Diff markup
1 /* 1 /*
2 * INET An implementation of the TCP/I 2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is imp 3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of comm 4 * interface as the means of communication with the user level.
5 * 5 *
6 * RAW - implementation of IP "ra 6 * RAW - implementation of IP "raw" sockets.
7 * 7 *
8 * Version: $Id: raw.c,v 1.64 2002/02/01 2 <<
9 * <<
10 * Authors: Ross Biro 8 * Authors: Ross Biro
11 * Fred N. van Kempen, <waltje@uW 9 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12 * 10 *
13 * Fixes: 11 * Fixes:
14 * Alan Cox : verify 12 * Alan Cox : verify_area() fixed up
15 * Alan Cox : ICMP e 13 * Alan Cox : ICMP error handling
16 * Alan Cox : EMSGSI 14 * Alan Cox : EMSGSIZE if you send too big a packet
17 * Alan Cox : Now us 15 * Alan Cox : Now uses generic datagrams and shared
18 * skbuff 16 * skbuff library. No more peek crashes,
19 * no mor 17 * no more backlogs
20 * Alan Cox : Checks 18 * Alan Cox : Checks sk->broadcast.
21 * Alan Cox : Uses s 19 * Alan Cox : Uses skb_free_datagram/skb_copy_datagram
22 * Alan Cox : Raw pa 20 * Alan Cox : Raw passes ip options too
23 * Alan Cox : Setsoc 21 * Alan Cox : Setsocketopt added
24 * Alan Cox : Fixed 22 * Alan Cox : Fixed error return for broadcasts
25 * Alan Cox : Remove 23 * Alan Cox : Removed wake_up calls
26 * Alan Cox : Use tt 24 * Alan Cox : Use ttl/tos
27 * Alan Cox : Cleane 25 * Alan Cox : Cleaned up old debugging
28 * Alan Cox : Use ne 26 * Alan Cox : Use new kernel side addresses
29 * Arnt Gulbrandsen : Fixed 27 * Arnt Gulbrandsen : Fixed MSG_DONTROUTE in raw sockets.
30 * Alan Cox : BSD st 28 * Alan Cox : BSD style RAW socket demultiplexing.
31 * Alan Cox : Beginn 29 * Alan Cox : Beginnings of mrouted support.
32 * Alan Cox : Added 30 * Alan Cox : Added IP_HDRINCL option.
33 * Alan Cox : Skip b 31 * Alan Cox : Skip broadcast check if BSDism set.
34 * David S. Miller : New so 32 * David S. Miller : New socket lookup architecture.
35 * 33 *
36 * This program is free software; 34 * This program is free software; you can redistribute it and/or
37 * modify it under the terms of t 35 * modify it under the terms of the GNU General Public License
38 * as published by the Free Softw 36 * as published by the Free Software Foundation; either version
39 * 2 of the License, or (at your 37 * 2 of the License, or (at your option) any later version.
40 */ 38 */
41 39
42 #include <linux/types.h> 40 #include <linux/types.h>
43 #include <asm/atomic.h> 41 #include <asm/atomic.h>
44 #include <asm/byteorder.h> 42 #include <asm/byteorder.h>
45 #include <asm/current.h> 43 #include <asm/current.h>
46 #include <asm/uaccess.h> 44 #include <asm/uaccess.h>
47 #include <asm/ioctls.h> 45 #include <asm/ioctls.h>
48 #include <linux/stddef.h> 46 #include <linux/stddef.h>
49 #include <linux/slab.h> 47 #include <linux/slab.h>
50 #include <linux/errno.h> 48 #include <linux/errno.h>
51 #include <linux/aio.h> 49 #include <linux/aio.h>
52 #include <linux/kernel.h> 50 #include <linux/kernel.h>
53 #include <linux/spinlock.h> 51 #include <linux/spinlock.h>
54 #include <linux/sockios.h> 52 #include <linux/sockios.h>
55 #include <linux/socket.h> 53 #include <linux/socket.h>
