Doug / smarc-fsl-linux-kernel

Download as
Browse Code ยป

Commit 3594698a1fb8e5ae60a92c72ce9ca280256939a7

Authored by Eric W. Biederman
Committed by David S. Miller
1 parent b51642f6d7
Exists in smarc-l5.0.0_1.0.0-ga and in 5 other branches imx_3.10.17_1.0.1_ga, imx_3.10.53_1.1.0_ga, imx_3.14.28_1.0.0_ga, smarc-imx_3.10.53_1.1.0_ga, smarc-imx_3.14.28_1.0.0_ga

net: Make CAP_NET_BIND_SERVICE per user namespace 

Allow privileged users in any user namespace to bind to
privileged sockets in network namespaces they control.

Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Showing 3 changed files with 10 additions and 6 deletions Inline Diff

1 /* 1 /*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX 2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket 3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level. 4 * interface as the means of communication with the user level.
5 * 5 *
6 * PF_INET protocol family socket handler. 6 * PF_INET protocol family socket handler.
7 * 7 *
8 * Authors: Ross Biro 8 * Authors: Ross Biro
9 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> 9 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
10 * Florian La Roche, <flla@stud.uni-sb.de> 10 * Florian La Roche, <flla@stud.uni-sb.de>
11 * Alan Cox, <A.Cox@swansea.ac.uk> 11 * Alan Cox, <A.Cox@swansea.ac.uk>
12 * 12 *
13 * Changes (see also sock.c) 13 * Changes (see also sock.c)
14 * 14 *
15 * piggy, 15 * piggy,
16 * Karl Knutson : Socket protocol table 16 * Karl Knutson : Socket protocol table
17 * A.N.Kuznetsov : Socket death error in accept(). 17 * A.N.Kuznetsov : Socket death error in accept().
18 * John Richardson : Fix non blocking error in connect() 18 * John Richardson : Fix non blocking error in connect()
19 * so sockets that fail to connect 19 * so sockets that fail to connect
20 * don't return -EINPROGRESS. 20 * don't return -EINPROGRESS.
21 * Alan Cox : Asynchronous I/O support 21 * Alan Cox : Asynchronous I/O support
22 * Alan Cox : Keep correct socket pointer on sock 22 * Alan Cox : Keep correct socket pointer on sock
23 * structures 23 * structures
24 * when accept() ed 24 * when accept() ed
25 * Alan Cox : Semantics of SO_LINGER aren't state 25 * Alan Cox : Semantics of SO_LINGER aren't state
26 * moved to close when you look carefully. 26 * moved to close when you look carefully.
27 * With this fixed and the accept bug fixed 27 * With this fixed and the accept bug fixed
28 * some RPC stuff seems happier. 28 * some RPC stuff seems happier.
29 * Niibe Yutaka : 4.4BSD style write async I/O 29 * Niibe Yutaka : 4.4BSD style write async I/O
30 * Alan Cox, 30 * Alan Cox,
31 * Tony Gale : Fixed reuse semantics. 31 * Tony Gale : Fixed reuse semantics.
32 * Alan Cox : bind() shouldn't abort existing but dead 32 * Alan Cox : bind() shouldn't abort existing but dead
33 * sockets. Stops FTP netin:.. I hope. 33 * sockets. Stops FTP netin:.. I hope.
34 * Alan Cox : bind() works correctly for RAW sockets. 34 * Alan Cox : bind() works correctly for RAW sockets.
35 * Note that FreeBSD at least was broken 35 * Note that FreeBSD at least was broken
36 * in this respect so be careful with 36 * in this respect so be careful with
37 * compatibility tests... 37 * compatibility tests...
38 * Alan Cox : routing cache support 38 * Alan Cox : routing cache support
39 * Alan Cox : memzero the socket structure for 39 * Alan Cox : memzero the socket structure for
40 * compactness. 40 * compactness.
41 * Matt Day : nonblock connect error handler 41 * Matt Day : nonblock connect error handler
42 * Alan Cox : Allow large numbers of pending sockets 42 * Alan Cox : Allow large numbers of pending sockets
43 * (eg for big web sites), but only if 43 * (eg for big web sites), but only if
44 * specifically application requested. 44 * specifically application requested.
45 * Alan Cox : New buffering throughout IP. Used 45 * Alan Cox : New buffering throughout IP. Used
46 * dumbly. 46 * dumbly.
47 * Alan Cox : New buffering now used smartly. 47 * Alan Cox : New buffering now used smartly.
48 * Alan Cox : BSD rather than common sense 48 * Alan Cox : BSD rather than common sense
49 * interpretation of listen. 49 * interpretation of listen.
50 * Germano Caronni : Assorted small races. 50 * Germano Caronni : Assorted small races.
51 * Alan Cox : sendmsg/recvmsg basic support. 51 * Alan Cox : sendmsg/recvmsg basic support.
52 * Alan Cox : Only sendmsg/recvmsg now supported. 52 * Alan Cox : Only sendmsg/recvmsg now supported.
53 * Alan Cox : Locked down bind (see security list). 53 * Alan Cox : Locked down bind (see security list).
54 * Alan Cox : Loosened bind a little. 54 * Alan Cox : Loosened bind a little.
55 * Mike McLagan : ADD/DEL DLCI Ioctls 55 * Mike McLagan : ADD/DEL DLCI Ioctls
56 * Willy Konynenberg : Transparent proxying support. 56 * Willy Konynenberg : Transparent proxying support.
57 * David S. Miller : New socket lookup architecture. 57 * David S. Miller : New socket lookup architecture.
58 * Some other random speedups. 58 * Some other random speedups.
59 * Cyrus Durgin : Cleaned up file for kmod hacks. 59 * Cyrus Durgin : Cleaned up file for kmod hacks.
60 * Andi Kleen : Fix inet_stream_connect TCP race. 60 * Andi Kleen : Fix inet_stream_connect TCP race.
61 * 61 *
62 * This program is free software; you can redistribute it and/or 62 * This program is free software; you can redistribute it and/or
63 * modify it under the terms of the GNU General Public License 63 * modify it under the terms of the GNU General Public License
64 * as published by the Free Software Foundation; either version 64 * as published by the Free Software Foundation; either version
65 * 2 of the License, or (at your option) any later version. 65 * 2 of the License, or (at your option) any later version.
66 */ 66 */
67 67
68 #define pr_fmt(fmt) "IPv4: " fmt 68 #define pr_fmt(fmt) "IPv4: " fmt
69 69
70 #include <linux/err.h> 70 #include <linux/err.h>
71 #include <linux/errno.h> 71 #include <linux/errno.h>
72 #include <linux/types.h> 72 #include <linux/types.h>
73 #include <linux/socket.h> 73 #include <linux/socket.h>
74 #include <linux/in.h> 74 #include <linux/in.h>
75 #include <linux/kernel.h> 75 #include <linux/kernel.h>
76 #include <linux/module.h> 76 #include <linux/module.h>
77 #include <linux/sched.h> 77 #include <linux/sched.h>
78 #include <linux/timer.h> 78 #include <linux/timer.h>
79 #include <linux/string.h> 79 #include <linux/string.h>
80 #include <linux/sockios.h> 80 #include <linux/sockios.h>
81 #include <linux/net.h> 81 #include <linux/net.h>
82 #include <linux/capability.h> 82 #include <linux/capability.h>
83 #include <linux/fcntl.h> 83 #include <linux/fcntl.h>
84 #include <linux/mm.h> 84 #include <linux/mm.h>
85 #include <linux/interrupt.h> 85 #include <linux/interrupt.h>
86 #include <linux/stat.h> 86 #include <linux/stat.h>
87 #include <linux/init.h> 87 #include <linux/init.h>
88 #include <linux/poll.h> 88 #include <linux/poll.h>
89 #include <linux/netfilter_ipv4.h> 89 #include <linux/netfilter_ipv4.h>
90 #include <linux/random.h> 90 #include <linux/random.h>
91 #include <linux/slab.h> 91 #include <linux/slab.h>
92 92
93 #include <asm/uaccess.h> 93 #include <asm/uaccess.h>
94 94
95 #include <linux/inet.h> 95 #include <linux/inet.h>
96 #include <linux/igmp.h> 96 #include <linux/igmp.h>
97 #include <linux/inetdevice.h> 97 #include <linux/inetdevice.h>
98 #include <linux/netdevice.h> 98 #include <linux/netdevice.h>
99 #include <net/checksum.h> 99 #include <net/checksum.h>
100 #include <net/ip.h> 100 #include <net/ip.h>
101 #include <net/protocol.h> 101 #include <net/protocol.h>
102 #include <net/arp.h> 102 #include <net/arp.h>
103 #include <net/route.h> 103 #include <net/route.h>
104 #include <net/ip_fib.h> 104 #include <net/ip_fib.h>
105 #include <net/inet_connection_sock.h> 105 #include <net/inet_connection_sock.h>
106 #include <net/tcp.h> 106 #include <net/tcp.h>
107 #include <net/udp.h> 107 #include <net/udp.h>
108 #include <net/udplite.h> 108 #include <net/udplite.h>
109 #include <net/ping.h> 109 #include <net/ping.h>
110 #include <linux/skbuff.h> 110 #include <linux/skbuff.h>
111 #include <net/sock.h> 111 #include <net/sock.h>
112 #include <net/raw.h> 112 #include <net/raw.h>
113 #include <net/icmp.h> 113 #include <net/icmp.h>
114 #include <net/ipip.h> 114 #include <net/ipip.h>
115 #include <net/inet_common.h> 115 #include <net/inet_common.h>
116 #include <net/xfrm.h> 116 #include <net/xfrm.h>
117 #include <net/net_namespace.h> 117 #include <net/net_namespace.h>
118 #ifdef CONFIG_IP_MROUTE 118 #ifdef CONFIG_IP_MROUTE
119 #include <linux/mroute.h> 119 #include <linux/mroute.h>
120 #endif 120 #endif
121 121
122 122
123 /* The inetsw table contains everything that inet_create needs to 123 /* The inetsw table contains everything that inet_create needs to
124 * build a new socket. 124 * build a new socket.
125 */ 125 */
126 static struct list_head inetsw[SOCK_MAX]; 126 static struct list_head inetsw[SOCK_MAX];
127 static DEFINE_SPINLOCK(inetsw_lock); 127 static DEFINE_SPINLOCK(inetsw_lock);
128 128
129 struct ipv4_config ipv4_config; 129 struct ipv4_config ipv4_config;
130 EXPORT_SYMBOL(ipv4_config); 130 EXPORT_SYMBOL(ipv4_config);
131 131
132 /* New destruction routine */ 132 /* New destruction routine */
133 133
134 void inet_sock_destruct(struct sock *sk) 134 void inet_sock_destruct(struct sock *sk)
135 { 135 {
136 struct inet_sock *inet = inet_sk(sk); 136 struct inet_sock *inet = inet_sk(sk);
137 137
138 __skb_queue_purge(&sk->sk_receive_queue); 138 __skb_queue_purge(&sk->sk_receive_queue);
139 __skb_queue_purge(&sk->sk_error_queue); 139 __skb_queue_purge(&sk->sk_error_queue);
140 140
141 sk_mem_reclaim(sk); 141 sk_mem_reclaim(sk);
142 142
143 if (sk->sk_type == SOCK_STREAM && sk->sk_state != TCP_CLOSE) { 143 if (sk->sk_type == SOCK_STREAM && sk->sk_state != TCP_CLOSE) {
144 pr_err("Attempt to release TCP socket in state %d %p\n", 144 pr_err("Attempt to release TCP socket in state %d %p\n",
145 sk->sk_state, sk); 145 sk->sk_state, sk);
146 return; 146 return;
147 } 147 }
148 if (!sock_flag(sk, SOCK_DEAD)) { 148 if (!sock_flag(sk, SOCK_DEAD)) {
149 pr_err("Attempt to release alive inet socket %p\n", sk); 149 pr_err("Attempt to release alive inet socket %p\n", sk);
150 return; 150 return;
151 } 151 }
152 152
153 WARN_ON(atomic_read(&sk->sk_rmem_alloc)); 153 WARN_ON(atomic_read(&sk->sk_rmem_alloc));
154 WARN_ON(atomic_read(&sk->sk_wmem_alloc)); 154 WARN_ON(atomic_read(&sk->sk_wmem_alloc));
155 WARN_ON(sk->sk_wmem_queued); 155 WARN_ON(sk->sk_wmem_queued);
156 WARN_ON(sk->sk_forward_alloc); 156 WARN_ON(sk->sk_forward_alloc);
157 157
158 kfree(rcu_dereference_protected(inet->inet_opt, 1)); 158 kfree(rcu_dereference_protected(inet->inet_opt, 1));
159 dst_release(rcu_dereference_check(sk->sk_dst_cache, 1)); 159 dst_release(rcu_dereference_check(sk->sk_dst_cache, 1));
160 dst_release(sk->sk_rx_dst); 160 dst_release(sk->sk_rx_dst);
161 sk_refcnt_debug_dec(sk); 161 sk_refcnt_debug_dec(sk);
162 } 162 }
163 EXPORT_SYMBOL(inet_sock_destruct); 163 EXPORT_SYMBOL(inet_sock_destruct);
164 164
165 /* 165 /*
166 * The routines beyond this point handle the behaviour of an AF_INET 166 * The routines beyond this point handle the behaviour of an AF_INET
167 * socket object. Mostly it punts to the subprotocols of IP to do 167 * socket object. Mostly it punts to the subprotocols of IP to do
168 * the work. 168 * the work.
169 */ 169 */
170 170
171 /* 171 /*
172 * Automatically bind an unbound socket. 172 * Automatically bind an unbound socket.
173 */ 173 */
174 174
175 static int inet_autobind(struct sock *sk) 175 static int inet_autobind(struct sock *sk)
176 { 176 {
177 struct inet_sock *inet; 177 struct inet_sock *inet;
178 /* We may need to bind the socket. */ 178 /* We may need to bind the socket. */
179 lock_sock(sk); 179 lock_sock(sk);
180 inet = inet_sk(sk); 180 inet = inet_sk(sk);
181 if (!inet->inet_num) { 181 if (!inet->inet_num) {
182 if (sk->sk_prot->get_port(sk, 0)) { 182 if (sk->sk_prot->get_port(sk, 0)) {
183 release_sock(sk); 183 release_sock(sk);
184 return -EAGAIN; 184 return -EAGAIN;
185 } 185 }
186 inet->inet_sport = htons(inet->inet_num); 186 inet->inet_sport = htons(inet->inet_num);
187 } 187 }
188 release_sock(sk); 188 release_sock(sk);
189 return 0; 189 return 0;
190 } 190 }
191 191
192 /* 192 /*
193 * Move a socket into listening state. 193 * Move a socket into listening state.
194 */ 194 */
195 int inet_listen(struct socket *sock, int backlog) 195 int inet_listen(struct socket *sock, int backlog)
196 { 196 {
197 struct sock *sk = sock->sk; 197 struct sock *sk = sock->sk;
198 unsigned char old_state; 198 unsigned char old_state;
199 int err; 199 int err;
200 200
201 lock_sock(sk); 201 lock_sock(sk);
202 202
203 err = -EINVAL; 203 err = -EINVAL;
204 if (sock->state != SS_UNCONNECTED || sock->type != SOCK_STREAM) 204 if (sock->state != SS_UNCONNECTED || sock->type != SOCK_STREAM)
205 goto out; 205 goto out;
206 206
207 old_state = sk->sk_state; 207 old_state = sk->sk_state;
208 if (!((1 << old_state) & (TCPF_CLOSE | TCPF_LISTEN))) 208 if (!((1 << old_state) & (TCPF_CLOSE | TCPF_LISTEN)))
209 goto out; 209 goto out;
210 210
211 /* Really, if the socket is already in listen state 211 /* Really, if the socket is already in listen state
212 * we can only allow the backlog to be adjusted. 212 * we can only allow the backlog to be adjusted.
213 */ 213 */
214 if (old_state != TCP_LISTEN) { 214 if (old_state != TCP_LISTEN) {
215 /* Check special setups for testing purpose to enable TFO w/o 215 /* Check special setups for testing purpose to enable TFO w/o
216 * requiring TCP_FASTOPEN sockopt. 216 * requiring TCP_FASTOPEN sockopt.
217 * Note that only TCP sockets (SOCK_STREAM) will reach here. 217 * Note that only TCP sockets (SOCK_STREAM) will reach here.
218 * Also fastopenq may already been allocated because this 218 * Also fastopenq may already been allocated because this
219 * socket was in TCP_LISTEN state previously but was 219 * socket was in TCP_LISTEN state previously but was
220 * shutdown() (rather than close()). 220 * shutdown() (rather than close()).
221 */ 221 */
222 if ((sysctl_tcp_fastopen & TFO_SERVER_ENABLE) != 0 && 222 if ((sysctl_tcp_fastopen & TFO_SERVER_ENABLE) != 0 &&
223 inet_csk(sk)->icsk_accept_queue.fastopenq == NULL) { 223 inet_csk(sk)->icsk_accept_queue.fastopenq == NULL) {
224 if ((sysctl_tcp_fastopen & TFO_SERVER_WO_SOCKOPT1) != 0) 224 if ((sysctl_tcp_fastopen & TFO_SERVER_WO_SOCKOPT1) != 0)
225 err = fastopen_init_queue(sk, backlog); 225 err = fastopen_init_queue(sk, backlog);
226 else if ((sysctl_tcp_fastopen & 226 else if ((sysctl_tcp_fastopen &
227 TFO_SERVER_WO_SOCKOPT2) != 0) 227 TFO_SERVER_WO_SOCKOPT2) != 0)
228 err = fastopen_init_queue(sk, 228 err = fastopen_init_queue(sk,
229 ((uint)sysctl_tcp_fastopen) >> 16); 229 ((uint)sysctl_tcp_fastopen) >> 16);
230 else 230 else
231 err = 0; 231 err = 0;
232 if (err) 232 if (err)
233 goto out; 233 goto out;
234 } 234 }
235 err = inet_csk_listen_start(sk, backlog); 235 err = inet_csk_listen_start(sk, backlog);
236 if (err) 236 if (err)
237 goto out; 237 goto out;
238 } 238 }
239 sk->sk_max_ack_backlog = backlog; 239 sk->sk_max_ack_backlog = backlog;
240 err = 0; 240 err = 0;
241 241
242 out: 242 out:
243 release_sock(sk); 243 release_sock(sk);
244 return err; 244 return err;
245 } 245 }
246 EXPORT_SYMBOL(inet_listen); 246 EXPORT_SYMBOL(inet_listen);
247 247
248 u32 inet_ehash_secret __read_mostly; 248 u32 inet_ehash_secret __read_mostly;
249 EXPORT_SYMBOL(inet_ehash_secret); 249 EXPORT_SYMBOL(inet_ehash_secret);
250 250
251 /* 251 /*
252 * inet_ehash_secret must be set exactly once 252 * inet_ehash_secret must be set exactly once
253 */ 253 */
254 void build_ehash_secret(void) 254 void build_ehash_secret(void)
255 { 255 {
256 u32 rnd; 256 u32 rnd;
257 257
258 do { 258 do {
259 get_random_bytes(&rnd, sizeof(rnd)); 259 get_random_bytes(&rnd, sizeof(rnd));
260 } while (rnd == 0); 260 } while (rnd == 0);
261 261
262 cmpxchg(&inet_ehash_secret, 0, rnd); 262 cmpxchg(&inet_ehash_secret, 0, rnd);
263 } 263 }
264 EXPORT_SYMBOL(build_ehash_secret); 264 EXPORT_SYMBOL(build_ehash_secret);
265 265
266 static inline int inet_netns_ok(struct net *net, __u8 protocol) 266 static inline int inet_netns_ok(struct net *net, __u8 protocol)
267 { 267 {
268 const struct net_protocol *ipprot; 268 const struct net_protocol *ipprot;
269 269
270 if (net_eq(net, &init_net)) 270 if (net_eq(net, &init_net))
271 return 1; 271 return 1;
272 272
273 ipprot = rcu_dereference(inet_protos[protocol]); 273 ipprot = rcu_dereference(inet_protos[protocol]);
274 if (ipprot == NULL) { 274 if (ipprot == NULL) {
275 /* raw IP is OK */ 275 /* raw IP is OK */
276 return 1; 276 return 1;
277 } 277 }
278 return ipprot->netns_ok; 278 return ipprot->netns_ok;
279 } 279 }
280 280
281 /* 281 /*
282 * Create an inet socket. 282 * Create an inet socket.
283 */ 283 */
284 284
285 static int inet_create(struct net *net, struct socket *sock, int protocol, 285 static int inet_create(struct net *net, struct socket *sock, int protocol,
286 int kern) 286 int kern)
287 { 287 {
288 struct sock *sk; 288 struct sock *sk;
289 struct inet_protosw *answer; 289 struct inet_protosw *answer;
290 struct inet_sock *inet; 290 struct inet_sock *inet;
291 struct proto *answer_prot; 291 struct proto *answer_prot;
292 unsigned char answer_flags; 292 unsigned char answer_flags;
293 char answer_no_check; 293 char answer_no_check;
294 int try_loading_module = 0; 294 int try_loading_module = 0;
295 int err; 295 int err;
296 296
297 if (unlikely(!inet_ehash_secret)) 297 if (unlikely(!inet_ehash_secret))
298 if (sock->type != SOCK_RAW && sock->type != SOCK_DGRAM) 298 if (sock->type != SOCK_RAW && sock->type != SOCK_DGRAM)
299 build_ehash_secret(); 299 build_ehash_secret();
300 300
301 sock->state = SS_UNCONNECTED; 301 sock->state = SS_UNCONNECTED;
302 302
303 /* Look for the requested type/protocol pair. */ 303 /* Look for the requested type/protocol pair. */
304 lookup_protocol: 304 lookup_protocol:
305 err = -ESOCKTNOSUPPORT; 305 err = -ESOCKTNOSUPPORT;
306 rcu_read_lock(); 306 rcu_read_lock();
307 list_for_each_entry_rcu(answer, &inetsw[sock->type], list) { 307 list_for_each_entry_rcu(answer, &inetsw[sock->type], list) {
308 308
309 err = 0; 309 err = 0;
310 /* Check the non-wild match. */ 310 /* Check the non-wild match. */
311 if (protocol == answer->protocol) { 311 if (protocol == answer->protocol) {
312 if (protocol != IPPROTO_IP) 312 if (protocol != IPPROTO_IP)
313 break; 313 break;
314 } else { 314 } else {
315 /* Check for the two wild cases. */ 315 /* Check for the two wild cases. */
316 if (IPPROTO_IP == protocol) { 316 if (IPPROTO_IP == protocol) {
317 protocol = answer->protocol; 317 protocol = answer->protocol;
318 break; 318 break;
319 } 319 }
320 if (IPPROTO_IP == answer->protocol) 320 if (IPPROTO_IP == answer->protocol)
321 break; 321 break;
322 } 322 }
323 err = -EPROTONOSUPPORT; 323 err = -EPROTONOSUPPORT;
324 } 324 }
325 325
326 if (unlikely(err)) { 326 if (unlikely(err)) {
327 if (try_loading_module < 2) { 327 if (try_loading_module < 2) {
328 rcu_read_unlock(); 328 rcu_read_unlock();
329 /* 329 /*
330 * Be more specific, e.g. net-pf-2-proto-132-type-1 330 * Be more specific, e.g. net-pf-2-proto-132-type-1
331 * (net-pf-PF_INET-proto-IPPROTO_SCTP-type-SOCK_STREAM) 331 * (net-pf-PF_INET-proto-IPPROTO_SCTP-type-SOCK_STREAM)
332 */ 332 */
333 if (++try_loading_module == 1) 333 if (++try_loading_module == 1)
334 request_module("net-pf-%d-proto-%d-type-%d", 334 request_module("net-pf-%d-proto-%d-type-%d",
335 PF_INET, protocol, sock->type); 335 PF_INET, protocol, sock->type);
336 /* 336 /*
337 * Fall back to generic, e.g. net-pf-2-proto-132 337 * Fall back to generic, e.g. net-pf-2-proto-132
338 * (net-pf-PF_INET-proto-IPPROTO_SCTP) 338 * (net-pf-PF_INET-proto-IPPROTO_SCTP)
339 */ 339 */
340 else 340 else
341 request_module("net-pf-%d-proto-%d", 341 request_module("net-pf-%d-proto-%d",
342 PF_INET, protocol); 342 PF_INET, protocol);
343 goto lookup_protocol; 343 goto lookup_protocol;
344 } else 344 } else
345 goto out_rcu_unlock; 345 goto out_rcu_unlock;
346 } 346 }
347 347
348 err = -EPERM; 348 err = -EPERM;
349 if (sock->type == SOCK_RAW && !kern && 349 if (sock->type == SOCK_RAW && !kern &&
350 !ns_capable(net->user_ns, CAP_NET_RAW)) 350 !ns_capable(net->user_ns, CAP_NET_RAW))
351 goto out_rcu_unlock; 351 goto out_rcu_unlock;
352 352
353 err = -EAFNOSUPPORT; 353 err = -EAFNOSUPPORT;
354 if (!inet_netns_ok(net, protocol)) 354 if (!inet_netns_ok(net, protocol))
355 goto out_rcu_unlock; 355 goto out_rcu_unlock;
356 356
357 sock->ops = answer->ops; 357 sock->ops = answer->ops;
358 answer_prot = answer->prot; 358 answer_prot = answer->prot;
359 answer_no_check = answer->no_check; 359 answer_no_check = answer->no_check;
360 answer_flags = answer->flags; 360 answer_flags = answer->flags;
361 rcu_read_unlock(); 361 rcu_read_unlock();
362 362
363 WARN_ON(answer_prot->slab == NULL); 363 WARN_ON(answer_prot->slab == NULL);
364 364
365 err = -ENOBUFS; 365 err = -ENOBUFS;
366 sk = sk_alloc(net, PF_INET, GFP_KERNEL, answer_prot); 366 sk = sk_alloc(net, PF_INET, GFP_KERNEL, answer_prot);
367 if (sk == NULL) 367 if (sk == NULL)
368 goto out; 368 goto out;
369 369
370 err = 0; 370 err = 0;
371 sk->sk_no_check = answer_no_check; 371 sk->sk_no_check = answer_no_check;
372 if (INET_PROTOSW_REUSE & answer_flags) 372 if (INET_PROTOSW_REUSE & answer_flags)
373 sk->sk_reuse = SK_CAN_REUSE; 373 sk->sk_reuse = SK_CAN_REUSE;
374 374
375 inet = inet_sk(sk); 375 inet = inet_sk(sk);
376 inet->is_icsk = (INET_PROTOSW_ICSK & answer_flags) != 0; 376 inet->is_icsk = (INET_PROTOSW_ICSK & answer_flags) != 0;
377 377
378 inet->nodefrag = 0; 378 inet->nodefrag = 0;
379 379
380 if (SOCK_RAW == sock->type) { 380 if (SOCK_RAW == sock->type) {
381 inet->inet_num = protocol; 381 inet->inet_num = protocol;
382 if (IPPROTO_RAW == protocol) 382 if (IPPROTO_RAW == protocol)
383 inet->hdrincl = 1; 383 inet->hdrincl = 1;
384 } 384 }
385 385
386 if (ipv4_config.no_pmtu_disc) 386 if (ipv4_config.no_pmtu_disc)
387 inet->pmtudisc = IP_PMTUDISC_DONT; 387 inet->pmtudisc = IP_PMTUDISC_DONT;
388 else 388 else
389 inet->pmtudisc = IP_PMTUDISC_WANT; 389 inet->pmtudisc = IP_PMTUDISC_WANT;
390 390
391 inet->inet_id = 0; 391 inet->inet_id = 0;
392 392
393 sock_init_data(sock, sk); 393 sock_init_data(sock, sk);
394 394
395 sk->sk_destruct = inet_sock_destruct; 395 sk->sk_destruct = inet_sock_destruct;
396 sk->sk_protocol = protocol; 396 sk->sk_protocol = protocol;
397 sk->sk_backlog_rcv = sk->sk_prot->backlog_rcv; 397 sk->sk_backlog_rcv = sk->sk_prot->backlog_rcv;
398 398
399 inet->uc_ttl = -1; 399 inet->uc_ttl = -1;
400 inet->mc_loop = 1; 400 inet->mc_loop = 1;
401 inet->mc_ttl = 1; 401 inet->mc_ttl = 1;
402 inet->mc_all = 1; 402 inet->mc_all = 1;
403 inet->mc_index = 0; 403 inet->mc_index = 0;
404 inet->mc_list = NULL; 404 inet->mc_list = NULL;
405 inet->rcv_tos = 0; 405 inet->rcv_tos = 0;
406 406
407 sk_refcnt_debug_inc(sk); 407 sk_refcnt_debug_inc(sk);
408 408
409 if (inet->inet_num) { 409 if (inet->inet_num) {
410 /* It assumes that any protocol which allows 410 /* It assumes that any protocol which allows
411 * the user to assign a number at socket 411 * the user to assign a number at socket
412 * creation time automatically 412 * creation time automatically
413 * shares. 413 * shares.
414 */ 414 */
415 inet->inet_sport = htons(inet->inet_num); 415 inet->inet_sport = htons(inet->inet_num);
416 /* Add to protocol hash chains. */ 416 /* Add to protocol hash chains. */
417 sk->sk_prot->hash(sk); 417 sk->sk_prot->hash(sk);
418 } 418 }
419 419
420 if (sk->sk_prot->init) { 420 if (sk->sk_prot->init) {
421 err = sk->sk_prot->init(sk); 421 err = sk->sk_prot->init(sk);
422 if (err) 422 if (err)
423 sk_common_release(sk); 423 sk_common_release(sk);
424 } 424 }
425 out: 425 out:
426 return err; 426 return err;
427 out_rcu_unlock: 427 out_rcu_unlock:
428 rcu_read_unlock(); 428 rcu_read_unlock();
429 goto out; 429 goto out;
430 } 430 }
431 431
432 432
433 /* 433 /*
434 * The peer socket should always be NULL (or else). When we call this 434 * The peer socket should always be NULL (or else). When we call this
435 * function we are destroying the object and from then on nobody 435 * function we are destroying the object and from then on nobody
436 * should refer to it. 436 * should refer to it.
437 */ 437 */
438 int inet_release(struct socket *sock) 438 int inet_release(struct socket *sock)
439 { 439 {
440 struct sock *sk = sock->sk; 440 struct sock *sk = sock->sk;
441 441
442 if (sk) { 442 if (sk) {
443 long timeout; 443 long timeout;
444 444
445 sock_rps_reset_flow(sk); 445 sock_rps_reset_flow(sk);
446 446
447 /* Applications forget to leave groups before exiting */ 447 /* Applications forget to leave groups before exiting */
448 ip_mc_drop_socket(sk); 448 ip_mc_drop_socket(sk);
449 449
450 /* If linger is set, we don't return until the close 450 /* If linger is set, we don't return until the close
451 * is complete. Otherwise we return immediately. The 451 * is complete. Otherwise we return immediately. The
452 * actually closing is done the same either way. 452 * actually closing is done the same either way.
453 * 453 *
454 * If the close is due to the process exiting, we never 454 * If the close is due to the process exiting, we never
455 * linger.. 455 * linger..
456 */ 456 */
457 timeout = 0; 457 timeout = 0;
458 if (sock_flag(sk, SOCK_LINGER) && 458 if (sock_flag(sk, SOCK_LINGER) &&
459 !(current->flags & PF_EXITING)) 459 !(current->flags & PF_EXITING))
460 timeout = sk->sk_lingertime; 460 timeout = sk->sk_lingertime;
461 sock->sk = NULL; 461 sock->sk = NULL;
462 sk->sk_prot->close(sk, timeout); 462 sk->sk_prot->close(sk, timeout);
463 } 463 }
464 return 0; 464 return 0;
465 } 465 }
466 EXPORT_SYMBOL(inet_release); 466 EXPORT_SYMBOL(inet_release);
467 467
468 /* It is off by default, see below. */ 468 /* It is off by default, see below. */
469 int sysctl_ip_nonlocal_bind __read_mostly; 469 int sysctl_ip_nonlocal_bind __read_mostly;
470 EXPORT_SYMBOL(sysctl_ip_nonlocal_bind); 470 EXPORT_SYMBOL(sysctl_ip_nonlocal_bind);
471 471
472 int inet_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) 472 int inet_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
473 { 473 {
474 struct sockaddr_in *addr = (struct sockaddr_in *)uaddr; 474 struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
475 struct sock *sk = sock->sk; 475 struct sock *sk = sock->sk;
476 struct inet_sock *inet = inet_sk(sk); 476 struct inet_sock *inet = inet_sk(sk);
477 struct net *net = sock_net(sk);
477 unsigned short snum; 478 unsigned short snum;
478 int chk_addr_ret; 479 int chk_addr_ret;
479 int err; 480 int err;
480 481
481 /* If the socket has its own bind function then use it. (RAW) */ 482 /* If the socket has its own bind function then use it. (RAW) */
482 if (sk->sk_prot->bind) { 483 if (sk->sk_prot->bind) {
483 err = sk->sk_prot->bind(sk, uaddr, addr_len); 484 err = sk->sk_prot->bind(sk, uaddr, addr_len);
484 goto out; 485 goto out;
485 } 486 }
486 err = -EINVAL; 487 err = -EINVAL;
487 if (addr_len < sizeof(struct sockaddr_in)) 488 if (addr_len < sizeof(struct sockaddr_in))
488 goto out; 489 goto out;
489 490
490 if (addr->sin_family != AF_INET) { 491 if (addr->sin_family != AF_INET) {
491 /* Compatibility games : accept AF_UNSPEC (mapped to AF_INET) 492 /* Compatibility games : accept AF_UNSPEC (mapped to AF_INET)
492 * only if s_addr is INADDR_ANY. 493 * only if s_addr is INADDR_ANY.
493 */ 494 */
494 err = -EAFNOSUPPORT; 495 err = -EAFNOSUPPORT;
495 if (addr->sin_family != AF_UNSPEC || 496 if (addr->sin_family != AF_UNSPEC ||
496 addr->sin_addr.s_addr != htonl(INADDR_ANY)) 497 addr->sin_addr.s_addr != htonl(INADDR_ANY))
497 goto out; 498 goto out;
498 } 499 }
499 500
500 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr); 501 chk_addr_ret = inet_addr_type(net, addr->sin_addr.s_addr);
501 502
502 /* Not specified by any standard per-se, however it breaks too 503 /* Not specified by any standard per-se, however it breaks too
503 * many applications when removed. It is unfortunate since 504 * many applications when removed. It is unfortunate since
504 * allowing applications to make a non-local bind solves 505 * allowing applications to make a non-local bind solves
505 * several problems with systems using dynamic addressing. 506 * several problems with systems using dynamic addressing.
506 * (ie. your servers still start up even if your ISDN link 507 * (ie. your servers still start up even if your ISDN link
507 * is temporarily down) 508 * is temporarily down)
508 */ 509 */
509 err = -EADDRNOTAVAIL; 510 err = -EADDRNOTAVAIL;
510 if (!sysctl_ip_nonlocal_bind && 511 if (!sysctl_ip_nonlocal_bind &&
511 !(inet->freebind || inet->transparent) && 512 !(inet->freebind || inet->transparent) &&
512 addr->sin_addr.s_addr != htonl(INADDR_ANY) && 513 addr->sin_addr.s_addr != htonl(INADDR_ANY) &&
513 chk_addr_ret != RTN_LOCAL && 514 chk_addr_ret != RTN_LOCAL &&
514 chk_addr_ret != RTN_MULTICAST && 515 chk_addr_ret != RTN_MULTICAST &&
515 chk_addr_ret != RTN_BROADCAST) 516 chk_addr_ret != RTN_BROADCAST)
516 goto out; 517 goto out;
517 518
518 snum = ntohs(addr->sin_port); 519 snum = ntohs(addr->sin_port);
519 err = -EACCES; 520 err = -EACCES;
520 if (snum && snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE)) 521 if (snum && snum < PROT_SOCK &&
522 !ns_capable(net->user_ns, CAP_NET_BIND_SERVICE))
521 goto out; 523 goto out;
522 524
523 /* We keep a pair of addresses. rcv_saddr is the one 525 /* We keep a pair of addresses. rcv_saddr is the one
524 * used by hash lookups, and saddr is used for transmit. 526 * used by hash lookups, and saddr is used for transmit.
525 * 527 *
526 * In the BSD API these are the same except where it 528 * In the BSD API these are the same except where it
527 * would be illegal to use them (multicast/broadcast) in 529 * would be illegal to use them (multicast/broadcast) in
528 * which case the sending device address is used. 530 * which case the sending device address is used.
529 */ 531 */
530 lock_sock(sk); 532 lock_sock(sk);
531 533
532 /* Check these errors (active socket, double bind). */ 534 /* Check these errors (active socket, double bind). */
533 err = -EINVAL; 535 err = -EINVAL;
534 if (sk->sk_state != TCP_CLOSE || inet->inet_num) 536 if (sk->sk_state != TCP_CLOSE || inet->inet_num)
535 goto out_release_sock; 537 goto out_release_sock;
536 538
537 inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr; 539 inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
538 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST) 540 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
539 inet->inet_saddr = 0; /* Use device */ 541 inet->inet_saddr = 0; /* Use device */
540 542
541 /* Make sure we are allowed to bind here. */ 543 /* Make sure we are allowed to bind here. */
542 if (sk->sk_prot->get_port(sk, snum)) { 544 if (sk->sk_prot->get_port(sk, snum)) {
543 inet->inet_saddr = inet->inet_rcv_saddr = 0; 545 inet->inet_saddr = inet->inet_rcv_saddr = 0;
544 err = -EADDRINUSE; 546 err = -EADDRINUSE;
545 goto out_release_sock; 547 goto out_release_sock;
546 } 548 }
547 549
548 if (inet->inet_rcv_saddr) 550 if (inet->inet_rcv_saddr)
549 sk->sk_userlocks |= SOCK_BINDADDR_LOCK; 551 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
550 if (snum) 552 if (snum)
551 sk->sk_userlocks |= SOCK_BINDPORT_LOCK; 553 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
552 inet->inet_sport = htons(inet->inet_num); 554 inet->inet_sport = htons(inet->inet_num);
553 inet->inet_daddr = 0; 555 inet->inet_daddr = 0;
554 inet->inet_dport = 0; 556 inet->inet_dport = 0;
555 sk_dst_reset(sk); 557 sk_dst_reset(sk);
556 err = 0; 558 err = 0;
557 out_release_sock: 559 out_release_sock:
558 release_sock(sk); 560 release_sock(sk);
559 out: 561 out:
560 return err; 562 return err;
561 } 563 }
562 EXPORT_SYMBOL(inet_bind); 564 EXPORT_SYMBOL(inet_bind);
563 565
564 int inet_dgram_connect(struct socket *sock, struct sockaddr *uaddr, 566 int inet_dgram_connect(struct socket *sock, struct sockaddr *uaddr,
565 int addr_len, int flags) 567 int addr_len, int flags)
566 { 568 {
567 struct sock *sk = sock->sk; 569 struct sock *sk = sock->sk;
568 570
569 if (addr_len < sizeof(uaddr->sa_family)) 571 if (addr_len < sizeof(uaddr->sa_family))
570 return -EINVAL; 572 return -EINVAL;
571 if (uaddr->sa_family == AF_UNSPEC) 573 if (uaddr->sa_family == AF_UNSPEC)
572 return sk->sk_prot->disconnect(sk, flags); 574 return sk->sk_prot->disconnect(sk, flags);
573 575
574 if (!inet_sk(sk)->inet_num && inet_autobind(sk)) 576 if (!inet_sk(sk)->inet_num && inet_autobind(sk))
575 return -EAGAIN; 577 return -EAGAIN;
576 return sk->sk_prot->connect(sk, uaddr, addr_len); 578 return sk->sk_prot->connect(sk, uaddr, addr_len);
577 } 579 }
578 EXPORT_SYMBOL(inet_dgram_connect); 580 EXPORT_SYMBOL(inet_dgram_connect);
579 581
580 static long inet_wait_for_connect(struct sock *sk, long timeo, int writebias) 582 static long inet_wait_for_connect(struct sock *sk, long timeo, int writebias)
581 { 583 {
582 DEFINE_WAIT(wait); 584 DEFINE_WAIT(wait);
583 585
584 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); 586 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
585 sk->sk_write_pending += writebias; 587 sk->sk_write_pending += writebias;
586 588
587 /* Basic assumption: if someone sets sk->sk_err, he _must_ 589 /* Basic assumption: if someone sets sk->sk_err, he _must_
588 * change state of the socket from TCP_SYN_*. 590 * change state of the socket from TCP_SYN_*.
589 * Connect() does not allow to get error notifications 591 * Connect() does not allow to get error notifications
590 * without closing the socket. 592 * without closing the socket.
591 */ 593 */
592 while ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) { 594 while ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) {
593 release_sock(sk); 595 release_sock(sk);
594 timeo = schedule_timeout(timeo); 596 timeo = schedule_timeout(timeo);
595 lock_sock(sk); 597 lock_sock(sk);
596 if (signal_pending(current) || !timeo) 598 if (signal_pending(current) || !timeo)
597 break; 599 break;
598 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); 600 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
599 } 601 }
600 finish_wait(sk_sleep(sk), &wait); 602 finish_wait(sk_sleep(sk), &wait);
601 sk->sk_write_pending -= writebias; 603 sk->sk_write_pending -= writebias;
602 return timeo; 604 return timeo;
603 } 605 }
604 606
605 /* 607 /*
606 * Connect to a remote host. There is regrettably still a little 608 * Connect to a remote host. There is regrettably still a little
607 * TCP 'magic' in here. 609 * TCP 'magic' in here.
608 */ 610 */
609 int __inet_stream_connect(struct socket *sock, struct sockaddr *uaddr, 611 int __inet_stream_connect(struct socket *sock, struct sockaddr *uaddr,
610 int addr_len, int flags) 612 int addr_len, int flags)
611 { 613 {
612 struct sock *sk = sock->sk; 614 struct sock *sk = sock->sk;
613 int err; 615 int err;
614 long timeo; 616 long timeo;
615 617
616 if (addr_len < sizeof(uaddr->sa_family)) 618 if (addr_len < sizeof(uaddr->sa_family))
617 return -EINVAL; 619 return -EINVAL;
618 620
619 if (uaddr->sa_family == AF_UNSPEC) { 621 if (uaddr->sa_family == AF_UNSPEC) {
620 err = sk->sk_prot->disconnect(sk, flags); 622 err = sk->sk_prot->disconnect(sk, flags);
621 sock->state = err ? SS_DISCONNECTING : SS_UNCONNECTED; 623 sock->state = err ? SS_DISCONNECTING : SS_UNCONNECTED;
622 goto out; 624 goto out;
623 } 625 }
624 626
625 switch (sock->state) { 627 switch (sock->state) {
626 default: 628 default:
627 err = -EINVAL; 629 err = -EINVAL;
628 goto out; 630 goto out;
629 case SS_CONNECTED: 631 case SS_CONNECTED:
630 err = -EISCONN; 632 err = -EISCONN;
631 goto out; 633 goto out;
632 case SS_CONNECTING: 634 case SS_CONNECTING:
633 err = -EALREADY; 635 err = -EALREADY;
634 /* Fall out of switch with err, set for this state */ 636 /* Fall out of switch with err, set for this state */
635 break; 637 break;
636 case SS_UNCONNECTED: 638 case SS_UNCONNECTED:
637 err = -EISCONN; 639 err = -EISCONN;
638 if (sk->sk_state != TCP_CLOSE) 640 if (sk->sk_state != TCP_CLOSE)
639 goto out; 641 goto out;
640 642
641 err = sk->sk_prot->connect(sk, uaddr, addr_len); 643 err = sk->sk_prot->connect(sk, uaddr, addr_len);
642 if (err < 0) 644 if (err < 0)
643 goto out; 645 goto out;
644 646
645 sock->state = SS_CONNECTING; 647 sock->state = SS_CONNECTING;
646 648
647 /* Just entered SS_CONNECTING state; the only 649 /* Just entered SS_CONNECTING state; the only
648 * difference is that return value in non-blocking 650 * difference is that return value in non-blocking
649 * case is EINPROGRESS, rather than EALREADY. 651 * case is EINPROGRESS, rather than EALREADY.
650 */ 652 */
651 err = -EINPROGRESS; 653 err = -EINPROGRESS;
652 break; 654 break;
653 } 655 }
654 656
655 timeo = sock_sndtimeo(sk, flags & O_NONBLOCK); 657 timeo = sock_sndtimeo(sk, flags & O_NONBLOCK);
656 658
657 if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) { 659 if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) {
658 int writebias = (sk->sk_protocol == IPPROTO_TCP) && 660 int writebias = (sk->sk_protocol == IPPROTO_TCP) &&
659 tcp_sk(sk)->fastopen_req && 661 tcp_sk(sk)->fastopen_req &&
660 tcp_sk(sk)->fastopen_req->data ? 1 : 0; 662 tcp_sk(sk)->fastopen_req->data ? 1 : 0;
661 663
662 /* Error code is set above */ 664 /* Error code is set above */
663 if (!timeo || !inet_wait_for_connect(sk, timeo, writebias)) 665 if (!timeo || !inet_wait_for_connect(sk, timeo, writebias))
664 goto out; 666 goto out;
665 667
666 err = sock_intr_errno(timeo); 668 err = sock_intr_errno(timeo);
667 if (signal_pending(current)) 669 if (signal_pending(current))
668 goto out; 670 goto out;
669 } 671 }
670 672
671 /* Connection was closed by RST, timeout, ICMP error 673 /* Connection was closed by RST, timeout, ICMP error
672 * or another process disconnected us. 674 * or another process disconnected us.
673 */ 675 */
674 if (sk->sk_state == TCP_CLOSE) 676 if (sk->sk_state == TCP_CLOSE)
675 goto sock_error; 677 goto sock_error;
676 678
677 /* sk->sk_err may be not zero now, if RECVERR was ordered by user 679 /* sk->sk_err may be not zero now, if RECVERR was ordered by user
678 * and error was received after socket entered established state. 680 * and error was received after socket entered established state.
679 * Hence, it is handled normally after connect() return successfully. 681 * Hence, it is handled normally after connect() return successfully.
680 */ 682 */
681 683
682 sock->state = SS_CONNECTED; 684 sock->state = SS_CONNECTED;
683 err = 0; 685 err = 0;
684 out: 686 out:
685 return err; 687 return err;
686 688
687 sock_error: 689 sock_error:
688 err = sock_error(sk) ? : -ECONNABORTED; 690 err = sock_error(sk) ? : -ECONNABORTED;
689 sock->state = SS_UNCONNECTED; 691 sock->state = SS_UNCONNECTED;
690 if (sk->sk_prot->disconnect(sk, flags)) 692 if (sk->sk_prot->disconnect(sk, flags))
691 sock->state = SS_DISCONNECTING; 693 sock->state = SS_DISCONNECTING;
692 goto out; 694 goto out;
693 } 695 }
694 EXPORT_SYMBOL(__inet_stream_connect); 696 EXPORT_SYMBOL(__inet_stream_connect);
695 697
696 int inet_stream_connect(struct socket *sock, struct sockaddr *uaddr, 698 int inet_stream_connect(struct socket *sock, struct sockaddr *uaddr,
697 int addr_len, int flags) 699 int addr_len, int flags)
698 { 700 {
699 int err; 701 int err;
700 702
701 lock_sock(sock->sk); 703 lock_sock(sock->sk);
702 err = __inet_stream_connect(sock, uaddr, addr_len, flags); 704 err = __inet_stream_connect(sock, uaddr, addr_len, flags);
703 release_sock(sock->sk); 705 release_sock(sock->sk);
704 return err; 706 return err;
705 } 707 }
706 EXPORT_SYMBOL(inet_stream_connect); 708 EXPORT_SYMBOL(inet_stream_connect);
707 709
708 /* 710 /*
709 * Accept a pending connection. The TCP layer now gives BSD semantics. 711 * Accept a pending connection. The TCP layer now gives BSD semantics.
710 */ 712 */
711 713
712 int inet_accept(struct socket *sock, struct socket *newsock, int flags) 714 int inet_accept(struct socket *sock, struct socket *newsock, int flags)
713 { 715 {
714 struct sock *sk1 = sock->sk; 716 struct sock *sk1 = sock->sk;
715 int err = -EINVAL; 717 int err = -EINVAL;
716 struct sock *sk2 = sk1->sk_prot->accept(sk1, flags, &err); 718 struct sock *sk2 = sk1->sk_prot->accept(sk1, flags, &err);
717 719
718 if (!sk2) 720 if (!sk2)
719 goto do_err; 721 goto do_err;
720 722
721 lock_sock(sk2); 723 lock_sock(sk2);
722 724
723 sock_rps_record_flow(sk2); 725 sock_rps_record_flow(sk2);
724 WARN_ON(!((1 << sk2->sk_state) & 726 WARN_ON(!((1 << sk2->sk_state) &
725 (TCPF_ESTABLISHED | TCPF_SYN_RECV | 727 (TCPF_ESTABLISHED | TCPF_SYN_RECV |
726 TCPF_CLOSE_WAIT | TCPF_CLOSE))); 728 TCPF_CLOSE_WAIT | TCPF_CLOSE)));
727 729
728 sock_graft(sk2, newsock); 730 sock_graft(sk2, newsock);
729 731
730 newsock->state = SS_CONNECTED; 732 newsock->state = SS_CONNECTED;
731 err = 0; 733 err = 0;
732 release_sock(sk2); 734 release_sock(sk2);
733 do_err: 735 do_err:
734 return err; 736 return err;
735 } 737 }
736 EXPORT_SYMBOL(inet_accept); 738 EXPORT_SYMBOL(inet_accept);
737 739
738 740
739 /* 741 /*
740 * This does both peername and sockname. 742 * This does both peername and sockname.
741 */ 743 */
742 int inet_getname(struct socket *sock, struct sockaddr *uaddr, 744 int inet_getname(struct socket *sock, struct sockaddr *uaddr,
743 int *uaddr_len, int peer) 745 int *uaddr_len, int peer)
744 { 746 {
745 struct sock *sk = sock->sk; 747 struct sock *sk = sock->sk;
746 struct inet_sock *inet = inet_sk(sk); 748 struct inet_sock *inet = inet_sk(sk);
747 DECLARE_SOCKADDR(struct sockaddr_in *, sin, uaddr); 749 DECLARE_SOCKADDR(struct sockaddr_in *, sin, uaddr);
748 750
749 sin->sin_family = AF_INET; 751 sin->sin_family = AF_INET;
750 if (peer) { 752 if (peer) {
751 if (!inet->inet_dport || 753 if (!inet->inet_dport ||
752 (((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_SYN_SENT)) && 754 (((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_SYN_SENT)) &&
753 peer == 1)) 755 peer == 1))
754 return -ENOTCONN; 756 return -ENOTCONN;
755 sin->sin_port = inet->inet_dport; 757 sin->sin_port = inet->inet_dport;
756 sin->sin_addr.s_addr = inet->inet_daddr; 758 sin->sin_addr.s_addr = inet->inet_daddr;
757 } else { 759 } else {
758 __be32 addr = inet->inet_rcv_saddr; 760 __be32 addr = inet->inet_rcv_saddr;
759 if (!addr) 761 if (!addr)
760 addr = inet->inet_saddr; 762 addr = inet->inet_saddr;
761 sin->sin_port = inet->inet_sport; 763 sin->sin_port = inet->inet_sport;
762 sin->sin_addr.s_addr = addr; 764 sin->sin_addr.s_addr = addr;
763 } 765 }
764 memset(sin->sin_zero, 0, sizeof(sin->sin_zero)); 766 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
765 *uaddr_len = sizeof(*sin); 767 *uaddr_len = sizeof(*sin);
766 return 0; 768 return 0;
767 } 769 }
768 EXPORT_SYMBOL(inet_getname); 770 EXPORT_SYMBOL(inet_getname);
769 771
770 int inet_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg, 772 int inet_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg,
771 size_t size) 773 size_t size)
772 { 774 {
773 struct sock *sk = sock->sk; 775 struct sock *sk = sock->sk;
774 776
775 sock_rps_record_flow(sk); 777 sock_rps_record_flow(sk);
776 778
777 /* We may need to bind the socket. */ 779 /* We may need to bind the socket. */
778 if (!inet_sk(sk)->inet_num && !sk->sk_prot->no_autobind && 780 if (!inet_sk(sk)->inet_num && !sk->sk_prot->no_autobind &&
779 inet_autobind(sk)) 781 inet_autobind(sk))
780 return -EAGAIN; 782 return -EAGAIN;
781 783
782 return sk->sk_prot->sendmsg(iocb, sk, msg, size); 784 return sk->sk_prot->sendmsg(iocb, sk, msg, size);
783 } 785 }
784 EXPORT_SYMBOL(inet_sendmsg); 786 EXPORT_SYMBOL(inet_sendmsg);
785 787
786 ssize_t inet_sendpage(struct socket *sock, struct page *page, int offset, 788 ssize_t inet_sendpage(struct socket *sock, struct page *page, int offset,
787 size_t size, int flags) 789 size_t size, int flags)
788 { 790 {
789 struct sock *sk = sock->sk; 791 struct sock *sk = sock->sk;
790 792
791 sock_rps_record_flow(sk); 793 sock_rps_record_flow(sk);
792 794
793 /* We may need to bind the socket. */ 795 /* We may need to bind the socket. */
794 if (!inet_sk(sk)->inet_num && !sk->sk_prot->no_autobind && 796 if (!inet_sk(sk)->inet_num && !sk->sk_prot->no_autobind &&
795 inet_autobind(sk)) 797 inet_autobind(sk))
796 return -EAGAIN; 798 return -EAGAIN;
797 799
798 if (sk->sk_prot->sendpage) 800 if (sk->sk_prot->sendpage)
799 return sk->sk_prot->sendpage(sk, page, offset, size, flags); 801 return sk->sk_prot->sendpage(sk, page, offset, size, flags);
800 return sock_no_sendpage(sock, page, offset, size, flags); 802 return sock_no_sendpage(sock, page, offset, size, flags);
801 } 803 }
802 EXPORT_SYMBOL(inet_sendpage); 804 EXPORT_SYMBOL(inet_sendpage);
803 805
804 int inet_recvmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg, 806 int inet_recvmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *msg,
805 size_t size, int flags) 807 size_t size, int flags)
806 { 808 {
807 struct sock *sk = sock->sk; 809 struct sock *sk = sock->sk;
808 int addr_len = 0; 810 int addr_len = 0;
809 int err; 811 int err;
810 812
811 sock_rps_record_flow(sk); 813 sock_rps_record_flow(sk);
812 814
813 err = sk->sk_prot->recvmsg(iocb, sk, msg, size, flags & MSG_DONTWAIT, 815 err = sk->sk_prot->recvmsg(iocb, sk, msg, size, flags & MSG_DONTWAIT,
814 flags & ~MSG_DONTWAIT, &addr_len); 816 flags & ~MSG_DONTWAIT, &addr_len);
815 if (err >= 0) 817 if (err >= 0)
816 msg->msg_namelen = addr_len; 818 msg->msg_namelen = addr_len;
817 return err; 819 return err;
818 } 820 }
819 EXPORT_SYMBOL(inet_recvmsg); 821 EXPORT_SYMBOL(inet_recvmsg);
820 822
821 int inet_shutdown(struct socket *sock, int how) 823 int inet_shutdown(struct socket *sock, int how)
822 { 824 {
823 struct sock *sk = sock->sk; 825 struct sock *sk = sock->sk;
824 int err = 0; 826 int err = 0;
825 827
826 /* This should really check to make sure 828 /* This should really check to make sure
827 * the socket is a TCP socket. (WHY AC...) 829 * the socket is a TCP socket. (WHY AC...)
828 */ 830 */
829 how++; /* maps 0->1 has the advantage of making bit 1 rcvs and 831 how++; /* maps 0->1 has the advantage of making bit 1 rcvs and
830 1->2 bit 2 snds. 832 1->2 bit 2 snds.
831 2->3 */ 833 2->3 */
832 if ((how & ~SHUTDOWN_MASK) || !how) /* MAXINT->0 */ 834 if ((how & ~SHUTDOWN_MASK) || !how) /* MAXINT->0 */
833 return -EINVAL; 835 return -EINVAL;
834 836
835 lock_sock(sk); 837 lock_sock(sk);
836 if (sock->state == SS_CONNECTING) { 838 if (sock->state == SS_CONNECTING) {
837 if ((1 << sk->sk_state) & 839 if ((1 << sk->sk_state) &
838 (TCPF_SYN_SENT | TCPF_SYN_RECV | TCPF_CLOSE)) 840 (TCPF_SYN_SENT | TCPF_SYN_RECV | TCPF_CLOSE))
839 sock->state = SS_DISCONNECTING; 841 sock->state = SS_DISCONNECTING;
840 else 842 else
841 sock->state = SS_CONNECTED; 843 sock->state = SS_CONNECTED;
842 } 844 }
843 845
844 switch (sk->sk_state) { 846 switch (sk->sk_state) {
845 case TCP_CLOSE: 847 case TCP_CLOSE:
846 err = -ENOTCONN; 848 err = -ENOTCONN;
847 /* Hack to wake up other listeners, who can poll for 849 /* Hack to wake up other listeners, who can poll for
848 POLLHUP, even on eg. unconnected UDP sockets -- RR */ 850 POLLHUP, even on eg. unconnected UDP sockets -- RR */
849 default: 851 default:
850 sk->sk_shutdown |= how; 852 sk->sk_shutdown |= how;
851 if (sk->sk_prot->shutdown) 853 if (sk->sk_prot->shutdown)
852 sk->sk_prot->shutdown(sk, how); 854 sk->sk_prot->shutdown(sk, how);
853 break; 855 break;
854 856
855 /* Remaining two branches are temporary solution for missing 857 /* Remaining two branches are temporary solution for missing
856 * close() in multithreaded environment. It is _not_ a good idea, 858 * close() in multithreaded environment. It is _not_ a good idea,
857 * but we have no choice until close() is repaired at VFS level. 859 * but we have no choice until close() is repaired at VFS level.
858 */ 860 */
859 case TCP_LISTEN: 861 case TCP_LISTEN:
860 if (!(how & RCV_SHUTDOWN)) 862 if (!(how & RCV_SHUTDOWN))
861 break; 863 break;
862 /* Fall through */ 864 /* Fall through */
863 case TCP_SYN_SENT: 865 case TCP_SYN_SENT:
864 err = sk->sk_prot->disconnect(sk, O_NONBLOCK); 866 err = sk->sk_prot->disconnect(sk, O_NONBLOCK);
865 sock->state = err ? SS_DISCONNECTING : SS_UNCONNECTED; 867 sock->state = err ? SS_DISCONNECTING : SS_UNCONNECTED;
866 break; 868 break;
867 } 869 }
868 870
869 /* Wake up anyone sleeping in poll. */ 871 /* Wake up anyone sleeping in poll. */
870 sk->sk_state_change(sk); 872 sk->sk_state_change(sk);
871 release_sock(sk); 873 release_sock(sk);
872 return err; 874 return err;
873 } 875 }
874 EXPORT_SYMBOL(inet_shutdown); 876 EXPORT_SYMBOL(inet_shutdown);
875 877
876 /* 878 /*
877 * ioctl() calls you can issue on an INET socket. Most of these are 879 * ioctl() calls you can issue on an INET socket. Most of these are
878 * device configuration and stuff and very rarely used. Some ioctls 880 * device configuration and stuff and very rarely used. Some ioctls
879 * pass on to the socket itself. 881 * pass on to the socket itself.
880 * 882 *
881 * NOTE: I like the idea of a module for the config stuff. ie ifconfig 883 * NOTE: I like the idea of a module for the config stuff. ie ifconfig
882 * loads the devconfigure module does its configuring and unloads it. 884 * loads the devconfigure module does its configuring and unloads it.
883 * There's a good 20K of config code hanging around the kernel. 885 * There's a good 20K of config code hanging around the kernel.
884 */ 886 */
885 887
886 int inet_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg) 888 int inet_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
887 { 889 {
888 struct sock *sk = sock->sk; 890 struct sock *sk = sock->sk;
889 int err = 0; 891 int err = 0;
890 struct net *net = sock_net(sk); 892 struct net *net = sock_net(sk);
891 893
892 switch (cmd) { 894 switch (cmd) {
893 case SIOCGSTAMP: 895 case SIOCGSTAMP:
894 err = sock_get_timestamp(sk, (struct timeval __user *)arg); 896 err = sock_get_timestamp(sk, (struct timeval __user *)arg);
895 break; 897 break;
896 case SIOCGSTAMPNS: 898 case SIOCGSTAMPNS:
897 err = sock_get_timestampns(sk, (struct timespec __user *)arg); 899 err = sock_get_timestampns(sk, (struct timespec __user *)arg);
898 break; 900 break;
899 case SIOCADDRT: 901 case SIOCADDRT:
900 case SIOCDELRT: 902 case SIOCDELRT:
901 case SIOCRTMSG: 903 case SIOCRTMSG:
902 err = ip_rt_ioctl(net, cmd, (void __user *)arg); 904 err = ip_rt_ioctl(net, cmd, (void __user *)arg);
903 break; 905 break;
904 case SIOCDARP: 906 case SIOCDARP:
905 case SIOCGARP: 907 case SIOCGARP:
906 case SIOCSARP: 908 case SIOCSARP:
907 err = arp_ioctl(net, cmd, (void __user *)arg); 909 err = arp_ioctl(net, cmd, (void __user *)arg);
908 break; 910 break;
909 case SIOCGIFADDR: 911 case SIOCGIFADDR:
910 case SIOCSIFADDR: 912 case SIOCSIFADDR:
911 case SIOCGIFBRDADDR: 913 case SIOCGIFBRDADDR:
912 case SIOCSIFBRDADDR: 914 case SIOCSIFBRDADDR:
913 case SIOCGIFNETMASK: 915 case SIOCGIFNETMASK:
914 case SIOCSIFNETMASK: 916 case SIOCSIFNETMASK:
915 case SIOCGIFDSTADDR: 917 case SIOCGIFDSTADDR:
916 case SIOCSIFDSTADDR: 918 case SIOCSIFDSTADDR:
917 case SIOCSIFPFLAGS: 919 case SIOCSIFPFLAGS:
918 case SIOCGIFPFLAGS: 920 case SIOCGIFPFLAGS:
919 case SIOCSIFFLAGS: 921 case SIOCSIFFLAGS:
920 err = devinet_ioctl(net, cmd, (void __user *)arg); 922 err = devinet_ioctl(net, cmd, (void __user *)arg);
921 break; 923 break;
922 default: 924 default:
923 if (sk->sk_prot->ioctl) 925 if (sk->sk_prot->ioctl)
924 err = sk->sk_prot->ioctl(sk, cmd, arg); 926 err = sk->sk_prot->ioctl(sk, cmd, arg);
925 else 927 else
926 err = -ENOIOCTLCMD; 928 err = -ENOIOCTLCMD;
927 break; 929 break;
928 } 930 }
929 return err; 931 return err;
930 } 932 }
931 EXPORT_SYMBOL(inet_ioctl); 933 EXPORT_SYMBOL(inet_ioctl);
932 934
933 #ifdef CONFIG_COMPAT 935 #ifdef CONFIG_COMPAT
934 static int inet_compat_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg) 936 static int inet_compat_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
935 { 937 {
936 struct sock *sk = sock->sk; 938 struct sock *sk = sock->sk;
937 int err = -ENOIOCTLCMD; 939 int err = -ENOIOCTLCMD;
938 940
939 if (sk->sk_prot->compat_ioctl) 941 if (sk->sk_prot->compat_ioctl)
940 err = sk->sk_prot->compat_ioctl(sk, cmd, arg); 942 err = sk->sk_prot->compat_ioctl(sk, cmd, arg);
941 943
942 return err; 944 return err;
943 } 945 }
944 #endif 946 #endif
945 947
946 const struct proto_ops inet_stream_ops = { 948 const struct proto_ops inet_stream_ops = {
947 .family = PF_INET, 949 .family = PF_INET,
948 .owner = THIS_MODULE, 950 .owner = THIS_MODULE,
949 .release = inet_release, 951 .release = inet_release,
950 .bind = inet_bind, 952 .bind = inet_bind,
951 .connect = inet_stream_connect, 953 .connect = inet_stream_connect,
952 .socketpair = sock_no_socketpair, 954 .socketpair = sock_no_socketpair,
953 .accept = inet_accept, 955 .accept = inet_accept,
954 .getname = inet_getname, 956 .getname = inet_getname,
955 .poll = tcp_poll, 957 .poll = tcp_poll,
956 .ioctl = inet_ioctl, 958 .ioctl = inet_ioctl,
957 .listen = inet_listen, 959 .listen = inet_listen,
958 .shutdown = inet_shutdown, 960 .shutdown = inet_shutdown,
959 .setsockopt = sock_common_setsockopt, 961 .setsockopt = sock_common_setsockopt,
960 .getsockopt = sock_common_getsockopt, 962 .getsockopt = sock_common_getsockopt,
961 .sendmsg = inet_sendmsg, 963 .sendmsg = inet_sendmsg,
962 .recvmsg = inet_recvmsg, 964 .recvmsg = inet_recvmsg,
963 .mmap = sock_no_mmap, 965 .mmap = sock_no_mmap,
964 .sendpage = inet_sendpage, 966 .sendpage = inet_sendpage,
965 .splice_read = tcp_splice_read, 967 .splice_read = tcp_splice_read,
966 #ifdef CONFIG_COMPAT 968 #ifdef CONFIG_COMPAT
967 .compat_setsockopt = compat_sock_common_setsockopt, 969 .compat_setsockopt = compat_sock_common_setsockopt,
968 .compat_getsockopt = compat_sock_common_getsockopt, 970 .compat_getsockopt = compat_sock_common_getsockopt,
969 .compat_ioctl = inet_compat_ioctl, 971 .compat_ioctl = inet_compat_ioctl,
970 #endif 972 #endif
971 }; 973 };
972 EXPORT_SYMBOL(inet_stream_ops); 974 EXPORT_SYMBOL(inet_stream_ops);
973 975
974 const struct proto_ops inet_dgram_ops = { 976 const struct proto_ops inet_dgram_ops = {
975 .family = PF_INET, 977 .family = PF_INET,
976 .owner = THIS_MODULE, 978 .owner = THIS_MODULE,
977 .release = inet_release, 979 .release = inet_release,
978 .bind = inet_bind, 980 .bind = inet_bind,
979 .connect = inet_dgram_connect, 981 .connect = inet_dgram_connect,
980 .socketpair = sock_no_socketpair, 982 .socketpair = sock_no_socketpair,
981 .accept = sock_no_accept, 983 .accept = sock_no_accept,
982 .getname = inet_getname, 984 .getname = inet_getname,
983 .poll = udp_poll, 985 .poll = udp_poll,
984 .ioctl = inet_ioctl, 986 .ioctl = inet_ioctl,
985 .listen = sock_no_listen, 987 .listen = sock_no_listen,
986 .shutdown = inet_shutdown, 988 .shutdown = inet_shutdown,
987 .setsockopt = sock_common_setsockopt, 989 .setsockopt = sock_common_setsockopt,
988 .getsockopt = sock_common_getsockopt, 990 .getsockopt = sock_common_getsockopt,
989 .sendmsg = inet_sendmsg, 991 .sendmsg = inet_sendmsg,
990 .recvmsg = inet_recvmsg, 992 .recvmsg = inet_recvmsg,
991 .mmap = sock_no_mmap, 993 .mmap = sock_no_mmap,
992 .sendpage = inet_sendpage, 994 .sendpage = inet_sendpage,
993 #ifdef CONFIG_COMPAT 995 #ifdef CONFIG_COMPAT
994 .compat_setsockopt = compat_sock_common_setsockopt, 996 .compat_setsockopt = compat_sock_common_setsockopt,
995 .compat_getsockopt = compat_sock_common_getsockopt, 997 .compat_getsockopt = compat_sock_common_getsockopt,
996 .compat_ioctl = inet_compat_ioctl, 998 .compat_ioctl = inet_compat_ioctl,
997 #endif 999 #endif
998 }; 1000 };
999 EXPORT_SYMBOL(inet_dgram_ops); 1001 EXPORT_SYMBOL(inet_dgram_ops);
1000 1002
1001 /* 1003 /*
1002 * For SOCK_RAW sockets; should be the same as inet_dgram_ops but without 1004 * For SOCK_RAW sockets; should be the same as inet_dgram_ops but without
1003 * udp_poll 1005 * udp_poll
1004 */ 1006 */
1005 static const struct proto_ops inet_sockraw_ops = { 1007 static const struct proto_ops inet_sockraw_ops = {
1006 .family = PF_INET, 1008 .family = PF_INET,
1007 .owner = THIS_MODULE, 1009 .owner = THIS_MODULE,
1008 .release = inet_release, 1010 .release = inet_release,
1009 .bind = inet_bind, 1011 .bind = inet_bind,
1010 .connect = inet_dgram_connect, 1012 .connect = inet_dgram_connect,
1011 .socketpair = sock_no_socketpair, 1013 .socketpair = sock_no_socketpair,
1012 .accept = sock_no_accept, 1014 .accept = sock_no_accept,
1013 .getname = inet_getname, 1015 .getname = inet_getname,
1014 .poll = datagram_poll, 1016 .poll = datagram_poll,
1015 .ioctl = inet_ioctl, 1017 .ioctl = inet_ioctl,
1016 .listen = sock_no_listen, 1018 .listen = sock_no_listen,
1017 .shutdown = inet_shutdown, 1019 .shutdown = inet_shutdown,
1018 .setsockopt = sock_common_setsockopt, 1020 .setsockopt = sock_common_setsockopt,
1019 .getsockopt = sock_common_getsockopt, 1021 .getsockopt = sock_common_getsockopt,
1020 .sendmsg = inet_sendmsg, 1022 .sendmsg = inet_sendmsg,
1021 .recvmsg = inet_recvmsg, 1023 .recvmsg = inet_recvmsg,
1022 .mmap = sock_no_mmap, 1024 .mmap = sock_no_mmap,
1023 .sendpage = inet_sendpage, 1025 .sendpage = inet_sendpage,
1024 #ifdef CONFIG_COMPAT 1026 #ifdef CONFIG_COMPAT
1025 .compat_setsockopt = compat_sock_common_setsockopt, 1027 .compat_setsockopt = compat_sock_common_setsockopt,
1026 .compat_getsockopt = compat_sock_common_getsockopt, 1028 .compat_getsockopt = compat_sock_common_getsockopt,
1027 .compat_ioctl = inet_compat_ioctl, 1029 .compat_ioctl = inet_compat_ioctl,
1028 #endif 1030 #endif
1029 }; 1031 };
1030 1032
1031 static const struct net_proto_family inet_family_ops = { 1033 static const struct net_proto_family inet_family_ops = {
1032 .family = PF_INET, 1034 .family = PF_INET,
1033 .create = inet_create, 1035 .create = inet_create,
1034 .owner = THIS_MODULE, 1036 .owner = THIS_MODULE,
1035 }; 1037 };
1036 1038
1037 /* Upon startup we insert all the elements in inetsw_array[] into 1039 /* Upon startup we insert all the elements in inetsw_array[] into
1038 * the linked list inetsw. 1040 * the linked list inetsw.
1039 */ 1041 */
1040 static struct inet_protosw inetsw_array[] = 1042 static struct inet_protosw inetsw_array[] =
1041 { 1043 {
1042 { 1044 {
1043 .type = SOCK_STREAM, 1045 .type = SOCK_STREAM,
1044 .protocol = IPPROTO_TCP, 1046 .protocol = IPPROTO_TCP,
1045 .prot = &tcp_prot, 1047 .prot = &tcp_prot,
1046 .ops = &inet_stream_ops, 1048 .ops = &inet_stream_ops,
1047 .no_check = 0, 1049 .no_check = 0,
1048 .flags = INET_PROTOSW_PERMANENT | 1050 .flags = INET_PROTOSW_PERMANENT |
1049 INET_PROTOSW_ICSK, 1051 INET_PROTOSW_ICSK,
1050 }, 1052 },
1051 1053
1052 { 1054 {
1053 .type = SOCK_DGRAM, 1055 .type = SOCK_DGRAM,
1054 .protocol = IPPROTO_UDP, 1056 .protocol = IPPROTO_UDP,
1055 .prot = &udp_prot, 1057 .prot = &udp_prot,
1056 .ops = &inet_dgram_ops, 1058 .ops = &inet_dgram_ops,
1057 .no_check = UDP_CSUM_DEFAULT, 1059 .no_check = UDP_CSUM_DEFAULT,
1058 .flags = INET_PROTOSW_PERMANENT, 1060 .flags = INET_PROTOSW_PERMANENT,
1059 }, 1061 },
1060 1062
1061 { 1063 {
1062 .type = SOCK_DGRAM, 1064 .type = SOCK_DGRAM,
1063 .protocol = IPPROTO_ICMP, 1065 .protocol = IPPROTO_ICMP,
1064 .prot = &ping_prot, 1066 .prot = &ping_prot,
1065 .ops = &inet_dgram_ops, 1067 .ops = &inet_dgram_ops,
1066 .no_check = UDP_CSUM_DEFAULT, 1068 .no_check = UDP_CSUM_DEFAULT,
1067 .flags = INET_PROTOSW_REUSE, 1069 .flags = INET_PROTOSW_REUSE,
1068 }, 1070 },
1069 1071
1070 { 1072 {
1071 .type = SOCK_RAW, 1073 .type = SOCK_RAW,
1072 .protocol = IPPROTO_IP, /* wild card */ 1074 .protocol = IPPROTO_IP, /* wild card */
1073 .prot = &raw_prot, 1075 .prot = &raw_prot,
1074 .ops = &inet_sockraw_ops, 1076 .ops = &inet_sockraw_ops,
1075 .no_check = UDP_CSUM_DEFAULT, 1077 .no_check = UDP_CSUM_DEFAULT,
1076 .flags = INET_PROTOSW_REUSE, 1078 .flags = INET_PROTOSW_REUSE,
1077 } 1079 }
1078 }; 1080 };
1079 1081
1080 #define INETSW_ARRAY_LEN ARRAY_SIZE(inetsw_array) 1082 #define INETSW_ARRAY_LEN ARRAY_SIZE(inetsw_array)
1081 1083
1082 void inet_register_protosw(struct inet_protosw *p) 1084 void inet_register_protosw(struct inet_protosw *p)
1083 { 1085 {
1084 struct list_head *lh; 1086 struct list_head *lh;
1085 struct inet_protosw *answer; 1087 struct inet_protosw *answer;
1086 int protocol = p->protocol; 1088 int protocol = p->protocol;
1087 struct list_head *last_perm; 1089 struct list_head *last_perm;
1088 1090
1089 spin_lock_bh(&inetsw_lock); 1091 spin_lock_bh(&inetsw_lock);
1090 1092
1091 if (p->type >= SOCK_MAX) 1093 if (p->type >= SOCK_MAX)
1092 goto out_illegal; 1094 goto out_illegal;
1093 1095
1094 /* If we are trying to override a permanent protocol, bail. */ 1096 /* If we are trying to override a permanent protocol, bail. */
1095 answer = NULL; 1097 answer = NULL;
1096 last_perm = &inetsw[p->type]; 1098 last_perm = &inetsw[p->type];
1097 list_for_each(lh, &inetsw[p->type]) { 1099 list_for_each(lh, &inetsw[p->type]) {
1098 answer = list_entry(lh, struct inet_protosw, list); 1100 answer = list_entry(lh, struct inet_protosw, list);
1099 1101
1100 /* Check only the non-wild match. */ 1102 /* Check only the non-wild match. */
1101 if (INET_PROTOSW_PERMANENT & answer->flags) { 1103 if (INET_PROTOSW_PERMANENT & answer->flags) {
1102 if (protocol == answer->protocol) 1104 if (protocol == answer->protocol)
1103 break; 1105 break;
1104 last_perm = lh; 1106 last_perm = lh;
1105 } 1107 }
1106 1108
1107 answer = NULL; 1109 answer = NULL;
1108 } 1110 }
1109 if (answer) 1111 if (answer)
1110 goto out_permanent; 1112 goto out_permanent;
1111 1113
1112 /* Add the new entry after the last permanent entry if any, so that 1114 /* Add the new entry after the last permanent entry if any, so that
1113 * the new entry does not override a permanent entry when matched with 1115 * the new entry does not override a permanent entry when matched with
1114 * a wild-card protocol. But it is allowed to override any existing 1116 * a wild-card protocol. But it is allowed to override any existing
1115 * non-permanent entry. This means that when we remove this entry, the 1117 * non-permanent entry. This means that when we remove this entry, the
1116 * system automatically returns to the old behavior. 1118 * system automatically returns to the old behavior.
1117 */ 1119 */
1118 list_add_rcu(&p->list, last_perm); 1120 list_add_rcu(&p->list, last_perm);
1119 out: 1121 out:
1120 spin_unlock_bh(&inetsw_lock); 1122 spin_unlock_bh(&inetsw_lock);
1121 1123
1122 return; 1124 return;
1123 1125
1124 out_permanent: 1126 out_permanent:
1125 pr_err("Attempt to override permanent protocol %d\n", protocol); 1127 pr_err("Attempt to override permanent protocol %d\n", protocol);
1126 goto out; 1128 goto out;
1127 1129
1128 out_illegal: 1130 out_illegal:
1129 pr_err("Ignoring attempt to register invalid socket type %d\n", 1131 pr_err("Ignoring attempt to register invalid socket type %d\n",
1130 p->type); 1132 p->type);
1131 goto out; 1133 goto out;
1132 } 1134 }
1133 EXPORT_SYMBOL(inet_register_protosw); 1135 EXPORT_SYMBOL(inet_register_protosw);
1134 1136
1135 void inet_unregister_protosw(struct inet_protosw *p) 1137 void inet_unregister_protosw(struct inet_protosw *p)
1136 { 1138 {
1137 if (INET_PROTOSW_PERMANENT & p->flags) { 1139 if (INET_PROTOSW_PERMANENT & p->flags) {
1138 pr_err("Attempt to unregister permanent protocol %d\n", 1140 pr_err("Attempt to unregister permanent protocol %d\n",
1139 p->protocol); 1141 p->protocol);
1140 } else { 1142 } else {
1141 spin_lock_bh(&inetsw_lock); 1143 spin_lock_bh(&inetsw_lock);
1142 list_del_rcu(&p->list); 1144 list_del_rcu(&p->list);
1143 spin_unlock_bh(&inetsw_lock); 1145 spin_unlock_bh(&inetsw_lock);
1144 1146
1145 synchronize_net(); 1147 synchronize_net();
1146 } 1148 }
1147 } 1149 }
1148 EXPORT_SYMBOL(inet_unregister_protosw); 1150 EXPORT_SYMBOL(inet_unregister_protosw);
1149 1151
1150 /* 1152 /*
1151 * Shall we try to damage output packets if routing dev changes? 1153 * Shall we try to damage output packets if routing dev changes?
1152 */ 1154 */
1153 1155
1154 int sysctl_ip_dynaddr __read_mostly; 1156 int sysctl_ip_dynaddr __read_mostly;
1155 1157
1156 static int inet_sk_reselect_saddr(struct sock *sk) 1158 static int inet_sk_reselect_saddr(struct sock *sk)
1157 { 1159 {
1158 struct inet_sock *inet = inet_sk(sk); 1160 struct inet_sock *inet = inet_sk(sk);
1159 __be32 old_saddr = inet->inet_saddr; 1161 __be32 old_saddr = inet->inet_saddr;
1160 __be32 daddr = inet->inet_daddr; 1162 __be32 daddr = inet->inet_daddr;
1161 struct flowi4 *fl4; 1163 struct flowi4 *fl4;
1162 struct rtable *rt; 1164 struct rtable *rt;
1163 __be32 new_saddr; 1165 __be32 new_saddr;
1164 struct ip_options_rcu *inet_opt; 1166 struct ip_options_rcu *inet_opt;
1165 1167
1166 inet_opt = rcu_dereference_protected(inet->inet_opt, 1168 inet_opt = rcu_dereference_protected(inet->inet_opt,
1167 sock_owned_by_user(sk)); 1169 sock_owned_by_user(sk));
1168 if (inet_opt && inet_opt->opt.srr) 1170 if (inet_opt && inet_opt->opt.srr)
1169 daddr = inet_opt->opt.faddr; 1171 daddr = inet_opt->opt.faddr;
1170 1172
1171 /* Query new route. */ 1173 /* Query new route. */
1172 fl4 = &inet->cork.fl.u.ip4; 1174 fl4 = &inet->cork.fl.u.ip4;
1173 rt = ip_route_connect(fl4, daddr, 0, RT_CONN_FLAGS(sk), 1175 rt = ip_route_connect(fl4, daddr, 0, RT_CONN_FLAGS(sk),
1174 sk->sk_bound_dev_if, sk->sk_protocol, 1176 sk->sk_bound_dev_if, sk->sk_protocol,
1175 inet->inet_sport, inet->inet_dport, sk, false); 1177 inet->inet_sport, inet->inet_dport, sk, false);
1176 if (IS_ERR(rt)) 1178 if (IS_ERR(rt))
1177 return PTR_ERR(rt); 1179 return PTR_ERR(rt);
1178 1180
1179 sk_setup_caps(sk, &rt->dst); 1181 sk_setup_caps(sk, &rt->dst);
1180 1182
1181 new_saddr = fl4->saddr; 1183 new_saddr = fl4->saddr;
1182 1184
1183 if (new_saddr == old_saddr) 1185 if (new_saddr == old_saddr)
1184 return 0; 1186 return 0;
1185 1187
1186 if (sysctl_ip_dynaddr > 1) { 1188 if (sysctl_ip_dynaddr > 1) {
1187 pr_info("%s(): shifting inet->saddr from %pI4 to %pI4\n", 1189 pr_info("%s(): shifting inet->saddr from %pI4 to %pI4\n",
1188 __func__, &old_saddr, &new_saddr); 1190 __func__, &old_saddr, &new_saddr);
1189 } 1191 }
1190 1192
1191 inet->inet_saddr = inet->inet_rcv_saddr = new_saddr; 1193 inet->inet_saddr = inet->inet_rcv_saddr = new_saddr;
1192 1194
1193 /* 1195 /*
1194 * XXX The only one ugly spot where we need to 1196 * XXX The only one ugly spot where we need to
1195 * XXX really change the sockets identity after 1197 * XXX really change the sockets identity after
1196 * XXX it has entered the hashes. -DaveM 1198 * XXX it has entered the hashes. -DaveM
1197 * 1199 *
1198 * Besides that, it does not check for connection 1200 * Besides that, it does not check for connection
1199 * uniqueness. Wait for troubles. 1201 * uniqueness. Wait for troubles.
1200 */ 1202 */
1201 __sk_prot_rehash(sk); 1203 __sk_prot_rehash(sk);
1202 return 0; 1204 return 0;
1203 } 1205 }
1204 1206
1205 int inet_sk_rebuild_header(struct sock *sk) 1207 int inet_sk_rebuild_header(struct sock *sk)
1206 { 1208 {
1207 struct inet_sock *inet = inet_sk(sk); 1209 struct inet_sock *inet = inet_sk(sk);
1208 struct rtable *rt = (struct rtable *)__sk_dst_check(sk, 0); 1210 struct rtable *rt = (struct rtable *)__sk_dst_check(sk, 0);
1209 __be32 daddr; 1211 __be32 daddr;
1210 struct ip_options_rcu *inet_opt; 1212 struct ip_options_rcu *inet_opt;
1211 struct flowi4 *fl4; 1213 struct flowi4 *fl4;
1212 int err; 1214 int err;
1213 1215
1214 /* Route is OK, nothing to do. */ 1216 /* Route is OK, nothing to do. */
1215 if (rt) 1217 if (rt)
1216 return 0; 1218 return 0;
1217 1219
1218 /* Reroute. */ 1220 /* Reroute. */
1219 rcu_read_lock(); 1221 rcu_read_lock();
1220 inet_opt = rcu_dereference(inet->inet_opt); 1222 inet_opt = rcu_dereference(inet->inet_opt);
1221 daddr = inet->inet_daddr; 1223 daddr = inet->inet_daddr;
1222 if (inet_opt && inet_opt->opt.srr) 1224 if (inet_opt && inet_opt->opt.srr)
1223 daddr = inet_opt->opt.faddr; 1225 daddr = inet_opt->opt.faddr;
1224 rcu_read_unlock(); 1226 rcu_read_unlock();
1225 fl4 = &inet->cork.fl.u.ip4; 1227 fl4 = &inet->cork.fl.u.ip4;
1226 rt = ip_route_output_ports(sock_net(sk), fl4, sk, daddr, inet->inet_saddr, 1228 rt = ip_route_output_ports(sock_net(sk), fl4, sk, daddr, inet->inet_saddr,
1227 inet->inet_dport, inet->inet_sport, 1229 inet->inet_dport, inet->inet_sport,
1228 sk->sk_protocol, RT_CONN_FLAGS(sk), 1230 sk->sk_protocol, RT_CONN_FLAGS(sk),
1229 sk->sk_bound_dev_if); 1231 sk->sk_bound_dev_if);
1230 if (!IS_ERR(rt)) { 1232 if (!IS_ERR(rt)) {
1231 err = 0; 1233 err = 0;
1232 sk_setup_caps(sk, &rt->dst); 1234 sk_setup_caps(sk, &rt->dst);
1233 } else { 1235 } else {
1234 err = PTR_ERR(rt); 1236 err = PTR_ERR(rt);
1235 1237
1236 /* Routing failed... */ 1238 /* Routing failed... */
1237 sk->sk_route_caps = 0; 1239 sk->sk_route_caps = 0;
1238 /* 1240 /*
1239 * Other protocols have to map its equivalent state to TCP_SYN_SENT. 1241 * Other protocols have to map its equivalent state to TCP_SYN_SENT.
