1, 0, 0xffff);
@@ -330,17 +318,6 @@ static inline struct tipc_msg *msg_get_wrapped(struct tipc_msg *m)
330 return (struct tipc_msg *)msg_data(m); 318 return (struct tipc_msg *)msg_data(m);
331} 319}
332 320
333static inline void msg_expand(struct tipc_msg *m, u32 destnode)
334{
335 if (!msg_short(m))
336 return;
337 msg_set_hdr_sz(m, LONG_H_SIZE);
338 msg_set_orignode(m, msg_prevnode(m));
339 msg_set_destnode(m, destnode);
340 memset(&m->hdr[8], 0, 12);
341}
342
343
344 321
345/* 322/*
346 TIPC internal message header format, version 2 323 TIPC internal message header format, version 2
@@ -388,7 +365,6 @@ static inline void msg_expand(struct tipc_msg *m, u32 destnode)
388#define NAME_DISTRIBUTOR 11 365#define NAME_DISTRIBUTOR 11
389#define MSG_FRAGMENTER 12 366#define MSG_FRAGMENTER 12
390#define LINK_CONFIG 13 367#define LINK_CONFIG 13
391#define INT_H_SIZE 40
392#define DSC_H_SIZE 40 368#define DSC_H_SIZE 40
393 369
394/* 370/*
diff --git a/net/tipc/port.c b/net/tipc/port.c
index f508614ca59b..e2646a96935d 100644
--- a/net/tipc/port.c
+++ b/net/tipc/port.c
@@ -242,7 +242,8 @@ u32 tipc_createport_raw(void *usr_handle,
242 p_ptr->publ.max_pkt = MAX_PKT_DEFAULT; 242 p_ptr->publ.max_pkt = MAX_PKT_DEFAULT;
243 p_ptr->publ.ref = ref; 243 p_ptr->publ.ref = ref;
244 msg = &p_ptr->publ.phdr; 244 msg = &p_ptr->publ.phdr;
245 msg_init(msg, DATA_LOW, TIPC_NAMED_MSG, TIPC_OK, LONG_H_SIZE, 0); 245 msg_init(msg, TIPC_LOW_IMPORTANCE, TIPC_NAMED_MSG, TIPC_OK, LONG_H_SIZE,
246 0);
246 msg_set_orignode(msg, tipc_own_addr); 247 msg_set_orignode(msg, tipc_own_addr);
247 msg_set_prevnode(msg, tipc_own_addr); 248 msg_set_prevnode(msg, tipc_own_addr);
248 msg_set_origport(msg, ref); 249 msg_set_origport(msg, ref);
@@ -413,13 +414,6 @@ static struct sk_buff *port_build_proto_msg(u32 destport, u32 destnode,
413 return buf; 414 return buf;
414} 415}
415 416
416int tipc_set_msg_option(struct tipc_port *tp_ptr, const char *opt, const u32 sz)
417{
418 msg_expand(&tp_ptr->phdr, msg_destnode(&tp_ptr->phdr));
419 msg_set_options(&tp_ptr->phdr, opt, sz);
420 return TIPC_OK;
421}
422
423int tipc_reject_msg(struct sk_buff *buf, u32 err) 417int tipc_reject_msg(struct sk_buff *buf, u32 err)
424{ 418{
425 struct tipc_msg *msg = buf_msg(buf); 419 struct tipc_msg *msg = buf_msg(buf);
@@ -632,7 +626,7 @@ void tipc_port_recv_proto_msg(struct sk_buff *buf)
632 msg_orignode(msg), 626 msg_orignode(msg),
633 msg_destport(msg), 627 msg_destport(msg),
634 tipc_own_addr, 628 tipc_own_addr,
635 DATA_HIGH, 629 TIPC_HIGH_IMPORTANCE,
636 TIPC_CONN_MSG, 630 TIPC_CONN_MSG,
637 err, 631 err,
638 0, 632 0,
diff --git a/net/tipc/socket.c b/net/tipc/socket.c
index 9ae8e9f74028..3220d5cb5b5d 100644
--- a/net/tipc/socket.c
+++ b/net/tipc/socket.c
@@ -1419,7 +1419,7 @@ exit:
1419/** 1419/**
1420 * shutdown - shutdown socket connection 1420 * shutdown - shutdown socket connection
1421 * @sock: socket structure 1421 * @sock: socket structure
1422 * @how: direction to close (unused; always treated as read + write) 1422 * @how: direction to close (must be SHUT_RDWR)
1423 * 1423 *
1424 * Terminates connection (if necessary), then purges socket's receive queue. 1424 * Terminates connection (if necessary), then purges socket's receive queue.
1425 * 1425 *
@@ -1432,7 +1432,8 @@ static int shutdown(struct socket *sock, int how)
1432 struct sk_buff *buf; 1432 struct sk_buff *buf;
1433 int res; 1433 int res;
1434 1434
1435 /* Could return -EINVAL for an invalid "how", but why bother? */ 1435 if (how != SHUT_RDWR)
1436 return -EINVAL;
1436 1437
1437 if (mutex_lock_interruptible(&tsock->lock)) 1438 if (mutex_lock_interruptible(&tsock->lock))
1438 return -ERESTARTSYS; 1439 return -ERESTARTSYS;
generated by cgit v1.2.2 (git 2.25.0) at 2025-01-27 15:30:57 -0500