56 #include <linux/in.h> 54 #include <linux/in.h>
57 #include <linux/mroute.h> 55 #include <linux/mroute.h>
58 #include <linux/netdevice.h> 56 #include <linux/netdevice.h>
59 #include <linux/in_route.h> 57 #include <linux/in_route.h>
60 #include <linux/route.h> 58 #include <linux/route.h>
61 #include <linux/skbuff.h> 59 #include <linux/skbuff.h>
62 #include <net/net_namespace.h> 60 #include <net/net_namespace.h>
63 #include <net/dst.h> 61 #include <net/dst.h>
64 #include <net/sock.h> 62 #include <net/sock.h>
65 #include <linux/gfp.h> 63 #include <linux/gfp.h>
66 #include <linux/ip.h> 64 #include <linux/ip.h>
67 #include <linux/net.h> 65 #include <linux/net.h>
68 #include <net/ip.h> 66 #include <net/ip.h>
69 #include <net/icmp.h> 67 #include <net/icmp.h>
70 #include <net/udp.h> 68 #include <net/udp.h>
71 #include <net/raw.h> 69 #include <net/raw.h>
72 #include <net/snmp.h> 70 #include <net/snmp.h>
73 #include <net/tcp_states.h> 71 #include <net/tcp_states.h>
74 #include <net/inet_common.h> 72 #include <net/inet_common.h>
75 #include <net/checksum.h> 73 #include <net/checksum.h>
76 #include <net/xfrm.h> 74 #include <net/xfrm.h>
77 #include <linux/rtnetlink.h> 75 #include <linux/rtnetlink.h>
78 #include <linux/proc_fs.h> 76 #include <linux/proc_fs.h>
79 #include <linux/seq_file.h> 77 #include <linux/seq_file.h>
80 #include <linux/netfilter.h> 78 #include <linux/netfilter.h>
81 #include <linux/netfilter_ipv4.h> 79 #include <linux/netfilter_ipv4.h>
82 80
83 static struct raw_hashinfo raw_v4_hashinfo = { 81 static struct raw_hashinfo raw_v4_hashinfo = {
84 .lock = __RW_LOCK_UNLOCKED(), !! 82 .lock = __RW_LOCK_UNLOCKED(raw_v4_hashinfo.lock),
85 }; 83 };
86 84
87 void raw_hash_sk(struct sock *sk, struct raw_h !! 85 void raw_hash_sk(struct sock *sk)
88 { 86 {
>> 87 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
89 struct hlist_head *head; 88 struct hlist_head *head;
90 89
91 head = &h->ht[inet_sk(sk)->num & (RAW_ 90 head = &h->ht[inet_sk(sk)->num & (RAW_HTABLE_SIZE - 1)];
92 91
93 write_lock_bh(&h->lock); 92 write_lock_bh(&h->lock);
94 sk_add_node(sk, head); 93 sk_add_node(sk, head);
95 sock_prot_inuse_add(sk->sk_prot, 1); !! 94 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
96 write_unlock_bh(&h->lock); 95 write_unlock_bh(&h->lock);
97 } 96 }
98 EXPORT_SYMBOL_GPL(raw_hash_sk); 97 EXPORT_SYMBOL_GPL(raw_hash_sk);
99 98
100 void raw_unhash_sk(struct sock *sk, struct raw !! 99 void raw_unhash_sk(struct sock *sk)
101 { 100 {
>> 101 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
>> 102
102 write_lock_bh(&h->lock); 103 write_lock_bh(&h->lock);
103 if (sk_del_node_init(sk)) 104 if (sk_del_node_init(sk))
104 sock_prot_inuse_add(sk->sk_pro !! 105 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
105 write_unlock_bh(&h->lock); 106 write_unlock_bh(&h->lock);
106 } 107 }
107 EXPORT_SYMBOL_GPL(raw_unhash_sk); 108 EXPORT_SYMBOL_GPL(raw_unhash_sk);
108 109
109 static void raw_v4_hash(struct sock *sk) <<
110 { <<
111 raw_hash_sk(sk, &raw_v4_hashinfo); <<
112 } <<
113 <<
114 static void raw_v4_unhash(struct sock *sk) <<
115 { <<
116 raw_unhash_sk(sk, &raw_v4_hashinfo); <<
117 } <<
118 <<