1240 * DCCP maps its DCCP_REQUESTING state to TCP_SYN_SENT. -acme 1242 * DCCP maps its DCCP_REQUESTING state to TCP_SYN_SENT. -acme
1241 */ 1243 */
1242 if (!sysctl_ip_dynaddr || 1244 if (!sysctl_ip_dynaddr ||
1243 sk->sk_state != TCP_SYN_SENT || 1245 sk->sk_state != TCP_SYN_SENT ||
1244 (sk->sk_userlocks & SOCK_BINDADDR_LOCK) || 1246 (sk->sk_userlocks & SOCK_BINDADDR_LOCK) ||
1245 (err = inet_sk_reselect_saddr(sk)) != 0) 1247 (err = inet_sk_reselect_saddr(sk)) != 0)
1246 sk->sk_err_soft = -err; 1248 sk->sk_err_soft = -err;
1247 } 1249 }
1248 1250
1249 return err; 1251 return err;
1250 } 1252 }
1251 EXPORT_SYMBOL(inet_sk_rebuild_header); 1253 EXPORT_SYMBOL(inet_sk_rebuild_header);
1252 1254
1253 static int inet_gso_send_check(struct sk_buff *skb) 1255 static int inet_gso_send_check(struct sk_buff *skb)
1254 { 1256 {
1255 const struct net_offload *ops; 1257 const struct net_offload *ops;
1256 const struct iphdr *iph; 1258 const struct iphdr *iph;
1257 int proto; 1259 int proto;
1258 int ihl; 1260 int ihl;
1259 int err = -EINVAL; 1261 int err = -EINVAL;
1260 1262
1261 if (unlikely(!pskb_may_pull(skb, sizeof(*iph)))) 1263 if (unlikely(!pskb_may_pull(skb, sizeof(*iph))))
1262 goto out; 1264 goto out;
1263 1265
1264 iph = ip_hdr(skb); 1266 iph = ip_hdr(skb);
1265 ihl = iph->ihl * 4; 1267 ihl = iph->ihl * 4;
1266 if (ihl < sizeof(*iph)) 1268 if (ihl < sizeof(*iph))
1267 goto out; 1269 goto out;
1268 1270
1269 if (unlikely(!pskb_may_pull(skb, ihl))) 1271 if (unlikely(!pskb_may_pull(skb, ihl)))
1270 goto out; 1272 goto out;
1271 1273
1272 __skb_pull(skb, ihl); 1274 __skb_pull(skb, ihl);
1273 skb_reset_transport_header(skb); 1275 skb_reset_transport_header(skb);
1274 iph = ip_hdr(skb); 1276 iph = ip_hdr(skb);
1275 proto = iph->protocol; 1277 proto = iph->protocol;
1276 err = -EPROTONOSUPPORT; 1278 err = -EPROTONOSUPPORT;
1277 1279
1278 rcu_read_lock(); 1280 rcu_read_lock();
1279 ops = rcu_dereference(inet_offloads[proto]); 1281 ops = rcu_dereference(inet_offloads[proto]);
1280 if (likely(ops && ops->callbacks.gso_send_check)) 1282 if (likely(ops && ops->callbacks.gso_send_check))
1281 err = ops->callbacks.gso_send_check(skb); 1283 err = ops->callbacks.gso_send_check(skb);
1282 rcu_read_unlock(); 1284 rcu_read_unlock();
1283 1285
1284 out: 1286 out:
1285 return err; 1287 return err;
1286 } 1288 }
1287 1289
1288 static struct sk_buff *inet_gso_segment(struct sk_buff *skb, 1290 static struct sk_buff *inet_gso_segment(struct sk_buff *skb,
1289 netdev_features_t features) 1291 netdev_features_t features)
1290 { 1292 {
1291 struct sk_buff *segs = ERR_PTR(-EINVAL); 1293 struct sk_buff *segs = ERR_PTR(-EINVAL);
1292 const struct net_offload *ops; 1294 const struct net_offload *ops;
1293 struct iphdr *iph; 1295 struct iphdr *iph;
1294 int proto; 1296 int proto;
1295 int ihl; 1297 int ihl;
1296 int id; 1298 int id;
1297 unsigned int offset = 0; 1299 unsigned int offset = 0;
1298 1300
1299 if (!(features & NETIF_F_V4_CSUM)) 1301 if (!(features & NETIF_F_V4_CSUM))
1300 features &= ~NETIF_F_SG; 1302 features &= ~NETIF_F_SG;
1301 1303
1302 if (unlikely(skb_shinfo(skb)->gso_type & 1304 if (unlikely(skb_shinfo(skb)->gso_type &
1303 ~(SKB_GSO_TCPV4 | 1305 ~(SKB_GSO_TCPV4 |
1304 SKB_GSO_UDP | 1306 SKB_GSO_UDP |
1305 SKB_GSO_DODGY | 1307 SKB_GSO_DODGY |
1306 SKB_GSO_TCP_ECN | 1308 SKB_GSO_TCP_ECN |
1307 0))) 1309 0)))
1308 goto out; 1310 goto out;
1309 1311
1310 if (unlikely(!pskb_may_pull(skb, sizeof(*iph)))) 1312 if (unlikely(!pskb_may_pull(skb, sizeof(*iph))))
1311 goto out; 1313 goto out;
1312 1314
1313 iph = ip_hdr(skb); 1315 iph = ip_hdr(skb);
1314 ihl = iph->ihl * 4; 1316 ihl = iph->ihl * 4;
1315 if (ihl < sizeof(*iph)) 1317 if (ihl < sizeof(*iph))
1316 goto out; 1318 goto out;
1317 1319
1318 if (unlikely(!pskb_may_pull(skb, ihl))) 1320 if (unlikely(!pskb_may_pull(skb, ihl)))
1319 goto out; 1321 goto out;
1320 1322
1321 __skb_pull(skb, ihl); 1323 __skb_pull(skb, ihl);
1322 skb_reset_transport_header(skb); 1324 skb_reset_transport_header(skb);
1323 iph = ip_hdr(skb); 1325 iph = ip_hdr(skb);
1324 id = ntohs(iph->id); 1326 id = ntohs(iph->id);
1325 proto = iph->protocol; 1327 proto = iph->protocol;
1326 segs = ERR_PTR(-EPROTONOSUPPORT); 1328 segs = ERR_PTR(-EPROTONOSUPPORT);
1327 1329
1328 rcu_read_lock(); 1330 rcu_read_lock();
1329 ops = rcu_dereference(inet_offloads[proto]); 1331 ops = rcu_dereference(inet_offloads[proto]);
1330 if (likely(ops && ops->callbacks.gso_segment)) 1332 if (likely(ops && ops->callbacks.gso_segment))
1331 segs = ops->callbacks.gso_segment(skb, features); 1333 segs = ops->callbacks.gso_segment(skb, features);
1332 rcu_read_unlock(); 1334 rcu_read_unlock();
1333 1335
1334 if (!segs || IS_ERR(segs)) 1336 if (!segs || IS_ERR(segs))
1335 goto out; 1337 goto out;
1336 1338
1337 skb = segs; 1339 skb = segs;
1338 do { 1340 do {
1339 iph = ip_hdr(skb); 1341 iph = ip_hdr(skb);
1340 if (proto == IPPROTO_UDP) { 1342 if (proto == IPPROTO_UDP) {
1341 iph->id = htons(id); 1343 iph->id = htons(id);
1342 iph->frag_off = htons(offset >> 3); 1344 iph->frag_off = htons(offset >> 3);
1343 if (skb->next != NULL) 1345 if (skb->next != NULL)
1344 iph->frag_off |= htons(IP_MF); 1346 iph->frag_off |= htons(IP_MF);
1345 offset += (skb->len - skb->mac_len - iph->ihl * 4); 1347 offset += (skb->len - skb->mac_len - iph->ihl * 4);
1346 } else 1348 } else
1347 iph->id = htons(id++); 1349 iph->id = htons(id++);
1348 iph->tot_len = htons(skb->len - skb->mac_len); 1350 iph->tot_len = htons(skb->len - skb->mac_len);
1349 iph->check = 0; 1351 iph->check = 0;
1350 iph->check = ip_fast_csum(skb_network_header(skb), iph->ihl); 1352 iph->check = ip_fast_csum(skb_network_header(skb), iph->ihl);
1351 } while ((skb = skb->next)); 1353 } while ((skb = skb->next));
1352 1354
1353 out: 1355 out:
1354 return segs; 1356 return segs;
1355 } 1357 }
1356 1358
1357 static struct sk_buff **inet_gro_receive(struct sk_buff **head, 1359 static struct sk_buff **inet_gro_receive(struct sk_buff **head,
1358 struct sk_buff *skb) 1360 struct sk_buff *skb)
1359 { 1361 {
1360 const struct net_offload *ops; 1362 const struct net_offload *ops;
1361 struct sk_buff **pp = NULL; 1363 struct sk_buff **pp = NULL;
1362 struct sk_buff *p; 1364 struct sk_buff *p;
1363 const struct iphdr *iph; 1365 const struct iphdr *iph;
1364 unsigned int hlen; 1366 unsigned int hlen;
1365 unsigned int off; 1367 unsigned int off;
1366 unsigned int id; 1368 unsigned int id;
1367 int flush = 1; 1369 int flush = 1;
1368 int proto; 1370 int proto;
1369 1371
1370 off = skb_gro_offset(skb); 1372 off = skb_gro_offset(skb);
1371 hlen = off + sizeof(*iph); 1373 hlen = off + sizeof(*iph);
1372 iph = skb_gro_header_fast(skb, off); 1374 iph = skb_gro_header_fast(skb, off);
1373 if (skb_gro_header_hard(skb, hlen)) { 1375 if (skb_gro_header_hard(skb, hlen)) {
1374 iph = skb_gro_header_slow(skb, hlen, off); 1376 iph = skb_gro_header_slow(skb, hlen, off);
1375 if (unlikely(!iph)) 1377 if (unlikely(!iph))
1376 goto out; 1378 goto out;
1377 } 1379 }
1378 1380
1379 proto = iph->protocol; 1381 proto = iph->protocol;
1380 1382
1381 rcu_read_lock(); 1383 rcu_read_lock();
1382 ops = rcu_dereference(inet_offloads[proto]); 1384 ops = rcu_dereference(inet_offloads[proto]);
1383 if (!ops || !ops->callbacks.gro_receive) 1385 if (!ops || !ops->callbacks.gro_receive)
1384 goto out_unlock; 1386 goto out_unlock;
1385 1387
1386 if (*(u8 *)iph != 0x45) 1388 if (*(u8 *)iph != 0x45)
1387 goto out_unlock; 1389 goto out_unlock;
1388 1390
1389 if (unlikely(ip_fast_csum((u8 *)iph, 5))) 1391 if (unlikely(ip_fast_csum((u8 *)iph, 5)))
1390 goto out_unlock; 1392 goto out_unlock;
1391 1393
1392 id = ntohl(*(__be32 *)&iph->id); 1394 id = ntohl(*(__be32 *)&iph->id);
1393 flush = (u16)((ntohl(*(__be32 *)iph) ^ skb_gro_len(skb)) | (id ^ IP_DF)); 1395 flush = (u16)((ntohl(*(__be32 *)iph) ^ skb_gro_len(skb)) | (id ^ IP_DF));
1394 id >>= 16; 1396 id >>= 16;
1395 1397
1396 for (p = *head; p; p = p->next) { 1398 for (p = *head; p; p = p->next) {
1397 struct iphdr *iph2; 1399 struct iphdr *iph2;
1398 1400
1399 if (!NAPI_GRO_CB(p)->same_flow) 1401 if (!NAPI_GRO_CB(p)->same_flow)
1400 continue; 1402 continue;
1401 1403
1402 iph2 = ip_hdr(p); 1404 iph2 = ip_hdr(p);
1403 1405
1404 if ((iph->protocol ^ iph2->protocol) | 1406 if ((iph->protocol ^ iph2->protocol) |
1405 ((__force u32)iph->saddr ^ (__force u32)iph2->saddr) | 1407 ((__force u32)iph->saddr ^ (__force u32)iph2->saddr) |
1406 ((__force u32)iph->daddr ^ (__force u32)iph2->daddr)) { 1408 ((__force u32)iph->daddr ^ (__force u32)iph2->daddr)) {
1407 NAPI_GRO_CB(p)->same_flow = 0; 1409 NAPI_GRO_CB(p)->same_flow = 0;
1408 continue; 1410 continue;
1409 } 1411 }
1410 1412
1411 /* All fields must match except length and checksum. */ 1413 /* All fields must match except length and checksum. */
1412 NAPI_GRO_CB(p)->flush |= 1414 NAPI_GRO_CB(p)->flush |=
1413 (iph->ttl ^ iph2->ttl) | 1415 (iph->ttl ^ iph2->ttl) |
1414 (iph->tos ^ iph2->tos) | 1416 (iph->tos ^ iph2->tos) |
1415 ((u16)(ntohs(iph2->id) + NAPI_GRO_CB(p)->count) ^ id); 1417 ((u16)(ntohs(iph2->id) + NAPI_GRO_CB(p)->count) ^ id);
1416 1418
1417 NAPI_GRO_CB(p)->flush |= flush; 1419 NAPI_GRO_CB(p)->flush |= flush;
1418 } 1420 }
1419 1421
1420 NAPI_GRO_CB(skb)->flush |= flush; 1422 NAPI_GRO_CB(skb)->flush |= flush;
1421 skb_gro_pull(skb, sizeof(*iph)); 1423 skb_gro_pull(skb, sizeof(*iph));
1422 skb_set_transport_header(skb, skb_gro_offset(skb)); 1424 skb_set_transport_header(skb, skb_gro_offset(skb));
1423 1425
1424 pp = ops->callbacks.gro_receive(head, skb); 1426 pp = ops->callbacks.gro_receive(head, skb);
1425 1427
1426 out_unlock: 1428 out_unlock:
1427 rcu_read_unlock(); 1429 rcu_read_unlock();
1428 1430
1429 out: 1431 out:
1430 NAPI_GRO_CB(skb)->flush |= flush; 1432 NAPI_GRO_CB(skb)->flush |= flush;
1431 1433
1432 return pp; 1434 return pp;
1433 } 1435 }
1434 1436
1435 static int inet_gro_complete(struct sk_buff *skb) 1437 static int inet_gro_complete(struct sk_buff *skb)
1436 { 1438 {
1437 __be16 newlen = htons(skb->len - skb_network_offset(skb)); 1439 __be16 newlen = htons(skb->len - skb_network_offset(skb));
1438 struct iphdr *iph = ip_hdr(skb); 1440 struct iphdr *iph = ip_hdr(skb);
1439 const struct net_offload *ops; 1441 const struct net_offload *ops;
1440 int proto = iph->protocol; 1442 int proto = iph->protocol;
1441 int err = -ENOSYS; 1443 int err = -ENOSYS;
1442 1444
1443 csum_replace2(&iph->check, iph->tot_len, newlen); 1445 csum_replace2(&iph->check, iph->tot_len, newlen);
1444 iph->tot_len = newlen; 1446 iph->tot_len = newlen;
1445 1447
1446 rcu_read_lock(); 1448 rcu_read_lock();
1447 ops = rcu_dereference(inet_offloads[proto]); 1449 ops = rcu_dereference(inet_offloads[proto]);
1448 if (WARN_ON(!ops || !ops->callbacks.gro_complete)) 1450 if (WARN_ON(!ops || !ops->callbacks.gro_complete))
1449 goto out_unlock; 1451 goto out_unlock;
1450 1452
1451 err = ops->callbacks.gro_complete(skb); 1453 err = ops->callbacks.gro_complete(skb);
1452 1454
1453 out_unlock: 1455 out_unlock:
1454 rcu_read_unlock(); 1456 rcu_read_unlock();
1455 1457
1456 return err; 1458 return err;
1457 } 1459 }
1458 1460
1459 int inet_ctl_sock_create(struct sock **sk, unsigned short family, 1461 int inet_ctl_sock_create(struct sock **sk, unsigned short family,
1460 unsigned short type, unsigned char protocol, 1462 unsigned short type, unsigned char protocol,
1461 struct net *net) 1463 struct net *net)
1462 { 1464 {
1463 struct socket *sock; 1465 struct socket *sock;
1464 int rc = sock_create_kern(family, type, protocol, &sock); 1466 int rc = sock_create_kern(family, type, protocol, &sock);
1465 1467
1466 if (rc == 0) { 1468 if (rc == 0) {
1467 *sk = sock->sk; 1469 *sk = sock->sk;
1468 (*sk)->sk_allocation = GFP_ATOMIC; 1470 (*sk)->sk_allocation = GFP_ATOMIC;
1469 /* 1471 /*
1470 * Unhash it so that IP input processing does not even see it, 1472 * Unhash it so that IP input processing does not even see it,
1471 * we do not wish this socket to see incoming packets. 1473 * we do not wish this socket to see incoming packets.
1472 */ 1474 */
1473 (*sk)->sk_prot->unhash(*sk); 1475 (*sk)->sk_prot->unhash(*sk);
1474 1476
1475 sk_change_net(*sk, net); 1477 sk_change_net(*sk, net);
1476 } 1478 }
1477 return rc; 1479 return rc;
1478 } 1480 }
1479 EXPORT_SYMBOL_GPL(inet_ctl_sock_create); 1481 EXPORT_SYMBOL_GPL(inet_ctl_sock_create);
1480 1482
1481 unsigned long snmp_fold_field(void __percpu *mib[], int offt) 1483 unsigned long snmp_fold_field(void __percpu *mib[], int offt)
1482 { 1484 {
1483 unsigned long res = 0; 1485 unsigned long res = 0;
1484 int i, j; 1486 int i, j;
1485 1487
1486 for_each_possible_cpu(i) { 1488 for_each_possible_cpu(i) {
1487 for (j = 0; j < SNMP_ARRAY_SZ; j++) 1489 for (j = 0; j < SNMP_ARRAY_SZ; j++)
1488 res += *(((unsigned long *) per_cpu_ptr(mib[j], i)) + offt); 1490 res += *(((unsigned long *) per_cpu_ptr(mib[j], i)) + offt);
1489 } 1491 }
1490 return res; 1492 return res;
1491 } 1493 }
1492 EXPORT_SYMBOL_GPL(snmp_fold_field); 1494 EXPORT_SYMBOL_GPL(snmp_fold_field);
1493 1495
1494 #if BITS_PER_LONG==32 1496 #if BITS_PER_LONG==32
1495 1497
1496 u64 snmp_fold_field64(void __percpu *mib[], int offt, size_t syncp_offset) 1498 u64 snmp_fold_field64(void __percpu *mib[], int offt, size_t syncp_offset)
1497 { 1499 {
1498 u64 res = 0; 1500 u64 res = 0;
1499 int cpu; 1501 int cpu;
1500 1502
1501 for_each_possible_cpu(cpu) { 1503 for_each_possible_cpu(cpu) {
1502 void *bhptr; 1504 void *bhptr;
1503 struct u64_stats_sync *syncp; 1505 struct u64_stats_sync *syncp;
1504 u64 v; 1506 u64 v;
1505 unsigned int start; 1507 unsigned int start;
1506 1508
1507 bhptr = per_cpu_ptr(mib[0], cpu); 1509 bhptr = per_cpu_ptr(mib[0], cpu);
1508 syncp = (struct u64_stats_sync *)(bhptr + syncp_offset); 1510 syncp = (struct u64_stats_sync *)(bhptr + syncp_offset);
1509 do { 1511 do {
1510 start = u64_stats_fetch_begin_bh(syncp); 1512 start = u64_stats_fetch_begin_bh(syncp);
1511 v = *(((u64 *) bhptr) + offt); 1513 v = *(((u64 *) bhptr) + offt);
1512 } while (u64_stats_fetch_retry_bh(syncp, start)); 1514 } while (u64_stats_fetch_retry_bh(syncp, start));
1513 1515
1514 res += v; 1516 res += v;
1515 } 1517 }
1516 return res; 1518 return res;
1517 } 1519 }
1518 EXPORT_SYMBOL_GPL(snmp_fold_field64); 1520 EXPORT_SYMBOL_GPL(snmp_fold_field64);
1519 #endif 1521 #endif
1520 1522
1521 int snmp_mib_init(void __percpu *ptr[2], size_t mibsize, size_t align) 1523 int snmp_mib_init(void __percpu *ptr[2], size_t mibsize, size_t align)
1522 { 1524 {
1523 BUG_ON(ptr == NULL); 1525 BUG_ON(ptr == NULL);
1524 ptr[0] = __alloc_percpu(mibsize, align); 1526 ptr[0] = __alloc_percpu(mibsize, align);
1525 if (!ptr[0]) 1527 if (!ptr[0])
1526 return -ENOMEM; 1528 return -ENOMEM;
1527 #if SNMP_ARRAY_SZ == 2 1529 #if SNMP_ARRAY_SZ == 2
1528 ptr[1] = __alloc_percpu(mibsize, align); 1530 ptr[1] = __alloc_percpu(mibsize, align);
1529 if (!ptr[1]) { 1531 if (!ptr[1]) {
1530 free_percpu(ptr[0]); 1532 free_percpu(ptr[0]);
1531 ptr[0] = NULL; 1533 ptr[0] = NULL;
1532 return -ENOMEM; 1534 return -ENOMEM;
1533 } 1535 }
1534 #endif 1536 #endif
1535 return 0; 1537 return 0;
1536 } 1538 }
1537 EXPORT_SYMBOL_GPL(snmp_mib_init); 1539 EXPORT_SYMBOL_GPL(snmp_mib_init);
1538 1540
1539 void snmp_mib_free(void __percpu *ptr[SNMP_ARRAY_SZ]) 1541 void snmp_mib_free(void __percpu *ptr[SNMP_ARRAY_SZ])
1540 { 1542 {
1541 int i; 1543 int i;
1542 1544
1543 BUG_ON(ptr == NULL); 1545 BUG_ON(ptr == NULL);
1544 for (i = 0; i < SNMP_ARRAY_SZ; i++) { 1546 for (i = 0; i < SNMP_ARRAY_SZ; i++) {
1545 free_percpu(ptr[i]); 1547 free_percpu(ptr[i]);
1546 ptr[i] = NULL; 1548 ptr[i] = NULL;
1547 } 1549 }
1548 } 1550 }
1549 EXPORT_SYMBOL_GPL(snmp_mib_free); 1551 EXPORT_SYMBOL_GPL(snmp_mib_free);
1550 1552
1551 #ifdef CONFIG_IP_MULTICAST 1553 #ifdef CONFIG_IP_MULTICAST
1552 static const struct net_protocol igmp_protocol = { 1554 static const struct net_protocol igmp_protocol = {
1553 .handler = igmp_rcv, 1555 .handler = igmp_rcv,
1554 .netns_ok = 1, 1556 .netns_ok = 1,
1555 }; 1557 };
1556 #endif 1558 #endif
1557 1559
1558 static const struct net_protocol tcp_protocol = { 1560 static const struct net_protocol tcp_protocol = {
1559 .early_demux = tcp_v4_early_demux, 1561 .early_demux = tcp_v4_early_demux,
1560 .handler = tcp_v4_rcv, 1562 .handler = tcp_v4_rcv,
1561 .err_handler = tcp_v4_err, 1563 .err_handler = tcp_v4_err,
1562 .no_policy = 1, 1564 .no_policy = 1,
1563 .netns_ok = 1, 1565 .netns_ok = 1,
1564 }; 1566 };
1565 1567
1566 static const struct net_offload tcp_offload = { 1568 static const struct net_offload tcp_offload = {
1567 .callbacks = { 1569 .callbacks = {
1568 .gso_send_check = tcp_v4_gso_send_check, 1570 .gso_send_check = tcp_v4_gso_send_check,
1569 .gso_segment = tcp_tso_segment, 1571 .gso_segment = tcp_tso_segment,
1570 .gro_receive = tcp4_gro_receive, 1572 .gro_receive = tcp4_gro_receive,
1571 .gro_complete = tcp4_gro_complete, 1573 .gro_complete = tcp4_gro_complete,
1572 }, 1574 },
1573 }; 1575 };
1574 1576
1575 static const struct net_protocol udp_protocol = { 1577 static const struct net_protocol udp_protocol = {
1576 .handler = udp_rcv, 1578 .handler = udp_rcv,
1577 .err_handler = udp_err, 1579 .err_handler = udp_err,
1578 .no_policy = 1, 1580 .no_policy = 1,
1579 .netns_ok = 1, 1581 .netns_ok = 1,
1580 }; 1582 };
1581 1583
1582 static const struct net_offload udp_offload = { 1584 static const struct net_offload udp_offload = {
1583 .callbacks = { 1585 .callbacks = {
1584 .gso_send_check = udp4_ufo_send_check, 1586 .gso_send_check = udp4_ufo_send_check,
1585 .gso_segment = udp4_ufo_fragment, 1587 .gso_segment = udp4_ufo_fragment,
1586 }, 1588 },
1587 }; 1589 };
1588 1590
1589 static const struct net_protocol icmp_protocol = { 1591 static const struct net_protocol icmp_protocol = {
1590 .handler = icmp_rcv, 1592 .handler = icmp_rcv,
1591 .err_handler = ping_err, 1593 .err_handler = ping_err,
1592 .no_policy = 1, 1594 .no_policy = 1,
1593 .netns_ok = 1, 1595 .netns_ok = 1,
1594 }; 1596 };
1595 1597
1596 static __net_init int ipv4_mib_init_net(struct net *net) 1598 static __net_init int ipv4_mib_init_net(struct net *net)
1597 { 1599 {
1598 if (snmp_mib_init((void __percpu **)net->mib.tcp_statistics, 1600 if (snmp_mib_init((void __percpu **)net->mib.tcp_statistics,
1599 sizeof(struct tcp_mib), 1601 sizeof(struct tcp_mib),
1600 __alignof__(struct tcp_mib)) < 0) 1602 __alignof__(struct tcp_mib)) < 0)
1601 goto err_tcp_mib; 1603 goto err_tcp_mib;
1602 if (snmp_mib_init((void __percpu **)net->mib.ip_statistics, 1604 if (snmp_mib_init((void __percpu **)net->mib.ip_statistics,
1603 sizeof(struct ipstats_mib), 1605 sizeof(struct ipstats_mib),
1604 __alignof__(struct ipstats_mib)) < 0) 1606 __alignof__(struct ipstats_mib)) < 0)
1605 goto err_ip_mib; 1607 goto err_ip_mib;
1606 if (snmp_mib_init((void __percpu **)net->mib.net_statistics, 1608 if (snmp_mib_init((void __percpu **)net->mib.net_statistics,
1607 sizeof(struct linux_mib), 1609 sizeof(struct linux_mib),
1608 __alignof__(struct linux_mib)) < 0) 1610 __alignof__(struct linux_mib)) < 0)
1609 goto err_net_mib; 1611 goto err_net_mib;
1610 if (snmp_mib_init((void __percpu **)net->mib.udp_statistics, 1612 if (snmp_mib_init((void __percpu **)net->mib.udp_statistics,
1611 sizeof(struct udp_mib), 1613 sizeof(struct udp_mib),
1612 __alignof__(struct udp_mib)) < 0) 1614 __alignof__(struct udp_mib)) < 0)
1613 goto err_udp_mib; 1615 goto err_udp_mib;
1614 if (snmp_mib_init((void __percpu **)net->mib.udplite_statistics, 1616 if (snmp_mib_init((void __percpu **)net->mib.udplite_statistics,
1615 sizeof(struct udp_mib), 1617 sizeof(struct udp_mib),
1616 __alignof__(struct udp_mib)) < 0) 1618 __alignof__(struct udp_mib)) < 0)
1617 goto err_udplite_mib; 1619 goto err_udplite_mib;
1618 if (snmp_mib_init((void __percpu **)net->mib.icmp_statistics, 1620 if (snmp_mib_init((void __percpu **)net->mib.icmp_statistics,
1619 sizeof(struct icmp_mib), 1621 sizeof(struct icmp_mib),
1620 __alignof__(struct icmp_mib)) < 0) 1622 __alignof__(struct icmp_mib)) < 0)
1621 goto err_icmp_mib; 1623 goto err_icmp_mib;
1622 net->mib.icmpmsg_statistics = kzalloc(sizeof(struct icmpmsg_mib), 1624 net->mib.icmpmsg_statistics = kzalloc(sizeof(struct icmpmsg_mib),
1623 GFP_KERNEL); 1625 GFP_KERNEL);
1624 if (!net->mib.icmpmsg_statistics) 1626 if (!net->mib.icmpmsg_statistics)
1625 goto err_icmpmsg_mib; 1627 goto err_icmpmsg_mib;
1626 1628
1627 tcp_mib_init(net); 1629 tcp_mib_init(net);
1628 return 0; 1630 return 0;
1629 1631
1630 err_icmpmsg_mib: 1632 err_icmpmsg_mib:
1631 snmp_mib_free((void __percpu **)net->mib.icmp_statistics); 1633 snmp_mib_free((void __percpu **)net->mib.icmp_statistics);
1632 err_icmp_mib: 1634 err_icmp_mib:
1633 snmp_mib_free((void __percpu **)net->mib.udplite_statistics); 1635 snmp_mib_free((void __percpu **)net->mib.udplite_statistics);
1634 err_udplite_mib: 1636 err_udplite_mib:
1635 snmp_mib_free((void __percpu **)net->mib.udp_statistics); 1637 snmp_mib_free((void __percpu **)net->mib.udp_statistics);
1636 err_udp_mib: 1638 err_udp_mib:
1637 snmp_mib_free((void __percpu **)net->mib.net_statistics); 1639 snmp_mib_free((void __percpu **)net->mib.net_statistics);
1638 err_net_mib: 1640 err_net_mib:
1639 snmp_mib_free((void __percpu **)net->mib.ip_statistics); 1641 snmp_mib_free((void __percpu **)net->mib.ip_statistics);
1640 err_ip_mib: 1642 err_ip_mib:
1641 snmp_mib_free((void __percpu **)net->mib.tcp_statistics); 1643 snmp_mib_free((void __percpu **)net->mib.tcp_statistics);
1642 err_tcp_mib: 1644 err_tcp_mib:
1643 return -ENOMEM; 1645 return -ENOMEM;
1644 } 1646 }
1645 1647
1646 static __net_exit void ipv4_mib_exit_net(struct net *net) 1648 static __net_exit void ipv4_mib_exit_net(struct net *net)
1647 { 1649 {
1648 kfree(net->mib.icmpmsg_statistics); 1650 kfree(net->mib.icmpmsg_statistics);
1649 snmp_mib_free((void __percpu **)net->mib.icmp_statistics); 1651 snmp_mib_free((void __percpu **)net->mib.icmp_statistics);
1650 snmp_mib_free((void __percpu **)net->mib.udplite_statistics); 1652 snmp_mib_free((void __percpu **)net->mib.udplite_statistics);
1651 snmp_mib_free((void __percpu **)net->mib.udp_statistics); 1653 snmp_mib_free((void __percpu **)net->mib.udp_statistics);
1652 snmp_mib_free((void __percpu **)net->mib.net_statistics); 1654 snmp_mib_free((void __percpu **)net->mib.net_statistics);
1653 snmp_mib_free((void __percpu **)net->mib.ip_statistics); 1655 snmp_mib_free((void __percpu **)net->mib.ip_statistics);
1654 snmp_mib_free((void __percpu **)net->mib.tcp_statistics); 1656 snmp_mib_free((void __percpu **)net->mib.tcp_statistics);
1655 } 1657 }
1656 1658
1657 static __net_initdata struct pernet_operations ipv4_mib_ops = { 1659 static __net_initdata struct pernet_operations ipv4_mib_ops = {
1658 .init = ipv4_mib_init_net, 1660 .init = ipv4_mib_init_net,
1659 .exit = ipv4_mib_exit_net, 1661 .exit = ipv4_mib_exit_net,
1660 }; 1662 };
1661 1663
1662 static int __init init_ipv4_mibs(void) 1664 static int __init init_ipv4_mibs(void)
1663 { 1665 {
1664 return register_pernet_subsys(&ipv4_mib_ops); 1666 return register_pernet_subsys(&ipv4_mib_ops);
1665 } 1667 }
1666 1668
1667 static int ipv4_proc_init(void); 1669 static int ipv4_proc_init(void);
1668 1670
1669 /* 1671 /*
1670 * IP protocol layer initialiser 1672 * IP protocol layer initialiser
1671 */ 1673 */
1672 1674
1673 static struct packet_offload ip_packet_offload __read_mostly = { 1675 static struct packet_offload ip_packet_offload __read_mostly = {
1674 .type = cpu_to_be16(ETH_P_IP), 1676 .type = cpu_to_be16(ETH_P_IP),
1675 .callbacks = { 1677 .callbacks = {
1676 .gso_send_check = inet_gso_send_check, 1678 .gso_send_check = inet_gso_send_check,
1677 .gso_segment = inet_gso_segment, 1679 .gso_segment = inet_gso_segment,
1678 .gro_receive = inet_gro_receive, 1680 .gro_receive = inet_gro_receive,
1679 .gro_complete = inet_gro_complete, 1681 .gro_complete = inet_gro_complete,
1680 }, 1682 },
1681 }; 1683 };
1682 1684
1683 static int __init ipv4_offload_init(void) 1685 static int __init ipv4_offload_init(void)
1684 { 1686 {
1685 /* 1687 /*
1686 * Add offloads 1688 * Add offloads
1687 */ 1689 */
1688 if (inet_add_offload(&udp_offload, IPPROTO_UDP) < 0) 1690 if (inet_add_offload(&udp_offload, IPPROTO_UDP) < 0)
1689 pr_crit("%s: Cannot add UDP protocol offload\n", __func__); 1691 pr_crit("%s: Cannot add UDP protocol offload\n", __func__);
1690 if (inet_add_offload(&tcp_offload, IPPROTO_TCP) < 0) 1692 if (inet_add_offload(&tcp_offload, IPPROTO_TCP) < 0)
1691 pr_crit("%s: Cannot add TCP protocol offlaod\n", __func__); 1693 pr_crit("%s: Cannot add TCP protocol offlaod\n", __func__);
1692 1694
1693 dev_add_offload(&ip_packet_offload); 1695 dev_add_offload(&ip_packet_offload);
1694 return 0; 1696 return 0;
1695 } 1697 }
1696 1698
1697 fs_initcall(ipv4_offload_init); 1699 fs_initcall(ipv4_offload_init);
1698 1700
1699 static struct packet_type ip_packet_type __read_mostly = { 1701 static struct packet_type ip_packet_type __read_mostly = {
1700 .type = cpu_to_be16(ETH_P_IP), 1702 .type = cpu_to_be16(ETH_P_IP),
1701 .func = ip_rcv, 1703 .func = ip_rcv,
1702 }; 1704 };
1703 1705
1704 static int __init inet_init(void) 1706 static int __init inet_init(void)
1705 { 1707 {
1706 struct sk_buff *dummy_skb; 1708 struct sk_buff *dummy_skb;
1707 struct inet_protosw *q; 1709 struct inet_protosw *q;
1708 struct list_head *r; 1710 struct list_head *r;
1709 int rc = -EINVAL; 1711 int rc = -EINVAL;
1710 1712
1711 BUILD_BUG_ON(sizeof(struct inet_skb_parm) > sizeof(dummy_skb->cb)); 1713 BUILD_BUG_ON(sizeof(struct inet_skb_parm) > sizeof(dummy_skb->cb));
1712 1714
1713 sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL); 1715 sysctl_local_reserved_ports = kzalloc(65536 / 8, GFP_KERNEL);
1714 if (!sysctl_local_reserved_ports) 1716 if (!sysctl_local_reserved_ports)
1715 goto out; 1717 goto out;
1716 1718
1717 rc = proto_register(&tcp_prot, 1); 1719 rc = proto_register(&tcp_prot, 1);
1718 if (rc) 1720 if (rc)
1719 goto out_free_reserved_ports; 1721 goto out_free_reserved_ports;
1720 1722
1721 rc = proto_register(&udp_prot, 1); 1723 rc = proto_register(&udp_prot, 1);
1722 if (rc) 1724 if (rc)
1723 goto out_unregister_tcp_proto; 1725 goto out_unregister_tcp_proto;
1724 1726
1725 rc = proto_register(&raw_prot, 1); 1727 rc = proto_register(&raw_prot, 1);
1726 if (rc) 1728 if (rc)
1727 goto out_unregister_udp_proto; 1729 goto out_unregister_udp_proto;
1728 1730
1729 rc = proto_register(&ping_prot, 1); 1731 rc = proto_register(&ping_prot, 1);
1730 if (rc) 1732 if (rc)
1731 goto out_unregister_raw_proto; 1733 goto out_unregister_raw_proto;
1732 1734
1733 /* 1735 /*
1734 * Tell SOCKET that we are alive... 1736 * Tell SOCKET that we are alive...
1735 */ 1737 */
1736 1738
1737 (void)sock_register(&inet_family_ops); 1739 (void)sock_register(&inet_family_ops);
1738 1740
1739 #ifdef CONFIG_SYSCTL 1741 #ifdef CONFIG_SYSCTL
1740 ip_static_sysctl_init(); 1742 ip_static_sysctl_init();
1741 #endif 1743 #endif
1742 1744
1743 tcp_prot.sysctl_mem = init_net.ipv4.sysctl_tcp_mem; 1745 tcp_prot.sysctl_mem = init_net.ipv4.sysctl_tcp_mem;
1744 1746
1745 /* 1747 /*
1746 * Add all the base protocols. 1748 * Add all the base protocols.
1747 */ 1749 */
1748 1750
1749 if (inet_add_protocol(&icmp_protocol, IPPROTO_ICMP) < 0) 1751 if (inet_add_protocol(&icmp_protocol, IPPROTO_ICMP) < 0)
1750 pr_crit("%s: Cannot add ICMP protocol\n", __func__); 1752 pr_crit("%s: Cannot add ICMP protocol\n", __func__);
1751 if (inet_add_protocol(&udp_protocol, IPPROTO_UDP) < 0) 1753 if (inet_add_protocol(&udp_protocol, IPPROTO_UDP) < 0)
1752 pr_crit("%s: Cannot add UDP protocol\n", __func__); 1754 pr_crit("%s: Cannot add UDP protocol\n", __func__);
1753 if (inet_add_protocol(&tcp_protocol, IPPROTO_TCP) < 0) 1755 if (inet_add_protocol(&tcp_protocol, IPPROTO_TCP) < 0)
1754 pr_crit("%s: Cannot add TCP protocol\n", __func__); 1756 pr_crit("%s: Cannot add TCP protocol\n", __func__);
1755 #ifdef CONFIG_IP_MULTICAST 1757 #ifdef CONFIG_IP_MULTICAST
1756 if (inet_add_protocol(&igmp_protocol, IPPROTO_IGMP) < 0) 1758 if (inet_add_protocol(&igmp_protocol, IPPROTO_IGMP) < 0)
1757 pr_crit("%s: Cannot add IGMP protocol\n", __func__); 1759 pr_crit("%s: Cannot add IGMP protocol\n", __func__);
1758 #endif 1760 #endif
1759 1761
1760 /* Register the socket-side information for inet_create. */ 1762 /* Register the socket-side information for inet_create. */
1761 for (r = &inetsw[0]; r < &inetsw[SOCK_MAX]; ++r) 1763 for (r = &inetsw[0]; r < &inetsw[SOCK_MAX]; ++r)
1762 INIT_LIST_HEAD(r); 1764 INIT_LIST_HEAD(r);
1763 1765
1764 for (q = inetsw_array; q < &inetsw_array[INETSW_ARRAY_LEN]; ++q) 1766 for (q = inetsw_array; q < &inetsw_array[INETSW_ARRAY_LEN]; ++q)
1765 inet_register_protosw(q); 1767 inet_register_protosw(q);
1766 1768
1767 /* 1769 /*
1768 * Set the ARP module up 1770 * Set the ARP module up
1769 */ 1771 */
1770 1772
1771 arp_init(); 1773 arp_init();
1772 1774
1773 /* 1775 /*
1774 * Set the IP module up 1776 * Set the IP module up
1775 */ 1777 */
1776 1778
1777 ip_init(); 1779 ip_init();
1778 1780
1779 tcp_v4_init(); 1781 tcp_v4_init();
1780 1782
1781 /* Setup TCP slab cache for open requests. */ 1783 /* Setup TCP slab cache for open requests. */
1782 tcp_init(); 1784 tcp_init();
1783 1785
1784 /* Setup UDP memory threshold */ 1786 /* Setup UDP memory threshold */
1785 udp_init(); 1787 udp_init();
1786 1788
1787 /* Add UDP-Lite (RFC 3828) */ 1789 /* Add UDP-Lite (RFC 3828) */
1788 udplite4_register(); 1790 udplite4_register();
1789 1791
1790 ping_init(); 1792 ping_init();
1791 1793
1792 /* 1794 /*
1793 * Set the ICMP layer up 1795 * Set the ICMP layer up
1794 */ 1796 */
1795 1797
1796 if (icmp_init() < 0) 1798 if (icmp_init() < 0)
1797 panic("Failed to create the ICMP control socket.\n"); 1799 panic("Failed to create the ICMP control socket.\n");
1798 1800
1799 /* 1801 /*
1800 * Initialise the multicast router 1802 * Initialise the multicast router
1801 */ 1803 */
1802 #if defined(CONFIG_IP_MROUTE) 1804 #if defined(CONFIG_IP_MROUTE)
1803 if (ip_mr_init()) 1805 if (ip_mr_init())
1804 pr_crit("%s: Cannot init ipv4 mroute\n", __func__); 1806 pr_crit("%s: Cannot init ipv4 mroute\n", __func__);
1805 #endif 1807 #endif
1806 /* 1808 /*
1807 * Initialise per-cpu ipv4 mibs 1809 * Initialise per-cpu ipv4 mibs
1808 */ 1810 */
1809 1811
1810 if (init_ipv4_mibs()) 1812 if (init_ipv4_mibs())
1811 pr_crit("%s: Cannot init ipv4 mibs\n", __func__); 1813 pr_crit("%s: Cannot init ipv4 mibs\n", __func__);
1812 1814
1813 ipv4_proc_init(); 1815 ipv4_proc_init();
1814 1816
1815 ipfrag_init(); 1817 ipfrag_init();
1816 1818
1817 dev_add_pack(&ip_packet_type); 1819 dev_add_pack(&ip_packet_type);
1818 1820
1819 rc = 0; 1821 rc = 0;
1820 out: 1822 out:
1821 return rc; 1823 return rc;
1822 out_unregister_raw_proto: 1824 out_unregister_raw_proto:
1823 proto_unregister(&raw_prot); 1825 proto_unregister(&raw_prot);
1824 out_unregister_udp_proto: 1826 out_unregister_udp_proto:
1825 proto_unregister(&udp_prot); 1827 proto_unregister(&udp_prot);
1826 out_unregister_tcp_proto: 1828 out_unregister_tcp_proto:
1827 proto_unregister(&tcp_prot); 1829 proto_unregister(&tcp_prot);
1828 out_free_reserved_ports: 1830 out_free_reserved_ports:
1829 kfree(sysctl_local_reserved_ports); 1831 kfree(sysctl_local_reserved_ports);
1830 goto out; 1832 goto out;
1831 } 1833 }
1832 1834
1833 fs_initcall(inet_init); 1835 fs_initcall(inet_init);
1834 1836
1835 /* ------------------------------------------------------------------------ */ 1837 /* ------------------------------------------------------------------------ */
1836 1838
1837 #ifdef CONFIG_PROC_FS 1839 #ifdef CONFIG_PROC_FS
1838 static int __init ipv4_proc_init(void) 1840 static int __init ipv4_proc_init(void)
1839 { 1841 {
1840 int rc = 0; 1842 int rc = 0;
1841 1843
1842 if (raw_proc_init()) 1844 if (raw_proc_init())
1843 goto out_raw; 1845 goto out_raw;
1844 if (tcp4_proc_init()) 1846 if (tcp4_proc_init())
1845 goto out_tcp; 1847 goto out_tcp;
1846 if (udp4_proc_init()) 1848 if (udp4_proc_init())
1847 goto out_udp; 1849 goto out_udp;
1848 if (ping_proc_init()) 1850 if (ping_proc_init())
1849 goto out_ping; 1851 goto out_ping;
1850 if (ip_misc_proc_init()) 1852 if (ip_misc_proc_init())
1851 goto out_misc; 1853 goto out_misc;
1852 out: 1854 out:
1853 return rc; 1855 return rc;
1854 out_misc: 1856 out_misc:
1855 ping_proc_exit(); 1857 ping_proc_exit();
1856 out_ping: 1858 out_ping:
1857 udp4_proc_exit(); 1859 udp4_proc_exit();
1858 out_udp: 1860 out_udp:
1859 tcp4_proc_exit(); 1861 tcp4_proc_exit();
1860 out_tcp: 1862 out_tcp:
1861 raw_proc_exit(); 1863 raw_proc_exit();
1862 out_raw: 1864 out_raw:
1863 rc = -ENOMEM; 1865 rc = -ENOMEM;
1864 goto out; 1866 goto out;
1865 } 1867 }
1866 1868
1867 #else /* CONFIG_PROC_FS */ 1869 #else /* CONFIG_PROC_FS */
1868 static int __init ipv4_proc_init(void) 1870 static int __init ipv4_proc_init(void)
1869 { 1871 {
1870 return 0; 1872 return 0;
1871 } 1873 }
1872 #endif /* CONFIG_PROC_FS */ 1874 #endif /* CONFIG_PROC_FS */
1873 1875
1874 MODULE_ALIAS_NETPROTO(PF_INET); 1876 MODULE_ALIAS_NETPROTO(PF_INET);
1875 1877
1876 1878
1 /* 1 /*
2 * PF_INET6 socket protocol family 2 * PF_INET6 socket protocol family
3 * Linux INET6 implementation 3 * Linux INET6 implementation
4 * 4 *
5 * Authors: 5 * Authors:
6 * Pedro Roque <roque@di.fc.ul.pt> 6 * Pedro Roque <roque@di.fc.ul.pt>
7 * 7 *
8 * Adapted from linux/net/ipv4/af_inet.c 8 * Adapted from linux/net/ipv4/af_inet.c
9 * 9 *
10 * Fixes: 10 * Fixes:
11 * piggy, Karl Knutson : Socket protocol table 11 * piggy, Karl Knutson : Socket protocol table
12 * Hideaki YOSHIFUJI : sin6_scope_id support 12 * Hideaki YOSHIFUJI : sin6_scope_id support
13 * Arnaldo Melo : check proc_net_create return, cleanups 13 * Arnaldo Melo : check proc_net_create return, cleanups
14 * 14 *
15 * This program is free software; you can redistribute it and/or 15 * This program is free software; you can redistribute it and/or
16 * modify it under the terms of the GNU General Public License 16 * modify it under the terms of the GNU General Public License
17 * as published by the Free Software Foundation; either version 17 * as published by the Free Software Foundation; either version
18 * 2 of the License, or (at your option) any later version. 18 * 2 of the License, or (at your option) any later version.
19 */ 19 */
20 20
21 #define pr_fmt(fmt) "IPv6: " fmt 21 #define pr_fmt(fmt) "IPv6: " fmt
22 22
23 #include <linux/module.h> 23 #include <linux/module.h>
24 #include <linux/capability.h> 24 #include <linux/capability.h>
25 #include <linux/errno.h> 25 #include <linux/errno.h>
26 #include <linux/types.h> 26 #include <linux/types.h>
27 #include <linux/socket.h> 27 #include <linux/socket.h>
28 #include <linux/in.h> 28 #include <linux/in.h>
29 #include <linux/kernel.h> 29 #include <linux/kernel.h>
30 #include <linux/timer.h> 30 #include <linux/timer.h>
31 #include <linux/string.h> 31 #include <linux/string.h>
32 #include <linux/sockios.h> 32 #include <linux/sockios.h>
33 #include <linux/net.h> 33 #include <linux/net.h>
34 #include <linux/fcntl.h> 34 #include <linux/fcntl.h>
35 #include <linux/mm.h> 35 #include <linux/mm.h>
36 #include <linux/interrupt.h> 36 #include <linux/interrupt.h>
37 #include <linux/proc_fs.h> 37 #include <linux/proc_fs.h>
38 #include <linux/stat.h> 38 #include <linux/stat.h>
39 #include <linux/init.h> 39 #include <linux/init.h>
40 #include <linux/slab.h> 40 #include <linux/slab.h>
41 41
42 #include <linux/inet.h> 42 #include <linux/inet.h>
43 #include <linux/netdevice.h> 43 #include <linux/netdevice.h>
44 #include <linux/icmpv6.h> 44 #include <linux/icmpv6.h>
45 #include <linux/netfilter_ipv6.h> 45 #include <linux/netfilter_ipv6.h>
46 46
47 #include <net/ip.h> 47 #include <net/ip.h>
48 #include <net/ipv6.h> 48 #include <net/ipv6.h>
49 #include <net/udp.h> 49 #include <net/udp.h>
50 #include <net/udplite.h> 50 #include <net/udplite.h>
51 #include <net/tcp.h> 51 #include <net/tcp.h>
52 #include <net/ipip.h> 52 #include <net/ipip.h>
53 #include <net/protocol.h> 53 #include <net/protocol.h>
54 #include <net/inet_common.h> 54 #include <net/inet_common.h>
55 #include <net/route.h> 55 #include <net/route.h>
56 #include <net/transp_v6.h> 56 #include <net/transp_v6.h>
57 #include <net/ip6_route.h> 57 #include <net/ip6_route.h>
58 #include <net/addrconf.h> 58 #include <net/addrconf.h>
59 #ifdef CONFIG_IPV6_TUNNEL 59 #ifdef CONFIG_IPV6_TUNNEL
60 #include <net/ip6_tunnel.h> 60 #include <net/ip6_tunnel.h>
61 #endif 61 #endif
62 62
63 #include <asm/uaccess.h> 63 #include <asm/uaccess.h>
64 #include <linux/mroute6.h> 64 #include <linux/mroute6.h>
65 65
66 MODULE_AUTHOR("Cast of dozens"); 66 MODULE_AUTHOR("Cast of dozens");
67 MODULE_DESCRIPTION("IPv6 protocol stack for Linux"); 67 MODULE_DESCRIPTION("IPv6 protocol stack for Linux");
68 MODULE_LICENSE("GPL"); 68 MODULE_LICENSE("GPL");
69 69
70 /* The inetsw6 table contains everything that inet6_create needs to 70 /* The inetsw6 table contains everything that inet6_create needs to
71 * build a new socket. 71 * build a new socket.
72 */ 72 */
73 static struct list_head inetsw6[SOCK_MAX]; 73 static struct list_head inetsw6[SOCK_MAX];
74 static DEFINE_SPINLOCK(inetsw6_lock); 74 static DEFINE_SPINLOCK(inetsw6_lock);
75 75
76 struct ipv6_params ipv6_defaults = { 76 struct ipv6_params ipv6_defaults = {
77 .disable_ipv6 = 0, 77 .disable_ipv6 = 0,
78 .autoconf = 1, 78 .autoconf = 1,
79 }; 79 };
80 80
81 static int disable_ipv6_mod; 81 static int disable_ipv6_mod;
82 82
83 module_param_named(disable, disable_ipv6_mod, int, 0444); 83 module_param_named(disable, disable_ipv6_mod, int, 0444);
84 MODULE_PARM_DESC(disable, "Disable IPv6 module such that it is non-functional"); 84 MODULE_PARM_DESC(disable, "Disable IPv6 module such that it is non-functional");
85 85
86 module_param_named(disable_ipv6, ipv6_defaults.disable_ipv6, int, 0444); 86 module_param_named(disable_ipv6, ipv6_defaults.disable_ipv6, int, 0444);
87 MODULE_PARM_DESC(disable_ipv6, "Disable IPv6 on all interfaces"); 87 MODULE_PARM_DESC(disable_ipv6, "Disable IPv6 on all interfaces");
88 88
89 module_param_named(autoconf, ipv6_defaults.autoconf, int, 0444); 89 module_param_named(autoconf, ipv6_defaults.autoconf, int, 0444);
90 MODULE_PARM_DESC(autoconf, "Enable IPv6 address autoconfiguration on all interfaces"); 90 MODULE_PARM_DESC(autoconf, "Enable IPv6 address autoconfiguration on all interfaces");
91 91
92 static __inline__ struct ipv6_pinfo *inet6_sk_generic(struct sock *sk) 92 static __inline__ struct ipv6_pinfo *inet6_sk_generic(struct sock *sk)
93 { 93 {
94 const int offset = sk->sk_prot->obj_size - sizeof(struct ipv6_pinfo); 94 const int offset = sk->sk_prot->obj_size - sizeof(struct ipv6_pinfo);
95 95
96 return (struct ipv6_pinfo *)(((u8 *)sk) + offset); 96 return (struct ipv6_pinfo *)(((u8 *)sk) + offset);
97 } 97 }
98 98
99 static int inet6_create(struct net *net, struct socket *sock, int protocol, 99 static int inet6_create(struct net *net, struct socket *sock, int protocol,
100 int kern) 100 int kern)
101 { 101 {
102 struct inet_sock *inet; 102 struct inet_sock *inet;
103 struct ipv6_pinfo *np; 103 struct ipv6_pinfo *np;
104 struct sock *sk; 104 struct sock *sk;
105 struct inet_protosw *answer; 105 struct inet_protosw *answer;
106 struct proto *answer_prot; 106 struct proto *answer_prot;
107 unsigned char answer_flags; 107 unsigned char answer_flags;
108 char answer_no_check; 108 char answer_no_check;
109 int try_loading_module = 0; 109 int try_loading_module = 0;
110 int err; 110 int err;
111 111
112 if (sock->type != SOCK_RAW && 112 if (sock->type != SOCK_RAW &&
113 sock->type != SOCK_DGRAM && 113 sock->type != SOCK_DGRAM &&
114 !inet_ehash_secret) 114 !inet_ehash_secret)
115 build_ehash_secret(); 115 build_ehash_secret();
116 116
117 /* Look for the requested type/protocol pair. */ 117 /* Look for the requested type/protocol pair. */
118 lookup_protocol: 118 lookup_protocol:
119 err = -ESOCKTNOSUPPORT; 119 err = -ESOCKTNOSUPPORT;
120 rcu_read_lock(); 120 rcu_read_lock();
121 list_for_each_entry_rcu(answer, &inetsw6[sock->type], list) { 121 list_for_each_entry_rcu(answer, &inetsw6[sock->type], list) {
122 122
123 err = 0; 123 err = 0;
124 /* Check the non-wild match. */ 124 /* Check the non-wild match. */
125 if (protocol == answer->protocol) { 125 if (protocol == answer->protocol) {
126 if (protocol != IPPROTO_IP) 126 if (protocol != IPPROTO_IP)
127 break; 127 break;
128 } else { 128 } else {
129 /* Check for the two wild cases. */ 129 /* Check for the two wild cases. */
130 if (IPPROTO_IP == protocol) { 130 if (IPPROTO_IP == protocol) {
131 protocol = answer->protocol; 131 protocol = answer->protocol;
132 break; 132 break;
133 } 133 }
134 if (IPPROTO_IP == answer->protocol) 134 if (IPPROTO_IP == answer->protocol)
135 break; 135 break;
136 } 136 }
137 err = -EPROTONOSUPPORT; 137 err = -EPROTONOSUPPORT;
138 } 138 }
139 139
140 if (err) { 140 if (err) {
141 if (try_loading_module < 2) { 141 if (try_loading_module < 2) {
142 rcu_read_unlock(); 142 rcu_read_unlock();
143 /* 143 /*
144 * Be more specific, e.g. net-pf-10-proto-132-type-1 144 * Be more specific, e.g. net-pf-10-proto-132-type-1
145 * (net-pf-PF_INET6-proto-IPPROTO_SCTP-type-SOCK_STREAM) 145 * (net-pf-PF_INET6-proto-IPPROTO_SCTP-type-SOCK_STREAM)
146 */ 146 */
147 if (++try_loading_module == 1) 147 if (++try_loading_module == 1)
148 request_module("net-pf-%d-proto-%d-type-%d", 148 request_module("net-pf-%d-proto-%d-type-%d",
149 PF_INET6, protocol, sock->type); 149 PF_INET6, protocol, sock->type);
150 /* 150 /*
151 * Fall back to generic, e.g. net-pf-10-proto-132 151 * Fall back to generic, e.g. net-pf-10-proto-132
152 * (net-pf-PF_INET6-proto-IPPROTO_SCTP) 152 * (net-pf-PF_INET6-proto-IPPROTO_SCTP)
153 */ 153 */
154 else 154 else
155 request_module("net-pf-%d-proto-%d", 155 request_module("net-pf-%d-proto-%d",
156 PF_INET6, protocol); 156 PF_INET6, protocol);
157 goto lookup_protocol; 157 goto lookup_protocol;
158 } else 158 } else
159 goto out_rcu_unlock; 159 goto out_rcu_unlock;
160 } 160 }
161 161
162 err = -EPERM; 162 err = -EPERM;
163 if (sock->type == SOCK_RAW && !kern && 163 if (sock->type == SOCK_RAW && !kern &&
164 !ns_capable(net->user_ns, CAP_NET_RAW)) 164 !ns_capable(net->user_ns, CAP_NET_RAW))
165 goto out_rcu_unlock; 165 goto out_rcu_unlock;
166 166
167 sock->ops = answer->ops; 167 sock->ops = answer->ops;
168 answer_prot = answer->prot; 168 answer_prot = answer->prot;
169 answer_no_check = answer->no_check; 169 answer_no_check = answer->no_check;
170 answer_flags = answer->flags; 170 answer_flags = answer->flags;
171 rcu_read_unlock(); 171 rcu_read_unlock();
172 172
173 WARN_ON(answer_prot->slab == NULL); 173 WARN_ON(answer_prot->slab == NULL);
174 174
175 err = -ENOBUFS; 175 err = -ENOBUFS;
176 sk = sk_alloc(net, PF_INET6, GFP_KERNEL, answer_prot); 176 sk = sk_alloc(net, PF_INET6, GFP_KERNEL, answer_prot);
177 if (sk == NULL) 177 if (sk == NULL)
178 goto out; 178 goto out;
179 179
180 sock_init_data(sock, sk); 180 sock_init_data(sock, sk);
181 181
182 err = 0; 182 err = 0;
183 sk->sk_no_check = answer_no_check; 183 sk->sk_no_check = answer_no_check;
184 if (INET_PROTOSW_REUSE & answer_flags) 184 if (INET_PROTOSW_REUSE & answer_flags)
185 sk->sk_reuse = SK_CAN_REUSE; 185 sk->sk_reuse = SK_CAN_REUSE;
186 186
187 inet = inet_sk(sk); 187 inet = inet_sk(sk);
188 inet->is_icsk = (INET_PROTOSW_ICSK & answer_flags) != 0; 188 inet->is_icsk = (INET_PROTOSW_ICSK & answer_flags) != 0;
189 189
190 if (SOCK_RAW == sock->type) { 190 if (SOCK_RAW == sock->type) {
191 inet->inet_num = protocol; 191 inet->inet_num = protocol;
192 if (IPPROTO_RAW == protocol) 192 if (IPPROTO_RAW == protocol)
193 inet->hdrincl = 1; 193 inet->hdrincl = 1;
194 } 194 }
195 195
196 sk->sk_destruct = inet_sock_destruct; 196 sk->sk_destruct = inet_sock_destruct;
197 sk->sk_family = PF_INET6; 197 sk->sk_family = PF_INET6;
198 sk->sk_protocol = protocol; 198 sk->sk_protocol = protocol;
199 199
200 sk->sk_backlog_rcv = answer->prot->backlog_rcv; 200 sk->sk_backlog_rcv = answer->prot->backlog_rcv;
201 201
202 inet_sk(sk)->pinet6 = np = inet6_sk_generic(sk); 202 inet_sk(sk)->pinet6 = np = inet6_sk_generic(sk);
203 np->hop_limit = -1; 203 np->hop_limit = -1;
204 np->mcast_hops = IPV6_DEFAULT_MCASTHOPS; 204 np->mcast_hops = IPV6_DEFAULT_MCASTHOPS;
205 np->mc_loop = 1; 205 np->mc_loop = 1;
206 np->pmtudisc = IPV6_PMTUDISC_WANT; 206 np->pmtudisc = IPV6_PMTUDISC_WANT;
207 np->ipv6only = net->ipv6.sysctl.bindv6only; 207 np->ipv6only = net->ipv6.sysctl.bindv6only;
208 208
209 /* Init the ipv4 part of the socket since we can have sockets 209 /* Init the ipv4 part of the socket since we can have sockets
210 * using v6 API for ipv4. 210 * using v6 API for ipv4.
211 */ 211 */
212 inet->uc_ttl = -1; 212 inet->uc_ttl = -1;
213 213
214 inet->mc_loop = 1; 214 inet->mc_loop = 1;
215 inet->mc_ttl = 1; 215 inet->mc_ttl = 1;
216 inet->mc_index = 0; 216 inet->mc_index = 0;
217 inet->mc_list = NULL; 217 inet->mc_list = NULL;
218 inet->rcv_tos = 0; 218 inet->rcv_tos = 0;
219 219
220 if (ipv4_config.no_pmtu_disc) 220 if (ipv4_config.no_pmtu_disc)
221 inet->pmtudisc = IP_PMTUDISC_DONT; 221 inet->pmtudisc = IP_PMTUDISC_DONT;
222 else 222 else
223 inet->pmtudisc = IP_PMTUDISC_WANT; 223 inet->pmtudisc = IP_PMTUDISC_WANT;
224 /* 224 /*
225 * Increment only the relevant sk_prot->socks debug field, this changes 225 * Increment only the relevant sk_prot->socks debug field, this changes
226 * the previous behaviour of incrementing both the equivalent to 226 * the previous behaviour of incrementing both the equivalent to
227 * answer->prot->socks (inet6_sock_nr) and inet_sock_nr. 227 * answer->prot->socks (inet6_sock_nr) and inet_sock_nr.
228 * 228 *
229 * This allows better debug granularity as we'll know exactly how many 229 * This allows better debug granularity as we'll know exactly how many
230 * UDPv6, TCPv6, etc socks were allocated, not the sum of all IPv6 230 * UDPv6, TCPv6, etc socks were allocated, not the sum of all IPv6
231 * transport protocol socks. -acme 231 * transport protocol socks. -acme
232 */ 232 */
233 sk_refcnt_debug_inc(sk); 233 sk_refcnt_debug_inc(sk);
234 234
235 if (inet->inet_num) { 235 if (inet->inet_num) {
236 /* It assumes that any protocol which allows 236 /* It assumes that any protocol which allows
237 * the user to assign a number at socket 237 * the user to assign a number at socket
238 * creation time automatically shares. 238 * creation time automatically shares.
239 */ 239 */
240 inet->inet_sport = htons(inet->inet_num); 240 inet->inet_sport = htons(inet->inet_num);
241 sk->sk_prot->hash(sk); 241 sk->sk_prot->hash(sk);
242 } 242 }
243 if (sk->sk_prot->init) { 243 if (sk->sk_prot->init) {
244 err = sk->sk_prot->init(sk); 244 err = sk->sk_prot->init(sk);
245 if (err) { 245 if (err) {
246 sk_common_release(sk); 246 sk_common_release(sk);
247 goto out; 247 goto out;
248 } 248 }
249 } 249 }
250 out: 250 out:
251 return err; 251 return err;
252 out_rcu_unlock: 252 out_rcu_unlock:
253 rcu_read_unlock(); 253 rcu_read_unlock();
254 goto out; 254 goto out;
255 } 255 }
256 256
257 257
258 /* bind for INET6 API */ 258 /* bind for INET6 API */
259 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) 259 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
260 { 260 {
261 struct sockaddr_in6 *addr = (struct sockaddr_in6 *)uaddr; 261 struct sockaddr_in6 *addr = (struct sockaddr_in6 *)uaddr;
262 struct sock *sk = sock->sk; 262 struct sock *sk = sock->sk;
263 struct inet_sock *inet = inet_sk(sk); 263 struct inet_sock *inet = inet_sk(sk);
264 struct ipv6_pinfo *np = inet6_sk(sk); 264 struct ipv6_pinfo *np = inet6_sk(sk);
265 struct net *net = sock_net(sk); 265 struct net *net = sock_net(sk);
266 __be32 v4addr = 0; 266 __be32 v4addr = 0;
267 unsigned short snum; 267 unsigned short snum;
268 int addr_type = 0; 268 int addr_type = 0;
269 int err = 0; 269 int err = 0;
270 270
271 /* If the socket has its own bind function then use it. */ 271 /* If the socket has its own bind function then use it. */
272 if (sk->sk_prot->bind) 272 if (sk->sk_prot->bind)
273 return sk->sk_prot->bind(sk, uaddr, addr_len); 273 return sk->sk_prot->bind(sk, uaddr, addr_len);
274 274
275 if (addr_len < SIN6_LEN_RFC2133) 275 if (addr_len < SIN6_LEN_RFC2133)
276 return -EINVAL; 276 return -EINVAL;
277 277
278 if (addr->sin6_family != AF_INET6) 278 if (addr->sin6_family != AF_INET6)
279 return -EAFNOSUPPORT; 279 return -EAFNOSUPPORT;
280 280
281 addr_type = ipv6_addr_type(&addr->sin6_addr); 281 addr_type = ipv6_addr_type(&addr->sin6_addr);
282 if ((addr_type & IPV6_ADDR_MULTICAST) && sock->type == SOCK_STREAM) 282 if ((addr_type & IPV6_ADDR_MULTICAST) && sock->type == SOCK_STREAM)
283 return -EINVAL; 283 return -EINVAL;
284 284
285 snum = ntohs(addr->sin6_port); 285 snum = ntohs(addr->sin6_port);
286 if (snum && snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE)) 286 if (snum && snum < PROT_SOCK && !ns_capable(net->user_ns, CAP_NET_BIND_SERVICE))
287 return -EACCES; 287 return -EACCES;
288 288
289 lock_sock(sk); 289 lock_sock(sk);
290 290
291 /* Check these errors (active socket, double bind). */ 291 /* Check these errors (active socket, double bind). */
292 if (sk->sk_state != TCP_CLOSE || inet->inet_num) { 292 if (sk->sk_state != TCP_CLOSE || inet->inet_num) {
293 err = -EINVAL; 293 err = -EINVAL;
294 goto out; 294 goto out;
295 } 295 }
296 296
297 /* Check if the address belongs to the host. */ 297 /* Check if the address belongs to the host. */
298 if (addr_type == IPV6_ADDR_MAPPED) { 298 if (addr_type == IPV6_ADDR_MAPPED) {
299 int chk_addr_ret; 299 int chk_addr_ret;
300 300
301 /* Binding to v4-mapped address on a v6-only socket 301 /* Binding to v4-mapped address on a v6-only socket
302 * makes no sense 302 * makes no sense
303 */ 303 */
304 if (np->ipv6only) { 304 if (np->ipv6only) {
305 err = -EINVAL; 305 err = -EINVAL;
306 goto out; 306 goto out;
307 } 307 }
308 308
309 /* Reproduce AF_INET checks to make the bindings consistent */ 309 /* Reproduce AF_INET checks to make the bindings consistent */
310 v4addr = addr->sin6_addr.s6_addr32[3]; 310 v4addr = addr->sin6_addr.s6_addr32[3];
311 chk_addr_ret = inet_addr_type(net, v4addr); 311 chk_addr_ret = inet_addr_type(net, v4addr);
312 if (!sysctl_ip_nonlocal_bind && 312 if (!sysctl_ip_nonlocal_bind &&
313 !(inet->freebind || inet->transparent) && 313 !(inet->freebind || inet->transparent) &&
314 v4addr != htonl(INADDR_ANY) && 314 v4addr != htonl(INADDR_ANY) &&
315 chk_addr_ret != RTN_LOCAL && 315 chk_addr_ret != RTN_LOCAL &&
316 chk_addr_ret != RTN_MULTICAST && 316 chk_addr_ret != RTN_MULTICAST &&
317 chk_addr_ret != RTN_BROADCAST) { 317 chk_addr_ret != RTN_BROADCAST) {
318 err = -EADDRNOTAVAIL; 318 err = -EADDRNOTAVAIL;
319 goto out; 319 goto out;
320 } 320 }
321 } else { 321 } else {
322 if (addr_type != IPV6_ADDR_ANY) { 322 if (addr_type != IPV6_ADDR_ANY) {
323 struct net_device *dev = NULL; 323 struct net_device *dev = NULL;
324 324
325 rcu_read_lock(); 325 rcu_read_lock();
326 if (addr_type & IPV6_ADDR_LINKLOCAL) { 326 if (addr_type & IPV6_ADDR_LINKLOCAL) {
327 if (addr_len >= sizeof(struct sockaddr_in6) && 327 if (addr_len >= sizeof(struct sockaddr_in6) &&
328 addr->sin6_scope_id) { 328 addr->sin6_scope_id) {
329 /* Override any existing binding, if another one 329 /* Override any existing binding, if another one
330 * is supplied by user. 330 * is supplied by user.
331 */ 331 */
332 sk->sk_bound_dev_if = addr->sin6_scope_id; 332 sk->sk_bound_dev_if = addr->sin6_scope_id;
333 } 333 }
334 334
335 /* Binding to link-local address requires an interface */ 335 /* Binding to link-local address requires an interface */
336 if (!sk->sk_bound_dev_if) { 336 if (!sk->sk_bound_dev_if) {
337 err = -EINVAL; 337 err = -EINVAL;
338 goto out_unlock; 338 goto out_unlock;
339 } 339 }
340 dev = dev_get_by_index_rcu(net, sk->sk_bound_dev_if); 340 dev = dev_get_by_index_rcu(net, sk->sk_bound_dev_if);
341 if (!dev) { 341 if (!dev) {
342 err = -ENODEV; 342 err = -ENODEV;
343 goto out_unlock; 343 goto out_unlock;
344 } 344 }
345 } 345 }
346 346
347 /* ipv4 addr of the socket is invalid. Only the 347 /* ipv4 addr of the socket is invalid. Only the
348 * unspecified and mapped address have a v4 equivalent. 348 * unspecified and mapped address have a v4 equivalent.
349 */ 349 */
350 v4addr = LOOPBACK4_IPV6; 350 v4addr = LOOPBACK4_IPV6;
351 if (!(addr_type & IPV6_ADDR_MULTICAST)) { 351 if (!(addr_type & IPV6_ADDR_MULTICAST)) {
352 if (!(inet->freebind || inet->transparent) && 352 if (!(inet->freebind || inet->transparent) &&
353 !ipv6_chk_addr(net, &addr->sin6_addr, 353 !ipv6_chk_addr(net, &addr->sin6_addr,
354 dev, 0)) { 354 dev, 0)) {
355 err = -EADDRNOTAVAIL; 355 err = -EADDRNOTAVAIL;
356 goto out_unlock; 356 goto out_unlock;
357 } 357 }
358 } 358 }
359 rcu_read_unlock(); 359 rcu_read_unlock();
360 } 360 }
361 } 361 }
362 362
363 inet->inet_rcv_saddr = v4addr; 363 inet->inet_rcv_saddr = v4addr;
364 inet->inet_saddr = v4addr; 364 inet->inet_saddr = v4addr;
365 365
366 np->rcv_saddr = addr->sin6_addr; 366 np->rcv_saddr = addr->sin6_addr;
367 367
368 if (!(addr_type & IPV6_ADDR_MULTICAST)) 368 if (!(addr_type & IPV6_ADDR_MULTICAST))
369 np->saddr = addr->sin6_addr; 369 np->saddr = addr->sin6_addr;
370 370
371 /* Make sure we are allowed to bind here. */ 371 /* Make sure we are allowed to bind here. */
372 if (sk->sk_prot->get_port(sk, snum)) { 372 if (sk->sk_prot->get_port(sk, snum)) {
373 inet_reset_saddr(sk); 373 inet_reset_saddr(sk);
374 err = -EADDRINUSE; 374 err = -EADDRINUSE;
375 goto out; 375 goto out;
376 } 376 }
377 377
378 if (addr_type != IPV6_ADDR_ANY) { 378 if (addr_type != IPV6_ADDR_ANY) {
379 sk->sk_userlocks |= SOCK_BINDADDR_LOCK; 379 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
380 if (addr_type != IPV6_ADDR_MAPPED) 380 if (addr_type != IPV6_ADDR_MAPPED)
381 np->ipv6only = 1; 381 np->ipv6only = 1;
382 } 382 }
383 if (snum) 383 if (snum)
384 sk->sk_userlocks |= SOCK_BINDPORT_LOCK; 384 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
385 inet->inet_sport = htons(inet->inet_num); 385 inet->inet_sport = htons(inet->inet_num);
386 inet->inet_dport = 0; 386 inet->inet_dport = 0;
387 inet->inet_daddr = 0; 387 inet->inet_daddr = 0;
388 out: 388 out:
389 release_sock(sk); 389 release_sock(sk);
390 return err; 390 return err;
391 out_unlock: 391 out_unlock:
392 rcu_read_unlock(); 392 rcu_read_unlock();
393 goto out; 393 goto out;
394 } 394 }
395 EXPORT_SYMBOL(inet6_bind); 395 EXPORT_SYMBOL(inet6_bind);
396 396
397 int inet6_release(struct socket *sock) 397 int inet6_release(struct socket *sock)
398 { 398 {
399 struct sock *sk = sock->sk; 399 struct sock *sk = sock->sk;
400 400
401 if (sk == NULL) 401 if (sk == NULL)
402 return -EINVAL; 402 return -EINVAL;
403 403
404 /* Free mc lists */ 404 /* Free mc lists */
405 ipv6_sock_mc_close(sk); 405 ipv6_sock_mc_close(sk);
406 406
407 /* Free ac lists */ 407 /* Free ac lists */
408 ipv6_sock_ac_close(sk); 408 ipv6_sock_ac_close(sk);
409 409
410 return inet_release(sock); 410 return inet_release(sock);
411 } 411 }
412 EXPORT_SYMBOL(inet6_release); 412 EXPORT_SYMBOL(inet6_release);
413 413
414 void inet6_destroy_sock(struct sock *sk) 414 void inet6_destroy_sock(struct sock *sk)
415 { 415 {
416 struct ipv6_pinfo *np = inet6_sk(sk); 416 struct ipv6_pinfo *np = inet6_sk(sk);
417 struct sk_buff *skb; 417 struct sk_buff *skb;
418 struct ipv6_txoptions *opt; 418 struct ipv6_txoptions *opt;
419 419
420 /* Release rx options */ 420 /* Release rx options */
421 421
422 skb = xchg(&np->pktoptions, NULL); 422 skb = xchg(&np->pktoptions, NULL);
423 if (skb != NULL) 423 if (skb != NULL)
424 kfree_skb(skb); 424 kfree_skb(skb);
425 425
426 skb = xchg(&np->rxpmtu, NULL); 426 skb = xchg(&np->rxpmtu, NULL);
427 if (skb != NULL) 427 if (skb != NULL)
428 kfree_skb(skb); 428 kfree_skb(skb);
429 429
430 /* Free flowlabels */ 430 /* Free flowlabels */
431 fl6_free_socklist(sk); 431 fl6_free_socklist(sk);
432 432
433 /* Free tx options */ 433 /* Free tx options */
434 434
435 opt = xchg(&np->opt, NULL); 435 opt = xchg(&np->opt, NULL);
436 if (opt != NULL) 436 if (opt != NULL)
437 sock_kfree_s(sk, opt, opt->tot_len); 437 sock_kfree_s(sk, opt, opt->tot_len);
438 } 438 }
439 EXPORT_SYMBOL_GPL(inet6_destroy_sock); 439 EXPORT_SYMBOL_GPL(inet6_destroy_sock);
440 440
441 /* 441 /*
442 * This does both peername and sockname. 442 * This does both peername and sockname.
443 */ 443 */
444 444
445 int inet6_getname(struct socket *sock, struct sockaddr *uaddr, 445 int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
446 int *uaddr_len, int peer) 446 int *uaddr_len, int peer)
447 { 447 {
448 struct sockaddr_in6 *sin = (struct sockaddr_in6 *)uaddr; 448 struct sockaddr_in6 *sin = (struct sockaddr_in6 *)uaddr;
449 struct sock *sk = sock->sk; 449 struct sock *sk = sock->sk;
450 struct inet_sock *inet = inet_sk(sk); 450 struct inet_sock *inet = inet_sk(sk);
451 struct ipv6_pinfo *np = inet6_sk(sk); 451 struct ipv6_pinfo *np = inet6_sk(sk);
452 452
453 sin->sin6_family = AF_INET6; 453 sin->sin6_family = AF_INET6;
454 sin->sin6_flowinfo = 0; 454 sin->sin6_flowinfo = 0;
455 sin->sin6_scope_id = 0; 455 sin->sin6_scope_id = 0;
456 if (peer) { 456 if (peer) {
457 if (!inet->inet_dport) 457 if (!inet->inet_dport)
458 return -ENOTCONN; 458 return -ENOTCONN;
459 if (((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_SYN_SENT)) && 459 if (((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_SYN_SENT)) &&
460 peer == 1) 460 peer == 1)
461 return -ENOTCONN; 461 return -ENOTCONN;
462 sin->sin6_port = inet->inet_dport; 462 sin->sin6_port = inet->inet_dport;
463 sin->sin6_addr = np->daddr; 463 sin->sin6_addr = np->daddr;
464 if (np->sndflow) 464 if (np->sndflow)
465 sin->sin6_flowinfo = np->flow_label; 465 sin->sin6_flowinfo = np->flow_label;
466 } else { 466 } else {
467 if (ipv6_addr_any(&np->rcv_saddr)) 467 if (ipv6_addr_any(&np->rcv_saddr))
468 sin->sin6_addr = np->saddr; 468 sin->sin6_addr = np->saddr;
469 else 469 else
470 sin->sin6_addr = np->rcv_saddr; 470 sin->sin6_addr = np->rcv_saddr;
471 471
472 sin->sin6_port = inet->inet_sport; 472 sin->sin6_port = inet->inet_sport;
473 } 473 }
474 if (ipv6_addr_type(&sin->sin6_addr) & IPV6_ADDR_LINKLOCAL) 474 if (ipv6_addr_type(&sin->sin6_addr) & IPV6_ADDR_LINKLOCAL)
475 sin->sin6_scope_id = sk->sk_bound_dev_if; 475 sin->sin6_scope_id = sk->sk_bound_dev_if;
476 *uaddr_len = sizeof(*sin); 476 *uaddr_len = sizeof(*sin);
477 return 0; 477 return 0;
478 } 478 }
479 EXPORT_SYMBOL(inet6_getname); 479 EXPORT_SYMBOL(inet6_getname);
480 480
481 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg) 481 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
482 { 482 {
483 struct sock *sk = sock->sk; 483 struct sock *sk = sock->sk;
484 struct net *net = sock_net(sk); 484 struct net *net = sock_net(sk);
485 485
486 switch (cmd) { 486 switch (cmd) {
487 case SIOCGSTAMP: 487 case SIOCGSTAMP:
488 return sock_get_timestamp(sk, (struct timeval __user *)arg); 488 return sock_get_timestamp(sk, (struct timeval __user *)arg);
489 489
490 case SIOCGSTAMPNS: 490 case SIOCGSTAMPNS:
491 return sock_get_timestampns(sk, (struct timespec __user *)arg); 491 return sock_get_timestampns(sk, (struct timespec __user *)arg);
492 492
493 case SIOCADDRT: 493 case SIOCADDRT:
494 case SIOCDELRT: 494 case SIOCDELRT:
495 495
496 return ipv6_route_ioctl(net, cmd, (void __user *)arg); 496 return ipv6_route_ioctl(net, cmd, (void __user *)arg);
497 497
498 case SIOCSIFADDR: 498 case SIOCSIFADDR:
499 return addrconf_add_ifaddr(net, (void __user *) arg); 499 return addrconf_add_ifaddr(net, (void __user *) arg);
500 case SIOCDIFADDR: 500 case SIOCDIFADDR:
501 return addrconf_del_ifaddr(net, (void __user *) arg); 501 return addrconf_del_ifaddr(net, (void __user *) arg);
502 case SIOCSIFDSTADDR: 502 case SIOCSIFDSTADDR:
503 return addrconf_set_dstaddr(net, (void __user *) arg); 503 return addrconf_set_dstaddr(net, (void __user *) arg);
504 default: 504 default:
505 if (!sk->sk_prot->ioctl) 505 if (!sk->sk_prot->ioctl)
506 return -ENOIOCTLCMD; 506 return -ENOIOCTLCMD;
507 return sk->sk_prot->ioctl(sk, cmd, arg); 507 return sk->sk_prot->ioctl(sk, cmd, arg);
508 } 508 }
509 /*NOTREACHED*/ 509 /*NOTREACHED*/
510 return 0; 510 return 0;
511 } 511 }
512 EXPORT_SYMBOL(inet6_ioctl); 512 EXPORT_SYMBOL(inet6_ioctl);
513 513
514 const struct proto_ops inet6_stream_ops = { 514 const struct proto_ops inet6_stream_ops = {
515 .family = PF_INET6, 515 .family = PF_INET6,
516 .owner = THIS_MODULE, 516 .owner = THIS_MODULE,
517 .release = inet6_release, 517 .release = inet6_release,
518 .bind = inet6_bind, 518 .bind = inet6_bind,
519 .connect = inet_stream_connect, /* ok */ 519 .connect = inet_stream_connect, /* ok */
520 .socketpair = sock_no_socketpair, /* a do nothing */ 520 .socketpair = sock_no_socketpair, /* a do nothing */
521 .accept = inet_accept, /* ok */ 521 .accept = inet_accept, /* ok */
522 .getname = inet6_getname, 522 .getname = inet6_getname,
523 .poll = tcp_poll, /* ok */ 523 .poll = tcp_poll, /* ok */
524 .ioctl = inet6_ioctl, /* must change */ 524 .ioctl = inet6_ioctl, /* must change */
525 .listen = inet_listen, /* ok */ 525 .listen = inet_listen, /* ok */
526 .shutdown = inet_shutdown, /* ok */ 526 .shutdown = inet_shutdown, /* ok */
527 .setsockopt = sock_common_setsockopt, /* ok */ 527 .setsockopt = sock_common_setsockopt, /* ok */
528 .getsockopt = sock_common_getsockopt, /* ok */ 528 .getsockopt = sock_common_getsockopt, /* ok */
529 .sendmsg = inet_sendmsg, /* ok */ 529 .sendmsg = inet_sendmsg, /* ok */
530 .recvmsg = inet_recvmsg, /* ok */ 530 .recvmsg = inet_recvmsg, /* ok */
531 .mmap = sock_no_mmap, 531 .mmap = sock_no_mmap,
532 .sendpage = inet_sendpage, 532 .sendpage = inet_sendpage,
533 .splice_read = tcp_splice_read, 533 .splice_read = tcp_splice_read,
534 #ifdef CONFIG_COMPAT 534 #ifdef CONFIG_COMPAT
535 .compat_setsockopt = compat_sock_common_setsockopt, 535 .compat_setsockopt = compat_sock_common_setsockopt,
536 .compat_getsockopt = compat_sock_common_getsockopt, 536 .compat_getsockopt = compat_sock_common_getsockopt,
537 #endif 537 #endif
538 }; 538 };
539 539
540 const struct proto_ops inet6_dgram_ops = { 540 const struct proto_ops inet6_dgram_ops = {
541 .family = PF_INET6, 541 .family = PF_INET6,
542 .owner = THIS_MODULE, 542 .owner = THIS_MODULE,
543 .release = inet6_release, 543 .release = inet6_release,
544 .bind = inet6_bind, 544 .bind = inet6_bind,
545 .connect = inet_dgram_connect, /* ok */ 545 .connect = inet_dgram_connect, /* ok */
546 .socketpair = sock_no_socketpair, /* a do nothing */ 546 .socketpair = sock_no_socketpair, /* a do nothing */
547 .accept = sock_no_accept, /* a do nothing */ 547 .accept = sock_no_accept, /* a do nothing */
548 .getname = inet6_getname, 548 .getname = inet6_getname,
549 .poll = udp_poll, /* ok */ 549 .poll = udp_poll, /* ok */
550 .ioctl = inet6_ioctl, /* must change */ 550 .ioctl = inet6_ioctl, /* must change */
551 .listen = sock_no_listen, /* ok */ 551 .listen = sock_no_listen, /* ok */
552 .shutdown = inet_shutdown, /* ok */ 552 .shutdown = inet_shutdown, /* ok */
553 .setsockopt = sock_common_setsockopt, /* ok */ 553 .setsockopt = sock_common_setsockopt, /* ok */
554 .getsockopt = sock_common_getsockopt, /* ok */ 554 .getsockopt = sock_common_getsockopt, /* ok */
555 .sendmsg = inet_sendmsg, /* ok */ 555 .sendmsg = inet_sendmsg, /* ok */
556 .recvmsg = inet_recvmsg, /* ok */ 556 .recvmsg = inet_recvmsg, /* ok */
557 .mmap = sock_no_mmap, 557 .mmap = sock_no_mmap,
558 .sendpage = sock_no_sendpage, 558 .sendpage = sock_no_sendpage,
559 #ifdef CONFIG_COMPAT 559 #ifdef CONFIG_COMPAT
560 .compat_setsockopt = compat_sock_common_setsockopt, 560 .compat_setsockopt = compat_sock_common_setsockopt,
561 .compat_getsockopt = compat_sock_common_getsockopt, 561 .compat_getsockopt = compat_sock_common_getsockopt,
562 #endif 562 #endif
563 }; 563 };
564 564
565 static const struct net_proto_family inet6_family_ops = { 565 static const struct net_proto_family inet6_family_ops = {
566 .family = PF_INET6, 566 .family = PF_INET6,
567 .create = inet6_create, 567 .create = inet6_create,
568 .owner = THIS_MODULE, 568 .owner = THIS_MODULE,
569 }; 569 };
570 570
571 int inet6_register_protosw(struct inet_protosw *p) 571 int inet6_register_protosw(struct inet_protosw *p)
572 { 572 {
573 struct list_head *lh; 573 struct list_head *lh;
574 struct inet_protosw *answer; 574 struct inet_protosw *answer;
575 struct list_head *last_perm; 575 struct list_head *last_perm;
576 int protocol = p->protocol; 576 int protocol = p->protocol;
577 int ret; 577 int ret;
578 578
579 spin_lock_bh(&inetsw6_lock); 579 spin_lock_bh(&inetsw6_lock);
580 580
581 ret = -EINVAL; 581 ret = -EINVAL;
582 if (p->type >= SOCK_MAX) 582 if (p->type >= SOCK_MAX)
583 goto out_illegal; 583 goto out_illegal;
584 584
585 /* If we are trying to override a permanent protocol, bail. */ 585 /* If we are trying to override a permanent protocol, bail. */
586 answer = NULL; 586 answer = NULL;
587 ret = -EPERM; 587 ret = -EPERM;
588 last_perm = &inetsw6[p->type]; 588 last_perm = &inetsw6[p->type];
589 list_for_each(lh, &inetsw6[p->type]) { 589 list_for_each(lh, &inetsw6[p->type]) {
590 answer = list_entry(lh, struct inet_protosw, list); 590 answer = list_entry(lh, struct inet_protosw, list);
591 591
592 /* Check only the non-wild match. */ 592 /* Check only the non-wild match. */
593 if (INET_PROTOSW_PERMANENT & answer->flags) { 593 if (INET_PROTOSW_PERMANENT & answer->flags) {
594 if (protocol == answer->protocol) 594 if (protocol == answer->protocol)
595 break; 595 break;
596 last_perm = lh; 596 last_perm = lh;
597 } 597 }
598 598
599 answer = NULL; 599 answer = NULL;
600 } 600 }
601 if (answer) 601 if (answer)
602 goto out_permanent; 602 goto out_permanent;
603 603
604 /* Add the new entry after the last permanent entry if any, so that 604 /* Add the new entry after the last permanent entry if any, so that
605 * the new entry does not override a permanent entry when matched with 605 * the new entry does not override a permanent entry when matched with
606 * a wild-card protocol. But it is allowed to override any existing 606 * a wild-card protocol. But it is allowed to override any existing
607 * non-permanent entry. This means that when we remove this entry, the 607 * non-permanent entry. This means that when we remove this entry, the
608 * system automatically returns to the old behavior. 608 * system automatically returns to the old behavior.
609 */ 609 */
610 list_add_rcu(&p->list, last_perm); 610 list_add_rcu(&p->list, last_perm);
611 ret = 0; 611 ret = 0;
612 out: 612 out:
613 spin_unlock_bh(&inetsw6_lock); 613 spin_unlock_bh(&inetsw6_lock);
614 return ret; 614 return ret;
615 615
616 out_permanent: 616 out_permanent:
617 pr_err("Attempt to override permanent protocol %d\n", protocol); 617 pr_err("Attempt to override permanent protocol %d\n", protocol);
618 goto out; 618 goto out;
619 619
620 out_illegal: 620 out_illegal:
621 pr_err("Ignoring attempt to register invalid socket type %d\n", 621 pr_err("Ignoring attempt to register invalid socket type %d\n",
622 p->type); 622 p->type);
623 goto out; 623 goto out;
624 } 624 }
625 EXPORT_SYMBOL(inet6_register_protosw); 625 EXPORT_SYMBOL(inet6_register_protosw);
626 626
627 void 627 void
628 inet6_unregister_protosw(struct inet_protosw *p) 628 inet6_unregister_protosw(struct inet_protosw *p)
629 { 629 {
630 if (INET_PROTOSW_PERMANENT & p->flags) { 630 if (INET_PROTOSW_PERMANENT & p->flags) {
631 pr_err("Attempt to unregister permanent protocol %d\n", 631 pr_err("Attempt to unregister permanent protocol %d\n",
632 p->protocol); 632 p->protocol);
633 } else { 633 } else {
634 spin_lock_bh(&inetsw6_lock); 634 spin_lock_bh(&inetsw6_lock);
635 list_del_rcu(&p->list); 635 list_del_rcu(&p->list);
636 spin_unlock_bh(&inetsw6_lock); 636 spin_unlock_bh(&inetsw6_lock);
637 637
638 synchronize_net(); 638 synchronize_net();
639 } 639 }
640 } 640 }
641 EXPORT_SYMBOL(inet6_unregister_protosw); 641 EXPORT_SYMBOL(inet6_unregister_protosw);
642 642
643 int inet6_sk_rebuild_header(struct sock *sk) 643 int inet6_sk_rebuild_header(struct sock *sk)
644 { 644 {
645 struct ipv6_pinfo *np = inet6_sk(sk); 645 struct ipv6_pinfo *np = inet6_sk(sk);
646 struct dst_entry *dst; 646 struct dst_entry *dst;
647 647
648 dst = __sk_dst_check(sk, np->dst_cookie); 648 dst = __sk_dst_check(sk, np->dst_cookie);
649 649
650 if (dst == NULL) { 650 if (dst == NULL) {
651 struct inet_sock *inet = inet_sk(sk); 651 struct inet_sock *inet = inet_sk(sk);
652 struct in6_addr *final_p, final; 652 struct in6_addr *final_p, final;
653 struct flowi6 fl6; 653 struct flowi6 fl6;
654 654
655 memset(&fl6, 0, sizeof(fl6)); 655 memset(&fl6, 0, sizeof(fl6));
656 fl6.flowi6_proto = sk->sk_protocol; 656 fl6.flowi6_proto = sk->sk_protocol;
657 fl6.daddr = np->daddr; 657 fl6.daddr = np->daddr;
658 fl6.saddr = np->saddr; 658 fl6.saddr = np->saddr;
659 fl6.flowlabel = np->flow_label; 659 fl6.flowlabel = np->flow_label;
660 fl6.flowi6_oif = sk->sk_bound_dev_if; 660 fl6.flowi6_oif = sk->sk_bound_dev_if;
661 fl6.flowi6_mark = sk->sk_mark; 661 fl6.flowi6_mark = sk->sk_mark;
662 fl6.fl6_dport = inet->inet_dport; 662 fl6.fl6_dport = inet->inet_dport;
663 fl6.fl6_sport = inet->inet_sport; 663 fl6.fl6_sport = inet->inet_sport;
664 security_sk_classify_flow(sk, flowi6_to_flowi(&fl6)); 664 security_sk_classify_flow(sk, flowi6_to_flowi(&fl6));
665 665
666 final_p = fl6_update_dst(&fl6, np->opt, &final); 666 final_p = fl6_update_dst(&fl6, np->opt, &final);
667 667
668 dst = ip6_dst_lookup_flow(sk, &fl6, final_p, false); 668 dst = ip6_dst_lookup_flow(sk, &fl6, final_p, false);
669 if (IS_ERR(dst)) { 669 if (IS_ERR(dst)) {
670 sk->sk_route_caps = 0; 670 sk->sk_route_caps = 0;
671 sk->sk_err_soft = -PTR_ERR(dst); 671 sk->sk_err_soft = -PTR_ERR(dst);
672 return PTR_ERR(dst); 672 return PTR_ERR(dst);
673 } 673 }
674 674
675 __ip6_dst_store(sk, dst, NULL, NULL); 675 __ip6_dst_store(sk, dst, NULL, NULL);
676 } 676 }
677 677
678 return 0; 678 return 0;
679 } 679 }
680 EXPORT_SYMBOL_GPL(inet6_sk_rebuild_header); 680 EXPORT_SYMBOL_GPL(inet6_sk_rebuild_header);
681 681
682 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb) 682 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb)
683 { 683 {
684 const struct ipv6_pinfo *np = inet6_sk(sk); 684 const struct ipv6_pinfo *np = inet6_sk(sk);
685 const struct inet6_skb_parm *opt = IP6CB(skb); 685 const struct inet6_skb_parm *opt = IP6CB(skb);
686 686
687 if (np->rxopt.all) { 687 if (np->rxopt.all) {
688 if ((opt->hop && (np->rxopt.bits.hopopts || 688 if ((opt->hop && (np->rxopt.bits.hopopts ||
689 np->rxopt.bits.ohopopts)) || 689 np->rxopt.bits.ohopopts)) ||
690 ((IPV6_FLOWINFO_MASK & 690 ((IPV6_FLOWINFO_MASK &
691 *(__be32 *)skb_network_header(skb)) && 691 *(__be32 *)skb_network_header(skb)) &&
692 np->rxopt.bits.rxflow) || 692 np->rxopt.bits.rxflow) ||
693 (opt->srcrt && (np->rxopt.bits.srcrt || 693 (opt->srcrt && (np->rxopt.bits.srcrt ||
694 np->rxopt.bits.osrcrt)) || 694 np->rxopt.bits.osrcrt)) ||
695 ((opt->dst1 || opt->dst0) && 695 ((opt->dst1 || opt->dst0) &&
696 (np->rxopt.bits.dstopts || np->rxopt.bits.odstopts))) 696 (np->rxopt.bits.dstopts || np->rxopt.bits.odstopts)))
697 return true; 697 return true;
698 } 698 }
699 return false; 699 return false;
700 } 700 }
701 EXPORT_SYMBOL_GPL(ipv6_opt_accepted); 701 EXPORT_SYMBOL_GPL(ipv6_opt_accepted);
702 702
703 static struct packet_type ipv6_packet_type __read_mostly = { 703 static struct packet_type ipv6_packet_type __read_mostly = {
704 .type = cpu_to_be16(ETH_P_IPV6), 704 .type = cpu_to_be16(ETH_P_IPV6),
705 .func = ipv6_rcv, 705 .func = ipv6_rcv,
706 }; 706 };
707 707
708 static int __init ipv6_packet_init(void) 708 static int __init ipv6_packet_init(void)
709 { 709 {
710 dev_add_pack(&ipv6_packet_type); 710 dev_add_pack(&ipv6_packet_type);
711 return 0; 711 return 0;
712 } 712 }
713 713
714 static void ipv6_packet_cleanup(void) 714 static void ipv6_packet_cleanup(void)
715 { 715 {
716 dev_remove_pack(&ipv6_packet_type); 716 dev_remove_pack(&ipv6_packet_type);
717 } 717 }
718 718
719 static int __net_init ipv6_init_mibs(struct net *net) 719 static int __net_init ipv6_init_mibs(struct net *net)
720 { 720 {
721 if (snmp_mib_init((void __percpu **)net->mib.udp_stats_in6, 721 if (snmp_mib_init((void __percpu **)net->mib.udp_stats_in6,
722 sizeof(struct udp_mib), 722 sizeof(struct udp_mib),
723 __alignof__(struct udp_mib)) < 0) 723 __alignof__(struct udp_mib)) < 0)
724 return -ENOMEM; 724 return -ENOMEM;
725 if (snmp_mib_init((void __percpu **)net->mib.udplite_stats_in6, 725 if (snmp_mib_init((void __percpu **)net->mib.udplite_stats_in6,
726 sizeof(struct udp_mib), 726 sizeof(struct udp_mib),
727 __alignof__(struct udp_mib)) < 0) 727 __alignof__(struct udp_mib)) < 0)
728 goto err_udplite_mib; 728 goto err_udplite_mib;
729 if (snmp_mib_init((void __percpu **)net->mib.ipv6_statistics, 729 if (snmp_mib_init((void __percpu **)net->mib.ipv6_statistics,
730 sizeof(struct ipstats_mib), 730 sizeof(struct ipstats_mib),
731 __alignof__(struct ipstats_mib)) < 0) 731 __alignof__(struct ipstats_mib)) < 0)
732 goto err_ip_mib; 732 goto err_ip_mib;
733 if (snmp_mib_init((void __percpu **)net->mib.icmpv6_statistics, 733 if (snmp_mib_init((void __percpu **)net->mib.icmpv6_statistics,
734 sizeof(struct icmpv6_mib), 734 sizeof(struct icmpv6_mib),
735 __alignof__(struct icmpv6_mib)) < 0) 735 __alignof__(struct icmpv6_mib)) < 0)
736 goto err_icmp_mib; 736 goto err_icmp_mib;
737 net->mib.icmpv6msg_statistics = kzalloc(sizeof(struct icmpv6msg_mib), 737 net->mib.icmpv6msg_statistics = kzalloc(sizeof(struct icmpv6msg_mib),
738 GFP_KERNEL); 738 GFP_KERNEL);
739 if (!net->mib.icmpv6msg_statistics) 739 if (!net->mib.icmpv6msg_statistics)
740 goto err_icmpmsg_mib; 740 goto err_icmpmsg_mib;
741 return 0; 741 return 0;
742 742
743 err_icmpmsg_mib: 743 err_icmpmsg_mib:
744 snmp_mib_free((void __percpu **)net->mib.icmpv6_statistics); 744 snmp_mib_free((void __percpu **)net->mib.icmpv6_statistics);
745 err_icmp_mib: 745 err_icmp_mib:
746 snmp_mib_free((void __percpu **)net->mib.ipv6_statistics); 746 snmp_mib_free((void __percpu **)net->mib.ipv6_statistics);
747 err_ip_mib: 747 err_ip_mib:
748 snmp_mib_free((void __percpu **)net->mib.udplite_stats_in6); 748 snmp_mib_free((void __percpu **)net->mib.udplite_stats_in6);
749 err_udplite_mib: 749 err_udplite_mib:
750 snmp_mib_free((void __percpu **)net->mib.udp_stats_in6); 750 snmp_mib_free((void __percpu **)net->mib.udp_stats_in6);
751 return -ENOMEM; 751 return -ENOMEM;
752 } 752 }
753 753
754 static void ipv6_cleanup_mibs(struct net *net) 754 static void ipv6_cleanup_mibs(struct net *net)
755 { 755 {
756 snmp_mib_free((void __percpu **)net->mib.udp_stats_in6); 756 snmp_mib_free((void __percpu **)net->mib.udp_stats_in6);
757 snmp_mib_free((void __percpu **)net->mib.udplite_stats_in6); 757 snmp_mib_free((void __percpu **)net->mib.udplite_stats_in6);
758 snmp_mib_free((void __percpu **)net->mib.ipv6_statistics); 758 snmp_mib_free((void __percpu **)net->mib.ipv6_statistics);
759 snmp_mib_free((void __percpu **)net->mib.icmpv6_statistics); 759 snmp_mib_free((void __percpu **)net->mib.icmpv6_statistics);
760 kfree(net->mib.icmpv6msg_statistics); 760 kfree(net->mib.icmpv6msg_statistics);
761 } 761 }
762 762
763 static int __net_init inet6_net_init(struct net *net) 763 static int __net_init inet6_net_init(struct net *net)
764 { 764 {
765 int err = 0; 765 int err = 0;
766 766
767 net->ipv6.sysctl.bindv6only = 0; 767 net->ipv6.sysctl.bindv6only = 0;
768 net->ipv6.sysctl.icmpv6_time = 1*HZ; 768 net->ipv6.sysctl.icmpv6_time = 1*HZ;
769 769
770 err = ipv6_init_mibs(net); 770 err = ipv6_init_mibs(net);
771 if (err) 771 if (err)
772 return err; 772 return err;
773 #ifdef CONFIG_PROC_FS 773 #ifdef CONFIG_PROC_FS
774 err = udp6_proc_init(net); 774 err = udp6_proc_init(net);
775 if (err) 775 if (err)
776 goto out; 776 goto out;
777 err = tcp6_proc_init(net); 777 err = tcp6_proc_init(net);
778 if (err) 778 if (err)
779 goto proc_tcp6_fail; 779 goto proc_tcp6_fail;
780 err = ac6_proc_init(net); 780 err = ac6_proc_init(net);
781 if (err) 781 if (err)
782 goto proc_ac6_fail; 782 goto proc_ac6_fail;
783 #endif 783 #endif
784 return err; 784 return err;
785 785
786 #ifdef CONFIG_PROC_FS 786 #ifdef CONFIG_PROC_FS
787 proc_ac6_fail: 787 proc_ac6_fail:
788 tcp6_proc_exit(net); 788 tcp6_proc_exit(net);
789 proc_tcp6_fail: 789 proc_tcp6_fail:
790 udp6_proc_exit(net); 790 udp6_proc_exit(net);
791 out: 791 out:
792 ipv6_cleanup_mibs(net); 792 ipv6_cleanup_mibs(net);
793 return err; 793 return err;
794 #endif 794 #endif
795 } 795 }
796 796
797 static void __net_exit inet6_net_exit(struct net *net) 797 static void __net_exit inet6_net_exit(struct net *net)
798 { 798 {
799 #ifdef CONFIG_PROC_FS 799 #ifdef CONFIG_PROC_FS
800 udp6_proc_exit(net); 800 udp6_proc_exit(net);
801 tcp6_proc_exit(net); 801 tcp6_proc_exit(net);
802 ac6_proc_exit(net); 802 ac6_proc_exit(net);
803 #endif 803 #endif
804 ipv6_cleanup_mibs(net); 804 ipv6_cleanup_mibs(net);
805 } 805 }
806 806
807 static struct pernet_operations inet6_net_ops = { 807 static struct pernet_operations inet6_net_ops = {
808 .init = inet6_net_init, 808 .init = inet6_net_init,
809 .exit = inet6_net_exit, 809 .exit = inet6_net_exit,
810 }; 810 };
811 811
812 static int __init inet6_init(void) 812 static int __init inet6_init(void)
813 { 813 {
814 struct sk_buff *dummy_skb; 814 struct sk_buff *dummy_skb;
815 struct list_head *r; 815 struct list_head *r;
816 int err = 0; 816 int err = 0;
817 817
818 BUILD_BUG_ON(sizeof(struct inet6_skb_parm) > sizeof(dummy_skb->cb)); 818 BUILD_BUG_ON(sizeof(struct inet6_skb_parm) > sizeof(dummy_skb->cb));
819 819
820 /* Register the socket-side information for inet6_create. */ 820 /* Register the socket-side information for inet6_create. */
821 for (r = &inetsw6[0]; r < &inetsw6[SOCK_MAX]; ++r) 821 for (r = &inetsw6[0]; r < &inetsw6[SOCK_MAX]; ++r)
822 INIT_LIST_HEAD(r); 822 INIT_LIST_HEAD(r);
823 823
824 if (disable_ipv6_mod) { 824 if (disable_ipv6_mod) {
825 pr_info("Loaded, but administratively disabled, reboot required to enable\n"); 825 pr_info("Loaded, but administratively disabled, reboot required to enable\n");
826 goto out; 826 goto out;
827 } 827 }
828 828
829 err = proto_register(&tcpv6_prot, 1); 829 err = proto_register(&tcpv6_prot, 1);
830 if (err) 830 if (err)
831 goto out; 831 goto out;
832 832
833 err = proto_register(&udpv6_prot, 1); 833 err = proto_register(&udpv6_prot, 1);
834 if (err) 834 if (err)
835 goto out_unregister_tcp_proto; 835 goto out_unregister_tcp_proto;
836 836
837 err = proto_register(&udplitev6_prot, 1); 837 err = proto_register(&udplitev6_prot, 1);
838 if (err) 838 if (err)
839 goto out_unregister_udp_proto; 839 goto out_unregister_udp_proto;
840 840
841 err = proto_register(&rawv6_prot, 1); 841 err = proto_register(&rawv6_prot, 1);
842 if (err) 842 if (err)
843 goto out_unregister_udplite_proto; 843 goto out_unregister_udplite_proto;
844 844
845 845
846 /* We MUST register RAW sockets before we create the ICMP6, 846 /* We MUST register RAW sockets before we create the ICMP6,
847 * IGMP6, or NDISC control sockets. 847 * IGMP6, or NDISC control sockets.