119 static struct sock *__raw_v4_lookup(struct net 110 static struct sock *__raw_v4_lookup(struct net *net, struct sock *sk,
120 unsigned short num, __be32 rad 111 unsigned short num, __be32 raddr, __be32 laddr, int dif)
121 { 112 {
122 struct hlist_node *node; 113 struct hlist_node *node;
123 114
124 sk_for_each_from(sk, node) { 115 sk_for_each_from(sk, node) {
125 struct inet_sock *inet = inet_ 116 struct inet_sock *inet = inet_sk(sk);
126 117
127 if (sk->sk_net == net && inet- !! 118 if (net_eq(sock_net(sk), net) && inet->num == num &&
128 !(inet->daddr && inet->dad 119 !(inet->daddr && inet->daddr != raddr) &&
129 !(inet->rcv_saddr && inet- 120 !(inet->rcv_saddr && inet->rcv_saddr != laddr) &&
130 !(sk->sk_bound_dev_if && s 121 !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
131 goto found; /* gotcha 122 goto found; /* gotcha */
132 } 123 }
133 sk = NULL; 124 sk = NULL;
134 found: 125 found:
135 return sk; 126 return sk;
136 } 127 }
137 128
138 /* 129 /*
139 * 0 - deliver 130 * 0 - deliver
140 * 1 - block 131 * 1 - block
141 */ 132 */
142 static __inline__ int icmp_filter(struct sock 133 static __inline__ int icmp_filter(struct sock *sk, struct sk_buff *skb)
143 { 134 {
144 int type; 135 int type;
145 136
146 if (!pskb_may_pull(skb, sizeof(struct 137 if (!pskb_may_pull(skb, sizeof(struct icmphdr)))
147 return 1; 138 return 1;
148 139
149 type = icmp_hdr(skb)->type; 140 type = icmp_hdr(skb)->type;
150 if (type < 32) { 141 if (type < 32) {
151 __u32 data = raw_sk(sk)->filte 142 __u32 data = raw_sk(sk)->filter.data;
152 143
153 return ((1 << type) & data) != 144 return ((1 << type) & data) != 0;
154 } 145 }
155 146
156 /* Do not block unknown ICMP types */ 147 /* Do not block unknown ICMP types */
157 return 0; 148 return 0;
158 } 149 }
159 150
160 /* IP input processing comes here for RAW sock 151 /* IP input processing comes here for RAW socket delivery.
161 * Caller owns SKB, so we must make clones. 152 * Caller owns SKB, so we must make clones.
162 * 153 *
163 * RFC 1122: SHOULD pass TOS value up to the t 154 * RFC 1122: SHOULD pass TOS value up to the transport layer.
164 * -> It does. And not only TOS, but all IP he 155 * -> It does. And not only TOS, but all IP header.
165 */ 156 */
166 static int raw_v4_input(struct sk_buff *skb, s 157 static int raw_v4_input(struct sk_buff *skb, struct iphdr *iph, int hash)
167 { 158 {
168 struct sock *sk; 159 struct sock *sk;
169 struct hlist_head *head; 160 struct hlist_head *head;
170 int delivered = 0; 161 int delivered = 0;
171 struct net *net; 162 struct net *net;
172 163
173 read_lock(&raw_v4_hashinfo.lock); 164 read_lock(&raw_v4_hashinfo.lock);
174 head = &raw_v4_hashinfo.ht[hash]; 165 head = &raw_v4_hashinfo.ht[hash];
175 if (hlist_empty(head)) 166 if (hlist_empty(head))
176 goto out; 167 goto out;
177 168
178 net = skb->dev->nd_net; !! 169 net = dev_net(skb->dev);
179 sk = __raw_v4_lookup(net, __sk_head(he 170 sk = __raw_v4_lookup(net, __sk_head(head), iph->protocol,
180 iph->saddr, iph-> 171 iph->saddr, iph->daddr,
181 skb->dev->ifindex 172 skb->dev->ifindex);
182 173
183 while (sk) { 174 while (sk) {
184 delivered = 1; 175 delivered = 1;
185 if (iph->protocol != IPPROTO_I 176 if (iph->protocol != IPPROTO_ICMP || !icmp_filter(sk, skb)) {