848 */ 848 */
849 err = rawv6_init(); 849 err = rawv6_init();
850 if (err) 850 if (err)
851 goto out_unregister_raw_proto; 851 goto out_unregister_raw_proto;
852 852
853 /* Register the family here so that the init calls below will 853 /* Register the family here so that the init calls below will
854 * be able to create sockets. (?? is this dangerous ??) 854 * be able to create sockets. (?? is this dangerous ??)
855 */ 855 */
856 err = sock_register(&inet6_family_ops); 856 err = sock_register(&inet6_family_ops);
857 if (err) 857 if (err)
858 goto out_sock_register_fail; 858 goto out_sock_register_fail;
859 859
860 tcpv6_prot.sysctl_mem = init_net.ipv4.sysctl_tcp_mem; 860 tcpv6_prot.sysctl_mem = init_net.ipv4.sysctl_tcp_mem;
861 861
862 /* 862 /*
863 * ipngwg API draft makes clear that the correct semantics 863 * ipngwg API draft makes clear that the correct semantics
864 * for TCP and UDP is to consider one TCP and UDP instance 864 * for TCP and UDP is to consider one TCP and UDP instance
865 * in a host available by both INET and INET6 APIs and 865 * in a host available by both INET and INET6 APIs and
866 * able to communicate via both network protocols. 866 * able to communicate via both network protocols.
867 */ 867 */
868 868
869 err = register_pernet_subsys(&inet6_net_ops); 869 err = register_pernet_subsys(&inet6_net_ops);
870 if (err) 870 if (err)
871 goto register_pernet_fail; 871 goto register_pernet_fail;
872 err = icmpv6_init(); 872 err = icmpv6_init();
873 if (err) 873 if (err)
874 goto icmp_fail; 874 goto icmp_fail;
875 err = ip6_mr_init(); 875 err = ip6_mr_init();
876 if (err) 876 if (err)
877 goto ipmr_fail; 877 goto ipmr_fail;
878 err = ndisc_init(); 878 err = ndisc_init();
879 if (err) 879 if (err)
880 goto ndisc_fail; 880 goto ndisc_fail;
881 err = igmp6_init(); 881 err = igmp6_init();
882 if (err) 882 if (err)
883 goto igmp_fail; 883 goto igmp_fail;
884 err = ipv6_netfilter_init(); 884 err = ipv6_netfilter_init();
885 if (err) 885 if (err)
886 goto netfilter_fail; 886 goto netfilter_fail;
887 /* Create /proc/foo6 entries. */ 887 /* Create /proc/foo6 entries. */
888 #ifdef CONFIG_PROC_FS 888 #ifdef CONFIG_PROC_FS
889 err = -ENOMEM; 889 err = -ENOMEM;
890 if (raw6_proc_init()) 890 if (raw6_proc_init())
891 goto proc_raw6_fail; 891 goto proc_raw6_fail;
892 if (udplite6_proc_init()) 892 if (udplite6_proc_init())
893 goto proc_udplite6_fail; 893 goto proc_udplite6_fail;
894 if (ipv6_misc_proc_init()) 894 if (ipv6_misc_proc_init())
895 goto proc_misc6_fail; 895 goto proc_misc6_fail;
896 if (if6_proc_init()) 896 if (if6_proc_init())
897 goto proc_if6_fail; 897 goto proc_if6_fail;
898 #endif 898 #endif
899 err = ip6_route_init(); 899 err = ip6_route_init();
900 if (err) 900 if (err)
901 goto ip6_route_fail; 901 goto ip6_route_fail;
902 err = ip6_flowlabel_init(); 902 err = ip6_flowlabel_init();
903 if (err) 903 if (err)
904 goto ip6_flowlabel_fail; 904 goto ip6_flowlabel_fail;
905 err = addrconf_init(); 905 err = addrconf_init();
906 if (err) 906 if (err)
907 goto addrconf_fail; 907 goto addrconf_fail;
908 908
909 /* Init v6 extension headers. */ 909 /* Init v6 extension headers. */
910 err = ipv6_exthdrs_init(); 910 err = ipv6_exthdrs_init();
911 if (err) 911 if (err)
912 goto ipv6_exthdrs_fail; 912 goto ipv6_exthdrs_fail;
913 913
914 err = ipv6_frag_init(); 914 err = ipv6_frag_init();
915 if (err) 915 if (err)
916 goto ipv6_frag_fail; 916 goto ipv6_frag_fail;
917 917
918 /* Init v6 transport protocols. */ 918 /* Init v6 transport protocols. */
919 err = udpv6_init(); 919 err = udpv6_init();
920 if (err) 920 if (err)
921 goto udpv6_fail; 921 goto udpv6_fail;
922 922
923 err = udplitev6_init(); 923 err = udplitev6_init();
924 if (err) 924 if (err)
925 goto udplitev6_fail; 925 goto udplitev6_fail;
926 926
927 err = tcpv6_init(); 927 err = tcpv6_init();
928 if (err) 928 if (err)
929 goto tcpv6_fail; 929 goto tcpv6_fail;
930 930
931 err = ipv6_packet_init(); 931 err = ipv6_packet_init();
932 if (err) 932 if (err)
933 goto ipv6_packet_fail; 933 goto ipv6_packet_fail;
934 934
935 #ifdef CONFIG_SYSCTL 935 #ifdef CONFIG_SYSCTL
936 err = ipv6_sysctl_register(); 936 err = ipv6_sysctl_register();
937 if (err) 937 if (err)
938 goto sysctl_fail; 938 goto sysctl_fail;
939 #endif 939 #endif
940 out: 940 out:
941 return err; 941 return err;
942 942
943 #ifdef CONFIG_SYSCTL 943 #ifdef CONFIG_SYSCTL
944 sysctl_fail: 944 sysctl_fail:
945 ipv6_packet_cleanup(); 945 ipv6_packet_cleanup();
946 #endif 946 #endif
947 ipv6_packet_fail: 947 ipv6_packet_fail:
948 tcpv6_exit(); 948 tcpv6_exit();
949 tcpv6_fail: 949 tcpv6_fail:
950 udplitev6_exit(); 950 udplitev6_exit();
951 udplitev6_fail: 951 udplitev6_fail:
952 udpv6_exit(); 952 udpv6_exit();
953 udpv6_fail: 953 udpv6_fail:
954 ipv6_frag_exit(); 954 ipv6_frag_exit();
955 ipv6_frag_fail: 955 ipv6_frag_fail:
956 ipv6_exthdrs_exit(); 956 ipv6_exthdrs_exit();
957 ipv6_exthdrs_fail: 957 ipv6_exthdrs_fail:
958 addrconf_cleanup(); 958 addrconf_cleanup();
959 addrconf_fail: 959 addrconf_fail:
960 ip6_flowlabel_cleanup(); 960 ip6_flowlabel_cleanup();
961 ip6_flowlabel_fail: 961 ip6_flowlabel_fail:
962 ip6_route_cleanup(); 962 ip6_route_cleanup();
963 ip6_route_fail: 963 ip6_route_fail:
964 #ifdef CONFIG_PROC_FS 964 #ifdef CONFIG_PROC_FS
965 if6_proc_exit(); 965 if6_proc_exit();
966 proc_if6_fail: 966 proc_if6_fail:
967 ipv6_misc_proc_exit(); 967 ipv6_misc_proc_exit();
968 proc_misc6_fail: 968 proc_misc6_fail:
969 udplite6_proc_exit(); 969 udplite6_proc_exit();
970 proc_udplite6_fail: 970 proc_udplite6_fail:
971 raw6_proc_exit(); 971 raw6_proc_exit();
972 proc_raw6_fail: 972 proc_raw6_fail:
973 #endif 973 #endif
974 ipv6_netfilter_fini(); 974 ipv6_netfilter_fini();
975 netfilter_fail: 975 netfilter_fail:
976 igmp6_cleanup(); 976 igmp6_cleanup();
977 igmp_fail: 977 igmp_fail:
978 ndisc_cleanup(); 978 ndisc_cleanup();
979 ndisc_fail: 979 ndisc_fail:
980 ip6_mr_cleanup(); 980 ip6_mr_cleanup();
981 ipmr_fail: 981 ipmr_fail:
982 icmpv6_cleanup(); 982 icmpv6_cleanup();
983 icmp_fail: 983 icmp_fail:
984 unregister_pernet_subsys(&inet6_net_ops); 984 unregister_pernet_subsys(&inet6_net_ops);
985 register_pernet_fail: 985 register_pernet_fail:
986 sock_unregister(PF_INET6); 986 sock_unregister(PF_INET6);
987 rtnl_unregister_all(PF_INET6); 987 rtnl_unregister_all(PF_INET6);
988 out_sock_register_fail: 988 out_sock_register_fail:
989 rawv6_exit(); 989 rawv6_exit();
990 out_unregister_raw_proto: 990 out_unregister_raw_proto:
991 proto_unregister(&rawv6_prot); 991 proto_unregister(&rawv6_prot);
992 out_unregister_udplite_proto: 992 out_unregister_udplite_proto:
993 proto_unregister(&udplitev6_prot); 993 proto_unregister(&udplitev6_prot);
994 out_unregister_udp_proto: 994 out_unregister_udp_proto:
995 proto_unregister(&udpv6_prot); 995 proto_unregister(&udpv6_prot);
996 out_unregister_tcp_proto: 996 out_unregister_tcp_proto:
997 proto_unregister(&tcpv6_prot); 997 proto_unregister(&tcpv6_prot);
998 goto out; 998 goto out;
999 } 999 }
1000 module_init(inet6_init); 1000 module_init(inet6_init);
1001 1001
1002 static void __exit inet6_exit(void) 1002 static void __exit inet6_exit(void)
1003 { 1003 {
1004 if (disable_ipv6_mod) 1004 if (disable_ipv6_mod)
1005 return; 1005 return;
1006 1006
1007 /* First of all disallow new sockets creation. */ 1007 /* First of all disallow new sockets creation. */
1008 sock_unregister(PF_INET6); 1008 sock_unregister(PF_INET6);
1009 /* Disallow any further netlink messages */ 1009 /* Disallow any further netlink messages */
1010 rtnl_unregister_all(PF_INET6); 1010 rtnl_unregister_all(PF_INET6);
1011 1011
1012 udpv6_exit(); 1012 udpv6_exit();
1013 udplitev6_exit(); 1013 udplitev6_exit();
1014 tcpv6_exit(); 1014 tcpv6_exit();
1015 1015
1016 /* Cleanup code parts. */ 1016 /* Cleanup code parts. */
1017 ipv6_packet_cleanup(); 1017 ipv6_packet_cleanup();
1018 ipv6_frag_exit(); 1018 ipv6_frag_exit();
1019 ipv6_exthdrs_exit(); 1019 ipv6_exthdrs_exit();
1020 addrconf_cleanup(); 1020 addrconf_cleanup();
1021 ip6_flowlabel_cleanup(); 1021 ip6_flowlabel_cleanup();
1022 ip6_route_cleanup(); 1022 ip6_route_cleanup();
1023 #ifdef CONFIG_PROC_FS 1023 #ifdef CONFIG_PROC_FS
1024 1024
1025 /* Cleanup code parts. */ 1025 /* Cleanup code parts. */
1026 if6_proc_exit(); 1026 if6_proc_exit();
1027 ipv6_misc_proc_exit(); 1027 ipv6_misc_proc_exit();
1028 udplite6_proc_exit(); 1028 udplite6_proc_exit();
1029 raw6_proc_exit(); 1029 raw6_proc_exit();
1030 #endif 1030 #endif
1031 ipv6_netfilter_fini(); 1031 ipv6_netfilter_fini();
1032 igmp6_cleanup(); 1032 igmp6_cleanup();
1033 ndisc_cleanup(); 1033 ndisc_cleanup();
1034 ip6_mr_cleanup(); 1034 ip6_mr_cleanup();
1035 icmpv6_cleanup(); 1035 icmpv6_cleanup();
1036 rawv6_exit(); 1036 rawv6_exit();
1037 1037
1038 unregister_pernet_subsys(&inet6_net_ops); 1038 unregister_pernet_subsys(&inet6_net_ops);
1039 proto_unregister(&rawv6_prot); 1039 proto_unregister(&rawv6_prot);
1040 proto_unregister(&udplitev6_prot); 1040 proto_unregister(&udplitev6_prot);
1041 proto_unregister(&udpv6_prot); 1041 proto_unregister(&udpv6_prot);
1042 proto_unregister(&tcpv6_prot); 1042 proto_unregister(&tcpv6_prot);
1043 1043
1044 rcu_barrier(); /* Wait for completion of call_rcu()'s */ 1044 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1045 } 1045 }
1046 module_exit(inet6_exit); 1046 module_exit(inet6_exit);
1047 1047
1048 MODULE_ALIAS_NETPROTO(PF_INET6); 1048 MODULE_ALIAS_NETPROTO(PF_INET6);
1049 1049
1 /* SCTP kernel implementation 1 /* SCTP kernel implementation
2 * (C) Copyright IBM Corp. 2001, 2004 2 * (C) Copyright IBM Corp. 2001, 2004
3 * Copyright (c) 1999-2000 Cisco, Inc. 3 * Copyright (c) 1999-2000 Cisco, Inc.
4 * Copyright (c) 1999-2001 Motorola, Inc. 4 * Copyright (c) 1999-2001 Motorola, Inc.
5 * Copyright (c) 2001-2003 Intel Corp. 5 * Copyright (c) 2001-2003 Intel Corp.
6 * Copyright (c) 2001-2002 Nokia, Inc. 6 * Copyright (c) 2001-2002 Nokia, Inc.
7 * Copyright (c) 2001 La Monte H.P. Yarroll 7 * Copyright (c) 2001 La Monte H.P. Yarroll
8 * 8 *
9 * This file is part of the SCTP kernel implementation 9 * This file is part of the SCTP kernel implementation
10 * 10 *
11 * These functions interface with the sockets layer to implement the 11 * These functions interface with the sockets layer to implement the
12 * SCTP Extensions for the Sockets API. 12 * SCTP Extensions for the Sockets API.
13 * 13 *
14 * Note that the descriptions from the specification are USER level 14 * Note that the descriptions from the specification are USER level
15 * functions--this file is the functions which populate the struct proto 15 * functions--this file is the functions which populate the struct proto
16 * for SCTP which is the BOTTOM of the sockets interface. 16 * for SCTP which is the BOTTOM of the sockets interface.
17 * 17 *
18 * This SCTP implementation is free software; 18 * This SCTP implementation is free software;
19 * you can redistribute it and/or modify it under the terms of 19 * you can redistribute it and/or modify it under the terms of
20 * the GNU General Public License as published by 20 * the GNU General Public License as published by
21 * the Free Software Foundation; either version 2, or (at your option) 21 * the Free Software Foundation; either version 2, or (at your option)
22 * any later version. 22 * any later version.
23 * 23 *
24 * This SCTP implementation is distributed in the hope that it 24 * This SCTP implementation is distributed in the hope that it
25 * will be useful, but WITHOUT ANY WARRANTY; without even the implied 25 * will be useful, but WITHOUT ANY WARRANTY; without even the implied
26 * ************************ 26 * ************************
27 * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. 27 * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
28 * See the GNU General Public License for more details. 28 * See the GNU General Public License for more details.
29 * 29 *
30 * You should have received a copy of the GNU General Public License 30 * You should have received a copy of the GNU General Public License
31 * along with GNU CC; see the file COPYING. If not, write to 31 * along with GNU CC; see the file COPYING. If not, write to
32 * the Free Software Foundation, 59 Temple Place - Suite 330, 32 * the Free Software Foundation, 59 Temple Place - Suite 330,
33 * Boston, MA 02111-1307, USA. 33 * Boston, MA 02111-1307, USA.
34 * 34 *
35 * Please send any bug reports or fixes you make to the 35 * Please send any bug reports or fixes you make to the
36 * email address(es): 36 * email address(es):
37 * lksctp developers <lksctp-developers@lists.sourceforge.net> 37 * lksctp developers <lksctp-developers@lists.sourceforge.net>
38 * 38 *
39 * Or submit a bug report through the following website: 39 * Or submit a bug report through the following website:
40 * http://www.sf.net/projects/lksctp 40 * http://www.sf.net/projects/lksctp
41 * 41 *
42 * Written or modified by: 42 * Written or modified by:
43 * La Monte H.P. Yarroll <piggy@acm.org> 43 * La Monte H.P. Yarroll <piggy@acm.org>
44 * Narasimha Budihal <narsi@refcode.org> 44 * Narasimha Budihal <narsi@refcode.org>
45 * Karl Knutson <karl@athena.chicago.il.us> 45 * Karl Knutson <karl@athena.chicago.il.us>
46 * Jon Grimm <jgrimm@us.ibm.com> 46 * Jon Grimm <jgrimm@us.ibm.com>
47 * Xingang Guo <xingang.guo@intel.com> 47 * Xingang Guo <xingang.guo@intel.com>
48 * Daisy Chang <daisyc@us.ibm.com> 48 * Daisy Chang <daisyc@us.ibm.com>
49 * Sridhar Samudrala <samudrala@us.ibm.com> 49 * Sridhar Samudrala <samudrala@us.ibm.com>
50 * Inaky Perez-Gonzalez <inaky.gonzalez@intel.com> 50 * Inaky Perez-Gonzalez <inaky.gonzalez@intel.com>
51 * Ardelle Fan <ardelle.fan@intel.com> 51 * Ardelle Fan <ardelle.fan@intel.com>
52 * Ryan Layer <rmlayer@us.ibm.com> 52 * Ryan Layer <rmlayer@us.ibm.com>
53 * Anup Pemmaiah <pemmaiah@cc.usu.edu> 53 * Anup Pemmaiah <pemmaiah@cc.usu.edu>
54 * Kevin Gao <kevin.gao@intel.com> 54 * Kevin Gao <kevin.gao@intel.com>
55 * 55 *
56 * Any bugs reported given to us we will try to fix... any fixes shared will 56 * Any bugs reported given to us we will try to fix... any fixes shared will
57 * be incorporated into the next SCTP release. 57 * be incorporated into the next SCTP release.
58 */ 58 */
59 59
60 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt 60 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
61 61
62 #include <linux/types.h> 62 #include <linux/types.h>
63 #include <linux/kernel.h> 63 #include <linux/kernel.h>
64 #include <linux/wait.h> 64 #include <linux/wait.h>
65 #include <linux/time.h> 65 #include <linux/time.h>
66 #include <linux/ip.h> 66 #include <linux/ip.h>
67 #include <linux/capability.h> 67 #include <linux/capability.h>
68 #include <linux/fcntl.h> 68 #include <linux/fcntl.h>
69 #include <linux/poll.h> 69 #include <linux/poll.h>
70 #include <linux/init.h> 70 #include <linux/init.h>
71 #include <linux/crypto.h> 71 #include <linux/crypto.h>
72 #include <linux/slab.h> 72 #include <linux/slab.h>
73 #include <linux/file.h> 73 #include <linux/file.h>
74 74
75 #include <net/ip.h> 75 #include <net/ip.h>
76 #include <net/icmp.h> 76 #include <net/icmp.h>
77 #include <net/route.h> 77 #include <net/route.h>
78 #include <net/ipv6.h> 78 #include <net/ipv6.h>
79 #include <net/inet_common.h> 79 #include <net/inet_common.h>
80 80
81 #include <linux/socket.h> /* for sa_family_t */ 81 #include <linux/socket.h> /* for sa_family_t */
82 #include <linux/export.h> 82 #include <linux/export.h>
83 #include <net/sock.h> 83 #include <net/sock.h>
84 #include <net/sctp/sctp.h> 84 #include <net/sctp/sctp.h>
85 #include <net/sctp/sm.h> 85 #include <net/sctp/sm.h>
86 86
87 /* WARNING: Please do not remove the SCTP_STATIC attribute to 87 /* WARNING: Please do not remove the SCTP_STATIC attribute to
88 * any of the functions below as they are used to export functions 88 * any of the functions below as they are used to export functions
89 * used by a project regression testsuite. 89 * used by a project regression testsuite.
90 */ 90 */
91 91
92 /* Forward declarations for internal helper functions. */ 92 /* Forward declarations for internal helper functions. */
93 static int sctp_writeable(struct sock *sk); 93 static int sctp_writeable(struct sock *sk);
94 static void sctp_wfree(struct sk_buff *skb); 94 static void sctp_wfree(struct sk_buff *skb);
95 static int sctp_wait_for_sndbuf(struct sctp_association *, long *timeo_p, 95 static int sctp_wait_for_sndbuf(struct sctp_association *, long *timeo_p,
96 size_t msg_len); 96 size_t msg_len);
97 static int sctp_wait_for_packet(struct sock * sk, int *err, long *timeo_p); 97 static int sctp_wait_for_packet(struct sock * sk, int *err, long *timeo_p);
98 static int sctp_wait_for_connect(struct sctp_association *, long *timeo_p); 98 static int sctp_wait_for_connect(struct sctp_association *, long *timeo_p);
99 static int sctp_wait_for_accept(struct sock *sk, long timeo); 99 static int sctp_wait_for_accept(struct sock *sk, long timeo);
100 static void sctp_wait_for_close(struct sock *sk, long timeo); 100 static void sctp_wait_for_close(struct sock *sk, long timeo);
101 static struct sctp_af *sctp_sockaddr_af(struct sctp_sock *opt, 101 static struct sctp_af *sctp_sockaddr_af(struct sctp_sock *opt,
102 union sctp_addr *addr, int len); 102 union sctp_addr *addr, int len);
103 static int sctp_bindx_add(struct sock *, struct sockaddr *, int); 103 static int sctp_bindx_add(struct sock *, struct sockaddr *, int);
104 static int sctp_bindx_rem(struct sock *, struct sockaddr *, int); 104 static int sctp_bindx_rem(struct sock *, struct sockaddr *, int);
105 static int sctp_send_asconf_add_ip(struct sock *, struct sockaddr *, int); 105 static int sctp_send_asconf_add_ip(struct sock *, struct sockaddr *, int);
106 static int sctp_send_asconf_del_ip(struct sock *, struct sockaddr *, int); 106 static int sctp_send_asconf_del_ip(struct sock *, struct sockaddr *, int);
107 static int sctp_send_asconf(struct sctp_association *asoc, 107 static int sctp_send_asconf(struct sctp_association *asoc,
108 struct sctp_chunk *chunk); 108 struct sctp_chunk *chunk);
109 static int sctp_do_bind(struct sock *, union sctp_addr *, int); 109 static int sctp_do_bind(struct sock *, union sctp_addr *, int);
110 static int sctp_autobind(struct sock *sk); 110 static int sctp_autobind(struct sock *sk);
111 static void sctp_sock_migrate(struct sock *, struct sock *, 111 static void sctp_sock_migrate(struct sock *, struct sock *,
112 struct sctp_association *, sctp_socket_type_t); 112 struct sctp_association *, sctp_socket_type_t);
113 113
114 extern struct kmem_cache *sctp_bucket_cachep; 114 extern struct kmem_cache *sctp_bucket_cachep;
115 extern long sysctl_sctp_mem[3]; 115 extern long sysctl_sctp_mem[3];
116 extern int sysctl_sctp_rmem[3]; 116 extern int sysctl_sctp_rmem[3];
117 extern int sysctl_sctp_wmem[3]; 117 extern int sysctl_sctp_wmem[3];
118 118
119 static int sctp_memory_pressure; 119 static int sctp_memory_pressure;
120 static atomic_long_t sctp_memory_allocated; 120 static atomic_long_t sctp_memory_allocated;
121 struct percpu_counter sctp_sockets_allocated; 121 struct percpu_counter sctp_sockets_allocated;
122 122
123 static void sctp_enter_memory_pressure(struct sock *sk) 123 static void sctp_enter_memory_pressure(struct sock *sk)
124 { 124 {
125 sctp_memory_pressure = 1; 125 sctp_memory_pressure = 1;
126 } 126 }
127 127
128 128
129 /* Get the sndbuf space available at the time on the association. */ 129 /* Get the sndbuf space available at the time on the association. */
130 static inline int sctp_wspace(struct sctp_association *asoc) 130 static inline int sctp_wspace(struct sctp_association *asoc)
131 { 131 {
132 int amt; 132 int amt;
133 133
134 if (asoc->ep->sndbuf_policy) 134 if (asoc->ep->sndbuf_policy)
135 amt = asoc->sndbuf_used; 135 amt = asoc->sndbuf_used;
136 else 136 else
137 amt = sk_wmem_alloc_get(asoc->base.sk); 137 amt = sk_wmem_alloc_get(asoc->base.sk);
138 138
139 if (amt >= asoc->base.sk->sk_sndbuf) { 139 if (amt >= asoc->base.sk->sk_sndbuf) {
140 if (asoc->base.sk->sk_userlocks & SOCK_SNDBUF_LOCK) 140 if (asoc->base.sk->sk_userlocks & SOCK_SNDBUF_LOCK)
141 amt = 0; 141 amt = 0;
142 else { 142 else {
143 amt = sk_stream_wspace(asoc->base.sk); 143 amt = sk_stream_wspace(asoc->base.sk);
144 if (amt < 0) 144 if (amt < 0)
145 amt = 0; 145 amt = 0;
146 } 146 }
147 } else { 147 } else {
148 amt = asoc->base.sk->sk_sndbuf - amt; 148 amt = asoc->base.sk->sk_sndbuf - amt;
149 } 149 }
150 return amt; 150 return amt;
151 } 151 }
152 152
153 /* Increment the used sndbuf space count of the corresponding association by 153 /* Increment the used sndbuf space count of the corresponding association by
154 * the size of the outgoing data chunk. 154 * the size of the outgoing data chunk.
155 * Also, set the skb destructor for sndbuf accounting later. 155 * Also, set the skb destructor for sndbuf accounting later.
156 * 156 *
157 * Since it is always 1-1 between chunk and skb, and also a new skb is always 157 * Since it is always 1-1 between chunk and skb, and also a new skb is always
158 * allocated for chunk bundling in sctp_packet_transmit(), we can use the 158 * allocated for chunk bundling in sctp_packet_transmit(), we can use the
159 * destructor in the data chunk skb for the purpose of the sndbuf space 159 * destructor in the data chunk skb for the purpose of the sndbuf space
160 * tracking. 160 * tracking.
161 */ 161 */
162 static inline void sctp_set_owner_w(struct sctp_chunk *chunk) 162 static inline void sctp_set_owner_w(struct sctp_chunk *chunk)
163 { 163 {
164 struct sctp_association *asoc = chunk->asoc; 164 struct sctp_association *asoc = chunk->asoc;
165 struct sock *sk = asoc->base.sk; 165 struct sock *sk = asoc->base.sk;
166 166
167 /* The sndbuf space is tracked per association. */ 167 /* The sndbuf space is tracked per association. */
168 sctp_association_hold(asoc); 168 sctp_association_hold(asoc);
169 169
170 skb_set_owner_w(chunk->skb, sk); 170 skb_set_owner_w(chunk->skb, sk);
171 171
172 chunk->skb->destructor = sctp_wfree; 172 chunk->skb->destructor = sctp_wfree;
173 /* Save the chunk pointer in skb for sctp_wfree to use later. */ 173 /* Save the chunk pointer in skb for sctp_wfree to use later. */
174 *((struct sctp_chunk **)(chunk->skb->cb)) = chunk; 174 *((struct sctp_chunk **)(chunk->skb->cb)) = chunk;
175 175
176 asoc->sndbuf_used += SCTP_DATA_SNDSIZE(chunk) + 176 asoc->sndbuf_used += SCTP_DATA_SNDSIZE(chunk) +
177 sizeof(struct sk_buff) + 177 sizeof(struct sk_buff) +
178 sizeof(struct sctp_chunk); 178 sizeof(struct sctp_chunk);
179 179
180 atomic_add(sizeof(struct sctp_chunk), &sk->sk_wmem_alloc); 180 atomic_add(sizeof(struct sctp_chunk), &sk->sk_wmem_alloc);
181 sk->sk_wmem_queued += chunk->skb->truesize; 181 sk->sk_wmem_queued += chunk->skb->truesize;
182 sk_mem_charge(sk, chunk->skb->truesize); 182 sk_mem_charge(sk, chunk->skb->truesize);
183 } 183 }
184 184
185 /* Verify that this is a valid address. */ 185 /* Verify that this is a valid address. */
186 static inline int sctp_verify_addr(struct sock *sk, union sctp_addr *addr, 186 static inline int sctp_verify_addr(struct sock *sk, union sctp_addr *addr,
187 int len) 187 int len)
188 { 188 {
189 struct sctp_af *af; 189 struct sctp_af *af;
190 190
191 /* Verify basic sockaddr. */ 191 /* Verify basic sockaddr. */
192 af = sctp_sockaddr_af(sctp_sk(sk), addr, len); 192 af = sctp_sockaddr_af(sctp_sk(sk), addr, len);
193 if (!af) 193 if (!af)
194 return -EINVAL; 194 return -EINVAL;
195 195
196 /* Is this a valid SCTP address? */ 196 /* Is this a valid SCTP address? */
197 if (!af->addr_valid(addr, sctp_sk(sk), NULL)) 197 if (!af->addr_valid(addr, sctp_sk(sk), NULL))
198 return -EINVAL; 198 return -EINVAL;
199 199
200 if (!sctp_sk(sk)->pf->send_verify(sctp_sk(sk), (addr))) 200 if (!sctp_sk(sk)->pf->send_verify(sctp_sk(sk), (addr)))
201 return -EINVAL; 201 return -EINVAL;
202 202
203 return 0; 203 return 0;
204 } 204 }
205 205
206 /* Look up the association by its id. If this is not a UDP-style 206 /* Look up the association by its id. If this is not a UDP-style
207 * socket, the ID field is always ignored. 207 * socket, the ID field is always ignored.
208 */ 208 */
209 struct sctp_association *sctp_id2assoc(struct sock *sk, sctp_assoc_t id) 209 struct sctp_association *sctp_id2assoc(struct sock *sk, sctp_assoc_t id)
210 { 210 {
211 struct sctp_association *asoc = NULL; 211 struct sctp_association *asoc = NULL;
212 212
213 /* If this is not a UDP-style socket, assoc id should be ignored. */ 213 /* If this is not a UDP-style socket, assoc id should be ignored. */
214 if (!sctp_style(sk, UDP)) { 214 if (!sctp_style(sk, UDP)) {
215 /* Return NULL if the socket state is not ESTABLISHED. It 215 /* Return NULL if the socket state is not ESTABLISHED. It
216 * could be a TCP-style listening socket or a socket which 216 * could be a TCP-style listening socket or a socket which
217 * hasn't yet called connect() to establish an association. 217 * hasn't yet called connect() to establish an association.
218 */ 218 */
219 if (!sctp_sstate(sk, ESTABLISHED)) 219 if (!sctp_sstate(sk, ESTABLISHED))
220 return NULL; 220 return NULL;
221 221
222 /* Get the first and the only association from the list. */ 222 /* Get the first and the only association from the list. */
223 if (!list_empty(&sctp_sk(sk)->ep->asocs)) 223 if (!list_empty(&sctp_sk(sk)->ep->asocs))
224 asoc = list_entry(sctp_sk(sk)->ep->asocs.next, 224 asoc = list_entry(sctp_sk(sk)->ep->asocs.next,
225 struct sctp_association, asocs); 225 struct sctp_association, asocs);
226 return asoc; 226 return asoc;
227 } 227 }
228 228
229 /* Otherwise this is a UDP-style socket. */ 229 /* Otherwise this is a UDP-style socket. */
230 if (!id || (id == (sctp_assoc_t)-1)) 230 if (!id || (id == (sctp_assoc_t)-1))
231 return NULL; 231 return NULL;
232 232
233 spin_lock_bh(&sctp_assocs_id_lock); 233 spin_lock_bh(&sctp_assocs_id_lock);
234 asoc = (struct sctp_association *)idr_find(&sctp_assocs_id, (int)id); 234 asoc = (struct sctp_association *)idr_find(&sctp_assocs_id, (int)id);
235 spin_unlock_bh(&sctp_assocs_id_lock); 235 spin_unlock_bh(&sctp_assocs_id_lock);
236 236
237 if (!asoc || (asoc->base.sk != sk) || asoc->base.dead) 237 if (!asoc || (asoc->base.sk != sk) || asoc->base.dead)
238 return NULL; 238 return NULL;
239 239
240 return asoc; 240 return asoc;
241 } 241 }
242 242
243 /* Look up the transport from an address and an assoc id. If both address and 243 /* Look up the transport from an address and an assoc id. If both address and
244 * id are specified, the associations matching the address and the id should be 244 * id are specified, the associations matching the address and the id should be
245 * the same. 245 * the same.
246 */ 246 */
247 static struct sctp_transport *sctp_addr_id2transport(struct sock *sk, 247 static struct sctp_transport *sctp_addr_id2transport(struct sock *sk,
248 struct sockaddr_storage *addr, 248 struct sockaddr_storage *addr,
249 sctp_assoc_t id) 249 sctp_assoc_t id)
250 { 250 {
251 struct sctp_association *addr_asoc = NULL, *id_asoc = NULL; 251 struct sctp_association *addr_asoc = NULL, *id_asoc = NULL;
252 struct sctp_transport *transport; 252 struct sctp_transport *transport;
253 union sctp_addr *laddr = (union sctp_addr *)addr; 253 union sctp_addr *laddr = (union sctp_addr *)addr;
254 254
255 addr_asoc = sctp_endpoint_lookup_assoc(sctp_sk(sk)->ep, 255 addr_asoc = sctp_endpoint_lookup_assoc(sctp_sk(sk)->ep,
256 laddr, 256 laddr,
257 &transport); 257 &transport);
258 258
259 if (!addr_asoc) 259 if (!addr_asoc)
260 return NULL; 260 return NULL;
261 261
262 id_asoc = sctp_id2assoc(sk, id); 262 id_asoc = sctp_id2assoc(sk, id);
263 if (id_asoc && (id_asoc != addr_asoc)) 263 if (id_asoc && (id_asoc != addr_asoc))
264 return NULL; 264 return NULL;
265 265
266 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sctp_sk(sk), 266 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sctp_sk(sk),
267 (union sctp_addr *)addr); 267 (union sctp_addr *)addr);
268 268
269 return transport; 269 return transport;
270 } 270 }
271 271
272 /* API 3.1.2 bind() - UDP Style Syntax 272 /* API 3.1.2 bind() - UDP Style Syntax
273 * The syntax of bind() is, 273 * The syntax of bind() is,
274 * 274 *
275 * ret = bind(int sd, struct sockaddr *addr, int addrlen); 275 * ret = bind(int sd, struct sockaddr *addr, int addrlen);
276 * 276 *
277 * sd - the socket descriptor returned by socket(). 277 * sd - the socket descriptor returned by socket().
278 * addr - the address structure (struct sockaddr_in or struct 278 * addr - the address structure (struct sockaddr_in or struct
279 * sockaddr_in6 [RFC 2553]), 279 * sockaddr_in6 [RFC 2553]),
280 * addr_len - the size of the address structure. 280 * addr_len - the size of the address structure.
281 */ 281 */
282 SCTP_STATIC int sctp_bind(struct sock *sk, struct sockaddr *addr, int addr_len) 282 SCTP_STATIC int sctp_bind(struct sock *sk, struct sockaddr *addr, int addr_len)
283 { 283 {
284 int retval = 0; 284 int retval = 0;
285 285
286 sctp_lock_sock(sk); 286 sctp_lock_sock(sk);
287 287
288 SCTP_DEBUG_PRINTK("sctp_bind(sk: %p, addr: %p, addr_len: %d)\n", 288 SCTP_DEBUG_PRINTK("sctp_bind(sk: %p, addr: %p, addr_len: %d)\n",
289 sk, addr, addr_len); 289 sk, addr, addr_len);
290 290
291 /* Disallow binding twice. */ 291 /* Disallow binding twice. */
292 if (!sctp_sk(sk)->ep->base.bind_addr.port) 292 if (!sctp_sk(sk)->ep->base.bind_addr.port)
293 retval = sctp_do_bind(sk, (union sctp_addr *)addr, 293 retval = sctp_do_bind(sk, (union sctp_addr *)addr,
294 addr_len); 294 addr_len);
295 else 295 else
296 retval = -EINVAL; 296 retval = -EINVAL;
297 297
298 sctp_release_sock(sk); 298 sctp_release_sock(sk);
299 299
300 return retval; 300 return retval;
301 } 301 }
302 302
303 static long sctp_get_port_local(struct sock *, union sctp_addr *); 303 static long sctp_get_port_local(struct sock *, union sctp_addr *);
304 304
305 /* Verify this is a valid sockaddr. */ 305 /* Verify this is a valid sockaddr. */
306 static struct sctp_af *sctp_sockaddr_af(struct sctp_sock *opt, 306 static struct sctp_af *sctp_sockaddr_af(struct sctp_sock *opt,
307 union sctp_addr *addr, int len) 307 union sctp_addr *addr, int len)
308 { 308 {
309 struct sctp_af *af; 309 struct sctp_af *af;
310 310
311 /* Check minimum size. */ 311 /* Check minimum size. */
312 if (len < sizeof (struct sockaddr)) 312 if (len < sizeof (struct sockaddr))
313 return NULL; 313 return NULL;
314 314
315 /* V4 mapped address are really of AF_INET family */ 315 /* V4 mapped address are really of AF_INET family */
316 if (addr->sa.sa_family == AF_INET6 && 316 if (addr->sa.sa_family == AF_INET6 &&
317 ipv6_addr_v4mapped(&addr->v6.sin6_addr)) { 317 ipv6_addr_v4mapped(&addr->v6.sin6_addr)) {
318 if (!opt->pf->af_supported(AF_INET, opt)) 318 if (!opt->pf->af_supported(AF_INET, opt))
319 return NULL; 319 return NULL;
320 } else { 320 } else {
321 /* Does this PF support this AF? */ 321 /* Does this PF support this AF? */
322 if (!opt->pf->af_supported(addr->sa.sa_family, opt)) 322 if (!opt->pf->af_supported(addr->sa.sa_family, opt))
323 return NULL; 323 return NULL;
324 } 324 }
325 325
326 /* If we get this far, af is valid. */ 326 /* If we get this far, af is valid. */
327 af = sctp_get_af_specific(addr->sa.sa_family); 327 af = sctp_get_af_specific(addr->sa.sa_family);
328 328
329 if (len < af->sockaddr_len) 329 if (len < af->sockaddr_len)
330 return NULL; 330 return NULL;
331 331
332 return af; 332 return af;
333 } 333 }
334 334
335 /* Bind a local address either to an endpoint or to an association. */ 335 /* Bind a local address either to an endpoint or to an association. */
336 SCTP_STATIC int sctp_do_bind(struct sock *sk, union sctp_addr *addr, int len) 336 SCTP_STATIC int sctp_do_bind(struct sock *sk, union sctp_addr *addr, int len)
337 { 337 {
338 struct net *net = sock_net(sk);
338 struct sctp_sock *sp = sctp_sk(sk); 339 struct sctp_sock *sp = sctp_sk(sk);
339 struct sctp_endpoint *ep = sp->ep; 340 struct sctp_endpoint *ep = sp->ep;
340 struct sctp_bind_addr *bp = &ep->base.bind_addr; 341 struct sctp_bind_addr *bp = &ep->base.bind_addr;
341 struct sctp_af *af; 342 struct sctp_af *af;
342 unsigned short snum; 343 unsigned short snum;
343 int ret = 0; 344 int ret = 0;
344 345
345 /* Common sockaddr verification. */ 346 /* Common sockaddr verification. */
346 af = sctp_sockaddr_af(sp, addr, len); 347 af = sctp_sockaddr_af(sp, addr, len);
347 if (!af) { 348 if (!af) {
348 SCTP_DEBUG_PRINTK("sctp_do_bind(sk: %p, newaddr: %p, len: %d) EINVAL\n", 349 SCTP_DEBUG_PRINTK("sctp_do_bind(sk: %p, newaddr: %p, len: %d) EINVAL\n",
349 sk, addr, len); 350 sk, addr, len);
350 return -EINVAL; 351 return -EINVAL;
351 } 352 }
352 353
353 snum = ntohs(addr->v4.sin_port); 354 snum = ntohs(addr->v4.sin_port);
354 355
355 SCTP_DEBUG_PRINTK_IPADDR("sctp_do_bind(sk: %p, new addr: ", 356 SCTP_DEBUG_PRINTK_IPADDR("sctp_do_bind(sk: %p, new addr: ",
356 ", port: %d, new port: %d, len: %d)\n", 357 ", port: %d, new port: %d, len: %d)\n",
357 sk, 358 sk,
358 addr, 359 addr,
359 bp->port, snum, 360 bp->port, snum,
360 len); 361 len);
361 362
362 /* PF specific bind() address verification. */ 363 /* PF specific bind() address verification. */
363 if (!sp->pf->bind_verify(sp, addr)) 364 if (!sp->pf->bind_verify(sp, addr))
364 return -EADDRNOTAVAIL; 365 return -EADDRNOTAVAIL;
365 366
366 /* We must either be unbound, or bind to the same port. 367 /* We must either be unbound, or bind to the same port.
367 * It's OK to allow 0 ports if we are already bound. 368 * It's OK to allow 0 ports if we are already bound.
368 * We'll just inhert an already bound port in this case 369 * We'll just inhert an already bound port in this case
369 */ 370 */
370 if (bp->port) { 371 if (bp->port) {
371 if (!snum) 372 if (!snum)
372 snum = bp->port; 373 snum = bp->port;
373 else if (snum != bp->port) { 374 else if (snum != bp->port) {
374 SCTP_DEBUG_PRINTK("sctp_do_bind:" 375 SCTP_DEBUG_PRINTK("sctp_do_bind:"
375 " New port %d does not match existing port " 376 " New port %d does not match existing port "
376 "%d.\n", snum, bp->port); 377 "%d.\n", snum, bp->port);
377 return -EINVAL; 378 return -EINVAL;
378 } 379 }
379 } 380 }
380 381
381 if (snum && snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE)) 382 if (snum && snum < PROT_SOCK &&
383 !ns_capable(net->user_ns, CAP_NET_BIND_SERVICE))
382 return -EACCES; 384 return -EACCES;
383 385
384 /* See if the address matches any of the addresses we may have 386 /* See if the address matches any of the addresses we may have
385 * already bound before checking against other endpoints. 387 * already bound before checking against other endpoints.
386 */ 388 */
387 if (sctp_bind_addr_match(bp, addr, sp)) 389 if (sctp_bind_addr_match(bp, addr, sp))
388 return -EINVAL; 390 return -EINVAL;
389 391
390 /* Make sure we are allowed to bind here. 392 /* Make sure we are allowed to bind here.
391 * The function sctp_get_port_local() does duplicate address 393 * The function sctp_get_port_local() does duplicate address
392 * detection. 394 * detection.
393 */ 395 */
394 addr->v4.sin_port = htons(snum); 396 addr->v4.sin_port = htons(snum);
395 if ((ret = sctp_get_port_local(sk, addr))) { 397 if ((ret = sctp_get_port_local(sk, addr))) {
396 return -EADDRINUSE; 398 return -EADDRINUSE;
397 } 399 }
398 400
399 /* Refresh ephemeral port. */ 401 /* Refresh ephemeral port. */
400 if (!bp->port) 402 if (!bp->port)
401 bp->port = inet_sk(sk)->inet_num; 403 bp->port = inet_sk(sk)->inet_num;
402 404
403 /* Add the address to the bind address list. 405 /* Add the address to the bind address list.
404 * Use GFP_ATOMIC since BHs will be disabled. 406 * Use GFP_ATOMIC since BHs will be disabled.
405 */ 407 */
406 ret = sctp_add_bind_addr(bp, addr, SCTP_ADDR_SRC, GFP_ATOMIC); 408 ret = sctp_add_bind_addr(bp, addr, SCTP_ADDR_SRC, GFP_ATOMIC);
407 409
408 /* Copy back into socket for getsockname() use. */ 410 /* Copy back into socket for getsockname() use. */
409 if (!ret) { 411 if (!ret) {
410 inet_sk(sk)->inet_sport = htons(inet_sk(sk)->inet_num); 412 inet_sk(sk)->inet_sport = htons(inet_sk(sk)->inet_num);
411 af->to_sk_saddr(addr, sk); 413 af->to_sk_saddr(addr, sk);
412 } 414 }
413 415
414 return ret; 416 return ret;
415 } 417 }
416 418
417 /* ADDIP Section 4.1.1 Congestion Control of ASCONF Chunks 419 /* ADDIP Section 4.1.1 Congestion Control of ASCONF Chunks
418 * 420 *
419 * R1) One and only one ASCONF Chunk MAY be in transit and unacknowledged 421 * R1) One and only one ASCONF Chunk MAY be in transit and unacknowledged
420 * at any one time. If a sender, after sending an ASCONF chunk, decides 422 * at any one time. If a sender, after sending an ASCONF chunk, decides
421 * it needs to transfer another ASCONF Chunk, it MUST wait until the 423 * it needs to transfer another ASCONF Chunk, it MUST wait until the
422 * ASCONF-ACK Chunk returns from the previous ASCONF Chunk before sending a 424 * ASCONF-ACK Chunk returns from the previous ASCONF Chunk before sending a
423 * subsequent ASCONF. Note this restriction binds each side, so at any 425 * subsequent ASCONF. Note this restriction binds each side, so at any
424 * time two ASCONF may be in-transit on any given association (one sent 426 * time two ASCONF may be in-transit on any given association (one sent
425 * from each endpoint). 427 * from each endpoint).
426 */ 428 */
427 static int sctp_send_asconf(struct sctp_association *asoc, 429 static int sctp_send_asconf(struct sctp_association *asoc,
428 struct sctp_chunk *chunk) 430 struct sctp_chunk *chunk)
429 { 431 {
430 struct net *net = sock_net(asoc->base.sk); 432 struct net *net = sock_net(asoc->base.sk);
431 int retval = 0; 433 int retval = 0;
432 434
433 /* If there is an outstanding ASCONF chunk, queue it for later 435 /* If there is an outstanding ASCONF chunk, queue it for later
434 * transmission. 436 * transmission.
435 */ 437 */
436 if (asoc->addip_last_asconf) { 438 if (asoc->addip_last_asconf) {
437 list_add_tail(&chunk->list, &asoc->addip_chunk_list); 439 list_add_tail(&chunk->list, &asoc->addip_chunk_list);
438 goto out; 440 goto out;
439 } 441 }
440 442
441 /* Hold the chunk until an ASCONF_ACK is received. */ 443 /* Hold the chunk until an ASCONF_ACK is received. */
442 sctp_chunk_hold(chunk); 444 sctp_chunk_hold(chunk);
443 retval = sctp_primitive_ASCONF(net, asoc, chunk); 445 retval = sctp_primitive_ASCONF(net, asoc, chunk);
444 if (retval) 446 if (retval)
445 sctp_chunk_free(chunk); 447 sctp_chunk_free(chunk);
446 else 448 else
447 asoc->addip_last_asconf = chunk; 449 asoc->addip_last_asconf = chunk;
448 450
449 out: 451 out:
450 return retval; 452 return retval;
451 } 453 }
452 454
453 /* Add a list of addresses as bind addresses to local endpoint or 455 /* Add a list of addresses as bind addresses to local endpoint or
454 * association. 456 * association.
455 * 457 *
456 * Basically run through each address specified in the addrs/addrcnt 458 * Basically run through each address specified in the addrs/addrcnt
457 * array/length pair, determine if it is IPv6 or IPv4 and call 459 * array/length pair, determine if it is IPv6 or IPv4 and call
458 * sctp_do_bind() on it. 460 * sctp_do_bind() on it.
459 * 461 *
460 * If any of them fails, then the operation will be reversed and the 462 * If any of them fails, then the operation will be reversed and the
461 * ones that were added will be removed. 463 * ones that were added will be removed.
462 * 464 *
463 * Only sctp_setsockopt_bindx() is supposed to call this function. 465 * Only sctp_setsockopt_bindx() is supposed to call this function.
464 */ 466 */
465 static int sctp_bindx_add(struct sock *sk, struct sockaddr *addrs, int addrcnt) 467 static int sctp_bindx_add(struct sock *sk, struct sockaddr *addrs, int addrcnt)
466 { 468 {
467 int cnt; 469 int cnt;
468 int retval = 0; 470 int retval = 0;
469 void *addr_buf; 471 void *addr_buf;
470 struct sockaddr *sa_addr; 472 struct sockaddr *sa_addr;
471 struct sctp_af *af; 473 struct sctp_af *af;
472 474
473 SCTP_DEBUG_PRINTK("sctp_bindx_add (sk: %p, addrs: %p, addrcnt: %d)\n", 475 SCTP_DEBUG_PRINTK("sctp_bindx_add (sk: %p, addrs: %p, addrcnt: %d)\n",
474 sk, addrs, addrcnt); 476 sk, addrs, addrcnt);
475 477
476 addr_buf = addrs; 478 addr_buf = addrs;
477 for (cnt = 0; cnt < addrcnt; cnt++) { 479 for (cnt = 0; cnt < addrcnt; cnt++) {
478 /* The list may contain either IPv4 or IPv6 address; 480 /* The list may contain either IPv4 or IPv6 address;
479 * determine the address length for walking thru the list. 481 * determine the address length for walking thru the list.
480 */ 482 */
481 sa_addr = addr_buf; 483 sa_addr = addr_buf;
482 af = sctp_get_af_specific(sa_addr->sa_family); 484 af = sctp_get_af_specific(sa_addr->sa_family);
483 if (!af) { 485 if (!af) {
484 retval = -EINVAL; 486 retval = -EINVAL;
485 goto err_bindx_add; 487 goto err_bindx_add;
486 } 488 }
487 489
488 retval = sctp_do_bind(sk, (union sctp_addr *)sa_addr, 490 retval = sctp_do_bind(sk, (union sctp_addr *)sa_addr,
489 af->sockaddr_len); 491 af->sockaddr_len);
490 492
491 addr_buf += af->sockaddr_len; 493 addr_buf += af->sockaddr_len;
492 494
493 err_bindx_add: 495 err_bindx_add:
494 if (retval < 0) { 496 if (retval < 0) {
495 /* Failed. Cleanup the ones that have been added */ 497 /* Failed. Cleanup the ones that have been added */
496 if (cnt > 0) 498 if (cnt > 0)
497 sctp_bindx_rem(sk, addrs, cnt); 499 sctp_bindx_rem(sk, addrs, cnt);
498 return retval; 500 return retval;
499 } 501 }
500 } 502 }
501 503
502 return retval; 504 return retval;
503 } 505 }
504 506
505 /* Send an ASCONF chunk with Add IP address parameters to all the peers of the 507 /* Send an ASCONF chunk with Add IP address parameters to all the peers of the
506 * associations that are part of the endpoint indicating that a list of local 508 * associations that are part of the endpoint indicating that a list of local
507 * addresses are added to the endpoint. 509 * addresses are added to the endpoint.
508 * 510 *
509 * If any of the addresses is already in the bind address list of the 511 * If any of the addresses is already in the bind address list of the
510 * association, we do not send the chunk for that association. But it will not 512 * association, we do not send the chunk for that association. But it will not
511 * affect other associations. 513 * affect other associations.
512 * 514 *
513 * Only sctp_setsockopt_bindx() is supposed to call this function. 515 * Only sctp_setsockopt_bindx() is supposed to call this function.
514 */ 516 */
515 static int sctp_send_asconf_add_ip(struct sock *sk, 517 static int sctp_send_asconf_add_ip(struct sock *sk,
516 struct sockaddr *addrs, 518 struct sockaddr *addrs,
517 int addrcnt) 519 int addrcnt)
518 { 520 {
519 struct net *net = sock_net(sk); 521 struct net *net = sock_net(sk);
520 struct sctp_sock *sp; 522 struct sctp_sock *sp;
521 struct sctp_endpoint *ep; 523 struct sctp_endpoint *ep;
522 struct sctp_association *asoc; 524 struct sctp_association *asoc;
523 struct sctp_bind_addr *bp; 525 struct sctp_bind_addr *bp;
524 struct sctp_chunk *chunk; 526 struct sctp_chunk *chunk;
525 struct sctp_sockaddr_entry *laddr; 527 struct sctp_sockaddr_entry *laddr;
526 union sctp_addr *addr; 528 union sctp_addr *addr;
527 union sctp_addr saveaddr; 529 union sctp_addr saveaddr;
528 void *addr_buf; 530 void *addr_buf;
529 struct sctp_af *af; 531 struct sctp_af *af;
530 struct list_head *p; 532 struct list_head *p;
531 int i; 533 int i;
532 int retval = 0; 534 int retval = 0;
533 535
534 if (!net->sctp.addip_enable) 536 if (!net->sctp.addip_enable)
535 return retval; 537 return retval;
536 538
537 sp = sctp_sk(sk); 539 sp = sctp_sk(sk);
538 ep = sp->ep; 540 ep = sp->ep;
539 541
540 SCTP_DEBUG_PRINTK("%s: (sk: %p, addrs: %p, addrcnt: %d)\n", 542 SCTP_DEBUG_PRINTK("%s: (sk: %p, addrs: %p, addrcnt: %d)\n",
541 __func__, sk, addrs, addrcnt); 543 __func__, sk, addrs, addrcnt);
542 544
543 list_for_each_entry(asoc, &ep->asocs, asocs) { 545 list_for_each_entry(asoc, &ep->asocs, asocs) {
544 546
545 if (!asoc->peer.asconf_capable) 547 if (!asoc->peer.asconf_capable)
546 continue; 548 continue;
547 549
548 if (asoc->peer.addip_disabled_mask & SCTP_PARAM_ADD_IP) 550 if (asoc->peer.addip_disabled_mask & SCTP_PARAM_ADD_IP)
549 continue; 551 continue;
550 552
551 if (!sctp_state(asoc, ESTABLISHED)) 553 if (!sctp_state(asoc, ESTABLISHED))
552 continue; 554 continue;
553 555
554 /* Check if any address in the packed array of addresses is 556 /* Check if any address in the packed array of addresses is
555 * in the bind address list of the association. If so, 557 * in the bind address list of the association. If so,
556 * do not send the asconf chunk to its peer, but continue with 558 * do not send the asconf chunk to its peer, but continue with
557 * other associations. 559 * other associations.
558 */ 560 */
559 addr_buf = addrs; 561 addr_buf = addrs;
560 for (i = 0; i < addrcnt; i++) { 562 for (i = 0; i < addrcnt; i++) {
561 addr = addr_buf; 563 addr = addr_buf;
562 af = sctp_get_af_specific(addr->v4.sin_family); 564 af = sctp_get_af_specific(addr->v4.sin_family);
563 if (!af) { 565 if (!af) {
564 retval = -EINVAL; 566 retval = -EINVAL;
565 goto out; 567 goto out;
566 } 568 }
567 569
568 if (sctp_assoc_lookup_laddr(asoc, addr)) 570 if (sctp_assoc_lookup_laddr(asoc, addr))
569 break; 571 break;
570 572
571 addr_buf += af->sockaddr_len; 573 addr_buf += af->sockaddr_len;
572 } 574 }
573 if (i < addrcnt) 575 if (i < addrcnt)
574 continue; 576 continue;
575 577
576 /* Use the first valid address in bind addr list of 578 /* Use the first valid address in bind addr list of
577 * association as Address Parameter of ASCONF CHUNK. 579 * association as Address Parameter of ASCONF CHUNK.
578 */ 580 */
579 bp = &asoc->base.bind_addr; 581 bp = &asoc->base.bind_addr;
580 p = bp->address_list.next; 582 p = bp->address_list.next;
581 laddr = list_entry(p, struct sctp_sockaddr_entry, list); 583 laddr = list_entry(p, struct sctp_sockaddr_entry, list);
582 chunk = sctp_make_asconf_update_ip(asoc, &laddr->a, addrs, 584 chunk = sctp_make_asconf_update_ip(asoc, &laddr->a, addrs,
583 addrcnt, SCTP_PARAM_ADD_IP); 585 addrcnt, SCTP_PARAM_ADD_IP);
584 if (!chunk) { 586 if (!chunk) {
585 retval = -ENOMEM; 587 retval = -ENOMEM;
586 goto out; 588 goto out;
587 } 589 }
588 590
589 /* Add the new addresses to the bind address list with 591 /* Add the new addresses to the bind address list with
590 * use_as_src set to 0. 592 * use_as_src set to 0.
591 */ 593 */
592 addr_buf = addrs; 594 addr_buf = addrs;
593 for (i = 0; i < addrcnt; i++) { 595 for (i = 0; i < addrcnt; i++) {
594 addr = addr_buf; 596 addr = addr_buf;
595 af = sctp_get_af_specific(addr->v4.sin_family); 597 af = sctp_get_af_specific(addr->v4.sin_family);
596 memcpy(&saveaddr, addr, af->sockaddr_len); 598 memcpy(&saveaddr, addr, af->sockaddr_len);
597 retval = sctp_add_bind_addr(bp, &saveaddr, 599 retval = sctp_add_bind_addr(bp, &saveaddr,
598 SCTP_ADDR_NEW, GFP_ATOMIC); 600 SCTP_ADDR_NEW, GFP_ATOMIC);
599 addr_buf += af->sockaddr_len; 601 addr_buf += af->sockaddr_len;
600 } 602 }
601 if (asoc->src_out_of_asoc_ok) { 603 if (asoc->src_out_of_asoc_ok) {
602 struct sctp_transport *trans; 604 struct sctp_transport *trans;
603 605
604 list_for_each_entry(trans, 606 list_for_each_entry(trans,
605 &asoc->peer.transport_addr_list, transports) { 607 &asoc->peer.transport_addr_list, transports) {
606 /* Clear the source and route cache */ 608 /* Clear the source and route cache */
607 dst_release(trans->dst); 609 dst_release(trans->dst);
608 trans->cwnd = min(4*asoc->pathmtu, max_t(__u32, 610 trans->cwnd = min(4*asoc->pathmtu, max_t(__u32,
609 2*asoc->pathmtu, 4380)); 611 2*asoc->pathmtu, 4380));
610 trans->ssthresh = asoc->peer.i.a_rwnd; 612 trans->ssthresh = asoc->peer.i.a_rwnd;
611 trans->rto = asoc->rto_initial; 613 trans->rto = asoc->rto_initial;
612 trans->rtt = trans->srtt = trans->rttvar = 0; 614 trans->rtt = trans->srtt = trans->rttvar = 0;
613 sctp_transport_route(trans, NULL, 615 sctp_transport_route(trans, NULL,
614 sctp_sk(asoc->base.sk)); 616 sctp_sk(asoc->base.sk));
615 } 617 }
616 } 618 }
617 retval = sctp_send_asconf(asoc, chunk); 619 retval = sctp_send_asconf(asoc, chunk);
618 } 620 }
619 621
620 out: 622 out:
621 return retval; 623 return retval;
622 } 624 }
623 625
624 /* Remove a list of addresses from bind addresses list. Do not remove the 626 /* Remove a list of addresses from bind addresses list. Do not remove the
625 * last address. 627 * last address.
626 * 628 *
627 * Basically run through each address specified in the addrs/addrcnt 629 * Basically run through each address specified in the addrs/addrcnt
628 * array/length pair, determine if it is IPv6 or IPv4 and call 630 * array/length pair, determine if it is IPv6 or IPv4 and call
629 * sctp_del_bind() on it. 631 * sctp_del_bind() on it.
630 * 632 *
631 * If any of them fails, then the operation will be reversed and the 633 * If any of them fails, then the operation will be reversed and the
632 * ones that were removed will be added back. 634 * ones that were removed will be added back.
633 * 635 *
634 * At least one address has to be left; if only one address is 636 * At least one address has to be left; if only one address is
635 * available, the operation will return -EBUSY. 637 * available, the operation will return -EBUSY.
636 * 638 *
637 * Only sctp_setsockopt_bindx() is supposed to call this function. 639 * Only sctp_setsockopt_bindx() is supposed to call this function.
638 */ 640 */
639 static int sctp_bindx_rem(struct sock *sk, struct sockaddr *addrs, int addrcnt) 641 static int sctp_bindx_rem(struct sock *sk, struct sockaddr *addrs, int addrcnt)
640 { 642 {
641 struct sctp_sock *sp = sctp_sk(sk); 643 struct sctp_sock *sp = sctp_sk(sk);
642 struct sctp_endpoint *ep = sp->ep; 644 struct sctp_endpoint *ep = sp->ep;
643 int cnt; 645 int cnt;
644 struct sctp_bind_addr *bp = &ep->base.bind_addr; 646 struct sctp_bind_addr *bp = &ep->base.bind_addr;
645 int retval = 0; 647 int retval = 0;
646 void *addr_buf; 648 void *addr_buf;
647 union sctp_addr *sa_addr; 649 union sctp_addr *sa_addr;
648 struct sctp_af *af; 650 struct sctp_af *af;
649 651
650 SCTP_DEBUG_PRINTK("sctp_bindx_rem (sk: %p, addrs: %p, addrcnt: %d)\n", 652 SCTP_DEBUG_PRINTK("sctp_bindx_rem (sk: %p, addrs: %p, addrcnt: %d)\n",
651 sk, addrs, addrcnt); 653 sk, addrs, addrcnt);
652 654
653 addr_buf = addrs; 655 addr_buf = addrs;
654 for (cnt = 0; cnt < addrcnt; cnt++) { 656 for (cnt = 0; cnt < addrcnt; cnt++) {
655 /* If the bind address list is empty or if there is only one 657 /* If the bind address list is empty or if there is only one
656 * bind address, there is nothing more to be removed (we need 658 * bind address, there is nothing more to be removed (we need
657 * at least one address here). 659 * at least one address here).
658 */ 660 */
659 if (list_empty(&bp->address_list) || 661 if (list_empty(&bp->address_list) ||
660 (sctp_list_single_entry(&bp->address_list))) { 662 (sctp_list_single_entry(&bp->address_list))) {
661 retval = -EBUSY; 663 retval = -EBUSY;
662 goto err_bindx_rem; 664 goto err_bindx_rem;
663 } 665 }
664 666
665 sa_addr = addr_buf; 667 sa_addr = addr_buf;
666 af = sctp_get_af_specific(sa_addr->sa.sa_family); 668 af = sctp_get_af_specific(sa_addr->sa.sa_family);
667 if (!af) { 669 if (!af) {
668 retval = -EINVAL; 670 retval = -EINVAL;
669 goto err_bindx_rem; 671 goto err_bindx_rem;
670 } 672 }
671 673
672 if (!af->addr_valid(sa_addr, sp, NULL)) { 674 if (!af->addr_valid(sa_addr, sp, NULL)) {
673 retval = -EADDRNOTAVAIL; 675 retval = -EADDRNOTAVAIL;
674 goto err_bindx_rem; 676 goto err_bindx_rem;
675 } 677 }
676 678
677 if (sa_addr->v4.sin_port && 679 if (sa_addr->v4.sin_port &&
678 sa_addr->v4.sin_port != htons(bp->port)) { 680 sa_addr->v4.sin_port != htons(bp->port)) {
679 retval = -EINVAL; 681 retval = -EINVAL;
680 goto err_bindx_rem; 682 goto err_bindx_rem;
681 } 683 }
682 684
683 if (!sa_addr->v4.sin_port) 685 if (!sa_addr->v4.sin_port)
684 sa_addr->v4.sin_port = htons(bp->port); 686 sa_addr->v4.sin_port = htons(bp->port);
685 687
686 /* FIXME - There is probably a need to check if sk->sk_saddr and 688 /* FIXME - There is probably a need to check if sk->sk_saddr and
687 * sk->sk_rcv_addr are currently set to one of the addresses to 689 * sk->sk_rcv_addr are currently set to one of the addresses to
688 * be removed. This is something which needs to be looked into 690 * be removed. This is something which needs to be looked into
689 * when we are fixing the outstanding issues with multi-homing 691 * when we are fixing the outstanding issues with multi-homing
690 * socket routing and failover schemes. Refer to comments in 692 * socket routing and failover schemes. Refer to comments in
691 * sctp_do_bind(). -daisy 693 * sctp_do_bind(). -daisy
692 */ 694 */
693 retval = sctp_del_bind_addr(bp, sa_addr); 695 retval = sctp_del_bind_addr(bp, sa_addr);
694 696
695 addr_buf += af->sockaddr_len; 697 addr_buf += af->sockaddr_len;
696 err_bindx_rem: 698 err_bindx_rem:
697 if (retval < 0) { 699 if (retval < 0) {
698 /* Failed. Add the ones that has been removed back */ 700 /* Failed. Add the ones that has been removed back */
699 if (cnt > 0) 701 if (cnt > 0)
700 sctp_bindx_add(sk, addrs, cnt); 702 sctp_bindx_add(sk, addrs, cnt);
701 return retval; 703 return retval;
702 } 704 }
703 } 705 }
704 706
705 return retval; 707 return retval;
706 } 708 }
707 709
708 /* Send an ASCONF chunk with Delete IP address parameters to all the peers of 710 /* Send an ASCONF chunk with Delete IP address parameters to all the peers of
709 * the associations that are part of the endpoint indicating that a list of 711 * the associations that are part of the endpoint indicating that a list of
710 * local addresses are removed from the endpoint. 712 * local addresses are removed from the endpoint.
711 * 713 *
712 * If any of the addresses is already in the bind address list of the 714 * If any of the addresses is already in the bind address list of the
713 * association, we do not send the chunk for that association. But it will not 715 * association, we do not send the chunk for that association. But it will not
714 * affect other associations. 716 * affect other associations.
715 * 717 *
716 * Only sctp_setsockopt_bindx() is supposed to call this function. 718 * Only sctp_setsockopt_bindx() is supposed to call this function.
717 */ 719 */
718 static int sctp_send_asconf_del_ip(struct sock *sk, 720 static int sctp_send_asconf_del_ip(struct sock *sk,
719 struct sockaddr *addrs, 721 struct sockaddr *addrs,
720 int addrcnt) 722 int addrcnt)
721 { 723 {
722 struct net *net = sock_net(sk); 724 struct net *net = sock_net(sk);
723 struct sctp_sock *sp; 725 struct sctp_sock *sp;
724 struct sctp_endpoint *ep; 726 struct sctp_endpoint *ep;
725 struct sctp_association *asoc; 727 struct sctp_association *asoc;
726 struct sctp_transport *transport; 728 struct sctp_transport *transport;
727 struct sctp_bind_addr *bp; 729 struct sctp_bind_addr *bp;
728 struct sctp_chunk *chunk; 730 struct sctp_chunk *chunk;
729 union sctp_addr *laddr; 731 union sctp_addr *laddr;
730 void *addr_buf; 732 void *addr_buf;
731 struct sctp_af *af; 733 struct sctp_af *af;
732 struct sctp_sockaddr_entry *saddr; 734 struct sctp_sockaddr_entry *saddr;
733 int i; 735 int i;
734 int retval = 0; 736 int retval = 0;
735 int stored = 0; 737 int stored = 0;
736 738
737 chunk = NULL; 739 chunk = NULL;
738 if (!net->sctp.addip_enable) 740 if (!net->sctp.addip_enable)
739 return retval; 741 return retval;
740 742
741 sp = sctp_sk(sk); 743 sp = sctp_sk(sk);
742 ep = sp->ep; 744 ep = sp->ep;
743 745
744 SCTP_DEBUG_PRINTK("%s: (sk: %p, addrs: %p, addrcnt: %d)\n", 746 SCTP_DEBUG_PRINTK("%s: (sk: %p, addrs: %p, addrcnt: %d)\n",
745 __func__, sk, addrs, addrcnt); 747 __func__, sk, addrs, addrcnt);
746 748
747 list_for_each_entry(asoc, &ep->asocs, asocs) { 749 list_for_each_entry(asoc, &ep->asocs, asocs) {
748 750
749 if (!asoc->peer.asconf_capable) 751 if (!asoc->peer.asconf_capable)
750 continue; 752 continue;
751 753
752 if (asoc->peer.addip_disabled_mask & SCTP_PARAM_DEL_IP) 754 if (asoc->peer.addip_disabled_mask & SCTP_PARAM_DEL_IP)
753 continue; 755 continue;
754 756
755 if (!sctp_state(asoc, ESTABLISHED)) 757 if (!sctp_state(asoc, ESTABLISHED))
756 continue; 758 continue;
757 759
758 /* Check if any address in the packed array of addresses is 760 /* Check if any address in the packed array of addresses is
759 * not present in the bind address list of the association. 761 * not present in the bind address list of the association.
760 * If so, do not send the asconf chunk to its peer, but 762 * If so, do not send the asconf chunk to its peer, but
761 * continue with other associations. 763 * continue with other associations.
762 */ 764 */
763 addr_buf = addrs; 765 addr_buf = addrs;
764 for (i = 0; i < addrcnt; i++) { 766 for (i = 0; i < addrcnt; i++) {
765 laddr = addr_buf; 767 laddr = addr_buf;
766 af = sctp_get_af_specific(laddr->v4.sin_family); 768 af = sctp_get_af_specific(laddr->v4.sin_family);
767 if (!af) { 769 if (!af) {
768 retval = -EINVAL; 770 retval = -EINVAL;
769 goto out; 771 goto out;
770 } 772 }
771 773
772 if (!sctp_assoc_lookup_laddr(asoc, laddr)) 774 if (!sctp_assoc_lookup_laddr(asoc, laddr))
773 break; 775 break;
774 776
775 addr_buf += af->sockaddr_len; 777 addr_buf += af->sockaddr_len;
776 } 778 }
777 if (i < addrcnt) 779 if (i < addrcnt)
778 continue; 780 continue;
779 781
780 /* Find one address in the association's bind address list 782 /* Find one address in the association's bind address list
781 * that is not in the packed array of addresses. This is to 783 * that is not in the packed array of addresses. This is to
782 * make sure that we do not delete all the addresses in the 784 * make sure that we do not delete all the addresses in the
783 * association. 785 * association.
784 */ 786 */
785 bp = &asoc->base.bind_addr; 787 bp = &asoc->base.bind_addr;
786 laddr = sctp_find_unmatch_addr(bp, (union sctp_addr *)addrs, 788 laddr = sctp_find_unmatch_addr(bp, (union sctp_addr *)addrs,
787 addrcnt, sp); 789 addrcnt, sp);
788 if ((laddr == NULL) && (addrcnt == 1)) { 790 if ((laddr == NULL) && (addrcnt == 1)) {
789 if (asoc->asconf_addr_del_pending) 791 if (asoc->asconf_addr_del_pending)
790 continue; 792 continue;
791 asoc->asconf_addr_del_pending = 793 asoc->asconf_addr_del_pending =
792 kzalloc(sizeof(union sctp_addr), GFP_ATOMIC); 794 kzalloc(sizeof(union sctp_addr), GFP_ATOMIC);
793 if (asoc->asconf_addr_del_pending == NULL) { 795 if (asoc->asconf_addr_del_pending == NULL) {
794 retval = -ENOMEM; 796 retval = -ENOMEM;
795 goto out; 797 goto out;
796 } 798 }
797 asoc->asconf_addr_del_pending->sa.sa_family = 799 asoc->asconf_addr_del_pending->sa.sa_family =
798 addrs->sa_family; 800 addrs->sa_family;
799 asoc->asconf_addr_del_pending->v4.sin_port = 801 asoc->asconf_addr_del_pending->v4.sin_port =
800 htons(bp->port); 802 htons(bp->port);
801 if (addrs->sa_family == AF_INET) { 803 if (addrs->sa_family == AF_INET) {
802 struct sockaddr_in *sin; 804 struct sockaddr_in *sin;
803 805
804 sin = (struct sockaddr_in *)addrs; 806 sin = (struct sockaddr_in *)addrs;
805 asoc->asconf_addr_del_pending->v4.sin_addr.s_addr = sin->sin_addr.s_addr; 807 asoc->asconf_addr_del_pending->v4.sin_addr.s_addr = sin->sin_addr.s_addr;
806 } else if (addrs->sa_family == AF_INET6) { 808 } else if (addrs->sa_family == AF_INET6) {
807 struct sockaddr_in6 *sin6; 809 struct sockaddr_in6 *sin6;
808 810
809 sin6 = (struct sockaddr_in6 *)addrs; 811 sin6 = (struct sockaddr_in6 *)addrs;
810 asoc->asconf_addr_del_pending->v6.sin6_addr = sin6->sin6_addr; 812 asoc->asconf_addr_del_pending->v6.sin6_addr = sin6->sin6_addr;
811 } 813 }
812 SCTP_DEBUG_PRINTK_IPADDR("send_asconf_del_ip: keep the last address asoc: %p ", 814 SCTP_DEBUG_PRINTK_IPADDR("send_asconf_del_ip: keep the last address asoc: %p ",
813 " at %p\n", asoc, asoc->asconf_addr_del_pending, 815 " at %p\n", asoc, asoc->asconf_addr_del_pending,
814 asoc->asconf_addr_del_pending); 816 asoc->asconf_addr_del_pending);
815 asoc->src_out_of_asoc_ok = 1; 817 asoc->src_out_of_asoc_ok = 1;
816 stored = 1; 818 stored = 1;
817 goto skip_mkasconf; 819 goto skip_mkasconf;
818 } 820 }
819 821
820 /* We do not need RCU protection throughout this loop 822 /* We do not need RCU protection throughout this loop
821 * because this is done under a socket lock from the 823 * because this is done under a socket lock from the
822 * setsockopt call. 824 * setsockopt call.
823 */ 825 */
824 chunk = sctp_make_asconf_update_ip(asoc, laddr, addrs, addrcnt, 826 chunk = sctp_make_asconf_update_ip(asoc, laddr, addrs, addrcnt,
825 SCTP_PARAM_DEL_IP); 827 SCTP_PARAM_DEL_IP);
826 if (!chunk) { 828 if (!chunk) {
827 retval = -ENOMEM; 829 retval = -ENOMEM;
828 goto out; 830 goto out;
829 } 831 }
830 832
831 skip_mkasconf: 833 skip_mkasconf:
832 /* Reset use_as_src flag for the addresses in the bind address 834 /* Reset use_as_src flag for the addresses in the bind address
833 * list that are to be deleted. 835 * list that are to be deleted.
834 */ 836 */
835 addr_buf = addrs; 837 addr_buf = addrs;
836 for (i = 0; i < addrcnt; i++) { 838 for (i = 0; i < addrcnt; i++) {
837 laddr = addr_buf; 839 laddr = addr_buf;
838 af = sctp_get_af_specific(laddr->v4.sin_family); 840 af = sctp_get_af_specific(laddr->v4.sin_family);
839 list_for_each_entry(saddr, &bp->address_list, list) { 841 list_for_each_entry(saddr, &bp->address_list, list) {
840 if (sctp_cmp_addr_exact(&saddr->a, laddr)) 842 if (sctp_cmp_addr_exact(&saddr->a, laddr))
841 saddr->state = SCTP_ADDR_DEL; 843 saddr->state = SCTP_ADDR_DEL;
842 } 844 }
843 addr_buf += af->sockaddr_len; 845 addr_buf += af->sockaddr_len;
844 } 846 }
845 847
846 /* Update the route and saddr entries for all the transports 848 /* Update the route and saddr entries for all the transports
847 * as some of the addresses in the bind address list are 849 * as some of the addresses in the bind address list are
848 * about to be deleted and cannot be used as source addresses. 850 * about to be deleted and cannot be used as source addresses.
849 */ 851 */
850 list_for_each_entry(transport, &asoc->peer.transport_addr_list, 852 list_for_each_entry(transport, &asoc->peer.transport_addr_list,
851 transports) { 853 transports) {
852 dst_release(transport->dst); 854 dst_release(transport->dst);
853 sctp_transport_route(transport, NULL, 855 sctp_transport_route(transport, NULL,
854 sctp_sk(asoc->base.sk)); 856 sctp_sk(asoc->base.sk));
855 } 857 }
856 858
857 if (stored) 859 if (stored)
858 /* We don't need to transmit ASCONF */ 860 /* We don't need to transmit ASCONF */
859 continue; 861 continue;
860 retval = sctp_send_asconf(asoc, chunk); 862 retval = sctp_send_asconf(asoc, chunk);
861 } 863 }
862 out: 864 out:
863 return retval; 865 return retval;
864 } 866 }
865 867
866 /* set addr events to assocs in the endpoint. ep and addr_wq must be locked */ 868 /* set addr events to assocs in the endpoint. ep and addr_wq must be locked */
867 int sctp_asconf_mgmt(struct sctp_sock *sp, struct sctp_sockaddr_entry *addrw) 869 int sctp_asconf_mgmt(struct sctp_sock *sp, struct sctp_sockaddr_entry *addrw)
868 { 870 {
869 struct sock *sk = sctp_opt2sk(sp); 871 struct sock *sk = sctp_opt2sk(sp);
870 union sctp_addr *addr; 872 union sctp_addr *addr;
871 struct sctp_af *af; 873 struct sctp_af *af;
872 874
873 /* It is safe to write port space in caller. */ 875 /* It is safe to write port space in caller. */
874 addr = &addrw->a; 876 addr = &addrw->a;
875 addr->v4.sin_port = htons(sp->ep->base.bind_addr.port); 877 addr->v4.sin_port = htons(sp->ep->base.bind_addr.port);
876 af = sctp_get_af_specific(addr->sa.sa_family); 878 af = sctp_get_af_specific(addr->sa.sa_family);
877 if (!af) 879 if (!af)
878 return -EINVAL; 880 return -EINVAL;
879 if (sctp_verify_addr(sk, addr, af->sockaddr_len)) 881 if (sctp_verify_addr(sk, addr, af->sockaddr_len))
880 return -EINVAL; 882 return -EINVAL;
881 883
882 if (addrw->state == SCTP_ADDR_NEW) 884 if (addrw->state == SCTP_ADDR_NEW)
883 return sctp_send_asconf_add_ip(sk, (struct sockaddr *)addr, 1); 885 return sctp_send_asconf_add_ip(sk, (struct sockaddr *)addr, 1);
884 else 886 else
885 return sctp_send_asconf_del_ip(sk, (struct sockaddr *)addr, 1); 887 return sctp_send_asconf_del_ip(sk, (struct sockaddr *)addr, 1);
886 } 888 }
887 889
888 /* Helper for tunneling sctp_bindx() requests through sctp_setsockopt() 890 /* Helper for tunneling sctp_bindx() requests through sctp_setsockopt()
889 * 891 *
890 * API 8.1 892 * API 8.1
891 * int sctp_bindx(int sd, struct sockaddr *addrs, int addrcnt, 893 * int sctp_bindx(int sd, struct sockaddr *addrs, int addrcnt,
892 * int flags); 894 * int flags);
893 * 895 *
894 * If sd is an IPv4 socket, the addresses passed must be IPv4 addresses. 896 * If sd is an IPv4 socket, the addresses passed must be IPv4 addresses.
895 * If the sd is an IPv6 socket, the addresses passed can either be IPv4 897 * If the sd is an IPv6 socket, the addresses passed can either be IPv4
896 * or IPv6 addresses. 898 * or IPv6 addresses.
897 * 899 *
898 * A single address may be specified as INADDR_ANY or IN6ADDR_ANY, see 900 * A single address may be specified as INADDR_ANY or IN6ADDR_ANY, see
899 * Section 3.1.2 for this usage. 901 * Section 3.1.2 for this usage.
900 * 902 *
901 * addrs is a pointer to an array of one or more socket addresses. Each 903 * addrs is a pointer to an array of one or more socket addresses. Each
902 * address is contained in its appropriate structure (i.e. struct 904 * address is contained in its appropriate structure (i.e. struct
903 * sockaddr_in or struct sockaddr_in6) the family of the address type 905 * sockaddr_in or struct sockaddr_in6) the family of the address type
904 * must be used to distinguish the address length (note that this 906 * must be used to distinguish the address length (note that this
905 * representation is termed a "packed array" of addresses). The caller 907 * representation is termed a "packed array" of addresses). The caller
906 * specifies the number of addresses in the array with addrcnt. 908 * specifies the number of addresses in the array with addrcnt.
907 * 909 *
908 * On success, sctp_bindx() returns 0. On failure, sctp_bindx() returns 910 * On success, sctp_bindx() returns 0. On failure, sctp_bindx() returns
909 * -1, and sets errno to the appropriate error code. 911 * -1, and sets errno to the appropriate error code.
910 * 912 *
911 * For SCTP, the port given in each socket address must be the same, or 913 * For SCTP, the port given in each socket address must be the same, or
912 * sctp_bindx() will fail, setting errno to EINVAL. 914 * sctp_bindx() will fail, setting errno to EINVAL.
913 * 915 *
914 * The flags parameter is formed from the bitwise OR of zero or more of 916 * The flags parameter is formed from the bitwise OR of zero or more of
915 * the following currently defined flags: 917 * the following currently defined flags:
916 * 918 *
917 * SCTP_BINDX_ADD_ADDR 919 * SCTP_BINDX_ADD_ADDR
918 * 920 *
919 * SCTP_BINDX_REM_ADDR 921 * SCTP_BINDX_REM_ADDR
920 * 922 *
921 * SCTP_BINDX_ADD_ADDR directs SCTP to add the given addresses to the 923 * SCTP_BINDX_ADD_ADDR directs SCTP to add the given addresses to the
922 * association, and SCTP_BINDX_REM_ADDR directs SCTP to remove the given 924 * association, and SCTP_BINDX_REM_ADDR directs SCTP to remove the given
923 * addresses from the association. The two flags are mutually exclusive; 925 * addresses from the association. The two flags are mutually exclusive;
924 * if both are given, sctp_bindx() will fail with EINVAL. A caller may 926 * if both are given, sctp_bindx() will fail with EINVAL. A caller may
925 * not remove all addresses from an association; sctp_bindx() will 927 * not remove all addresses from an association; sctp_bindx() will
926 * reject such an attempt with EINVAL. 928 * reject such an attempt with EINVAL.
927 * 929 *
928 * An application can use sctp_bindx(SCTP_BINDX_ADD_ADDR) to associate 930 * An application can use sctp_bindx(SCTP_BINDX_ADD_ADDR) to associate
929 * additional addresses with an endpoint after calling bind(). Or use 931 * additional addresses with an endpoint after calling bind(). Or use
930 * sctp_bindx(SCTP_BINDX_REM_ADDR) to remove some addresses a listening 932 * sctp_bindx(SCTP_BINDX_REM_ADDR) to remove some addresses a listening
931 * socket is associated with so that no new association accepted will be 933 * socket is associated with so that no new association accepted will be
932 * associated with those addresses. If the endpoint supports dynamic 934 * associated with those addresses. If the endpoint supports dynamic
933 * address a SCTP_BINDX_REM_ADDR or SCTP_BINDX_ADD_ADDR may cause a 935 * address a SCTP_BINDX_REM_ADDR or SCTP_BINDX_ADD_ADDR may cause a
934 * endpoint to send the appropriate message to the peer to change the 936 * endpoint to send the appropriate message to the peer to change the
935 * peers address lists. 937 * peers address lists.
936 * 938 *
937 * Adding and removing addresses from a connected association is 939 * Adding and removing addresses from a connected association is
938 * optional functionality. Implementations that do not support this 940 * optional functionality. Implementations that do not support this
939 * functionality should return EOPNOTSUPP. 941 * functionality should return EOPNOTSUPP.
940 * 942 *
941 * Basically do nothing but copying the addresses from user to kernel 943 * Basically do nothing but copying the addresses from user to kernel
942 * land and invoking either sctp_bindx_add() or sctp_bindx_rem() on the sk. 944 * land and invoking either sctp_bindx_add() or sctp_bindx_rem() on the sk.
943 * This is used for tunneling the sctp_bindx() request through sctp_setsockopt() 945 * This is used for tunneling the sctp_bindx() request through sctp_setsockopt()
944 * from userspace. 946 * from userspace.
945 * 947 *
946 * We don't use copy_from_user() for optimization: we first do the 948 * We don't use copy_from_user() for optimization: we first do the
947 * sanity checks (buffer size -fast- and access check-healthy 949 * sanity checks (buffer size -fast- and access check-healthy
948 * pointer); if all of those succeed, then we can alloc the memory 950 * pointer); if all of those succeed, then we can alloc the memory
949 * (expensive operation) needed to copy the data to kernel. Then we do 951 * (expensive operation) needed to copy the data to kernel. Then we do
950 * the copying without checking the user space area 952 * the copying without checking the user space area
951 * (__copy_from_user()). 953 * (__copy_from_user()).
952 * 954 *
953 * On exit there is no need to do sockfd_put(), sys_setsockopt() does 955 * On exit there is no need to do sockfd_put(), sys_setsockopt() does
954 * it. 956 * it.
955 * 957 *
956 * sk The sk of the socket 958 * sk The sk of the socket
957 * addrs The pointer to the addresses in user land 959 * addrs The pointer to the addresses in user land
958 * addrssize Size of the addrs buffer 960 * addrssize Size of the addrs buffer
959 * op Operation to perform (add or remove, see the flags of 961 * op Operation to perform (add or remove, see the flags of
960 * sctp_bindx) 962 * sctp_bindx)
961 * 963 *
962 * Returns 0 if ok, <0 errno code on error. 964 * Returns 0 if ok, <0 errno code on error.