186 struct sk_buff *clone 177 struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
187 178
188 /* Not releasing hash 179 /* Not releasing hash table! */
189 if (clone) 180 if (clone)
190 raw_rcv(sk, cl 181 raw_rcv(sk, clone);
191 } 182 }
192 sk = __raw_v4_lookup(net, sk_n 183 sk = __raw_v4_lookup(net, sk_next(sk), iph->protocol,
193 iph->sadd 184 iph->saddr, iph->daddr,
194 skb->dev- 185 skb->dev->ifindex);
195 } 186 }
196 out: 187 out:
197 read_unlock(&raw_v4_hashinfo.lock); 188 read_unlock(&raw_v4_hashinfo.lock);
198 return delivered; 189 return delivered;
199 } 190 }
200 191
201 int raw_local_deliver(struct sk_buff *skb, int 192 int raw_local_deliver(struct sk_buff *skb, int protocol)
202 { 193 {
203 int hash; 194 int hash;
204 struct sock *raw_sk; 195 struct sock *raw_sk;
205 196
206 hash = protocol & (RAW_HTABLE_SIZE - 1 197 hash = protocol & (RAW_HTABLE_SIZE - 1);
207 raw_sk = sk_head(&raw_v4_hashinfo.ht[h 198 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
208 199
209 /* If there maybe a raw socket we must 200 /* If there maybe a raw socket we must check - if not we
210 * don't care less 201 * don't care less
211 */ 202 */
212 if (raw_sk && !raw_v4_input(skb, ip_hd 203 if (raw_sk && !raw_v4_input(skb, ip_hdr(skb), hash))
213 raw_sk = NULL; 204 raw_sk = NULL;
214 205
215 return raw_sk != NULL; 206 return raw_sk != NULL;
216 207
217 } 208 }
218 209
219 static void raw_err(struct sock *sk, struct sk 210 static void raw_err(struct sock *sk, struct sk_buff *skb, u32 info)
220 { 211 {
221 struct inet_sock *inet = inet_sk(sk); 212 struct inet_sock *inet = inet_sk(sk);
222 const int type = icmp_hdr(skb)->type; 213 const int type = icmp_hdr(skb)->type;
223 const int code = icmp_hdr(skb)->code; 214 const int code = icmp_hdr(skb)->code;
224 int err = 0; 215 int err = 0;
225 int harderr = 0; 216 int harderr = 0;
226 217
227 /* Report error on raw socket, if: 218 /* Report error on raw socket, if:
228 1. User requested ip_recverr. 219 1. User requested ip_recverr.
229 2. Socket is connected (otherwise t 220 2. Socket is connected (otherwise the error indication
230 is useless without ip_recverr an 221 is useless without ip_recverr and error is hard.
231 */ 222 */
232 if (!inet->recverr && sk->sk_state != 223 if (!inet->recverr && sk->sk_state != TCP_ESTABLISHED)
233 return; 224 return;
234 225
235 switch (type) { 226 switch (type) {
236 default: 227 default:
237 case ICMP_TIME_EXCEEDED: 228 case ICMP_TIME_EXCEEDED:
238 err = EHOSTUNREACH; 229 err = EHOSTUNREACH;
239 break; 230 break;
240 case ICMP_SOURCE_QUENCH: 231 case ICMP_SOURCE_QUENCH:
241 return; 232 return;
242 case ICMP_PARAMETERPROB: 233 case ICMP_PARAMETERPROB:
243 err = EPROTO; 234 err = EPROTO;
244 harderr = 1; 235 harderr = 1;
245 break; 236 break;
246 case ICMP_DEST_UNREACH: 237 case ICMP_DEST_UNREACH:
247 err = EHOSTUNREACH; 238 err = EHOSTUNREACH;
248 if (code > NR_ICMP_UNREACH) 239 if (code > NR_ICMP_UNREACH)
249 break; 240 break;
250 err = icmp_err_convert[code].e 241 err = icmp_err_convert[code].errno;
251 harderr = icmp_err_convert[cod 242 harderr = icmp_err_convert[code].fatal;
252 if (code == ICMP_FRAG_NEEDED) 243 if (code == ICMP_FRAG_NEEDED) {
253 harderr = inet->pmtudi 244 harderr = inet->pmtudisc != IP_PMTUDISC_DONT;
254 err = EMSGSIZE; 245 err = EMSGSIZE;
255 } 246 }
256 } 247 }
257 248
258 if (inet->recverr) { 249 if (inet->recverr) {
259 struct iphdr *iph = (struct ip !! 250 struct iphdr *iph = (struct iphdr *)skb->data;
260 u8 *payload = skb->data + (iph 251 u8 *payload = skb->data + (iph->ihl << 2);
261 252
262 if (inet->hdrincl) 253 if (inet->hdrincl)
263 payload = skb->data; 254 payload = skb->data;
264 ip_icmp_error(sk, skb, err, 0, 255 ip_icmp_error(sk, skb, err, 0, info, payload);
265 } 256 }
266 257
267 if (inet->recverr || harderr) { 258 if (inet->recverr || harderr) {
268 sk->sk_err = err; 259 sk->sk_err = err;
269 sk->sk_error_report(sk); 260 sk->sk_error_report(sk);
270 } 261 }
271 } 262 }
272 263
273 void raw_icmp_error(struct sk_buff *skb, int p 264 void raw_icmp_error(struct sk_buff *skb, int protocol, u32 info)
274 { 265 {
275 int hash; 266 int hash;
276 struct sock *raw_sk; 267 struct sock *raw_sk;
277 struct iphdr *iph; 268 struct iphdr *iph;
278 struct net *net; 269 struct net *net;
279 270
280 hash = protocol & (RAW_HTABLE_SIZE - 1 271 hash = protocol & (RAW_HTABLE_SIZE - 1);
281 272
282 read_lock(&raw_v4_hashinfo.lock); 273 read_lock(&raw_v4_hashinfo.lock);
283 raw_sk = sk_head(&raw_v4_hashinfo.ht[h 274 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
284 if (raw_sk != NULL) { 275 if (raw_sk != NULL) {
285 iph = (struct iphdr *)skb->dat 276 iph = (struct iphdr *)skb->data;
286 net = skb->dev->nd_net; !! 277 net = dev_net(skb->dev);
287 278
288 while ((raw_sk = __raw_v4_look 279 while ((raw_sk = __raw_v4_lookup(net, raw_sk, protocol,
289 280 iph->daddr, iph->saddr,
290 281 skb->dev->ifindex)) != NULL) {
291 raw_err(raw_sk, skb, i 282 raw_err(raw_sk, skb, info);
292 raw_sk = sk_next(raw_s 283 raw_sk = sk_next(raw_sk);
293 iph = (struct iphdr *) 284 iph = (struct iphdr *)skb->data;
294 } 285 }
295 } 286 }
296 read_unlock(&raw_v4_hashinfo.lock); 287 read_unlock(&raw_v4_hashinfo.lock);
297 } 288 }
298 289
299 static int raw_rcv_skb(struct sock * sk, struc 290 static int raw_rcv_skb(struct sock * sk, struct sk_buff * skb)
300 { 291 {
301 /* Charge it to the socket. */ 292 /* Charge it to the socket. */
302 293
303 if (sock_queue_rcv_skb(sk, skb) < 0) { 294 if (sock_queue_rcv_skb(sk, skb) < 0) {
304 atomic_inc(&sk->sk_drops); 295 atomic_inc(&sk->sk_drops);
305 kfree_skb(skb); 296 kfree_skb(skb);
306 return NET_RX_DROP; 297 return NET_RX_DROP;
307 } 298 }
308 299
309 return NET_RX_SUCCESS; 300 return NET_RX_SUCCESS;
310 } 301 }
311 302
312 int raw_rcv(struct sock *sk, struct sk_buff *s 303 int raw_rcv(struct sock *sk, struct sk_buff *skb)
313 { 304 {
314 if (!xfrm4_policy_check(sk, XFRM_POLIC 305 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
315 atomic_inc(&sk->sk_drops); 306 atomic_inc(&sk->sk_drops);
316 kfree_skb(skb); 307 kfree_skb(skb);
317 return NET_RX_DROP; 308 return NET_RX_DROP;
318 } 309 }
319 nf_reset(skb); 310 nf_reset(skb);
320 311
321 skb_push(skb, skb->data - skb_network_ 312 skb_push(skb, skb->data - skb_network_header(skb));
322 313
323 raw_rcv_skb(sk, skb); 314 raw_rcv_skb(sk, skb);
324 return 0; 315 return 0;
325 } 316 }
326 317