963 */ 965 */
964 SCTP_STATIC int sctp_setsockopt_bindx(struct sock* sk, 966 SCTP_STATIC int sctp_setsockopt_bindx(struct sock* sk,
965 struct sockaddr __user *addrs, 967 struct sockaddr __user *addrs,
966 int addrs_size, int op) 968 int addrs_size, int op)
967 { 969 {
968 struct sockaddr *kaddrs; 970 struct sockaddr *kaddrs;
969 int err; 971 int err;
970 int addrcnt = 0; 972 int addrcnt = 0;
971 int walk_size = 0; 973 int walk_size = 0;
972 struct sockaddr *sa_addr; 974 struct sockaddr *sa_addr;
973 void *addr_buf; 975 void *addr_buf;
974 struct sctp_af *af; 976 struct sctp_af *af;
975 977
976 SCTP_DEBUG_PRINTK("sctp_setsockopt_bindx: sk %p addrs %p" 978 SCTP_DEBUG_PRINTK("sctp_setsockopt_bindx: sk %p addrs %p"
977 " addrs_size %d opt %d\n", sk, addrs, addrs_size, op); 979 " addrs_size %d opt %d\n", sk, addrs, addrs_size, op);
978 980
979 if (unlikely(addrs_size <= 0)) 981 if (unlikely(addrs_size <= 0))
980 return -EINVAL; 982 return -EINVAL;
981 983
982 /* Check the user passed a healthy pointer. */ 984 /* Check the user passed a healthy pointer. */
983 if (unlikely(!access_ok(VERIFY_READ, addrs, addrs_size))) 985 if (unlikely(!access_ok(VERIFY_READ, addrs, addrs_size)))
984 return -EFAULT; 986 return -EFAULT;
985 987
986 /* Alloc space for the address array in kernel memory. */ 988 /* Alloc space for the address array in kernel memory. */
987 kaddrs = kmalloc(addrs_size, GFP_KERNEL); 989 kaddrs = kmalloc(addrs_size, GFP_KERNEL);
988 if (unlikely(!kaddrs)) 990 if (unlikely(!kaddrs))
989 return -ENOMEM; 991 return -ENOMEM;
990 992
991 if (__copy_from_user(kaddrs, addrs, addrs_size)) { 993 if (__copy_from_user(kaddrs, addrs, addrs_size)) {
992 kfree(kaddrs); 994 kfree(kaddrs);
993 return -EFAULT; 995 return -EFAULT;
994 } 996 }
995 997
996 /* Walk through the addrs buffer and count the number of addresses. */ 998 /* Walk through the addrs buffer and count the number of addresses. */
997 addr_buf = kaddrs; 999 addr_buf = kaddrs;
998 while (walk_size < addrs_size) { 1000 while (walk_size < addrs_size) {
999 if (walk_size + sizeof(sa_family_t) > addrs_size) { 1001 if (walk_size + sizeof(sa_family_t) > addrs_size) {
1000 kfree(kaddrs); 1002 kfree(kaddrs);
1001 return -EINVAL; 1003 return -EINVAL;
1002 } 1004 }
1003 1005
1004 sa_addr = addr_buf; 1006 sa_addr = addr_buf;
1005 af = sctp_get_af_specific(sa_addr->sa_family); 1007 af = sctp_get_af_specific(sa_addr->sa_family);
1006 1008
1007 /* If the address family is not supported or if this address 1009 /* If the address family is not supported or if this address
1008 * causes the address buffer to overflow return EINVAL. 1010 * causes the address buffer to overflow return EINVAL.
1009 */ 1011 */
1010 if (!af || (walk_size + af->sockaddr_len) > addrs_size) { 1012 if (!af || (walk_size + af->sockaddr_len) > addrs_size) {
1011 kfree(kaddrs); 1013 kfree(kaddrs);
1012 return -EINVAL; 1014 return -EINVAL;
1013 } 1015 }
1014 addrcnt++; 1016 addrcnt++;
1015 addr_buf += af->sockaddr_len; 1017 addr_buf += af->sockaddr_len;
1016 walk_size += af->sockaddr_len; 1018 walk_size += af->sockaddr_len;
1017 } 1019 }
1018 1020
1019 /* Do the work. */ 1021 /* Do the work. */
1020 switch (op) { 1022 switch (op) {
1021 case SCTP_BINDX_ADD_ADDR: 1023 case SCTP_BINDX_ADD_ADDR:
1022 err = sctp_bindx_add(sk, kaddrs, addrcnt); 1024 err = sctp_bindx_add(sk, kaddrs, addrcnt);
1023 if (err) 1025 if (err)
1024 goto out; 1026 goto out;
1025 err = sctp_send_asconf_add_ip(sk, kaddrs, addrcnt); 1027 err = sctp_send_asconf_add_ip(sk, kaddrs, addrcnt);
1026 break; 1028 break;
1027 1029
1028 case SCTP_BINDX_REM_ADDR: 1030 case SCTP_BINDX_REM_ADDR:
1029 err = sctp_bindx_rem(sk, kaddrs, addrcnt); 1031 err = sctp_bindx_rem(sk, kaddrs, addrcnt);
1030 if (err) 1032 if (err)
1031 goto out; 1033 goto out;
1032 err = sctp_send_asconf_del_ip(sk, kaddrs, addrcnt); 1034 err = sctp_send_asconf_del_ip(sk, kaddrs, addrcnt);
1033 break; 1035 break;
1034 1036
1035 default: 1037 default:
1036 err = -EINVAL; 1038 err = -EINVAL;
1037 break; 1039 break;
1038 } 1040 }
1039 1041
1040 out: 1042 out:
1041 kfree(kaddrs); 1043 kfree(kaddrs);
1042 1044
1043 return err; 1045 return err;
1044 } 1046 }
1045 1047
1046 /* __sctp_connect(struct sock* sk, struct sockaddr *kaddrs, int addrs_size) 1048 /* __sctp_connect(struct sock* sk, struct sockaddr *kaddrs, int addrs_size)
1047 * 1049 *
1048 * Common routine for handling connect() and sctp_connectx(). 1050 * Common routine for handling connect() and sctp_connectx().
1049 * Connect will come in with just a single address. 1051 * Connect will come in with just a single address.
1050 */ 1052 */
1051 static int __sctp_connect(struct sock* sk, 1053 static int __sctp_connect(struct sock* sk,
1052 struct sockaddr *kaddrs, 1054 struct sockaddr *kaddrs,
1053 int addrs_size, 1055 int addrs_size,
1054 sctp_assoc_t *assoc_id) 1056 sctp_assoc_t *assoc_id)
1055 { 1057 {
1056 struct net *net = sock_net(sk); 1058 struct net *net = sock_net(sk);
1057 struct sctp_sock *sp; 1059 struct sctp_sock *sp;
1058 struct sctp_endpoint *ep; 1060 struct sctp_endpoint *ep;
1059 struct sctp_association *asoc = NULL; 1061 struct sctp_association *asoc = NULL;
1060 struct sctp_association *asoc2; 1062 struct sctp_association *asoc2;
1061 struct sctp_transport *transport; 1063 struct sctp_transport *transport;
1062 union sctp_addr to; 1064 union sctp_addr to;
1063 struct sctp_af *af; 1065 struct sctp_af *af;
1064 sctp_scope_t scope; 1066 sctp_scope_t scope;
1065 long timeo; 1067 long timeo;
1066 int err = 0; 1068 int err = 0;
1067 int addrcnt = 0; 1069 int addrcnt = 0;
1068 int walk_size = 0; 1070 int walk_size = 0;
1069 union sctp_addr *sa_addr = NULL; 1071 union sctp_addr *sa_addr = NULL;
1070 void *addr_buf; 1072 void *addr_buf;
1071 unsigned short port; 1073 unsigned short port;
1072 unsigned int f_flags = 0; 1074 unsigned int f_flags = 0;
1073 1075
1074 sp = sctp_sk(sk); 1076 sp = sctp_sk(sk);
1075 ep = sp->ep; 1077 ep = sp->ep;
1076 1078
1077 /* connect() cannot be done on a socket that is already in ESTABLISHED 1079 /* connect() cannot be done on a socket that is already in ESTABLISHED
1078 * state - UDP-style peeled off socket or a TCP-style socket that 1080 * state - UDP-style peeled off socket or a TCP-style socket that
1079 * is already connected. 1081 * is already connected.
1080 * It cannot be done even on a TCP-style listening socket. 1082 * It cannot be done even on a TCP-style listening socket.
1081 */ 1083 */
1082 if (sctp_sstate(sk, ESTABLISHED) || 1084 if (sctp_sstate(sk, ESTABLISHED) ||
1083 (sctp_style(sk, TCP) && sctp_sstate(sk, LISTENING))) { 1085 (sctp_style(sk, TCP) && sctp_sstate(sk, LISTENING))) {
1084 err = -EISCONN; 1086 err = -EISCONN;
1085 goto out_free; 1087 goto out_free;
1086 } 1088 }
1087 1089
1088 /* Walk through the addrs buffer and count the number of addresses. */ 1090 /* Walk through the addrs buffer and count the number of addresses. */
1089 addr_buf = kaddrs; 1091 addr_buf = kaddrs;
1090 while (walk_size < addrs_size) { 1092 while (walk_size < addrs_size) {
1091 if (walk_size + sizeof(sa_family_t) > addrs_size) { 1093 if (walk_size + sizeof(sa_family_t) > addrs_size) {
1092 err = -EINVAL; 1094 err = -EINVAL;
1093 goto out_free; 1095 goto out_free;
1094 } 1096 }
1095 1097
1096 sa_addr = addr_buf; 1098 sa_addr = addr_buf;
1097 af = sctp_get_af_specific(sa_addr->sa.sa_family); 1099 af = sctp_get_af_specific(sa_addr->sa.sa_family);
1098 1100
1099 /* If the address family is not supported or if this address 1101 /* If the address family is not supported or if this address
1100 * causes the address buffer to overflow return EINVAL. 1102 * causes the address buffer to overflow return EINVAL.
1101 */ 1103 */
1102 if (!af || (walk_size + af->sockaddr_len) > addrs_size) { 1104 if (!af || (walk_size + af->sockaddr_len) > addrs_size) {
1103 err = -EINVAL; 1105 err = -EINVAL;
1104 goto out_free; 1106 goto out_free;
1105 } 1107 }
1106 1108
1107 port = ntohs(sa_addr->v4.sin_port); 1109 port = ntohs(sa_addr->v4.sin_port);
1108 1110
1109 /* Save current address so we can work with it */ 1111 /* Save current address so we can work with it */
1110 memcpy(&to, sa_addr, af->sockaddr_len); 1112 memcpy(&to, sa_addr, af->sockaddr_len);
1111 1113
1112 err = sctp_verify_addr(sk, &to, af->sockaddr_len); 1114 err = sctp_verify_addr(sk, &to, af->sockaddr_len);
1113 if (err) 1115 if (err)
1114 goto out_free; 1116 goto out_free;
1115 1117
1116 /* Make sure the destination port is correctly set 1118 /* Make sure the destination port is correctly set
1117 * in all addresses. 1119 * in all addresses.
1118 */ 1120 */
1119 if (asoc && asoc->peer.port && asoc->peer.port != port) 1121 if (asoc && asoc->peer.port && asoc->peer.port != port)
1120 goto out_free; 1122 goto out_free;
1121 1123
1122 1124
1123 /* Check if there already is a matching association on the 1125 /* Check if there already is a matching association on the
1124 * endpoint (other than the one created here). 1126 * endpoint (other than the one created here).
1125 */ 1127 */
1126 asoc2 = sctp_endpoint_lookup_assoc(ep, &to, &transport); 1128 asoc2 = sctp_endpoint_lookup_assoc(ep, &to, &transport);
1127 if (asoc2 && asoc2 != asoc) { 1129 if (asoc2 && asoc2 != asoc) {
1128 if (asoc2->state >= SCTP_STATE_ESTABLISHED) 1130 if (asoc2->state >= SCTP_STATE_ESTABLISHED)
1129 err = -EISCONN; 1131 err = -EISCONN;
1130 else 1132 else
1131 err = -EALREADY; 1133 err = -EALREADY;
1132 goto out_free; 1134 goto out_free;
1133 } 1135 }
1134 1136
1135 /* If we could not find a matching association on the endpoint, 1137 /* If we could not find a matching association on the endpoint,
1136 * make sure that there is no peeled-off association matching 1138 * make sure that there is no peeled-off association matching
1137 * the peer address even on another socket. 1139 * the peer address even on another socket.
1138 */ 1140 */
1139 if (sctp_endpoint_is_peeled_off(ep, &to)) { 1141 if (sctp_endpoint_is_peeled_off(ep, &to)) {
1140 err = -EADDRNOTAVAIL; 1142 err = -EADDRNOTAVAIL;
1141 goto out_free; 1143 goto out_free;
1142 } 1144 }
1143 1145
1144 if (!asoc) { 1146 if (!asoc) {
1145 /* If a bind() or sctp_bindx() is not called prior to 1147 /* If a bind() or sctp_bindx() is not called prior to
1146 * an sctp_connectx() call, the system picks an 1148 * an sctp_connectx() call, the system picks an
1147 * ephemeral port and will choose an address set 1149 * ephemeral port and will choose an address set
1148 * equivalent to binding with a wildcard address. 1150 * equivalent to binding with a wildcard address.
1149 */ 1151 */
1150 if (!ep->base.bind_addr.port) { 1152 if (!ep->base.bind_addr.port) {
1151 if (sctp_autobind(sk)) { 1153 if (sctp_autobind(sk)) {
1152 err = -EAGAIN; 1154 err = -EAGAIN;
1153 goto out_free; 1155 goto out_free;
1154 } 1156 }
1155 } else { 1157 } else {
1156 /* 1158 /*
1157 * If an unprivileged user inherits a 1-many 1159 * If an unprivileged user inherits a 1-many
1158 * style socket with open associations on a 1160 * style socket with open associations on a
1159 * privileged port, it MAY be permitted to 1161 * privileged port, it MAY be permitted to
1160 * accept new associations, but it SHOULD NOT 1162 * accept new associations, but it SHOULD NOT
1161 * be permitted to open new associations. 1163 * be permitted to open new associations.
1162 */ 1164 */
1163 if (ep->base.bind_addr.port < PROT_SOCK && 1165 if (ep->base.bind_addr.port < PROT_SOCK &&
1164 !capable(CAP_NET_BIND_SERVICE)) { 1166 !ns_capable(net->user_ns, CAP_NET_BIND_SERVICE)) {
1165 err = -EACCES; 1167 err = -EACCES;
1166 goto out_free; 1168 goto out_free;
1167 } 1169 }
1168 } 1170 }
1169 1171
1170 scope = sctp_scope(&to); 1172 scope = sctp_scope(&to);
1171 asoc = sctp_association_new(ep, sk, scope, GFP_KERNEL); 1173 asoc = sctp_association_new(ep, sk, scope, GFP_KERNEL);
1172 if (!asoc) { 1174 if (!asoc) {
1173 err = -ENOMEM; 1175 err = -ENOMEM;
1174 goto out_free; 1176 goto out_free;
1175 } 1177 }
1176 1178
1177 err = sctp_assoc_set_bind_addr_from_ep(asoc, scope, 1179 err = sctp_assoc_set_bind_addr_from_ep(asoc, scope,
1178 GFP_KERNEL); 1180 GFP_KERNEL);
1179 if (err < 0) { 1181 if (err < 0) {
1180 goto out_free; 1182 goto out_free;
1181 } 1183 }
1182 1184
1183 } 1185 }
1184 1186
1185 /* Prime the peer's transport structures. */ 1187 /* Prime the peer's transport structures. */
1186 transport = sctp_assoc_add_peer(asoc, &to, GFP_KERNEL, 1188 transport = sctp_assoc_add_peer(asoc, &to, GFP_KERNEL,
1187 SCTP_UNKNOWN); 1189 SCTP_UNKNOWN);
1188 if (!transport) { 1190 if (!transport) {
1189 err = -ENOMEM; 1191 err = -ENOMEM;
1190 goto out_free; 1192 goto out_free;
1191 } 1193 }
1192 1194
1193 addrcnt++; 1195 addrcnt++;
1194 addr_buf += af->sockaddr_len; 1196 addr_buf += af->sockaddr_len;
1195 walk_size += af->sockaddr_len; 1197 walk_size += af->sockaddr_len;
1196 } 1198 }
1197 1199
1198 /* In case the user of sctp_connectx() wants an association 1200 /* In case the user of sctp_connectx() wants an association
1199 * id back, assign one now. 1201 * id back, assign one now.
1200 */ 1202 */
1201 if (assoc_id) { 1203 if (assoc_id) {
1202 err = sctp_assoc_set_id(asoc, GFP_KERNEL); 1204 err = sctp_assoc_set_id(asoc, GFP_KERNEL);
1203 if (err < 0) 1205 if (err < 0)
1204 goto out_free; 1206 goto out_free;
1205 } 1207 }
1206 1208
1207 err = sctp_primitive_ASSOCIATE(net, asoc, NULL); 1209 err = sctp_primitive_ASSOCIATE(net, asoc, NULL);
1208 if (err < 0) { 1210 if (err < 0) {
1209 goto out_free; 1211 goto out_free;
1210 } 1212 }
1211 1213
1212 /* Initialize sk's dport and daddr for getpeername() */ 1214 /* Initialize sk's dport and daddr for getpeername() */
1213 inet_sk(sk)->inet_dport = htons(asoc->peer.port); 1215 inet_sk(sk)->inet_dport = htons(asoc->peer.port);
1214 af = sctp_get_af_specific(sa_addr->sa.sa_family); 1216 af = sctp_get_af_specific(sa_addr->sa.sa_family);
1215 af->to_sk_daddr(sa_addr, sk); 1217 af->to_sk_daddr(sa_addr, sk);
1216 sk->sk_err = 0; 1218 sk->sk_err = 0;
1217 1219
1218 /* in-kernel sockets don't generally have a file allocated to them 1220 /* in-kernel sockets don't generally have a file allocated to them
1219 * if all they do is call sock_create_kern(). 1221 * if all they do is call sock_create_kern().
1220 */ 1222 */
1221 if (sk->sk_socket->file) 1223 if (sk->sk_socket->file)
1222 f_flags = sk->sk_socket->file->f_flags; 1224 f_flags = sk->sk_socket->file->f_flags;
1223 1225
1224 timeo = sock_sndtimeo(sk, f_flags & O_NONBLOCK); 1226 timeo = sock_sndtimeo(sk, f_flags & O_NONBLOCK);
1225 1227
1226 err = sctp_wait_for_connect(asoc, &timeo); 1228 err = sctp_wait_for_connect(asoc, &timeo);
1227 if ((err == 0 || err == -EINPROGRESS) && assoc_id) 1229 if ((err == 0 || err == -EINPROGRESS) && assoc_id)
1228 *assoc_id = asoc->assoc_id; 1230 *assoc_id = asoc->assoc_id;
1229 1231
1230 /* Don't free association on exit. */ 1232 /* Don't free association on exit. */
1231 asoc = NULL; 1233 asoc = NULL;
1232 1234
1233 out_free: 1235 out_free:
1234 1236
1235 SCTP_DEBUG_PRINTK("About to exit __sctp_connect() free asoc: %p" 1237 SCTP_DEBUG_PRINTK("About to exit __sctp_connect() free asoc: %p"
1236 " kaddrs: %p err: %d\n", 1238 " kaddrs: %p err: %d\n",
1237 asoc, kaddrs, err); 1239 asoc, kaddrs, err);
1238 if (asoc) { 1240 if (asoc) {
1239 /* sctp_primitive_ASSOCIATE may have added this association 1241 /* sctp_primitive_ASSOCIATE may have added this association
1240 * To the hash table, try to unhash it, just in case, its a noop 1242 * To the hash table, try to unhash it, just in case, its a noop
1241 * if it wasn't hashed so we're safe 1243 * if it wasn't hashed so we're safe
1242 */ 1244 */
1243 sctp_unhash_established(asoc); 1245 sctp_unhash_established(asoc);
1244 sctp_association_free(asoc); 1246 sctp_association_free(asoc);
1245 } 1247 }
1246 return err; 1248 return err;
1247 } 1249 }
1248 1250
1249 /* Helper for tunneling sctp_connectx() requests through sctp_setsockopt() 1251 /* Helper for tunneling sctp_connectx() requests through sctp_setsockopt()
1250 * 1252 *
1251 * API 8.9 1253 * API 8.9
1252 * int sctp_connectx(int sd, struct sockaddr *addrs, int addrcnt, 1254 * int sctp_connectx(int sd, struct sockaddr *addrs, int addrcnt,
1253 * sctp_assoc_t *asoc); 1255 * sctp_assoc_t *asoc);
1254 * 1256 *
1255 * If sd is an IPv4 socket, the addresses passed must be IPv4 addresses. 1257 * If sd is an IPv4 socket, the addresses passed must be IPv4 addresses.
1256 * If the sd is an IPv6 socket, the addresses passed can either be IPv4 1258 * If the sd is an IPv6 socket, the addresses passed can either be IPv4
1257 * or IPv6 addresses. 1259 * or IPv6 addresses.
1258 * 1260 *
1259 * A single address may be specified as INADDR_ANY or IN6ADDR_ANY, see 1261 * A single address may be specified as INADDR_ANY or IN6ADDR_ANY, see
1260 * Section 3.1.2 for this usage. 1262 * Section 3.1.2 for this usage.
1261 * 1263 *
1262 * addrs is a pointer to an array of one or more socket addresses. Each 1264 * addrs is a pointer to an array of one or more socket addresses. Each
1263 * address is contained in its appropriate structure (i.e. struct 1265 * address is contained in its appropriate structure (i.e. struct
1264 * sockaddr_in or struct sockaddr_in6) the family of the address type 1266 * sockaddr_in or struct sockaddr_in6) the family of the address type
1265 * must be used to distengish the address length (note that this 1267 * must be used to distengish the address length (note that this
1266 * representation is termed a "packed array" of addresses). The caller 1268 * representation is termed a "packed array" of addresses). The caller
1267 * specifies the number of addresses in the array with addrcnt. 1269 * specifies the number of addresses in the array with addrcnt.
1268 * 1270 *
1269 * On success, sctp_connectx() returns 0. It also sets the assoc_id to 1271 * On success, sctp_connectx() returns 0. It also sets the assoc_id to
1270 * the association id of the new association. On failure, sctp_connectx() 1272 * the association id of the new association. On failure, sctp_connectx()
1271 * returns -1, and sets errno to the appropriate error code. The assoc_id 1273 * returns -1, and sets errno to the appropriate error code. The assoc_id
1272 * is not touched by the kernel. 1274 * is not touched by the kernel.
1273 * 1275 *
1274 * For SCTP, the port given in each socket address must be the same, or 1276 * For SCTP, the port given in each socket address must be the same, or
1275 * sctp_connectx() will fail, setting errno to EINVAL. 1277 * sctp_connectx() will fail, setting errno to EINVAL.
1276 * 1278 *
1277 * An application can use sctp_connectx to initiate an association with 1279 * An application can use sctp_connectx to initiate an association with
1278 * an endpoint that is multi-homed. Much like sctp_bindx() this call 1280 * an endpoint that is multi-homed. Much like sctp_bindx() this call
1279 * allows a caller to specify multiple addresses at which a peer can be 1281 * allows a caller to specify multiple addresses at which a peer can be
1280 * reached. The way the SCTP stack uses the list of addresses to set up 1282 * reached. The way the SCTP stack uses the list of addresses to set up
1281 * the association is implementation dependent. This function only 1283 * the association is implementation dependent. This function only
1282 * specifies that the stack will try to make use of all the addresses in 1284 * specifies that the stack will try to make use of all the addresses in
1283 * the list when needed. 1285 * the list when needed.
1284 * 1286 *
1285 * Note that the list of addresses passed in is only used for setting up 1287 * Note that the list of addresses passed in is only used for setting up
1286 * the association. It does not necessarily equal the set of addresses 1288 * the association. It does not necessarily equal the set of addresses
1287 * the peer uses for the resulting association. If the caller wants to 1289 * the peer uses for the resulting association. If the caller wants to
1288 * find out the set of peer addresses, it must use sctp_getpaddrs() to 1290 * find out the set of peer addresses, it must use sctp_getpaddrs() to
1289 * retrieve them after the association has been set up. 1291 * retrieve them after the association has been set up.
1290 * 1292 *
1291 * Basically do nothing but copying the addresses from user to kernel 1293 * Basically do nothing but copying the addresses from user to kernel
1292 * land and invoking either sctp_connectx(). This is used for tunneling 1294 * land and invoking either sctp_connectx(). This is used for tunneling
1293 * the sctp_connectx() request through sctp_setsockopt() from userspace. 1295 * the sctp_connectx() request through sctp_setsockopt() from userspace.
1294 * 1296 *
1295 * We don't use copy_from_user() for optimization: we first do the 1297 * We don't use copy_from_user() for optimization: we first do the
1296 * sanity checks (buffer size -fast- and access check-healthy 1298 * sanity checks (buffer size -fast- and access check-healthy
1297 * pointer); if all of those succeed, then we can alloc the memory 1299 * pointer); if all of those succeed, then we can alloc the memory
1298 * (expensive operation) needed to copy the data to kernel. Then we do 1300 * (expensive operation) needed to copy the data to kernel. Then we do
1299 * the copying without checking the user space area 1301 * the copying without checking the user space area
1300 * (__copy_from_user()). 1302 * (__copy_from_user()).
1301 * 1303 *
1302 * On exit there is no need to do sockfd_put(), sys_setsockopt() does 1304 * On exit there is no need to do sockfd_put(), sys_setsockopt() does
1303 * it. 1305 * it.
1304 * 1306 *
1305 * sk The sk of the socket 1307 * sk The sk of the socket
1306 * addrs The pointer to the addresses in user land 1308 * addrs The pointer to the addresses in user land
1307 * addrssize Size of the addrs buffer 1309 * addrssize Size of the addrs buffer
1308 * 1310 *
1309 * Returns >=0 if ok, <0 errno code on error. 1311 * Returns >=0 if ok, <0 errno code on error.
1310 */ 1312 */
1311 SCTP_STATIC int __sctp_setsockopt_connectx(struct sock* sk, 1313 SCTP_STATIC int __sctp_setsockopt_connectx(struct sock* sk,
1312 struct sockaddr __user *addrs, 1314 struct sockaddr __user *addrs,
1313 int addrs_size, 1315 int addrs_size,
1314 sctp_assoc_t *assoc_id) 1316 sctp_assoc_t *assoc_id)
1315 { 1317 {
1316 int err = 0; 1318 int err = 0;
1317 struct sockaddr *kaddrs; 1319 struct sockaddr *kaddrs;
1318 1320
1319 SCTP_DEBUG_PRINTK("%s - sk %p addrs %p addrs_size %d\n", 1321 SCTP_DEBUG_PRINTK("%s - sk %p addrs %p addrs_size %d\n",
1320 __func__, sk, addrs, addrs_size); 1322 __func__, sk, addrs, addrs_size);
1321 1323
1322 if (unlikely(addrs_size <= 0)) 1324 if (unlikely(addrs_size <= 0))
1323 return -EINVAL; 1325 return -EINVAL;
1324 1326
1325 /* Check the user passed a healthy pointer. */ 1327 /* Check the user passed a healthy pointer. */
1326 if (unlikely(!access_ok(VERIFY_READ, addrs, addrs_size))) 1328 if (unlikely(!access_ok(VERIFY_READ, addrs, addrs_size)))
1327 return -EFAULT; 1329 return -EFAULT;
1328 1330
1329 /* Alloc space for the address array in kernel memory. */ 1331 /* Alloc space for the address array in kernel memory. */
1330 kaddrs = kmalloc(addrs_size, GFP_KERNEL); 1332 kaddrs = kmalloc(addrs_size, GFP_KERNEL);
1331 if (unlikely(!kaddrs)) 1333 if (unlikely(!kaddrs))
1332 return -ENOMEM; 1334 return -ENOMEM;
1333 1335
1334 if (__copy_from_user(kaddrs, addrs, addrs_size)) { 1336 if (__copy_from_user(kaddrs, addrs, addrs_size)) {
1335 err = -EFAULT; 1337 err = -EFAULT;
1336 } else { 1338 } else {
1337 err = __sctp_connect(sk, kaddrs, addrs_size, assoc_id); 1339 err = __sctp_connect(sk, kaddrs, addrs_size, assoc_id);
1338 } 1340 }
1339 1341
1340 kfree(kaddrs); 1342 kfree(kaddrs);
1341 1343
1342 return err; 1344 return err;
1343 } 1345 }
1344 1346
1345 /* 1347 /*
1346 * This is an older interface. It's kept for backward compatibility 1348 * This is an older interface. It's kept for backward compatibility
1347 * to the option that doesn't provide association id. 1349 * to the option that doesn't provide association id.
1348 */ 1350 */
1349 SCTP_STATIC int sctp_setsockopt_connectx_old(struct sock* sk, 1351 SCTP_STATIC int sctp_setsockopt_connectx_old(struct sock* sk,
1350 struct sockaddr __user *addrs, 1352 struct sockaddr __user *addrs,
1351 int addrs_size) 1353 int addrs_size)
1352 { 1354 {
1353 return __sctp_setsockopt_connectx(sk, addrs, addrs_size, NULL); 1355 return __sctp_setsockopt_connectx(sk, addrs, addrs_size, NULL);
1354 } 1356 }
1355 1357
1356 /* 1358 /*
1357 * New interface for the API. The since the API is done with a socket 1359 * New interface for the API. The since the API is done with a socket
1358 * option, to make it simple we feed back the association id is as a return 1360 * option, to make it simple we feed back the association id is as a return
1359 * indication to the call. Error is always negative and association id is 1361 * indication to the call. Error is always negative and association id is
1360 * always positive. 1362 * always positive.
1361 */ 1363 */
1362 SCTP_STATIC int sctp_setsockopt_connectx(struct sock* sk, 1364 SCTP_STATIC int sctp_setsockopt_connectx(struct sock* sk,
1363 struct sockaddr __user *addrs, 1365 struct sockaddr __user *addrs,
1364 int addrs_size) 1366 int addrs_size)
1365 { 1367 {
1366 sctp_assoc_t assoc_id = 0; 1368 sctp_assoc_t assoc_id = 0;
1367 int err = 0; 1369 int err = 0;
1368 1370
1369 err = __sctp_setsockopt_connectx(sk, addrs, addrs_size, &assoc_id); 1371 err = __sctp_setsockopt_connectx(sk, addrs, addrs_size, &assoc_id);
1370 1372
1371 if (err) 1373 if (err)
1372 return err; 1374 return err;
1373 else 1375 else
1374 return assoc_id; 1376 return assoc_id;
1375 } 1377 }
1376 1378
1377 /* 1379 /*
1378 * New (hopefully final) interface for the API. 1380 * New (hopefully final) interface for the API.
1379 * We use the sctp_getaddrs_old structure so that use-space library 1381 * We use the sctp_getaddrs_old structure so that use-space library
1380 * can avoid any unnecessary allocations. The only defferent part 1382 * can avoid any unnecessary allocations. The only defferent part
1381 * is that we store the actual length of the address buffer into the 1383 * is that we store the actual length of the address buffer into the
1382 * addrs_num structure member. That way we can re-use the existing 1384 * addrs_num structure member. That way we can re-use the existing
1383 * code. 1385 * code.
1384 */ 1386 */
1385 SCTP_STATIC int sctp_getsockopt_connectx3(struct sock* sk, int len, 1387 SCTP_STATIC int sctp_getsockopt_connectx3(struct sock* sk, int len,
1386 char __user *optval, 1388 char __user *optval,
1387 int __user *optlen) 1389 int __user *optlen)
1388 { 1390 {
1389 struct sctp_getaddrs_old param; 1391 struct sctp_getaddrs_old param;
1390 sctp_assoc_t assoc_id = 0; 1392 sctp_assoc_t assoc_id = 0;
1391 int err = 0; 1393 int err = 0;
1392 1394
1393 if (len < sizeof(param)) 1395 if (len < sizeof(param))
1394 return -EINVAL; 1396 return -EINVAL;
1395 1397
1396 if (copy_from_user(&param, optval, sizeof(param))) 1398 if (copy_from_user(&param, optval, sizeof(param)))
1397 return -EFAULT; 1399 return -EFAULT;
1398 1400
1399 err = __sctp_setsockopt_connectx(sk, 1401 err = __sctp_setsockopt_connectx(sk,
1400 (struct sockaddr __user *)param.addrs, 1402 (struct sockaddr __user *)param.addrs,
1401 param.addr_num, &assoc_id); 1403 param.addr_num, &assoc_id);
1402 1404
1403 if (err == 0 || err == -EINPROGRESS) { 1405 if (err == 0 || err == -EINPROGRESS) {
1404 if (copy_to_user(optval, &assoc_id, sizeof(assoc_id))) 1406 if (copy_to_user(optval, &assoc_id, sizeof(assoc_id)))
1405 return -EFAULT; 1407 return -EFAULT;
1406 if (put_user(sizeof(assoc_id), optlen)) 1408 if (put_user(sizeof(assoc_id), optlen))
1407 return -EFAULT; 1409 return -EFAULT;
1408 } 1410 }
1409 1411
1410 return err; 1412 return err;
1411 } 1413 }
1412 1414
1413 /* API 3.1.4 close() - UDP Style Syntax 1415 /* API 3.1.4 close() - UDP Style Syntax
1414 * Applications use close() to perform graceful shutdown (as described in 1416 * Applications use close() to perform graceful shutdown (as described in
1415 * Section 10.1 of [SCTP]) on ALL the associations currently represented 1417 * Section 10.1 of [SCTP]) on ALL the associations currently represented
1416 * by a UDP-style socket. 1418 * by a UDP-style socket.
1417 * 1419 *
1418 * The syntax is 1420 * The syntax is
1419 * 1421 *
1420 * ret = close(int sd); 1422 * ret = close(int sd);
1421 * 1423 *
1422 * sd - the socket descriptor of the associations to be closed. 1424 * sd - the socket descriptor of the associations to be closed.
1423 * 1425 *
1424 * To gracefully shutdown a specific association represented by the 1426 * To gracefully shutdown a specific association represented by the
1425 * UDP-style socket, an application should use the sendmsg() call, 1427 * UDP-style socket, an application should use the sendmsg() call,
1426 * passing no user data, but including the appropriate flag in the 1428 * passing no user data, but including the appropriate flag in the
1427 * ancillary data (see Section xxxx). 1429 * ancillary data (see Section xxxx).
1428 * 1430 *
1429 * If sd in the close() call is a branched-off socket representing only 1431 * If sd in the close() call is a branched-off socket representing only
1430 * one association, the shutdown is performed on that association only. 1432 * one association, the shutdown is performed on that association only.
1431 * 1433 *
1432 * 4.1.6 close() - TCP Style Syntax 1434 * 4.1.6 close() - TCP Style Syntax
1433 * 1435 *
1434 * Applications use close() to gracefully close down an association. 1436 * Applications use close() to gracefully close down an association.
1435 * 1437 *
1436 * The syntax is: 1438 * The syntax is:
1437 * 1439 *
1438 * int close(int sd); 1440 * int close(int sd);
1439 * 1441 *
1440 * sd - the socket descriptor of the association to be closed. 1442 * sd - the socket descriptor of the association to be closed.
1441 * 1443 *
1442 * After an application calls close() on a socket descriptor, no further 1444 * After an application calls close() on a socket descriptor, no further
1443 * socket operations will succeed on that descriptor. 1445 * socket operations will succeed on that descriptor.
1444 * 1446 *
1445 * API 7.1.4 SO_LINGER 1447 * API 7.1.4 SO_LINGER
1446 * 1448 *
1447 * An application using the TCP-style socket can use this option to 1449 * An application using the TCP-style socket can use this option to
1448 * perform the SCTP ABORT primitive. The linger option structure is: 1450 * perform the SCTP ABORT primitive. The linger option structure is:
1449 * 1451 *
1450 * struct linger { 1452 * struct linger {
1451 * int l_onoff; // option on/off 1453 * int l_onoff; // option on/off
1452 * int l_linger; // linger time 1454 * int l_linger; // linger time
1453 * }; 1455 * };
1454 * 1456 *
1455 * To enable the option, set l_onoff to 1. If the l_linger value is set 1457 * To enable the option, set l_onoff to 1. If the l_linger value is set
1456 * to 0, calling close() is the same as the ABORT primitive. If the 1458 * to 0, calling close() is the same as the ABORT primitive. If the
1457 * value is set to a negative value, the setsockopt() call will return 1459 * value is set to a negative value, the setsockopt() call will return
1458 * an error. If the value is set to a positive value linger_time, the 1460 * an error. If the value is set to a positive value linger_time, the
1459 * close() can be blocked for at most linger_time ms. If the graceful 1461 * close() can be blocked for at most linger_time ms. If the graceful
1460 * shutdown phase does not finish during this period, close() will 1462 * shutdown phase does not finish during this period, close() will
1461 * return but the graceful shutdown phase continues in the system. 1463 * return but the graceful shutdown phase continues in the system.
1462 */ 1464 */
1463 SCTP_STATIC void sctp_close(struct sock *sk, long timeout) 1465 SCTP_STATIC void sctp_close(struct sock *sk, long timeout)
1464 { 1466 {
1465 struct net *net = sock_net(sk); 1467 struct net *net = sock_net(sk);
1466 struct sctp_endpoint *ep; 1468 struct sctp_endpoint *ep;
1467 struct sctp_association *asoc; 1469 struct sctp_association *asoc;
1468 struct list_head *pos, *temp; 1470 struct list_head *pos, *temp;
1469 unsigned int data_was_unread; 1471 unsigned int data_was_unread;
1470 1472
1471 SCTP_DEBUG_PRINTK("sctp_close(sk: 0x%p, timeout:%ld)\n", sk, timeout); 1473 SCTP_DEBUG_PRINTK("sctp_close(sk: 0x%p, timeout:%ld)\n", sk, timeout);
1472 1474
1473 sctp_lock_sock(sk); 1475 sctp_lock_sock(sk);
1474 sk->sk_shutdown = SHUTDOWN_MASK; 1476 sk->sk_shutdown = SHUTDOWN_MASK;
1475 sk->sk_state = SCTP_SS_CLOSING; 1477 sk->sk_state = SCTP_SS_CLOSING;
1476 1478
1477 ep = sctp_sk(sk)->ep; 1479 ep = sctp_sk(sk)->ep;
1478 1480
1479 /* Clean up any skbs sitting on the receive queue. */ 1481 /* Clean up any skbs sitting on the receive queue. */
1480 data_was_unread = sctp_queue_purge_ulpevents(&sk->sk_receive_queue); 1482 data_was_unread = sctp_queue_purge_ulpevents(&sk->sk_receive_queue);
1481 data_was_unread += sctp_queue_purge_ulpevents(&sctp_sk(sk)->pd_lobby); 1483 data_was_unread += sctp_queue_purge_ulpevents(&sctp_sk(sk)->pd_lobby);
1482 1484
1483 /* Walk all associations on an endpoint. */ 1485 /* Walk all associations on an endpoint. */
1484 list_for_each_safe(pos, temp, &ep->asocs) { 1486 list_for_each_safe(pos, temp, &ep->asocs) {
1485 asoc = list_entry(pos, struct sctp_association, asocs); 1487 asoc = list_entry(pos, struct sctp_association, asocs);
1486 1488
1487 if (sctp_style(sk, TCP)) { 1489 if (sctp_style(sk, TCP)) {
1488 /* A closed association can still be in the list if 1490 /* A closed association can still be in the list if
1489 * it belongs to a TCP-style listening socket that is 1491 * it belongs to a TCP-style listening socket that is
1490 * not yet accepted. If so, free it. If not, send an 1492 * not yet accepted. If so, free it. If not, send an
1491 * ABORT or SHUTDOWN based on the linger options. 1493 * ABORT or SHUTDOWN based on the linger options.
1492 */ 1494 */
1493 if (sctp_state(asoc, CLOSED)) { 1495 if (sctp_state(asoc, CLOSED)) {
1494 sctp_unhash_established(asoc); 1496 sctp_unhash_established(asoc);
1495 sctp_association_free(asoc); 1497 sctp_association_free(asoc);
1496 continue; 1498 continue;
1497 } 1499 }
1498 } 1500 }
1499 1501
1500 if (data_was_unread || !skb_queue_empty(&asoc->ulpq.lobby) || 1502 if (data_was_unread || !skb_queue_empty(&asoc->ulpq.lobby) ||
1501 !skb_queue_empty(&asoc->ulpq.reasm) || 1503 !skb_queue_empty(&asoc->ulpq.reasm) ||
1502 (sock_flag(sk, SOCK_LINGER) && !sk->sk_lingertime)) { 1504 (sock_flag(sk, SOCK_LINGER) && !sk->sk_lingertime)) {
1503 struct sctp_chunk *chunk; 1505 struct sctp_chunk *chunk;
1504 1506
1505 chunk = sctp_make_abort_user(asoc, NULL, 0); 1507 chunk = sctp_make_abort_user(asoc, NULL, 0);
1506 if (chunk) 1508 if (chunk)
1507 sctp_primitive_ABORT(net, asoc, chunk); 1509 sctp_primitive_ABORT(net, asoc, chunk);
1508 } else 1510 } else
1509 sctp_primitive_SHUTDOWN(net, asoc, NULL); 1511 sctp_primitive_SHUTDOWN(net, asoc, NULL);
1510 } 1512 }
1511 1513
1512 /* On a TCP-style socket, block for at most linger_time if set. */ 1514 /* On a TCP-style socket, block for at most linger_time if set. */
1513 if (sctp_style(sk, TCP) && timeout) 1515 if (sctp_style(sk, TCP) && timeout)
1514 sctp_wait_for_close(sk, timeout); 1516 sctp_wait_for_close(sk, timeout);
1515 1517
1516 /* This will run the backlog queue. */ 1518 /* This will run the backlog queue. */
1517 sctp_release_sock(sk); 1519 sctp_release_sock(sk);
1518 1520
1519 /* Supposedly, no process has access to the socket, but 1521 /* Supposedly, no process has access to the socket, but
1520 * the net layers still may. 1522 * the net layers still may.
1521 */ 1523 */
1522 sctp_local_bh_disable(); 1524 sctp_local_bh_disable();
1523 sctp_bh_lock_sock(sk); 1525 sctp_bh_lock_sock(sk);
1524 1526
1525 /* Hold the sock, since sk_common_release() will put sock_put() 1527 /* Hold the sock, since sk_common_release() will put sock_put()
1526 * and we have just a little more cleanup. 1528 * and we have just a little more cleanup.
1527 */ 1529 */
1528 sock_hold(sk); 1530 sock_hold(sk);
1529 sk_common_release(sk); 1531 sk_common_release(sk);
1530 1532
1531 sctp_bh_unlock_sock(sk); 1533 sctp_bh_unlock_sock(sk);
1532 sctp_local_bh_enable(); 1534 sctp_local_bh_enable();
1533 1535
1534 sock_put(sk); 1536 sock_put(sk);
1535 1537
1536 SCTP_DBG_OBJCNT_DEC(sock); 1538 SCTP_DBG_OBJCNT_DEC(sock);
1537 } 1539 }
1538 1540
1539 /* Handle EPIPE error. */ 1541 /* Handle EPIPE error. */
1540 static int sctp_error(struct sock *sk, int flags, int err) 1542 static int sctp_error(struct sock *sk, int flags, int err)
1541 { 1543 {
1542 if (err == -EPIPE) 1544 if (err == -EPIPE)
1543 err = sock_error(sk) ? : -EPIPE; 1545 err = sock_error(sk) ? : -EPIPE;
1544 if (err == -EPIPE && !(flags & MSG_NOSIGNAL)) 1546 if (err == -EPIPE && !(flags & MSG_NOSIGNAL))
1545 send_sig(SIGPIPE, current, 0); 1547 send_sig(SIGPIPE, current, 0);
1546 return err; 1548 return err;
1547 } 1549 }
1548 1550
1549 /* API 3.1.3 sendmsg() - UDP Style Syntax 1551 /* API 3.1.3 sendmsg() - UDP Style Syntax
1550 * 1552 *
1551 * An application uses sendmsg() and recvmsg() calls to transmit data to 1553 * An application uses sendmsg() and recvmsg() calls to transmit data to
1552 * and receive data from its peer. 1554 * and receive data from its peer.
1553 * 1555 *
1554 * ssize_t sendmsg(int socket, const struct msghdr *message, 1556 * ssize_t sendmsg(int socket, const struct msghdr *message,
1555 * int flags); 1557 * int flags);
1556 * 1558 *
1557 * socket - the socket descriptor of the endpoint. 1559 * socket - the socket descriptor of the endpoint.
1558 * message - pointer to the msghdr structure which contains a single 1560 * message - pointer to the msghdr structure which contains a single
1559 * user message and possibly some ancillary data. 1561 * user message and possibly some ancillary data.
1560 * 1562 *
1561 * See Section 5 for complete description of the data 1563 * See Section 5 for complete description of the data
1562 * structures. 1564 * structures.
1563 * 1565 *
1564 * flags - flags sent or received with the user message, see Section 1566 * flags - flags sent or received with the user message, see Section
1565 * 5 for complete description of the flags. 1567 * 5 for complete description of the flags.
1566 * 1568 *
1567 * Note: This function could use a rewrite especially when explicit 1569 * Note: This function could use a rewrite especially when explicit
1568 * connect support comes in. 1570 * connect support comes in.
1569 */ 1571 */
1570 /* BUG: We do not implement the equivalent of sk_stream_wait_memory(). */ 1572 /* BUG: We do not implement the equivalent of sk_stream_wait_memory(). */
1571 1573
1572 SCTP_STATIC int sctp_msghdr_parse(const struct msghdr *, sctp_cmsgs_t *); 1574 SCTP_STATIC int sctp_msghdr_parse(const struct msghdr *, sctp_cmsgs_t *);
1573 1575
1574 SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk, 1576 SCTP_STATIC int sctp_sendmsg(struct kiocb *iocb, struct sock *sk,
1575 struct msghdr *msg, size_t msg_len) 1577 struct msghdr *msg, size_t msg_len)
1576 { 1578 {
1577 struct net *net = sock_net(sk); 1579 struct net *net = sock_net(sk);
1578 struct sctp_sock *sp; 1580 struct sctp_sock *sp;
1579 struct sctp_endpoint *ep; 1581 struct sctp_endpoint *ep;
1580 struct sctp_association *new_asoc=NULL, *asoc=NULL; 1582 struct sctp_association *new_asoc=NULL, *asoc=NULL;
1581 struct sctp_transport *transport, *chunk_tp; 1583 struct sctp_transport *transport, *chunk_tp;
1582 struct sctp_chunk *chunk; 1584 struct sctp_chunk *chunk;
1583 union sctp_addr to; 1585 union sctp_addr to;
1584 struct sockaddr *msg_name = NULL; 1586 struct sockaddr *msg_name = NULL;
1585 struct sctp_sndrcvinfo default_sinfo; 1587 struct sctp_sndrcvinfo default_sinfo;
1586 struct sctp_sndrcvinfo *sinfo; 1588 struct sctp_sndrcvinfo *sinfo;
1587 struct sctp_initmsg *sinit; 1589 struct sctp_initmsg *sinit;
1588 sctp_assoc_t associd = 0; 1590 sctp_assoc_t associd = 0;
1589 sctp_cmsgs_t cmsgs = { NULL }; 1591 sctp_cmsgs_t cmsgs = { NULL };
1590 int err; 1592 int err;
1591 sctp_scope_t scope; 1593 sctp_scope_t scope;
1592 long timeo; 1594 long timeo;
1593 __u16 sinfo_flags = 0; 1595 __u16 sinfo_flags = 0;
1594 struct sctp_datamsg *datamsg; 1596 struct sctp_datamsg *datamsg;
1595 int msg_flags = msg->msg_flags; 1597 int msg_flags = msg->msg_flags;
1596 1598
1597 SCTP_DEBUG_PRINTK("sctp_sendmsg(sk: %p, msg: %p, msg_len: %zu)\n", 1599 SCTP_DEBUG_PRINTK("sctp_sendmsg(sk: %p, msg: %p, msg_len: %zu)\n",
1598 sk, msg, msg_len); 1600 sk, msg, msg_len);
1599 1601
1600 err = 0; 1602 err = 0;
1601 sp = sctp_sk(sk); 1603 sp = sctp_sk(sk);
1602 ep = sp->ep; 1604 ep = sp->ep;
1603 1605
1604 SCTP_DEBUG_PRINTK("Using endpoint: %p.\n", ep); 1606 SCTP_DEBUG_PRINTK("Using endpoint: %p.\n", ep);
1605 1607
1606 /* We cannot send a message over a TCP-style listening socket. */ 1608 /* We cannot send a message over a TCP-style listening socket. */
1607 if (sctp_style(sk, TCP) && sctp_sstate(sk, LISTENING)) { 1609 if (sctp_style(sk, TCP) && sctp_sstate(sk, LISTENING)) {
1608 err = -EPIPE; 1610 err = -EPIPE;
1609 goto out_nounlock; 1611 goto out_nounlock;
1610 } 1612 }
1611 1613
1612 /* Parse out the SCTP CMSGs. */ 1614 /* Parse out the SCTP CMSGs. */
1613 err = sctp_msghdr_parse(msg, &cmsgs); 1615 err = sctp_msghdr_parse(msg, &cmsgs);
1614 1616
1615 if (err) { 1617 if (err) {
1616 SCTP_DEBUG_PRINTK("msghdr parse err = %x\n", err); 1618 SCTP_DEBUG_PRINTK("msghdr parse err = %x\n", err);
1617 goto out_nounlock; 1619 goto out_nounlock;
1618 } 1620 }
1619 1621
1620 /* Fetch the destination address for this packet. This 1622 /* Fetch the destination address for this packet. This
1621 * address only selects the association--it is not necessarily 1623 * address only selects the association--it is not necessarily
1622 * the address we will send to. 1624 * the address we will send to.
1623 * For a peeled-off socket, msg_name is ignored. 1625 * For a peeled-off socket, msg_name is ignored.
1624 */ 1626 */
1625 if (!sctp_style(sk, UDP_HIGH_BANDWIDTH) && msg->msg_name) { 1627 if (!sctp_style(sk, UDP_HIGH_BANDWIDTH) && msg->msg_name) {
1626 int msg_namelen = msg->msg_namelen; 1628 int msg_namelen = msg->msg_namelen;
1627 1629
1628 err = sctp_verify_addr(sk, (union sctp_addr *)msg->msg_name, 1630 err = sctp_verify_addr(sk, (union sctp_addr *)msg->msg_name,
1629 msg_namelen); 1631 msg_namelen);
1630 if (err) 1632 if (err)
1631 return err; 1633 return err;
1632 1634
1633 if (msg_namelen > sizeof(to)) 1635 if (msg_namelen > sizeof(to))
1634 msg_namelen = sizeof(to); 1636 msg_namelen = sizeof(to);
1635 memcpy(&to, msg->msg_name, msg_namelen); 1637 memcpy(&to, msg->msg_name, msg_namelen);
1636 msg_name = msg->msg_name; 1638 msg_name = msg->msg_name;
1637 } 1639 }
1638 1640
1639 sinfo = cmsgs.info; 1641 sinfo = cmsgs.info;
1640 sinit = cmsgs.init; 1642 sinit = cmsgs.init;
1641 1643
1642 /* Did the user specify SNDRCVINFO? */ 1644 /* Did the user specify SNDRCVINFO? */
1643 if (sinfo) { 1645 if (sinfo) {
1644 sinfo_flags = sinfo->sinfo_flags; 1646 sinfo_flags = sinfo->sinfo_flags;
1645 associd = sinfo->sinfo_assoc_id; 1647 associd = sinfo->sinfo_assoc_id;
1646 } 1648 }
1647 1649
1648 SCTP_DEBUG_PRINTK("msg_len: %zu, sinfo_flags: 0x%x\n", 1650 SCTP_DEBUG_PRINTK("msg_len: %zu, sinfo_flags: 0x%x\n",
1649 msg_len, sinfo_flags); 1651 msg_len, sinfo_flags);
1650 1652
1651 /* SCTP_EOF or SCTP_ABORT cannot be set on a TCP-style socket. */ 1653 /* SCTP_EOF or SCTP_ABORT cannot be set on a TCP-style socket. */
1652 if (sctp_style(sk, TCP) && (sinfo_flags & (SCTP_EOF | SCTP_ABORT))) { 1654 if (sctp_style(sk, TCP) && (sinfo_flags & (SCTP_EOF | SCTP_ABORT))) {
1653 err = -EINVAL; 1655 err = -EINVAL;
1654 goto out_nounlock; 1656 goto out_nounlock;
1655 } 1657 }
1656 1658
1657 /* If SCTP_EOF is set, no data can be sent. Disallow sending zero 1659 /* If SCTP_EOF is set, no data can be sent. Disallow sending zero
1658 * length messages when SCTP_EOF|SCTP_ABORT is not set. 1660 * length messages when SCTP_EOF|SCTP_ABORT is not set.
1659 * If SCTP_ABORT is set, the message length could be non zero with 1661 * If SCTP_ABORT is set, the message length could be non zero with
1660 * the msg_iov set to the user abort reason. 1662 * the msg_iov set to the user abort reason.
1661 */ 1663 */
1662 if (((sinfo_flags & SCTP_EOF) && (msg_len > 0)) || 1664 if (((sinfo_flags & SCTP_EOF) && (msg_len > 0)) ||
1663 (!(sinfo_flags & (SCTP_EOF|SCTP_ABORT)) && (msg_len == 0))) { 1665 (!(sinfo_flags & (SCTP_EOF|SCTP_ABORT)) && (msg_len == 0))) {
1664 err = -EINVAL; 1666 err = -EINVAL;
1665 goto out_nounlock; 1667 goto out_nounlock;
1666 } 1668 }
1667 1669
1668 /* If SCTP_ADDR_OVER is set, there must be an address 1670 /* If SCTP_ADDR_OVER is set, there must be an address
1669 * specified in msg_name. 1671 * specified in msg_name.
1670 */ 1672 */
1671 if ((sinfo_flags & SCTP_ADDR_OVER) && (!msg->msg_name)) { 1673 if ((sinfo_flags & SCTP_ADDR_OVER) && (!msg->msg_name)) {
1672 err = -EINVAL; 1674 err = -EINVAL;
1673 goto out_nounlock; 1675 goto out_nounlock;
1674 } 1676 }
1675 1677
1676 transport = NULL; 1678 transport = NULL;
1677 1679
1678 SCTP_DEBUG_PRINTK("About to look up association.\n"); 1680 SCTP_DEBUG_PRINTK("About to look up association.\n");
1679 1681
1680 sctp_lock_sock(sk); 1682 sctp_lock_sock(sk);
1681 1683
1682 /* If a msg_name has been specified, assume this is to be used. */ 1684 /* If a msg_name has been specified, assume this is to be used. */
1683 if (msg_name) { 1685 if (msg_name) {
1684 /* Look for a matching association on the endpoint. */ 1686 /* Look for a matching association on the endpoint. */
1685 asoc = sctp_endpoint_lookup_assoc(ep, &to, &transport); 1687 asoc = sctp_endpoint_lookup_assoc(ep, &to, &transport);
1686 if (!asoc) { 1688 if (!asoc) {
1687 /* If we could not find a matching association on the 1689 /* If we could not find a matching association on the
1688 * endpoint, make sure that it is not a TCP-style 1690 * endpoint, make sure that it is not a TCP-style
1689 * socket that already has an association or there is 1691 * socket that already has an association or there is
1690 * no peeled-off association on another socket. 1692 * no peeled-off association on another socket.
1691 */ 1693 */
1692 if ((sctp_style(sk, TCP) && 1694 if ((sctp_style(sk, TCP) &&
1693 sctp_sstate(sk, ESTABLISHED)) || 1695 sctp_sstate(sk, ESTABLISHED)) ||
1694 sctp_endpoint_is_peeled_off(ep, &to)) { 1696 sctp_endpoint_is_peeled_off(ep, &to)) {
1695 err = -EADDRNOTAVAIL; 1697 err = -EADDRNOTAVAIL;
1696 goto out_unlock; 1698 goto out_unlock;
1697 } 1699 }
1698 } 1700 }
1699 } else { 1701 } else {
1700 asoc = sctp_id2assoc(sk, associd); 1702 asoc = sctp_id2assoc(sk, associd);
1701 if (!asoc) { 1703 if (!asoc) {
1702 err = -EPIPE; 1704 err = -EPIPE;
1703 goto out_unlock; 1705 goto out_unlock;
1704 } 1706 }
1705 } 1707 }
1706 1708
1707 if (asoc) { 1709 if (asoc) {
1708 SCTP_DEBUG_PRINTK("Just looked up association: %p.\n", asoc); 1710 SCTP_DEBUG_PRINTK("Just looked up association: %p.\n", asoc);
1709 1711
1710 /* We cannot send a message on a TCP-style SCTP_SS_ESTABLISHED 1712 /* We cannot send a message on a TCP-style SCTP_SS_ESTABLISHED
1711 * socket that has an association in CLOSED state. This can 1713 * socket that has an association in CLOSED state. This can
1712 * happen when an accepted socket has an association that is 1714 * happen when an accepted socket has an association that is
1713 * already CLOSED. 1715 * already CLOSED.
1714 */ 1716 */
1715 if (sctp_state(asoc, CLOSED) && sctp_style(sk, TCP)) { 1717 if (sctp_state(asoc, CLOSED) && sctp_style(sk, TCP)) {
1716 err = -EPIPE; 1718 err = -EPIPE;
1717 goto out_unlock; 1719 goto out_unlock;
1718 } 1720 }
1719 1721
1720 if (sinfo_flags & SCTP_EOF) { 1722 if (sinfo_flags & SCTP_EOF) {
1721 SCTP_DEBUG_PRINTK("Shutting down association: %p\n", 1723 SCTP_DEBUG_PRINTK("Shutting down association: %p\n",
1722 asoc); 1724 asoc);
1723 sctp_primitive_SHUTDOWN(net, asoc, NULL); 1725 sctp_primitive_SHUTDOWN(net, asoc, NULL);
1724 err = 0; 1726 err = 0;
1725 goto out_unlock; 1727 goto out_unlock;
1726 } 1728 }
1727 if (sinfo_flags & SCTP_ABORT) { 1729 if (sinfo_flags & SCTP_ABORT) {
1728 1730
1729 chunk = sctp_make_abort_user(asoc, msg, msg_len); 1731 chunk = sctp_make_abort_user(asoc, msg, msg_len);
1730 if (!chunk) { 1732 if (!chunk) {
1731 err = -ENOMEM; 1733 err = -ENOMEM;
1732 goto out_unlock; 1734 goto out_unlock;
1733 } 1735 }
1734 1736
1735 SCTP_DEBUG_PRINTK("Aborting association: %p\n", asoc); 1737 SCTP_DEBUG_PRINTK("Aborting association: %p\n", asoc);
1736 sctp_primitive_ABORT(net, asoc, chunk); 1738 sctp_primitive_ABORT(net, asoc, chunk);
1737 err = 0; 1739 err = 0;
1738 goto out_unlock; 1740 goto out_unlock;
1739 } 1741 }
1740 } 1742 }
1741 1743
1742 /* Do we need to create the association? */ 1744 /* Do we need to create the association? */
1743 if (!asoc) { 1745 if (!asoc) {
1744 SCTP_DEBUG_PRINTK("There is no association yet.\n"); 1746 SCTP_DEBUG_PRINTK("There is no association yet.\n");
1745 1747
1746 if (sinfo_flags & (SCTP_EOF | SCTP_ABORT)) { 1748 if (sinfo_flags & (SCTP_EOF | SCTP_ABORT)) {
1747 err = -EINVAL; 1749 err = -EINVAL;
1748 goto out_unlock; 1750 goto out_unlock;
1749 } 1751 }
1750 1752
1751 /* Check for invalid stream against the stream counts, 1753 /* Check for invalid stream against the stream counts,
1752 * either the default or the user specified stream counts. 1754 * either the default or the user specified stream counts.
1753 */ 1755 */
1754 if (sinfo) { 1756 if (sinfo) {
1755 if (!sinit || (sinit && !sinit->sinit_num_ostreams)) { 1757 if (!sinit || (sinit && !sinit->sinit_num_ostreams)) {
1756 /* Check against the defaults. */ 1758 /* Check against the defaults. */
1757 if (sinfo->sinfo_stream >= 1759 if (sinfo->sinfo_stream >=
1758 sp->initmsg.sinit_num_ostreams) { 1760 sp->initmsg.sinit_num_ostreams) {
1759 err = -EINVAL; 1761 err = -EINVAL;
1760 goto out_unlock; 1762 goto out_unlock;
1761 } 1763 }
1762 } else { 1764 } else {
1763 /* Check against the requested. */ 1765 /* Check against the requested. */
1764 if (sinfo->sinfo_stream >= 1766 if (sinfo->sinfo_stream >=
1765 sinit->sinit_num_ostreams) { 1767 sinit->sinit_num_ostreams) {
1766 err = -EINVAL; 1768 err = -EINVAL;
1767 goto out_unlock; 1769 goto out_unlock;
1768 } 1770 }
1769 } 1771 }
1770 } 1772 }
1771 1773
1772 /* 1774 /*
1773 * API 3.1.2 bind() - UDP Style Syntax 1775 * API 3.1.2 bind() - UDP Style Syntax
1774 * If a bind() or sctp_bindx() is not called prior to a 1776 * If a bind() or sctp_bindx() is not called prior to a
1775 * sendmsg() call that initiates a new association, the 1777 * sendmsg() call that initiates a new association, the
1776 * system picks an ephemeral port and will choose an address 1778 * system picks an ephemeral port and will choose an address
1777 * set equivalent to binding with a wildcard address. 1779 * set equivalent to binding with a wildcard address.
1778 */ 1780 */
1779 if (!ep->base.bind_addr.port) { 1781 if (!ep->base.bind_addr.port) {
1780 if (sctp_autobind(sk)) { 1782 if (sctp_autobind(sk)) {
1781 err = -EAGAIN; 1783 err = -EAGAIN;
1782 goto out_unlock; 1784 goto out_unlock;
1783 } 1785 }
1784 } else { 1786 } else {
1785 /* 1787 /*
1786 * If an unprivileged user inherits a one-to-many 1788 * If an unprivileged user inherits a one-to-many
1787 * style socket with open associations on a privileged 1789 * style socket with open associations on a privileged
1788 * port, it MAY be permitted to accept new associations, 1790 * port, it MAY be permitted to accept new associations,
1789 * but it SHOULD NOT be permitted to open new 1791 * but it SHOULD NOT be permitted to open new
1790 * associations. 1792 * associations.
1791 */ 1793 */
1792 if (ep->base.bind_addr.port < PROT_SOCK && 1794 if (ep->base.bind_addr.port < PROT_SOCK &&
1793 !capable(CAP_NET_BIND_SERVICE)) { 1795 !ns_capable(net->user_ns, CAP_NET_BIND_SERVICE)) {
1794 err = -EACCES; 1796 err = -EACCES;
1795 goto out_unlock; 1797 goto out_unlock;
1796 } 1798 }
1797 } 1799 }
1798 1800
1799 scope = sctp_scope(&to); 1801 scope = sctp_scope(&to);
1800 new_asoc = sctp_association_new(ep, sk, scope, GFP_KERNEL); 1802 new_asoc = sctp_association_new(ep, sk, scope, GFP_KERNEL);
1801 if (!new_asoc) { 1803 if (!new_asoc) {
1802 err = -ENOMEM; 1804 err = -ENOMEM;
1803 goto out_unlock; 1805 goto out_unlock;
1804 } 1806 }
1805 asoc = new_asoc; 1807 asoc = new_asoc;
1806 err = sctp_assoc_set_bind_addr_from_ep(asoc, scope, GFP_KERNEL); 1808 err = sctp_assoc_set_bind_addr_from_ep(asoc, scope, GFP_KERNEL);
1807 if (err < 0) { 1809 if (err < 0) {
1808 err = -ENOMEM; 1810 err = -ENOMEM;
1809 goto out_free; 1811 goto out_free;
1810 } 1812 }
1811 1813
1812 /* If the SCTP_INIT ancillary data is specified, set all 1814 /* If the SCTP_INIT ancillary data is specified, set all
1813 * the association init values accordingly. 1815 * the association init values accordingly.
1814 */ 1816 */
1815 if (sinit) { 1817 if (sinit) {
1816 if (sinit->sinit_num_ostreams) { 1818 if (sinit->sinit_num_ostreams) {
1817 asoc->c.sinit_num_ostreams = 1819 asoc->c.sinit_num_ostreams =
1818 sinit->sinit_num_ostreams; 1820 sinit->sinit_num_ostreams;
1819 } 1821 }
1820 if (sinit->sinit_max_instreams) { 1822 if (sinit->sinit_max_instreams) {
1821 asoc->c.sinit_max_instreams = 1823 asoc->c.sinit_max_instreams =
1822 sinit->sinit_max_instreams; 1824 sinit->sinit_max_instreams;
1823 } 1825 }
1824 if (sinit->sinit_max_attempts) { 1826 if (sinit->sinit_max_attempts) {
1825 asoc->max_init_attempts 1827 asoc->max_init_attempts
1826 = sinit->sinit_max_attempts; 1828 = sinit->sinit_max_attempts;
1827 } 1829 }
1828 if (sinit->sinit_max_init_timeo) { 1830 if (sinit->sinit_max_init_timeo) {
1829 asoc->max_init_timeo = 1831 asoc->max_init_timeo =
1830 msecs_to_jiffies(sinit->sinit_max_init_timeo); 1832 msecs_to_jiffies(sinit->sinit_max_init_timeo);
1831 } 1833 }
1832 } 1834 }
1833 1835
1834 /* Prime the peer's transport structures. */ 1836 /* Prime the peer's transport structures. */
1835 transport = sctp_assoc_add_peer(asoc, &to, GFP_KERNEL, SCTP_UNKNOWN); 1837 transport = sctp_assoc_add_peer(asoc, &to, GFP_KERNEL, SCTP_UNKNOWN);
1836 if (!transport) { 1838 if (!transport) {
1837 err = -ENOMEM; 1839 err = -ENOMEM;
1838 goto out_free; 1840 goto out_free;
1839 } 1841 }
1840 } 1842 }
1841 1843
1842 /* ASSERT: we have a valid association at this point. */ 1844 /* ASSERT: we have a valid association at this point. */
1843 SCTP_DEBUG_PRINTK("We have a valid association.\n"); 1845 SCTP_DEBUG_PRINTK("We have a valid association.\n");
1844 1846
1845 if (!sinfo) { 1847 if (!sinfo) {
1846 /* If the user didn't specify SNDRCVINFO, make up one with 1848 /* If the user didn't specify SNDRCVINFO, make up one with
1847 * some defaults. 1849 * some defaults.
1848 */ 1850 */
1849 memset(&default_sinfo, 0, sizeof(default_sinfo)); 1851 memset(&default_sinfo, 0, sizeof(default_sinfo));
1850 default_sinfo.sinfo_stream = asoc->default_stream; 1852 default_sinfo.sinfo_stream = asoc->default_stream;
1851 default_sinfo.sinfo_flags = asoc->default_flags; 1853 default_sinfo.sinfo_flags = asoc->default_flags;
1852 default_sinfo.sinfo_ppid = asoc->default_ppid; 1854 default_sinfo.sinfo_ppid = asoc->default_ppid;
1853 default_sinfo.sinfo_context = asoc->default_context; 1855 default_sinfo.sinfo_context = asoc->default_context;
1854 default_sinfo.sinfo_timetolive = asoc->default_timetolive; 1856 default_sinfo.sinfo_timetolive = asoc->default_timetolive;
1855 default_sinfo.sinfo_assoc_id = sctp_assoc2id(asoc); 1857 default_sinfo.sinfo_assoc_id = sctp_assoc2id(asoc);
1856 sinfo = &default_sinfo; 1858 sinfo = &default_sinfo;
1857 } 1859 }
1858 1860
1859 /* API 7.1.7, the sndbuf size per association bounds the 1861 /* API 7.1.7, the sndbuf size per association bounds the
1860 * maximum size of data that can be sent in a single send call. 1862 * maximum size of data that can be sent in a single send call.
1861 */ 1863 */
1862 if (msg_len > sk->sk_sndbuf) { 1864 if (msg_len > sk->sk_sndbuf) {
1863 err = -EMSGSIZE; 1865 err = -EMSGSIZE;
1864 goto out_free; 1866 goto out_free;
1865 } 1867 }
1866 1868
1867 if (asoc->pmtu_pending) 1869 if (asoc->pmtu_pending)
1868 sctp_assoc_pending_pmtu(sk, asoc); 1870 sctp_assoc_pending_pmtu(sk, asoc);
1869 1871
1870 /* If fragmentation is disabled and the message length exceeds the 1872 /* If fragmentation is disabled and the message length exceeds the
1871 * association fragmentation point, return EMSGSIZE. The I-D 1873 * association fragmentation point, return EMSGSIZE. The I-D
1872 * does not specify what this error is, but this looks like 1874 * does not specify what this error is, but this looks like
1873 * a great fit. 1875 * a great fit.
1874 */ 1876 */
1875 if (sctp_sk(sk)->disable_fragments && (msg_len > asoc->frag_point)) { 1877 if (sctp_sk(sk)->disable_fragments && (msg_len > asoc->frag_point)) {
1876 err = -EMSGSIZE; 1878 err = -EMSGSIZE;
1877 goto out_free; 1879 goto out_free;
1878 } 1880 }
1879 1881
1880 /* Check for invalid stream. */ 1882 /* Check for invalid stream. */
1881 if (sinfo->sinfo_stream >= asoc->c.sinit_num_ostreams) { 1883 if (sinfo->sinfo_stream >= asoc->c.sinit_num_ostreams) {
1882 err = -EINVAL; 1884 err = -EINVAL;
1883 goto out_free; 1885 goto out_free;
1884 } 1886 }
1885 1887
1886 timeo = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT); 1888 timeo = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);
1887 if (!sctp_wspace(asoc)) { 1889 if (!sctp_wspace(asoc)) {
1888 err = sctp_wait_for_sndbuf(asoc, &timeo, msg_len); 1890 err = sctp_wait_for_sndbuf(asoc, &timeo, msg_len);
1889 if (err) 1891 if (err)
1890 goto out_free; 1892 goto out_free;
1891 } 1893 }
1892 1894
1893 /* If an address is passed with the sendto/sendmsg call, it is used 1895 /* If an address is passed with the sendto/sendmsg call, it is used
1894 * to override the primary destination address in the TCP model, or 1896 * to override the primary destination address in the TCP model, or
1895 * when SCTP_ADDR_OVER flag is set in the UDP model. 1897 * when SCTP_ADDR_OVER flag is set in the UDP model.
1896 */ 1898 */
1897 if ((sctp_style(sk, TCP) && msg_name) || 1899 if ((sctp_style(sk, TCP) && msg_name) ||
1898 (sinfo_flags & SCTP_ADDR_OVER)) { 1900 (sinfo_flags & SCTP_ADDR_OVER)) {
1899 chunk_tp = sctp_assoc_lookup_paddr(asoc, &to); 1901 chunk_tp = sctp_assoc_lookup_paddr(asoc, &to);
1900 if (!chunk_tp) { 1902 if (!chunk_tp) {
1901 err = -EINVAL; 1903 err = -EINVAL;
1902 goto out_free; 1904 goto out_free;
1903 } 1905 }
1904 } else 1906 } else
1905 chunk_tp = NULL; 1907 chunk_tp = NULL;
1906 1908
1907 /* Auto-connect, if we aren't connected already. */ 1909 /* Auto-connect, if we aren't connected already. */
1908 if (sctp_state(asoc, CLOSED)) { 1910 if (sctp_state(asoc, CLOSED)) {
1909 err = sctp_primitive_ASSOCIATE(net, asoc, NULL); 1911 err = sctp_primitive_ASSOCIATE(net, asoc, NULL);
1910 if (err < 0) 1912 if (err < 0)
1911 goto out_free; 1913 goto out_free;
1912 SCTP_DEBUG_PRINTK("We associated primitively.\n"); 1914 SCTP_DEBUG_PRINTK("We associated primitively.\n");
1913 } 1915 }
1914 1916
1915 /* Break the message into multiple chunks of maximum size. */ 1917 /* Break the message into multiple chunks of maximum size. */
1916 datamsg = sctp_datamsg_from_user(asoc, sinfo, msg, msg_len); 1918 datamsg = sctp_datamsg_from_user(asoc, sinfo, msg, msg_len);
1917 if (!datamsg) { 1919 if (!datamsg) {
1918 err = -ENOMEM; 1920 err = -ENOMEM;
1919 goto out_free; 1921 goto out_free;
1920 } 1922 }
1921 1923
1922 /* Now send the (possibly) fragmented message. */ 1924 /* Now send the (possibly) fragmented message. */
1923 list_for_each_entry(chunk, &datamsg->chunks, frag_list) { 1925 list_for_each_entry(chunk, &datamsg->chunks, frag_list) {
1924 sctp_chunk_hold(chunk); 1926 sctp_chunk_hold(chunk);
1925 1927
1926 /* Do accounting for the write space. */ 1928 /* Do accounting for the write space. */
1927 sctp_set_owner_w(chunk); 1929 sctp_set_owner_w(chunk);
1928 1930
1929 chunk->transport = chunk_tp; 1931 chunk->transport = chunk_tp;
1930 } 1932 }
1931 1933
1932 /* Send it to the lower layers. Note: all chunks 1934 /* Send it to the lower layers. Note: all chunks
1933 * must either fail or succeed. The lower layer 1935 * must either fail or succeed. The lower layer
1934 * works that way today. Keep it that way or this 1936 * works that way today. Keep it that way or this
1935 * breaks. 1937 * breaks.
1936 */ 1938 */
1937 err = sctp_primitive_SEND(net, asoc, datamsg); 1939 err = sctp_primitive_SEND(net, asoc, datamsg);
1938 /* Did the lower layer accept the chunk? */ 1940 /* Did the lower layer accept the chunk? */
1939 if (err) 1941 if (err)
1940 sctp_datamsg_free(datamsg); 1942 sctp_datamsg_free(datamsg);
1941 else 1943 else
1942 sctp_datamsg_put(datamsg); 1944 sctp_datamsg_put(datamsg);
1943 1945
1944 SCTP_DEBUG_PRINTK("We sent primitively.\n"); 1946 SCTP_DEBUG_PRINTK("We sent primitively.\n");
1945 1947
1946 if (err) 1948 if (err)
1947 goto out_free; 1949 goto out_free;
1948 else 1950 else
1949 err = msg_len; 1951 err = msg_len;
1950 1952
1951 /* If we are already past ASSOCIATE, the lower 1953 /* If we are already past ASSOCIATE, the lower
1952 * layers are responsible for association cleanup. 1954 * layers are responsible for association cleanup.
1953 */ 1955 */
1954 goto out_unlock; 1956 goto out_unlock;
1955 1957
1956 out_free: 1958 out_free:
1957 if (new_asoc) { 1959 if (new_asoc) {
1958 sctp_unhash_established(asoc); 1960 sctp_unhash_established(asoc);
1959 sctp_association_free(asoc); 1961 sctp_association_free(asoc);
1960 } 1962 }
1961 out_unlock: 1963 out_unlock:
1962 sctp_release_sock(sk); 1964 sctp_release_sock(sk);
1963 1965
1964 out_nounlock: 1966 out_nounlock:
1965 return sctp_error(sk, msg_flags, err); 1967 return sctp_error(sk, msg_flags, err);
1966 1968
1967 #if 0 1969 #if 0
1968 do_sock_err: 1970 do_sock_err:
1969 if (msg_len) 1971 if (msg_len)
1970 err = msg_len; 1972 err = msg_len;
1971 else 1973 else
1972 err = sock_error(sk); 1974 err = sock_error(sk);
1973 goto out; 1975 goto out;
1974 1976
1975 do_interrupted: 1977 do_interrupted:
1976 if (msg_len) 1978 if (msg_len)
1977 err = msg_len; 1979 err = msg_len;
1978 goto out; 1980 goto out;
1979 #endif /* 0 */ 1981 #endif /* 0 */
1980 } 1982 }
1981 1983
1982 /* This is an extended version of skb_pull() that removes the data from the 1984 /* This is an extended version of skb_pull() that removes the data from the
1983 * start of a skb even when data is spread across the list of skb's in the 1985 * start of a skb even when data is spread across the list of skb's in the
1984 * frag_list. len specifies the total amount of data that needs to be removed. 1986 * frag_list. len specifies the total amount of data that needs to be removed.
1985 * when 'len' bytes could be removed from the skb, it returns 0. 1987 * when 'len' bytes could be removed from the skb, it returns 0.
1986 * If 'len' exceeds the total skb length, it returns the no. of bytes that 1988 * If 'len' exceeds the total skb length, it returns the no. of bytes that
1987 * could not be removed. 1989 * could not be removed.
1988 */ 1990 */
1989 static int sctp_skb_pull(struct sk_buff *skb, int len) 1991 static int sctp_skb_pull(struct sk_buff *skb, int len)
1990 { 1992 {
1991 struct sk_buff *list; 1993 struct sk_buff *list;
1992 int skb_len = skb_headlen(skb); 1994 int skb_len = skb_headlen(skb);
1993 int rlen; 1995 int rlen;
1994 1996
1995 if (len <= skb_len) { 1997 if (len <= skb_len) {
1996 __skb_pull(skb, len); 1998 __skb_pull(skb, len);
1997 return 0; 1999 return 0;
1998 } 2000 }
1999 len -= skb_len; 2001 len -= skb_len;
2000 __skb_pull(skb, skb_len); 2002 __skb_pull(skb, skb_len);
2001 2003
2002 skb_walk_frags(skb, list) { 2004 skb_walk_frags(skb, list) {
2003 rlen = sctp_skb_pull(list, len); 2005 rlen = sctp_skb_pull(list, len);
2004 skb->len -= (len-rlen); 2006 skb->len -= (len-rlen);
2005 skb->data_len -= (len-rlen); 2007 skb->data_len -= (len-rlen);
2006 2008
2007 if (!rlen) 2009 if (!rlen)
2008 return 0; 2010 return 0;
2009 2011
2010 len = rlen; 2012 len = rlen;
2011 } 2013 }
2012 2014
2013 return len; 2015 return len;
2014 } 2016 }
2015 2017
2016 /* API 3.1.3 recvmsg() - UDP Style Syntax 2018 /* API 3.1.3 recvmsg() - UDP Style Syntax
2017 * 2019 *
2018 * ssize_t recvmsg(int socket, struct msghdr *message, 2020 * ssize_t recvmsg(int socket, struct msghdr *message,
2019 * int flags); 2021 * int flags);
2020 * 2022 *
2021 * socket - the socket descriptor of the endpoint. 2023 * socket - the socket descriptor of the endpoint.
2022 * message - pointer to the msghdr structure which contains a single 2024 * message - pointer to the msghdr structure which contains a single
2023 * user message and possibly some ancillary data. 2025 * user message and possibly some ancillary data.
2024 * 2026 *
2025 * See Section 5 for complete description of the data 2027 * See Section 5 for complete description of the data
2026 * structures. 2028 * structures.
2027 * 2029 *
2028 * flags - flags sent or received with the user message, see Section 2030 * flags - flags sent or received with the user message, see Section
2029 * 5 for complete description of the flags. 2031 * 5 for complete description of the flags.
2030 */ 2032 */
2031 static struct sk_buff *sctp_skb_recv_datagram(struct sock *, int, int, int *); 2033 static struct sk_buff *sctp_skb_recv_datagram(struct sock *, int, int, int *);
2032 2034
2033 SCTP_STATIC int sctp_recvmsg(struct kiocb *iocb, struct sock *sk, 2035 SCTP_STATIC int sctp_recvmsg(struct kiocb *iocb, struct sock *sk,
2034 struct msghdr *msg, size_t len, int noblock, 2036 struct msghdr *msg, size_t len, int noblock,
2035 int flags, int *addr_len) 2037 int flags, int *addr_len)
2036 { 2038 {
2037 struct sctp_ulpevent *event = NULL; 2039 struct sctp_ulpevent *event = NULL;
2038 struct sctp_sock *sp = sctp_sk(sk); 2040 struct sctp_sock *sp = sctp_sk(sk);
2039 struct sk_buff *skb; 2041 struct sk_buff *skb;
2040 int copied; 2042 int copied;
2041 int err = 0; 2043 int err = 0;
2042 int skb_len; 2044 int skb_len;
2043 2045
2044 SCTP_DEBUG_PRINTK("sctp_recvmsg(%s: %p, %s: %p, %s: %zd, %s: %d, %s: " 2046 SCTP_DEBUG_PRINTK("sctp_recvmsg(%s: %p, %s: %p, %s: %zd, %s: %d, %s: "
2045 "0x%x, %s: %p)\n", "sk", sk, "msghdr", msg, 2047 "0x%x, %s: %p)\n", "sk", sk, "msghdr", msg,
2046 "len", len, "knoblauch", noblock, 2048 "len", len, "knoblauch", noblock,
2047 "flags", flags, "addr_len", addr_len); 2049 "flags", flags, "addr_len", addr_len);
2048 2050
2049 sctp_lock_sock(sk); 2051 sctp_lock_sock(sk);
2050 2052
2051 if (sctp_style(sk, TCP) && !sctp_sstate(sk, ESTABLISHED)) { 2053 if (sctp_style(sk, TCP) && !sctp_sstate(sk, ESTABLISHED)) {
2052 err = -ENOTCONN; 2054 err = -ENOTCONN;
2053 goto out; 2055 goto out;
2054 } 2056 }
2055 2057
2056 skb = sctp_skb_recv_datagram(sk, flags, noblock, &err); 2058 skb = sctp_skb_recv_datagram(sk, flags, noblock, &err);
2057 if (!skb) 2059 if (!skb)
2058 goto out; 2060 goto out;
2059 2061
2060 /* Get the total length of the skb including any skb's in the 2062 /* Get the total length of the skb including any skb's in the
2061 * frag_list. 2063 * frag_list.
2062 */ 2064 */
2063 skb_len = skb->len; 2065 skb_len = skb->len;
2064 2066
2065 copied = skb_len; 2067 copied = skb_len;
2066 if (copied > len) 2068 if (copied > len)
2067 copied = len; 2069 copied = len;
2068 2070
2069 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied); 2071 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
2070 2072
2071 event = sctp_skb2event(skb); 2073 event = sctp_skb2event(skb);
2072 2074
2073 if (err) 2075 if (err)
2074 goto out_free; 2076 goto out_free;
2075 2077
2076 sock_recv_ts_and_drops(msg, sk, skb); 2078 sock_recv_ts_and_drops(msg, sk, skb);
2077 if (sctp_ulpevent_is_notification(event)) { 2079 if (sctp_ulpevent_is_notification(event)) {
2078 msg->msg_flags |= MSG_NOTIFICATION; 2080 msg->msg_flags |= MSG_NOTIFICATION;
2079 sp->pf->event_msgname(event, msg->msg_name, addr_len); 2081 sp->pf->event_msgname(event, msg->msg_name, addr_len);
2080 } else { 2082 } else {
2081 sp->pf->skb_msgname(skb, msg->msg_name, addr_len); 2083 sp->pf->skb_msgname(skb, msg->msg_name, addr_len);
2082 } 2084 }
2083 2085
2084 /* Check if we allow SCTP_SNDRCVINFO. */ 2086 /* Check if we allow SCTP_SNDRCVINFO. */
2085 if (sp->subscribe.sctp_data_io_event) 2087 if (sp->subscribe.sctp_data_io_event)
2086 sctp_ulpevent_read_sndrcvinfo(event, msg); 2088 sctp_ulpevent_read_sndrcvinfo(event, msg);
2087 #if 0 2089 #if 0
2088 /* FIXME: we should be calling IP/IPv6 layers. */ 2090 /* FIXME: we should be calling IP/IPv6 layers. */
2089 if (sk->sk_protinfo.af_inet.cmsg_flags) 2091 if (sk->sk_protinfo.af_inet.cmsg_flags)
2090 ip_cmsg_recv(msg, skb); 2092 ip_cmsg_recv(msg, skb);
2091 #endif 2093 #endif
2092 2094
2093 err = copied; 2095 err = copied;
2094 2096
2095 /* If skb's length exceeds the user's buffer, update the skb and 2097 /* If skb's length exceeds the user's buffer, update the skb and
2096 * push it back to the receive_queue so that the next call to 2098 * push it back to the receive_queue so that the next call to
2097 * recvmsg() will return the remaining data. Don't set MSG_EOR. 2099 * recvmsg() will return the remaining data. Don't set MSG_EOR.
2098 */ 2100 */
2099 if (skb_len > copied) { 2101 if (skb_len > copied) {
2100 msg->msg_flags &= ~MSG_EOR; 2102 msg->msg_flags &= ~MSG_EOR;
2101 if (flags & MSG_PEEK) 2103 if (flags & MSG_PEEK)
2102 goto out_free; 2104 goto out_free;
2103 sctp_skb_pull(skb, copied); 2105 sctp_skb_pull(skb, copied);
2104 skb_queue_head(&sk->sk_receive_queue, skb); 2106 skb_queue_head(&sk->sk_receive_queue, skb);
2105 2107
2106 /* When only partial message is copied to the user, increase 2108 /* When only partial message is copied to the user, increase
2107 * rwnd by that amount. If all the data in the skb is read, 2109 * rwnd by that amount. If all the data in the skb is read,
2108 * rwnd is updated when the event is freed. 2110 * rwnd is updated when the event is freed.
2109 */ 2111 */
2110 if (!sctp_ulpevent_is_notification(event)) 2112 if (!sctp_ulpevent_is_notification(event))
2111 sctp_assoc_rwnd_increase(event->asoc, copied); 2113 sctp_assoc_rwnd_increase(event->asoc, copied);
2112 goto out; 2114 goto out;
2113 } else if ((event->msg_flags & MSG_NOTIFICATION) || 2115 } else if ((event->msg_flags & MSG_NOTIFICATION) ||
2114 (event->msg_flags & MSG_EOR)) 2116 (event->msg_flags & MSG_EOR))
2115 msg->msg_flags |= MSG_EOR; 2117 msg->msg_flags |= MSG_EOR;
2116 else 2118 else
2117 msg->msg_flags &= ~MSG_EOR; 2119 msg->msg_flags &= ~MSG_EOR;
2118 2120
2119 out_free: 2121 out_free:
2120 if (flags & MSG_PEEK) { 2122 if (flags & MSG_PEEK) {
2121 /* Release the skb reference acquired after peeking the skb in 2123 /* Release the skb reference acquired after peeking the skb in
2122 * sctp_skb_recv_datagram(). 2124 * sctp_skb_recv_datagram().
2123 */ 2125 */
2124 kfree_skb(skb); 2126 kfree_skb(skb);
2125 } else { 2127 } else {
2126 /* Free the event which includes releasing the reference to 2128 /* Free the event which includes releasing the reference to
2127 * the owner of the skb, freeing the skb and updating the 2129 * the owner of the skb, freeing the skb and updating the
2128 * rwnd. 2130 * rwnd.
2129 */ 2131 */
2130 sctp_ulpevent_free(event); 2132 sctp_ulpevent_free(event);
2131 } 2133 }
2132 out: 2134 out:
2133 sctp_release_sock(sk); 2135 sctp_release_sock(sk);
2134 return err; 2136 return err;
2135 } 2137 }
2136 2138
2137 /* 7.1.12 Enable/Disable message fragmentation (SCTP_DISABLE_FRAGMENTS) 2139 /* 7.1.12 Enable/Disable message fragmentation (SCTP_DISABLE_FRAGMENTS)
2138 * 2140 *
2139 * This option is a on/off flag. If enabled no SCTP message 2141 * This option is a on/off flag. If enabled no SCTP message
2140 * fragmentation will be performed. Instead if a message being sent 2142 * fragmentation will be performed. Instead if a message being sent
2141 * exceeds the current PMTU size, the message will NOT be sent and 2143 * exceeds the current PMTU size, the message will NOT be sent and
2142 * instead a error will be indicated to the user. 2144 * instead a error will be indicated to the user.
2143 */ 2145 */
2144 static int sctp_setsockopt_disable_fragments(struct sock *sk, 2146 static int sctp_setsockopt_disable_fragments(struct sock *sk,
2145 char __user *optval, 2147 char __user *optval,
2146 unsigned int optlen) 2148 unsigned int optlen)
2147 { 2149 {
2148 int val; 2150 int val;
2149 2151
2150 if (optlen < sizeof(int)) 2152 if (optlen < sizeof(int))
2151 return -EINVAL; 2153 return -EINVAL;
2152 2154
2153 if (get_user(val, (int __user *)optval)) 2155 if (get_user(val, (int __user *)optval))
2154 return -EFAULT; 2156 return -EFAULT;
2155 2157
2156 sctp_sk(sk)->disable_fragments = (val == 0) ? 0 : 1; 2158 sctp_sk(sk)->disable_fragments = (val == 0) ? 0 : 1;
2157 2159
2158 return 0; 2160 return 0;
2159 } 2161 }
2160 2162
2161 static int sctp_setsockopt_events(struct sock *sk, char __user *optval, 2163 static int sctp_setsockopt_events(struct sock *sk, char __user *optval,
2162 unsigned int optlen) 2164 unsigned int optlen)
2163 { 2165 {
2164 struct sctp_association *asoc; 2166 struct sctp_association *asoc;
2165 struct sctp_ulpevent *event; 2167 struct sctp_ulpevent *event;
2166 2168
2167 if (optlen > sizeof(struct sctp_event_subscribe)) 2169 if (optlen > sizeof(struct sctp_event_subscribe))
2168 return -EINVAL; 2170 return -EINVAL;
2169 if (copy_from_user(&sctp_sk(sk)->subscribe, optval, optlen)) 2171 if (copy_from_user(&sctp_sk(sk)->subscribe, optval, optlen))
2170 return -EFAULT; 2172 return -EFAULT;
2171 2173
2172 /* 2174 /*
2173 * At the time when a user app subscribes to SCTP_SENDER_DRY_EVENT, 2175 * At the time when a user app subscribes to SCTP_SENDER_DRY_EVENT,
2174 * if there is no data to be sent or retransmit, the stack will 2176 * if there is no data to be sent or retransmit, the stack will
2175 * immediately send up this notification. 2177 * immediately send up this notification.
2176 */ 2178 */
2177 if (sctp_ulpevent_type_enabled(SCTP_SENDER_DRY_EVENT, 2179 if (sctp_ulpevent_type_enabled(SCTP_SENDER_DRY_EVENT,
2178 &sctp_sk(sk)->subscribe)) { 2180 &sctp_sk(sk)->subscribe)) {
2179 asoc = sctp_id2assoc(sk, 0); 2181 asoc = sctp_id2assoc(sk, 0);
2180 2182
2181 if (asoc && sctp_outq_is_empty(&asoc->outqueue)) { 2183 if (asoc && sctp_outq_is_empty(&asoc->outqueue)) {
2182 event = sctp_ulpevent_make_sender_dry_event(asoc, 2184 event = sctp_ulpevent_make_sender_dry_event(asoc,
2183 GFP_ATOMIC); 2185 GFP_ATOMIC);
2184 if (!event) 2186 if (!event)
2185 return -ENOMEM; 2187 return -ENOMEM;
2186 2188
2187 sctp_ulpq_tail_event(&asoc->ulpq, event); 2189 sctp_ulpq_tail_event(&asoc->ulpq, event);
2188 } 2190 }
2189 } 2191 }
2190 2192
2191 return 0; 2193 return 0;
2192 } 2194 }
2193 2195
2194 /* 7.1.8 Automatic Close of associations (SCTP_AUTOCLOSE) 2196 /* 7.1.8 Automatic Close of associations (SCTP_AUTOCLOSE)
2195 * 2197 *
2196 * This socket option is applicable to the UDP-style socket only. When 2198 * This socket option is applicable to the UDP-style socket only. When
2197 * set it will cause associations that are idle for more than the 2199 * set it will cause associations that are idle for more than the
2198 * specified number of seconds to automatically close. An association 2200 * specified number of seconds to automatically close. An association
2199 * being idle is defined an association that has NOT sent or received 2201 * being idle is defined an association that has NOT sent or received
2200 * user data. The special value of '0' indicates that no automatic 2202 * user data. The special value of '0' indicates that no automatic
2201 * close of any associations should be performed. The option expects an 2203 * close of any associations should be performed. The option expects an
2202 * integer defining the number of seconds of idle time before an 2204 * integer defining the number of seconds of idle time before an
2203 * association is closed. 2205 * association is closed.