327 static int raw_send_hdrinc(struct sock *sk, vo 318 static int raw_send_hdrinc(struct sock *sk, void *from, size_t length,
328 struct rtable *rt, 319 struct rtable *rt,
329 unsigned int flags) 320 unsigned int flags)
330 { 321 {
331 struct inet_sock *inet = inet_sk(sk); 322 struct inet_sock *inet = inet_sk(sk);
332 int hh_len; !! 323 struct net *net = sock_net(sk);
333 struct iphdr *iph; 324 struct iphdr *iph;
334 struct sk_buff *skb; 325 struct sk_buff *skb;
335 unsigned int iphlen; 326 unsigned int iphlen;
336 int err; 327 int err;
337 328
338 if (length > rt->u.dst.dev->mtu) { 329 if (length > rt->u.dst.dev->mtu) {
339 ip_local_error(sk, EMSGSIZE, r 330 ip_local_error(sk, EMSGSIZE, rt->rt_dst, inet->dport,
340 rt->u.dst.dev-> 331 rt->u.dst.dev->mtu);
341 return -EMSGSIZE; 332 return -EMSGSIZE;
342 } 333 }
343 if (flags&MSG_PROBE) 334 if (flags&MSG_PROBE)
344 goto out; 335 goto out;
345 336
346 hh_len = LL_RESERVED_SPACE(rt->u.dst.d !! 337 skb = sock_alloc_send_skb(sk,
347 !! 338 length + LL_ALLOCATED_SPACE(rt->u.dst.dev) + 15,
348 skb = sock_alloc_send_skb(sk, length+h !! 339 flags & MSG_DONTWAIT, &err);
349 flags&MSG_DO <<
350 if (skb == NULL) 340 if (skb == NULL)
351 goto error; 341 goto error;
352 skb_reserve(skb, hh_len); !! 342 skb_reserve(skb, LL_RESERVED_SPACE(rt->u.dst.dev));
353 343
354 skb->priority = sk->sk_priority; 344 skb->priority = sk->sk_priority;
355 skb->mark = sk->sk_mark; 345 skb->mark = sk->sk_mark;
356 skb->dst = dst_clone(&rt->u.dst); !! 346 skb_dst_set(skb, dst_clone(&rt->u.dst));
357 347
358 skb_reset_network_header(skb); 348 skb_reset_network_header(skb);
359 iph = ip_hdr(skb); 349 iph = ip_hdr(skb);
360 skb_put(skb, length); 350 skb_put(skb, length);
361 351
362 skb->ip_summed = CHECKSUM_NONE; 352 skb->ip_summed = CHECKSUM_NONE;
363 353
364 skb->transport_header = skb->network_h 354 skb->transport_header = skb->network_header;
365 err = memcpy_fromiovecend((void *)iph, 355 err = memcpy_fromiovecend((void *)iph, from, 0, length);
366 if (err) 356 if (err)
367 goto error_fault; 357 goto error_fault;
368 358
369 /* We don't modify invalid header */ 359 /* We don't modify invalid header */
370 iphlen = iph->ihl * 4; 360 iphlen = iph->ihl * 4;
371 if (iphlen >= sizeof(*iph) && iphlen < 361 if (iphlen >= sizeof(*iph) && iphlen <= length) {
372 if (!iph->saddr) 362 if (!iph->saddr)
373 iph->saddr = rt->rt_sr 363 iph->saddr = rt->rt_src;
374 iph->check = 0; 364 iph->check = 0;
375 iph->tot_len = htons(length); 365 iph->tot_len = htons(length);
376 if (!iph->id) 366 if (!iph->id)
377 ip_select_ident(iph, & 367 ip_select_ident(iph, &rt->u.dst, NULL);
378 368
379 iph->check = ip_fast_csum((uns 369 iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
380 } 370 }
381 if (iph->protocol == IPPROTO_ICMP) 371 if (iph->protocol == IPPROTO_ICMP)
382 icmp_out_count(((struct icmphd !! 372 icmp_out_count(net, ((struct icmphdr *)
383 skb_transport_header(s 373 skb_transport_header(skb))->type);
384 374
385 err = NF_HOOK(PF_INET, NF_INET_LOCAL_O 375 err = NF_HOOK(PF_INET, NF_INET_LOCAL_OUT, skb, NULL, rt->u.dst.dev,
386 dst_output); 376 dst_output);
387 if (err > 0) 377 if (err > 0)
388 err = inet->recverr ? net_xmit 378 err = inet->recverr ? net_xmit_errno(err) : 0;