2204 */ 2206 */
2205 static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval, 2207 static int sctp_setsockopt_autoclose(struct sock *sk, char __user *optval,
2206 unsigned int optlen) 2208 unsigned int optlen)
2207 { 2209 {
2208 struct sctp_sock *sp = sctp_sk(sk); 2210 struct sctp_sock *sp = sctp_sk(sk);
2209 2211
2210 /* Applicable to UDP-style socket only */ 2212 /* Applicable to UDP-style socket only */
2211 if (sctp_style(sk, TCP)) 2213 if (sctp_style(sk, TCP))
2212 return -EOPNOTSUPP; 2214 return -EOPNOTSUPP;
2213 if (optlen != sizeof(int)) 2215 if (optlen != sizeof(int))
2214 return -EINVAL; 2216 return -EINVAL;
2215 if (copy_from_user(&sp->autoclose, optval, optlen)) 2217 if (copy_from_user(&sp->autoclose, optval, optlen))
2216 return -EFAULT; 2218 return -EFAULT;
2217 2219
2218 return 0; 2220 return 0;
2219 } 2221 }
2220 2222
2221 /* 7.1.13 Peer Address Parameters (SCTP_PEER_ADDR_PARAMS) 2223 /* 7.1.13 Peer Address Parameters (SCTP_PEER_ADDR_PARAMS)
2222 * 2224 *
2223 * Applications can enable or disable heartbeats for any peer address of 2225 * Applications can enable or disable heartbeats for any peer address of
2224 * an association, modify an address's heartbeat interval, force a 2226 * an association, modify an address's heartbeat interval, force a
2225 * heartbeat to be sent immediately, and adjust the address's maximum 2227 * heartbeat to be sent immediately, and adjust the address's maximum
2226 * number of retransmissions sent before an address is considered 2228 * number of retransmissions sent before an address is considered
2227 * unreachable. The following structure is used to access and modify an 2229 * unreachable. The following structure is used to access and modify an
2228 * address's parameters: 2230 * address's parameters:
2229 * 2231 *
2230 * struct sctp_paddrparams { 2232 * struct sctp_paddrparams {
2231 * sctp_assoc_t spp_assoc_id; 2233 * sctp_assoc_t spp_assoc_id;
2232 * struct sockaddr_storage spp_address; 2234 * struct sockaddr_storage spp_address;
2233 * uint32_t spp_hbinterval; 2235 * uint32_t spp_hbinterval;
2234 * uint16_t spp_pathmaxrxt; 2236 * uint16_t spp_pathmaxrxt;
2235 * uint32_t spp_pathmtu; 2237 * uint32_t spp_pathmtu;
2236 * uint32_t spp_sackdelay; 2238 * uint32_t spp_sackdelay;
2237 * uint32_t spp_flags; 2239 * uint32_t spp_flags;
2238 * }; 2240 * };
2239 * 2241 *
2240 * spp_assoc_id - (one-to-many style socket) This is filled in the 2242 * spp_assoc_id - (one-to-many style socket) This is filled in the
2241 * application, and identifies the association for 2243 * application, and identifies the association for
2242 * this query. 2244 * this query.
2243 * spp_address - This specifies which address is of interest. 2245 * spp_address - This specifies which address is of interest.
2244 * spp_hbinterval - This contains the value of the heartbeat interval, 2246 * spp_hbinterval - This contains the value of the heartbeat interval,
2245 * in milliseconds. If a value of zero 2247 * in milliseconds. If a value of zero
2246 * is present in this field then no changes are to 2248 * is present in this field then no changes are to
2247 * be made to this parameter. 2249 * be made to this parameter.
2248 * spp_pathmaxrxt - This contains the maximum number of 2250 * spp_pathmaxrxt - This contains the maximum number of
2249 * retransmissions before this address shall be 2251 * retransmissions before this address shall be
2250 * considered unreachable. If a value of zero 2252 * considered unreachable. If a value of zero
2251 * is present in this field then no changes are to 2253 * is present in this field then no changes are to
2252 * be made to this parameter. 2254 * be made to this parameter.
2253 * spp_pathmtu - When Path MTU discovery is disabled the value 2255 * spp_pathmtu - When Path MTU discovery is disabled the value
2254 * specified here will be the "fixed" path mtu. 2256 * specified here will be the "fixed" path mtu.
2255 * Note that if the spp_address field is empty 2257 * Note that if the spp_address field is empty
2256 * then all associations on this address will 2258 * then all associations on this address will
2257 * have this fixed path mtu set upon them. 2259 * have this fixed path mtu set upon them.
2258 * 2260 *
2259 * spp_sackdelay - When delayed sack is enabled, this value specifies 2261 * spp_sackdelay - When delayed sack is enabled, this value specifies
2260 * the number of milliseconds that sacks will be delayed 2262 * the number of milliseconds that sacks will be delayed
2261 * for. This value will apply to all addresses of an 2263 * for. This value will apply to all addresses of an
2262 * association if the spp_address field is empty. Note 2264 * association if the spp_address field is empty. Note
2263 * also, that if delayed sack is enabled and this 2265 * also, that if delayed sack is enabled and this
2264 * value is set to 0, no change is made to the last 2266 * value is set to 0, no change is made to the last
2265 * recorded delayed sack timer value. 2267 * recorded delayed sack timer value.
2266 * 2268 *
2267 * spp_flags - These flags are used to control various features 2269 * spp_flags - These flags are used to control various features
2268 * on an association. The flag field may contain 2270 * on an association. The flag field may contain
2269 * zero or more of the following options. 2271 * zero or more of the following options.
2270 * 2272 *
2271 * SPP_HB_ENABLE - Enable heartbeats on the 2273 * SPP_HB_ENABLE - Enable heartbeats on the
2272 * specified address. Note that if the address 2274 * specified address. Note that if the address
2273 * field is empty all addresses for the association 2275 * field is empty all addresses for the association
2274 * have heartbeats enabled upon them. 2276 * have heartbeats enabled upon them.
2275 * 2277 *
2276 * SPP_HB_DISABLE - Disable heartbeats on the 2278 * SPP_HB_DISABLE - Disable heartbeats on the
2277 * speicifed address. Note that if the address 2279 * speicifed address. Note that if the address
2278 * field is empty all addresses for the association 2280 * field is empty all addresses for the association
2279 * will have their heartbeats disabled. Note also 2281 * will have their heartbeats disabled. Note also
2280 * that SPP_HB_ENABLE and SPP_HB_DISABLE are 2282 * that SPP_HB_ENABLE and SPP_HB_DISABLE are
2281 * mutually exclusive, only one of these two should 2283 * mutually exclusive, only one of these two should
2282 * be specified. Enabling both fields will have 2284 * be specified. Enabling both fields will have
2283 * undetermined results. 2285 * undetermined results.
2284 * 2286 *
2285 * SPP_HB_DEMAND - Request a user initiated heartbeat 2287 * SPP_HB_DEMAND - Request a user initiated heartbeat
2286 * to be made immediately. 2288 * to be made immediately.
2287 * 2289 *
2288 * SPP_HB_TIME_IS_ZERO - Specify's that the time for 2290 * SPP_HB_TIME_IS_ZERO - Specify's that the time for
2289 * heartbeat delayis to be set to the value of 0 2291 * heartbeat delayis to be set to the value of 0
2290 * milliseconds. 2292 * milliseconds.
2291 * 2293 *
2292 * SPP_PMTUD_ENABLE - This field will enable PMTU 2294 * SPP_PMTUD_ENABLE - This field will enable PMTU
2293 * discovery upon the specified address. Note that 2295 * discovery upon the specified address. Note that
2294 * if the address feild is empty then all addresses 2296 * if the address feild is empty then all addresses
2295 * on the association are effected. 2297 * on the association are effected.
2296 * 2298 *
2297 * SPP_PMTUD_DISABLE - This field will disable PMTU 2299 * SPP_PMTUD_DISABLE - This field will disable PMTU
2298 * discovery upon the specified address. Note that 2300 * discovery upon the specified address. Note that
2299 * if the address feild is empty then all addresses 2301 * if the address feild is empty then all addresses
2300 * on the association are effected. Not also that 2302 * on the association are effected. Not also that
2301 * SPP_PMTUD_ENABLE and SPP_PMTUD_DISABLE are mutually 2303 * SPP_PMTUD_ENABLE and SPP_PMTUD_DISABLE are mutually
2302 * exclusive. Enabling both will have undetermined 2304 * exclusive. Enabling both will have undetermined
2303 * results. 2305 * results.
2304 * 2306 *
2305 * SPP_SACKDELAY_ENABLE - Setting this flag turns 2307 * SPP_SACKDELAY_ENABLE - Setting this flag turns
2306 * on delayed sack. The time specified in spp_sackdelay 2308 * on delayed sack. The time specified in spp_sackdelay
2307 * is used to specify the sack delay for this address. Note 2309 * is used to specify the sack delay for this address. Note
2308 * that if spp_address is empty then all addresses will 2310 * that if spp_address is empty then all addresses will
2309 * enable delayed sack and take on the sack delay 2311 * enable delayed sack and take on the sack delay
2310 * value specified in spp_sackdelay. 2312 * value specified in spp_sackdelay.
2311 * SPP_SACKDELAY_DISABLE - Setting this flag turns 2313 * SPP_SACKDELAY_DISABLE - Setting this flag turns
2312 * off delayed sack. If the spp_address field is blank then 2314 * off delayed sack. If the spp_address field is blank then
2313 * delayed sack is disabled for the entire association. Note 2315 * delayed sack is disabled for the entire association. Note
2314 * also that this field is mutually exclusive to 2316 * also that this field is mutually exclusive to
2315 * SPP_SACKDELAY_ENABLE, setting both will have undefined 2317 * SPP_SACKDELAY_ENABLE, setting both will have undefined
2316 * results. 2318 * results.
2317 */ 2319 */
2318 static int sctp_apply_peer_addr_params(struct sctp_paddrparams *params, 2320 static int sctp_apply_peer_addr_params(struct sctp_paddrparams *params,
2319 struct sctp_transport *trans, 2321 struct sctp_transport *trans,
2320 struct sctp_association *asoc, 2322 struct sctp_association *asoc,
2321 struct sctp_sock *sp, 2323 struct sctp_sock *sp,
2322 int hb_change, 2324 int hb_change,
2323 int pmtud_change, 2325 int pmtud_change,
2324 int sackdelay_change) 2326 int sackdelay_change)
2325 { 2327 {
2326 int error; 2328 int error;
2327 2329
2328 if (params->spp_flags & SPP_HB_DEMAND && trans) { 2330 if (params->spp_flags & SPP_HB_DEMAND && trans) {
2329 struct net *net = sock_net(trans->asoc->base.sk); 2331 struct net *net = sock_net(trans->asoc->base.sk);
2330 2332
2331 error = sctp_primitive_REQUESTHEARTBEAT(net, trans->asoc, trans); 2333 error = sctp_primitive_REQUESTHEARTBEAT(net, trans->asoc, trans);
2332 if (error) 2334 if (error)
2333 return error; 2335 return error;
2334 } 2336 }
2335 2337
2336 /* Note that unless the spp_flag is set to SPP_HB_ENABLE the value of 2338 /* Note that unless the spp_flag is set to SPP_HB_ENABLE the value of
2337 * this field is ignored. Note also that a value of zero indicates 2339 * this field is ignored. Note also that a value of zero indicates
2338 * the current setting should be left unchanged. 2340 * the current setting should be left unchanged.
2339 */ 2341 */
2340 if (params->spp_flags & SPP_HB_ENABLE) { 2342 if (params->spp_flags & SPP_HB_ENABLE) {
2341 2343
2342 /* Re-zero the interval if the SPP_HB_TIME_IS_ZERO is 2344 /* Re-zero the interval if the SPP_HB_TIME_IS_ZERO is
2343 * set. This lets us use 0 value when this flag 2345 * set. This lets us use 0 value when this flag
2344 * is set. 2346 * is set.
2345 */ 2347 */
2346 if (params->spp_flags & SPP_HB_TIME_IS_ZERO) 2348 if (params->spp_flags & SPP_HB_TIME_IS_ZERO)
2347 params->spp_hbinterval = 0; 2349 params->spp_hbinterval = 0;
2348 2350
2349 if (params->spp_hbinterval || 2351 if (params->spp_hbinterval ||
2350 (params->spp_flags & SPP_HB_TIME_IS_ZERO)) { 2352 (params->spp_flags & SPP_HB_TIME_IS_ZERO)) {
2351 if (trans) { 2353 if (trans) {
2352 trans->hbinterval = 2354 trans->hbinterval =
2353 msecs_to_jiffies(params->spp_hbinterval); 2355 msecs_to_jiffies(params->spp_hbinterval);
2354 } else if (asoc) { 2356 } else if (asoc) {
2355 asoc->hbinterval = 2357 asoc->hbinterval =
2356 msecs_to_jiffies(params->spp_hbinterval); 2358 msecs_to_jiffies(params->spp_hbinterval);
2357 } else { 2359 } else {
2358 sp->hbinterval = params->spp_hbinterval; 2360 sp->hbinterval = params->spp_hbinterval;
2359 } 2361 }
2360 } 2362 }
2361 } 2363 }
2362 2364
2363 if (hb_change) { 2365 if (hb_change) {
2364 if (trans) { 2366 if (trans) {
2365 trans->param_flags = 2367 trans->param_flags =
2366 (trans->param_flags & ~SPP_HB) | hb_change; 2368 (trans->param_flags & ~SPP_HB) | hb_change;
2367 } else if (asoc) { 2369 } else if (asoc) {
2368 asoc->param_flags = 2370 asoc->param_flags =
2369 (asoc->param_flags & ~SPP_HB) | hb_change; 2371 (asoc->param_flags & ~SPP_HB) | hb_change;
2370 } else { 2372 } else {
2371 sp->param_flags = 2373 sp->param_flags =
2372 (sp->param_flags & ~SPP_HB) | hb_change; 2374 (sp->param_flags & ~SPP_HB) | hb_change;
2373 } 2375 }
2374 } 2376 }
2375 2377
2376 /* When Path MTU discovery is disabled the value specified here will 2378 /* When Path MTU discovery is disabled the value specified here will
2377 * be the "fixed" path mtu (i.e. the value of the spp_flags field must 2379 * be the "fixed" path mtu (i.e. the value of the spp_flags field must
2378 * include the flag SPP_PMTUD_DISABLE for this field to have any 2380 * include the flag SPP_PMTUD_DISABLE for this field to have any
2379 * effect). 2381 * effect).
2380 */ 2382 */
2381 if ((params->spp_flags & SPP_PMTUD_DISABLE) && params->spp_pathmtu) { 2383 if ((params->spp_flags & SPP_PMTUD_DISABLE) && params->spp_pathmtu) {
2382 if (trans) { 2384 if (trans) {
2383 trans->pathmtu = params->spp_pathmtu; 2385 trans->pathmtu = params->spp_pathmtu;
2384 sctp_assoc_sync_pmtu(sctp_opt2sk(sp), asoc); 2386 sctp_assoc_sync_pmtu(sctp_opt2sk(sp), asoc);
2385 } else if (asoc) { 2387 } else if (asoc) {
2386 asoc->pathmtu = params->spp_pathmtu; 2388 asoc->pathmtu = params->spp_pathmtu;
2387 sctp_frag_point(asoc, params->spp_pathmtu); 2389 sctp_frag_point(asoc, params->spp_pathmtu);
2388 } else { 2390 } else {
2389 sp->pathmtu = params->spp_pathmtu; 2391 sp->pathmtu = params->spp_pathmtu;
2390 } 2392 }
2391 } 2393 }
2392 2394
2393 if (pmtud_change) { 2395 if (pmtud_change) {
2394 if (trans) { 2396 if (trans) {
2395 int update = (trans->param_flags & SPP_PMTUD_DISABLE) && 2397 int update = (trans->param_flags & SPP_PMTUD_DISABLE) &&
2396 (params->spp_flags & SPP_PMTUD_ENABLE); 2398 (params->spp_flags & SPP_PMTUD_ENABLE);
2397 trans->param_flags = 2399 trans->param_flags =
2398 (trans->param_flags & ~SPP_PMTUD) | pmtud_change; 2400 (trans->param_flags & ~SPP_PMTUD) | pmtud_change;
2399 if (update) { 2401 if (update) {
2400 sctp_transport_pmtu(trans, sctp_opt2sk(sp)); 2402 sctp_transport_pmtu(trans, sctp_opt2sk(sp));
2401 sctp_assoc_sync_pmtu(sctp_opt2sk(sp), asoc); 2403 sctp_assoc_sync_pmtu(sctp_opt2sk(sp), asoc);
2402 } 2404 }
2403 } else if (asoc) { 2405 } else if (asoc) {
2404 asoc->param_flags = 2406 asoc->param_flags =
2405 (asoc->param_flags & ~SPP_PMTUD) | pmtud_change; 2407 (asoc->param_flags & ~SPP_PMTUD) | pmtud_change;
2406 } else { 2408 } else {
2407 sp->param_flags = 2409 sp->param_flags =
2408 (sp->param_flags & ~SPP_PMTUD) | pmtud_change; 2410 (sp->param_flags & ~SPP_PMTUD) | pmtud_change;
2409 } 2411 }
2410 } 2412 }
2411 2413
2412 /* Note that unless the spp_flag is set to SPP_SACKDELAY_ENABLE the 2414 /* Note that unless the spp_flag is set to SPP_SACKDELAY_ENABLE the
2413 * value of this field is ignored. Note also that a value of zero 2415 * value of this field is ignored. Note also that a value of zero
2414 * indicates the current setting should be left unchanged. 2416 * indicates the current setting should be left unchanged.
2415 */ 2417 */
2416 if ((params->spp_flags & SPP_SACKDELAY_ENABLE) && params->spp_sackdelay) { 2418 if ((params->spp_flags & SPP_SACKDELAY_ENABLE) && params->spp_sackdelay) {
2417 if (trans) { 2419 if (trans) {
2418 trans->sackdelay = 2420 trans->sackdelay =
2419 msecs_to_jiffies(params->spp_sackdelay); 2421 msecs_to_jiffies(params->spp_sackdelay);
2420 } else if (asoc) { 2422 } else if (asoc) {
2421 asoc->sackdelay = 2423 asoc->sackdelay =
2422 msecs_to_jiffies(params->spp_sackdelay); 2424 msecs_to_jiffies(params->spp_sackdelay);
2423 } else { 2425 } else {
2424 sp->sackdelay = params->spp_sackdelay; 2426 sp->sackdelay = params->spp_sackdelay;
2425 } 2427 }
2426 } 2428 }
2427 2429
2428 if (sackdelay_change) { 2430 if (sackdelay_change) {
2429 if (trans) { 2431 if (trans) {
2430 trans->param_flags = 2432 trans->param_flags =
2431 (trans->param_flags & ~SPP_SACKDELAY) | 2433 (trans->param_flags & ~SPP_SACKDELAY) |
2432 sackdelay_change; 2434 sackdelay_change;
2433 } else if (asoc) { 2435 } else if (asoc) {
2434 asoc->param_flags = 2436 asoc->param_flags =
2435 (asoc->param_flags & ~SPP_SACKDELAY) | 2437 (asoc->param_flags & ~SPP_SACKDELAY) |
2436 sackdelay_change; 2438 sackdelay_change;
2437 } else { 2439 } else {
2438 sp->param_flags = 2440 sp->param_flags =
2439 (sp->param_flags & ~SPP_SACKDELAY) | 2441 (sp->param_flags & ~SPP_SACKDELAY) |
2440 sackdelay_change; 2442 sackdelay_change;
2441 } 2443 }
2442 } 2444 }
2443 2445
2444 /* Note that a value of zero indicates the current setting should be 2446 /* Note that a value of zero indicates the current setting should be
2445 left unchanged. 2447 left unchanged.
2446 */ 2448 */
2447 if (params->spp_pathmaxrxt) { 2449 if (params->spp_pathmaxrxt) {
2448 if (trans) { 2450 if (trans) {
2449 trans->pathmaxrxt = params->spp_pathmaxrxt; 2451 trans->pathmaxrxt = params->spp_pathmaxrxt;
2450 } else if (asoc) { 2452 } else if (asoc) {
2451 asoc->pathmaxrxt = params->spp_pathmaxrxt; 2453 asoc->pathmaxrxt = params->spp_pathmaxrxt;
2452 } else { 2454 } else {
2453 sp->pathmaxrxt = params->spp_pathmaxrxt; 2455 sp->pathmaxrxt = params->spp_pathmaxrxt;
2454 } 2456 }
2455 } 2457 }
2456 2458
2457 return 0; 2459 return 0;
2458 } 2460 }
2459 2461
2460 static int sctp_setsockopt_peer_addr_params(struct sock *sk, 2462 static int sctp_setsockopt_peer_addr_params(struct sock *sk,
2461 char __user *optval, 2463 char __user *optval,
2462 unsigned int optlen) 2464 unsigned int optlen)
2463 { 2465 {
2464 struct sctp_paddrparams params; 2466 struct sctp_paddrparams params;
2465 struct sctp_transport *trans = NULL; 2467 struct sctp_transport *trans = NULL;
2466 struct sctp_association *asoc = NULL; 2468 struct sctp_association *asoc = NULL;
2467 struct sctp_sock *sp = sctp_sk(sk); 2469 struct sctp_sock *sp = sctp_sk(sk);
2468 int error; 2470 int error;
2469 int hb_change, pmtud_change, sackdelay_change; 2471 int hb_change, pmtud_change, sackdelay_change;
2470 2472
2471 if (optlen != sizeof(struct sctp_paddrparams)) 2473 if (optlen != sizeof(struct sctp_paddrparams))
2472 return - EINVAL; 2474 return - EINVAL;
2473 2475
2474 if (copy_from_user(&params, optval, optlen)) 2476 if (copy_from_user(&params, optval, optlen))
2475 return -EFAULT; 2477 return -EFAULT;
2476 2478
2477 /* Validate flags and value parameters. */ 2479 /* Validate flags and value parameters. */
2478 hb_change = params.spp_flags & SPP_HB; 2480 hb_change = params.spp_flags & SPP_HB;
2479 pmtud_change = params.spp_flags & SPP_PMTUD; 2481 pmtud_change = params.spp_flags & SPP_PMTUD;
2480 sackdelay_change = params.spp_flags & SPP_SACKDELAY; 2482 sackdelay_change = params.spp_flags & SPP_SACKDELAY;
2481 2483
2482 if (hb_change == SPP_HB || 2484 if (hb_change == SPP_HB ||
2483 pmtud_change == SPP_PMTUD || 2485 pmtud_change == SPP_PMTUD ||
2484 sackdelay_change == SPP_SACKDELAY || 2486 sackdelay_change == SPP_SACKDELAY ||
2485 params.spp_sackdelay > 500 || 2487 params.spp_sackdelay > 500 ||
2486 (params.spp_pathmtu && 2488 (params.spp_pathmtu &&
2487 params.spp_pathmtu < SCTP_DEFAULT_MINSEGMENT)) 2489 params.spp_pathmtu < SCTP_DEFAULT_MINSEGMENT))
2488 return -EINVAL; 2490 return -EINVAL;
2489 2491
2490 /* If an address other than INADDR_ANY is specified, and 2492 /* If an address other than INADDR_ANY is specified, and
2491 * no transport is found, then the request is invalid. 2493 * no transport is found, then the request is invalid.
2492 */ 2494 */
2493 if (!sctp_is_any(sk, ( union sctp_addr *)&params.spp_address)) { 2495 if (!sctp_is_any(sk, ( union sctp_addr *)&params.spp_address)) {
2494 trans = sctp_addr_id2transport(sk, &params.spp_address, 2496 trans = sctp_addr_id2transport(sk, &params.spp_address,
2495 params.spp_assoc_id); 2497 params.spp_assoc_id);
2496 if (!trans) 2498 if (!trans)
2497 return -EINVAL; 2499 return -EINVAL;
2498 } 2500 }
2499 2501
2500 /* Get association, if assoc_id != 0 and the socket is a one 2502 /* Get association, if assoc_id != 0 and the socket is a one
2501 * to many style socket, and an association was not found, then 2503 * to many style socket, and an association was not found, then
2502 * the id was invalid. 2504 * the id was invalid.
2503 */ 2505 */
2504 asoc = sctp_id2assoc(sk, params.spp_assoc_id); 2506 asoc = sctp_id2assoc(sk, params.spp_assoc_id);
2505 if (!asoc && params.spp_assoc_id && sctp_style(sk, UDP)) 2507 if (!asoc && params.spp_assoc_id && sctp_style(sk, UDP))
2506 return -EINVAL; 2508 return -EINVAL;
2507 2509
2508 /* Heartbeat demand can only be sent on a transport or 2510 /* Heartbeat demand can only be sent on a transport or
2509 * association, but not a socket. 2511 * association, but not a socket.
2510 */ 2512 */
2511 if (params.spp_flags & SPP_HB_DEMAND && !trans && !asoc) 2513 if (params.spp_flags & SPP_HB_DEMAND && !trans && !asoc)
2512 return -EINVAL; 2514 return -EINVAL;
2513 2515
2514 /* Process parameters. */ 2516 /* Process parameters. */
2515 error = sctp_apply_peer_addr_params(&params, trans, asoc, sp, 2517 error = sctp_apply_peer_addr_params(&params, trans, asoc, sp,
2516 hb_change, pmtud_change, 2518 hb_change, pmtud_change,
2517 sackdelay_change); 2519 sackdelay_change);
2518 2520
2519 if (error) 2521 if (error)
2520 return error; 2522 return error;
2521 2523
2522 /* If changes are for association, also apply parameters to each 2524 /* If changes are for association, also apply parameters to each
2523 * transport. 2525 * transport.
2524 */ 2526 */
2525 if (!trans && asoc) { 2527 if (!trans && asoc) {
2526 list_for_each_entry(trans, &asoc->peer.transport_addr_list, 2528 list_for_each_entry(trans, &asoc->peer.transport_addr_list,
2527 transports) { 2529 transports) {
2528 sctp_apply_peer_addr_params(&params, trans, asoc, sp, 2530 sctp_apply_peer_addr_params(&params, trans, asoc, sp,
2529 hb_change, pmtud_change, 2531 hb_change, pmtud_change,
2530 sackdelay_change); 2532 sackdelay_change);
2531 } 2533 }
2532 } 2534 }
2533 2535
2534 return 0; 2536 return 0;
2535 } 2537 }
2536 2538
2537 /* 2539 /*
2538 * 7.1.23. Get or set delayed ack timer (SCTP_DELAYED_SACK) 2540 * 7.1.23. Get or set delayed ack timer (SCTP_DELAYED_SACK)
2539 * 2541 *
2540 * This option will effect the way delayed acks are performed. This 2542 * This option will effect the way delayed acks are performed. This
2541 * option allows you to get or set the delayed ack time, in 2543 * option allows you to get or set the delayed ack time, in
2542 * milliseconds. It also allows changing the delayed ack frequency. 2544 * milliseconds. It also allows changing the delayed ack frequency.
2543 * Changing the frequency to 1 disables the delayed sack algorithm. If 2545 * Changing the frequency to 1 disables the delayed sack algorithm. If
2544 * the assoc_id is 0, then this sets or gets the endpoints default 2546 * the assoc_id is 0, then this sets or gets the endpoints default
2545 * values. If the assoc_id field is non-zero, then the set or get 2547 * values. If the assoc_id field is non-zero, then the set or get
2546 * effects the specified association for the one to many model (the 2548 * effects the specified association for the one to many model (the
2547 * assoc_id field is ignored by the one to one model). Note that if 2549 * assoc_id field is ignored by the one to one model). Note that if
2548 * sack_delay or sack_freq are 0 when setting this option, then the 2550 * sack_delay or sack_freq are 0 when setting this option, then the
2549 * current values will remain unchanged. 2551 * current values will remain unchanged.
2550 * 2552 *
2551 * struct sctp_sack_info { 2553 * struct sctp_sack_info {
2552 * sctp_assoc_t sack_assoc_id; 2554 * sctp_assoc_t sack_assoc_id;
2553 * uint32_t sack_delay; 2555 * uint32_t sack_delay;
2554 * uint32_t sack_freq; 2556 * uint32_t sack_freq;
2555 * }; 2557 * };
2556 * 2558 *
2557 * sack_assoc_id - This parameter, indicates which association the user 2559 * sack_assoc_id - This parameter, indicates which association the user
2558 * is performing an action upon. Note that if this field's value is 2560 * is performing an action upon. Note that if this field's value is
2559 * zero then the endpoints default value is changed (effecting future 2561 * zero then the endpoints default value is changed (effecting future
2560 * associations only). 2562 * associations only).
2561 * 2563 *
2562 * sack_delay - This parameter contains the number of milliseconds that 2564 * sack_delay - This parameter contains the number of milliseconds that
2563 * the user is requesting the delayed ACK timer be set to. Note that 2565 * the user is requesting the delayed ACK timer be set to. Note that
2564 * this value is defined in the standard to be between 200 and 500 2566 * this value is defined in the standard to be between 200 and 500
2565 * milliseconds. 2567 * milliseconds.
2566 * 2568 *
2567 * sack_freq - This parameter contains the number of packets that must 2569 * sack_freq - This parameter contains the number of packets that must
2568 * be received before a sack is sent without waiting for the delay 2570 * be received before a sack is sent without waiting for the delay
2569 * timer to expire. The default value for this is 2, setting this 2571 * timer to expire. The default value for this is 2, setting this
2570 * value to 1 will disable the delayed sack algorithm. 2572 * value to 1 will disable the delayed sack algorithm.
2571 */ 2573 */
2572 2574
2573 static int sctp_setsockopt_delayed_ack(struct sock *sk, 2575 static int sctp_setsockopt_delayed_ack(struct sock *sk,
2574 char __user *optval, unsigned int optlen) 2576 char __user *optval, unsigned int optlen)
2575 { 2577 {
2576 struct sctp_sack_info params; 2578 struct sctp_sack_info params;
2577 struct sctp_transport *trans = NULL; 2579 struct sctp_transport *trans = NULL;
2578 struct sctp_association *asoc = NULL; 2580 struct sctp_association *asoc = NULL;
2579 struct sctp_sock *sp = sctp_sk(sk); 2581 struct sctp_sock *sp = sctp_sk(sk);
2580 2582
2581 if (optlen == sizeof(struct sctp_sack_info)) { 2583 if (optlen == sizeof(struct sctp_sack_info)) {
2582 if (copy_from_user(&params, optval, optlen)) 2584 if (copy_from_user(&params, optval, optlen))
2583 return -EFAULT; 2585 return -EFAULT;
2584 2586
2585 if (params.sack_delay == 0 && params.sack_freq == 0) 2587 if (params.sack_delay == 0 && params.sack_freq == 0)
2586 return 0; 2588 return 0;
2587 } else if (optlen == sizeof(struct sctp_assoc_value)) { 2589 } else if (optlen == sizeof(struct sctp_assoc_value)) {
2588 pr_warn("Use of struct sctp_assoc_value in delayed_ack socket option deprecated\n"); 2590 pr_warn("Use of struct sctp_assoc_value in delayed_ack socket option deprecated\n");
2589 pr_warn("Use struct sctp_sack_info instead\n"); 2591 pr_warn("Use struct sctp_sack_info instead\n");
2590 if (copy_from_user(&params, optval, optlen)) 2592 if (copy_from_user(&params, optval, optlen))
2591 return -EFAULT; 2593 return -EFAULT;
2592 2594
2593 if (params.sack_delay == 0) 2595 if (params.sack_delay == 0)
2594 params.sack_freq = 1; 2596 params.sack_freq = 1;
2595 else 2597 else
2596 params.sack_freq = 0; 2598 params.sack_freq = 0;
2597 } else 2599 } else
2598 return - EINVAL; 2600 return - EINVAL;
2599 2601
2600 /* Validate value parameter. */ 2602 /* Validate value parameter. */
2601 if (params.sack_delay > 500) 2603 if (params.sack_delay > 500)
2602 return -EINVAL; 2604 return -EINVAL;
2603 2605
2604 /* Get association, if sack_assoc_id != 0 and the socket is a one 2606 /* Get association, if sack_assoc_id != 0 and the socket is a one
2605 * to many style socket, and an association was not found, then 2607 * to many style socket, and an association was not found, then
2606 * the id was invalid. 2608 * the id was invalid.
2607 */ 2609 */
2608 asoc = sctp_id2assoc(sk, params.sack_assoc_id); 2610 asoc = sctp_id2assoc(sk, params.sack_assoc_id);
2609 if (!asoc && params.sack_assoc_id && sctp_style(sk, UDP)) 2611 if (!asoc && params.sack_assoc_id && sctp_style(sk, UDP))
2610 return -EINVAL; 2612 return -EINVAL;
2611 2613
2612 if (params.sack_delay) { 2614 if (params.sack_delay) {
2613 if (asoc) { 2615 if (asoc) {
2614 asoc->sackdelay = 2616 asoc->sackdelay =
2615 msecs_to_jiffies(params.sack_delay); 2617 msecs_to_jiffies(params.sack_delay);
2616 asoc->param_flags = 2618 asoc->param_flags =
2617 (asoc->param_flags & ~SPP_SACKDELAY) | 2619 (asoc->param_flags & ~SPP_SACKDELAY) |
2618 SPP_SACKDELAY_ENABLE; 2620 SPP_SACKDELAY_ENABLE;
2619 } else { 2621 } else {
2620 sp->sackdelay = params.sack_delay; 2622 sp->sackdelay = params.sack_delay;
2621 sp->param_flags = 2623 sp->param_flags =
2622 (sp->param_flags & ~SPP_SACKDELAY) | 2624 (sp->param_flags & ~SPP_SACKDELAY) |
2623 SPP_SACKDELAY_ENABLE; 2625 SPP_SACKDELAY_ENABLE;
2624 } 2626 }
2625 } 2627 }
2626 2628
2627 if (params.sack_freq == 1) { 2629 if (params.sack_freq == 1) {
2628 if (asoc) { 2630 if (asoc) {
2629 asoc->param_flags = 2631 asoc->param_flags =
2630 (asoc->param_flags & ~SPP_SACKDELAY) | 2632 (asoc->param_flags & ~SPP_SACKDELAY) |
2631 SPP_SACKDELAY_DISABLE; 2633 SPP_SACKDELAY_DISABLE;
2632 } else { 2634 } else {
2633 sp->param_flags = 2635 sp->param_flags =
2634 (sp->param_flags & ~SPP_SACKDELAY) | 2636 (sp->param_flags & ~SPP_SACKDELAY) |
2635 SPP_SACKDELAY_DISABLE; 2637 SPP_SACKDELAY_DISABLE;
2636 } 2638 }
2637 } else if (params.sack_freq > 1) { 2639 } else if (params.sack_freq > 1) {
2638 if (asoc) { 2640 if (asoc) {
2639 asoc->sackfreq = params.sack_freq; 2641 asoc->sackfreq = params.sack_freq;
2640 asoc->param_flags = 2642 asoc->param_flags =
2641 (asoc->param_flags & ~SPP_SACKDELAY) | 2643 (asoc->param_flags & ~SPP_SACKDELAY) |
2642 SPP_SACKDELAY_ENABLE; 2644 SPP_SACKDELAY_ENABLE;
2643 } else { 2645 } else {
2644 sp->sackfreq = params.sack_freq; 2646 sp->sackfreq = params.sack_freq;
2645 sp->param_flags = 2647 sp->param_flags =
2646 (sp->param_flags & ~SPP_SACKDELAY) | 2648 (sp->param_flags & ~SPP_SACKDELAY) |
2647 SPP_SACKDELAY_ENABLE; 2649 SPP_SACKDELAY_ENABLE;
2648 } 2650 }
2649 } 2651 }
2650 2652
2651 /* If change is for association, also apply to each transport. */ 2653 /* If change is for association, also apply to each transport. */
2652 if (asoc) { 2654 if (asoc) {
2653 list_for_each_entry(trans, &asoc->peer.transport_addr_list, 2655 list_for_each_entry(trans, &asoc->peer.transport_addr_list,
2654 transports) { 2656 transports) {
2655 if (params.sack_delay) { 2657 if (params.sack_delay) {
2656 trans->sackdelay = 2658 trans->sackdelay =
2657 msecs_to_jiffies(params.sack_delay); 2659 msecs_to_jiffies(params.sack_delay);
2658 trans->param_flags = 2660 trans->param_flags =
2659 (trans->param_flags & ~SPP_SACKDELAY) | 2661 (trans->param_flags & ~SPP_SACKDELAY) |
2660 SPP_SACKDELAY_ENABLE; 2662 SPP_SACKDELAY_ENABLE;
2661 } 2663 }
2662 if (params.sack_freq == 1) { 2664 if (params.sack_freq == 1) {
2663 trans->param_flags = 2665 trans->param_flags =
2664 (trans->param_flags & ~SPP_SACKDELAY) | 2666 (trans->param_flags & ~SPP_SACKDELAY) |
2665 SPP_SACKDELAY_DISABLE; 2667 SPP_SACKDELAY_DISABLE;
2666 } else if (params.sack_freq > 1) { 2668 } else if (params.sack_freq > 1) {
2667 trans->sackfreq = params.sack_freq; 2669 trans->sackfreq = params.sack_freq;
2668 trans->param_flags = 2670 trans->param_flags =
2669 (trans->param_flags & ~SPP_SACKDELAY) | 2671 (trans->param_flags & ~SPP_SACKDELAY) |
2670 SPP_SACKDELAY_ENABLE; 2672 SPP_SACKDELAY_ENABLE;
2671 } 2673 }
2672 } 2674 }
2673 } 2675 }
2674 2676
2675 return 0; 2677 return 0;
2676 } 2678 }
2677 2679
2678 /* 7.1.3 Initialization Parameters (SCTP_INITMSG) 2680 /* 7.1.3 Initialization Parameters (SCTP_INITMSG)
2679 * 2681 *
2680 * Applications can specify protocol parameters for the default association 2682 * Applications can specify protocol parameters for the default association
2681 * initialization. The option name argument to setsockopt() and getsockopt() 2683 * initialization. The option name argument to setsockopt() and getsockopt()
2682 * is SCTP_INITMSG. 2684 * is SCTP_INITMSG.
2683 * 2685 *
2684 * Setting initialization parameters is effective only on an unconnected 2686 * Setting initialization parameters is effective only on an unconnected
2685 * socket (for UDP-style sockets only future associations are effected 2687 * socket (for UDP-style sockets only future associations are effected
2686 * by the change). With TCP-style sockets, this option is inherited by 2688 * by the change). With TCP-style sockets, this option is inherited by
2687 * sockets derived from a listener socket. 2689 * sockets derived from a listener socket.
2688 */ 2690 */
2689 static int sctp_setsockopt_initmsg(struct sock *sk, char __user *optval, unsigned int optlen) 2691 static int sctp_setsockopt_initmsg(struct sock *sk, char __user *optval, unsigned int optlen)
2690 { 2692 {
2691 struct sctp_initmsg sinit; 2693 struct sctp_initmsg sinit;
2692 struct sctp_sock *sp = sctp_sk(sk); 2694 struct sctp_sock *sp = sctp_sk(sk);
2693 2695
2694 if (optlen != sizeof(struct sctp_initmsg)) 2696 if (optlen != sizeof(struct sctp_initmsg))
2695 return -EINVAL; 2697 return -EINVAL;
2696 if (copy_from_user(&sinit, optval, optlen)) 2698 if (copy_from_user(&sinit, optval, optlen))
2697 return -EFAULT; 2699 return -EFAULT;
2698 2700
2699 if (sinit.sinit_num_ostreams) 2701 if (sinit.sinit_num_ostreams)
2700 sp->initmsg.sinit_num_ostreams = sinit.sinit_num_ostreams; 2702 sp->initmsg.sinit_num_ostreams = sinit.sinit_num_ostreams;
2701 if (sinit.sinit_max_instreams) 2703 if (sinit.sinit_max_instreams)
2702 sp->initmsg.sinit_max_instreams = sinit.sinit_max_instreams; 2704 sp->initmsg.sinit_max_instreams = sinit.sinit_max_instreams;
2703 if (sinit.sinit_max_attempts) 2705 if (sinit.sinit_max_attempts)
2704 sp->initmsg.sinit_max_attempts = sinit.sinit_max_attempts; 2706 sp->initmsg.sinit_max_attempts = sinit.sinit_max_attempts;
2705 if (sinit.sinit_max_init_timeo) 2707 if (sinit.sinit_max_init_timeo)
2706 sp->initmsg.sinit_max_init_timeo = sinit.sinit_max_init_timeo; 2708 sp->initmsg.sinit_max_init_timeo = sinit.sinit_max_init_timeo;
2707 2709
2708 return 0; 2710 return 0;
2709 } 2711 }
2710 2712
2711 /* 2713 /*
2712 * 7.1.14 Set default send parameters (SCTP_DEFAULT_SEND_PARAM) 2714 * 7.1.14 Set default send parameters (SCTP_DEFAULT_SEND_PARAM)
2713 * 2715 *
2714 * Applications that wish to use the sendto() system call may wish to 2716 * Applications that wish to use the sendto() system call may wish to
2715 * specify a default set of parameters that would normally be supplied 2717 * specify a default set of parameters that would normally be supplied
2716 * through the inclusion of ancillary data. This socket option allows 2718 * through the inclusion of ancillary data. This socket option allows
2717 * such an application to set the default sctp_sndrcvinfo structure. 2719 * such an application to set the default sctp_sndrcvinfo structure.
2718 * The application that wishes to use this socket option simply passes 2720 * The application that wishes to use this socket option simply passes
2719 * in to this call the sctp_sndrcvinfo structure defined in Section 2721 * in to this call the sctp_sndrcvinfo structure defined in Section
2720 * 5.2.2) The input parameters accepted by this call include 2722 * 5.2.2) The input parameters accepted by this call include
2721 * sinfo_stream, sinfo_flags, sinfo_ppid, sinfo_context, 2723 * sinfo_stream, sinfo_flags, sinfo_ppid, sinfo_context,
2722 * sinfo_timetolive. The user must provide the sinfo_assoc_id field in 2724 * sinfo_timetolive. The user must provide the sinfo_assoc_id field in
2723 * to this call if the caller is using the UDP model. 2725 * to this call if the caller is using the UDP model.
2724 */ 2726 */
2725 static int sctp_setsockopt_default_send_param(struct sock *sk, 2727 static int sctp_setsockopt_default_send_param(struct sock *sk,
2726 char __user *optval, 2728 char __user *optval,
2727 unsigned int optlen) 2729 unsigned int optlen)
2728 { 2730 {
2729 struct sctp_sndrcvinfo info; 2731 struct sctp_sndrcvinfo info;
2730 struct sctp_association *asoc; 2732 struct sctp_association *asoc;
2731 struct sctp_sock *sp = sctp_sk(sk); 2733 struct sctp_sock *sp = sctp_sk(sk);
2732 2734
2733 if (optlen != sizeof(struct sctp_sndrcvinfo)) 2735 if (optlen != sizeof(struct sctp_sndrcvinfo))
2734 return -EINVAL; 2736 return -EINVAL;
2735 if (copy_from_user(&info, optval, optlen)) 2737 if (copy_from_user(&info, optval, optlen))
2736 return -EFAULT; 2738 return -EFAULT;
2737 2739
2738 asoc = sctp_id2assoc(sk, info.sinfo_assoc_id); 2740 asoc = sctp_id2assoc(sk, info.sinfo_assoc_id);
2739 if (!asoc && info.sinfo_assoc_id && sctp_style(sk, UDP)) 2741 if (!asoc && info.sinfo_assoc_id && sctp_style(sk, UDP))
2740 return -EINVAL; 2742 return -EINVAL;
2741 2743
2742 if (asoc) { 2744 if (asoc) {
2743 asoc->default_stream = info.sinfo_stream; 2745 asoc->default_stream = info.sinfo_stream;
2744 asoc->default_flags = info.sinfo_flags; 2746 asoc->default_flags = info.sinfo_flags;
2745 asoc->default_ppid = info.sinfo_ppid; 2747 asoc->default_ppid = info.sinfo_ppid;
2746 asoc->default_context = info.sinfo_context; 2748 asoc->default_context = info.sinfo_context;
2747 asoc->default_timetolive = info.sinfo_timetolive; 2749 asoc->default_timetolive = info.sinfo_timetolive;
2748 } else { 2750 } else {
2749 sp->default_stream = info.sinfo_stream; 2751 sp->default_stream = info.sinfo_stream;
2750 sp->default_flags = info.sinfo_flags; 2752 sp->default_flags = info.sinfo_flags;
2751 sp->default_ppid = info.sinfo_ppid; 2753 sp->default_ppid = info.sinfo_ppid;
2752 sp->default_context = info.sinfo_context; 2754 sp->default_context = info.sinfo_context;
2753 sp->default_timetolive = info.sinfo_timetolive; 2755 sp->default_timetolive = info.sinfo_timetolive;
2754 } 2756 }
2755 2757
2756 return 0; 2758 return 0;
2757 } 2759 }
2758 2760
2759 /* 7.1.10 Set Primary Address (SCTP_PRIMARY_ADDR) 2761 /* 7.1.10 Set Primary Address (SCTP_PRIMARY_ADDR)
2760 * 2762 *
2761 * Requests that the local SCTP stack use the enclosed peer address as 2763 * Requests that the local SCTP stack use the enclosed peer address as
2762 * the association primary. The enclosed address must be one of the 2764 * the association primary. The enclosed address must be one of the
2763 * association peer's addresses. 2765 * association peer's addresses.
2764 */ 2766 */
2765 static int sctp_setsockopt_primary_addr(struct sock *sk, char __user *optval, 2767 static int sctp_setsockopt_primary_addr(struct sock *sk, char __user *optval,
2766 unsigned int optlen) 2768 unsigned int optlen)
2767 { 2769 {
2768 struct sctp_prim prim; 2770 struct sctp_prim prim;
2769 struct sctp_transport *trans; 2771 struct sctp_transport *trans;
2770 2772
2771 if (optlen != sizeof(struct sctp_prim)) 2773 if (optlen != sizeof(struct sctp_prim))
2772 return -EINVAL; 2774 return -EINVAL;
2773 2775
2774 if (copy_from_user(&prim, optval, sizeof(struct sctp_prim))) 2776 if (copy_from_user(&prim, optval, sizeof(struct sctp_prim)))
2775 return -EFAULT; 2777 return -EFAULT;
2776 2778
2777 trans = sctp_addr_id2transport(sk, &prim.ssp_addr, prim.ssp_assoc_id); 2779 trans = sctp_addr_id2transport(sk, &prim.ssp_addr, prim.ssp_assoc_id);
2778 if (!trans) 2780 if (!trans)
2779 return -EINVAL; 2781 return -EINVAL;
2780 2782
2781 sctp_assoc_set_primary(trans->asoc, trans); 2783 sctp_assoc_set_primary(trans->asoc, trans);
2782 2784
2783 return 0; 2785 return 0;
2784 } 2786 }
2785 2787
2786 /* 2788 /*
2787 * 7.1.5 SCTP_NODELAY 2789 * 7.1.5 SCTP_NODELAY
2788 * 2790 *
2789 * Turn on/off any Nagle-like algorithm. This means that packets are 2791 * Turn on/off any Nagle-like algorithm. This means that packets are
2790 * generally sent as soon as possible and no unnecessary delays are 2792 * generally sent as soon as possible and no unnecessary delays are
2791 * introduced, at the cost of more packets in the network. Expects an 2793 * introduced, at the cost of more packets in the network. Expects an
2792 * integer boolean flag. 2794 * integer boolean flag.
2793 */ 2795 */
2794 static int sctp_setsockopt_nodelay(struct sock *sk, char __user *optval, 2796 static int sctp_setsockopt_nodelay(struct sock *sk, char __user *optval,
2795 unsigned int optlen) 2797 unsigned int optlen)
2796 { 2798 {
2797 int val; 2799 int val;
2798 2800
2799 if (optlen < sizeof(int)) 2801 if (optlen < sizeof(int))
2800 return -EINVAL; 2802 return -EINVAL;
2801 if (get_user(val, (int __user *)optval)) 2803 if (get_user(val, (int __user *)optval))
2802 return -EFAULT; 2804 return -EFAULT;
2803 2805
2804 sctp_sk(sk)->nodelay = (val == 0) ? 0 : 1; 2806 sctp_sk(sk)->nodelay = (val == 0) ? 0 : 1;
2805 return 0; 2807 return 0;
2806 } 2808 }
2807 2809
2808 /* 2810 /*
2809 * 2811 *
2810 * 7.1.1 SCTP_RTOINFO 2812 * 7.1.1 SCTP_RTOINFO
2811 * 2813 *
2812 * The protocol parameters used to initialize and bound retransmission 2814 * The protocol parameters used to initialize and bound retransmission
2813 * timeout (RTO) are tunable. sctp_rtoinfo structure is used to access 2815 * timeout (RTO) are tunable. sctp_rtoinfo structure is used to access
2814 * and modify these parameters. 2816 * and modify these parameters.
2815 * All parameters are time values, in milliseconds. A value of 0, when 2817 * All parameters are time values, in milliseconds. A value of 0, when
2816 * modifying the parameters, indicates that the current value should not 2818 * modifying the parameters, indicates that the current value should not
2817 * be changed. 2819 * be changed.
2818 * 2820 *
2819 */ 2821 */
2820 static int sctp_setsockopt_rtoinfo(struct sock *sk, char __user *optval, unsigned int optlen) 2822 static int sctp_setsockopt_rtoinfo(struct sock *sk, char __user *optval, unsigned int optlen)
2821 { 2823 {
2822 struct sctp_rtoinfo rtoinfo; 2824 struct sctp_rtoinfo rtoinfo;
2823 struct sctp_association *asoc; 2825 struct sctp_association *asoc;
2824 2826
2825 if (optlen != sizeof (struct sctp_rtoinfo)) 2827 if (optlen != sizeof (struct sctp_rtoinfo))
2826 return -EINVAL; 2828 return -EINVAL;
2827 2829
2828 if (copy_from_user(&rtoinfo, optval, optlen)) 2830 if (copy_from_user(&rtoinfo, optval, optlen))
2829 return -EFAULT; 2831 return -EFAULT;
2830 2832
2831 asoc = sctp_id2assoc(sk, rtoinfo.srto_assoc_id); 2833 asoc = sctp_id2assoc(sk, rtoinfo.srto_assoc_id);
2832 2834
2833 /* Set the values to the specific association */ 2835 /* Set the values to the specific association */
2834 if (!asoc && rtoinfo.srto_assoc_id && sctp_style(sk, UDP)) 2836 if (!asoc && rtoinfo.srto_assoc_id && sctp_style(sk, UDP))
2835 return -EINVAL; 2837 return -EINVAL;
2836 2838
2837 if (asoc) { 2839 if (asoc) {
2838 if (rtoinfo.srto_initial != 0) 2840 if (rtoinfo.srto_initial != 0)
2839 asoc->rto_initial = 2841 asoc->rto_initial =
2840 msecs_to_jiffies(rtoinfo.srto_initial); 2842 msecs_to_jiffies(rtoinfo.srto_initial);
2841 if (rtoinfo.srto_max != 0) 2843 if (rtoinfo.srto_max != 0)
2842 asoc->rto_max = msecs_to_jiffies(rtoinfo.srto_max); 2844 asoc->rto_max = msecs_to_jiffies(rtoinfo.srto_max);
2843 if (rtoinfo.srto_min != 0) 2845 if (rtoinfo.srto_min != 0)
2844 asoc->rto_min = msecs_to_jiffies(rtoinfo.srto_min); 2846 asoc->rto_min = msecs_to_jiffies(rtoinfo.srto_min);
2845 } else { 2847 } else {
2846 /* If there is no association or the association-id = 0 2848 /* If there is no association or the association-id = 0
2847 * set the values to the endpoint. 2849 * set the values to the endpoint.
2848 */ 2850 */
2849 struct sctp_sock *sp = sctp_sk(sk); 2851 struct sctp_sock *sp = sctp_sk(sk);
2850 2852
2851 if (rtoinfo.srto_initial != 0) 2853 if (rtoinfo.srto_initial != 0)
2852 sp->rtoinfo.srto_initial = rtoinfo.srto_initial; 2854 sp->rtoinfo.srto_initial = rtoinfo.srto_initial;
2853 if (rtoinfo.srto_max != 0) 2855 if (rtoinfo.srto_max != 0)
2854 sp->rtoinfo.srto_max = rtoinfo.srto_max; 2856 sp->rtoinfo.srto_max = rtoinfo.srto_max;
2855 if (rtoinfo.srto_min != 0) 2857 if (rtoinfo.srto_min != 0)
2856 sp->rtoinfo.srto_min = rtoinfo.srto_min; 2858 sp->rtoinfo.srto_min = rtoinfo.srto_min;
2857 } 2859 }
2858 2860
2859 return 0; 2861 return 0;
2860 } 2862 }
2861 2863
2862 /* 2864 /*
2863 * 2865 *
2864 * 7.1.2 SCTP_ASSOCINFO 2866 * 7.1.2 SCTP_ASSOCINFO
2865 * 2867 *
2866 * This option is used to tune the maximum retransmission attempts 2868 * This option is used to tune the maximum retransmission attempts
2867 * of the association. 2869 * of the association.
2868 * Returns an error if the new association retransmission value is 2870 * Returns an error if the new association retransmission value is
2869 * greater than the sum of the retransmission value of the peer. 2871 * greater than the sum of the retransmission value of the peer.
2870 * See [SCTP] for more information. 2872 * See [SCTP] for more information.
2871 * 2873 *
2872 */ 2874 */
2873 static int sctp_setsockopt_associnfo(struct sock *sk, char __user *optval, unsigned int optlen) 2875 static int sctp_setsockopt_associnfo(struct sock *sk, char __user *optval, unsigned int optlen)
2874 { 2876 {
2875 2877
2876 struct sctp_assocparams assocparams; 2878 struct sctp_assocparams assocparams;
2877 struct sctp_association *asoc; 2879 struct sctp_association *asoc;
2878 2880
2879 if (optlen != sizeof(struct sctp_assocparams)) 2881 if (optlen != sizeof(struct sctp_assocparams))
2880 return -EINVAL; 2882 return -EINVAL;
2881 if (copy_from_user(&assocparams, optval, optlen)) 2883 if (copy_from_user(&assocparams, optval, optlen))
2882 return -EFAULT; 2884 return -EFAULT;
2883 2885
2884 asoc = sctp_id2assoc(sk, assocparams.sasoc_assoc_id); 2886 asoc = sctp_id2assoc(sk, assocparams.sasoc_assoc_id);
2885 2887
2886 if (!asoc && assocparams.sasoc_assoc_id && sctp_style(sk, UDP)) 2888 if (!asoc && assocparams.sasoc_assoc_id && sctp_style(sk, UDP))
2887 return -EINVAL; 2889 return -EINVAL;
2888 2890
2889 /* Set the values to the specific association */ 2891 /* Set the values to the specific association */
2890 if (asoc) { 2892 if (asoc) {
2891 if (assocparams.sasoc_asocmaxrxt != 0) { 2893 if (assocparams.sasoc_asocmaxrxt != 0) {
2892 __u32 path_sum = 0; 2894 __u32 path_sum = 0;
2893 int paths = 0; 2895 int paths = 0;
2894 struct sctp_transport *peer_addr; 2896 struct sctp_transport *peer_addr;
2895 2897
2896 list_for_each_entry(peer_addr, &asoc->peer.transport_addr_list, 2898 list_for_each_entry(peer_addr, &asoc->peer.transport_addr_list,
2897 transports) { 2899 transports) {
2898 path_sum += peer_addr->pathmaxrxt; 2900 path_sum += peer_addr->pathmaxrxt;
2899 paths++; 2901 paths++;
2900 } 2902 }
2901 2903
2902 /* Only validate asocmaxrxt if we have more than 2904 /* Only validate asocmaxrxt if we have more than
2903 * one path/transport. We do this because path 2905 * one path/transport. We do this because path
2904 * retransmissions are only counted when we have more 2906 * retransmissions are only counted when we have more
2905 * then one path. 2907 * then one path.
2906 */ 2908 */
2907 if (paths > 1 && 2909 if (paths > 1 &&
2908 assocparams.sasoc_asocmaxrxt > path_sum) 2910 assocparams.sasoc_asocmaxrxt > path_sum)
2909 return -EINVAL; 2911 return -EINVAL;
2910 2912
2911 asoc->max_retrans = assocparams.sasoc_asocmaxrxt; 2913 asoc->max_retrans = assocparams.sasoc_asocmaxrxt;
2912 } 2914 }
2913 2915
2914 if (assocparams.sasoc_cookie_life != 0) { 2916 if (assocparams.sasoc_cookie_life != 0) {
2915 asoc->cookie_life.tv_sec = 2917 asoc->cookie_life.tv_sec =
2916 assocparams.sasoc_cookie_life / 1000; 2918 assocparams.sasoc_cookie_life / 1000;
2917 asoc->cookie_life.tv_usec = 2919 asoc->cookie_life.tv_usec =
2918 (assocparams.sasoc_cookie_life % 1000) 2920 (assocparams.sasoc_cookie_life % 1000)
2919 * 1000; 2921 * 1000;
2920 } 2922 }
2921 } else { 2923 } else {
2922 /* Set the values to the endpoint */ 2924 /* Set the values to the endpoint */
2923 struct sctp_sock *sp = sctp_sk(sk); 2925 struct sctp_sock *sp = sctp_sk(sk);
2924 2926
2925 if (assocparams.sasoc_asocmaxrxt != 0) 2927 if (assocparams.sasoc_asocmaxrxt != 0)
2926 sp->assocparams.sasoc_asocmaxrxt = 2928 sp->assocparams.sasoc_asocmaxrxt =
2927 assocparams.sasoc_asocmaxrxt; 2929 assocparams.sasoc_asocmaxrxt;
2928 if (assocparams.sasoc_cookie_life != 0) 2930 if (assocparams.sasoc_cookie_life != 0)
2929 sp->assocparams.sasoc_cookie_life = 2931 sp->assocparams.sasoc_cookie_life =
2930 assocparams.sasoc_cookie_life; 2932 assocparams.sasoc_cookie_life;
2931 } 2933 }
2932 return 0; 2934 return 0;
2933 } 2935 }
2934 2936
2935 /* 2937 /*
2936 * 7.1.16 Set/clear IPv4 mapped addresses (SCTP_I_WANT_MAPPED_V4_ADDR) 2938 * 7.1.16 Set/clear IPv4 mapped addresses (SCTP_I_WANT_MAPPED_V4_ADDR)
2937 * 2939 *
2938 * This socket option is a boolean flag which turns on or off mapped V4 2940 * This socket option is a boolean flag which turns on or off mapped V4
2939 * addresses. If this option is turned on and the socket is type 2941 * addresses. If this option is turned on and the socket is type
2940 * PF_INET6, then IPv4 addresses will be mapped to V6 representation. 2942 * PF_INET6, then IPv4 addresses will be mapped to V6 representation.
2941 * If this option is turned off, then no mapping will be done of V4 2943 * If this option is turned off, then no mapping will be done of V4
2942 * addresses and a user will receive both PF_INET6 and PF_INET type 2944 * addresses and a user will receive both PF_INET6 and PF_INET type
2943 * addresses on the socket. 2945 * addresses on the socket.
2944 */ 2946 */
2945 static int sctp_setsockopt_mappedv4(struct sock *sk, char __user *optval, unsigned int optlen) 2947 static int sctp_setsockopt_mappedv4(struct sock *sk, char __user *optval, unsigned int optlen)
2946 { 2948 {
2947 int val; 2949 int val;
2948 struct sctp_sock *sp = sctp_sk(sk); 2950 struct sctp_sock *sp = sctp_sk(sk);
2949 2951
2950 if (optlen < sizeof(int)) 2952 if (optlen < sizeof(int))
2951 return -EINVAL; 2953 return -EINVAL;
2952 if (get_user(val, (int __user *)optval)) 2954 if (get_user(val, (int __user *)optval))
2953 return -EFAULT; 2955 return -EFAULT;
2954 if (val) 2956 if (val)
2955 sp->v4mapped = 1; 2957 sp->v4mapped = 1;
2956 else 2958 else
2957 sp->v4mapped = 0; 2959 sp->v4mapped = 0;
2958 2960
2959 return 0; 2961 return 0;
2960 } 2962 }
2961 2963
2962 /* 2964 /*
2963 * 8.1.16. Get or Set the Maximum Fragmentation Size (SCTP_MAXSEG) 2965 * 8.1.16. Get or Set the Maximum Fragmentation Size (SCTP_MAXSEG)
2964 * This option will get or set the maximum size to put in any outgoing 2966 * This option will get or set the maximum size to put in any outgoing
2965 * SCTP DATA chunk. If a message is larger than this size it will be 2967 * SCTP DATA chunk. If a message is larger than this size it will be
2966 * fragmented by SCTP into the specified size. Note that the underlying 2968 * fragmented by SCTP into the specified size. Note that the underlying
2967 * SCTP implementation may fragment into smaller sized chunks when the 2969 * SCTP implementation may fragment into smaller sized chunks when the
2968 * PMTU of the underlying association is smaller than the value set by 2970 * PMTU of the underlying association is smaller than the value set by
2969 * the user. The default value for this option is '0' which indicates 2971 * the user. The default value for this option is '0' which indicates
2970 * the user is NOT limiting fragmentation and only the PMTU will effect 2972 * the user is NOT limiting fragmentation and only the PMTU will effect
2971 * SCTP's choice of DATA chunk size. Note also that values set larger 2973 * SCTP's choice of DATA chunk size. Note also that values set larger
2972 * than the maximum size of an IP datagram will effectively let SCTP 2974 * than the maximum size of an IP datagram will effectively let SCTP
2973 * control fragmentation (i.e. the same as setting this option to 0). 2975 * control fragmentation (i.e. the same as setting this option to 0).
2974 * 2976 *
2975 * The following structure is used to access and modify this parameter: 2977 * The following structure is used to access and modify this parameter:
2976 * 2978 *
2977 * struct sctp_assoc_value { 2979 * struct sctp_assoc_value {
2978 * sctp_assoc_t assoc_id; 2980 * sctp_assoc_t assoc_id;
2979 * uint32_t assoc_value; 2981 * uint32_t assoc_value;
2980 * }; 2982 * };
2981 * 2983 *
2982 * assoc_id: This parameter is ignored for one-to-one style sockets. 2984 * assoc_id: This parameter is ignored for one-to-one style sockets.
2983 * For one-to-many style sockets this parameter indicates which 2985 * For one-to-many style sockets this parameter indicates which
2984 * association the user is performing an action upon. Note that if 2986 * association the user is performing an action upon. Note that if
2985 * this field's value is zero then the endpoints default value is 2987 * this field's value is zero then the endpoints default value is
2986 * changed (effecting future associations only). 2988 * changed (effecting future associations only).
2987 * assoc_value: This parameter specifies the maximum size in bytes. 2989 * assoc_value: This parameter specifies the maximum size in bytes.
2988 */ 2990 */
2989 static int sctp_setsockopt_maxseg(struct sock *sk, char __user *optval, unsigned int optlen) 2991 static int sctp_setsockopt_maxseg(struct sock *sk, char __user *optval, unsigned int optlen)
2990 { 2992 {
2991 struct sctp_assoc_value params; 2993 struct sctp_assoc_value params;
2992 struct sctp_association *asoc; 2994 struct sctp_association *asoc;
2993 struct sctp_sock *sp = sctp_sk(sk); 2995 struct sctp_sock *sp = sctp_sk(sk);
2994 int val; 2996 int val;
2995 2997
2996 if (optlen == sizeof(int)) { 2998 if (optlen == sizeof(int)) {
2997 pr_warn("Use of int in maxseg socket option deprecated\n"); 2999 pr_warn("Use of int in maxseg socket option deprecated\n");
2998 pr_warn("Use struct sctp_assoc_value instead\n"); 3000 pr_warn("Use struct sctp_assoc_value instead\n");
2999 if (copy_from_user(&val, optval, optlen)) 3001 if (copy_from_user(&val, optval, optlen))
3000 return -EFAULT; 3002 return -EFAULT;
3001 params.assoc_id = 0; 3003 params.assoc_id = 0;
3002 } else if (optlen == sizeof(struct sctp_assoc_value)) { 3004 } else if (optlen == sizeof(struct sctp_assoc_value)) {
3003 if (copy_from_user(&params, optval, optlen)) 3005 if (copy_from_user(&params, optval, optlen))
3004 return -EFAULT; 3006 return -EFAULT;
3005 val = params.assoc_value; 3007 val = params.assoc_value;
3006 } else 3008 } else
3007 return -EINVAL; 3009 return -EINVAL;
3008 3010
3009 if ((val != 0) && ((val < 8) || (val > SCTP_MAX_CHUNK_LEN))) 3011 if ((val != 0) && ((val < 8) || (val > SCTP_MAX_CHUNK_LEN)))
3010 return -EINVAL; 3012 return -EINVAL;
3011 3013
3012 asoc = sctp_id2assoc(sk, params.assoc_id); 3014 asoc = sctp_id2assoc(sk, params.assoc_id);
3013 if (!asoc && params.assoc_id && sctp_style(sk, UDP)) 3015 if (!asoc && params.assoc_id && sctp_style(sk, UDP))
3014 return -EINVAL; 3016 return -EINVAL;
3015 3017
3016 if (asoc) { 3018 if (asoc) {
3017 if (val == 0) { 3019 if (val == 0) {
3018 val = asoc->pathmtu; 3020 val = asoc->pathmtu;
3019 val -= sp->pf->af->net_header_len; 3021 val -= sp->pf->af->net_header_len;
3020 val -= sizeof(struct sctphdr) + 3022 val -= sizeof(struct sctphdr) +
3021 sizeof(struct sctp_data_chunk); 3023 sizeof(struct sctp_data_chunk);
3022 } 3024 }
3023 asoc->user_frag = val; 3025 asoc->user_frag = val;
3024 asoc->frag_point = sctp_frag_point(asoc, asoc->pathmtu); 3026 asoc->frag_point = sctp_frag_point(asoc, asoc->pathmtu);
3025 } else { 3027 } else {
3026 sp->user_frag = val; 3028 sp->user_frag = val;
3027 } 3029 }
3028 3030
3029 return 0; 3031 return 0;
3030 } 3032 }
3031 3033
3032 3034
3033 /* 3035 /*
3034 * 7.1.9 Set Peer Primary Address (SCTP_SET_PEER_PRIMARY_ADDR) 3036 * 7.1.9 Set Peer Primary Address (SCTP_SET_PEER_PRIMARY_ADDR)
3035 * 3037 *
3036 * Requests that the peer mark the enclosed address as the association 3038 * Requests that the peer mark the enclosed address as the association
3037 * primary. The enclosed address must be one of the association's 3039 * primary. The enclosed address must be one of the association's
3038 * locally bound addresses. The following structure is used to make a 3040 * locally bound addresses. The following structure is used to make a
3039 * set primary request: 3041 * set primary request:
3040 */ 3042 */
3041 static int sctp_setsockopt_peer_primary_addr(struct sock *sk, char __user *optval, 3043 static int sctp_setsockopt_peer_primary_addr(struct sock *sk, char __user *optval,
3042 unsigned int optlen) 3044 unsigned int optlen)
3043 { 3045 {
3044 struct net *net = sock_net(sk); 3046 struct net *net = sock_net(sk);
3045 struct sctp_sock *sp; 3047 struct sctp_sock *sp;
3046 struct sctp_association *asoc = NULL; 3048 struct sctp_association *asoc = NULL;
3047 struct sctp_setpeerprim prim; 3049 struct sctp_setpeerprim prim;
3048 struct sctp_chunk *chunk; 3050 struct sctp_chunk *chunk;
3049 struct sctp_af *af; 3051 struct sctp_af *af;
3050 int err; 3052 int err;
3051 3053
3052 sp = sctp_sk(sk); 3054 sp = sctp_sk(sk);
3053 3055
3054 if (!net->sctp.addip_enable) 3056 if (!net->sctp.addip_enable)
3055 return -EPERM; 3057 return -EPERM;
3056 3058
3057 if (optlen != sizeof(struct sctp_setpeerprim)) 3059 if (optlen != sizeof(struct sctp_setpeerprim))
3058 return -EINVAL; 3060 return -EINVAL;
3059 3061
3060 if (copy_from_user(&prim, optval, optlen)) 3062 if (copy_from_user(&prim, optval, optlen))
3061 return -EFAULT; 3063 return -EFAULT;
3062 3064
3063 asoc = sctp_id2assoc(sk, prim.sspp_assoc_id); 3065 asoc = sctp_id2assoc(sk, prim.sspp_assoc_id);
3064 if (!asoc) 3066 if (!asoc)
3065 return -EINVAL; 3067 return -EINVAL;
3066 3068
3067 if (!asoc->peer.asconf_capable) 3069 if (!asoc->peer.asconf_capable)
3068 return -EPERM; 3070 return -EPERM;
3069 3071
3070 if (asoc->peer.addip_disabled_mask & SCTP_PARAM_SET_PRIMARY) 3072 if (asoc->peer.addip_disabled_mask & SCTP_PARAM_SET_PRIMARY)
3071 return -EPERM; 3073 return -EPERM;
3072 3074
3073 if (!sctp_state(asoc, ESTABLISHED)) 3075 if (!sctp_state(asoc, ESTABLISHED))
3074 return -ENOTCONN; 3076 return -ENOTCONN;
3075 3077
3076 af = sctp_get_af_specific(prim.sspp_addr.ss_family); 3078 af = sctp_get_af_specific(prim.sspp_addr.ss_family);
3077 if (!af) 3079 if (!af)
3078 return -EINVAL; 3080 return -EINVAL;
3079 3081
3080 if (!af->addr_valid((union sctp_addr *)&prim.sspp_addr, sp, NULL)) 3082 if (!af->addr_valid((union sctp_addr *)&prim.sspp_addr, sp, NULL))
3081 return -EADDRNOTAVAIL; 3083 return -EADDRNOTAVAIL;
3082 3084
3083 if (!sctp_assoc_lookup_laddr(asoc, (union sctp_addr *)&prim.sspp_addr)) 3085 if (!sctp_assoc_lookup_laddr(asoc, (union sctp_addr *)&prim.sspp_addr))
3084 return -EADDRNOTAVAIL; 3086 return -EADDRNOTAVAIL;
3085 3087
3086 /* Create an ASCONF chunk with SET_PRIMARY parameter */ 3088 /* Create an ASCONF chunk with SET_PRIMARY parameter */
3087 chunk = sctp_make_asconf_set_prim(asoc, 3089 chunk = sctp_make_asconf_set_prim(asoc,
3088 (union sctp_addr *)&prim.sspp_addr); 3090 (union sctp_addr *)&prim.sspp_addr);
3089 if (!chunk) 3091 if (!chunk)
3090 return -ENOMEM; 3092 return -ENOMEM;
3091 3093
3092 err = sctp_send_asconf(asoc, chunk); 3094 err = sctp_send_asconf(asoc, chunk);
3093 3095
3094 SCTP_DEBUG_PRINTK("We set peer primary addr primitively.\n"); 3096 SCTP_DEBUG_PRINTK("We set peer primary addr primitively.\n");
3095 3097
3096 return err; 3098 return err;
3097 } 3099 }
3098 3100
3099 static int sctp_setsockopt_adaptation_layer(struct sock *sk, char __user *optval, 3101 static int sctp_setsockopt_adaptation_layer(struct sock *sk, char __user *optval,
3100 unsigned int optlen) 3102 unsigned int optlen)
3101 { 3103 {
3102 struct sctp_setadaptation adaptation; 3104 struct sctp_setadaptation adaptation;
3103 3105
3104 if (optlen != sizeof(struct sctp_setadaptation)) 3106 if (optlen != sizeof(struct sctp_setadaptation))
3105 return -EINVAL; 3107 return -EINVAL;
3106 if (copy_from_user(&adaptation, optval, optlen)) 3108 if (copy_from_user(&adaptation, optval, optlen))
3107 return -EFAULT; 3109 return -EFAULT;
3108 3110
3109 sctp_sk(sk)->adaptation_ind = adaptation.ssb_adaptation_ind; 3111 sctp_sk(sk)->adaptation_ind = adaptation.ssb_adaptation_ind;
3110 3112
3111 return 0; 3113 return 0;
3112 } 3114 }
3113 3115
3114 /* 3116 /*
3115 * 7.1.29. Set or Get the default context (SCTP_CONTEXT) 3117 * 7.1.29. Set or Get the default context (SCTP_CONTEXT)
3116 * 3118 *
3117 * The context field in the sctp_sndrcvinfo structure is normally only 3119 * The context field in the sctp_sndrcvinfo structure is normally only
3118 * used when a failed message is retrieved holding the value that was 3120 * used when a failed message is retrieved holding the value that was
3119 * sent down on the actual send call. This option allows the setting of 3121 * sent down on the actual send call. This option allows the setting of
3120 * a default context on an association basis that will be received on 3122 * a default context on an association basis that will be received on
3121 * reading messages from the peer. This is especially helpful in the 3123 * reading messages from the peer. This is especially helpful in the
3122 * one-2-many model for an application to keep some reference to an 3124 * one-2-many model for an application to keep some reference to an
3123 * internal state machine that is processing messages on the 3125 * internal state machine that is processing messages on the
3124 * association. Note that the setting of this value only effects 3126 * association. Note that the setting of this value only effects
3125 * received messages from the peer and does not effect the value that is 3127 * received messages from the peer and does not effect the value that is
3126 * saved with outbound messages. 3128 * saved with outbound messages.
3127 */ 3129 */
3128 static int sctp_setsockopt_context(struct sock *sk, char __user *optval, 3130 static int sctp_setsockopt_context(struct sock *sk, char __user *optval,
3129 unsigned int optlen) 3131 unsigned int optlen)
3130 { 3132 {
3131 struct sctp_assoc_value params; 3133 struct sctp_assoc_value params;
3132 struct sctp_sock *sp; 3134 struct sctp_sock *sp;
3133 struct sctp_association *asoc; 3135 struct sctp_association *asoc;
3134 3136
3135 if (optlen != sizeof(struct sctp_assoc_value)) 3137 if (optlen != sizeof(struct sctp_assoc_value))
3136 return -EINVAL; 3138 return -EINVAL;
3137 if (copy_from_user(&params, optval, optlen)) 3139 if (copy_from_user(&params, optval, optlen))
3138 return -EFAULT; 3140 return -EFAULT;
3139 3141
3140 sp = sctp_sk(sk); 3142 sp = sctp_sk(sk);
3141 3143
3142 if (params.assoc_id != 0) { 3144 if (params.assoc_id != 0) {
3143 asoc = sctp_id2assoc(sk, params.assoc_id); 3145 asoc = sctp_id2assoc(sk, params.assoc_id);
3144 if (!asoc) 3146 if (!asoc)
3145 return -EINVAL; 3147 return -EINVAL;
3146 asoc->default_rcv_context = params.assoc_value; 3148 asoc->default_rcv_context = params.assoc_value;
3147 } else { 3149 } else {
3148 sp->default_rcv_context = params.assoc_value; 3150 sp->default_rcv_context = params.assoc_value;
3149 } 3151 }
3150 3152
3151 return 0; 3153 return 0;
3152 } 3154 }
3153 3155
3154 /* 3156 /*
3155 * 7.1.24. Get or set fragmented interleave (SCTP_FRAGMENT_INTERLEAVE) 3157 * 7.1.24. Get or set fragmented interleave (SCTP_FRAGMENT_INTERLEAVE)
3156 * 3158 *
3157 * This options will at a minimum specify if the implementation is doing 3159 * This options will at a minimum specify if the implementation is doing
3158 * fragmented interleave. Fragmented interleave, for a one to many 3160 * fragmented interleave. Fragmented interleave, for a one to many
3159 * socket, is when subsequent calls to receive a message may return 3161 * socket, is when subsequent calls to receive a message may return
3160 * parts of messages from different associations. Some implementations 3162 * parts of messages from different associations. Some implementations
3161 * may allow you to turn this value on or off. If so, when turned off, 3163 * may allow you to turn this value on or off. If so, when turned off,
3162 * no fragment interleave will occur (which will cause a head of line 3164 * no fragment interleave will occur (which will cause a head of line
3163 * blocking amongst multiple associations sharing the same one to many 3165 * blocking amongst multiple associations sharing the same one to many
3164 * socket). When this option is turned on, then each receive call may 3166 * socket). When this option is turned on, then each receive call may
3165 * come from a different association (thus the user must receive data 3167 * come from a different association (thus the user must receive data
3166 * with the extended calls (e.g. sctp_recvmsg) to keep track of which 3168 * with the extended calls (e.g. sctp_recvmsg) to keep track of which
3167 * association each receive belongs to. 3169 * association each receive belongs to.
3168 * 3170 *
3169 * This option takes a boolean value. A non-zero value indicates that 3171 * This option takes a boolean value. A non-zero value indicates that
3170 * fragmented interleave is on. A value of zero indicates that 3172 * fragmented interleave is on. A value of zero indicates that
3171 * fragmented interleave is off. 3173 * fragmented interleave is off.
3172 * 3174 *
3173 * Note that it is important that an implementation that allows this 3175 * Note that it is important that an implementation that allows this
3174 * option to be turned on, have it off by default. Otherwise an unaware 3176 * option to be turned on, have it off by default. Otherwise an unaware
3175 * application using the one to many model may become confused and act 3177 * application using the one to many model may become confused and act
3176 * incorrectly. 3178 * incorrectly.
3177 */ 3179 */
3178 static int sctp_setsockopt_fragment_interleave(struct sock *sk, 3180 static int sctp_setsockopt_fragment_interleave(struct sock *sk,
3179 char __user *optval, 3181 char __user *optval,
3180 unsigned int optlen) 3182 unsigned int optlen)
3181 { 3183 {
3182 int val; 3184 int val;
3183 3185
3184 if (optlen != sizeof(int)) 3186 if (optlen != sizeof(int))
3185 return -EINVAL; 3187 return -EINVAL;
3186 if (get_user(val, (int __user *)optval)) 3188 if (get_user(val, (int __user *)optval))
3187 return -EFAULT; 3189 return -EFAULT;
3188 3190
3189 sctp_sk(sk)->frag_interleave = (val == 0) ? 0 : 1; 3191 sctp_sk(sk)->frag_interleave = (val == 0) ? 0 : 1;
3190 3192
3191 return 0; 3193 return 0;
3192 } 3194 }
3193 3195
3194 /* 3196 /*
3195 * 8.1.21. Set or Get the SCTP Partial Delivery Point 3197 * 8.1.21. Set or Get the SCTP Partial Delivery Point
3196 * (SCTP_PARTIAL_DELIVERY_POINT) 3198 * (SCTP_PARTIAL_DELIVERY_POINT)
3197 * 3199 *
3198 * This option will set or get the SCTP partial delivery point. This 3200 * This option will set or get the SCTP partial delivery point. This
3199 * point is the size of a message where the partial delivery API will be 3201 * point is the size of a message where the partial delivery API will be
3200 * invoked to help free up rwnd space for the peer. Setting this to a 3202 * invoked to help free up rwnd space for the peer. Setting this to a
3201 * lower value will cause partial deliveries to happen more often. The 3203 * lower value will cause partial deliveries to happen more often. The
3202 * calls argument is an integer that sets or gets the partial delivery 3204 * calls argument is an integer that sets or gets the partial delivery
3203 * point. Note also that the call will fail if the user attempts to set 3205 * point. Note also that the call will fail if the user attempts to set
3204 * this value larger than the socket receive buffer size. 3206 * this value larger than the socket receive buffer size.
3205 * 3207 *
3206 * Note that any single message having a length smaller than or equal to 3208 * Note that any single message having a length smaller than or equal to
3207 * the SCTP partial delivery point will be delivered in one single read 3209 * the SCTP partial delivery point will be delivered in one single read
3208 * call as long as the user provided buffer is large enough to hold the 3210 * call as long as the user provided buffer is large enough to hold the
3209 * message. 3211 * message.
3210 */ 3212 */
3211 static int sctp_setsockopt_partial_delivery_point(struct sock *sk, 3213 static int sctp_setsockopt_partial_delivery_point(struct sock *sk,
3212 char __user *optval, 3214 char __user *optval,
3213 unsigned int optlen) 3215 unsigned int optlen)
3214 { 3216 {
3215 u32 val; 3217 u32 val;
3216 3218
3217 if (optlen != sizeof(u32)) 3219 if (optlen != sizeof(u32))
3218 return -EINVAL; 3220 return -EINVAL;
3219 if (get_user(val, (int __user *)optval)) 3221 if (get_user(val, (int __user *)optval))
3220 return -EFAULT; 3222 return -EFAULT;
3221 3223
3222 /* Note: We double the receive buffer from what the user sets 3224 /* Note: We double the receive buffer from what the user sets
3223 * it to be, also initial rwnd is based on rcvbuf/2. 3225 * it to be, also initial rwnd is based on rcvbuf/2.
3224 */ 3226 */
3225 if (val > (sk->sk_rcvbuf >> 1)) 3227 if (val > (sk->sk_rcvbuf >> 1))
3226 return -EINVAL; 3228 return -EINVAL;
3227 3229
3228 sctp_sk(sk)->pd_point = val; 3230 sctp_sk(sk)->pd_point = val;
3229 3231
3230 return 0; /* is this the right error code? */ 3232 return 0; /* is this the right error code? */
3231 } 3233 }
3232 3234
3233 /* 3235 /*
3234 * 7.1.28. Set or Get the maximum burst (SCTP_MAX_BURST) 3236 * 7.1.28. Set or Get the maximum burst (SCTP_MAX_BURST)
3235 * 3237 *
3236 * This option will allow a user to change the maximum burst of packets 3238 * This option will allow a user to change the maximum burst of packets
3237 * that can be emitted by this association. Note that the default value 3239 * that can be emitted by this association. Note that the default value
3238 * is 4, and some implementations may restrict this setting so that it 3240 * is 4, and some implementations may restrict this setting so that it
3239 * can only be lowered. 3241 * can only be lowered.
3240 * 3242 *
3241 * NOTE: This text doesn't seem right. Do this on a socket basis with 3243 * NOTE: This text doesn't seem right. Do this on a socket basis with
3242 * future associations inheriting the socket value. 3244 * future associations inheriting the socket value.
3243 */ 3245 */
3244 static int sctp_setsockopt_maxburst(struct sock *sk, 3246 static int sctp_setsockopt_maxburst(struct sock *sk,
3245 char __user *optval, 3247 char __user *optval,
3246 unsigned int optlen) 3248 unsigned int optlen)
3247 { 3249 {
3248 struct sctp_assoc_value params; 3250 struct sctp_assoc_value params;
3249 struct sctp_sock *sp; 3251 struct sctp_sock *sp;
3250 struct sctp_association *asoc; 3252 struct sctp_association *asoc;
3251 int val; 3253 int val;
3252 int assoc_id = 0; 3254 int assoc_id = 0;
3253 3255
3254 if (optlen == sizeof(int)) { 3256 if (optlen == sizeof(int)) {
3255 pr_warn("Use of int in max_burst socket option deprecated\n"); 3257 pr_warn("Use of int in max_burst socket option deprecated\n");
3256 pr_warn("Use struct sctp_assoc_value instead\n"); 3258 pr_warn("Use struct sctp_assoc_value instead\n");
3257 if (copy_from_user(&val, optval, optlen)) 3259 if (copy_from_user(&val, optval, optlen))
3258 return -EFAULT; 3260 return -EFAULT;
3259 } else if (optlen == sizeof(struct sctp_assoc_value)) { 3261 } else if (optlen == sizeof(struct sctp_assoc_value)) {
3260 if (copy_from_user(&params, optval, optlen)) 3262 if (copy_from_user(&params, optval, optlen))
3261 return -EFAULT; 3263 return -EFAULT;
3262 val = params.assoc_value; 3264 val = params.assoc_value;
3263 assoc_id = params.assoc_id; 3265 assoc_id = params.assoc_id;
3264 } else 3266 } else
3265 return -EINVAL; 3267 return -EINVAL;
3266 3268
3267 sp = sctp_sk(sk); 3269 sp = sctp_sk(sk);
3268 3270
3269 if (assoc_id != 0) { 3271 if (assoc_id != 0) {
3270 asoc = sctp_id2assoc(sk, assoc_id); 3272 asoc = sctp_id2assoc(sk, assoc_id);
3271 if (!asoc) 3273 if (!asoc)
3272 return -EINVAL; 3274 return -EINVAL;
3273 asoc->max_burst = val; 3275 asoc->max_burst = val;
3274 } else 3276 } else
3275 sp->max_burst = val; 3277 sp->max_burst = val;
3276 3278
3277 return 0; 3279 return 0;
3278 } 3280 }
3279 3281
3280 /* 3282 /*
3281 * 7.1.18. Add a chunk that must be authenticated (SCTP_AUTH_CHUNK) 3283 * 7.1.18. Add a chunk that must be authenticated (SCTP_AUTH_CHUNK)
3282 * 3284 *
3283 * This set option adds a chunk type that the user is requesting to be 3285 * This set option adds a chunk type that the user is requesting to be
3284 * received only in an authenticated way. Changes to the list of chunks 3286 * received only in an authenticated way. Changes to the list of chunks
3285 * will only effect future associations on the socket. 3287 * will only effect future associations on the socket.
3286 */ 3288 */
3287 static int sctp_setsockopt_auth_chunk(struct sock *sk, 3289 static int sctp_setsockopt_auth_chunk(struct sock *sk,
3288 char __user *optval, 3290 char __user *optval,
3289 unsigned int optlen) 3291 unsigned int optlen)
3290 { 3292 {
3291 struct net *net = sock_net(sk); 3293 struct net *net = sock_net(sk);
3292 struct sctp_authchunk val; 3294 struct sctp_authchunk val;
3293 3295
3294 if (!net->sctp.auth_enable) 3296 if (!net->sctp.auth_enable)
3295 return -EACCES; 3297 return -EACCES;
3296 3298
3297 if (optlen != sizeof(struct sctp_authchunk)) 3299 if (optlen != sizeof(struct sctp_authchunk))
3298 return -EINVAL; 3300 return -EINVAL;
3299 if (copy_from_user(&val, optval, optlen)) 3301 if (copy_from_user(&val, optval, optlen))
3300 return -EFAULT; 3302 return -EFAULT;
3301 3303
3302 switch (val.sauth_chunk) { 3304 switch (val.sauth_chunk) {
3303 case SCTP_CID_INIT: 3305 case SCTP_CID_INIT:
3304 case SCTP_CID_INIT_ACK: 3306 case SCTP_CID_INIT_ACK:
3305 case SCTP_CID_SHUTDOWN_COMPLETE: 3307 case SCTP_CID_SHUTDOWN_COMPLETE:
3306 case SCTP_CID_AUTH: 3308 case SCTP_CID_AUTH:
3307 return -EINVAL; 3309 return -EINVAL;
3308 } 3310 }
3309 3311
3310 /* add this chunk id to the endpoint */ 3312 /* add this chunk id to the endpoint */
3311 return sctp_auth_ep_add_chunkid(sctp_sk(sk)->ep, val.sauth_chunk); 3313 return sctp_auth_ep_add_chunkid(sctp_sk(sk)->ep, val.sauth_chunk);
3312 } 3314 }
3313 3315
3314 /* 3316 /*
3315 * 7.1.19. Get or set the list of supported HMAC Identifiers (SCTP_HMAC_IDENT) 3317 * 7.1.19. Get or set the list of supported HMAC Identifiers (SCTP_HMAC_IDENT)
3316 * 3318 *
3317 * This option gets or sets the list of HMAC algorithms that the local 3319 * This option gets or sets the list of HMAC algorithms that the local
3318 * endpoint requires the peer to use. 3320 * endpoint requires the peer to use.
3319 */ 3321 */
3320 static int sctp_setsockopt_hmac_ident(struct sock *sk, 3322 static int sctp_setsockopt_hmac_ident(struct sock *sk,
3321 char __user *optval, 3323 char __user *optval,
3322 unsigned int optlen) 3324 unsigned int optlen)
3323 { 3325 {
3324 struct net *net = sock_net(sk); 3326 struct net *net = sock_net(sk);
3325 struct sctp_hmacalgo *hmacs; 3327 struct sctp_hmacalgo *hmacs;
3326 u32 idents; 3328 u32 idents;
3327 int err; 3329 int err;
3328 3330
3329 if (!net->sctp.auth_enable) 3331 if (!net->sctp.auth_enable)
3330 return -EACCES; 3332 return -EACCES;
3331 3333
3332 if (optlen < sizeof(struct sctp_hmacalgo)) 3334 if (optlen < sizeof(struct sctp_hmacalgo))
3333 return -EINVAL; 3335 return -EINVAL;
3334 3336
3335 hmacs= memdup_user(optval, optlen); 3337 hmacs= memdup_user(optval, optlen);
3336 if (IS_ERR(hmacs)) 3338 if (IS_ERR(hmacs))
3337 return PTR_ERR(hmacs); 3339 return PTR_ERR(hmacs);
3338 3340
3339 idents = hmacs->shmac_num_idents; 3341 idents = hmacs->shmac_num_idents;
3340 if (idents == 0 || idents > SCTP_AUTH_NUM_HMACS || 3342 if (idents == 0 || idents > SCTP_AUTH_NUM_HMACS ||
3341 (idents * sizeof(u16)) > (optlen - sizeof(struct sctp_hmacalgo))) { 3343 (idents * sizeof(u16)) > (optlen - sizeof(struct sctp_hmacalgo))) {
3342 err = -EINVAL; 3344 err = -EINVAL;
3343 goto out; 3345 goto out;
3344 } 3346 }
3345 3347
3346 err = sctp_auth_ep_set_hmacs(sctp_sk(sk)->ep, hmacs); 3348 err = sctp_auth_ep_set_hmacs(sctp_sk(sk)->ep, hmacs);
3347 out: 3349 out:
3348 kfree(hmacs); 3350 kfree(hmacs);
3349 return err; 3351 return err;
3350 } 3352 }
3351 3353
3352 /* 3354 /*
3353 * 7.1.20. Set a shared key (SCTP_AUTH_KEY) 3355 * 7.1.20. Set a shared key (SCTP_AUTH_KEY)
3354 * 3356 *
3355 * This option will set a shared secret key which is used to build an 3357 * This option will set a shared secret key which is used to build an
3356 * association shared key. 3358 * association shared key.
3357 */ 3359 */
3358 static int sctp_setsockopt_auth_key(struct sock *sk, 3360 static int sctp_setsockopt_auth_key(struct sock *sk,
3359 char __user *optval, 3361 char __user *optval,
3360 unsigned int optlen) 3362 unsigned int optlen)
3361 { 3363 {
3362 struct net *net = sock_net(sk); 3364 struct net *net = sock_net(sk);
3363 struct sctp_authkey *authkey; 3365 struct sctp_authkey *authkey;
3364 struct sctp_association *asoc; 3366 struct sctp_association *asoc;
3365 int ret; 3367 int ret;
3366 3368
3367 if (!net->sctp.auth_enable) 3369 if (!net->sctp.auth_enable)
3368 return -EACCES; 3370 return -EACCES;
3369 3371
3370 if (optlen <= sizeof(struct sctp_authkey)) 3372 if (optlen <= sizeof(struct sctp_authkey))
3371 return -EINVAL; 3373 return -EINVAL;
3372 3374
3373 authkey= memdup_user(optval, optlen); 3375 authkey= memdup_user(optval, optlen);
3374 if (IS_ERR(authkey)) 3376 if (IS_ERR(authkey))
3375 return PTR_ERR(authkey); 3377 return PTR_ERR(authkey);
3376 3378
3377 if (authkey->sca_keylength > optlen - sizeof(struct sctp_authkey)) { 3379 if (authkey->sca_keylength > optlen - sizeof(struct sctp_authkey)) {
3378 ret = -EINVAL; 3380 ret = -EINVAL;
3379 goto out; 3381 goto out;
3380 } 3382 }
3381 3383
3382 asoc = sctp_id2assoc(sk, authkey->sca_assoc_id); 3384 asoc = sctp_id2assoc(sk, authkey->sca_assoc_id);
3383 if (!asoc && authkey->sca_assoc_id && sctp_style(sk, UDP)) { 3385 if (!asoc && authkey->sca_assoc_id && sctp_style(sk, UDP)) {
3384 ret = -EINVAL; 3386 ret = -EINVAL;
3385 goto out; 3387 goto out;
3386 } 3388 }
3387 3389
3388 ret = sctp_auth_set_key(sctp_sk(sk)->ep, asoc, authkey); 3390 ret = sctp_auth_set_key(sctp_sk(sk)->ep, asoc, authkey);
3389 out: 3391 out:
3390 kfree(authkey); 3392 kfree(authkey);
3391 return ret; 3393 return ret;
3392 } 3394 }
3393 3395
3394 /* 3396 /*
3395 * 7.1.21. Get or set the active shared key (SCTP_AUTH_ACTIVE_KEY) 3397 * 7.1.21. Get or set the active shared key (SCTP_AUTH_ACTIVE_KEY)
3396 * 3398 *
3397 * This option will get or set the active shared key to be used to build 3399 * This option will get or set the active shared key to be used to build
3398 * the association shared key. 3400 * the association shared key.
3399 */ 3401 */
3400 static int sctp_setsockopt_active_key(struct sock *sk, 3402 static int sctp_setsockopt_active_key(struct sock *sk,
3401 char __user *optval, 3403 char __user *optval,
3402 unsigned int optlen) 3404 unsigned int optlen)
3403 { 3405 {
3404 struct net *net = sock_net(sk); 3406 struct net *net = sock_net(sk);
3405 struct sctp_authkeyid val; 3407 struct sctp_authkeyid val;
3406 struct sctp_association *asoc; 3408 struct sctp_association *asoc;
3407 3409
3408 if (!net->sctp.auth_enable) 3410 if (!net->sctp.auth_enable)
3409 return -EACCES; 3411 return -EACCES;
3410 3412
3411 if (optlen != sizeof(struct sctp_authkeyid)) 3413 if (optlen != sizeof(struct sctp_authkeyid))
3412 return -EINVAL; 3414 return -EINVAL;
3413 if (copy_from_user(&val, optval, optlen)) 3415 if (copy_from_user(&val, optval, optlen))
3414 return -EFAULT; 3416 return -EFAULT;
3415 3417
3416 asoc = sctp_id2assoc(sk, val.scact_assoc_id); 3418 asoc = sctp_id2assoc(sk, val.scact_assoc_id);
3417 if (!asoc && val.scact_assoc_id && sctp_style(sk, UDP)) 3419 if (!asoc && val.scact_assoc_id && sctp_style(sk, UDP))
3418 return -EINVAL; 3420 return -EINVAL;
3419 3421
3420 return sctp_auth_set_active_key(sctp_sk(sk)->ep, asoc, 3422 return sctp_auth_set_active_key(sctp_sk(sk)->ep, asoc,
3421 val.scact_keynumber); 3423 val.scact_keynumber);
3422 } 3424 }
3423 3425
3424 /* 3426 /*
3425 * 7.1.22. Delete a shared key (SCTP_AUTH_DELETE_KEY) 3427 * 7.1.22. Delete a shared key (SCTP_AUTH_DELETE_KEY)
3426 * 3428 *
3427 * This set option will delete a shared secret key from use. 3429 * This set option will delete a shared secret key from use.
3428 */ 3430 */
3429 static int sctp_setsockopt_del_key(struct sock *sk, 3431 static int sctp_setsockopt_del_key(struct sock *sk,
3430 char __user *optval, 3432 char __user *optval,
3431 unsigned int optlen) 3433 unsigned int optlen)
3432 { 3434 {
3433 struct net *net = sock_net(sk); 3435 struct net *net = sock_net(sk);
3434 struct sctp_authkeyid val; 3436 struct sctp_authkeyid val;
3435 struct sctp_association *asoc; 3437 struct sctp_association *asoc;
3436 3438
3437 if (!net->sctp.auth_enable) 3439 if (!net->sctp.auth_enable)
3438 return -EACCES; 3440 return -EACCES;
3439 3441
3440 if (optlen != sizeof(struct sctp_authkeyid)) 3442 if (optlen != sizeof(struct sctp_authkeyid))
3441 return -EINVAL; 3443 return -EINVAL;
3442 if (copy_from_user(&val, optval, optlen)) 3444 if (copy_from_user(&val, optval, optlen))
3443 return -EFAULT; 3445 return -EFAULT;
3444 3446
3445 asoc = sctp_id2assoc(sk, val.scact_assoc_id); 3447 asoc = sctp_id2assoc(sk, val.scact_assoc_id);
3446 if (!asoc && val.scact_assoc_id && sctp_style(sk, UDP)) 3448 if (!asoc && val.scact_assoc_id && sctp_style(sk, UDP))
3447 return -EINVAL; 3449 return -EINVAL;
3448 3450
3449 return sctp_auth_del_key_id(sctp_sk(sk)->ep, asoc, 3451 return sctp_auth_del_key_id(sctp_sk(sk)->ep, asoc,
3450 val.scact_keynumber); 3452 val.scact_keynumber);
3451 3453
3452 } 3454 }
3453 3455
3454 /* 3456 /*
3455 * 8.1.23 SCTP_AUTO_ASCONF 3457 * 8.1.23 SCTP_AUTO_ASCONF
3456 * 3458 *
3457 * This option will enable or disable the use of the automatic generation of 3459 * This option will enable or disable the use of the automatic generation of
3458 * ASCONF chunks to add and delete addresses to an existing association. Note 3460 * ASCONF chunks to add and delete addresses to an existing association. Note
3459 * that this option has two caveats namely: a) it only affects sockets that 3461 * that this option has two caveats namely: a) it only affects sockets that
3460 * are bound to all addresses available to the SCTP stack, and b) the system 3462 * are bound to all addresses available to the SCTP stack, and b) the system
3461 * administrator may have an overriding control that turns the ASCONF feature 3463 * administrator may have an overriding control that turns the ASCONF feature
3462 * off no matter what setting the socket option may have. 3464 * off no matter what setting the socket option may have.
3463 * This option expects an integer boolean flag, where a non-zero value turns on 3465 * This option expects an integer boolean flag, where a non-zero value turns on
3464 * the option, and a zero value turns off the option. 3466 * the option, and a zero value turns off the option.
3465 * Note. In this implementation, socket operation overrides default parameter 3467 * Note. In this implementation, socket operation overrides default parameter
3466 * being set by sysctl as well as FreeBSD implementation 3468 * being set by sysctl as well as FreeBSD implementation
3467 */ 3469 */
3468 static int sctp_setsockopt_auto_asconf(struct sock *sk, char __user *optval, 3470 static int sctp_setsockopt_auto_asconf(struct sock *sk, char __user *optval,
3469 unsigned int optlen) 3471 unsigned int optlen)
3470 { 3472 {
3471 int val; 3473 int val;
3472 struct sctp_sock *sp = sctp_sk(sk); 3474 struct sctp_sock *sp = sctp_sk(sk);
3473 3475
3474 if (optlen < sizeof(int)) 3476 if (optlen < sizeof(int))
3475 return -EINVAL; 3477 return -EINVAL;
3476 if (get_user(val, (int __user *)optval)) 3478 if (get_user(val, (int __user *)optval))
3477 return -EFAULT; 3479 return -EFAULT;
3478 if (!sctp_is_ep_boundall(sk) && val) 3480 if (!sctp_is_ep_boundall(sk) && val)
3479 return -EINVAL; 3481 return -EINVAL;
3480 if ((val && sp->do_auto_asconf) || (!val && !sp->do_auto_asconf)) 3482 if ((val && sp->do_auto_asconf) || (!val && !sp->do_auto_asconf))
3481 return 0; 3483 return 0;
3482 3484
3483 if (val == 0 && sp->do_auto_asconf) { 3485 if (val == 0 && sp->do_auto_asconf) {
3484 list_del(&sp->auto_asconf_list); 3486 list_del(&sp->auto_asconf_list);
3485 sp->do_auto_asconf = 0; 3487 sp->do_auto_asconf = 0;
3486 } else if (val && !sp->do_auto_asconf) { 3488 } else if (val && !sp->do_auto_asconf) {
3487 list_add_tail(&sp->auto_asconf_list, 3489 list_add_tail(&sp->auto_asconf_list,
3488 &sock_net(sk)->sctp.auto_asconf_splist); 3490 &sock_net(sk)->sctp.auto_asconf_splist);
3489 sp->do_auto_asconf = 1; 3491 sp->do_auto_asconf = 1;
3490 } 3492 }
3491 return 0; 3493 return 0;
3492 } 3494 }
3493 3495
3494 3496
3495 /* 3497 /*
3496 * SCTP_PEER_ADDR_THLDS 3498 * SCTP_PEER_ADDR_THLDS
3497 * 3499 *
3498 * This option allows us to alter the partially failed threshold for one or all 3500 * This option allows us to alter the partially failed threshold for one or all
3499 * transports in an association. See Section 6.1 of: 3501 * transports in an association. See Section 6.1 of:
3500 * http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt 3502 * http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
3501 */ 3503 */
3502 static int sctp_setsockopt_paddr_thresholds(struct sock *sk, 3504 static int sctp_setsockopt_paddr_thresholds(struct sock *sk,
3503 char __user *optval, 3505 char __user *optval,
3504 unsigned int optlen) 3506 unsigned int optlen)
3505 { 3507 {
3506 struct sctp_paddrthlds val; 3508 struct sctp_paddrthlds val;
3507 struct sctp_transport *trans; 3509 struct sctp_transport *trans;
3508 struct sctp_association *asoc; 3510 struct sctp_association *asoc;
3509 3511
3510 if (optlen < sizeof(struct sctp_paddrthlds)) 3512 if (optlen < sizeof(struct sctp_paddrthlds))
3511 return -EINVAL; 3513 return -EINVAL;
3512 if (copy_from_user(&val, (struct sctp_paddrthlds __user *)optval, 3514 if (copy_from_user(&val, (struct sctp_paddrthlds __user *)optval,
3513 sizeof(struct sctp_paddrthlds))) 3515 sizeof(struct sctp_paddrthlds)))
3514 return -EFAULT; 3516 return -EFAULT;
3515 3517
3516 3518
3517 if (sctp_is_any(sk, (const union sctp_addr *)&val.spt_address)) { 3519 if (sctp_is_any(sk, (const union sctp_addr *)&val.spt_address)) {
3518 asoc = sctp_id2assoc(sk, val.spt_assoc_id); 3520 asoc = sctp_id2assoc(sk, val.spt_assoc_id);
3519 if (!asoc) 3521 if (!asoc)
3520 return -ENOENT; 3522 return -ENOENT;
3521 list_for_each_entry(trans, &asoc->peer.transport_addr_list, 3523 list_for_each_entry(trans, &asoc->peer.transport_addr_list,
3522 transports) { 3524 transports) {
3523 if (val.spt_pathmaxrxt) 3525 if (val.spt_pathmaxrxt)
3524 trans->pathmaxrxt = val.spt_pathmaxrxt; 3526 trans->pathmaxrxt = val.spt_pathmaxrxt;
3525 trans->pf_retrans = val.spt_pathpfthld; 3527 trans->pf_retrans = val.spt_pathpfthld;
3526 } 3528 }
3527 3529
3528 if (val.spt_pathmaxrxt) 3530 if (val.spt_pathmaxrxt)
3529 asoc->pathmaxrxt = val.spt_pathmaxrxt; 3531 asoc->pathmaxrxt = val.spt_pathmaxrxt;
3530 asoc->pf_retrans = val.spt_pathpfthld; 3532 asoc->pf_retrans = val.spt_pathpfthld;
3531 } else { 3533 } else {
3532 trans = sctp_addr_id2transport(sk, &val.spt_address, 3534 trans = sctp_addr_id2transport(sk, &val.spt_address,
3533 val.spt_assoc_id); 3535 val.spt_assoc_id);
3534 if (!trans) 3536 if (!trans)
3535 return -ENOENT; 3537 return -ENOENT;
3536 3538
3537 if (val.spt_pathmaxrxt) 3539 if (val.spt_pathmaxrxt)
3538 trans->pathmaxrxt = val.spt_pathmaxrxt; 3540 trans->pathmaxrxt = val.spt_pathmaxrxt;
3539 trans->pf_retrans = val.spt_pathpfthld; 3541 trans->pf_retrans = val.spt_pathpfthld;
3540 } 3542 }
3541 3543
3542 return 0; 3544 return 0;
3543 } 3545 }
3544 3546
3545 /* API 6.2 setsockopt(), getsockopt() 3547 /* API 6.2 setsockopt(), getsockopt()
3546 * 3548 *
3547 * Applications use setsockopt() and getsockopt() to set or retrieve 3549 * Applications use setsockopt() and getsockopt() to set or retrieve
3548 * socket options. Socket options are used to change the default 3550 * socket options. Socket options are used to change the default
3549 * behavior of sockets calls. They are described in Section 7. 3551 * behavior of sockets calls. They are described in Section 7.
3550 * 3552 *
3551 * The syntax is: 3553 * The syntax is:
3552 * 3554 *
3553 * ret = getsockopt(int sd, int level, int optname, void __user *optval, 3555 * ret = getsockopt(int sd, int level, int optname, void __user *optval,
3554 * int __user *optlen); 3556 * int __user *optlen);
3555 * ret = setsockopt(int sd, int level, int optname, const void __user *optval, 3557 * ret = setsockopt(int sd, int level, int optname, const void __user *optval,
3556 * int optlen); 3558 * int optlen);
3557 * 3559 *
3558 * sd - the socket descript. 3560 * sd - the socket descript.
3559 * level - set to IPPROTO_SCTP for all SCTP options. 3561 * level - set to IPPROTO_SCTP for all SCTP options.
3560 * optname - the option name. 3562 * optname - the option name.
3561 * optval - the buffer to store the value of the option. 3563 * optval - the buffer to store the value of the option.
3562 * optlen - the size of the buffer. 3564 * optlen - the size of the buffer.
3563 */ 3565 */
3564 SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname, 3566 SCTP_STATIC int sctp_setsockopt(struct sock *sk, int level, int optname,
3565 char __user *optval, unsigned int optlen) 3567 char __user *optval, unsigned int optlen)
3566 { 3568 {
3567 int retval = 0; 3569 int retval = 0;
3568 3570
3569 SCTP_DEBUG_PRINTK("sctp_setsockopt(sk: %p... optname: %d)\n", 3571 SCTP_DEBUG_PRINTK("sctp_setsockopt(sk: %p... optname: %d)\n",
3570 sk, optname); 3572 sk, optname);
3571 3573
3572 /* I can hardly begin to describe how wrong this is. This is 3574 /* I can hardly begin to describe how wrong this is. This is
3573 * so broken as to be worse than useless. The API draft 3575 * so broken as to be worse than useless. The API draft
3574 * REALLY is NOT helpful here... I am not convinced that the 3576 * REALLY is NOT helpful here... I am not convinced that the
3575 * semantics of setsockopt() with a level OTHER THAN SOL_SCTP 3577 * semantics of setsockopt() with a level OTHER THAN SOL_SCTP
3576 * are at all well-founded. 3578 * are at all well-founded.
3577 */ 3579 */
3578 if (level != SOL_SCTP) { 3580 if (level != SOL_SCTP) {
3579 struct sctp_af *af = sctp_sk(sk)->pf->af; 3581 struct sctp_af *af = sctp_sk(sk)->pf->af;
3580 retval = af->setsockopt(sk, level, optname, optval, optlen); 3582 retval = af->setsockopt(sk, level, optname, optval, optlen);
3581 goto out_nounlock; 3583 goto out_nounlock;
3582 } 3584 }
3583 3585
3584 sctp_lock_sock(sk); 3586 sctp_lock_sock(sk);
3585 3587
3586 switch (optname) { 3588 switch (optname) {
3587 case SCTP_SOCKOPT_BINDX_ADD: 3589 case SCTP_SOCKOPT_BINDX_ADD:
3588 /* 'optlen' is the size of the addresses buffer. */ 3590 /* 'optlen' is the size of the addresses buffer. */
3589 retval = sctp_setsockopt_bindx(sk, (struct sockaddr __user *)optval, 3591 retval = sctp_setsockopt_bindx(sk, (struct sockaddr __user *)optval,
3590 optlen, SCTP_BINDX_ADD_ADDR); 3592 optlen, SCTP_BINDX_ADD_ADDR);
3591 break; 3593 break;
3592 3594
3593 case SCTP_SOCKOPT_BINDX_REM: 3595 case SCTP_SOCKOPT_BINDX_REM:
3594 /* 'optlen' is the size of the addresses buffer. */ 3596 /* 'optlen' is the size of the addresses buffer. */
3595 retval = sctp_setsockopt_bindx(sk, (struct sockaddr __user *)optval, 3597 retval = sctp_setsockopt_bindx(sk, (struct sockaddr __user *)optval,
3596 optlen, SCTP_BINDX_REM_ADDR); 3598 optlen, SCTP_BINDX_REM_ADDR);
3597 break; 3599 break;
3598 3600
3599 case SCTP_SOCKOPT_CONNECTX_OLD: 3601 case SCTP_SOCKOPT_CONNECTX_OLD:
3600 /* 'optlen' is the size of the addresses buffer. */ 3602 /* 'optlen' is the size of the addresses buffer. */
3601 retval = sctp_setsockopt_connectx_old(sk, 3603 retval = sctp_setsockopt_connectx_old(sk,
3602 (struct sockaddr __user *)optval, 3604 (struct sockaddr __user *)optval,
3603 optlen); 3605 optlen);
3604 break; 3606 break;
3605 3607
3606 case SCTP_SOCKOPT_CONNECTX: 3608 case SCTP_SOCKOPT_CONNECTX:
3607 /* 'optlen' is the size of the addresses buffer. */ 3609 /* 'optlen' is the size of the addresses buffer. */
3608 retval = sctp_setsockopt_connectx(sk, 3610 retval = sctp_setsockopt_connectx(sk,
3609 (struct sockaddr __user *)optval, 3611 (struct sockaddr __user *)optval,
3610 optlen); 3612 optlen);
3611 break; 3613 break;
3612 3614
3613 case SCTP_DISABLE_FRAGMENTS: 3615 case SCTP_DISABLE_FRAGMENTS:
3614 retval = sctp_setsockopt_disable_fragments(sk, optval, optlen); 3616 retval = sctp_setsockopt_disable_fragments(sk, optval, optlen);
3615 break; 3617 break;
3616 3618
3617 case SCTP_EVENTS: 3619 case SCTP_EVENTS:
3618 retval = sctp_setsockopt_events(sk, optval, optlen); 3620 retval = sctp_setsockopt_events(sk, optval, optlen);
3619 break; 3621 break;
3620 3622
3621 case SCTP_AUTOCLOSE: 3623 case SCTP_AUTOCLOSE:
3622 retval = sctp_setsockopt_autoclose(sk, optval, optlen); 3624 retval = sctp_setsockopt_autoclose(sk, optval, optlen);
3623 break; 3625 break;
3624 3626
3625 case SCTP_PEER_ADDR_PARAMS: 3627 case SCTP_PEER_ADDR_PARAMS:
3626 retval = sctp_setsockopt_peer_addr_params(sk, optval, optlen); 3628 retval = sctp_setsockopt_peer_addr_params(sk, optval, optlen);
3627 break; 3629 break;
3628 3630
3629 case SCTP_DELAYED_SACK: 3631 case SCTP_DELAYED_SACK:
3630 retval = sctp_setsockopt_delayed_ack(sk, optval, optlen); 3632 retval = sctp_setsockopt_delayed_ack(sk, optval, optlen);
3631 break; 3633 break;
3632 case SCTP_PARTIAL_DELIVERY_POINT: 3634 case SCTP_PARTIAL_DELIVERY_POINT:
3633 retval = sctp_setsockopt_partial_delivery_point(sk, optval, optlen); 3635 retval = sctp_setsockopt_partial_delivery_point(sk, optval, optlen);
3634 break; 3636 break;
3635 3637
3636 case SCTP_INITMSG: 3638 case SCTP_INITMSG:
3637 retval = sctp_setsockopt_initmsg(sk, optval, optlen); 3639 retval = sctp_setsockopt_initmsg(sk, optval, optlen);
3638 break; 3640 break;
3639 case SCTP_DEFAULT_SEND_PARAM: 3641 case SCTP_DEFAULT_SEND_PARAM:
3640 retval = sctp_setsockopt_default_send_param(sk, optval, 3642 retval = sctp_setsockopt_default_send_param(sk, optval,
3641 optlen); 3643 optlen);
3642 break; 3644 break;
3643 case SCTP_PRIMARY_ADDR: 3645 case SCTP_PRIMARY_ADDR:
3644 retval = sctp_setsockopt_primary_addr(sk, optval, optlen); 3646 retval = sctp_setsockopt_primary_addr(sk, optval, optlen);
3645 break; 3647 break;
3646 case SCTP_SET_PEER_PRIMARY_ADDR: 3648 case SCTP_SET_PEER_PRIMARY_ADDR:
3647 retval = sctp_setsockopt_peer_primary_addr(sk, optval, optlen); 3649 retval = sctp_setsockopt_peer_primary_addr(sk, optval, optlen);
3648 break; 3650 break;
3649 case SCTP_NODELAY: 3651 case SCTP_NODELAY:
3650 retval = sctp_setsockopt_nodelay(sk, optval, optlen); 3652 retval = sctp_setsockopt_nodelay(sk, optval, optlen);
3651 break; 3653 break;
3652 case SCTP_RTOINFO: 3654 case SCTP_RTOINFO:
3653 retval = sctp_setsockopt_rtoinfo(sk, optval, optlen); 3655 retval = sctp_setsockopt_rtoinfo(sk, optval, optlen);
3654 break; 3656 break;
3655 case SCTP_ASSOCINFO: 3657 case SCTP_ASSOCINFO:
3656 retval = sctp_setsockopt_associnfo(sk, optval, optlen); 3658 retval = sctp_setsockopt_associnfo(sk, optval, optlen);
3657 break; 3659 break;
3658 case SCTP_I_WANT_MAPPED_V4_ADDR: 3660 case SCTP_I_WANT_MAPPED_V4_ADDR:
3659 retval = sctp_setsockopt_mappedv4(sk, optval, optlen); 3661 retval = sctp_setsockopt_mappedv4(sk, optval, optlen);
3660 break; 3662 break;
3661 case SCTP_MAXSEG: 3663 case SCTP_MAXSEG:
3662 retval = sctp_setsockopt_maxseg(sk, optval, optlen); 3664 retval = sctp_setsockopt_maxseg(sk, optval, optlen);
3663 break; 3665 break;
3664 case SCTP_ADAPTATION_LAYER: 3666 case SCTP_ADAPTATION_LAYER:
3665 retval = sctp_setsockopt_adaptation_layer(sk, optval, optlen); 3667 retval = sctp_setsockopt_adaptation_layer(sk, optval, optlen);
3666 break; 3668 break;
3667 case SCTP_CONTEXT: 3669 case SCTP_CONTEXT:
3668 retval = sctp_setsockopt_context(sk, optval, optlen); 3670 retval = sctp_setsockopt_context(sk, optval, optlen);
3669 break; 3671 break;
3670 case SCTP_FRAGMENT_INTERLEAVE: 3672 case SCTP_FRAGMENT_INTERLEAVE:
3671 retval = sctp_setsockopt_fragment_interleave(sk, optval, optlen); 3673 retval = sctp_setsockopt_fragment_interleave(sk, optval, optlen);
3672 break; 3674 break;
3673 case SCTP_MAX_BURST: 3675 case SCTP_MAX_BURST:
3674 retval = sctp_setsockopt_maxburst(sk, optval, optlen); 3676 retval = sctp_setsockopt_maxburst(sk, optval, optlen);
3675 break; 3677 break;
3676 case SCTP_AUTH_CHUNK: 3678 case SCTP_AUTH_CHUNK:
3677 retval = sctp_setsockopt_auth_chunk(sk, optval, optlen); 3679 retval = sctp_setsockopt_auth_chunk(sk, optval, optlen);
3678 break; 3680 break;
3679 case SCTP_HMAC_IDENT: 3681 case SCTP_HMAC_IDENT:
3680 retval = sctp_setsockopt_hmac_ident(sk, optval, optlen); 3682 retval = sctp_setsockopt_hmac_ident(sk, optval, optlen);
3681 break; 3683 break;
3682 case SCTP_AUTH_KEY: 3684 case SCTP_AUTH_KEY:
3683 retval = sctp_setsockopt_auth_key(sk, optval, optlen); 3685 retval = sctp_setsockopt_auth_key(sk, optval, optlen);
3684 break; 3686 break;
3685 case SCTP_AUTH_ACTIVE_KEY: 3687 case SCTP_AUTH_ACTIVE_KEY:
3686 retval = sctp_setsockopt_active_key(sk, optval, optlen); 3688 retval = sctp_setsockopt_active_key(sk, optval, optlen);
3687 break; 3689 break;
3688 case SCTP_AUTH_DELETE_KEY: 3690 case SCTP_AUTH_DELETE_KEY:
3689 retval = sctp_setsockopt_del_key(sk, optval, optlen); 3691 retval = sctp_setsockopt_del_key(sk, optval, optlen);
3690 break; 3692 break;
3691 case SCTP_AUTO_ASCONF: 3693 case SCTP_AUTO_ASCONF:
3692 retval = sctp_setsockopt_auto_asconf(sk, optval, optlen); 3694 retval = sctp_setsockopt_auto_asconf(sk, optval, optlen);
3693 break; 3695 break;
3694 case SCTP_PEER_ADDR_THLDS: 3696 case SCTP_PEER_ADDR_THLDS:
3695 retval = sctp_setsockopt_paddr_thresholds(sk, optval, optlen); 3697 retval = sctp_setsockopt_paddr_thresholds(sk, optval, optlen);
3696 break; 3698 break;
3697 default: 3699 default:
3698 retval = -ENOPROTOOPT; 3700 retval = -ENOPROTOOPT;
3699 break; 3701 break;
3700 } 3702 }
3701 3703
3702 sctp_release_sock(sk); 3704 sctp_release_sock(sk);
3703 3705
3704 out_nounlock: 3706 out_nounlock:
3705 return retval; 3707 return retval;
3706 } 3708 }
3707 3709
3708 /* API 3.1.6 connect() - UDP Style Syntax 3710 /* API 3.1.6 connect() - UDP Style Syntax
3709 * 3711 *
3710 * An application may use the connect() call in the UDP model to initiate an 3712 * An application may use the connect() call in the UDP model to initiate an
3711 * association without sending data. 3713 * association without sending data.
3712 * 3714 *
3713 * The syntax is: 3715 * The syntax is:
3714 * 3716 *
3715 * ret = connect(int sd, const struct sockaddr *nam, socklen_t len); 3717 * ret = connect(int sd, const struct sockaddr *nam, socklen_t len);
3716 * 3718 *
3717 * sd: the socket descriptor to have a new association added to. 3719 * sd: the socket descriptor to have a new association added to.
3718 * 3720 *
3719 * nam: the address structure (either struct sockaddr_in or struct 3721 * nam: the address structure (either struct sockaddr_in or struct
3720 * sockaddr_in6 defined in RFC2553 [7]). 3722 * sockaddr_in6 defined in RFC2553 [7]).
3721 * 3723 *
3722 * len: the size of the address. 3724 * len: the size of the address.
3723 */ 3725 */
3724 SCTP_STATIC int sctp_connect(struct sock *sk, struct sockaddr *addr, 3726 SCTP_STATIC int sctp_connect(struct sock *sk, struct sockaddr *addr,
3725 int addr_len) 3727 int addr_len)
3726 { 3728 {
3727 int err = 0; 3729 int err = 0;
3728 struct sctp_af *af; 3730 struct sctp_af *af;
3729 3731
3730 sctp_lock_sock(sk); 3732 sctp_lock_sock(sk);
3731 3733
3732 SCTP_DEBUG_PRINTK("%s - sk: %p, sockaddr: %p, addr_len: %d\n", 3734 SCTP_DEBUG_PRINTK("%s - sk: %p, sockaddr: %p, addr_len: %d\n",
3733 __func__, sk, addr, addr_len); 3735 __func__, sk, addr, addr_len);
3734 3736
3735 /* Validate addr_len before calling common connect/connectx routine. */ 3737 /* Validate addr_len before calling common connect/connectx routine. */
3736 af = sctp_get_af_specific(addr->sa_family); 3738 af = sctp_get_af_specific(addr->sa_family);
3737 if (!af || addr_len < af->sockaddr_len) { 3739 if (!af || addr_len < af->sockaddr_len) {
3738 err = -EINVAL; 3740 err = -EINVAL;
3739 } else { 3741 } else {
3740 /* Pass correct addr len to common routine (so it knows there 3742 /* Pass correct addr len to common routine (so it knows there
3741 * is only one address being passed. 3743 * is only one address being passed.
3742 */ 3744 */
3743 err = __sctp_connect(sk, addr, af->sockaddr_len, NULL); 3745 err = __sctp_connect(sk, addr, af->sockaddr_len, NULL);
3744 } 3746 }
3745 3747
3746 sctp_release_sock(sk); 3748 sctp_release_sock(sk);
3747 return err; 3749 return err;
3748 } 3750 }
3749 3751
3750 /* FIXME: Write comments. */ 3752 /* FIXME: Write comments. */
3751 SCTP_STATIC int sctp_disconnect(struct sock *sk, int flags) 3753 SCTP_STATIC int sctp_disconnect(struct sock *sk, int flags)
3752 { 3754 {
3753 return -EOPNOTSUPP; /* STUB */ 3755 return -EOPNOTSUPP; /* STUB */
3754 } 3756 }
3755 3757
3756 /* 4.1.4 accept() - TCP Style Syntax 3758 /* 4.1.4 accept() - TCP Style Syntax
3757 * 3759 *
3758 * Applications use accept() call to remove an established SCTP 3760 * Applications use accept() call to remove an established SCTP
3759 * association from the accept queue of the endpoint. A new socket 3761 * association from the accept queue of the endpoint. A new socket
3760 * descriptor will be returned from accept() to represent the newly 3762 * descriptor will be returned from accept() to represent the newly
3761 * formed association. 3763 * formed association.
3762 */ 3764 */
3763 SCTP_STATIC struct sock *sctp_accept(struct sock *sk, int flags, int *err) 3765 SCTP_STATIC struct sock *sctp_accept(struct sock *sk, int flags, int *err)
3764 { 3766 {
3765 struct sctp_sock *sp; 3767 struct sctp_sock *sp;
3766 struct sctp_endpoint *ep; 3768 struct sctp_endpoint *ep;
3767 struct sock *newsk = NULL; 3769 struct sock *newsk = NULL;
3768 struct sctp_association *asoc; 3770 struct sctp_association *asoc;
3769 long timeo; 3771 long timeo;
3770 int error = 0; 3772 int error = 0;
3771 3773
3772 sctp_lock_sock(sk); 3774 sctp_lock_sock(sk);
3773 3775
3774 sp = sctp_sk(sk); 3776 sp = sctp_sk(sk);
3775 ep = sp->ep; 3777 ep = sp->ep;
3776 3778
3777 if (!sctp_style(sk, TCP)) { 3779 if (!sctp_style(sk, TCP)) {
3778 error = -EOPNOTSUPP; 3780 error = -EOPNOTSUPP;
3779 goto out; 3781 goto out;
3780 } 3782 }
3781 3783
3782 if (!sctp_sstate(sk, LISTENING)) { 3784 if (!sctp_sstate(sk, LISTENING)) {
3783 error = -EINVAL; 3785 error = -EINVAL;
3784 goto out; 3786 goto out;
3785 } 3787 }
3786 3788
3787 timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK); 3789 timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
3788 3790
3789 error = sctp_wait_for_accept(sk, timeo); 3791 error = sctp_wait_for_accept(sk, timeo);
3790 if (error) 3792 if (error)
3791 goto out; 3793 goto out;
3792 3794
3793 /* We treat the list of associations on the endpoint as the accept 3795 /* We treat the list of associations on the endpoint as the accept
3794 * queue and pick the first association on the list. 3796 * queue and pick the first association on the list.
3795 */ 3797 */
3796 asoc = list_entry(ep->asocs.next, struct sctp_association, asocs); 3798 asoc = list_entry(ep->asocs.next, struct sctp_association, asocs);
3797 3799
3798 newsk = sp->pf->create_accept_sk(sk, asoc); 3800 newsk = sp->pf->create_accept_sk(sk, asoc);
3799 if (!newsk) { 3801 if (!newsk) {
3800 error = -ENOMEM; 3802 error = -ENOMEM;
3801 goto out; 3803 goto out;
3802 } 3804 }
3803 3805
3804 /* Populate the fields of the newsk from the oldsk and migrate the 3806 /* Populate the fields of the newsk from the oldsk and migrate the
3805 * asoc to the newsk. 3807 * asoc to the newsk.
3806 */ 3808 */
3807 sctp_sock_migrate(sk, newsk, asoc, SCTP_SOCKET_TCP); 3809 sctp_sock_migrate(sk, newsk, asoc, SCTP_SOCKET_TCP);
3808 3810
3809 out: 3811 out:
3810 sctp_release_sock(sk); 3812 sctp_release_sock(sk);
3811 *err = error; 3813 *err = error;
3812 return newsk; 3814 return newsk;
3813 } 3815 }
3814 3816
3815 /* The SCTP ioctl handler. */ 3817 /* The SCTP ioctl handler. */
3816 SCTP_STATIC int sctp_ioctl(struct sock *sk, int cmd, unsigned long arg) 3818 SCTP_STATIC int sctp_ioctl(struct sock *sk, int cmd, unsigned long arg)
3817 { 3819 {
3818 int rc = -ENOTCONN; 3820 int rc = -ENOTCONN;
3819 3821
3820 sctp_lock_sock(sk); 3822 sctp_lock_sock(sk);
3821 3823
3822 /* 3824 /*
3823 * SEQPACKET-style sockets in LISTENING state are valid, for 3825 * SEQPACKET-style sockets in LISTENING state are valid, for
3824 * SCTP, so only discard TCP-style sockets in LISTENING state. 3826 * SCTP, so only discard TCP-style sockets in LISTENING state.
3825 */ 3827 */
3826 if (sctp_style(sk, TCP) && sctp_sstate(sk, LISTENING)) 3828 if (sctp_style(sk, TCP) && sctp_sstate(sk, LISTENING))
3827 goto out; 3829 goto out;
3828 3830
3829 switch (cmd) { 3831 switch (cmd) {
3830 case SIOCINQ: { 3832 case SIOCINQ: {
3831 struct sk_buff *skb; 3833 struct sk_buff *skb;
3832 unsigned int amount = 0; 3834 unsigned int amount = 0;
3833 3835
3834 skb = skb_peek(&sk->sk_receive_queue); 3836 skb = skb_peek(&sk->sk_receive_queue);
3835 if (skb != NULL) { 3837 if (skb != NULL) {
3836 /* 3838 /*
3837 * We will only return the amount of this packet since 3839 * We will only return the amount of this packet since
3838 * that is all that will be read. 3840 * that is all that will be read.
3839 */ 3841 */
3840 amount = skb->len; 3842 amount = skb->len;
3841 } 3843 }
3842 rc = put_user(amount, (int __user *)arg); 3844 rc = put_user(amount, (int __user *)arg);
3843 break; 3845 break;
3844 } 3846 }
3845 default: 3847 default:
3846 rc = -ENOIOCTLCMD; 3848 rc = -ENOIOCTLCMD;
3847 break; 3849 break;
3848 } 3850 }
3849 out: 3851 out:
3850 sctp_release_sock(sk); 3852 sctp_release_sock(sk);
3851 return rc; 3853 return rc;
3852 } 3854 }
3853 3855
3854 /* This is the function which gets called during socket creation to 3856 /* This is the function which gets called during socket creation to
3855 * initialized the SCTP-specific portion of the sock. 3857 * initialized the SCTP-specific portion of the sock.
3856 * The sock structure should already be zero-filled memory. 3858 * The sock structure should already be zero-filled memory.
3857 */ 3859 */
3858 SCTP_STATIC int sctp_init_sock(struct sock *sk) 3860 SCTP_STATIC int sctp_init_sock(struct sock *sk)
3859 { 3861 {
3860 struct net *net = sock_net(sk); 3862 struct net *net = sock_net(sk);
3861 struct sctp_endpoint *ep; 3863 struct sctp_endpoint *ep;
3862 struct sctp_sock *sp; 3864 struct sctp_sock *sp;
3863 3865
3864 SCTP_DEBUG_PRINTK("sctp_init_sock(sk: %p)\n", sk); 3866 SCTP_DEBUG_PRINTK("sctp_init_sock(sk: %p)\n", sk);
3865 3867
3866 sp = sctp_sk(sk); 3868 sp = sctp_sk(sk);
3867 3869
3868 /* Initialize the SCTP per socket area. */ 3870 /* Initialize the SCTP per socket area. */
3869 switch (sk->sk_type) { 3871 switch (sk->sk_type) {
3870 case SOCK_SEQPACKET: 3872 case SOCK_SEQPACKET:
3871 sp->type = SCTP_SOCKET_UDP; 3873 sp->type = SCTP_SOCKET_UDP;
3872 break; 3874 break;
3873 case SOCK_STREAM: 3875 case SOCK_STREAM:
3874 sp->type = SCTP_SOCKET_TCP; 3876 sp->type = SCTP_SOCKET_TCP;
3875 break; 3877 break;
3876 default: 3878 default:
3877 return -ESOCKTNOSUPPORT; 3879 return -ESOCKTNOSUPPORT;
3878 } 3880 }
3879 3881
3880 /* Initialize default send parameters. These parameters can be 3882 /* Initialize default send parameters. These parameters can be
3881 * modified with the SCTP_DEFAULT_SEND_PARAM socket option. 3883 * modified with the SCTP_DEFAULT_SEND_PARAM socket option.
3882 */ 3884 */
3883 sp->default_stream = 0; 3885 sp->default_stream = 0;
3884 sp->default_ppid = 0; 3886 sp->default_ppid = 0;
3885 sp->default_flags = 0; 3887 sp->default_flags = 0;
3886 sp->default_context = 0; 3888 sp->default_context = 0;
3887 sp->default_timetolive = 0; 3889 sp->default_timetolive = 0;
3888 3890
3889 sp->default_rcv_context = 0; 3891 sp->default_rcv_context = 0;
3890 sp->max_burst = net->sctp.max_burst; 3892 sp->max_burst = net->sctp.max_burst;
3891 3893
3892 sp->sctp_hmac_alg = net->sctp.sctp_hmac_alg; 3894 sp->sctp_hmac_alg = net->sctp.sctp_hmac_alg;
3893 3895
3894 /* Initialize default setup parameters. These parameters 3896 /* Initialize default setup parameters. These parameters
3895 * can be modified with the SCTP_INITMSG socket option or 3897 * can be modified with the SCTP_INITMSG socket option or
3896 * overridden by the SCTP_INIT CMSG. 3898 * overridden by the SCTP_INIT CMSG.
3897 */ 3899 */
3898 sp->initmsg.sinit_num_ostreams = sctp_max_outstreams; 3900 sp->initmsg.sinit_num_ostreams = sctp_max_outstreams;
3899 sp->initmsg.sinit_max_instreams = sctp_max_instreams; 3901 sp->initmsg.sinit_max_instreams = sctp_max_instreams;
3900 sp->initmsg.sinit_max_attempts = net->sctp.max_retrans_init; 3902 sp->initmsg.sinit_max_attempts = net->sctp.max_retrans_init;
3901 sp->initmsg.sinit_max_init_timeo = net->sctp.rto_max; 3903 sp->initmsg.sinit_max_init_timeo = net->sctp.rto_max;
3902 3904
3903 /* Initialize default RTO related parameters. These parameters can 3905 /* Initialize default RTO related parameters. These parameters can
3904 * be modified for with the SCTP_RTOINFO socket option. 3906 * be modified for with the SCTP_RTOINFO socket option.
3905 */ 3907 */
3906 sp->rtoinfo.srto_initial = net->sctp.rto_initial; 3908 sp->rtoinfo.srto_initial = net->sctp.rto_initial;
3907 sp->rtoinfo.srto_max = net->sctp.rto_max; 3909 sp->rtoinfo.srto_max = net->sctp.rto_max;
3908 sp->rtoinfo.srto_min = net->sctp.rto_min; 3910 sp->rtoinfo.srto_min = net->sctp.rto_min;
3909 3911
3910 /* Initialize default association related parameters. These parameters 3912 /* Initialize default association related parameters. These parameters
3911 * can be modified with the SCTP_ASSOCINFO socket option. 3913 * can be modified with the SCTP_ASSOCINFO socket option.
3912 */ 3914 */
3913 sp->assocparams.sasoc_asocmaxrxt = net->sctp.max_retrans_association; 3915 sp->assocparams.sasoc_asocmaxrxt = net->sctp.max_retrans_association;
3914 sp->assocparams.sasoc_number_peer_destinations = 0; 3916 sp->assocparams.sasoc_number_peer_destinations = 0;
3915 sp->assocparams.sasoc_peer_rwnd = 0; 3917 sp->assocparams.sasoc_peer_rwnd = 0;
3916 sp->assocparams.sasoc_local_rwnd = 0; 3918 sp->assocparams.sasoc_local_rwnd = 0;
3917 sp->assocparams.sasoc_cookie_life = net->sctp.valid_cookie_life; 3919 sp->assocparams.sasoc_cookie_life = net->sctp.valid_cookie_life;
3918 3920
3919 /* Initialize default event subscriptions. By default, all the 3921 /* Initialize default event subscriptions. By default, all the
3920 * options are off. 3922 * options are off.
3921 */ 3923 */
3922 memset(&sp->subscribe, 0, sizeof(struct sctp_event_subscribe)); 3924 memset(&sp->subscribe, 0, sizeof(struct sctp_event_subscribe));
3923 3925
3924 /* Default Peer Address Parameters. These defaults can 3926 /* Default Peer Address Parameters. These defaults can
3925 * be modified via SCTP_PEER_ADDR_PARAMS 3927 * be modified via SCTP_PEER_ADDR_PARAMS
3926 */ 3928 */
3927 sp->hbinterval = net->sctp.hb_interval; 3929 sp->hbinterval = net->sctp.hb_interval;
3928 sp->pathmaxrxt = net->sctp.max_retrans_path; 3930 sp->pathmaxrxt = net->sctp.max_retrans_path;
3929 sp->pathmtu = 0; // allow default discovery 3931 sp->pathmtu = 0; // allow default discovery
3930 sp->sackdelay = net->sctp.sack_timeout; 3932 sp->sackdelay = net->sctp.sack_timeout;
3931 sp->sackfreq = 2; 3933 sp->sackfreq = 2;
3932 sp->param_flags = SPP_HB_ENABLE | 3934 sp->param_flags = SPP_HB_ENABLE |
3933 SPP_PMTUD_ENABLE | 3935 SPP_PMTUD_ENABLE |
3934 SPP_SACKDELAY_ENABLE; 3936 SPP_SACKDELAY_ENABLE;
3935 3937
3936 /* If enabled no SCTP message fragmentation will be performed. 3938 /* If enabled no SCTP message fragmentation will be performed.
3937 * Configure through SCTP_DISABLE_FRAGMENTS socket option. 3939 * Configure through SCTP_DISABLE_FRAGMENTS socket option.
3938 */ 3940 */
3939 sp->disable_fragments = 0; 3941 sp->disable_fragments = 0;
3940 3942
3941 /* Enable Nagle algorithm by default. */ 3943 /* Enable Nagle algorithm by default. */
3942 sp->nodelay = 0; 3944 sp->nodelay = 0;
3943 3945
3944 /* Enable by default. */ 3946 /* Enable by default. */
3945 sp->v4mapped = 1; 3947 sp->v4mapped = 1;
3946 3948
3947 /* Auto-close idle associations after the configured 3949 /* Auto-close idle associations after the configured
3948 * number of seconds. A value of 0 disables this 3950 * number of seconds. A value of 0 disables this
3949 * feature. Configure through the SCTP_AUTOCLOSE socket option, 3951 * feature. Configure through the SCTP_AUTOCLOSE socket option,
3950 * for UDP-style sockets only. 3952 * for UDP-style sockets only.
3951 */ 3953 */
3952 sp->autoclose = 0; 3954 sp->autoclose = 0;
3953 3955
3954 /* User specified fragmentation limit. */ 3956 /* User specified fragmentation limit. */
3955 sp->user_frag = 0; 3957 sp->user_frag = 0;
3956 3958
3957 sp->adaptation_ind = 0; 3959 sp->adaptation_ind = 0;
3958 3960
3959 sp->pf = sctp_get_pf_specific(sk->sk_family); 3961 sp->pf = sctp_get_pf_specific(sk->sk_family);
3960 3962
3961 /* Control variables for partial data delivery. */ 3963 /* Control variables for partial data delivery. */
3962 atomic_set(&sp->pd_mode, 0); 3964 atomic_set(&sp->pd_mode, 0);
3963 skb_queue_head_init(&sp->pd_lobby); 3965 skb_queue_head_init(&sp->pd_lobby);
3964 sp->frag_interleave = 0; 3966 sp->frag_interleave = 0;
3965 3967
3966 /* Create a per socket endpoint structure. Even if we 3968 /* Create a per socket endpoint structure. Even if we
3967 * change the data structure relationships, this may still 3969 * change the data structure relationships, this may still
3968 * be useful for storing pre-connect address information. 3970 * be useful for storing pre-connect address information.
3969 */ 3971 */
3970 ep = sctp_endpoint_new(sk, GFP_KERNEL); 3972 ep = sctp_endpoint_new(sk, GFP_KERNEL);
3971 if (!ep) 3973 if (!ep)
3972 return -ENOMEM; 3974 return -ENOMEM;
3973 3975
3974 sp->ep = ep; 3976 sp->ep = ep;
3975 sp->hmac = NULL; 3977 sp->hmac = NULL;
3976 3978
3977 SCTP_DBG_OBJCNT_INC(sock); 3979 SCTP_DBG_OBJCNT_INC(sock);
3978 3980
3979 local_bh_disable(); 3981 local_bh_disable();
3980 percpu_counter_inc(&sctp_sockets_allocated); 3982 percpu_counter_inc(&sctp_sockets_allocated);
3981 sock_prot_inuse_add(net, sk->sk_prot, 1); 3983 sock_prot_inuse_add(net, sk->sk_prot, 1);
3982 if (net->sctp.default_auto_asconf) { 3984 if (net->sctp.default_auto_asconf) {
3983 list_add_tail(&sp->auto_asconf_list, 3985 list_add_tail(&sp->auto_asconf_list,
3984 &net->sctp.auto_asconf_splist); 3986 &net->sctp.auto_asconf_splist);
3985 sp->do_auto_asconf = 1; 3987 sp->do_auto_asconf = 1;
3986 } else 3988 } else
3987 sp->do_auto_asconf = 0; 3989 sp->do_auto_asconf = 0;
3988 local_bh_enable(); 3990 local_bh_enable();
3989 3991
3990 return 0; 3992 return 0;
3991 } 3993 }
3992 3994
3993 /* Cleanup any SCTP per socket resources. */ 3995 /* Cleanup any SCTP per socket resources. */
3994 SCTP_STATIC void sctp_destroy_sock(struct sock *sk) 3996 SCTP_STATIC void sctp_destroy_sock(struct sock *sk)
3995 { 3997 {
3996 struct sctp_sock *sp; 3998 struct sctp_sock *sp;
3997 3999
3998 SCTP_DEBUG_PRINTK("sctp_destroy_sock(sk: %p)\n", sk); 4000 SCTP_DEBUG_PRINTK("sctp_destroy_sock(sk: %p)\n", sk);
3999 4001
4000 /* Release our hold on the endpoint. */ 4002 /* Release our hold on the endpoint. */
4001 sp = sctp_sk(sk); 4003 sp = sctp_sk(sk);
4002 if (sp->do_auto_asconf) { 4004 if (sp->do_auto_asconf) {
4003 sp->do_auto_asconf = 0; 4005 sp->do_auto_asconf = 0;
4004 list_del(&sp->auto_asconf_list); 4006 list_del(&sp->auto_asconf_list);
4005 } 4007 }
4006 sctp_endpoint_free(sp->ep); 4008 sctp_endpoint_free(sp->ep);
4007 local_bh_disable(); 4009 local_bh_disable();
4008 percpu_counter_dec(&sctp_sockets_allocated); 4010 percpu_counter_dec(&sctp_sockets_allocated);
4009 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1); 4011 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
4010 local_bh_enable(); 4012 local_bh_enable();
4011 } 4013 }
4012 4014
4013 /* API 4.1.7 shutdown() - TCP Style Syntax 4015 /* API 4.1.7 shutdown() - TCP Style Syntax
4014 * int shutdown(int socket, int how); 4016 * int shutdown(int socket, int how);
4015 * 4017 *
4016 * sd - the socket descriptor of the association to be closed. 4018 * sd - the socket descriptor of the association to be closed.
4017 * how - Specifies the type of shutdown. The values are 4019 * how - Specifies the type of shutdown. The values are
4018 * as follows: 4020 * as follows:
4019 * SHUT_RD 4021 * SHUT_RD
4020 * Disables further receive operations. No SCTP 4022 * Disables further receive operations. No SCTP
4021 * protocol action is taken. 4023 * protocol action is taken.
4022 * SHUT_WR 4024 * SHUT_WR
4023 * Disables further send operations, and initiates 4025 * Disables further send operations, and initiates
4024 * the SCTP shutdown sequence. 4026 * the SCTP shutdown sequence.
4025 * SHUT_RDWR 4027 * SHUT_RDWR
4026 * Disables further send and receive operations 4028 * Disables further send and receive operations
4027 * and initiates the SCTP shutdown sequence. 4029 * and initiates the SCTP shutdown sequence.
4028 */ 4030 */
4029 SCTP_STATIC void sctp_shutdown(struct sock *sk, int how) 4031 SCTP_STATIC void sctp_shutdown(struct sock *sk, int how)
4030 { 4032 {
4031 struct net *net = sock_net(sk); 4033 struct net *net = sock_net(sk);
4032 struct sctp_endpoint *ep; 4034 struct sctp_endpoint *ep;
4033 struct sctp_association *asoc; 4035 struct sctp_association *asoc;
4034 4036
4035 if (!sctp_style(sk, TCP)) 4037 if (!sctp_style(sk, TCP))
4036 return; 4038 return;
4037 4039
4038 if (how & SEND_SHUTDOWN) { 4040 if (how & SEND_SHUTDOWN) {
4039 ep = sctp_sk(sk)->ep; 4041 ep = sctp_sk(sk)->ep;
4040 if (!list_empty(&ep->asocs)) { 4042 if (!list_empty(&ep->asocs)) {
4041 asoc = list_entry(ep->asocs.next, 4043 asoc = list_entry(ep->asocs.next,
4042 struct sctp_association, asocs); 4044 struct sctp_association, asocs);
4043 sctp_primitive_SHUTDOWN(net, asoc, NULL); 4045 sctp_primitive_SHUTDOWN(net, asoc, NULL);
4044 } 4046 }
4045 } 4047 }
4046 } 4048 }
4047 4049
4048 /* 7.2.1 Association Status (SCTP_STATUS) 4050 /* 7.2.1 Association Status (SCTP_STATUS)
4049 4051
4050 * Applications can retrieve current status information about an 4052 * Applications can retrieve current status information about an
4051 * association, including association state, peer receiver window size, 4053 * association, including association state, peer receiver window size,
4052 * number of unacked data chunks, and number of data chunks pending 4054 * number of unacked data chunks, and number of data chunks pending
4053 * receipt. This information is read-only. 4055 * receipt. This information is read-only.
4054 */ 4056 */
4055 static int sctp_getsockopt_sctp_status(struct sock *sk, int len, 4057 static int sctp_getsockopt_sctp_status(struct sock *sk, int len,
4056 char __user *optval, 4058 char __user *optval,
4057 int __user *optlen) 4059 int __user *optlen)
4058 { 4060 {
4059 struct sctp_status status; 4061 struct sctp_status status;
4060 struct sctp_association *asoc = NULL; 4062 struct sctp_association *asoc = NULL;
4061 struct sctp_transport *transport; 4063 struct sctp_transport *transport;
4062 sctp_assoc_t associd; 4064 sctp_assoc_t associd;
4063 int retval = 0; 4065 int retval = 0;
4064 4066
4065 if (len < sizeof(status)) { 4067 if (len < sizeof(status)) {
4066 retval = -EINVAL; 4068 retval = -EINVAL;
4067 goto out; 4069 goto out;
4068 } 4070 }
4069 4071
4070 len = sizeof(status); 4072 len = sizeof(status);
4071 if (copy_from_user(&status, optval, len)) { 4073 if (copy_from_user(&status, optval, len)) {
4072 retval = -EFAULT; 4074 retval = -EFAULT;
4073 goto out; 4075 goto out;
4074 } 4076 }
4075 4077
4076 associd = status.sstat_assoc_id; 4078 associd = status.sstat_assoc_id;
4077 asoc = sctp_id2assoc(sk, associd); 4079 asoc = sctp_id2assoc(sk, associd);
4078 if (!asoc) { 4080 if (!asoc) {
4079 retval = -EINVAL; 4081 retval = -EINVAL;
4080 goto out; 4082 goto out;
4081 } 4083 }
4082 4084
4083 transport = asoc->peer.primary_path; 4085 transport = asoc->peer.primary_path;
4084 4086
4085 status.sstat_assoc_id = sctp_assoc2id(asoc); 4087 status.sstat_assoc_id = sctp_assoc2id(asoc);
4086 status.sstat_state = asoc->state; 4088 status.sstat_state = asoc->state;
4087 status.sstat_rwnd = asoc->peer.rwnd; 4089 status.sstat_rwnd = asoc->peer.rwnd;
4088 status.sstat_unackdata = asoc->unack_data; 4090 status.sstat_unackdata = asoc->unack_data;
4089 4091
4090 status.sstat_penddata = sctp_tsnmap_pending(&asoc->peer.tsn_map); 4092 status.sstat_penddata = sctp_tsnmap_pending(&asoc->peer.tsn_map);
4091 status.sstat_instrms = asoc->c.sinit_max_instreams; 4093 status.sstat_instrms = asoc->c.sinit_max_instreams;
4092 status.sstat_outstrms = asoc->c.sinit_num_ostreams; 4094 status.sstat_outstrms = asoc->c.sinit_num_ostreams;
4093 status.sstat_fragmentation_point = asoc->frag_point; 4095 status.sstat_fragmentation_point = asoc->frag_point;
4094 status.sstat_primary.spinfo_assoc_id = sctp_assoc2id(transport->asoc); 4096 status.sstat_primary.spinfo_assoc_id = sctp_assoc2id(transport->asoc);
4095 memcpy(&status.sstat_primary.spinfo_address, &transport->ipaddr, 4097 memcpy(&status.sstat_primary.spinfo_address, &transport->ipaddr,
4096 transport->af_specific->sockaddr_len); 4098 transport->af_specific->sockaddr_len);
4097 /* Map ipv4 address into v4-mapped-on-v6 address. */ 4099 /* Map ipv4 address into v4-mapped-on-v6 address. */
4098 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sctp_sk(sk), 4100 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sctp_sk(sk),
4099 (union sctp_addr *)&status.sstat_primary.spinfo_address); 4101 (union sctp_addr *)&status.sstat_primary.spinfo_address);
4100 status.sstat_primary.spinfo_state = transport->state; 4102 status.sstat_primary.spinfo_state = transport->state;
4101 status.sstat_primary.spinfo_cwnd = transport->cwnd; 4103 status.sstat_primary.spinfo_cwnd = transport->cwnd;
4102 status.sstat_primary.spinfo_srtt = transport->srtt; 4104 status.sstat_primary.spinfo_srtt = transport->srtt;
4103 status.sstat_primary.spinfo_rto = jiffies_to_msecs(transport->rto); 4105 status.sstat_primary.spinfo_rto = jiffies_to_msecs(transport->rto);
4104 status.sstat_primary.spinfo_mtu = transport->pathmtu; 4106 status.sstat_primary.spinfo_mtu = transport->pathmtu;
4105 4107
4106 if (status.sstat_primary.spinfo_state == SCTP_UNKNOWN) 4108 if (status.sstat_primary.spinfo_state == SCTP_UNKNOWN)
4107 status.sstat_primary.spinfo_state = SCTP_ACTIVE; 4109 status.sstat_primary.spinfo_state = SCTP_ACTIVE;
4108 4110
4109 if (put_user(len, optlen)) { 4111 if (put_user(len, optlen)) {
4110 retval = -EFAULT; 4112 retval = -EFAULT;
4111 goto out; 4113 goto out;
4112 } 4114 }
4113 4115
4114 SCTP_DEBUG_PRINTK("sctp_getsockopt_sctp_status(%d): %d %d %d\n", 4116 SCTP_DEBUG_PRINTK("sctp_getsockopt_sctp_status(%d): %d %d %d\n",
4115 len, status.sstat_state, status.sstat_rwnd, 4117 len, status.sstat_state, status.sstat_rwnd,
4116 status.sstat_assoc_id); 4118 status.sstat_assoc_id);
4117 4119
4118 if (copy_to_user(optval, &status, len)) { 4120 if (copy_to_user(optval, &status, len)) {
4119 retval = -EFAULT; 4121 retval = -EFAULT;
4120 goto out; 4122 goto out;
4121 } 4123 }
4122 4124
4123 out: 4125 out:
4124 return retval; 4126 return retval;
4125 } 4127 }
4126 4128
4127 4129
4128 /* 7.2.2 Peer Address Information (SCTP_GET_PEER_ADDR_INFO) 4130 /* 7.2.2 Peer Address Information (SCTP_GET_PEER_ADDR_INFO)
4129 * 4131 *
4130 * Applications can retrieve information about a specific peer address 4132 * Applications can retrieve information about a specific peer address
4131 * of an association, including its reachability state, congestion 4133 * of an association, including its reachability state, congestion
4132 * window, and retransmission timer values. This information is 4134 * window, and retransmission timer values. This information is
4133 * read-only. 4135 * read-only.
4134 */ 4136 */
4135 static int sctp_getsockopt_peer_addr_info(struct sock *sk, int len, 4137 static int sctp_getsockopt_peer_addr_info(struct sock *sk, int len,
4136 char __user *optval, 4138 char __user *optval,
4137 int __user *optlen) 4139 int __user *optlen)
4138 { 4140 {
4139 struct sctp_paddrinfo pinfo; 4141 struct sctp_paddrinfo pinfo;
4140 struct sctp_transport *transport; 4142 struct sctp_transport *transport;
4141 int retval = 0; 4143 int retval = 0;
4142 4144
4143 if (len < sizeof(pinfo)) { 4145 if (len < sizeof(pinfo)) {
4144 retval = -EINVAL; 4146 retval = -EINVAL;
4145 goto out; 4147 goto out;
4146 } 4148 }
4147 4149
4148 len = sizeof(pinfo); 4150 len = sizeof(pinfo);
4149 if (copy_from_user(&pinfo, optval, len)) { 4151 if (copy_from_user(&pinfo, optval, len)) {
4150 retval = -EFAULT; 4152 retval = -EFAULT;
4151 goto out; 4153 goto out;
4152 } 4154 }
4153 4155
4154 transport = sctp_addr_id2transport(sk, &pinfo.spinfo_address, 4156 transport = sctp_addr_id2transport(sk, &pinfo.spinfo_address,
4155 pinfo.spinfo_assoc_id); 4157 pinfo.spinfo_assoc_id);
4156 if (!transport) 4158 if (!transport)
4157 return -EINVAL; 4159 return -EINVAL;
4158 4160
4159 pinfo.spinfo_assoc_id = sctp_assoc2id(transport->asoc); 4161 pinfo.spinfo_assoc_id = sctp_assoc2id(transport->asoc);
4160 pinfo.spinfo_state = transport->state; 4162 pinfo.spinfo_state = transport->state;
4161 pinfo.spinfo_cwnd = transport->cwnd; 4163 pinfo.spinfo_cwnd = transport->cwnd;
4162 pinfo.spinfo_srtt = transport->srtt; 4164 pinfo.spinfo_srtt = transport->srtt;
4163 pinfo.spinfo_rto = jiffies_to_msecs(transport->rto); 4165 pinfo.spinfo_rto = jiffies_to_msecs(transport->rto);
4164 pinfo.spinfo_mtu = transport->pathmtu; 4166 pinfo.spinfo_mtu = transport->pathmtu;
4165 4167
4166 if (pinfo.spinfo_state == SCTP_UNKNOWN) 4168 if (pinfo.spinfo_state == SCTP_UNKNOWN)
4167 pinfo.spinfo_state = SCTP_ACTIVE; 4169 pinfo.spinfo_state = SCTP_ACTIVE;
4168 4170
4169 if (put_user(len, optlen)) { 4171 if (put_user(len, optlen)) {
4170 retval = -EFAULT; 4172 retval = -EFAULT;
4171 goto out; 4173 goto out;
4172 } 4174 }
4173 4175
4174 if (copy_to_user(optval, &pinfo, len)) { 4176 if (copy_to_user(optval, &pinfo, len)) {
4175 retval = -EFAULT; 4177 retval = -EFAULT;
4176 goto out; 4178 goto out;
4177 } 4179 }
4178 4180
4179 out: 4181 out:
4180 return retval; 4182 return retval;
4181 } 4183 }
4182 4184
4183 /* 7.1.12 Enable/Disable message fragmentation (SCTP_DISABLE_FRAGMENTS) 4185 /* 7.1.12 Enable/Disable message fragmentation (SCTP_DISABLE_FRAGMENTS)
4184 * 4186 *
4185 * This option is a on/off flag. If enabled no SCTP message 4187 * This option is a on/off flag. If enabled no SCTP message
4186 * fragmentation will be performed. Instead if a message being sent 4188 * fragmentation will be performed. Instead if a message being sent
4187 * exceeds the current PMTU size, the message will NOT be sent and 4189 * exceeds the current PMTU size, the message will NOT be sent and
4188 * instead a error will be indicated to the user. 4190 * instead a error will be indicated to the user.
4189 */ 4191 */
4190 static int sctp_getsockopt_disable_fragments(struct sock *sk, int len, 4192 static int sctp_getsockopt_disable_fragments(struct sock *sk, int len,
4191 char __user *optval, int __user *optlen) 4193 char __user *optval, int __user *optlen)
4192 { 4194 {
4193 int val; 4195 int val;
4194 4196
4195 if (len < sizeof(int)) 4197 if (len < sizeof(int))
4196 return -EINVAL; 4198 return -EINVAL;
4197 4199
4198 len = sizeof(int); 4200 len = sizeof(int);
4199 val = (sctp_sk(sk)->disable_fragments == 1); 4201 val = (sctp_sk(sk)->disable_fragments == 1);
4200 if (put_user(len, optlen)) 4202 if (put_user(len, optlen))
4201 return -EFAULT; 4203 return -EFAULT;
4202 if (copy_to_user(optval, &val, len)) 4204 if (copy_to_user(optval, &val, len))
4203 return -EFAULT; 4205 return -EFAULT;
4204 return 0; 4206 return 0;
4205 } 4207 }
4206 4208
4207 /* 7.1.15 Set notification and ancillary events (SCTP_EVENTS) 4209 /* 7.1.15 Set notification and ancillary events (SCTP_EVENTS)
4208 * 4210 *
4209 * This socket option is used to specify various notifications and 4211 * This socket option is used to specify various notifications and
4210 * ancillary data the user wishes to receive. 4212 * ancillary data the user wishes to receive.
4211 */ 4213 */
4212 static int sctp_getsockopt_events(struct sock *sk, int len, char __user *optval, 4214 static int sctp_getsockopt_events(struct sock *sk, int len, char __user *optval,
4213 int __user *optlen) 4215 int __user *optlen)
4214 { 4216 {
4215 if (len <= 0) 4217 if (len <= 0)
4216 return -EINVAL; 4218 return -EINVAL;
4217 if (len > sizeof(struct sctp_event_subscribe)) 4219 if (len > sizeof(struct sctp_event_subscribe))
4218 len = sizeof(struct sctp_event_subscribe); 4220 len = sizeof(struct sctp_event_subscribe);
4219 if (put_user(len, optlen)) 4221 if (put_user(len, optlen))
4220 return -EFAULT; 4222 return -EFAULT;
4221 if (copy_to_user(optval, &sctp_sk(sk)->subscribe, len)) 4223 if (copy_to_user(optval, &sctp_sk(sk)->subscribe, len))
4222 return -EFAULT; 4224 return -EFAULT;
4223 return 0; 4225 return 0;
4224 } 4226 }
4225 4227
4226 /* 7.1.8 Automatic Close of associations (SCTP_AUTOCLOSE) 4228 /* 7.1.8 Automatic Close of associations (SCTP_AUTOCLOSE)
4227 * 4229 *
4228 * This socket option is applicable to the UDP-style socket only. When 4230 * This socket option is applicable to the UDP-style socket only. When
4229 * set it will cause associations that are idle for more than the 4231 * set it will cause associations that are idle for more than the
4230 * specified number of seconds to automatically close. An association 4232 * specified number of seconds to automatically close. An association
4231 * being idle is defined an association that has NOT sent or received 4233 * being idle is defined an association that has NOT sent or received
4232 * user data. The special value of '0' indicates that no automatic 4234 * user data. The special value of '0' indicates that no automatic
4233 * close of any associations should be performed. The option expects an 4235 * close of any associations should be performed. The option expects an
4234 * integer defining the number of seconds of idle time before an 4236 * integer defining the number of seconds of idle time before an
4235 * association is closed. 4237 * association is closed.
4236 */ 4238 */
4237 static int sctp_getsockopt_autoclose(struct sock *sk, int len, char __user *optval, int __user *optlen) 4239 static int sctp_getsockopt_autoclose(struct sock *sk, int len, char __user *optval, int __user *optlen)
4238 { 4240 {
4239 /* Applicable to UDP-style socket only */ 4241 /* Applicable to UDP-style socket only */
4240 if (sctp_style(sk, TCP)) 4242 if (sctp_style(sk, TCP))
4241 return -EOPNOTSUPP; 4243 return -EOPNOTSUPP;
4242 if (len < sizeof(int)) 4244 if (len < sizeof(int))
4243 return -EINVAL; 4245 return -EINVAL;
4244 len = sizeof(int); 4246 len = sizeof(int);
4245 if (put_user(len, optlen)) 4247 if (put_user(len, optlen))
4246 return -EFAULT; 4248 return -EFAULT;
4247 if (copy_to_user(optval, &sctp_sk(sk)->autoclose, sizeof(int))) 4249 if (copy_to_user(optval, &sctp_sk(sk)->autoclose, sizeof(int)))
4248 return -EFAULT; 4250 return -EFAULT;
4249 return 0; 4251 return 0;
4250 } 4252 }
4251 4253
4252 /* Helper routine to branch off an association to a new socket. */ 4254 /* Helper routine to branch off an association to a new socket. */
4253 int sctp_do_peeloff(struct sock *sk, sctp_assoc_t id, struct socket **sockp) 4255 int sctp_do_peeloff(struct sock *sk, sctp_assoc_t id, struct socket **sockp)
4254 { 4256 {
4255 struct sctp_association *asoc = sctp_id2assoc(sk, id); 4257 struct sctp_association *asoc = sctp_id2assoc(sk, id);
4256 struct socket *sock; 4258 struct socket *sock;
4257 struct sctp_af *af; 4259 struct sctp_af *af;
4258 int err = 0; 4260 int err = 0;
4259 4261
4260 if (!asoc) 4262 if (!asoc)
4261 return -EINVAL; 4263 return -EINVAL;
4262 4264
4263 /* An association cannot be branched off from an already peeled-off 4265 /* An association cannot be branched off from an already peeled-off
4264 * socket, nor is this supported for tcp style sockets. 4266 * socket, nor is this supported for tcp style sockets.
4265 */ 4267 */
4266 if (!sctp_style(sk, UDP)) 4268 if (!sctp_style(sk, UDP))
4267 return -EINVAL; 4269 return -EINVAL;
4268 4270
4269 /* Create a new socket. */ 4271 /* Create a new socket. */
4270 err = sock_create(sk->sk_family, SOCK_SEQPACKET, IPPROTO_SCTP, &sock); 4272 err = sock_create(sk->sk_family, SOCK_SEQPACKET, IPPROTO_SCTP, &sock);
4271 if (err < 0) 4273 if (err < 0)
4272 return err; 4274 return err;
4273 4275
4274 sctp_copy_sock(sock->sk, sk, asoc); 4276 sctp_copy_sock(sock->sk, sk, asoc);
4275 4277
4276 /* Make peeled-off sockets more like 1-1 accepted sockets. 4278 /* Make peeled-off sockets more like 1-1 accepted sockets.
4277 * Set the daddr and initialize id to something more random 4279 * Set the daddr and initialize id to something more random
4278 */ 4280 */
4279 af = sctp_get_af_specific(asoc->peer.primary_addr.sa.sa_family); 4281 af = sctp_get_af_specific(asoc->peer.primary_addr.sa.sa_family);
4280 af->to_sk_daddr(&asoc->peer.primary_addr, sk); 4282 af->to_sk_daddr(&asoc->peer.primary_addr, sk);
4281 4283
4282 /* Populate the fields of the newsk from the oldsk and migrate the 4284 /* Populate the fields of the newsk from the oldsk and migrate the
4283 * asoc to the newsk. 4285 * asoc to the newsk.
4284 */ 4286 */
4285 sctp_sock_migrate(sk, sock->sk, asoc, SCTP_SOCKET_UDP_HIGH_BANDWIDTH); 4287 sctp_sock_migrate(sk, sock->sk, asoc, SCTP_SOCKET_UDP_HIGH_BANDWIDTH);
4286 4288
4287 *sockp = sock; 4289 *sockp = sock;
4288 4290
4289 return err; 4291 return err;
4290 } 4292 }
4291 EXPORT_SYMBOL(sctp_do_peeloff); 4293 EXPORT_SYMBOL(sctp_do_peeloff);
4292 4294
4293 static int sctp_getsockopt_peeloff(struct sock *sk, int len, char __user *optval, int __user *optlen) 4295 static int sctp_getsockopt_peeloff(struct sock *sk, int len, char __user *optval, int __user *optlen)
4294 { 4296 {
4295 sctp_peeloff_arg_t peeloff; 4297 sctp_peeloff_arg_t peeloff;
4296 struct socket *newsock; 4298 struct socket *newsock;
4297 struct file *newfile; 4299 struct file *newfile;
4298 int retval = 0; 4300 int retval = 0;
4299 4301
4300 if (len < sizeof(sctp_peeloff_arg_t)) 4302 if (len < sizeof(sctp_peeloff_arg_t))
4301 return -EINVAL; 4303 return -EINVAL;
4302 len = sizeof(sctp_peeloff_arg_t); 4304 len = sizeof(sctp_peeloff_arg_t);
4303 if (copy_from_user(&peeloff, optval, len)) 4305 if (copy_from_user(&peeloff, optval, len))
4304 return -EFAULT; 4306 return -EFAULT;
4305 4307
4306 retval = sctp_do_peeloff(sk, peeloff.associd, &newsock); 4308 retval = sctp_do_peeloff(sk, peeloff.associd, &newsock);
4307 if (retval < 0) 4309 if (retval < 0)
4308 goto out; 4310 goto out;
4309 4311
4310 /* Map the socket to an unused fd that can be returned to the user. */ 4312 /* Map the socket to an unused fd that can be returned to the user. */
4311 retval = get_unused_fd(); 4313 retval = get_unused_fd();
4312 if (retval < 0) { 4314 if (retval < 0) {
4313 sock_release(newsock); 4315 sock_release(newsock);
4314 goto out; 4316 goto out;
4315 } 4317 }
4316 4318
4317 newfile = sock_alloc_file(newsock, 0, NULL); 4319 newfile = sock_alloc_file(newsock, 0, NULL);
4318 if (unlikely(IS_ERR(newfile))) { 4320 if (unlikely(IS_ERR(newfile))) {
4319 put_unused_fd(retval); 4321 put_unused_fd(retval);
4320 sock_release(newsock); 4322 sock_release(newsock);
4321 return PTR_ERR(newfile); 4323 return PTR_ERR(newfile);
4322 } 4324 }
4323 4325
4324 SCTP_DEBUG_PRINTK("%s: sk: %p newsk: %p sd: %d\n", 4326 SCTP_DEBUG_PRINTK("%s: sk: %p newsk: %p sd: %d\n",
4325 __func__, sk, newsock->sk, retval); 4327 __func__, sk, newsock->sk, retval);
4326 4328
4327 /* Return the fd mapped to the new socket. */ 4329 /* Return the fd mapped to the new socket. */
4328 if (put_user(len, optlen)) { 4330 if (put_user(len, optlen)) {
4329 fput(newfile); 4331 fput(newfile);
4330 put_unused_fd(retval); 4332 put_unused_fd(retval);
4331 return -EFAULT; 4333 return -EFAULT;
4332 } 4334 }
4333 peeloff.sd = retval; 4335 peeloff.sd = retval;
4334 if (copy_to_user(optval, &peeloff, len)) { 4336 if (copy_to_user(optval, &peeloff, len)) {
4335 fput(newfile); 4337 fput(newfile);
4336 put_unused_fd(retval); 4338 put_unused_fd(retval);
4337 return -EFAULT; 4339 return -EFAULT;
4338 } 4340 }
4339 fd_install(retval, newfile); 4341 fd_install(retval, newfile);
4340 out: 4342 out:
4341 return retval; 4343 return retval;
4342 } 4344 }
4343 4345
4344 /* 7.1.13 Peer Address Parameters (SCTP_PEER_ADDR_PARAMS) 4346 /* 7.1.13 Peer Address Parameters (SCTP_PEER_ADDR_PARAMS)
4345 * 4347 *
4346 * Applications can enable or disable heartbeats for any peer address of 4348 * Applications can enable or disable heartbeats for any peer address of
4347 * an association, modify an address's heartbeat interval, force a 4349 * an association, modify an address's heartbeat interval, force a
4348 * heartbeat to be sent immediately, and adjust the address's maximum 4350 * heartbeat to be sent immediately, and adjust the address's maximum
4349 * number of retransmissions sent before an address is considered 4351 * number of retransmissions sent before an address is considered
4350 * unreachable. The following structure is used to access and modify an 4352 * unreachable. The following structure is used to access and modify an
4351 * address's parameters: 4353 * address's parameters:
4352 * 4354 *
4353 * struct sctp_paddrparams { 4355 * struct sctp_paddrparams {
4354 * sctp_assoc_t spp_assoc_id; 4356 * sctp_assoc_t spp_assoc_id;
4355 * struct sockaddr_storage spp_address; 4357 * struct sockaddr_storage spp_address;
4356 * uint32_t spp_hbinterval; 4358 * uint32_t spp_hbinterval;
4357 * uint16_t spp_pathmaxrxt; 4359 * uint16_t spp_pathmaxrxt;
4358 * uint32_t spp_pathmtu; 4360 * uint32_t spp_pathmtu;
4359 * uint32_t spp_sackdelay; 4361 * uint32_t spp_sackdelay;
4360 * uint32_t spp_flags; 4362 * uint32_t spp_flags;
4361 * }; 4363 * };
4362 * 4364 *
4363 * spp_assoc_id - (one-to-many style socket) This is filled in the 4365 * spp_assoc_id - (one-to-many style socket) This is filled in the
4364 * application, and identifies the association for 4366 * application, and identifies the association for
4365 * this query. 4367 * this query.
4366 * spp_address - This specifies which address is of interest. 4368 * spp_address - This specifies which address is of interest.
4367 * spp_hbinterval - This contains the value of the heartbeat interval, 4369 * spp_hbinterval - This contains the value of the heartbeat interval,
4368 * in milliseconds. If a value of zero 4370 * in milliseconds. If a value of zero
4369 * is present in this field then no changes are to 4371 * is present in this field then no changes are to
4370 * be made to this parameter. 4372 * be made to this parameter.
4371 * spp_pathmaxrxt - This contains the maximum number of 4373 * spp_pathmaxrxt - This contains the maximum number of
4372 * retransmissions before this address shall be 4374 * retransmissions before this address shall be
4373 * considered unreachable. If a value of zero 4375 * considered unreachable. If a value of zero
4374 * is present in this field then no changes are to 4376 * is present in this field then no changes are to
4375 * be made to this parameter. 4377 * be made to this parameter.
4376 * spp_pathmtu - When Path MTU discovery is disabled the value 4378 * spp_pathmtu - When Path MTU discovery is disabled the value
4377 * specified here will be the "fixed" path mtu. 4379 * specified here will be the "fixed" path mtu.
4378 * Note that if the spp_address field is empty 4380 * Note that if the spp_address field is empty
4379 * then all associations on this address will 4381 * then all associations on this address will
4380 * have this fixed path mtu set upon them. 4382 * have this fixed path mtu set upon them.
4381 * 4383 *
4382 * spp_sackdelay - When delayed sack is enabled, this value specifies 4384 * spp_sackdelay - When delayed sack is enabled, this value specifies
4383 * the number of milliseconds that sacks will be delayed 4385 * the number of milliseconds that sacks will be delayed
4384 * for. This value will apply to all addresses of an 4386 * for. This value will apply to all addresses of an
4385 * association if the spp_address field is empty. Note 4387 * association if the spp_address field is empty. Note
4386 * also, that if delayed sack is enabled and this 4388 * also, that if delayed sack is enabled and this
4387 * value is set to 0, no change is made to the last 4389 * value is set to 0, no change is made to the last
4388 * recorded delayed sack timer value. 4390 * recorded delayed sack timer value.
4389 * 4391 *
4390 * spp_flags - These flags are used to control various features 4392 * spp_flags - These flags are used to control various features
4391 * on an association. The flag field may contain 4393 * on an association. The flag field may contain
4392 * zero or more of the following options. 4394 * zero or more of the following options.
4393 * 4395 *
4394 * SPP_HB_ENABLE - Enable heartbeats on the 4396 * SPP_HB_ENABLE - Enable heartbeats on the
4395 * specified address. Note that if the address 4397 * specified address. Note that if the address
4396 * field is empty all addresses for the association 4398 * field is empty all addresses for the association
4397 * have heartbeats enabled upon them. 4399 * have heartbeats enabled upon them.
4398 * 4400 *
4399 * SPP_HB_DISABLE - Disable heartbeats on the 4401 * SPP_HB_DISABLE - Disable heartbeats on the
4400 * speicifed address. Note that if the address 4402 * speicifed address. Note that if the address
4401 * field is empty all addresses for the association 4403 * field is empty all addresses for the association
4402 * will have their heartbeats disabled. Note also 4404 * will have their heartbeats disabled. Note also
4403 * that SPP_HB_ENABLE and SPP_HB_DISABLE are 4405 * that SPP_HB_ENABLE and SPP_HB_DISABLE are
4404 * mutually exclusive, only one of these two should 4406 * mutually exclusive, only one of these two should
4405 * be specified. Enabling both fields will have 4407 * be specified. Enabling both fields will have
4406 * undetermined results. 4408 * undetermined results.
4407 * 4409 *
4408 * SPP_HB_DEMAND - Request a user initiated heartbeat 4410 * SPP_HB_DEMAND - Request a user initiated heartbeat
4409 * to be made immediately. 4411 * to be made immediately.
4410 * 4412 *
4411 * SPP_PMTUD_ENABLE - This field will enable PMTU 4413 * SPP_PMTUD_ENABLE - This field will enable PMTU
4412 * discovery upon the specified address. Note that 4414 * discovery upon the specified address. Note that
4413 * if the address feild is empty then all addresses 4415 * if the address feild is empty then all addresses
4414 * on the association are effected. 4416 * on the association are effected.
4415 * 4417 *
4416 * SPP_PMTUD_DISABLE - This field will disable PMTU 4418 * SPP_PMTUD_DISABLE - This field will disable PMTU
4417 * discovery upon the specified address. Note that 4419 * discovery upon the specified address. Note that
4418 * if the address feild is empty then all addresses 4420 * if the address feild is empty then all addresses
4419 * on the association are effected. Not also that 4421 * on the association are effected. Not also that
4420 * SPP_PMTUD_ENABLE and SPP_PMTUD_DISABLE are mutually 4422 * SPP_PMTUD_ENABLE and SPP_PMTUD_DISABLE are mutually
4421 * exclusive. Enabling both will have undetermined 4423 * exclusive. Enabling both will have undetermined
4422 * results. 4424 * results.
4423 * 4425 *
4424 * SPP_SACKDELAY_ENABLE - Setting this flag turns 4426 * SPP_SACKDELAY_ENABLE - Setting this flag turns
4425 * on delayed sack. The time specified in spp_sackdelay 4427 * on delayed sack. The time specified in spp_sackdelay
4426 * is used to specify the sack delay for this address. Note 4428 * is used to specify the sack delay for this address. Note
4427 * that if spp_address is empty then all addresses will 4429 * that if spp_address is empty then all addresses will
4428 * enable delayed sack and take on the sack delay 4430 * enable delayed sack and take on the sack delay
4429 * value specified in spp_sackdelay. 4431 * value specified in spp_sackdelay.
4430 * SPP_SACKDELAY_DISABLE - Setting this flag turns 4432 * SPP_SACKDELAY_DISABLE - Setting this flag turns
4431 * off delayed sack. If the spp_address field is blank then 4433 * off delayed sack. If the spp_address field is blank then
4432 * delayed sack is disabled for the entire association. Note 4434 * delayed sack is disabled for the entire association. Note
4433 * also that this field is mutually exclusive to 4435 * also that this field is mutually exclusive to
4434 * SPP_SACKDELAY_ENABLE, setting both will have undefined 4436 * SPP_SACKDELAY_ENABLE, setting both will have undefined
4435 * results. 4437 * results.
4436 */ 4438 */
4437 static int sctp_getsockopt_peer_addr_params(struct sock *sk, int len, 4439 static int sctp_getsockopt_peer_addr_params(struct sock *sk, int len,
4438 char __user *optval, int __user *optlen) 4440 char __user *optval, int __user *optlen)
4439 { 4441 {
4440 struct sctp_paddrparams params; 4442 struct sctp_paddrparams params;
4441 struct sctp_transport *trans = NULL; 4443 struct sctp_transport *trans = NULL;
4442 struct sctp_association *asoc = NULL; 4444 struct sctp_association *asoc = NULL;
4443 struct sctp_sock *sp = sctp_sk(sk); 4445 struct sctp_sock *sp = sctp_sk(sk);
4444 4446
4445 if (len < sizeof(struct sctp_paddrparams)) 4447 if (len < sizeof(struct sctp_paddrparams))
4446 return -EINVAL; 4448 return -EINVAL;
4447 len = sizeof(struct sctp_paddrparams); 4449 len = sizeof(struct sctp_paddrparams);
4448 if (copy_from_user(&params, optval, len)) 4450 if (copy_from_user(&params, optval, len))
4449 return -EFAULT; 4451 return -EFAULT;
4450 4452
4451 /* If an address other than INADDR_ANY is specified, and 4453 /* If an address other than INADDR_ANY is specified, and
4452 * no transport is found, then the request is invalid. 4454 * no transport is found, then the request is invalid.
4453 */ 4455 */
4454 if (!sctp_is_any(sk, ( union sctp_addr *)&params.spp_address)) { 4456 if (!sctp_is_any(sk, ( union sctp_addr *)&params.spp_address)) {
4455 trans = sctp_addr_id2transport(sk, &params.spp_address, 4457 trans = sctp_addr_id2transport(sk, &params.spp_address,
4456 params.spp_assoc_id); 4458 params.spp_assoc_id);
4457 if (!trans) { 4459 if (!trans) {
4458 SCTP_DEBUG_PRINTK("Failed no transport\n"); 4460 SCTP_DEBUG_PRINTK("Failed no transport\n");
4459 return -EINVAL; 4461 return -EINVAL;
4460 } 4462 }
4461 } 4463 }
4462 4464
4463 /* Get association, if assoc_id != 0 and the socket is a one 4465 /* Get association, if assoc_id != 0 and the socket is a one
4464 * to many style socket, and an association was not found, then 4466 * to many style socket, and an association was not found, then
4465 * the id was invalid. 4467 * the id was invalid.
4466 */ 4468 */
4467 asoc = sctp_id2assoc(sk, params.spp_assoc_id); 4469 asoc = sctp_id2assoc(sk, params.spp_assoc_id);
4468 if (!asoc && params.spp_assoc_id && sctp_style(sk, UDP)) { 4470 if (!asoc && params.spp_assoc_id && sctp_style(sk, UDP)) {
4469 SCTP_DEBUG_PRINTK("Failed no association\n"); 4471 SCTP_DEBUG_PRINTK("Failed no association\n");
4470 return -EINVAL; 4472 return -EINVAL;
4471 } 4473 }
4472 4474
4473 if (trans) { 4475 if (trans) {
4474 /* Fetch transport values. */ 4476 /* Fetch transport values. */
4475 params.spp_hbinterval = jiffies_to_msecs(trans->hbinterval); 4477 params.spp_hbinterval = jiffies_to_msecs(trans->hbinterval);
4476 params.spp_pathmtu = trans->pathmtu; 4478 params.spp_pathmtu = trans->pathmtu;
4477 params.spp_pathmaxrxt = trans->pathmaxrxt; 4479 params.spp_pathmaxrxt = trans->pathmaxrxt;
4478 params.spp_sackdelay = jiffies_to_msecs(trans->sackdelay); 4480 params.spp_sackdelay = jiffies_to_msecs(trans->sackdelay);
4479 4481
4480 /*draft-11 doesn't say what to return in spp_flags*/ 4482 /*draft-11 doesn't say what to return in spp_flags*/
4481 params.spp_flags = trans->param_flags; 4483 params.spp_flags = trans->param_flags;
4482 } else if (asoc) { 4484 } else if (asoc) {
4483 /* Fetch association values. */ 4485 /* Fetch association values. */
4484 params.spp_hbinterval = jiffies_to_msecs(asoc->hbinterval); 4486 params.spp_hbinterval = jiffies_to_msecs(asoc->hbinterval);
4485 params.spp_pathmtu = asoc->pathmtu; 4487 params.spp_pathmtu = asoc->pathmtu;
4486 params.spp_pathmaxrxt = asoc->pathmaxrxt; 4488 params.spp_pathmaxrxt = asoc->pathmaxrxt;
4487 params.spp_sackdelay = jiffies_to_msecs(asoc->sackdelay); 4489 params.spp_sackdelay = jiffies_to_msecs(asoc->sackdelay);
4488 4490
4489 /*draft-11 doesn't say what to return in spp_flags*/ 4491 /*draft-11 doesn't say what to return in spp_flags*/
4490 params.spp_flags = asoc->param_flags; 4492 params.spp_flags = asoc->param_flags;
4491 } else { 4493 } else {
4492 /* Fetch socket values. */ 4494 /* Fetch socket values. */
4493 params.spp_hbinterval = sp->hbinterval; 4495 params.spp_hbinterval = sp->hbinterval;
4494 params.spp_pathmtu = sp->pathmtu; 4496 params.spp_pathmtu = sp->pathmtu;
4495 params.spp_sackdelay = sp->sackdelay; 4497 params.spp_sackdelay = sp->sackdelay;
4496 params.spp_pathmaxrxt = sp->pathmaxrxt; 4498 params.spp_pathmaxrxt = sp->pathmaxrxt;
4497 4499
4498 /*draft-11 doesn't say what to return in spp_flags*/ 4500 /*draft-11 doesn't say what to return in spp_flags*/
4499 params.spp_flags = sp->param_flags; 4501 params.spp_flags = sp->param_flags;
4500 } 4502 }
4501 4503
4502 if (copy_to_user(optval, &params, len)) 4504 if (copy_to_user(optval, &params, len))
4503 return -EFAULT; 4505 return -EFAULT;
4504 4506
4505 if (put_user(len, optlen)) 4507 if (put_user(len, optlen))
4506 return -EFAULT; 4508 return -EFAULT;
4507 4509
4508 return 0; 4510 return 0;
4509 } 4511 }
4510 4512
4511 /* 4513 /*
4512 * 7.1.23. Get or set delayed ack timer (SCTP_DELAYED_SACK) 4514 * 7.1.23. Get or set delayed ack timer (SCTP_DELAYED_SACK)
4513 * 4515 *
4514 * This option will effect the way delayed acks are performed. This 4516 * This option will effect the way delayed acks are performed. This
4515 * option allows you to get or set the delayed ack time, in 4517 * option allows you to get or set the delayed ack time, in
4516 * milliseconds. It also allows changing the delayed ack frequency. 4518 * milliseconds. It also allows changing the delayed ack frequency.
4517 * Changing the frequency to 1 disables the delayed sack algorithm. If 4519 * Changing the frequency to 1 disables the delayed sack algorithm. If
4518 * the assoc_id is 0, then this sets or gets the endpoints default 4520 * the assoc_id is 0, then this sets or gets the endpoints default
4519 * values. If the assoc_id field is non-zero, then the set or get 4521 * values. If the assoc_id field is non-zero, then the set or get
4520 * effects the specified association for the one to many model (the 4522 * effects the specified association for the one to many model (the
4521 * assoc_id field is ignored by the one to one model). Note that if 4523 * assoc_id field is ignored by the one to one model). Note that if
4522 * sack_delay or sack_freq are 0 when setting this option, then the 4524 * sack_delay or sack_freq are 0 when setting this option, then the
4523 * current values will remain unchanged. 4525 * current values will remain unchanged.
4524 * 4526 *
4525 * struct sctp_sack_info { 4527 * struct sctp_sack_info {
4526 * sctp_assoc_t sack_assoc_id; 4528 * sctp_assoc_t sack_assoc_id;
4527 * uint32_t sack_delay; 4529 * uint32_t sack_delay;
4528 * uint32_t sack_freq; 4530 * uint32_t sack_freq;
4529 * }; 4531 * };
4530 * 4532 *
4531 * sack_assoc_id - This parameter, indicates which association the user 4533 * sack_assoc_id - This parameter, indicates which association the user
4532 * is performing an action upon. Note that if this field's value is 4534 * is performing an action upon. Note that if this field's value is
4533 * zero then the endpoints default value is changed (effecting future 4535 * zero then the endpoints default value is changed (effecting future
4534 * associations only). 4536 * associations only).
4535 * 4537 *
4536 * sack_delay - This parameter contains the number of milliseconds that 4538 * sack_delay - This parameter contains the number of milliseconds that
4537 * the user is requesting the delayed ACK timer be set to. Note that 4539 * the user is requesting the delayed ACK timer be set to. Note that
4538 * this value is defined in the standard to be between 200 and 500 4540 * this value is defined in the standard to be between 200 and 500
4539 * milliseconds. 4541 * milliseconds.
4540 * 4542 *
4541 * sack_freq - This parameter contains the number of packets that must 4543 * sack_freq - This parameter contains the number of packets that must
4542 * be received before a sack is sent without waiting for the delay 4544 * be received before a sack is sent without waiting for the delay
4543 * timer to expire. The default value for this is 2, setting this 4545 * timer to expire. The default value for this is 2, setting this
4544 * value to 1 will disable the delayed sack algorithm. 4546 * value to 1 will disable the delayed sack algorithm.
4545 */ 4547 */
4546 static int sctp_getsockopt_delayed_ack(struct sock *sk, int len, 4548 static int sctp_getsockopt_delayed_ack(struct sock *sk, int len,
4547 char __user *optval, 4549 char __user *optval,
4548 int __user *optlen) 4550 int __user *optlen)
4549 { 4551 {
4550 struct sctp_sack_info params; 4552 struct sctp_sack_info params;
4551 struct sctp_association *asoc = NULL; 4553 struct sctp_association *asoc = NULL;
4552 struct sctp_sock *sp = sctp_sk(sk); 4554 struct sctp_sock *sp = sctp_sk(sk);
4553 4555
4554 if (len >= sizeof(struct sctp_sack_info)) { 4556 if (len >= sizeof(struct sctp_sack_info)) {
4555 len = sizeof(struct sctp_sack_info); 4557 len = sizeof(struct sctp_sack_info);
4556 4558
4557 if (copy_from_user(&params, optval, len)) 4559 if (copy_from_user(&params, optval, len))
4558 return -EFAULT; 4560 return -EFAULT;
4559 } else if (len == sizeof(struct sctp_assoc_value)) { 4561 } else if (len == sizeof(struct sctp_assoc_value)) {
4560 pr_warn("Use of struct sctp_assoc_value in delayed_ack socket option deprecated\n"); 4562 pr_warn("Use of struct sctp_assoc_value in delayed_ack socket option deprecated\n");
4561 pr_warn("Use struct sctp_sack_info instead\n"); 4563 pr_warn("Use struct sctp_sack_info instead\n");
4562 if (copy_from_user(&params, optval, len)) 4564 if (copy_from_user(&params, optval, len))
4563 return -EFAULT; 4565 return -EFAULT;
4564 } else 4566 } else
4565 return - EINVAL; 4567 return - EINVAL;
4566 4568
4567 /* Get association, if sack_assoc_id != 0 and the socket is a one 4569 /* Get association, if sack_assoc_id != 0 and the socket is a one
4568 * to many style socket, and an association was not found, then 4570 * to many style socket, and an association was not found, then
4569 * the id was invalid. 4571 * the id was invalid.