389 if (err) 379 if (err)
390 goto error; 380 goto error;
391 out: 381 out:
392 return 0; 382 return 0;
393 383
394 error_fault: 384 error_fault:
395 err = -EFAULT; 385 err = -EFAULT;
396 kfree_skb(skb); 386 kfree_skb(skb);
397 error: 387 error:
398 IP_INC_STATS(IPSTATS_MIB_OUTDISCARDS); !! 388 IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
399 return err; 389 return err;
400 } 390 }
401 391
402 static int raw_probe_proto_opt(struct flowi *f 392 static int raw_probe_proto_opt(struct flowi *fl, struct msghdr *msg)
403 { 393 {
404 struct iovec *iov; 394 struct iovec *iov;
405 u8 __user *type = NULL; 395 u8 __user *type = NULL;
406 u8 __user *code = NULL; 396 u8 __user *code = NULL;
407 int probed = 0; 397 int probed = 0;
408 unsigned int i; 398 unsigned int i;
409 399
410 if (!msg->msg_iov) 400 if (!msg->msg_iov)
411 return 0; 401 return 0;
412 402
413 for (i = 0; i < msg->msg_iovlen; i++) 403 for (i = 0; i < msg->msg_iovlen; i++) {
414 iov = &msg->msg_iov[i]; 404 iov = &msg->msg_iov[i];
415 if (!iov) 405 if (!iov)
416 continue; 406 continue;
417 407
418 switch (fl->proto) { 408 switch (fl->proto) {
419 case IPPROTO_ICMP: 409 case IPPROTO_ICMP:
420 /* check if one-byte f 410 /* check if one-byte field is readable or not. */
421 if (iov->iov_base && i 411 if (iov->iov_base && iov->iov_len < 1)
422 break; 412 break;
423 413
424 if (!type) { 414 if (!type) {
425 type = iov->io 415 type = iov->iov_base;
426 /* check if co 416 /* check if code field is readable or not. */
427 if (iov->iov_l 417 if (iov->iov_len > 1)
428 code = 418 code = type + 1;
429 } else if (!code) 419 } else if (!code)
430 code = iov->io 420 code = iov->iov_base;
431 421
432 if (type && code) { 422 if (type && code) {
433 if (get_user(f 423 if (get_user(fl->fl_icmp_type, type) ||
434 get_user(f 424 get_user(fl->fl_icmp_code, code))
435 return 425 return -EFAULT;
436 probed = 1; 426 probed = 1;
437 } 427 }
438 break; 428 break;
439 default: 429 default:
440 probed = 1; 430 probed = 1;
441 break; 431 break;
442 } 432 }
443 if (probed) 433 if (probed)
444 break; 434 break;
445 } 435 }
446 return 0; 436 return 0;
447 } 437 }
448 438
449 static int raw_sendmsg(struct kiocb *iocb, str 439 static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
450 size_t len) 440 size_t len)
451 { 441 {
452 struct inet_sock *inet = inet_sk(sk); 442 struct inet_sock *inet = inet_sk(sk);
453 struct ipcm_cookie ipc; 443 struct ipcm_cookie ipc;
454 struct rtable *rt = NULL; 444 struct rtable *rt = NULL;
455 int free = 0; 445 int free = 0;
456 __be32 daddr; 446 __be32 daddr;
457 __be32 saddr; 447 __be32 saddr;
458 u8 tos; 448 u8 tos;
459 int err; 449 int err;
460 450
461 err = -EMSGSIZE; 451 err = -EMSGSIZE;
462 if (len > 0xFFFF) 452 if (len > 0xFFFF)
463 goto out; 453 goto out;
464 454
465 /* 455 /*
466 * Check the flags. 456 * Check the flags.
467 */ 457 */
468 458
469 err = -EOPNOTSUPP; 459 err = -EOPNOTSUPP;
470 if (msg->msg_flags & MSG_OOB) /* Mir 460 if (msg->msg_flags & MSG_OOB) /* Mirror BSD error message */
471 goto out; /* com 461 goto out; /* compatibility */
472 462
473 /* 463 /*
474 * Get and verify the address. 464 * Get and verify the address.