4570 */ 4572 */
4571 asoc = sctp_id2assoc(sk, params.sack_assoc_id); 4573 asoc = sctp_id2assoc(sk, params.sack_assoc_id);
4572 if (!asoc && params.sack_assoc_id && sctp_style(sk, UDP)) 4574 if (!asoc && params.sack_assoc_id && sctp_style(sk, UDP))
4573 return -EINVAL; 4575 return -EINVAL;
4574 4576
4575 if (asoc) { 4577 if (asoc) {
4576 /* Fetch association values. */ 4578 /* Fetch association values. */
4577 if (asoc->param_flags & SPP_SACKDELAY_ENABLE) { 4579 if (asoc->param_flags & SPP_SACKDELAY_ENABLE) {
4578 params.sack_delay = jiffies_to_msecs( 4580 params.sack_delay = jiffies_to_msecs(
4579 asoc->sackdelay); 4581 asoc->sackdelay);
4580 params.sack_freq = asoc->sackfreq; 4582 params.sack_freq = asoc->sackfreq;
4581 4583
4582 } else { 4584 } else {
4583 params.sack_delay = 0; 4585 params.sack_delay = 0;
4584 params.sack_freq = 1; 4586 params.sack_freq = 1;
4585 } 4587 }
4586 } else { 4588 } else {
4587 /* Fetch socket values. */ 4589 /* Fetch socket values. */
4588 if (sp->param_flags & SPP_SACKDELAY_ENABLE) { 4590 if (sp->param_flags & SPP_SACKDELAY_ENABLE) {
4589 params.sack_delay = sp->sackdelay; 4591 params.sack_delay = sp->sackdelay;
4590 params.sack_freq = sp->sackfreq; 4592 params.sack_freq = sp->sackfreq;
4591 } else { 4593 } else {
4592 params.sack_delay = 0; 4594 params.sack_delay = 0;
4593 params.sack_freq = 1; 4595 params.sack_freq = 1;
4594 } 4596 }
4595 } 4597 }
4596 4598
4597 if (copy_to_user(optval, &params, len)) 4599 if (copy_to_user(optval, &params, len))
4598 return -EFAULT; 4600 return -EFAULT;
4599 4601
4600 if (put_user(len, optlen)) 4602 if (put_user(len, optlen))
4601 return -EFAULT; 4603 return -EFAULT;
4602 4604
4603 return 0; 4605 return 0;
4604 } 4606 }
4605 4607
4606 /* 7.1.3 Initialization Parameters (SCTP_INITMSG) 4608 /* 7.1.3 Initialization Parameters (SCTP_INITMSG)
4607 * 4609 *
4608 * Applications can specify protocol parameters for the default association 4610 * Applications can specify protocol parameters for the default association
4609 * initialization. The option name argument to setsockopt() and getsockopt() 4611 * initialization. The option name argument to setsockopt() and getsockopt()
4610 * is SCTP_INITMSG. 4612 * is SCTP_INITMSG.
4611 * 4613 *
4612 * Setting initialization parameters is effective only on an unconnected 4614 * Setting initialization parameters is effective only on an unconnected
4613 * socket (for UDP-style sockets only future associations are effected 4615 * socket (for UDP-style sockets only future associations are effected
4614 * by the change). With TCP-style sockets, this option is inherited by 4616 * by the change). With TCP-style sockets, this option is inherited by
4615 * sockets derived from a listener socket. 4617 * sockets derived from a listener socket.
4616 */ 4618 */
4617 static int sctp_getsockopt_initmsg(struct sock *sk, int len, char __user *optval, int __user *optlen) 4619 static int sctp_getsockopt_initmsg(struct sock *sk, int len, char __user *optval, int __user *optlen)
4618 { 4620 {
4619 if (len < sizeof(struct sctp_initmsg)) 4621 if (len < sizeof(struct sctp_initmsg))
4620 return -EINVAL; 4622 return -EINVAL;
4621 len = sizeof(struct sctp_initmsg); 4623 len = sizeof(struct sctp_initmsg);
4622 if (put_user(len, optlen)) 4624 if (put_user(len, optlen))
4623 return -EFAULT; 4625 return -EFAULT;
4624 if (copy_to_user(optval, &sctp_sk(sk)->initmsg, len)) 4626 if (copy_to_user(optval, &sctp_sk(sk)->initmsg, len))
4625 return -EFAULT; 4627 return -EFAULT;
4626 return 0; 4628 return 0;
4627 } 4629 }
4628 4630
4629 4631
4630 static int sctp_getsockopt_peer_addrs(struct sock *sk, int len, 4632 static int sctp_getsockopt_peer_addrs(struct sock *sk, int len,
4631 char __user *optval, int __user *optlen) 4633 char __user *optval, int __user *optlen)
4632 { 4634 {
4633 struct sctp_association *asoc; 4635 struct sctp_association *asoc;
4634 int cnt = 0; 4636 int cnt = 0;
4635 struct sctp_getaddrs getaddrs; 4637 struct sctp_getaddrs getaddrs;
4636 struct sctp_transport *from; 4638 struct sctp_transport *from;
4637 void __user *to; 4639 void __user *to;
4638 union sctp_addr temp; 4640 union sctp_addr temp;
4639 struct sctp_sock *sp = sctp_sk(sk); 4641 struct sctp_sock *sp = sctp_sk(sk);
4640 int addrlen; 4642 int addrlen;
4641 size_t space_left; 4643 size_t space_left;
4642 int bytes_copied; 4644 int bytes_copied;
4643 4645
4644 if (len < sizeof(struct sctp_getaddrs)) 4646 if (len < sizeof(struct sctp_getaddrs))
4645 return -EINVAL; 4647 return -EINVAL;
4646 4648
4647 if (copy_from_user(&getaddrs, optval, sizeof(struct sctp_getaddrs))) 4649 if (copy_from_user(&getaddrs, optval, sizeof(struct sctp_getaddrs)))
4648 return -EFAULT; 4650 return -EFAULT;
4649 4651
4650 /* For UDP-style sockets, id specifies the association to query. */ 4652 /* For UDP-style sockets, id specifies the association to query. */
4651 asoc = sctp_id2assoc(sk, getaddrs.assoc_id); 4653 asoc = sctp_id2assoc(sk, getaddrs.assoc_id);
4652 if (!asoc) 4654 if (!asoc)
4653 return -EINVAL; 4655 return -EINVAL;
4654 4656
4655 to = optval + offsetof(struct sctp_getaddrs,addrs); 4657 to = optval + offsetof(struct sctp_getaddrs,addrs);
4656 space_left = len - offsetof(struct sctp_getaddrs,addrs); 4658 space_left = len - offsetof(struct sctp_getaddrs,addrs);
4657 4659
4658 list_for_each_entry(from, &asoc->peer.transport_addr_list, 4660 list_for_each_entry(from, &asoc->peer.transport_addr_list,
4659 transports) { 4661 transports) {
4660 memcpy(&temp, &from->ipaddr, sizeof(temp)); 4662 memcpy(&temp, &from->ipaddr, sizeof(temp));
4661 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sp, &temp); 4663 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sp, &temp);
4662 addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len; 4664 addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len;
4663 if (space_left < addrlen) 4665 if (space_left < addrlen)
4664 return -ENOMEM; 4666 return -ENOMEM;
4665 if (copy_to_user(to, &temp, addrlen)) 4667 if (copy_to_user(to, &temp, addrlen))
4666 return -EFAULT; 4668 return -EFAULT;
4667 to += addrlen; 4669 to += addrlen;
4668 cnt++; 4670 cnt++;
4669 space_left -= addrlen; 4671 space_left -= addrlen;
4670 } 4672 }
4671 4673
4672 if (put_user(cnt, &((struct sctp_getaddrs __user *)optval)->addr_num)) 4674 if (put_user(cnt, &((struct sctp_getaddrs __user *)optval)->addr_num))
4673 return -EFAULT; 4675 return -EFAULT;
4674 bytes_copied = ((char __user *)to) - optval; 4676 bytes_copied = ((char __user *)to) - optval;
4675 if (put_user(bytes_copied, optlen)) 4677 if (put_user(bytes_copied, optlen))
4676 return -EFAULT; 4678 return -EFAULT;
4677 4679
4678 return 0; 4680 return 0;
4679 } 4681 }
4680 4682
4681 static int sctp_copy_laddrs(struct sock *sk, __u16 port, void *to, 4683 static int sctp_copy_laddrs(struct sock *sk, __u16 port, void *to,
4682 size_t space_left, int *bytes_copied) 4684 size_t space_left, int *bytes_copied)
4683 { 4685 {
4684 struct sctp_sockaddr_entry *addr; 4686 struct sctp_sockaddr_entry *addr;
4685 union sctp_addr temp; 4687 union sctp_addr temp;
4686 int cnt = 0; 4688 int cnt = 0;
4687 int addrlen; 4689 int addrlen;
4688 struct net *net = sock_net(sk); 4690 struct net *net = sock_net(sk);
4689 4691
4690 rcu_read_lock(); 4692 rcu_read_lock();
4691 list_for_each_entry_rcu(addr, &net->sctp.local_addr_list, list) { 4693 list_for_each_entry_rcu(addr, &net->sctp.local_addr_list, list) {
4692 if (!addr->valid) 4694 if (!addr->valid)
4693 continue; 4695 continue;
4694 4696
4695 if ((PF_INET == sk->sk_family) && 4697 if ((PF_INET == sk->sk_family) &&
4696 (AF_INET6 == addr->a.sa.sa_family)) 4698 (AF_INET6 == addr->a.sa.sa_family))
4697 continue; 4699 continue;
4698 if ((PF_INET6 == sk->sk_family) && 4700 if ((PF_INET6 == sk->sk_family) &&
4699 inet_v6_ipv6only(sk) && 4701 inet_v6_ipv6only(sk) &&
4700 (AF_INET == addr->a.sa.sa_family)) 4702 (AF_INET == addr->a.sa.sa_family))
4701 continue; 4703 continue;
4702 memcpy(&temp, &addr->a, sizeof(temp)); 4704 memcpy(&temp, &addr->a, sizeof(temp));
4703 if (!temp.v4.sin_port) 4705 if (!temp.v4.sin_port)
4704 temp.v4.sin_port = htons(port); 4706 temp.v4.sin_port = htons(port);
4705 4707
4706 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sctp_sk(sk), 4708 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sctp_sk(sk),
4707 &temp); 4709 &temp);
4708 addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len; 4710 addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len;
4709 if (space_left < addrlen) { 4711 if (space_left < addrlen) {
4710 cnt = -ENOMEM; 4712 cnt = -ENOMEM;
4711 break; 4713 break;
4712 } 4714 }
4713 memcpy(to, &temp, addrlen); 4715 memcpy(to, &temp, addrlen);
4714 4716
4715 to += addrlen; 4717 to += addrlen;
4716 cnt ++; 4718 cnt ++;
4717 space_left -= addrlen; 4719 space_left -= addrlen;
4718 *bytes_copied += addrlen; 4720 *bytes_copied += addrlen;
4719 } 4721 }
4720 rcu_read_unlock(); 4722 rcu_read_unlock();
4721 4723
4722 return cnt; 4724 return cnt;
4723 } 4725 }
4724 4726
4725 4727
4726 static int sctp_getsockopt_local_addrs(struct sock *sk, int len, 4728 static int sctp_getsockopt_local_addrs(struct sock *sk, int len,
4727 char __user *optval, int __user *optlen) 4729 char __user *optval, int __user *optlen)
4728 { 4730 {
4729 struct sctp_bind_addr *bp; 4731 struct sctp_bind_addr *bp;
4730 struct sctp_association *asoc; 4732 struct sctp_association *asoc;
4731 int cnt = 0; 4733 int cnt = 0;
4732 struct sctp_getaddrs getaddrs; 4734 struct sctp_getaddrs getaddrs;
4733 struct sctp_sockaddr_entry *addr; 4735 struct sctp_sockaddr_entry *addr;
4734 void __user *to; 4736 void __user *to;
4735 union sctp_addr temp; 4737 union sctp_addr temp;
4736 struct sctp_sock *sp = sctp_sk(sk); 4738 struct sctp_sock *sp = sctp_sk(sk);
4737 int addrlen; 4739 int addrlen;
4738 int err = 0; 4740 int err = 0;
4739 size_t space_left; 4741 size_t space_left;
4740 int bytes_copied = 0; 4742 int bytes_copied = 0;
4741 void *addrs; 4743 void *addrs;
4742 void *buf; 4744 void *buf;
4743 4745
4744 if (len < sizeof(struct sctp_getaddrs)) 4746 if (len < sizeof(struct sctp_getaddrs))
4745 return -EINVAL; 4747 return -EINVAL;
4746 4748
4747 if (copy_from_user(&getaddrs, optval, sizeof(struct sctp_getaddrs))) 4749 if (copy_from_user(&getaddrs, optval, sizeof(struct sctp_getaddrs)))
4748 return -EFAULT; 4750 return -EFAULT;
4749 4751
4750 /* 4752 /*
4751 * For UDP-style sockets, id specifies the association to query. 4753 * For UDP-style sockets, id specifies the association to query.
4752 * If the id field is set to the value '0' then the locally bound 4754 * If the id field is set to the value '0' then the locally bound
4753 * addresses are returned without regard to any particular 4755 * addresses are returned without regard to any particular
4754 * association. 4756 * association.
4755 */ 4757 */
4756 if (0 == getaddrs.assoc_id) { 4758 if (0 == getaddrs.assoc_id) {
4757 bp = &sctp_sk(sk)->ep->base.bind_addr; 4759 bp = &sctp_sk(sk)->ep->base.bind_addr;
4758 } else { 4760 } else {
4759 asoc = sctp_id2assoc(sk, getaddrs.assoc_id); 4761 asoc = sctp_id2assoc(sk, getaddrs.assoc_id);
4760 if (!asoc) 4762 if (!asoc)
4761 return -EINVAL; 4763 return -EINVAL;
4762 bp = &asoc->base.bind_addr; 4764 bp = &asoc->base.bind_addr;
4763 } 4765 }
4764 4766
4765 to = optval + offsetof(struct sctp_getaddrs,addrs); 4767 to = optval + offsetof(struct sctp_getaddrs,addrs);
4766 space_left = len - offsetof(struct sctp_getaddrs,addrs); 4768 space_left = len - offsetof(struct sctp_getaddrs,addrs);
4767 4769
4768 addrs = kmalloc(space_left, GFP_KERNEL); 4770 addrs = kmalloc(space_left, GFP_KERNEL);
4769 if (!addrs) 4771 if (!addrs)
4770 return -ENOMEM; 4772 return -ENOMEM;
4771 4773
4772 /* If the endpoint is bound to 0.0.0.0 or ::0, get the valid 4774 /* If the endpoint is bound to 0.0.0.0 or ::0, get the valid
4773 * addresses from the global local address list. 4775 * addresses from the global local address list.
4774 */ 4776 */
4775 if (sctp_list_single_entry(&bp->address_list)) { 4777 if (sctp_list_single_entry(&bp->address_list)) {
4776 addr = list_entry(bp->address_list.next, 4778 addr = list_entry(bp->address_list.next,
4777 struct sctp_sockaddr_entry, list); 4779 struct sctp_sockaddr_entry, list);
4778 if (sctp_is_any(sk, &addr->a)) { 4780 if (sctp_is_any(sk, &addr->a)) {
4779 cnt = sctp_copy_laddrs(sk, bp->port, addrs, 4781 cnt = sctp_copy_laddrs(sk, bp->port, addrs,
4780 space_left, &bytes_copied); 4782 space_left, &bytes_copied);
4781 if (cnt < 0) { 4783 if (cnt < 0) {
4782 err = cnt; 4784 err = cnt;
4783 goto out; 4785 goto out;
4784 } 4786 }
4785 goto copy_getaddrs; 4787 goto copy_getaddrs;
4786 } 4788 }
4787 } 4789 }
4788 4790
4789 buf = addrs; 4791 buf = addrs;
4790 /* Protection on the bound address list is not needed since 4792 /* Protection on the bound address list is not needed since
4791 * in the socket option context we hold a socket lock and 4793 * in the socket option context we hold a socket lock and
4792 * thus the bound address list can't change. 4794 * thus the bound address list can't change.
4793 */ 4795 */
4794 list_for_each_entry(addr, &bp->address_list, list) { 4796 list_for_each_entry(addr, &bp->address_list, list) {
4795 memcpy(&temp, &addr->a, sizeof(temp)); 4797 memcpy(&temp, &addr->a, sizeof(temp));
4796 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sp, &temp); 4798 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sp, &temp);
4797 addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len; 4799 addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len;
4798 if (space_left < addrlen) { 4800 if (space_left < addrlen) {
4799 err = -ENOMEM; /*fixme: right error?*/ 4801 err = -ENOMEM; /*fixme: right error?*/
4800 goto out; 4802 goto out;
4801 } 4803 }
4802 memcpy(buf, &temp, addrlen); 4804 memcpy(buf, &temp, addrlen);
4803 buf += addrlen; 4805 buf += addrlen;
4804 bytes_copied += addrlen; 4806 bytes_copied += addrlen;
4805 cnt ++; 4807 cnt ++;
4806 space_left -= addrlen; 4808 space_left -= addrlen;
4807 } 4809 }
4808 4810
4809 copy_getaddrs: 4811 copy_getaddrs:
4810 if (copy_to_user(to, addrs, bytes_copied)) { 4812 if (copy_to_user(to, addrs, bytes_copied)) {
4811 err = -EFAULT; 4813 err = -EFAULT;
4812 goto out; 4814 goto out;
4813 } 4815 }
4814 if (put_user(cnt, &((struct sctp_getaddrs __user *)optval)->addr_num)) { 4816 if (put_user(cnt, &((struct sctp_getaddrs __user *)optval)->addr_num)) {
4815 err = -EFAULT; 4817 err = -EFAULT;
4816 goto out; 4818 goto out;
4817 } 4819 }
4818 if (put_user(bytes_copied, optlen)) 4820 if (put_user(bytes_copied, optlen))
4819 err = -EFAULT; 4821 err = -EFAULT;
4820 out: 4822 out:
4821 kfree(addrs); 4823 kfree(addrs);
4822 return err; 4824 return err;
4823 } 4825 }
4824 4826
4825 /* 7.1.10 Set Primary Address (SCTP_PRIMARY_ADDR) 4827 /* 7.1.10 Set Primary Address (SCTP_PRIMARY_ADDR)
4826 * 4828 *
4827 * Requests that the local SCTP stack use the enclosed peer address as 4829 * Requests that the local SCTP stack use the enclosed peer address as
4828 * the association primary. The enclosed address must be one of the 4830 * the association primary. The enclosed address must be one of the
4829 * association peer's addresses. 4831 * association peer's addresses.
4830 */ 4832 */
4831 static int sctp_getsockopt_primary_addr(struct sock *sk, int len, 4833 static int sctp_getsockopt_primary_addr(struct sock *sk, int len,
4832 char __user *optval, int __user *optlen) 4834 char __user *optval, int __user *optlen)
4833 { 4835 {
4834 struct sctp_prim prim; 4836 struct sctp_prim prim;
4835 struct sctp_association *asoc; 4837 struct sctp_association *asoc;
4836 struct sctp_sock *sp = sctp_sk(sk); 4838 struct sctp_sock *sp = sctp_sk(sk);
4837 4839
4838 if (len < sizeof(struct sctp_prim)) 4840 if (len < sizeof(struct sctp_prim))
4839 return -EINVAL; 4841 return -EINVAL;
4840 4842
4841 len = sizeof(struct sctp_prim); 4843 len = sizeof(struct sctp_prim);
4842 4844
4843 if (copy_from_user(&prim, optval, len)) 4845 if (copy_from_user(&prim, optval, len))
4844 return -EFAULT; 4846 return -EFAULT;
4845 4847
4846 asoc = sctp_id2assoc(sk, prim.ssp_assoc_id); 4848 asoc = sctp_id2assoc(sk, prim.ssp_assoc_id);
4847 if (!asoc) 4849 if (!asoc)
4848 return -EINVAL; 4850 return -EINVAL;
4849 4851
4850 if (!asoc->peer.primary_path) 4852 if (!asoc->peer.primary_path)
4851 return -ENOTCONN; 4853 return -ENOTCONN;
4852 4854
4853 memcpy(&prim.ssp_addr, &asoc->peer.primary_path->ipaddr, 4855 memcpy(&prim.ssp_addr, &asoc->peer.primary_path->ipaddr,
4854 asoc->peer.primary_path->af_specific->sockaddr_len); 4856 asoc->peer.primary_path->af_specific->sockaddr_len);
4855 4857
4856 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sp, 4858 sctp_get_pf_specific(sk->sk_family)->addr_v4map(sp,
4857 (union sctp_addr *)&prim.ssp_addr); 4859 (union sctp_addr *)&prim.ssp_addr);
4858 4860
4859 if (put_user(len, optlen)) 4861 if (put_user(len, optlen))
4860 return -EFAULT; 4862 return -EFAULT;
4861 if (copy_to_user(optval, &prim, len)) 4863 if (copy_to_user(optval, &prim, len))
4862 return -EFAULT; 4864 return -EFAULT;
4863 4865
4864 return 0; 4866 return 0;
4865 } 4867 }
4866 4868
4867 /* 4869 /*
4868 * 7.1.11 Set Adaptation Layer Indicator (SCTP_ADAPTATION_LAYER) 4870 * 7.1.11 Set Adaptation Layer Indicator (SCTP_ADAPTATION_LAYER)
4869 * 4871 *
4870 * Requests that the local endpoint set the specified Adaptation Layer 4872 * Requests that the local endpoint set the specified Adaptation Layer
4871 * Indication parameter for all future INIT and INIT-ACK exchanges. 4873 * Indication parameter for all future INIT and INIT-ACK exchanges.
4872 */ 4874 */
4873 static int sctp_getsockopt_adaptation_layer(struct sock *sk, int len, 4875 static int sctp_getsockopt_adaptation_layer(struct sock *sk, int len,
4874 char __user *optval, int __user *optlen) 4876 char __user *optval, int __user *optlen)
4875 { 4877 {
4876 struct sctp_setadaptation adaptation; 4878 struct sctp_setadaptation adaptation;
4877 4879
4878 if (len < sizeof(struct sctp_setadaptation)) 4880 if (len < sizeof(struct sctp_setadaptation))
4879 return -EINVAL; 4881 return -EINVAL;
4880 4882
4881 len = sizeof(struct sctp_setadaptation); 4883 len = sizeof(struct sctp_setadaptation);
4882 4884
4883 adaptation.ssb_adaptation_ind = sctp_sk(sk)->adaptation_ind; 4885 adaptation.ssb_adaptation_ind = sctp_sk(sk)->adaptation_ind;
4884 4886
4885 if (put_user(len, optlen)) 4887 if (put_user(len, optlen))
4886 return -EFAULT; 4888 return -EFAULT;
4887 if (copy_to_user(optval, &adaptation, len)) 4889 if (copy_to_user(optval, &adaptation, len))
4888 return -EFAULT; 4890 return -EFAULT;
4889 4891
4890 return 0; 4892 return 0;
4891 } 4893 }
4892 4894
4893 /* 4895 /*
4894 * 4896 *
4895 * 7.1.14 Set default send parameters (SCTP_DEFAULT_SEND_PARAM) 4897 * 7.1.14 Set default send parameters (SCTP_DEFAULT_SEND_PARAM)
4896 * 4898 *
4897 * Applications that wish to use the sendto() system call may wish to 4899 * Applications that wish to use the sendto() system call may wish to
4898 * specify a default set of parameters that would normally be supplied 4900 * specify a default set of parameters that would normally be supplied
4899 * through the inclusion of ancillary data. This socket option allows 4901 * through the inclusion of ancillary data. This socket option allows
4900 * such an application to set the default sctp_sndrcvinfo structure. 4902 * such an application to set the default sctp_sndrcvinfo structure.
4901 4903
4902 4904
4903 * The application that wishes to use this socket option simply passes 4905 * The application that wishes to use this socket option simply passes
4904 * in to this call the sctp_sndrcvinfo structure defined in Section 4906 * in to this call the sctp_sndrcvinfo structure defined in Section
4905 * 5.2.2) The input parameters accepted by this call include 4907 * 5.2.2) The input parameters accepted by this call include
4906 * sinfo_stream, sinfo_flags, sinfo_ppid, sinfo_context, 4908 * sinfo_stream, sinfo_flags, sinfo_ppid, sinfo_context,
4907 * sinfo_timetolive. The user must provide the sinfo_assoc_id field in 4909 * sinfo_timetolive. The user must provide the sinfo_assoc_id field in
4908 * to this call if the caller is using the UDP model. 4910 * to this call if the caller is using the UDP model.
4909 * 4911 *
4910 * For getsockopt, it get the default sctp_sndrcvinfo structure. 4912 * For getsockopt, it get the default sctp_sndrcvinfo structure.
4911 */ 4913 */
4912 static int sctp_getsockopt_default_send_param(struct sock *sk, 4914 static int sctp_getsockopt_default_send_param(struct sock *sk,
4913 int len, char __user *optval, 4915 int len, char __user *optval,
4914 int __user *optlen) 4916 int __user *optlen)
4915 { 4917 {
4916 struct sctp_sndrcvinfo info; 4918 struct sctp_sndrcvinfo info;
4917 struct sctp_association *asoc; 4919 struct sctp_association *asoc;
4918 struct sctp_sock *sp = sctp_sk(sk); 4920 struct sctp_sock *sp = sctp_sk(sk);
4919 4921
4920 if (len < sizeof(struct sctp_sndrcvinfo)) 4922 if (len < sizeof(struct sctp_sndrcvinfo))
4921 return -EINVAL; 4923 return -EINVAL;
4922 4924
4923 len = sizeof(struct sctp_sndrcvinfo); 4925 len = sizeof(struct sctp_sndrcvinfo);
4924 4926
4925 if (copy_from_user(&info, optval, len)) 4927 if (copy_from_user(&info, optval, len))
4926 return -EFAULT; 4928 return -EFAULT;
4927 4929
4928 asoc = sctp_id2assoc(sk, info.sinfo_assoc_id); 4930 asoc = sctp_id2assoc(sk, info.sinfo_assoc_id);
4929 if (!asoc && info.sinfo_assoc_id && sctp_style(sk, UDP)) 4931 if (!asoc && info.sinfo_assoc_id && sctp_style(sk, UDP))
4930 return -EINVAL; 4932 return -EINVAL;
4931 4933
4932 if (asoc) { 4934 if (asoc) {
4933 info.sinfo_stream = asoc->default_stream; 4935 info.sinfo_stream = asoc->default_stream;
4934 info.sinfo_flags = asoc->default_flags; 4936 info.sinfo_flags = asoc->default_flags;
4935 info.sinfo_ppid = asoc->default_ppid; 4937 info.sinfo_ppid = asoc->default_ppid;
4936 info.sinfo_context = asoc->default_context; 4938 info.sinfo_context = asoc->default_context;
4937 info.sinfo_timetolive = asoc->default_timetolive; 4939 info.sinfo_timetolive = asoc->default_timetolive;
4938 } else { 4940 } else {
4939 info.sinfo_stream = sp->default_stream; 4941 info.sinfo_stream = sp->default_stream;
4940 info.sinfo_flags = sp->default_flags; 4942 info.sinfo_flags = sp->default_flags;
4941 info.sinfo_ppid = sp->default_ppid; 4943 info.sinfo_ppid = sp->default_ppid;
4942 info.sinfo_context = sp->default_context; 4944 info.sinfo_context = sp->default_context;
4943 info.sinfo_timetolive = sp->default_timetolive; 4945 info.sinfo_timetolive = sp->default_timetolive;
4944 } 4946 }
4945 4947
4946 if (put_user(len, optlen)) 4948 if (put_user(len, optlen))
4947 return -EFAULT; 4949 return -EFAULT;
4948 if (copy_to_user(optval, &info, len)) 4950 if (copy_to_user(optval, &info, len))
4949 return -EFAULT; 4951 return -EFAULT;
4950 4952
4951 return 0; 4953 return 0;
4952 } 4954 }
4953 4955
4954 /* 4956 /*
4955 * 4957 *
4956 * 7.1.5 SCTP_NODELAY 4958 * 7.1.5 SCTP_NODELAY
4957 * 4959 *
4958 * Turn on/off any Nagle-like algorithm. This means that packets are 4960 * Turn on/off any Nagle-like algorithm. This means that packets are
4959 * generally sent as soon as possible and no unnecessary delays are 4961 * generally sent as soon as possible and no unnecessary delays are
4960 * introduced, at the cost of more packets in the network. Expects an 4962 * introduced, at the cost of more packets in the network. Expects an
4961 * integer boolean flag. 4963 * integer boolean flag.
4962 */ 4964 */
4963 4965
4964 static int sctp_getsockopt_nodelay(struct sock *sk, int len, 4966 static int sctp_getsockopt_nodelay(struct sock *sk, int len,
4965 char __user *optval, int __user *optlen) 4967 char __user *optval, int __user *optlen)
4966 { 4968 {
4967 int val; 4969 int val;
4968 4970
4969 if (len < sizeof(int)) 4971 if (len < sizeof(int))
4970 return -EINVAL; 4972 return -EINVAL;
4971 4973
4972 len = sizeof(int); 4974 len = sizeof(int);
4973 val = (sctp_sk(sk)->nodelay == 1); 4975 val = (sctp_sk(sk)->nodelay == 1);
4974 if (put_user(len, optlen)) 4976 if (put_user(len, optlen))
4975 return -EFAULT; 4977 return -EFAULT;
4976 if (copy_to_user(optval, &val, len)) 4978 if (copy_to_user(optval, &val, len))
4977 return -EFAULT; 4979 return -EFAULT;
4978 return 0; 4980 return 0;
4979 } 4981 }
4980 4982
4981 /* 4983 /*
4982 * 4984 *
4983 * 7.1.1 SCTP_RTOINFO 4985 * 7.1.1 SCTP_RTOINFO
4984 * 4986 *
4985 * The protocol parameters used to initialize and bound retransmission 4987 * The protocol parameters used to initialize and bound retransmission
4986 * timeout (RTO) are tunable. sctp_rtoinfo structure is used to access 4988 * timeout (RTO) are tunable. sctp_rtoinfo structure is used to access
4987 * and modify these parameters. 4989 * and modify these parameters.
4988 * All parameters are time values, in milliseconds. A value of 0, when 4990 * All parameters are time values, in milliseconds. A value of 0, when
4989 * modifying the parameters, indicates that the current value should not 4991 * modifying the parameters, indicates that the current value should not
4990 * be changed. 4992 * be changed.
4991 * 4993 *
4992 */ 4994 */
4993 static int sctp_getsockopt_rtoinfo(struct sock *sk, int len, 4995 static int sctp_getsockopt_rtoinfo(struct sock *sk, int len,
4994 char __user *optval, 4996 char __user *optval,
4995 int __user *optlen) { 4997 int __user *optlen) {
4996 struct sctp_rtoinfo rtoinfo; 4998 struct sctp_rtoinfo rtoinfo;
4997 struct sctp_association *asoc; 4999 struct sctp_association *asoc;
4998 5000
4999 if (len < sizeof (struct sctp_rtoinfo)) 5001 if (len < sizeof (struct sctp_rtoinfo))
5000 return -EINVAL; 5002 return -EINVAL;
5001 5003
5002 len = sizeof(struct sctp_rtoinfo); 5004 len = sizeof(struct sctp_rtoinfo);
5003 5005
5004 if (copy_from_user(&rtoinfo, optval, len)) 5006 if (copy_from_user(&rtoinfo, optval, len))
5005 return -EFAULT; 5007 return -EFAULT;
5006 5008
5007 asoc = sctp_id2assoc(sk, rtoinfo.srto_assoc_id); 5009 asoc = sctp_id2assoc(sk, rtoinfo.srto_assoc_id);
5008 5010
5009 if (!asoc && rtoinfo.srto_assoc_id && sctp_style(sk, UDP)) 5011 if (!asoc && rtoinfo.srto_assoc_id && sctp_style(sk, UDP))
5010 return -EINVAL; 5012 return -EINVAL;
5011 5013
5012 /* Values corresponding to the specific association. */ 5014 /* Values corresponding to the specific association. */
5013 if (asoc) { 5015 if (asoc) {
5014 rtoinfo.srto_initial = jiffies_to_msecs(asoc->rto_initial); 5016 rtoinfo.srto_initial = jiffies_to_msecs(asoc->rto_initial);
5015 rtoinfo.srto_max = jiffies_to_msecs(asoc->rto_max); 5017 rtoinfo.srto_max = jiffies_to_msecs(asoc->rto_max);
5016 rtoinfo.srto_min = jiffies_to_msecs(asoc->rto_min); 5018 rtoinfo.srto_min = jiffies_to_msecs(asoc->rto_min);
5017 } else { 5019 } else {
5018 /* Values corresponding to the endpoint. */ 5020 /* Values corresponding to the endpoint. */
5019 struct sctp_sock *sp = sctp_sk(sk); 5021 struct sctp_sock *sp = sctp_sk(sk);
5020 5022
5021 rtoinfo.srto_initial = sp->rtoinfo.srto_initial; 5023 rtoinfo.srto_initial = sp->rtoinfo.srto_initial;
5022 rtoinfo.srto_max = sp->rtoinfo.srto_max; 5024 rtoinfo.srto_max = sp->rtoinfo.srto_max;
5023 rtoinfo.srto_min = sp->rtoinfo.srto_min; 5025 rtoinfo.srto_min = sp->rtoinfo.srto_min;
5024 } 5026 }
5025 5027
5026 if (put_user(len, optlen)) 5028 if (put_user(len, optlen))
5027 return -EFAULT; 5029 return -EFAULT;
5028 5030
5029 if (copy_to_user(optval, &rtoinfo, len)) 5031 if (copy_to_user(optval, &rtoinfo, len))
5030 return -EFAULT; 5032 return -EFAULT;
5031 5033
5032 return 0; 5034 return 0;
5033 } 5035 }
5034 5036
5035 /* 5037 /*
5036 * 5038 *
5037 * 7.1.2 SCTP_ASSOCINFO 5039 * 7.1.2 SCTP_ASSOCINFO
5038 * 5040 *
5039 * This option is used to tune the maximum retransmission attempts 5041 * This option is used to tune the maximum retransmission attempts
5040 * of the association. 5042 * of the association.
5041 * Returns an error if the new association retransmission value is 5043 * Returns an error if the new association retransmission value is
5042 * greater than the sum of the retransmission value of the peer. 5044 * greater than the sum of the retransmission value of the peer.
5043 * See [SCTP] for more information. 5045 * See [SCTP] for more information.
5044 * 5046 *
5045 */ 5047 */
5046 static int sctp_getsockopt_associnfo(struct sock *sk, int len, 5048 static int sctp_getsockopt_associnfo(struct sock *sk, int len,
5047 char __user *optval, 5049 char __user *optval,
5048 int __user *optlen) 5050 int __user *optlen)
5049 { 5051 {
5050 5052
5051 struct sctp_assocparams assocparams; 5053 struct sctp_assocparams assocparams;
5052 struct sctp_association *asoc; 5054 struct sctp_association *asoc;
5053 struct list_head *pos; 5055 struct list_head *pos;
5054 int cnt = 0; 5056 int cnt = 0;
5055 5057
5056 if (len < sizeof (struct sctp_assocparams)) 5058 if (len < sizeof (struct sctp_assocparams))
5057 return -EINVAL; 5059 return -EINVAL;
5058 5060
5059 len = sizeof(struct sctp_assocparams); 5061 len = sizeof(struct sctp_assocparams);
5060 5062
5061 if (copy_from_user(&assocparams, optval, len)) 5063 if (copy_from_user(&assocparams, optval, len))
5062 return -EFAULT; 5064 return -EFAULT;
5063 5065
5064 asoc = sctp_id2assoc(sk, assocparams.sasoc_assoc_id); 5066 asoc = sctp_id2assoc(sk, assocparams.sasoc_assoc_id);
5065 5067
5066 if (!asoc && assocparams.sasoc_assoc_id && sctp_style(sk, UDP)) 5068 if (!asoc && assocparams.sasoc_assoc_id && sctp_style(sk, UDP))
5067 return -EINVAL; 5069 return -EINVAL;
5068 5070
5069 /* Values correspoinding to the specific association */ 5071 /* Values correspoinding to the specific association */
5070 if (asoc) { 5072 if (asoc) {
5071 assocparams.sasoc_asocmaxrxt = asoc->max_retrans; 5073 assocparams.sasoc_asocmaxrxt = asoc->max_retrans;
5072 assocparams.sasoc_peer_rwnd = asoc->peer.rwnd; 5074 assocparams.sasoc_peer_rwnd = asoc->peer.rwnd;
5073 assocparams.sasoc_local_rwnd = asoc->a_rwnd; 5075 assocparams.sasoc_local_rwnd = asoc->a_rwnd;
5074 assocparams.sasoc_cookie_life = (asoc->cookie_life.tv_sec 5076 assocparams.sasoc_cookie_life = (asoc->cookie_life.tv_sec
5075 * 1000) + 5077 * 1000) +
5076 (asoc->cookie_life.tv_usec 5078 (asoc->cookie_life.tv_usec
5077 / 1000); 5079 / 1000);
5078 5080
5079 list_for_each(pos, &asoc->peer.transport_addr_list) { 5081 list_for_each(pos, &asoc->peer.transport_addr_list) {
5080 cnt ++; 5082 cnt ++;
5081 } 5083 }
5082 5084
5083 assocparams.sasoc_number_peer_destinations = cnt; 5085 assocparams.sasoc_number_peer_destinations = cnt;
5084 } else { 5086 } else {
5085 /* Values corresponding to the endpoint */ 5087 /* Values corresponding to the endpoint */
5086 struct sctp_sock *sp = sctp_sk(sk); 5088 struct sctp_sock *sp = sctp_sk(sk);
5087 5089
5088 assocparams.sasoc_asocmaxrxt = sp->assocparams.sasoc_asocmaxrxt; 5090 assocparams.sasoc_asocmaxrxt = sp->assocparams.sasoc_asocmaxrxt;
5089 assocparams.sasoc_peer_rwnd = sp->assocparams.sasoc_peer_rwnd; 5091 assocparams.sasoc_peer_rwnd = sp->assocparams.sasoc_peer_rwnd;
5090 assocparams.sasoc_local_rwnd = sp->assocparams.sasoc_local_rwnd; 5092 assocparams.sasoc_local_rwnd = sp->assocparams.sasoc_local_rwnd;
5091 assocparams.sasoc_cookie_life = 5093 assocparams.sasoc_cookie_life =
5092 sp->assocparams.sasoc_cookie_life; 5094 sp->assocparams.sasoc_cookie_life;
5093 assocparams.sasoc_number_peer_destinations = 5095 assocparams.sasoc_number_peer_destinations =
5094 sp->assocparams. 5096 sp->assocparams.
5095 sasoc_number_peer_destinations; 5097 sasoc_number_peer_destinations;
5096 } 5098 }
5097 5099
5098 if (put_user(len, optlen)) 5100 if (put_user(len, optlen))
5099 return -EFAULT; 5101 return -EFAULT;
5100 5102
5101 if (copy_to_user(optval, &assocparams, len)) 5103 if (copy_to_user(optval, &assocparams, len))
5102 return -EFAULT; 5104 return -EFAULT;
5103 5105
5104 return 0; 5106 return 0;
5105 } 5107 }
5106 5108
5107 /* 5109 /*
5108 * 7.1.16 Set/clear IPv4 mapped addresses (SCTP_I_WANT_MAPPED_V4_ADDR) 5110 * 7.1.16 Set/clear IPv4 mapped addresses (SCTP_I_WANT_MAPPED_V4_ADDR)
5109 * 5111 *
5110 * This socket option is a boolean flag which turns on or off mapped V4 5112 * This socket option is a boolean flag which turns on or off mapped V4
5111 * addresses. If this option is turned on and the socket is type 5113 * addresses. If this option is turned on and the socket is type
5112 * PF_INET6, then IPv4 addresses will be mapped to V6 representation. 5114 * PF_INET6, then IPv4 addresses will be mapped to V6 representation.
5113 * If this option is turned off, then no mapping will be done of V4 5115 * If this option is turned off, then no mapping will be done of V4
5114 * addresses and a user will receive both PF_INET6 and PF_INET type 5116 * addresses and a user will receive both PF_INET6 and PF_INET type
5115 * addresses on the socket. 5117 * addresses on the socket.
5116 */ 5118 */
5117 static int sctp_getsockopt_mappedv4(struct sock *sk, int len, 5119 static int sctp_getsockopt_mappedv4(struct sock *sk, int len,
5118 char __user *optval, int __user *optlen) 5120 char __user *optval, int __user *optlen)
5119 { 5121 {
5120 int val; 5122 int val;
5121 struct sctp_sock *sp = sctp_sk(sk); 5123 struct sctp_sock *sp = sctp_sk(sk);
5122 5124
5123 if (len < sizeof(int)) 5125 if (len < sizeof(int))
5124 return -EINVAL; 5126 return -EINVAL;
5125 5127
5126 len = sizeof(int); 5128 len = sizeof(int);
5127 val = sp->v4mapped; 5129 val = sp->v4mapped;
5128 if (put_user(len, optlen)) 5130 if (put_user(len, optlen))
5129 return -EFAULT; 5131 return -EFAULT;
5130 if (copy_to_user(optval, &val, len)) 5132 if (copy_to_user(optval, &val, len))
5131 return -EFAULT; 5133 return -EFAULT;
5132 5134
5133 return 0; 5135 return 0;
5134 } 5136 }
5135 5137
5136 /* 5138 /*
5137 * 7.1.29. Set or Get the default context (SCTP_CONTEXT) 5139 * 7.1.29. Set or Get the default context (SCTP_CONTEXT)
5138 * (chapter and verse is quoted at sctp_setsockopt_context()) 5140 * (chapter and verse is quoted at sctp_setsockopt_context())
5139 */ 5141 */
5140 static int sctp_getsockopt_context(struct sock *sk, int len, 5142 static int sctp_getsockopt_context(struct sock *sk, int len,
5141 char __user *optval, int __user *optlen) 5143 char __user *optval, int __user *optlen)
5142 { 5144 {
5143 struct sctp_assoc_value params; 5145 struct sctp_assoc_value params;
5144 struct sctp_sock *sp; 5146 struct sctp_sock *sp;
5145 struct sctp_association *asoc; 5147 struct sctp_association *asoc;
5146 5148
5147 if (len < sizeof(struct sctp_assoc_value)) 5149 if (len < sizeof(struct sctp_assoc_value))
5148 return -EINVAL; 5150 return -EINVAL;
5149 5151
5150 len = sizeof(struct sctp_assoc_value); 5152 len = sizeof(struct sctp_assoc_value);
5151 5153
5152 if (copy_from_user(&params, optval, len)) 5154 if (copy_from_user(&params, optval, len))
5153 return -EFAULT; 5155 return -EFAULT;
5154 5156
5155 sp = sctp_sk(sk); 5157 sp = sctp_sk(sk);
5156 5158
5157 if (params.assoc_id != 0) { 5159 if (params.assoc_id != 0) {
5158 asoc = sctp_id2assoc(sk, params.assoc_id); 5160 asoc = sctp_id2assoc(sk, params.assoc_id);
5159 if (!asoc) 5161 if (!asoc)
5160 return -EINVAL; 5162 return -EINVAL;
5161 params.assoc_value = asoc->default_rcv_context; 5163 params.assoc_value = asoc->default_rcv_context;
5162 } else { 5164 } else {
5163 params.assoc_value = sp->default_rcv_context; 5165 params.assoc_value = sp->default_rcv_context;
5164 } 5166 }
5165 5167
5166 if (put_user(len, optlen)) 5168 if (put_user(len, optlen))
5167 return -EFAULT; 5169 return -EFAULT;
5168 if (copy_to_user(optval, &params, len)) 5170 if (copy_to_user(optval, &params, len))
5169 return -EFAULT; 5171 return -EFAULT;
5170 5172
5171 return 0; 5173 return 0;
5172 } 5174 }
5173 5175
5174 /* 5176 /*
5175 * 8.1.16. Get or Set the Maximum Fragmentation Size (SCTP_MAXSEG) 5177 * 8.1.16. Get or Set the Maximum Fragmentation Size (SCTP_MAXSEG)
5176 * This option will get or set the maximum size to put in any outgoing 5178 * This option will get or set the maximum size to put in any outgoing
5177 * SCTP DATA chunk. If a message is larger than this size it will be 5179 * SCTP DATA chunk. If a message is larger than this size it will be
5178 * fragmented by SCTP into the specified size. Note that the underlying 5180 * fragmented by SCTP into the specified size. Note that the underlying
5179 * SCTP implementation may fragment into smaller sized chunks when the 5181 * SCTP implementation may fragment into smaller sized chunks when the
5180 * PMTU of the underlying association is smaller than the value set by 5182 * PMTU of the underlying association is smaller than the value set by
5181 * the user. The default value for this option is '0' which indicates 5183 * the user. The default value for this option is '0' which indicates
5182 * the user is NOT limiting fragmentation and only the PMTU will effect 5184 * the user is NOT limiting fragmentation and only the PMTU will effect
5183 * SCTP's choice of DATA chunk size. Note also that values set larger 5185 * SCTP's choice of DATA chunk size. Note also that values set larger
5184 * than the maximum size of an IP datagram will effectively let SCTP 5186 * than the maximum size of an IP datagram will effectively let SCTP
5185 * control fragmentation (i.e. the same as setting this option to 0). 5187 * control fragmentation (i.e. the same as setting this option to 0).
5186 * 5188 *
5187 * The following structure is used to access and modify this parameter: 5189 * The following structure is used to access and modify this parameter:
5188 * 5190 *
5189 * struct sctp_assoc_value { 5191 * struct sctp_assoc_value {
5190 * sctp_assoc_t assoc_id; 5192 * sctp_assoc_t assoc_id;
5191 * uint32_t assoc_value; 5193 * uint32_t assoc_value;
5192 * }; 5194 * };
5193 * 5195 *
5194 * assoc_id: This parameter is ignored for one-to-one style sockets. 5196 * assoc_id: This parameter is ignored for one-to-one style sockets.
5195 * For one-to-many style sockets this parameter indicates which 5197 * For one-to-many style sockets this parameter indicates which
5196 * association the user is performing an action upon. Note that if 5198 * association the user is performing an action upon. Note that if
5197 * this field's value is zero then the endpoints default value is 5199 * this field's value is zero then the endpoints default value is
5198 * changed (effecting future associations only). 5200 * changed (effecting future associations only).
5199 * assoc_value: This parameter specifies the maximum size in bytes. 5201 * assoc_value: This parameter specifies the maximum size in bytes.
5200 */ 5202 */
5201 static int sctp_getsockopt_maxseg(struct sock *sk, int len, 5203 static int sctp_getsockopt_maxseg(struct sock *sk, int len,
5202 char __user *optval, int __user *optlen) 5204 char __user *optval, int __user *optlen)
5203 { 5205 {
5204 struct sctp_assoc_value params; 5206 struct sctp_assoc_value params;
5205 struct sctp_association *asoc; 5207 struct sctp_association *asoc;
5206 5208
5207 if (len == sizeof(int)) { 5209 if (len == sizeof(int)) {
5208 pr_warn("Use of int in maxseg socket option deprecated\n"); 5210 pr_warn("Use of int in maxseg socket option deprecated\n");
5209 pr_warn("Use struct sctp_assoc_value instead\n"); 5211 pr_warn("Use struct sctp_assoc_value instead\n");
5210 params.assoc_id = 0; 5212 params.assoc_id = 0;
5211 } else if (len >= sizeof(struct sctp_assoc_value)) { 5213 } else if (len >= sizeof(struct sctp_assoc_value)) {
5212 len = sizeof(struct sctp_assoc_value); 5214 len = sizeof(struct sctp_assoc_value);
5213 if (copy_from_user(&params, optval, sizeof(params))) 5215 if (copy_from_user(&params, optval, sizeof(params)))
5214 return -EFAULT; 5216 return -EFAULT;
5215 } else 5217 } else
5216 return -EINVAL; 5218 return -EINVAL;
5217 5219
5218 asoc = sctp_id2assoc(sk, params.assoc_id); 5220 asoc = sctp_id2assoc(sk, params.assoc_id);
5219 if (!asoc && params.assoc_id && sctp_style(sk, UDP)) 5221 if (!asoc && params.assoc_id && sctp_style(sk, UDP))
5220 return -EINVAL; 5222 return -EINVAL;
5221 5223
5222 if (asoc) 5224 if (asoc)
5223 params.assoc_value = asoc->frag_point; 5225 params.assoc_value = asoc->frag_point;
5224 else 5226 else
5225 params.assoc_value = sctp_sk(sk)->user_frag; 5227 params.assoc_value = sctp_sk(sk)->user_frag;
5226 5228
5227 if (put_user(len, optlen)) 5229 if (put_user(len, optlen))
5228 return -EFAULT; 5230 return -EFAULT;
5229 if (len == sizeof(int)) { 5231 if (len == sizeof(int)) {
5230 if (copy_to_user(optval, &params.assoc_value, len)) 5232 if (copy_to_user(optval, &params.assoc_value, len))
5231 return -EFAULT; 5233 return -EFAULT;
5232 } else { 5234 } else {
5233 if (copy_to_user(optval, &params, len)) 5235 if (copy_to_user(optval, &params, len))
5234 return -EFAULT; 5236 return -EFAULT;
5235 } 5237 }
5236 5238
5237 return 0; 5239 return 0;
5238 } 5240 }
5239 5241
5240 /* 5242 /*
5241 * 7.1.24. Get or set fragmented interleave (SCTP_FRAGMENT_INTERLEAVE) 5243 * 7.1.24. Get or set fragmented interleave (SCTP_FRAGMENT_INTERLEAVE)
5242 * (chapter and verse is quoted at sctp_setsockopt_fragment_interleave()) 5244 * (chapter and verse is quoted at sctp_setsockopt_fragment_interleave())
5243 */ 5245 */
5244 static int sctp_getsockopt_fragment_interleave(struct sock *sk, int len, 5246 static int sctp_getsockopt_fragment_interleave(struct sock *sk, int len,
5245 char __user *optval, int __user *optlen) 5247 char __user *optval, int __user *optlen)
5246 { 5248 {
5247 int val; 5249 int val;
5248 5250
5249 if (len < sizeof(int)) 5251 if (len < sizeof(int))
5250 return -EINVAL; 5252 return -EINVAL;
5251 5253
5252 len = sizeof(int); 5254 len = sizeof(int);
5253 5255
5254 val = sctp_sk(sk)->frag_interleave; 5256 val = sctp_sk(sk)->frag_interleave;
5255 if (put_user(len, optlen)) 5257 if (put_user(len, optlen))
5256 return -EFAULT; 5258 return -EFAULT;
5257 if (copy_to_user(optval, &val, len)) 5259 if (copy_to_user(optval, &val, len))
5258 return -EFAULT; 5260 return -EFAULT;
5259 5261
5260 return 0; 5262 return 0;
5261 } 5263 }
5262 5264
5263 /* 5265 /*
5264 * 7.1.25. Set or Get the sctp partial delivery point 5266 * 7.1.25. Set or Get the sctp partial delivery point
5265 * (chapter and verse is quoted at sctp_setsockopt_partial_delivery_point()) 5267 * (chapter and verse is quoted at sctp_setsockopt_partial_delivery_point())
5266 */ 5268 */
5267 static int sctp_getsockopt_partial_delivery_point(struct sock *sk, int len, 5269 static int sctp_getsockopt_partial_delivery_point(struct sock *sk, int len,
5268 char __user *optval, 5270 char __user *optval,
5269 int __user *optlen) 5271 int __user *optlen)
5270 { 5272 {
5271 u32 val; 5273 u32 val;
5272 5274
5273 if (len < sizeof(u32)) 5275 if (len < sizeof(u32))
5274 return -EINVAL; 5276 return -EINVAL;
5275 5277
5276 len = sizeof(u32); 5278 len = sizeof(u32);
5277 5279
5278 val = sctp_sk(sk)->pd_point; 5280 val = sctp_sk(sk)->pd_point;
5279 if (put_user(len, optlen)) 5281 if (put_user(len, optlen))
5280 return -EFAULT; 5282 return -EFAULT;
5281 if (copy_to_user(optval, &val, len)) 5283 if (copy_to_user(optval, &val, len))
5282 return -EFAULT; 5284 return -EFAULT;
5283 5285
5284 return 0; 5286 return 0;
5285 } 5287 }
5286 5288
5287 /* 5289 /*
5288 * 7.1.28. Set or Get the maximum burst (SCTP_MAX_BURST) 5290 * 7.1.28. Set or Get the maximum burst (SCTP_MAX_BURST)
5289 * (chapter and verse is quoted at sctp_setsockopt_maxburst()) 5291 * (chapter and verse is quoted at sctp_setsockopt_maxburst())
5290 */ 5292 */
5291 static int sctp_getsockopt_maxburst(struct sock *sk, int len, 5293 static int sctp_getsockopt_maxburst(struct sock *sk, int len,
5292 char __user *optval, 5294 char __user *optval,
5293 int __user *optlen) 5295 int __user *optlen)
5294 { 5296 {
5295 struct sctp_assoc_value params; 5297 struct sctp_assoc_value params;
5296 struct sctp_sock *sp; 5298 struct sctp_sock *sp;
5297 struct sctp_association *asoc; 5299 struct sctp_association *asoc;
5298 5300
5299 if (len == sizeof(int)) { 5301 if (len == sizeof(int)) {
5300 pr_warn("Use of int in max_burst socket option deprecated\n"); 5302 pr_warn("Use of int in max_burst socket option deprecated\n");
5301 pr_warn("Use struct sctp_assoc_value instead\n"); 5303 pr_warn("Use struct sctp_assoc_value instead\n");
5302 params.assoc_id = 0; 5304 params.assoc_id = 0;
5303 } else if (len >= sizeof(struct sctp_assoc_value)) { 5305 } else if (len >= sizeof(struct sctp_assoc_value)) {
5304 len = sizeof(struct sctp_assoc_value); 5306 len = sizeof(struct sctp_assoc_value);
5305 if (copy_from_user(&params, optval, len)) 5307 if (copy_from_user(&params, optval, len))
5306 return -EFAULT; 5308 return -EFAULT;
5307 } else 5309 } else
5308 return -EINVAL; 5310 return -EINVAL;
5309 5311
5310 sp = sctp_sk(sk); 5312 sp = sctp_sk(sk);
5311 5313
5312 if (params.assoc_id != 0) { 5314 if (params.assoc_id != 0) {
5313 asoc = sctp_id2assoc(sk, params.assoc_id); 5315 asoc = sctp_id2assoc(sk, params.assoc_id);
5314 if (!asoc) 5316 if (!asoc)
5315 return -EINVAL; 5317 return -EINVAL;
5316 params.assoc_value = asoc->max_burst; 5318 params.assoc_value = asoc->max_burst;
5317 } else 5319 } else
5318 params.assoc_value = sp->max_burst; 5320 params.assoc_value = sp->max_burst;
5319 5321
5320 if (len == sizeof(int)) { 5322 if (len == sizeof(int)) {
5321 if (copy_to_user(optval, &params.assoc_value, len)) 5323 if (copy_to_user(optval, &params.assoc_value, len))
5322 return -EFAULT; 5324 return -EFAULT;
5323 } else { 5325 } else {
5324 if (copy_to_user(optval, &params, len)) 5326 if (copy_to_user(optval, &params, len))
5325 return -EFAULT; 5327 return -EFAULT;
5326 } 5328 }
5327 5329
5328 return 0; 5330 return 0;
5329 5331
5330 } 5332 }
5331 5333
5332 static int sctp_getsockopt_hmac_ident(struct sock *sk, int len, 5334 static int sctp_getsockopt_hmac_ident(struct sock *sk, int len,
5333 char __user *optval, int __user *optlen) 5335 char __user *optval, int __user *optlen)
5334 { 5336 {
5335 struct net *net = sock_net(sk); 5337 struct net *net = sock_net(sk);
5336 struct sctp_hmacalgo __user *p = (void __user *)optval; 5338 struct sctp_hmacalgo __user *p = (void __user *)optval;
5337 struct sctp_hmac_algo_param *hmacs; 5339 struct sctp_hmac_algo_param *hmacs;
5338 __u16 data_len = 0; 5340 __u16 data_len = 0;
5339 u32 num_idents; 5341 u32 num_idents;
5340 5342
5341 if (!net->sctp.auth_enable) 5343 if (!net->sctp.auth_enable)
5342 return -EACCES; 5344 return -EACCES;
5343 5345
5344 hmacs = sctp_sk(sk)->ep->auth_hmacs_list; 5346 hmacs = sctp_sk(sk)->ep->auth_hmacs_list;
5345 data_len = ntohs(hmacs->param_hdr.length) - sizeof(sctp_paramhdr_t); 5347 data_len = ntohs(hmacs->param_hdr.length) - sizeof(sctp_paramhdr_t);
5346 5348
5347 if (len < sizeof(struct sctp_hmacalgo) + data_len) 5349 if (len < sizeof(struct sctp_hmacalgo) + data_len)
5348 return -EINVAL; 5350 return -EINVAL;
5349 5351
5350 len = sizeof(struct sctp_hmacalgo) + data_len; 5352 len = sizeof(struct sctp_hmacalgo) + data_len;
5351 num_idents = data_len / sizeof(u16); 5353 num_idents = data_len / sizeof(u16);
5352 5354
5353 if (put_user(len, optlen)) 5355 if (put_user(len, optlen))
5354 return -EFAULT; 5356 return -EFAULT;
5355 if (put_user(num_idents, &p->shmac_num_idents)) 5357 if (put_user(num_idents, &p->shmac_num_idents))
5356 return -EFAULT; 5358 return -EFAULT;
5357 if (copy_to_user(p->shmac_idents, hmacs->hmac_ids, data_len)) 5359 if (copy_to_user(p->shmac_idents, hmacs->hmac_ids, data_len))
5358 return -EFAULT; 5360 return -EFAULT;
5359 return 0; 5361 return 0;
5360 } 5362 }
5361 5363
5362 static int sctp_getsockopt_active_key(struct sock *sk, int len, 5364 static int sctp_getsockopt_active_key(struct sock *sk, int len,
5363 char __user *optval, int __user *optlen) 5365 char __user *optval, int __user *optlen)
5364 { 5366 {
5365 struct net *net = sock_net(sk); 5367 struct net *net = sock_net(sk);
5366 struct sctp_authkeyid val; 5368 struct sctp_authkeyid val;
5367 struct sctp_association *asoc; 5369 struct sctp_association *asoc;
5368 5370
5369 if (!net->sctp.auth_enable) 5371 if (!net->sctp.auth_enable)
5370 return -EACCES; 5372 return -EACCES;
5371 5373
5372 if (len < sizeof(struct sctp_authkeyid)) 5374 if (len < sizeof(struct sctp_authkeyid))
5373 return -EINVAL; 5375 return -EINVAL;
5374 if (copy_from_user(&val, optval, sizeof(struct sctp_authkeyid))) 5376 if (copy_from_user(&val, optval, sizeof(struct sctp_authkeyid)))
5375 return -EFAULT; 5377 return -EFAULT;
5376 5378
5377 asoc = sctp_id2assoc(sk, val.scact_assoc_id); 5379 asoc = sctp_id2assoc(sk, val.scact_assoc_id);
5378 if (!asoc && val.scact_assoc_id && sctp_style(sk, UDP)) 5380 if (!asoc && val.scact_assoc_id && sctp_style(sk, UDP))
5379 return -EINVAL; 5381 return -EINVAL;
5380 5382
5381 if (asoc) 5383 if (asoc)
5382 val.scact_keynumber = asoc->active_key_id; 5384 val.scact_keynumber = asoc->active_key_id;
5383 else 5385 else
5384 val.scact_keynumber = sctp_sk(sk)->ep->active_key_id; 5386 val.scact_keynumber = sctp_sk(sk)->ep->active_key_id;
5385 5387
5386 len = sizeof(struct sctp_authkeyid); 5388 len = sizeof(struct sctp_authkeyid);
5387 if (put_user(len, optlen)) 5389 if (put_user(len, optlen))
5388 return -EFAULT; 5390 return -EFAULT;
5389 if (copy_to_user(optval, &val, len)) 5391 if (copy_to_user(optval, &val, len))
5390 return -EFAULT; 5392 return -EFAULT;
5391 5393
5392 return 0; 5394 return 0;
5393 } 5395 }
5394 5396
5395 static int sctp_getsockopt_peer_auth_chunks(struct sock *sk, int len, 5397 static int sctp_getsockopt_peer_auth_chunks(struct sock *sk, int len,
5396 char __user *optval, int __user *optlen) 5398 char __user *optval, int __user *optlen)
5397 { 5399 {
5398 struct net *net = sock_net(sk); 5400 struct net *net = sock_net(sk);
5399 struct sctp_authchunks __user *p = (void __user *)optval; 5401 struct sctp_authchunks __user *p = (void __user *)optval;
5400 struct sctp_authchunks val; 5402 struct sctp_authchunks val;
5401 struct sctp_association *asoc; 5403 struct sctp_association *asoc;
5402 struct sctp_chunks_param *ch; 5404 struct sctp_chunks_param *ch;
5403 u32 num_chunks = 0; 5405 u32 num_chunks = 0;
5404 char __user *to; 5406 char __user *to;
5405 5407
5406 if (!net->sctp.auth_enable) 5408 if (!net->sctp.auth_enable)
5407 return -EACCES; 5409 return -EACCES;
5408 5410
5409 if (len < sizeof(struct sctp_authchunks)) 5411 if (len < sizeof(struct sctp_authchunks))
5410 return -EINVAL; 5412 return -EINVAL;
5411 5413
5412 if (copy_from_user(&val, optval, sizeof(struct sctp_authchunks))) 5414 if (copy_from_user(&val, optval, sizeof(struct sctp_authchunks)))
5413 return -EFAULT; 5415 return -EFAULT;
5414 5416
5415 to = p->gauth_chunks; 5417 to = p->gauth_chunks;
5416 asoc = sctp_id2assoc(sk, val.gauth_assoc_id); 5418 asoc = sctp_id2assoc(sk, val.gauth_assoc_id);
5417 if (!asoc) 5419 if (!asoc)
5418 return -EINVAL; 5420 return -EINVAL;
5419 5421
5420 ch = asoc->peer.peer_chunks; 5422 ch = asoc->peer.peer_chunks;
5421 if (!ch) 5423 if (!ch)
5422 goto num; 5424 goto num;
5423 5425
5424 /* See if the user provided enough room for all the data */ 5426 /* See if the user provided enough room for all the data */
5425 num_chunks = ntohs(ch->param_hdr.length) - sizeof(sctp_paramhdr_t); 5427 num_chunks = ntohs(ch->param_hdr.length) - sizeof(sctp_paramhdr_t);
5426 if (len < num_chunks) 5428 if (len < num_chunks)
5427 return -EINVAL; 5429 return -EINVAL;
5428 5430
5429 if (copy_to_user(to, ch->chunks, num_chunks)) 5431 if (copy_to_user(to, ch->chunks, num_chunks))
5430 return -EFAULT; 5432 return -EFAULT;
5431 num: 5433 num:
5432 len = sizeof(struct sctp_authchunks) + num_chunks; 5434 len = sizeof(struct sctp_authchunks) + num_chunks;
5433 if (put_user(len, optlen)) return -EFAULT; 5435 if (put_user(len, optlen)) return -EFAULT;
5434 if (put_user(num_chunks, &p->gauth_number_of_chunks)) 5436 if (put_user(num_chunks, &p->gauth_number_of_chunks))
5435 return -EFAULT; 5437 return -EFAULT;
5436 return 0; 5438 return 0;
5437 } 5439 }
5438 5440
5439 static int sctp_getsockopt_local_auth_chunks(struct sock *sk, int len, 5441 static int sctp_getsockopt_local_auth_chunks(struct sock *sk, int len,
5440 char __user *optval, int __user *optlen) 5442 char __user *optval, int __user *optlen)
5441 { 5443 {
5442 struct net *net = sock_net(sk); 5444 struct net *net = sock_net(sk);
5443 struct sctp_authchunks __user *p = (void __user *)optval; 5445 struct sctp_authchunks __user *p = (void __user *)optval;
5444 struct sctp_authchunks val; 5446 struct sctp_authchunks val;
5445 struct sctp_association *asoc; 5447 struct sctp_association *asoc;
5446 struct sctp_chunks_param *ch; 5448 struct sctp_chunks_param *ch;
5447 u32 num_chunks = 0; 5449 u32 num_chunks = 0;
5448 char __user *to; 5450 char __user *to;
5449 5451
5450 if (!net->sctp.auth_enable) 5452 if (!net->sctp.auth_enable)
5451 return -EACCES; 5453 return -EACCES;
5452 5454
5453 if (len < sizeof(struct sctp_authchunks)) 5455 if (len < sizeof(struct sctp_authchunks))
5454 return -EINVAL; 5456 return -EINVAL;
5455 5457
5456 if (copy_from_user(&val, optval, sizeof(struct sctp_authchunks))) 5458 if (copy_from_user(&val, optval, sizeof(struct sctp_authchunks)))
5457 return -EFAULT; 5459 return -EFAULT;
5458 5460
5459 to = p->gauth_chunks; 5461 to = p->gauth_chunks;
5460 asoc = sctp_id2assoc(sk, val.gauth_assoc_id); 5462 asoc = sctp_id2assoc(sk, val.gauth_assoc_id);
5461 if (!asoc && val.gauth_assoc_id && sctp_style(sk, UDP)) 5463 if (!asoc && val.gauth_assoc_id && sctp_style(sk, UDP))
5462 return -EINVAL; 5464 return -EINVAL;
5463 5465
5464 if (asoc) 5466 if (asoc)
5465 ch = (struct sctp_chunks_param*)asoc->c.auth_chunks; 5467 ch = (struct sctp_chunks_param*)asoc->c.auth_chunks;
5466 else 5468 else
5467 ch = sctp_sk(sk)->ep->auth_chunk_list; 5469 ch = sctp_sk(sk)->ep->auth_chunk_list;
5468 5470
5469 if (!ch) 5471 if (!ch)
5470 goto num; 5472 goto num;
5471 5473
5472 num_chunks = ntohs(ch->param_hdr.length) - sizeof(sctp_paramhdr_t); 5474 num_chunks = ntohs(ch->param_hdr.length) - sizeof(sctp_paramhdr_t);
5473 if (len < sizeof(struct sctp_authchunks) + num_chunks) 5475 if (len < sizeof(struct sctp_authchunks) + num_chunks)
5474 return -EINVAL; 5476 return -EINVAL;
5475 5477
5476 if (copy_to_user(to, ch->chunks, num_chunks)) 5478 if (copy_to_user(to, ch->chunks, num_chunks))
5477 return -EFAULT; 5479 return -EFAULT;
5478 num: 5480 num:
5479 len = sizeof(struct sctp_authchunks) + num_chunks; 5481 len = sizeof(struct sctp_authchunks) + num_chunks;
5480 if (put_user(len, optlen)) 5482 if (put_user(len, optlen))
5481 return -EFAULT; 5483 return -EFAULT;
5482 if (put_user(num_chunks, &p->gauth_number_of_chunks)) 5484 if (put_user(num_chunks, &p->gauth_number_of_chunks))
5483 return -EFAULT; 5485 return -EFAULT;
5484 5486
5485 return 0; 5487 return 0;
5486 } 5488 }
5487 5489
5488 /* 5490 /*
5489 * 8.2.5. Get the Current Number of Associations (SCTP_GET_ASSOC_NUMBER) 5491 * 8.2.5. Get the Current Number of Associations (SCTP_GET_ASSOC_NUMBER)
5490 * This option gets the current number of associations that are attached 5492 * This option gets the current number of associations that are attached
5491 * to a one-to-many style socket. The option value is an uint32_t. 5493 * to a one-to-many style socket. The option value is an uint32_t.
5492 */ 5494 */
5493 static int sctp_getsockopt_assoc_number(struct sock *sk, int len, 5495 static int sctp_getsockopt_assoc_number(struct sock *sk, int len,
5494 char __user *optval, int __user *optlen) 5496 char __user *optval, int __user *optlen)
5495 { 5497 {
5496 struct sctp_sock *sp = sctp_sk(sk); 5498 struct sctp_sock *sp = sctp_sk(sk);
5497 struct sctp_association *asoc; 5499 struct sctp_association *asoc;
5498 u32 val = 0; 5500 u32 val = 0;
5499 5501
5500 if (sctp_style(sk, TCP)) 5502 if (sctp_style(sk, TCP))
5501 return -EOPNOTSUPP; 5503 return -EOPNOTSUPP;
5502 5504
5503 if (len < sizeof(u32)) 5505 if (len < sizeof(u32))
5504 return -EINVAL; 5506 return -EINVAL;
5505 5507
5506 len = sizeof(u32); 5508 len = sizeof(u32);
5507 5509
5508 list_for_each_entry(asoc, &(sp->ep->asocs), asocs) { 5510 list_for_each_entry(asoc, &(sp->ep->asocs), asocs) {
5509 val++; 5511 val++;
5510 } 5512 }
5511 5513
5512 if (put_user(len, optlen)) 5514 if (put_user(len, optlen))
5513 return -EFAULT; 5515 return -EFAULT;
5514 if (copy_to_user(optval, &val, len)) 5516 if (copy_to_user(optval, &val, len))
5515 return -EFAULT; 5517 return -EFAULT;
5516 5518
5517 return 0; 5519 return 0;
5518 } 5520 }
5519 5521
5520 /* 5522 /*
5521 * 8.1.23 SCTP_AUTO_ASCONF 5523 * 8.1.23 SCTP_AUTO_ASCONF
5522 * See the corresponding setsockopt entry as description 5524 * See the corresponding setsockopt entry as description
5523 */ 5525 */
5524 static int sctp_getsockopt_auto_asconf(struct sock *sk, int len, 5526 static int sctp_getsockopt_auto_asconf(struct sock *sk, int len,
5525 char __user *optval, int __user *optlen) 5527 char __user *optval, int __user *optlen)
5526 { 5528 {
5527 int val = 0; 5529 int val = 0;
5528 5530
5529 if (len < sizeof(int)) 5531 if (len < sizeof(int))
5530 return -EINVAL; 5532 return -EINVAL;
5531 5533
5532 len = sizeof(int); 5534 len = sizeof(int);
5533 if (sctp_sk(sk)->do_auto_asconf && sctp_is_ep_boundall(sk)) 5535 if (sctp_sk(sk)->do_auto_asconf && sctp_is_ep_boundall(sk))
5534 val = 1; 5536 val = 1;
5535 if (put_user(len, optlen)) 5537 if (put_user(len, optlen))
5536 return -EFAULT; 5538 return -EFAULT;
5537 if (copy_to_user(optval, &val, len)) 5539 if (copy_to_user(optval, &val, len))
5538 return -EFAULT; 5540 return -EFAULT;
5539 return 0; 5541 return 0;
5540 } 5542 }
5541 5543
5542 /* 5544 /*
5543 * 8.2.6. Get the Current Identifiers of Associations 5545 * 8.2.6. Get the Current Identifiers of Associations
5544 * (SCTP_GET_ASSOC_ID_LIST) 5546 * (SCTP_GET_ASSOC_ID_LIST)
5545 * 5547 *
5546 * This option gets the current list of SCTP association identifiers of 5548 * This option gets the current list of SCTP association identifiers of
5547 * the SCTP associations handled by a one-to-many style socket. 5549 * the SCTP associations handled by a one-to-many style socket.
5548 */ 5550 */
5549 static int sctp_getsockopt_assoc_ids(struct sock *sk, int len, 5551 static int sctp_getsockopt_assoc_ids(struct sock *sk, int len,
5550 char __user *optval, int __user *optlen) 5552 char __user *optval, int __user *optlen)
5551 { 5553 {
5552 struct sctp_sock *sp = sctp_sk(sk); 5554 struct sctp_sock *sp = sctp_sk(sk);
5553 struct sctp_association *asoc; 5555 struct sctp_association *asoc;
5554 struct sctp_assoc_ids *ids; 5556 struct sctp_assoc_ids *ids;
5555 u32 num = 0; 5557 u32 num = 0;
5556 5558
5557 if (sctp_style(sk, TCP)) 5559 if (sctp_style(sk, TCP))
5558 return -EOPNOTSUPP; 5560 return -EOPNOTSUPP;
5559 5561
5560 if (len < sizeof(struct sctp_assoc_ids)) 5562 if (len < sizeof(struct sctp_assoc_ids))
5561 return -EINVAL; 5563 return -EINVAL;
5562 5564
5563 list_for_each_entry(asoc, &(sp->ep->asocs), asocs) { 5565 list_for_each_entry(asoc, &(sp->ep->asocs), asocs) {
5564 num++; 5566 num++;
5565 } 5567 }
5566 5568
5567 if (len < sizeof(struct sctp_assoc_ids) + sizeof(sctp_assoc_t) * num) 5569 if (len < sizeof(struct sctp_assoc_ids) + sizeof(sctp_assoc_t) * num)
5568 return -EINVAL; 5570 return -EINVAL;
5569 5571
5570 len = sizeof(struct sctp_assoc_ids) + sizeof(sctp_assoc_t) * num; 5572 len = sizeof(struct sctp_assoc_ids) + sizeof(sctp_assoc_t) * num;
5571 5573
5572 ids = kmalloc(len, GFP_KERNEL); 5574 ids = kmalloc(len, GFP_KERNEL);
5573 if (unlikely(!ids)) 5575 if (unlikely(!ids))
5574 return -ENOMEM; 5576 return -ENOMEM;
5575 5577
5576 ids->gaids_number_of_ids = num; 5578 ids->gaids_number_of_ids = num;
5577 num = 0; 5579 num = 0;
5578 list_for_each_entry(asoc, &(sp->ep->asocs), asocs) { 5580 list_for_each_entry(asoc, &(sp->ep->asocs), asocs) {
5579 ids->gaids_assoc_id[num++] = asoc->assoc_id; 5581 ids->gaids_assoc_id[num++] = asoc->assoc_id;
5580 } 5582 }
5581 5583
5582 if (put_user(len, optlen) || copy_to_user(optval, ids, len)) { 5584 if (put_user(len, optlen) || copy_to_user(optval, ids, len)) {
5583 kfree(ids); 5585 kfree(ids);
5584 return -EFAULT; 5586 return -EFAULT;
5585 } 5587 }
5586 5588
5587 kfree(ids); 5589 kfree(ids);
5588 return 0; 5590 return 0;
5589 } 5591 }
5590 5592
5591 /* 5593 /*
5592 * SCTP_PEER_ADDR_THLDS 5594 * SCTP_PEER_ADDR_THLDS
5593 * 5595 *
5594 * This option allows us to fetch the partially failed threshold for one or all 5596 * This option allows us to fetch the partially failed threshold for one or all
5595 * transports in an association. See Section 6.1 of: 5597 * transports in an association. See Section 6.1 of:
5596 * http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt 5598 * http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
5597 */ 5599 */
5598 static int sctp_getsockopt_paddr_thresholds(struct sock *sk, 5600 static int sctp_getsockopt_paddr_thresholds(struct sock *sk,
5599 char __user *optval, 5601 char __user *optval,
5600 int len, 5602 int len,
5601 int __user *optlen) 5603 int __user *optlen)
5602 { 5604 {
5603 struct sctp_paddrthlds val; 5605 struct sctp_paddrthlds val;
5604 struct sctp_transport *trans; 5606 struct sctp_transport *trans;
5605 struct sctp_association *asoc; 5607 struct sctp_association *asoc;
5606 5608
5607 if (len < sizeof(struct sctp_paddrthlds)) 5609 if (len < sizeof(struct sctp_paddrthlds))
5608 return -EINVAL; 5610 return -EINVAL;
5609 len = sizeof(struct sctp_paddrthlds); 5611 len = sizeof(struct sctp_paddrthlds);
5610 if (copy_from_user(&val, (struct sctp_paddrthlds __user *)optval, len)) 5612 if (copy_from_user(&val, (struct sctp_paddrthlds __user *)optval, len))
5611 return -EFAULT; 5613 return -EFAULT;
5612 5614
5613 if (sctp_is_any(sk, (const union sctp_addr *)&val.spt_address)) { 5615 if (sctp_is_any(sk, (const union sctp_addr *)&val.spt_address)) {
5614 asoc = sctp_id2assoc(sk, val.spt_assoc_id); 5616 asoc = sctp_id2assoc(sk, val.spt_assoc_id);
5615 if (!asoc) 5617 if (!asoc)
5616 return -ENOENT; 5618 return -ENOENT;
5617 5619
5618 val.spt_pathpfthld = asoc->pf_retrans; 5620 val.spt_pathpfthld = asoc->pf_retrans;
5619 val.spt_pathmaxrxt = asoc->pathmaxrxt; 5621 val.spt_pathmaxrxt = asoc->pathmaxrxt;
5620 } else { 5622 } else {
5621 trans = sctp_addr_id2transport(sk, &val.spt_address, 5623 trans = sctp_addr_id2transport(sk, &val.spt_address,
5622 val.spt_assoc_id); 5624 val.spt_assoc_id);
5623 if (!trans) 5625 if (!trans)
5624 return -ENOENT; 5626 return -ENOENT;
5625 5627
5626 val.spt_pathmaxrxt = trans->pathmaxrxt; 5628 val.spt_pathmaxrxt = trans->pathmaxrxt;
5627 val.spt_pathpfthld = trans->pf_retrans; 5629 val.spt_pathpfthld = trans->pf_retrans;
5628 } 5630 }
5629 5631
5630 if (put_user(len, optlen) || copy_to_user(optval, &val, len)) 5632 if (put_user(len, optlen) || copy_to_user(optval, &val, len))
5631 return -EFAULT; 5633 return -EFAULT;
5632 5634
5633 return 0; 5635 return 0;
5634 } 5636 }
5635 5637
5636 SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname, 5638 SCTP_STATIC int sctp_getsockopt(struct sock *sk, int level, int optname,
5637 char __user *optval, int __user *optlen) 5639 char __user *optval, int __user *optlen)
5638 { 5640 {
5639 int retval = 0; 5641 int retval = 0;
5640 int len; 5642 int len;
5641 5643
5642 SCTP_DEBUG_PRINTK("sctp_getsockopt(sk: %p... optname: %d)\n", 5644 SCTP_DEBUG_PRINTK("sctp_getsockopt(sk: %p... optname: %d)\n",
5643 sk, optname); 5645 sk, optname);
5644 5646
5645 /* I can hardly begin to describe how wrong this is. This is 5647 /* I can hardly begin to describe how wrong this is. This is
5646 * so broken as to be worse than useless. The API draft 5648 * so broken as to be worse than useless. The API draft
5647 * REALLY is NOT helpful here... I am not convinced that the 5649 * REALLY is NOT helpful here... I am not convinced that the
5648 * semantics of getsockopt() with a level OTHER THAN SOL_SCTP 5650 * semantics of getsockopt() with a level OTHER THAN SOL_SCTP
5649 * are at all well-founded. 5651 * are at all well-founded.
5650 */ 5652 */
5651 if (level != SOL_SCTP) { 5653 if (level != SOL_SCTP) {
5652 struct sctp_af *af = sctp_sk(sk)->pf->af; 5654 struct sctp_af *af = sctp_sk(sk)->pf->af;
5653 5655
5654 retval = af->getsockopt(sk, level, optname, optval, optlen); 5656 retval = af->getsockopt(sk, level, optname, optval, optlen);
5655 return retval; 5657 return retval;
5656 } 5658 }
5657 5659
5658 if (get_user(len, optlen)) 5660 if (get_user(len, optlen))
5659 return -EFAULT; 5661 return -EFAULT;
5660 5662
5661 sctp_lock_sock(sk); 5663 sctp_lock_sock(sk);
5662 5664
5663 switch (optname) { 5665 switch (optname) {
5664 case SCTP_STATUS: 5666 case SCTP_STATUS:
5665 retval = sctp_getsockopt_sctp_status(sk, len, optval, optlen); 5667 retval = sctp_getsockopt_sctp_status(sk, len, optval, optlen);
5666 break; 5668 break;
5667 case SCTP_DISABLE_FRAGMENTS: 5669 case SCTP_DISABLE_FRAGMENTS:
5668 retval = sctp_getsockopt_disable_fragments(sk, len, optval, 5670 retval = sctp_getsockopt_disable_fragments(sk, len, optval,
5669 optlen); 5671 optlen);
5670 break; 5672 break;
5671 case SCTP_EVENTS: 5673 case SCTP_EVENTS:
5672 retval = sctp_getsockopt_events(sk, len, optval, optlen); 5674 retval = sctp_getsockopt_events(sk, len, optval, optlen);
5673 break; 5675 break;
5674 case SCTP_AUTOCLOSE: 5676 case SCTP_AUTOCLOSE:
5675 retval = sctp_getsockopt_autoclose(sk, len, optval, optlen); 5677 retval = sctp_getsockopt_autoclose(sk, len, optval, optlen);
5676 break; 5678 break;
5677 case SCTP_SOCKOPT_PEELOFF: 5679 case SCTP_SOCKOPT_PEELOFF:
5678 retval = sctp_getsockopt_peeloff(sk, len, optval, optlen); 5680 retval = sctp_getsockopt_peeloff(sk, len, optval, optlen);
5679 break; 5681 break;
5680 case SCTP_PEER_ADDR_PARAMS: 5682 case SCTP_PEER_ADDR_PARAMS:
5681 retval = sctp_getsockopt_peer_addr_params(sk, len, optval, 5683 retval = sctp_getsockopt_peer_addr_params(sk, len, optval,
5682 optlen); 5684 optlen);
5683 break; 5685 break;
5684 case SCTP_DELAYED_SACK: 5686 case SCTP_DELAYED_SACK:
5685 retval = sctp_getsockopt_delayed_ack(sk, len, optval, 5687 retval = sctp_getsockopt_delayed_ack(sk, len, optval,
5686 optlen); 5688 optlen);
5687 break; 5689 break;
5688 case SCTP_INITMSG: 5690 case SCTP_INITMSG:
5689 retval = sctp_getsockopt_initmsg(sk, len, optval, optlen); 5691 retval = sctp_getsockopt_initmsg(sk, len, optval, optlen);
5690 break; 5692 break;
5691 case SCTP_GET_PEER_ADDRS: 5693 case SCTP_GET_PEER_ADDRS:
5692 retval = sctp_getsockopt_peer_addrs(sk, len, optval, 5694 retval = sctp_getsockopt_peer_addrs(sk, len, optval,
5693 optlen); 5695 optlen);
5694 break; 5696 break;
5695 case SCTP_GET_LOCAL_ADDRS: 5697 case SCTP_GET_LOCAL_ADDRS:
5696 retval = sctp_getsockopt_local_addrs(sk, len, optval, 5698 retval = sctp_getsockopt_local_addrs(sk, len, optval,
5697 optlen); 5699 optlen);
5698 break; 5700 break;
5699 case SCTP_SOCKOPT_CONNECTX3: 5701 case SCTP_SOCKOPT_CONNECTX3:
5700 retval = sctp_getsockopt_connectx3(sk, len, optval, optlen); 5702 retval = sctp_getsockopt_connectx3(sk, len, optval, optlen);
5701 break; 5703 break;
5702 case SCTP_DEFAULT_SEND_PARAM: 5704 case SCTP_DEFAULT_SEND_PARAM:
5703 retval = sctp_getsockopt_default_send_param(sk, len, 5705 retval = sctp_getsockopt_default_send_param(sk, len,
5704 optval, optlen); 5706 optval, optlen);
5705 break; 5707 break;
5706 case SCTP_PRIMARY_ADDR: 5708 case SCTP_PRIMARY_ADDR:
5707 retval = sctp_getsockopt_primary_addr(sk, len, optval, optlen); 5709 retval = sctp_getsockopt_primary_addr(sk, len, optval, optlen);
5708 break; 5710 break;
5709 case SCTP_NODELAY: 5711 case SCTP_NODELAY:
5710 retval = sctp_getsockopt_nodelay(sk, len, optval, optlen); 5712 retval = sctp_getsockopt_nodelay(sk, len, optval, optlen);
5711 break; 5713 break;
5712 case SCTP_RTOINFO: 5714 case SCTP_RTOINFO:
5713 retval = sctp_getsockopt_rtoinfo(sk, len, optval, optlen); 5715 retval = sctp_getsockopt_rtoinfo(sk, len, optval, optlen);
5714 break; 5716 break;
5715 case SCTP_ASSOCINFO: 5717 case SCTP_ASSOCINFO:
5716 retval = sctp_getsockopt_associnfo(sk, len, optval, optlen); 5718 retval = sctp_getsockopt_associnfo(sk, len, optval, optlen);
5717 break; 5719 break;
5718 case SCTP_I_WANT_MAPPED_V4_ADDR: 5720 case SCTP_I_WANT_MAPPED_V4_ADDR:
5719 retval = sctp_getsockopt_mappedv4(sk, len, optval, optlen); 5721 retval = sctp_getsockopt_mappedv4(sk, len, optval, optlen);
5720 break; 5722 break;
5721 case SCTP_MAXSEG: 5723 case SCTP_MAXSEG:
5722 retval = sctp_getsockopt_maxseg(sk, len, optval, optlen); 5724 retval = sctp_getsockopt_maxseg(sk, len, optval, optlen);
5723 break; 5725 break;
5724 case SCTP_GET_PEER_ADDR_INFO: 5726 case SCTP_GET_PEER_ADDR_INFO:
5725 retval = sctp_getsockopt_peer_addr_info(sk, len, optval, 5727 retval = sctp_getsockopt_peer_addr_info(sk, len, optval,
5726 optlen); 5728 optlen);
5727 break; 5729 break;
5728 case SCTP_ADAPTATION_LAYER: 5730 case SCTP_ADAPTATION_LAYER:
5729 retval = sctp_getsockopt_adaptation_layer(sk, len, optval, 5731 retval = sctp_getsockopt_adaptation_layer(sk, len, optval,
5730 optlen); 5732 optlen);
5731 break; 5733 break;
5732 case SCTP_CONTEXT: 5734 case SCTP_CONTEXT:
5733 retval = sctp_getsockopt_context(sk, len, optval, optlen); 5735 retval = sctp_getsockopt_context(sk, len, optval, optlen);
5734 break; 5736 break;
5735 case SCTP_FRAGMENT_INTERLEAVE: 5737 case SCTP_FRAGMENT_INTERLEAVE:
5736 retval = sctp_getsockopt_fragment_interleave(sk, len, optval, 5738 retval = sctp_getsockopt_fragment_interleave(sk, len, optval,
5737 optlen); 5739 optlen);
5738 break; 5740 break;
5739 case SCTP_PARTIAL_DELIVERY_POINT: 5741 case SCTP_PARTIAL_DELIVERY_POINT:
5740 retval = sctp_getsockopt_partial_delivery_point(sk, len, optval, 5742 retval = sctp_getsockopt_partial_delivery_point(sk, len, optval,
5741 optlen); 5743 optlen);
5742 break; 5744 break;
5743 case SCTP_MAX_BURST: 5745 case SCTP_MAX_BURST:
5744 retval = sctp_getsockopt_maxburst(sk, len, optval, optlen); 5746 retval = sctp_getsockopt_maxburst(sk, len, optval, optlen);
5745 break; 5747 break;
5746 case SCTP_AUTH_KEY: 5748 case SCTP_AUTH_KEY:
5747 case SCTP_AUTH_CHUNK: 5749 case SCTP_AUTH_CHUNK:
5748 case SCTP_AUTH_DELETE_KEY: 5750 case SCTP_AUTH_DELETE_KEY:
5749 retval = -EOPNOTSUPP; 5751 retval = -EOPNOTSUPP;
5750 break; 5752 break;
5751 case SCTP_HMAC_IDENT: 5753 case SCTP_HMAC_IDENT:
5752 retval = sctp_getsockopt_hmac_ident(sk, len, optval, optlen); 5754 retval = sctp_getsockopt_hmac_ident(sk, len, optval, optlen);
5753 break; 5755 break;
5754 case SCTP_AUTH_ACTIVE_KEY: 5756 case SCTP_AUTH_ACTIVE_KEY:
5755 retval = sctp_getsockopt_active_key(sk, len, optval, optlen); 5757 retval = sctp_getsockopt_active_key(sk, len, optval, optlen);
5756 break; 5758 break;
5757 case SCTP_PEER_AUTH_CHUNKS: 5759 case SCTP_PEER_AUTH_CHUNKS:
5758 retval = sctp_getsockopt_peer_auth_chunks(sk, len, optval, 5760 retval = sctp_getsockopt_peer_auth_chunks(sk, len, optval,
5759 optlen); 5761 optlen);
5760 break; 5762 break;
5761 case SCTP_LOCAL_AUTH_CHUNKS: 5763 case SCTP_LOCAL_AUTH_CHUNKS:
5762 retval = sctp_getsockopt_local_auth_chunks(sk, len, optval, 5764 retval = sctp_getsockopt_local_auth_chunks(sk, len, optval,
5763 optlen); 5765 optlen);
5764 break; 5766 break;
5765 case SCTP_GET_ASSOC_NUMBER: 5767 case SCTP_GET_ASSOC_NUMBER:
5766 retval = sctp_getsockopt_assoc_number(sk, len, optval, optlen); 5768 retval = sctp_getsockopt_assoc_number(sk, len, optval, optlen);
5767 break; 5769 break;
5768 case SCTP_GET_ASSOC_ID_LIST: 5770 case SCTP_GET_ASSOC_ID_LIST:
5769 retval = sctp_getsockopt_assoc_ids(sk, len, optval, optlen); 5771 retval = sctp_getsockopt_assoc_ids(sk, len, optval, optlen);
5770 break; 5772 break;
5771 case SCTP_AUTO_ASCONF: 5773 case SCTP_AUTO_ASCONF:
5772 retval = sctp_getsockopt_auto_asconf(sk, len, optval, optlen); 5774 retval = sctp_getsockopt_auto_asconf(sk, len, optval, optlen);
5773 break; 5775 break;
5774 case SCTP_PEER_ADDR_THLDS: 5776 case SCTP_PEER_ADDR_THLDS:
5775 retval = sctp_getsockopt_paddr_thresholds(sk, optval, len, optlen); 5777 retval = sctp_getsockopt_paddr_thresholds(sk, optval, len, optlen);
5776 break; 5778 break;
5777 default: 5779 default:
5778 retval = -ENOPROTOOPT; 5780 retval = -ENOPROTOOPT;
5779 break; 5781 break;
5780 } 5782 }
5781 5783
5782 sctp_release_sock(sk); 5784 sctp_release_sock(sk);
5783 return retval; 5785 return retval;
5784 } 5786 }
5785 5787
5786 static void sctp_hash(struct sock *sk) 5788 static void sctp_hash(struct sock *sk)
5787 { 5789 {
5788 /* STUB */ 5790 /* STUB */
5789 } 5791 }
5790 5792
5791 static void sctp_unhash(struct sock *sk) 5793 static void sctp_unhash(struct sock *sk)
5792 { 5794 {
5793 /* STUB */ 5795 /* STUB */
5794 } 5796 }
5795 5797
5796 /* Check if port is acceptable. Possibly find first available port. 5798 /* Check if port is acceptable. Possibly find first available port.
5797 * 5799 *
5798 * The port hash table (contained in the 'global' SCTP protocol storage 5800 * The port hash table (contained in the 'global' SCTP protocol storage
5799 * returned by struct sctp_protocol *sctp_get_protocol()). The hash 5801 * returned by struct sctp_protocol *sctp_get_protocol()). The hash
5800 * table is an array of 4096 lists (sctp_bind_hashbucket). Each 5802 * table is an array of 4096 lists (sctp_bind_hashbucket). Each
5801 * list (the list number is the port number hashed out, so as you 5803 * list (the list number is the port number hashed out, so as you
5802 * would expect from a hash function, all the ports in a given list have 5804 * would expect from a hash function, all the ports in a given list have
5803 * such a number that hashes out to the same list number; you were 5805 * such a number that hashes out to the same list number; you were
5804 * expecting that, right?); so each list has a set of ports, with a 5806 * expecting that, right?); so each list has a set of ports, with a
5805 * link to the socket (struct sock) that uses it, the port number and 5807 * link to the socket (struct sock) that uses it, the port number and
5806 * a fastreuse flag (FIXME: NPI ipg). 5808 * a fastreuse flag (FIXME: NPI ipg).
5807 */ 5809 */
5808 static struct sctp_bind_bucket *sctp_bucket_create( 5810 static struct sctp_bind_bucket *sctp_bucket_create(
5809 struct sctp_bind_hashbucket *head, struct net *, unsigned short snum); 5811 struct sctp_bind_hashbucket *head, struct net *, unsigned short snum);
5810 5812
5811 static long sctp_get_port_local(struct sock *sk, union sctp_addr *addr) 5813 static long sctp_get_port_local(struct sock *sk, union sctp_addr *addr)
5812 { 5814 {
5813 struct sctp_bind_hashbucket *head; /* hash list */ 5815 struct sctp_bind_hashbucket *head; /* hash list */
5814 struct sctp_bind_bucket *pp; /* hash list port iterator */ 5816 struct sctp_bind_bucket *pp; /* hash list port iterator */
5815 struct hlist_node *node; 5817 struct hlist_node *node;
5816 unsigned short snum; 5818 unsigned short snum;
5817 int ret; 5819 int ret;
5818 5820
5819 snum = ntohs(addr->v4.sin_port); 5821 snum = ntohs(addr->v4.sin_port);
5820 5822
5821 SCTP_DEBUG_PRINTK("sctp_get_port() begins, snum=%d\n", snum); 5823 SCTP_DEBUG_PRINTK("sctp_get_port() begins, snum=%d\n", snum);
5822 sctp_local_bh_disable(); 5824 sctp_local_bh_disable();
5823 5825
5824 if (snum == 0) { 5826 if (snum == 0) {
5825 /* Search for an available port. */ 5827 /* Search for an available port. */
5826 int low, high, remaining, index; 5828 int low, high, remaining, index;
5827 unsigned int rover; 5829 unsigned int rover;
5828 5830
5829 inet_get_local_port_range(&low, &high); 5831 inet_get_local_port_range(&low, &high);
5830 remaining = (high - low) + 1; 5832 remaining = (high - low) + 1;
5831 rover = net_random() % remaining + low; 5833 rover = net_random() % remaining + low;
5832 5834
5833 do { 5835 do {
5834 rover++; 5836 rover++;
5835 if ((rover < low) || (rover > high)) 5837 if ((rover < low) || (rover > high))
5836 rover = low; 5838 rover = low;
5837 if (inet_is_reserved_local_port(rover)) 5839 if (inet_is_reserved_local_port(rover))
5838 continue; 5840 continue;
5839 index = sctp_phashfn(sock_net(sk), rover); 5841 index = sctp_phashfn(sock_net(sk), rover);
5840 head = &sctp_port_hashtable[index]; 5842 head = &sctp_port_hashtable[index];
5841 sctp_spin_lock(&head->lock); 5843 sctp_spin_lock(&head->lock);
5842 sctp_for_each_hentry(pp, node, &head->chain) 5844 sctp_for_each_hentry(pp, node, &head->chain)
5843 if ((pp->port == rover) && 5845 if ((pp->port == rover) &&
5844 net_eq(sock_net(sk), pp->net)) 5846 net_eq(sock_net(sk), pp->net))
5845 goto next; 5847 goto next;
5846 break; 5848 break;
5847 next: 5849 next:
5848 sctp_spin_unlock(&head->lock); 5850 sctp_spin_unlock(&head->lock);
5849 } while (--remaining > 0); 5851 } while (--remaining > 0);
5850 5852
5851 /* Exhausted local port range during search? */ 5853 /* Exhausted local port range during search? */
5852 ret = 1; 5854 ret = 1;
5853 if (remaining <= 0) 5855 if (remaining <= 0)
5854 goto fail; 5856 goto fail;
5855 5857
5856 /* OK, here is the one we will use. HEAD (the port 5858 /* OK, here is the one we will use. HEAD (the port
5857 * hash table list entry) is non-NULL and we hold it's 5859 * hash table list entry) is non-NULL and we hold it's
5858 * mutex. 5860 * mutex.
5859 */ 5861 */
5860 snum = rover; 5862 snum = rover;
5861 } else { 5863 } else {
5862 /* We are given an specific port number; we verify 5864 /* We are given an specific port number; we verify
5863 * that it is not being used. If it is used, we will 5865 * that it is not being used. If it is used, we will
5864 * exahust the search in the hash list corresponding 5866 * exahust the search in the hash list corresponding
5865 * to the port number (snum) - we detect that with the 5867 * to the port number (snum) - we detect that with the
5866 * port iterator, pp being NULL. 5868 * port iterator, pp being NULL.
5867 */ 5869 */
5868 head = &sctp_port_hashtable[sctp_phashfn(sock_net(sk), snum)]; 5870 head = &sctp_port_hashtable[sctp_phashfn(sock_net(sk), snum)];
5869 sctp_spin_lock(&head->lock); 5871 sctp_spin_lock(&head->lock);
5870 sctp_for_each_hentry(pp, node, &head->chain) { 5872 sctp_for_each_hentry(pp, node, &head->chain) {
5871 if ((pp->port == snum) && net_eq(pp->net, sock_net(sk))) 5873 if ((pp->port == snum) && net_eq(pp->net, sock_net(sk)))
5872 goto pp_found; 5874 goto pp_found;
5873 } 5875 }
5874 } 5876 }
5875 pp = NULL; 5877 pp = NULL;
5876 goto pp_not_found; 5878 goto pp_not_found;
5877 pp_found: 5879 pp_found:
5878 if (!hlist_empty(&pp->owner)) { 5880 if (!hlist_empty(&pp->owner)) {
5879 /* We had a port hash table hit - there is an 5881 /* We had a port hash table hit - there is an
5880 * available port (pp != NULL) and it is being 5882 * available port (pp != NULL) and it is being
5881 * used by other socket (pp->owner not empty); that other 5883 * used by other socket (pp->owner not empty); that other
5882 * socket is going to be sk2. 5884 * socket is going to be sk2.
5883 */ 5885 */
5884 int reuse = sk->sk_reuse; 5886 int reuse = sk->sk_reuse;
5885 struct sock *sk2; 5887 struct sock *sk2;
5886 5888
5887 SCTP_DEBUG_PRINTK("sctp_get_port() found a possible match\n"); 5889 SCTP_DEBUG_PRINTK("sctp_get_port() found a possible match\n");
5888 if (pp->fastreuse && sk->sk_reuse && 5890 if (pp->fastreuse && sk->sk_reuse &&
5889 sk->sk_state != SCTP_SS_LISTENING) 5891 sk->sk_state != SCTP_SS_LISTENING)
5890 goto success; 5892 goto success;
5891 5893
5892 /* Run through the list of sockets bound to the port 5894 /* Run through the list of sockets bound to the port
5893 * (pp->port) [via the pointers bind_next and 5895 * (pp->port) [via the pointers bind_next and
5894 * bind_pprev in the struct sock *sk2 (pp->sk)]. On each one, 5896 * bind_pprev in the struct sock *sk2 (pp->sk)]. On each one,
5895 * we get the endpoint they describe and run through 5897 * we get the endpoint they describe and run through
5896 * the endpoint's list of IP (v4 or v6) addresses, 5898 * the endpoint's list of IP (v4 or v6) addresses,
5897 * comparing each of the addresses with the address of 5899 * comparing each of the addresses with the address of
5898 * the socket sk. If we find a match, then that means 5900 * the socket sk. If we find a match, then that means
5899 * that this port/socket (sk) combination are already 5901 * that this port/socket (sk) combination are already
5900 * in an endpoint. 5902 * in an endpoint.
5901 */ 5903 */
5902 sk_for_each_bound(sk2, node, &pp->owner) { 5904 sk_for_each_bound(sk2, node, &pp->owner) {
5903 struct sctp_endpoint *ep2; 5905 struct sctp_endpoint *ep2;
5904 ep2 = sctp_sk(sk2)->ep; 5906 ep2 = sctp_sk(sk2)->ep;
5905 5907
5906 if (sk == sk2 || 5908 if (sk == sk2 ||
5907 (reuse && sk2->sk_reuse && 5909 (reuse && sk2->sk_reuse &&
5908 sk2->sk_state != SCTP_SS_LISTENING)) 5910 sk2->sk_state != SCTP_SS_LISTENING))
5909 continue; 5911 continue;
5910 5912
5911 if (sctp_bind_addr_conflict(&ep2->base.bind_addr, addr, 5913 if (sctp_bind_addr_conflict(&ep2->base.bind_addr, addr,
5912 sctp_sk(sk2), sctp_sk(sk))) { 5914 sctp_sk(sk2), sctp_sk(sk))) {
5913 ret = (long)sk2; 5915 ret = (long)sk2;
5914 goto fail_unlock; 5916 goto fail_unlock;
5915 } 5917 }
5916 } 5918 }
5917 SCTP_DEBUG_PRINTK("sctp_get_port(): Found a match\n"); 5919 SCTP_DEBUG_PRINTK("sctp_get_port(): Found a match\n");
5918 } 5920 }
5919 pp_not_found: 5921 pp_not_found:
5920 /* If there was a hash table miss, create a new port. */ 5922 /* If there was a hash table miss, create a new port. */
5921 ret = 1; 5923 ret = 1;
5922 if (!pp && !(pp = sctp_bucket_create(head, sock_net(sk), snum))) 5924 if (!pp && !(pp = sctp_bucket_create(head, sock_net(sk), snum)))
5923 goto fail_unlock; 5925 goto fail_unlock;
5924 5926
5925 /* In either case (hit or miss), make sure fastreuse is 1 only 5927 /* In either case (hit or miss), make sure fastreuse is 1 only
5926 * if sk->sk_reuse is too (that is, if the caller requested 5928 * if sk->sk_reuse is too (that is, if the caller requested
5927 * SO_REUSEADDR on this socket -sk-). 5929 * SO_REUSEADDR on this socket -sk-).
5928 */ 5930 */
5929 if (hlist_empty(&pp->owner)) { 5931 if (hlist_empty(&pp->owner)) {
5930 if (sk->sk_reuse && sk->sk_state != SCTP_SS_LISTENING) 5932 if (sk->sk_reuse && sk->sk_state != SCTP_SS_LISTENING)
5931 pp->fastreuse = 1; 5933 pp->fastreuse = 1;
5932 else 5934 else
5933 pp->fastreuse = 0; 5935 pp->fastreuse = 0;
5934 } else if (pp->fastreuse && 5936 } else if (pp->fastreuse &&
5935 (!sk->sk_reuse || sk->sk_state == SCTP_SS_LISTENING)) 5937 (!sk->sk_reuse || sk->sk_state == SCTP_SS_LISTENING))
5936 pp->fastreuse = 0; 5938 pp->fastreuse = 0;
5937 5939
5938 /* We are set, so fill up all the data in the hash table 5940 /* We are set, so fill up all the data in the hash table
5939 * entry, tie the socket list information with the rest of the 5941 * entry, tie the socket list information with the rest of the
5940 * sockets FIXME: Blurry, NPI (ipg). 5942 * sockets FIXME: Blurry, NPI (ipg).
5941 */ 5943 */
5942 success: 5944 success:
5943 if (!sctp_sk(sk)->bind_hash) { 5945 if (!sctp_sk(sk)->bind_hash) {
5944 inet_sk(sk)->inet_num = snum; 5946 inet_sk(sk)->inet_num = snum;
5945 sk_add_bind_node(sk, &pp->owner); 5947 sk_add_bind_node(sk, &pp->owner);
5946 sctp_sk(sk)->bind_hash = pp; 5948 sctp_sk(sk)->bind_hash = pp;
5947 } 5949 }
5948 ret = 0; 5950 ret = 0;
5949 5951
5950 fail_unlock: 5952 fail_unlock:
5951 sctp_spin_unlock(&head->lock); 5953 sctp_spin_unlock(&head->lock);
5952 5954
5953 fail: 5955 fail:
5954 sctp_local_bh_enable(); 5956 sctp_local_bh_enable();
5955 return ret; 5957 return ret;
5956 } 5958 }
5957 5959
5958 /* Assign a 'snum' port to the socket. If snum == 0, an ephemeral 5960 /* Assign a 'snum' port to the socket. If snum == 0, an ephemeral
5959 * port is requested. 5961 * port is requested.
5960 */ 5962 */
5961 static int sctp_get_port(struct sock *sk, unsigned short snum) 5963 static int sctp_get_port(struct sock *sk, unsigned short snum)
5962 { 5964 {
5963 long ret; 5965 long ret;
5964 union sctp_addr addr; 5966 union sctp_addr addr;
5965 struct sctp_af *af = sctp_sk(sk)->pf->af; 5967 struct sctp_af *af = sctp_sk(sk)->pf->af;
5966 5968
5967 /* Set up a dummy address struct from the sk. */ 5969 /* Set up a dummy address struct from the sk. */
5968 af->from_sk(&addr, sk); 5970 af->from_sk(&addr, sk);
5969 addr.v4.sin_port = htons(snum); 5971 addr.v4.sin_port = htons(snum);
5970 5972
5971 /* Note: sk->sk_num gets filled in if ephemeral port request. */ 5973 /* Note: sk->sk_num gets filled in if ephemeral port request. */
5972 ret = sctp_get_port_local(sk, &addr); 5974 ret = sctp_get_port_local(sk, &addr);
5973 5975
5974 return ret ? 1 : 0; 5976 return ret ? 1 : 0;
5975 } 5977 }
5976 5978
5977 /* 5979 /*
5978 * Move a socket to LISTENING state. 5980 * Move a socket to LISTENING state.
5979 */ 5981 */
5980 SCTP_STATIC int sctp_listen_start(struct sock *sk, int backlog) 5982 SCTP_STATIC int sctp_listen_start(struct sock *sk, int backlog)
5981 { 5983 {
5982 struct sctp_sock *sp = sctp_sk(sk); 5984 struct sctp_sock *sp = sctp_sk(sk);
5983 struct sctp_endpoint *ep = sp->ep; 5985 struct sctp_endpoint *ep = sp->ep;
5984 struct crypto_hash *tfm = NULL; 5986 struct crypto_hash *tfm = NULL;
5985 char alg[32]; 5987 char alg[32];
5986 5988
5987 /* Allocate HMAC for generating cookie. */ 5989 /* Allocate HMAC for generating cookie. */
5988 if (!sp->hmac && sp->sctp_hmac_alg) { 5990 if (!sp->hmac && sp->sctp_hmac_alg) {
5989 sprintf(alg, "hmac(%s)", sp->sctp_hmac_alg); 5991 sprintf(alg, "hmac(%s)", sp->sctp_hmac_alg);
5990 tfm = crypto_alloc_hash(alg, 0, CRYPTO_ALG_ASYNC); 5992 tfm = crypto_alloc_hash(alg, 0, CRYPTO_ALG_ASYNC);
5991 if (IS_ERR(tfm)) { 5993 if (IS_ERR(tfm)) {
5992 net_info_ratelimited("failed to load transform for %s: %ld\n", 5994 net_info_ratelimited("failed to load transform for %s: %ld\n",
5993 sp->sctp_hmac_alg, PTR_ERR(tfm)); 5995 sp->sctp_hmac_alg, PTR_ERR(tfm));
5994 return -ENOSYS; 5996 return -ENOSYS;
5995 } 5997 }
5996 sctp_sk(sk)->hmac = tfm; 5998 sctp_sk(sk)->hmac = tfm;
5997 } 5999 }
5998 6000
5999 /* 6001 /*
6000 * If a bind() or sctp_bindx() is not called prior to a listen() 6002 * If a bind() or sctp_bindx() is not called prior to a listen()
6001 * call that allows new associations to be accepted, the system 6003 * call that allows new associations to be accepted, the system
6002 * picks an ephemeral port and will choose an address set equivalent 6004 * picks an ephemeral port and will choose an address set equivalent
6003 * to binding with a wildcard address. 6005 * to binding with a wildcard address.
6004 * 6006 *
6005 * This is not currently spelled out in the SCTP sockets 6007 * This is not currently spelled out in the SCTP sockets
6006 * extensions draft, but follows the practice as seen in TCP 6008 * extensions draft, but follows the practice as seen in TCP
6007 * sockets. 6009 * sockets.
6008 * 6010 *
6009 */ 6011 */
6010 sk->sk_state = SCTP_SS_LISTENING; 6012 sk->sk_state = SCTP_SS_LISTENING;
6011 if (!ep->base.bind_addr.port) { 6013 if (!ep->base.bind_addr.port) {
6012 if (sctp_autobind(sk)) 6014 if (sctp_autobind(sk))
6013 return -EAGAIN; 6015 return -EAGAIN;
6014 } else { 6016 } else {
6015 if (sctp_get_port(sk, inet_sk(sk)->inet_num)) { 6017 if (sctp_get_port(sk, inet_sk(sk)->inet_num)) {
6016 sk->sk_state = SCTP_SS_CLOSED; 6018 sk->sk_state = SCTP_SS_CLOSED;
6017 return -EADDRINUSE; 6019 return -EADDRINUSE;
6018 } 6020 }
6019 } 6021 }
6020 6022
6021 sk->sk_max_ack_backlog = backlog; 6023 sk->sk_max_ack_backlog = backlog;
6022 sctp_hash_endpoint(ep); 6024 sctp_hash_endpoint(ep);
6023 return 0; 6025 return 0;
6024 } 6026 }
6025 6027
6026 /* 6028 /*
6027 * 4.1.3 / 5.1.3 listen() 6029 * 4.1.3 / 5.1.3 listen()
6028 * 6030 *
6029 * By default, new associations are not accepted for UDP style sockets. 6031 * By default, new associations are not accepted for UDP style sockets.
6030 * An application uses listen() to mark a socket as being able to 6032 * An application uses listen() to mark a socket as being able to
6031 * accept new associations. 6033 * accept new associations.
6032 * 6034 *
6033 * On TCP style sockets, applications use listen() to ready the SCTP 6035 * On TCP style sockets, applications use listen() to ready the SCTP
6034 * endpoint for accepting inbound associations. 6036 * endpoint for accepting inbound associations.
6035 * 6037 *
6036 * On both types of endpoints a backlog of '0' disables listening. 6038 * On both types of endpoints a backlog of '0' disables listening.
6037 * 6039 *
6038 * Move a socket to LISTENING state. 6040 * Move a socket to LISTENING state.
6039 */ 6041 */
6040 int sctp_inet_listen(struct socket *sock, int backlog) 6042 int sctp_inet_listen(struct socket *sock, int backlog)
6041 { 6043 {
6042 struct sock *sk = sock->sk; 6044 struct sock *sk = sock->sk;
6043 struct sctp_endpoint *ep = sctp_sk(sk)->ep; 6045 struct sctp_endpoint *ep = sctp_sk(sk)->ep;
6044 int err = -EINVAL; 6046 int err = -EINVAL;
6045 6047
6046 if (unlikely(backlog < 0)) 6048 if (unlikely(backlog < 0))
6047 return err; 6049 return err;
6048 6050
6049 sctp_lock_sock(sk); 6051 sctp_lock_sock(sk);
6050 6052
6051 /* Peeled-off sockets are not allowed to listen(). */ 6053 /* Peeled-off sockets are not allowed to listen(). */
6052 if (sctp_style(sk, UDP_HIGH_BANDWIDTH)) 6054 if (sctp_style(sk, UDP_HIGH_BANDWIDTH))
6053 goto out; 6055 goto out;
6054 6056
6055 if (sock->state != SS_UNCONNECTED) 6057 if (sock->state != SS_UNCONNECTED)
6056 goto out; 6058 goto out;
6057 6059
6058 /* If backlog is zero, disable listening. */ 6060 /* If backlog is zero, disable listening. */
6059 if (!backlog) { 6061 if (!backlog) {
6060 if (sctp_sstate(sk, CLOSED)) 6062 if (sctp_sstate(sk, CLOSED))
6061 goto out; 6063 goto out;
6062 6064
6063 err = 0; 6065 err = 0;
6064 sctp_unhash_endpoint(ep); 6066 sctp_unhash_endpoint(ep);
6065 sk->sk_state = SCTP_SS_CLOSED; 6067 sk->sk_state = SCTP_SS_CLOSED;
6066 if (sk->sk_reuse) 6068 if (sk->sk_reuse)
6067 sctp_sk(sk)->bind_hash->fastreuse = 1; 6069 sctp_sk(sk)->bind_hash->fastreuse = 1;
6068 goto out; 6070 goto out;
6069 } 6071 }
6070 6072
6071 /* If we are already listening, just update the backlog */ 6073 /* If we are already listening, just update the backlog */
6072 if (sctp_sstate(sk, LISTENING)) 6074 if (sctp_sstate(sk, LISTENING))
6073 sk->sk_max_ack_backlog = backlog; 6075 sk->sk_max_ack_backlog = backlog;
6074 else { 6076 else {
6075 err = sctp_listen_start(sk, backlog); 6077 err = sctp_listen_start(sk, backlog);
6076 if (err) 6078 if (err)
6077 goto out; 6079 goto out;
6078 } 6080 }
6079 6081
6080 err = 0; 6082 err = 0;
6081 out: 6083 out:
6082 sctp_release_sock(sk); 6084 sctp_release_sock(sk);
6083 return err; 6085 return err;
6084 } 6086 }
6085 6087
6086 /* 6088 /*
6087 * This function is done by modeling the current datagram_poll() and the 6089 * This function is done by modeling the current datagram_poll() and the
6088 * tcp_poll(). Note that, based on these implementations, we don't 6090 * tcp_poll(). Note that, based on these implementations, we don't
6089 * lock the socket in this function, even though it seems that, 6091 * lock the socket in this function, even though it seems that,
6090 * ideally, locking or some other mechanisms can be used to ensure 6092 * ideally, locking or some other mechanisms can be used to ensure
6091 * the integrity of the counters (sndbuf and wmem_alloc) used 6093 * the integrity of the counters (sndbuf and wmem_alloc) used
6092 * in this place. We assume that we don't need locks either until proven 6094 * in this place. We assume that we don't need locks either until proven
6093 * otherwise. 6095 * otherwise.
6094 * 6096 *
6095 * Another thing to note is that we include the Async I/O support 6097 * Another thing to note is that we include the Async I/O support
6096 * here, again, by modeling the current TCP/UDP code. We don't have 6098 * here, again, by modeling the current TCP/UDP code. We don't have
6097 * a good way to test with it yet. 6099 * a good way to test with it yet.
6098 */ 6100 */
6099 unsigned int sctp_poll(struct file *file, struct socket *sock, poll_table *wait) 6101 unsigned int sctp_poll(struct file *file, struct socket *sock, poll_table *wait)
6100 { 6102 {
6101 struct sock *sk = sock->sk; 6103 struct sock *sk = sock->sk;
6102 struct sctp_sock *sp = sctp_sk(sk); 6104 struct sctp_sock *sp = sctp_sk(sk);
6103 unsigned int mask; 6105 unsigned int mask;
6104 6106
6105 poll_wait(file, sk_sleep(sk), wait); 6107 poll_wait(file, sk_sleep(sk), wait);
6106 6108
6107 /* A TCP-style listening socket becomes readable when the accept queue 6109 /* A TCP-style listening socket becomes readable when the accept queue
6108 * is not empty. 6110 * is not empty.
6109 */ 6111 */
6110 if (sctp_style(sk, TCP) && sctp_sstate(sk, LISTENING)) 6112 if (sctp_style(sk, TCP) && sctp_sstate(sk, LISTENING))
6111 return (!list_empty(&sp->ep->asocs)) ? 6113 return (!list_empty(&sp->ep->asocs)) ?
6112 (POLLIN | POLLRDNORM) : 0; 6114 (POLLIN | POLLRDNORM) : 0;
6113 6115
6114 mask = 0; 6116 mask = 0;
6115 6117
6116 /* Is there any exceptional events? */ 6118 /* Is there any exceptional events? */
6117 if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue)) 6119 if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
6118 mask |= POLLERR; 6120 mask |= POLLERR;
6119 if (sk->sk_shutdown & RCV_SHUTDOWN) 6121 if (sk->sk_shutdown & RCV_SHUTDOWN)
6120 mask |= POLLRDHUP | POLLIN | POLLRDNORM; 6122 mask |= POLLRDHUP | POLLIN | POLLRDNORM;
6121 if (sk->sk_shutdown == SHUTDOWN_MASK) 6123 if (sk->sk_shutdown == SHUTDOWN_MASK)
6122 mask |= POLLHUP; 6124 mask |= POLLHUP;
6123 6125
6124 /* Is it readable? Reconsider this code with TCP-style support. */ 6126 /* Is it readable? Reconsider this code with TCP-style support. */
6125 if (!skb_queue_empty(&sk->sk_receive_queue)) 6127 if (!skb_queue_empty(&sk->sk_receive_queue))
6126 mask |= POLLIN | POLLRDNORM; 6128 mask |= POLLIN | POLLRDNORM;
6127 6129
6128 /* The association is either gone or not ready. */ 6130 /* The association is either gone or not ready. */
6129 if (!sctp_style(sk, UDP) && sctp_sstate(sk, CLOSED)) 6131 if (!sctp_style(sk, UDP) && sctp_sstate(sk, CLOSED))
6130 return mask; 6132 return mask;
6131 6133
6132 /* Is it writable? */ 6134 /* Is it writable? */
6133 if (sctp_writeable(sk)) { 6135 if (sctp_writeable(sk)) {
6134 mask |= POLLOUT | POLLWRNORM; 6136 mask |= POLLOUT | POLLWRNORM;
6135 } else { 6137 } else {
6136 set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags); 6138 set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);
6137 /* 6139 /*
6138 * Since the socket is not locked, the buffer 6140 * Since the socket is not locked, the buffer
6139 * might be made available after the writeable check and 6141 * might be made available after the writeable check and
6140 * before the bit is set. This could cause a lost I/O 6142 * before the bit is set. This could cause a lost I/O
6141 * signal. tcp_poll() has a race breaker for this race 6143 * signal. tcp_poll() has a race breaker for this race
6142 * condition. Based on their implementation, we put 6144 * condition. Based on their implementation, we put
6143 * in the following code to cover it as well. 6145 * in the following code to cover it as well.
6144 */ 6146 */
6145 if (sctp_writeable(sk)) 6147 if (sctp_writeable(sk))
6146 mask |= POLLOUT | POLLWRNORM; 6148 mask |= POLLOUT | POLLWRNORM;
6147 } 6149 }
6148 return mask; 6150 return mask;
6149 } 6151 }
6150 6152
6151 /******************************************************************** 6153 /********************************************************************
6152 * 2nd Level Abstractions 6154 * 2nd Level Abstractions
6153 ********************************************************************/ 6155 ********************************************************************/
6154 6156
6155 static struct sctp_bind_bucket *sctp_bucket_create( 6157 static struct sctp_bind_bucket *sctp_bucket_create(
6156 struct sctp_bind_hashbucket *head, struct net *net, unsigned short snum) 6158 struct sctp_bind_hashbucket *head, struct net *net, unsigned short snum)
6157 { 6159 {
6158 struct sctp_bind_bucket *pp; 6160 struct sctp_bind_bucket *pp;
6159 6161
6160 pp = kmem_cache_alloc(sctp_bucket_cachep, GFP_ATOMIC); 6162 pp = kmem_cache_alloc(sctp_bucket_cachep, GFP_ATOMIC);
6161 if (pp) { 6163 if (pp) {
6162 SCTP_DBG_OBJCNT_INC(bind_bucket); 6164 SCTP_DBG_OBJCNT_INC(bind_bucket);
6163 pp->port = snum; 6165 pp->port = snum;
6164 pp->fastreuse = 0; 6166 pp->fastreuse = 0;
6165 INIT_HLIST_HEAD(&pp->owner); 6167 INIT_HLIST_HEAD(&pp->owner);
6166 pp->net = net; 6168 pp->net = net;
6167 hlist_add_head(&pp->node, &head->chain); 6169 hlist_add_head(&pp->node, &head->chain);
6168 } 6170 }
6169 return pp; 6171 return pp;
6170 } 6172 }
6171 6173
6172 /* Caller must hold hashbucket lock for this tb with local BH disabled */ 6174 /* Caller must hold hashbucket lock for this tb with local BH disabled */
6173 static void sctp_bucket_destroy(struct sctp_bind_bucket *pp) 6175 static void sctp_bucket_destroy(struct sctp_bind_bucket *pp)
6174 { 6176 {
6175 if (pp && hlist_empty(&pp->owner)) { 6177 if (pp && hlist_empty(&pp->owner)) {
6176 __hlist_del(&pp->node); 6178 __hlist_del(&pp->node);
6177 kmem_cache_free(sctp_bucket_cachep, pp); 6179 kmem_cache_free(sctp_bucket_cachep, pp);
6178 SCTP_DBG_OBJCNT_DEC(bind_bucket); 6180 SCTP_DBG_OBJCNT_DEC(bind_bucket);
6179 } 6181 }
6180 } 6182 }
6181 6183
6182 /* Release this socket's reference to a local port. */ 6184 /* Release this socket's reference to a local port. */
6183 static inline void __sctp_put_port(struct sock *sk) 6185 static inline void __sctp_put_port(struct sock *sk)
6184 { 6186 {
6185 struct sctp_bind_hashbucket *head = 6187 struct sctp_bind_hashbucket *head =
6186 &sctp_port_hashtable[sctp_phashfn(sock_net(sk), 6188 &sctp_port_hashtable[sctp_phashfn(sock_net(sk),
6187 inet_sk(sk)->inet_num)]; 6189 inet_sk(sk)->inet_num)];
6188 struct sctp_bind_bucket *pp; 6190 struct sctp_bind_bucket *pp;
6189 6191
6190 sctp_spin_lock(&head->lock); 6192 sctp_spin_lock(&head->lock);
6191 pp = sctp_sk(sk)->bind_hash; 6193 pp = sctp_sk(sk)->bind_hash;
6192 __sk_del_bind_node(sk); 6194 __sk_del_bind_node(sk);
6193 sctp_sk(sk)->bind_hash = NULL; 6195 sctp_sk(sk)->bind_hash = NULL;
6194 inet_sk(sk)->inet_num = 0; 6196 inet_sk(sk)->inet_num = 0;
6195 sctp_bucket_destroy(pp); 6197 sctp_bucket_destroy(pp);
6196 sctp_spin_unlock(&head->lock); 6198 sctp_spin_unlock(&head->lock);
6197 } 6199 }
6198 6200
6199 void sctp_put_port(struct sock *sk) 6201 void sctp_put_port(struct sock *sk)
6200 { 6202 {
6201 sctp_local_bh_disable(); 6203 sctp_local_bh_disable();
6202 __sctp_put_port(sk); 6204 __sctp_put_port(sk);
6203 sctp_local_bh_enable(); 6205 sctp_local_bh_enable();
6204 } 6206 }
6205 6207
6206 /* 6208 /*
6207 * The system picks an ephemeral port and choose an address set equivalent 6209 * The system picks an ephemeral port and choose an address set equivalent
6208 * to binding with a wildcard address. 6210 * to binding with a wildcard address.
6209 * One of those addresses will be the primary address for the association. 6211 * One of those addresses will be the primary address for the association.
6210 * This automatically enables the multihoming capability of SCTP. 6212 * This automatically enables the multihoming capability of SCTP.
6211 */ 6213 */
6212 static int sctp_autobind(struct sock *sk) 6214 static int sctp_autobind(struct sock *sk)
6213 { 6215 {
6214 union sctp_addr autoaddr; 6216 union sctp_addr autoaddr;
6215 struct sctp_af *af; 6217 struct sctp_af *af;
6216 __be16 port; 6218 __be16 port;
6217 6219
6218 /* Initialize a local sockaddr structure to INADDR_ANY. */ 6220 /* Initialize a local sockaddr structure to INADDR_ANY. */
6219 af = sctp_sk(sk)->pf->af; 6221 af = sctp_sk(sk)->pf->af;
6220 6222
6221 port = htons(inet_sk(sk)->inet_num); 6223 port = htons(inet_sk(sk)->inet_num);
6222 af->inaddr_any(&autoaddr, port); 6224 af->inaddr_any(&autoaddr, port);
6223 6225
6224 return sctp_do_bind(sk, &autoaddr, af->sockaddr_len); 6226 return sctp_do_bind(sk, &autoaddr, af->sockaddr_len);
6225 } 6227 }
6226 6228
6227 /* Parse out IPPROTO_SCTP CMSG headers. Perform only minimal validation. 6229 /* Parse out IPPROTO_SCTP CMSG headers. Perform only minimal validation.
6228 * 6230 *
6229 * From RFC 2292 6231 * From RFC 2292
6230 * 4.2 The cmsghdr Structure * 6232 * 4.2 The cmsghdr Structure *
6231 * 6233 *
6232 * When ancillary data is sent or received, any number of ancillary data 6234 * When ancillary data is sent or received, any number of ancillary data
6233 * objects can be specified by the msg_control and msg_controllen members of 6235 * objects can be specified by the msg_control and msg_controllen members of
6234 * the msghdr structure, because each object is preceded by 6236 * the msghdr structure, because each object is preceded by
6235 * a cmsghdr structure defining the object's length (the cmsg_len member). 6237 * a cmsghdr structure defining the object's length (the cmsg_len member).
6236 * Historically Berkeley-derived implementations have passed only one object 6238 * Historically Berkeley-derived implementations have passed only one object
6237 * at a time, but this API allows multiple objects to be 6239 * at a time, but this API allows multiple objects to be
6238 * passed in a single call to sendmsg() or recvmsg(). The following example 6240 * passed in a single call to sendmsg() or recvmsg(). The following example
6239 * shows two ancillary data objects in a control buffer. 6241 * shows two ancillary data objects in a control buffer.
6240 * 6242 *
6241 * |<--------------------------- msg_controllen -------------------------->| 6243 * |<--------------------------- msg_controllen -------------------------->|
6242 * | | 6244 * | |
6243 * 6245 *
6244 * |<----- ancillary data object ----->|<----- ancillary data object ----->| 6246 * |<----- ancillary data object ----->|<----- ancillary data object ----->|
6245 * 6247 *
6246 * |<---------- CMSG_SPACE() --------->|<---------- CMSG_SPACE() --------->| 6248 * |<---------- CMSG_SPACE() --------->|<---------- CMSG_SPACE() --------->|
6247 * | | | 6249 * | | |
6248 * 6250 *
6249 * |<---------- cmsg_len ---------->| |<--------- cmsg_len ----------->| | 6251 * |<---------- cmsg_len ---------->| |<--------- cmsg_len ----------->| |
6250 * 6252 *
6251 * |<--------- CMSG_LEN() --------->| |<-------- CMSG_LEN() ---------->| | 6253 * |<--------- CMSG_LEN() --------->| |<-------- CMSG_LEN() ---------->| |
6252 * | | | | | 6254 * | | | | |
6253 * 6255 *
6254 * +-----+-----+-----+--+-----------+--+-----+-----+-----+--+-----------+--+ 6256 * +-----+-----+-----+--+-----------+--+-----+-----+-----+--+-----------+--+
6255 * |cmsg_|cmsg_|cmsg_|XX| |XX|cmsg_|cmsg_|cmsg_|XX| |XX| 6257 * |cmsg_|cmsg_|cmsg_|XX| |XX|cmsg_|cmsg_|cmsg_|XX| |XX|
6256 * 6258 *
6257 * |len |level|type |XX|cmsg_data[]|XX|len |level|type |XX|cmsg_data[]|XX| 6259 * |len |level|type |XX|cmsg_data[]|XX|len |level|type |XX|cmsg_data[]|XX|
6258 * 6260 *
6259 * +-----+-----+-----+--+-----------+--+-----+-----+-----+--+-----------+--+ 6261 * +-----+-----+-----+--+-----------+--+-----+-----+-----+--+-----------+--+
6260 * ^ 6262 * ^
6261 * | 6263 * |
6262 * 6264 *
6263 * msg_control 6265 * msg_control
6264 * points here 6266 * points here
6265 */ 6267 */
6266 SCTP_STATIC int sctp_msghdr_parse(const struct msghdr *msg, 6268 SCTP_STATIC int sctp_msghdr_parse(const struct msghdr *msg,
6267 sctp_cmsgs_t *cmsgs) 6269 sctp_cmsgs_t *cmsgs)
6268 { 6270 {
6269 struct cmsghdr *cmsg; 6271 struct cmsghdr *cmsg;
6270 struct msghdr *my_msg = (struct msghdr *)msg; 6272 struct msghdr *my_msg = (struct msghdr *)msg;
6271 6273
6272 for (cmsg = CMSG_FIRSTHDR(msg); 6274 for (cmsg = CMSG_FIRSTHDR(msg);
6273 cmsg != NULL; 6275 cmsg != NULL;
6274 cmsg = CMSG_NXTHDR(my_msg, cmsg)) { 6276 cmsg = CMSG_NXTHDR(my_msg, cmsg)) {
6275 if (!CMSG_OK(my_msg, cmsg)) 6277 if (!CMSG_OK(my_msg, cmsg))
6276 return -EINVAL; 6278 return -EINVAL;
6277 6279
6278 /* Should we parse this header or ignore? */ 6280 /* Should we parse this header or ignore? */
6279 if (cmsg->cmsg_level != IPPROTO_SCTP) 6281 if (cmsg->cmsg_level != IPPROTO_SCTP)
6280 continue; 6282 continue;
6281 6283
6282 /* Strictly check lengths following example in SCM code. */ 6284 /* Strictly check lengths following example in SCM code. */
6283 switch (cmsg->cmsg_type) { 6285 switch (cmsg->cmsg_type) {
6284 case SCTP_INIT: 6286 case SCTP_INIT:
6285 /* SCTP Socket API Extension 6287 /* SCTP Socket API Extension
6286 * 5.2.1 SCTP Initiation Structure (SCTP_INIT) 6288 * 5.2.1 SCTP Initiation Structure (SCTP_INIT)
6287 * 6289 *
6288 * This cmsghdr structure provides information for 6290 * This cmsghdr structure provides information for
6289 * initializing new SCTP associations with sendmsg(). 6291 * initializing new SCTP associations with sendmsg().
6290 * The SCTP_INITMSG socket option uses this same data 6292 * The SCTP_INITMSG socket option uses this same data
6291 * structure. This structure is not used for 6293 * structure. This structure is not used for
6292 * recvmsg(). 6294 * recvmsg().
6293 * 6295 *
6294 * cmsg_level cmsg_type cmsg_data[] 6296 * cmsg_level cmsg_type cmsg_data[]
6295 * ------------ ------------ ---------------------- 6297 * ------------ ------------ ----------------------
6296 * IPPROTO_SCTP SCTP_INIT struct sctp_initmsg 6298 * IPPROTO_SCTP SCTP_INIT struct sctp_initmsg
6297 */ 6299 */
6298 if (cmsg->cmsg_len != 6300 if (cmsg->cmsg_len !=
6299 CMSG_LEN(sizeof(struct sctp_initmsg))) 6301 CMSG_LEN(sizeof(struct sctp_initmsg)))
6300 return -EINVAL; 6302 return -EINVAL;
6301 cmsgs->init = (struct sctp_initmsg *)CMSG_DATA(cmsg); 6303 cmsgs->init = (struct sctp_initmsg *)CMSG_DATA(cmsg);
6302 break; 6304 break;
6303 6305
6304 case SCTP_SNDRCV: 6306 case SCTP_SNDRCV:
6305 /* SCTP Socket API Extension 6307 /* SCTP Socket API Extension
6306 * 5.2.2 SCTP Header Information Structure(SCTP_SNDRCV) 6308 * 5.2.2 SCTP Header Information Structure(SCTP_SNDRCV)
6307 * 6309 *
6308 * This cmsghdr structure specifies SCTP options for 6310 * This cmsghdr structure specifies SCTP options for
6309 * sendmsg() and describes SCTP header information 6311 * sendmsg() and describes SCTP header information
6310 * about a received message through recvmsg(). 6312 * about a received message through recvmsg().
6311 * 6313 *
6312 * cmsg_level cmsg_type cmsg_data[] 6314 * cmsg_level cmsg_type cmsg_data[]
6313 * ------------ ------------ ---------------------- 6315 * ------------ ------------ ----------------------
6314 * IPPROTO_SCTP SCTP_SNDRCV struct sctp_sndrcvinfo 6316 * IPPROTO_SCTP SCTP_SNDRCV struct sctp_sndrcvinfo
6315 */ 6317 */
6316 if (cmsg->cmsg_len != 6318 if (cmsg->cmsg_len !=
6317 CMSG_LEN(sizeof(struct sctp_sndrcvinfo))) 6319 CMSG_LEN(sizeof(struct sctp_sndrcvinfo)))
6318 return -EINVAL; 6320 return -EINVAL;
6319 6321
6320 cmsgs->info = 6322 cmsgs->info =
6321 (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg); 6323 (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
6322 6324
6323 /* Minimally, validate the sinfo_flags. */ 6325 /* Minimally, validate the sinfo_flags. */
6324 if (cmsgs->info->sinfo_flags & 6326 if (cmsgs->info->sinfo_flags &
6325 ~(SCTP_UNORDERED | SCTP_ADDR_OVER | 6327 ~(SCTP_UNORDERED | SCTP_ADDR_OVER |
6326 SCTP_ABORT | SCTP_EOF)) 6328 SCTP_ABORT | SCTP_EOF))
6327 return -EINVAL; 6329 return -EINVAL;
6328 break; 6330 break;
6329 6331
6330 default: 6332 default:
6331 return -EINVAL; 6333 return -EINVAL;
6332 } 6334 }
6333 } 6335 }
6334 return 0; 6336 return 0;
6335 } 6337 }
6336 6338
6337 /* 6339 /*
6338 * Wait for a packet.. 6340 * Wait for a packet..
6339 * Note: This function is the same function as in core/datagram.c 6341 * Note: This function is the same function as in core/datagram.c
6340 * with a few modifications to make lksctp work. 6342 * with a few modifications to make lksctp work.
6341 */ 6343 */
6342 static int sctp_wait_for_packet(struct sock * sk, int *err, long *timeo_p) 6344 static int sctp_wait_for_packet(struct sock * sk, int *err, long *timeo_p)
6343 { 6345 {
6344 int error; 6346 int error;
6345 DEFINE_WAIT(wait); 6347 DEFINE_WAIT(wait);
6346 6348
6347 prepare_to_wait_exclusive(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); 6349 prepare_to_wait_exclusive(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
6348 6350
6349 /* Socket errors? */ 6351 /* Socket errors? */
6350 error = sock_error(sk); 6352 error = sock_error(sk);
6351 if (error) 6353 if (error)
6352 goto out; 6354 goto out;
6353 6355
6354 if (!skb_queue_empty(&sk->sk_receive_queue)) 6356 if (!skb_queue_empty(&sk->sk_receive_queue))
6355 goto ready; 6357 goto ready;
6356 6358
6357 /* Socket shut down? */ 6359 /* Socket shut down? */
6358 if (sk->sk_shutdown & RCV_SHUTDOWN) 6360 if (sk->sk_shutdown & RCV_SHUTDOWN)
6359 goto out; 6361 goto out;
6360 6362
6361 /* Sequenced packets can come disconnected. If so we report the 6363 /* Sequenced packets can come disconnected. If so we report the
6362 * problem. 6364 * problem.
6363 */ 6365 */
6364 error = -ENOTCONN; 6366 error = -ENOTCONN;
6365 6367
6366 /* Is there a good reason to think that we may receive some data? */ 6368 /* Is there a good reason to think that we may receive some data? */
6367 if (list_empty(&sctp_sk(sk)->ep->asocs) && !sctp_sstate(sk, LISTENING)) 6369 if (list_empty(&sctp_sk(sk)->ep->asocs) && !sctp_sstate(sk, LISTENING))
6368 goto out; 6370 goto out;
6369 6371
6370 /* Handle signals. */ 6372 /* Handle signals. */
6371 if (signal_pending(current)) 6373 if (signal_pending(current))
6372 goto interrupted; 6374 goto interrupted;
6373 6375
6374 /* Let another process have a go. Since we are going to sleep 6376 /* Let another process have a go. Since we are going to sleep
6375 * anyway. Note: This may cause odd behaviors if the message 6377 * anyway. Note: This may cause odd behaviors if the message
6376 * does not fit in the user's buffer, but this seems to be the 6378 * does not fit in the user's buffer, but this seems to be the
6377 * only way to honor MSG_DONTWAIT realistically. 6379 * only way to honor MSG_DONTWAIT realistically.
6378 */ 6380 */
6379 sctp_release_sock(sk); 6381 sctp_release_sock(sk);
6380 *timeo_p = schedule_timeout(*timeo_p); 6382 *timeo_p = schedule_timeout(*timeo_p);
6381 sctp_lock_sock(sk); 6383 sctp_lock_sock(sk);
6382 6384
6383 ready: 6385 ready:
6384 finish_wait(sk_sleep(sk), &wait); 6386 finish_wait(sk_sleep(sk), &wait);
6385 return 0; 6387 return 0;
6386 6388
6387 interrupted: 6389 interrupted:
6388 error = sock_intr_errno(*timeo_p); 6390 error = sock_intr_errno(*timeo_p);
6389 6391
6390 out: 6392 out:
6391 finish_wait(sk_sleep(sk), &wait); 6393 finish_wait(sk_sleep(sk), &wait);
6392 *err = error; 6394 *err = error;
6393 return error; 6395 return error;
6394 } 6396 }
6395 6397
6396 /* Receive a datagram. 6398 /* Receive a datagram.
6397 * Note: This is pretty much the same routine as in core/datagram.c 6399 * Note: This is pretty much the same routine as in core/datagram.c
6398 * with a few changes to make lksctp work. 6400 * with a few changes to make lksctp work.
6399 */ 6401 */
6400 static struct sk_buff *sctp_skb_recv_datagram(struct sock *sk, int flags, 6402 static struct sk_buff *sctp_skb_recv_datagram(struct sock *sk, int flags,
6401 int noblock, int *err) 6403 int noblock, int *err)
6402 { 6404 {
6403 int error; 6405 int error;
6404 struct sk_buff *skb; 6406 struct sk_buff *skb;
6405 long timeo; 6407 long timeo;
6406 6408
6407 timeo = sock_rcvtimeo(sk, noblock); 6409 timeo = sock_rcvtimeo(sk, noblock);
6408 6410
6409 SCTP_DEBUG_PRINTK("Timeout: timeo: %ld, MAX: %ld.\n", 6411 SCTP_DEBUG_PRINTK("Timeout: timeo: %ld, MAX: %ld.\n",
6410 timeo, MAX_SCHEDULE_TIMEOUT); 6412 timeo, MAX_SCHEDULE_TIMEOUT);
6411 6413
6412 do { 6414 do {
6413 /* Again only user level code calls this function, 6415 /* Again only user level code calls this function,
6414 * so nothing interrupt level 6416 * so nothing interrupt level
6415 * will suddenly eat the receive_queue. 6417 * will suddenly eat the receive_queue.
6416 * 6418 *
6417 * Look at current nfs client by the way... 6419 * Look at current nfs client by the way...
6418 * However, this function was correct in any case. 8) 6420 * However, this function was correct in any case. 8)
6419 */ 6421 */
6420 if (flags & MSG_PEEK) { 6422 if (flags & MSG_PEEK) {
6421 spin_lock_bh(&sk->sk_receive_queue.lock); 6423 spin_lock_bh(&sk->sk_receive_queue.lock);
6422 skb = skb_peek(&sk->sk_receive_queue); 6424 skb = skb_peek(&sk->sk_receive_queue);
6423 if (skb) 6425 if (skb)
6424 atomic_inc(&skb->users); 6426 atomic_inc(&skb->users);
6425 spin_unlock_bh(&sk->sk_receive_queue.lock); 6427 spin_unlock_bh(&sk->sk_receive_queue.lock);
6426 } else { 6428 } else {
6427 skb = skb_dequeue(&sk->sk_receive_queue); 6429 skb = skb_dequeue(&sk->sk_receive_queue);
6428 } 6430 }
6429 6431
6430 if (skb) 6432 if (skb)
6431 return skb; 6433 return skb;
6432 6434
6433 /* Caller is allowed not to check sk->sk_err before calling. */ 6435 /* Caller is allowed not to check sk->sk_err before calling. */
6434 error = sock_error(sk); 6436 error = sock_error(sk);
6435 if (error) 6437 if (error)
6436 goto no_packet; 6438 goto no_packet;
6437 6439
6438 if (sk->sk_shutdown & RCV_SHUTDOWN) 6440 if (sk->sk_shutdown & RCV_SHUTDOWN)
6439 break; 6441 break;
6440 6442
6441 /* User doesn't want to wait. */ 6443 /* User doesn't want to wait. */
6442 error = -EAGAIN; 6444 error = -EAGAIN;
6443 if (!timeo) 6445 if (!timeo)
6444 goto no_packet; 6446 goto no_packet;
6445 } while (sctp_wait_for_packet(sk, err, &timeo) == 0); 6447 } while (sctp_wait_for_packet(sk, err, &timeo) == 0);
6446 6448
6447 return NULL; 6449 return NULL;
6448 6450
6449 no_packet: 6451 no_packet:
6450 *err = error; 6452 *err = error;
6451 return NULL; 6453 return NULL;
6452 } 6454 }
6453 6455
6454 /* If sndbuf has changed, wake up per association sndbuf waiters. */ 6456 /* If sndbuf has changed, wake up per association sndbuf waiters. */
6455 static void __sctp_write_space(struct sctp_association *asoc) 6457 static void __sctp_write_space(struct sctp_association *asoc)
6456 { 6458 {
6457 struct sock *sk = asoc->base.sk; 6459 struct sock *sk = asoc->base.sk;
6458 struct socket *sock = sk->sk_socket; 6460 struct socket *sock = sk->sk_socket;
6459 6461
6460 if ((sctp_wspace(asoc) > 0) && sock) { 6462 if ((sctp_wspace(asoc) > 0) && sock) {
6461 if (waitqueue_active(&asoc->wait)) 6463 if (waitqueue_active(&asoc->wait))
6462 wake_up_interruptible(&asoc->wait); 6464 wake_up_interruptible(&asoc->wait);
6463 6465
6464 if (sctp_writeable(sk)) { 6466 if (sctp_writeable(sk)) {
6465 wait_queue_head_t *wq = sk_sleep(sk); 6467 wait_queue_head_t *wq = sk_sleep(sk);
6466 6468
6467 if (wq && waitqueue_active(wq)) 6469 if (wq && waitqueue_active(wq))
6468 wake_up_interruptible(wq); 6470 wake_up_interruptible(wq);
6469 6471
6470 /* Note that we try to include the Async I/O support 6472 /* Note that we try to include the Async I/O support
6471 * here by modeling from the current TCP/UDP code. 6473 * here by modeling from the current TCP/UDP code.
6472 * We have not tested with it yet. 6474 * We have not tested with it yet.
6473 */ 6475 */
6474 if (!(sk->sk_shutdown & SEND_SHUTDOWN)) 6476 if (!(sk->sk_shutdown & SEND_SHUTDOWN))
6475 sock_wake_async(sock, 6477 sock_wake_async(sock,
6476 SOCK_WAKE_SPACE, POLL_OUT); 6478 SOCK_WAKE_SPACE, POLL_OUT);
6477 } 6479 }
6478 } 6480 }
6479 } 6481 }
6480 6482
6481 /* Do accounting for the sndbuf space. 6483 /* Do accounting for the sndbuf space.
6482 * Decrement the used sndbuf space of the corresponding association by the 6484 * Decrement the used sndbuf space of the corresponding association by the
6483 * data size which was just transmitted(freed). 6485 * data size which was just transmitted(freed).
6484 */ 6486 */
6485 static void sctp_wfree(struct sk_buff *skb) 6487 static void sctp_wfree(struct sk_buff *skb)
6486 { 6488 {
6487 struct sctp_association *asoc; 6489 struct sctp_association *asoc;
6488 struct sctp_chunk *chunk; 6490 struct sctp_chunk *chunk;
6489 struct sock *sk; 6491 struct sock *sk;
6490 6492
6491 /* Get the saved chunk pointer. */ 6493 /* Get the saved chunk pointer. */
6492 chunk = *((struct sctp_chunk **)(skb->cb)); 6494 chunk = *((struct sctp_chunk **)(skb->cb));
6493 asoc = chunk->asoc; 6495 asoc = chunk->asoc;
6494 sk = asoc->base.sk; 6496 sk = asoc->base.sk;
6495 asoc->sndbuf_used -= SCTP_DATA_SNDSIZE(chunk) + 6497 asoc->sndbuf_used -= SCTP_DATA_SNDSIZE(chunk) +
6496 sizeof(struct sk_buff) + 6498 sizeof(struct sk_buff) +
6497 sizeof(struct sctp_chunk); 6499 sizeof(struct sctp_chunk);
6498 6500
6499 atomic_sub(sizeof(struct sctp_chunk), &sk->sk_wmem_alloc); 6501 atomic_sub(sizeof(struct sctp_chunk), &sk->sk_wmem_alloc);
6500 6502
6501 /* 6503 /*
6502 * This undoes what is done via sctp_set_owner_w and sk_mem_charge 6504 * This undoes what is done via sctp_set_owner_w and sk_mem_charge
6503 */ 6505 */
6504 sk->sk_wmem_queued -= skb->truesize; 6506 sk->sk_wmem_queued -= skb->truesize;
6505 sk_mem_uncharge(sk, skb->truesize); 6507 sk_mem_uncharge(sk, skb->truesize);
6506 6508
6507 sock_wfree(skb); 6509 sock_wfree(skb);
6508 __sctp_write_space(asoc); 6510 __sctp_write_space(asoc);
6509 6511
6510 sctp_association_put(asoc); 6512 sctp_association_put(asoc);
6511 } 6513 }
6512 6514
6513 /* Do accounting for the receive space on the socket. 6515 /* Do accounting for the receive space on the socket.
6514 * Accounting for the association is done in ulpevent.c 6516 * Accounting for the association is done in ulpevent.c
6515 * We set this as a destructor for the cloned data skbs so that 6517 * We set this as a destructor for the cloned data skbs so that
6516 * accounting is done at the correct time. 6518 * accounting is done at the correct time.
6517 */ 6519 */
6518 void sctp_sock_rfree(struct sk_buff *skb) 6520 void sctp_sock_rfree(struct sk_buff *skb)
6519 { 6521 {
6520 struct sock *sk = skb->sk; 6522 struct sock *sk = skb->sk;
6521 struct sctp_ulpevent *event = sctp_skb2event(skb); 6523 struct sctp_ulpevent *event = sctp_skb2event(skb);
6522 6524
6523 atomic_sub(event->rmem_len, &sk->sk_rmem_alloc); 6525 atomic_sub(event->rmem_len, &sk->sk_rmem_alloc);
6524 6526
6525 /* 6527 /*
6526 * Mimic the behavior of sock_rfree 6528 * Mimic the behavior of sock_rfree
6527 */ 6529 */
6528 sk_mem_uncharge(sk, event->rmem_len); 6530 sk_mem_uncharge(sk, event->rmem_len);
6529 } 6531 }
6530 6532
6531 6533
6532 /* Helper function to wait for space in the sndbuf. */ 6534 /* Helper function to wait for space in the sndbuf. */
6533 static int sctp_wait_for_sndbuf(struct sctp_association *asoc, long *timeo_p, 6535 static int sctp_wait_for_sndbuf(struct sctp_association *asoc, long *timeo_p,
6534 size_t msg_len) 6536 size_t msg_len)
6535 { 6537 {
6536 struct sock *sk = asoc->base.sk; 6538 struct sock *sk = asoc->base.sk;
6537 int err = 0; 6539 int err = 0;
6538 long current_timeo = *timeo_p; 6540 long current_timeo = *timeo_p;
6539 DEFINE_WAIT(wait); 6541 DEFINE_WAIT(wait);
6540 6542
6541 SCTP_DEBUG_PRINTK("wait_for_sndbuf: asoc=%p, timeo=%ld, msg_len=%zu\n", 6543 SCTP_DEBUG_PRINTK("wait_for_sndbuf: asoc=%p, timeo=%ld, msg_len=%zu\n",
6542 asoc, (long)(*timeo_p), msg_len); 6544 asoc, (long)(*timeo_p), msg_len);
6543 6545
6544 /* Increment the association's refcnt. */ 6546 /* Increment the association's refcnt. */
6545 sctp_association_hold(asoc); 6547 sctp_association_hold(asoc);
6546 6548
6547 /* Wait on the association specific sndbuf space. */ 6549 /* Wait on the association specific sndbuf space. */
6548 for (;;) { 6550 for (;;) {
6549 prepare_to_wait_exclusive(&asoc->wait, &wait, 6551 prepare_to_wait_exclusive(&asoc->wait, &wait,
6550 TASK_INTERRUPTIBLE); 6552 TASK_INTERRUPTIBLE);
6551 if (!*timeo_p) 6553 if (!*timeo_p)
6552 goto do_nonblock; 6554 goto do_nonblock;
6553 if (sk->sk_err || asoc->state >= SCTP_STATE_SHUTDOWN_PENDING || 6555 if (sk->sk_err || asoc->state >= SCTP_STATE_SHUTDOWN_PENDING ||
6554 asoc->base.dead) 6556 asoc->base.dead)
6555 goto do_error; 6557 goto do_error;
6556 if (signal_pending(current)) 6558 if (signal_pending(current))
6557 goto do_interrupted; 6559 goto do_interrupted;
6558 if (msg_len <= sctp_wspace(asoc)) 6560 if (msg_len <= sctp_wspace(asoc))
6559 break; 6561 break;
6560 6562
6561 /* Let another process have a go. Since we are going 6563 /* Let another process have a go. Since we are going
6562 * to sleep anyway. 6564 * to sleep anyway.
6563 */ 6565 */
6564 sctp_release_sock(sk); 6566 sctp_release_sock(sk);
6565 current_timeo = schedule_timeout(current_timeo); 6567 current_timeo = schedule_timeout(current_timeo);
6566 BUG_ON(sk != asoc->base.sk); 6568 BUG_ON(sk != asoc->base.sk);
6567 sctp_lock_sock(sk); 6569 sctp_lock_sock(sk);
6568 6570
6569 *timeo_p = current_timeo; 6571 *timeo_p = current_timeo;
6570 } 6572 }
6571 6573
6572 out: 6574 out:
6573 finish_wait(&asoc->wait, &wait); 6575 finish_wait(&asoc->wait, &wait);
6574 6576
6575 /* Release the association's refcnt. */ 6577 /* Release the association's refcnt. */
6576 sctp_association_put(asoc); 6578 sctp_association_put(asoc);
6577 6579
6578 return err; 6580 return err;
6579 6581
6580 do_error: 6582 do_error:
6581 err = -EPIPE; 6583 err = -EPIPE;
6582 goto out; 6584 goto out;
6583 6585
6584 do_interrupted: 6586 do_interrupted:
6585 err = sock_intr_errno(*timeo_p); 6587 err = sock_intr_errno(*timeo_p);
6586 goto out; 6588 goto out;
6587 6589
6588 do_nonblock: 6590 do_nonblock:
6589 err = -EAGAIN; 6591 err = -EAGAIN;
6590 goto out; 6592 goto out;
6591 } 6593 }
6592 6594
6593 void sctp_data_ready(struct sock *sk, int len) 6595 void sctp_data_ready(struct sock *sk, int len)
6594 { 6596 {
6595 struct socket_wq *wq; 6597 struct socket_wq *wq;
6596 6598
6597 rcu_read_lock(); 6599 rcu_read_lock();
6598 wq = rcu_dereference(sk->sk_wq); 6600 wq = rcu_dereference(sk->sk_wq);
6599 if (wq_has_sleeper(wq)) 6601 if (wq_has_sleeper(wq))
6600 wake_up_interruptible_sync_poll(&wq->wait, POLLIN | 6602 wake_up_interruptible_sync_poll(&wq->wait, POLLIN |
6601 POLLRDNORM | POLLRDBAND); 6603 POLLRDNORM | POLLRDBAND);
6602 sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN); 6604 sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
6603 rcu_read_unlock(); 6605 rcu_read_unlock();
6604 } 6606 }
6605 6607
6606 /* If socket sndbuf has changed, wake up all per association waiters. */ 6608 /* If socket sndbuf has changed, wake up all per association waiters. */
6607 void sctp_write_space(struct sock *sk) 6609 void sctp_write_space(struct sock *sk)
6608 { 6610 {
6609 struct sctp_association *asoc; 6611 struct sctp_association *asoc;
6610 6612
6611 /* Wake up the tasks in each wait queue. */ 6613 /* Wake up the tasks in each wait queue. */
6612 list_for_each_entry(asoc, &((sctp_sk(sk))->ep->asocs), asocs) { 6614 list_for_each_entry(asoc, &((sctp_sk(sk))->ep->asocs), asocs) {
6613 __sctp_write_space(asoc); 6615 __sctp_write_space(asoc);
6614 } 6616 }
6615 } 6617 }
6616 6618
6617 /* Is there any sndbuf space available on the socket? 6619 /* Is there any sndbuf space available on the socket?
6618 * 6620 *
6619 * Note that sk_wmem_alloc is the sum of the send buffers on all of the 6621 * Note that sk_wmem_alloc is the sum of the send buffers on all of the
6620 * associations on the same socket. For a UDP-style socket with 6622 * associations on the same socket. For a UDP-style socket with
6621 * multiple associations, it is possible for it to be "unwriteable" 6623 * multiple associations, it is possible for it to be "unwriteable"
6622 * prematurely. I assume that this is acceptable because 6624 * prematurely. I assume that this is acceptable because
6623 * a premature "unwriteable" is better than an accidental "writeable" which 6625 * a premature "unwriteable" is better than an accidental "writeable" which
6624 * would cause an unwanted block under certain circumstances. For the 1-1 6626 * would cause an unwanted block under certain circumstances. For the 1-1
6625 * UDP-style sockets or TCP-style sockets, this code should work. 6627 * UDP-style sockets or TCP-style sockets, this code should work.
6626 * - Daisy 6628 * - Daisy
6627 */ 6629 */
6628 static int sctp_writeable(struct sock *sk) 6630 static int sctp_writeable(struct sock *sk)
6629 { 6631 {
6630 int amt = 0; 6632 int amt = 0;
6631 6633
6632 amt = sk->sk_sndbuf - sk_wmem_alloc_get(sk); 6634 amt = sk->sk_sndbuf - sk_wmem_alloc_get(sk);
6633 if (amt < 0) 6635 if (amt < 0)
6634 amt = 0; 6636 amt = 0;
6635 return amt; 6637 return amt;
6636 } 6638 }
6637 6639
6638 /* Wait for an association to go into ESTABLISHED state. If timeout is 0, 6640 /* Wait for an association to go into ESTABLISHED state. If timeout is 0,
6639 * returns immediately with EINPROGRESS. 6641 * returns immediately with EINPROGRESS.
6640 */ 6642 */
6641 static int sctp_wait_for_connect(struct sctp_association *asoc, long *timeo_p) 6643 static int sctp_wait_for_connect(struct sctp_association *asoc, long *timeo_p)
6642 { 6644 {
6643 struct sock *sk = asoc->base.sk; 6645 struct sock *sk = asoc->base.sk;
6644 int err = 0; 6646 int err = 0;
6645 long current_timeo = *timeo_p; 6647 long current_timeo = *timeo_p;
6646 DEFINE_WAIT(wait); 6648 DEFINE_WAIT(wait);
6647 6649
6648 SCTP_DEBUG_PRINTK("%s: asoc=%p, timeo=%ld\n", __func__, asoc, 6650 SCTP_DEBUG_PRINTK("%s: asoc=%p, timeo=%ld\n", __func__, asoc,
6649 (long)(*timeo_p)); 6651 (long)(*timeo_p));
6650 6652
6651 /* Increment the association's refcnt. */ 6653 /* Increment the association's refcnt. */
6652 sctp_association_hold(asoc); 6654 sctp_association_hold(asoc);
6653 6655
6654 for (;;) { 6656 for (;;) {
6655 prepare_to_wait_exclusive(&asoc->wait, &wait, 6657 prepare_to_wait_exclusive(&asoc->wait, &wait,
6656 TASK_INTERRUPTIBLE); 6658 TASK_INTERRUPTIBLE);
6657 if (!*timeo_p) 6659 if (!*timeo_p)
6658 goto do_nonblock; 6660 goto do_nonblock;
6659 if (sk->sk_shutdown & RCV_SHUTDOWN) 6661 if (sk->sk_shutdown & RCV_SHUTDOWN)
6660 break; 6662 break;
6661 if (sk->sk_err || asoc->state >= SCTP_STATE_SHUTDOWN_PENDING || 6663 if (sk->sk_err || asoc->state >= SCTP_STATE_SHUTDOWN_PENDING ||
6662 asoc->base.dead) 6664 asoc->base.dead)
6663 goto do_error; 6665 goto do_error;
6664 if (signal_pending(current)) 6666 if (signal_pending(current))
6665 goto do_interrupted; 6667 goto do_interrupted;
6666 6668
6667 if (sctp_state(asoc, ESTABLISHED)) 6669 if (sctp_state(asoc, ESTABLISHED))
6668 break; 6670 break;
6669 6671
6670 /* Let another process have a go. Since we are going 6672 /* Let another process have a go. Since we are going
6671 * to sleep anyway. 6673 * to sleep anyway.
6672 */ 6674 */
6673 sctp_release_sock(sk); 6675 sctp_release_sock(sk);
6674 current_timeo = schedule_timeout(current_timeo); 6676 current_timeo = schedule_timeout(current_timeo);
6675 sctp_lock_sock(sk); 6677 sctp_lock_sock(sk);
6676 6678
6677 *timeo_p = current_timeo; 6679 *timeo_p = current_timeo;
6678 } 6680 }
6679 6681
6680 out: 6682 out:
6681 finish_wait(&asoc->wait, &wait); 6683 finish_wait(&asoc->wait, &wait);
6682 6684
6683 /* Release the association's refcnt. */ 6685 /* Release the association's refcnt. */
6684 sctp_association_put(asoc); 6686 sctp_association_put(asoc);
6685 6687
6686 return err; 6688 return err;
6687 6689
6688 do_error: 6690 do_error:
6689 if (asoc->init_err_counter + 1 > asoc->max_init_attempts) 6691 if (asoc->init_err_counter + 1 > asoc->max_init_attempts)
6690 err = -ETIMEDOUT; 6692 err = -ETIMEDOUT;
6691 else 6693 else
6692 err = -ECONNREFUSED; 6694 err = -ECONNREFUSED;
6693 goto out; 6695 goto out;
6694 6696
6695 do_interrupted: 6697 do_interrupted:
6696 err = sock_intr_errno(*timeo_p); 6698 err = sock_intr_errno(*timeo_p);
6697 goto out; 6699 goto out;
6698 6700
6699 do_nonblock: 6701 do_nonblock:
6700 err = -EINPROGRESS; 6702 err = -EINPROGRESS;
6701 goto out; 6703 goto out;
6702 } 6704 }
6703 6705
6704 static int sctp_wait_for_accept(struct sock *sk, long timeo) 6706 static int sctp_wait_for_accept(struct sock *sk, long timeo)
6705 { 6707 {
6706 struct sctp_endpoint *ep; 6708 struct sctp_endpoint *ep;
6707 int err = 0; 6709 int err = 0;
6708 DEFINE_WAIT(wait); 6710 DEFINE_WAIT(wait);
6709 6711
6710 ep = sctp_sk(sk)->ep; 6712 ep = sctp_sk(sk)->ep;
6711 6713
6712 6714
6713 for (;;) { 6715 for (;;) {
6714 prepare_to_wait_exclusive(sk_sleep(sk), &wait, 6716 prepare_to_wait_exclusive(sk_sleep(sk), &wait,
6715 TASK_INTERRUPTIBLE); 6717 TASK_INTERRUPTIBLE);
6716 6718
6717 if (list_empty(&ep->asocs)) { 6719 if (list_empty(&ep->asocs)) {
6718 sctp_release_sock(sk); 6720 sctp_release_sock(sk);
6719 timeo = schedule_timeout(timeo); 6721 timeo = schedule_timeout(timeo);
6720 sctp_lock_sock(sk); 6722 sctp_lock_sock(sk);
6721 } 6723 }
6722 6724
6723 err = -EINVAL; 6725 err = -EINVAL;
6724 if (!sctp_sstate(sk, LISTENING)) 6726 if (!sctp_sstate(sk, LISTENING))
6725 break; 6727 break;
6726 6728
6727 err = 0; 6729 err = 0;
6728 if (!list_empty(&ep->asocs)) 6730 if (!list_empty(&ep->asocs))
6729 break; 6731 break;
6730 6732
6731 err = sock_intr_errno(timeo); 6733 err = sock_intr_errno(timeo);
6732 if (signal_pending(current)) 6734 if (signal_pending(current))
6733 break; 6735 break;
6734 6736
6735 err = -EAGAIN; 6737 err = -EAGAIN;
6736 if (!timeo) 6738 if (!timeo)
6737 break; 6739 break;
6738 } 6740 }
6739 6741
6740 finish_wait(sk_sleep(sk), &wait); 6742 finish_wait(sk_sleep(sk), &wait);
6741 6743
6742 return err; 6744 return err;
6743 } 6745 }
6744 6746
6745 static void sctp_wait_for_close(struct sock *sk, long timeout) 6747 static void sctp_wait_for_close(struct sock *sk, long timeout)
6746 { 6748 {
6747 DEFINE_WAIT(wait); 6749 DEFINE_WAIT(wait);
6748 6750
6749 do { 6751 do {
6750 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); 6752 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
6751 if (list_empty(&sctp_sk(sk)->ep->asocs)) 6753 if (list_empty(&sctp_sk(sk)->ep->asocs))
6752 break; 6754 break;
6753 sctp_release_sock(sk); 6755 sctp_release_sock(sk);
6754 timeout = schedule_timeout(timeout); 6756 timeout = schedule_timeout(timeout);
6755 sctp_lock_sock(sk); 6757 sctp_lock_sock(sk);
6756 } while (!signal_pending(current) && timeout); 6758 } while (!signal_pending(current) && timeout);
6757 6759
6758 finish_wait(sk_sleep(sk), &wait); 6760 finish_wait(sk_sleep(sk), &wait);
6759 } 6761 }
6760 6762
6761 static void sctp_skb_set_owner_r_frag(struct sk_buff *skb, struct sock *sk) 6763 static void sctp_skb_set_owner_r_frag(struct sk_buff *skb, struct sock *sk)
6762 { 6764 {
6763 struct sk_buff *frag; 6765 struct sk_buff *frag;
6764 6766
6765 if (!skb->data_len) 6767 if (!skb->data_len)
6766 goto done; 6768 goto done;
6767 6769
6768 /* Don't forget the fragments. */ 6770 /* Don't forget the fragments. */
6769 skb_walk_frags(skb, frag) 6771 skb_walk_frags(skb, frag)
6770 sctp_skb_set_owner_r_frag(frag, sk); 6772 sctp_skb_set_owner_r_frag(frag, sk);
6771 6773
6772 done: 6774 done:
6773 sctp_skb_set_owner_r(skb, sk); 6775 sctp_skb_set_owner_r(skb, sk);
6774 } 6776 }
6775 6777
6776 void sctp_copy_sock(struct sock *newsk, struct sock *sk, 6778 void sctp_copy_sock(struct sock *newsk, struct sock *sk,
6777 struct sctp_association *asoc) 6779 struct sctp_association *asoc)
6778 { 6780 {
6779 struct inet_sock *inet = inet_sk(sk); 6781 struct inet_sock *inet = inet_sk(sk);
6780 struct inet_sock *newinet; 6782 struct inet_sock *newinet;
6781 6783
6782 newsk->sk_type = sk->sk_type; 6784 newsk->sk_type = sk->sk_type;
6783 newsk->sk_bound_dev_if = sk->sk_bound_dev_if; 6785 newsk->sk_bound_dev_if = sk->sk_bound_dev_if;
6784 newsk->sk_flags = sk->sk_flags; 6786 newsk->sk_flags = sk->sk_flags;
6785 newsk->sk_no_check = sk->sk_no_check; 6787 newsk->sk_no_check = sk->sk_no_check;
6786 newsk->sk_reuse = sk->sk_reuse; 6788 newsk->sk_reuse = sk->sk_reuse;
6787 6789
6788 newsk->sk_shutdown = sk->sk_shutdown; 6790 newsk->sk_shutdown = sk->sk_shutdown;
6789 newsk->sk_destruct = inet_sock_destruct; 6791 newsk->sk_destruct = inet_sock_destruct;
6790 newsk->sk_family = sk->sk_family; 6792 newsk->sk_family = sk->sk_family;
6791 newsk->sk_protocol = IPPROTO_SCTP; 6793 newsk->sk_protocol = IPPROTO_SCTP;
6792 newsk->sk_backlog_rcv = sk->sk_prot->backlog_rcv; 6794 newsk->sk_backlog_rcv = sk->sk_prot->backlog_rcv;
6793 newsk->sk_sndbuf = sk->sk_sndbuf; 6795 newsk->sk_sndbuf = sk->sk_sndbuf;
6794 newsk->sk_rcvbuf = sk->sk_rcvbuf; 6796 newsk->sk_rcvbuf = sk->sk_rcvbuf;
6795 newsk->sk_lingertime = sk->sk_lingertime; 6797 newsk->sk_lingertime = sk->sk_lingertime;
6796 newsk->sk_rcvtimeo = sk->sk_rcvtimeo; 6798 newsk->sk_rcvtimeo = sk->sk_rcvtimeo;
6797 newsk->sk_sndtimeo = sk->sk_sndtimeo; 6799 newsk->sk_sndtimeo = sk->sk_sndtimeo;
6798 6800
6799 newinet = inet_sk(newsk); 6801 newinet = inet_sk(newsk);
6800 6802
6801 /* Initialize sk's sport, dport, rcv_saddr and daddr for 6803 /* Initialize sk's sport, dport, rcv_saddr and daddr for
6802 * getsockname() and getpeername() 6804 * getsockname() and getpeername()
6803 */ 6805 */
6804 newinet->inet_sport = inet->inet_sport; 6806 newinet->inet_sport = inet->inet_sport;
6805 newinet->inet_saddr = inet->inet_saddr; 6807 newinet->inet_saddr = inet->inet_saddr;
6806 newinet->inet_rcv_saddr = inet->inet_rcv_saddr; 6808 newinet->inet_rcv_saddr = inet->inet_rcv_saddr;
6807 newinet->inet_dport = htons(asoc->peer.port); 6809 newinet->inet_dport = htons(asoc->peer.port);
6808 newinet->pmtudisc = inet->pmtudisc; 6810 newinet->pmtudisc = inet->pmtudisc;
6809 newinet->inet_id = asoc->next_tsn ^ jiffies; 6811 newinet->inet_id = asoc->next_tsn ^ jiffies;
6810 6812
6811 newinet->uc_ttl = inet->uc_ttl; 6813 newinet->uc_ttl = inet->uc_ttl;
6812 newinet->mc_loop = 1; 6814 newinet->mc_loop = 1;
6813 newinet->mc_ttl = 1; 6815 newinet->mc_ttl = 1;
6814 newinet->mc_index = 0; 6816 newinet->mc_index = 0;
6815 newinet->mc_list = NULL; 6817 newinet->mc_list = NULL;
6816 } 6818 }
6817 6819
6818 /* Populate the fields of the newsk from the oldsk and migrate the assoc 6820 /* Populate the fields of the newsk from the oldsk and migrate the assoc
6819 * and its messages to the newsk. 6821 * and its messages to the newsk.
6820 */ 6822 */
6821 static void sctp_sock_migrate(struct sock *oldsk, struct sock *newsk, 6823 static void sctp_sock_migrate(struct sock *oldsk, struct sock *newsk,
6822 struct sctp_association *assoc, 6824 struct sctp_association *assoc,
6823 sctp_socket_type_t type) 6825 sctp_socket_type_t type)
6824 { 6826 {
6825 struct sctp_sock *oldsp = sctp_sk(oldsk); 6827 struct sctp_sock *oldsp = sctp_sk(oldsk);
6826 struct sctp_sock *newsp = sctp_sk(newsk); 6828 struct sctp_sock *newsp = sctp_sk(newsk);
6827 struct sctp_bind_bucket *pp; /* hash list port iterator */ 6829 struct sctp_bind_bucket *pp; /* hash list port iterator */
6828 struct sctp_endpoint *newep = newsp->ep; 6830 struct sctp_endpoint *newep = newsp->ep;
6829 struct sk_buff *skb, *tmp; 6831 struct sk_buff *skb, *tmp;
6830 struct sctp_ulpevent *event; 6832 struct sctp_ulpevent *event;
6831 struct sctp_bind_hashbucket *head; 6833 struct sctp_bind_hashbucket *head;
6832 struct list_head tmplist; 6834 struct list_head tmplist;
6833 6835
6834 /* Migrate socket buffer sizes and all the socket level options to the 6836 /* Migrate socket buffer sizes and all the socket level options to the
6835 * new socket. 6837 * new socket.
6836 */ 6838 */
6837 newsk->sk_sndbuf = oldsk->sk_sndbuf; 6839 newsk->sk_sndbuf = oldsk->sk_sndbuf;
6838 newsk->sk_rcvbuf = oldsk->sk_rcvbuf; 6840 newsk->sk_rcvbuf = oldsk->sk_rcvbuf;
6839 /* Brute force copy old sctp opt. */ 6841 /* Brute force copy old sctp opt. */
6840 if (oldsp->do_auto_asconf) { 6842 if (oldsp->do_auto_asconf) {
6841 memcpy(&tmplist, &newsp->auto_asconf_list, sizeof(tmplist)); 6843 memcpy(&tmplist, &newsp->auto_asconf_list, sizeof(tmplist));
6842 inet_sk_copy_descendant(newsk, oldsk); 6844 inet_sk_copy_descendant(newsk, oldsk);
6843 memcpy(&newsp->auto_asconf_list, &tmplist, sizeof(tmplist)); 6845 memcpy(&newsp->auto_asconf_list, &tmplist, sizeof(tmplist));
6844 } else 6846 } else
6845 inet_sk_copy_descendant(newsk, oldsk); 6847 inet_sk_copy_descendant(newsk, oldsk);
6846 6848
6847 /* Restore the ep value that was overwritten with the above structure 6849 /* Restore the ep value that was overwritten with the above structure
6848 * copy. 6850 * copy.
6849 */ 6851 */
6850 newsp->ep = newep; 6852 newsp->ep = newep;
6851 newsp->hmac = NULL; 6853 newsp->hmac = NULL;
6852 6854
6853 /* Hook this new socket in to the bind_hash list. */ 6855 /* Hook this new socket in to the bind_hash list. */
6854 head = &sctp_port_hashtable[sctp_phashfn(sock_net(oldsk), 6856 head = &sctp_port_hashtable[sctp_phashfn(sock_net(oldsk),
6855 inet_sk(oldsk)->inet_num)]; 6857 inet_sk(oldsk)->inet_num)];
6856 sctp_local_bh_disable(); 6858 sctp_local_bh_disable();
6857 sctp_spin_lock(&head->lock); 6859 sctp_spin_lock(&head->lock);
6858 pp = sctp_sk(oldsk)->bind_hash; 6860 pp = sctp_sk(oldsk)->bind_hash;
6859 sk_add_bind_node(newsk, &pp->owner); 6861 sk_add_bind_node(newsk, &pp->owner);
6860 sctp_sk(newsk)->bind_hash = pp; 6862 sctp_sk(newsk)->bind_hash = pp;
6861 inet_sk(newsk)->inet_num = inet_sk(oldsk)->inet_num; 6863 inet_sk(newsk)->inet_num = inet_sk(oldsk)->inet_num;
6862 sctp_spin_unlock(&head->lock); 6864 sctp_spin_unlock(&head->lock);
6863 sctp_local_bh_enable(); 6865 sctp_local_bh_enable();
6864 6866
6865 /* Copy the bind_addr list from the original endpoint to the new 6867 /* Copy the bind_addr list from the original endpoint to the new
6866 * endpoint so that we can handle restarts properly 6868 * endpoint so that we can handle restarts properly
6867 */ 6869 */
6868 sctp_bind_addr_dup(&newsp->ep->base.bind_addr, 6870 sctp_bind_addr_dup(&newsp->ep->base.bind_addr,
6869 &oldsp->ep->base.bind_addr, GFP_KERNEL); 6871 &oldsp->ep->base.bind_addr, GFP_KERNEL);
6870 6872
6871 /* Move any messages in the old socket's receive queue that are for the 6873 /* Move any messages in the old socket's receive queue that are for the
6872 * peeled off association to the new socket's receive queue. 6874 * peeled off association to the new socket's receive queue.
6873 */ 6875 */
6874 sctp_skb_for_each(skb, &oldsk->sk_receive_queue, tmp) { 6876 sctp_skb_for_each(skb, &oldsk->sk_receive_queue, tmp) {
6875 event = sctp_skb2event(skb); 6877 event = sctp_skb2event(skb);
6876 if (event->asoc == assoc) { 6878 if (event->asoc == assoc) {
6877 __skb_unlink(skb, &oldsk->sk_receive_queue); 6879 __skb_unlink(skb, &oldsk->sk_receive_queue);
6878 __skb_queue_tail(&newsk->sk_receive_queue, skb); 6880 __skb_queue_tail(&newsk->sk_receive_queue, skb);
6879 sctp_skb_set_owner_r_frag(skb, newsk); 6881 sctp_skb_set_owner_r_frag(skb, newsk);
6880 } 6882 }
6881 } 6883 }
6882 6884
6883 /* Clean up any messages pending delivery due to partial 6885 /* Clean up any messages pending delivery due to partial
6884 * delivery. Three cases: 6886 * delivery. Three cases:
6885 * 1) No partial deliver; no work. 6887 * 1) No partial deliver; no work.
6886 * 2) Peeling off partial delivery; keep pd_lobby in new pd_lobby. 6888 * 2) Peeling off partial delivery; keep pd_lobby in new pd_lobby.
6887 * 3) Peeling off non-partial delivery; move pd_lobby to receive_queue. 6889 * 3) Peeling off non-partial delivery; move pd_lobby to receive_queue.
6888 */ 6890 */
6889 skb_queue_head_init(&newsp->pd_lobby); 6891 skb_queue_head_init(&newsp->pd_lobby);
6890 atomic_set(&sctp_sk(newsk)->pd_mode, assoc->ulpq.pd_mode); 6892 atomic_set(&sctp_sk(newsk)->pd_mode, assoc->ulpq.pd_mode);
6891 6893
6892 if (atomic_read(&sctp_sk(oldsk)->pd_mode)) { 6894 if (atomic_read(&sctp_sk(oldsk)->pd_mode)) {
6893 struct sk_buff_head *queue; 6895 struct sk_buff_head *queue;
6894 6896
6895 /* Decide which queue to move pd_lobby skbs to. */ 6897 /* Decide which queue to move pd_lobby skbs to. */
6896 if (assoc->ulpq.pd_mode) { 6898 if (assoc->ulpq.pd_mode) {
6897 queue = &newsp->pd_lobby; 6899 queue = &newsp->pd_lobby;
6898 } else 6900 } else
6899 queue = &newsk->sk_receive_queue; 6901 queue = &newsk->sk_receive_queue;
6900 6902
6901 /* Walk through the pd_lobby, looking for skbs that 6903 /* Walk through the pd_lobby, looking for skbs that
6902 * need moved to the new socket. 6904 * need moved to the new socket.
6903 */ 6905 */
6904 sctp_skb_for_each(skb, &oldsp->pd_lobby, tmp) { 6906 sctp_skb_for_each(skb, &oldsp->pd_lobby, tmp) {
6905 event = sctp_skb2event(skb); 6907 event = sctp_skb2event(skb);
6906 if (event->asoc == assoc) { 6908 if (event->asoc == assoc) {
6907 __skb_unlink(skb, &oldsp->pd_lobby); 6909 __skb_unlink(skb, &oldsp->pd_lobby);
6908 __skb_queue_tail(queue, skb); 6910 __skb_queue_tail(queue, skb);
6909 sctp_skb_set_owner_r_frag(skb, newsk); 6911 sctp_skb_set_owner_r_frag(skb, newsk);
6910 } 6912 }
6911 } 6913 }
6912 6914
6913 /* Clear up any skbs waiting for the partial 6915 /* Clear up any skbs waiting for the partial
6914 * delivery to finish. 6916 * delivery to finish.
6915 */ 6917 */
6916 if (assoc->ulpq.pd_mode) 6918 if (assoc->ulpq.pd_mode)
6917 sctp_clear_pd(oldsk, NULL); 6919 sctp_clear_pd(oldsk, NULL);
6918 6920
6919 } 6921 }
6920 6922
6921 sctp_skb_for_each(skb, &assoc->ulpq.reasm, tmp) 6923 sctp_skb_for_each(skb, &assoc->ulpq.reasm, tmp)
6922 sctp_skb_set_owner_r_frag(skb, newsk); 6924 sctp_skb_set_owner_r_frag(skb, newsk);
6923 6925
6924 sctp_skb_for_each(skb, &assoc->ulpq.lobby, tmp) 6926 sctp_skb_for_each(skb, &assoc->ulpq.lobby, tmp)
6925 sctp_skb_set_owner_r_frag(skb, newsk); 6927 sctp_skb_set_owner_r_frag(skb, newsk);
6926 6928
6927 /* Set the type of socket to indicate that it is peeled off from the 6929 /* Set the type of socket to indicate that it is peeled off from the
6928 * original UDP-style socket or created with the accept() call on a 6930 * original UDP-style socket or created with the accept() call on a
6929 * TCP-style socket.. 6931 * TCP-style socket..
6930 */ 6932 */
6931 newsp->type = type; 6933 newsp->type = type;
6932 6934
6933 /* Mark the new socket "in-use" by the user so that any packets 6935 /* Mark the new socket "in-use" by the user so that any packets
6934 * that may arrive on the association after we've moved it are 6936 * that may arrive on the association after we've moved it are
6935 * queued to the backlog. This prevents a potential race between 6937 * queued to the backlog. This prevents a potential race between
6936 * backlog processing on the old socket and new-packet processing 6938 * backlog processing on the old socket and new-packet processing
6937 * on the new socket. 6939 * on the new socket.
6938 * 6940 *
6939 * The caller has just allocated newsk so we can guarantee that other 6941 * The caller has just allocated newsk so we can guarantee that other
6940 * paths won't try to lock it and then oldsk. 6942 * paths won't try to lock it and then oldsk.
6941 */ 6943 */
6942 lock_sock_nested(newsk, SINGLE_DEPTH_NESTING); 6944 lock_sock_nested(newsk, SINGLE_DEPTH_NESTING);
6943 sctp_assoc_migrate(assoc, newsk); 6945 sctp_assoc_migrate(assoc, newsk);
6944 6946
6945 /* If the association on the newsk is already closed before accept() 6947 /* If the association on the newsk is already closed before accept()
6946 * is called, set RCV_SHUTDOWN flag. 6948 * is called, set RCV_SHUTDOWN flag.
6947 */ 6949 */
6948 if (sctp_state(assoc, CLOSED) && sctp_style(newsk, TCP)) 6950 if (sctp_state(assoc, CLOSED) && sctp_style(newsk, TCP))
6949 newsk->sk_shutdown |= RCV_SHUTDOWN; 6951 newsk->sk_shutdown |= RCV_SHUTDOWN;
6950 6952
6951 newsk->sk_state = SCTP_SS_ESTABLISHED; 6953 newsk->sk_state = SCTP_SS_ESTABLISHED;
6952 sctp_release_sock(newsk); 6954 sctp_release_sock(newsk);
6953 } 6955 }
6954 6956
6955 6957
6956 /* This proto struct describes the ULP interface for SCTP. */ 6958 /* This proto struct describes the ULP interface for SCTP. */
6957 struct proto sctp_prot = { 6959 struct proto sctp_prot = {
6958 .name = "SCTP", 6960 .name = "SCTP",
6959 .owner = THIS_MODULE, 6961 .owner = THIS_MODULE,
6960 .close = sctp_close, 6962 .close = sctp_close,
6961 .connect = sctp_connect, 6963 .connect = sctp_connect,
6962 .disconnect = sctp_disconnect, 6964 .disconnect = sctp_disconnect,
6963 .accept = sctp_accept, 6965 .accept = sctp_accept,
6964 .ioctl = sctp_ioctl, 6966 .ioctl = sctp_ioctl,
6965 .init = sctp_init_sock, 6967 .init = sctp_init_sock,
6966 .destroy = sctp_destroy_sock, 6968 .destroy = sctp_destroy_sock,
6967 .shutdown = sctp_shutdown, 6969 .shutdown = sctp_shutdown,
6968 .setsockopt = sctp_setsockopt, 6970 .setsockopt = sctp_setsockopt,
6969 .getsockopt = sctp_getsockopt, 6971 .getsockopt = sctp_getsockopt,
6970 .sendmsg = sctp_sendmsg, 6972 .sendmsg = sctp_sendmsg,
6971 .recvmsg = sctp_recvmsg, 6973 .recvmsg = sctp_recvmsg,
6972 .bind = sctp_bind, 6974 .bind = sctp_bind,
6973 .backlog_rcv = sctp_backlog_rcv, 6975 .backlog_rcv = sctp_backlog_rcv,
6974 .hash = sctp_hash, 6976 .hash = sctp_hash,
6975 .unhash = sctp_unhash, 6977 .unhash = sctp_unhash,
6976 .get_port = sctp_get_port, 6978 .get_port = sctp_get_port,
6977 .obj_size = sizeof(struct sctp_sock), 6979 .obj_size = sizeof(struct sctp_sock),
6978 .sysctl_mem = sysctl_sctp_mem, 6980 .sysctl_mem = sysctl_sctp_mem,
6979 .sysctl_rmem = sysctl_sctp_rmem, 6981 .sysctl_rmem = sysctl_sctp_rmem,
6980 .sysctl_wmem = sysctl_sctp_wmem, 6982 .sysctl_wmem = sysctl_sctp_wmem,
6981 .memory_pressure = &sctp_memory_pressure, 6983 .memory_pressure = &sctp_memory_pressure,
6982 .enter_memory_pressure = sctp_enter_memory_pressure, 6984 .enter_memory_pressure = sctp_enter_memory_pressure,
6983 .memory_allocated = &sctp_memory_allocated, 6985 .memory_allocated = &sctp_memory_allocated,
6984 .sockets_allocated = &sctp_sockets_allocated, 6986 .sockets_allocated = &sctp_sockets_allocated,
6985 }; 6987 };
6986 6988
6987 #if IS_ENABLED(CONFIG_IPV6) 6989 #if IS_ENABLED(CONFIG_IPV6)
6988 6990
6989 struct proto sctpv6_prot = { 6991 struct proto sctpv6_prot = {
6990 .name = "SCTPv6", 6992 .name = "SCTPv6",
6991 .owner = THIS_MODULE, 6993 .owner = THIS_MODULE,
6992 .close = sctp_close, 6994 .close = sctp_close,
6993 .connect = sctp_connect, 6995 .connect = sctp_connect,
6994 .disconnect = sctp_disconnect, 6996 .disconnect = sctp_disconnect,
6995 .accept = sctp_accept, 6997 .accept = sctp_accept,
6996 .ioctl = sctp_ioctl, 6998 .ioctl = sctp_ioctl,
6997 .init = sctp_init_sock, 6999 .init = sctp_init_sock,
6998 .destroy = sctp_destroy_sock, 7000 .destroy = sctp_destroy_sock,
6999 .shutdown = sctp_shutdown, 7001 .shutdown = sctp_shutdown,
7000 .setsockopt = sctp_setsockopt, 7002 .setsockopt = sctp_setsockopt,
7001 .getsockopt = sctp_getsockopt, 7003 .getsockopt = sctp_getsockopt,
7002 .sendmsg = sctp_sendmsg, 7004 .sendmsg = sctp_sendmsg,
7003 .recvmsg = sctp_recvmsg, 7005 .recvmsg = sctp_recvmsg,
7004 .bind = sctp_bind, 7006 .bind = sctp_bind,
7005 .backlog_rcv = sctp_backlog_rcv, 7007 .backlog_rcv = sctp_backlog_rcv,
7006 .hash = sctp_hash, 7008 .hash = sctp_hash,
7007 .unhash = sctp_unhash, 7009 .unhash = sctp_unhash,
7008 .get_port = sctp_get_port, 7010 .get_port = sctp_get_port,
7009 .obj_size = sizeof(struct sctp6_sock), 7011 .obj_size = sizeof(struct sctp6_sock),
7010 .sysctl_mem = sysctl_sctp_mem, 7012 .sysctl_mem = sysctl_sctp_mem,
7011 .sysctl_rmem = sysctl_sctp_rmem, 7013 .sysctl_rmem = sysctl_sctp_rmem,
7012 .sysctl_wmem = sysctl_sctp_wmem, 7014 .sysctl_wmem = sysctl_sctp_wmem,
7013 .memory_pressure = &sctp_memory_pressure, 7015 .memory_pressure = &sctp_memory_pressure,
7014 .enter_memory_pressure = sctp_enter_memory_pressure, 7016 .enter_memory_pressure = sctp_enter_memory_pressure,
7015 .memory_allocated = &sctp_memory_allocated, 7017 .memory_allocated = &sctp_memory_allocated,
7016 .sockets_allocated = &sctp_sockets_allocated, 7018 .sockets_allocated = &sctp_sockets_allocated,
7017 }; 7019 };
7018 #endif /* IS_ENABLED(CONFIG_IPV6) */ 7020 #endif /* IS_ENABLED(CONFIG_IPV6) */
7019 7021