475 */ 465 */
476 466
477 if (msg->msg_namelen) { 467 if (msg->msg_namelen) {
478 struct sockaddr_in *usin = (st !! 468 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
479 err = -EINVAL; 469 err = -EINVAL;
480 if (msg->msg_namelen < sizeof( 470 if (msg->msg_namelen < sizeof(*usin))
481 goto out; 471 goto out;
482 if (usin->sin_family != AF_INE 472 if (usin->sin_family != AF_INET) {
483 static int complained; 473 static int complained;
484 if (!complained++) 474 if (!complained++)
485 printk(KERN_IN 475 printk(KERN_INFO "%s forgot to set AF_INET in "
486 476 "raw sendmsg. Fix it!\n",
487 477 current->comm);
488 err = -EAFNOSUPPORT; 478 err = -EAFNOSUPPORT;
489 if (usin->sin_family) 479 if (usin->sin_family)
490 goto out; 480 goto out;
491 } 481 }
492 daddr = usin->sin_addr.s_addr; 482 daddr = usin->sin_addr.s_addr;
493 /* ANK: I did not forget to ge 483 /* ANK: I did not forget to get protocol from port field.
494 * I just do not know, who use 484 * I just do not know, who uses this weirdness.
495 * IP_HDRINCL is much more con 485 * IP_HDRINCL is much more convenient.
496 */ 486 */
497 } else { 487 } else {
498 err = -EDESTADDRREQ; 488 err = -EDESTADDRREQ;
499 if (sk->sk_state != TCP_ESTABL 489 if (sk->sk_state != TCP_ESTABLISHED)
500 goto out; 490 goto out;
501 daddr = inet->daddr; 491 daddr = inet->daddr;
502 } 492 }
503 493
504 ipc.addr = inet->saddr; 494 ipc.addr = inet->saddr;
505 ipc.opt = NULL; 495 ipc.opt = NULL;
>> 496 ipc.shtx.flags = 0;
506 ipc.oif = sk->sk_bound_dev_if; 497 ipc.oif = sk->sk_bound_dev_if;
507 498
508 if (msg->msg_controllen) { 499 if (msg->msg_controllen) {
509 err = ip_cmsg_send(msg, &ipc); !! 500 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
510 if (err) 501 if (err)
511 goto out; 502 goto out;
512 if (ipc.opt) 503 if (ipc.opt)
513 free = 1; 504 free = 1;
514 } 505 }
515 506
516 saddr = ipc.addr; 507 saddr = ipc.addr;
517 ipc.addr = daddr; 508 ipc.addr = daddr;
518 509
519 if (!ipc.opt) 510 if (!ipc.opt)
520 ipc.opt = inet->opt; 511 ipc.opt = inet->opt;
521 512
522 if (ipc.opt) { 513 if (ipc.opt) {
523 err = -EINVAL; 514 err = -EINVAL;
524 /* Linux does not mangle heade 515 /* Linux does not mangle headers on raw sockets,
525 * so that IP options + IP_HDR 516 * so that IP options + IP_HDRINCL is non-sense.
526 */ 517 */
527 if (inet->hdrincl) 518 if (inet->hdrincl)
528 goto done; 519 goto done;
529 if (ipc.opt->srr) { 520 if (ipc.opt->srr) {
530 if (!daddr) 521 if (!daddr)
531 goto done; 522 goto done;
532 daddr = ipc.opt->faddr 523 daddr = ipc.opt->faddr;
533 } 524 }
534 } 525 }
535 tos = RT_CONN_FLAGS(sk); 526 tos = RT_CONN_FLAGS(sk);
536 if (msg->msg_flags & MSG_DONTROUTE) 527 if (msg->msg_flags & MSG_DONTROUTE)
537 tos |= RTO_ONLINK; 528 tos |= RTO_ONLINK;
538 529
539 if (ipv4_is_multicast(daddr)) { 530 if (ipv4_is_multicast(daddr)) {
540 if (!ipc.oif) 531 if (!ipc.oif)
541 ipc.oif = inet->mc_ind 532 ipc.oif = inet->mc_index;
542 if (!saddr) 533 if (!saddr)
543 saddr = inet->mc_addr; 534 saddr = inet->mc_addr;
544 } 535 }
545 536
546 { 537 {
547 struct flowi fl = { .oif = ipc 538 struct flowi fl = { .oif = ipc.oif,
548 .mark = sk 539 .mark = sk->sk_mark,
549 .nl_u = { 540 .nl_u = { .ip4_u =
550 541 { .daddr = daddr,
551 542 .saddr = saddr,
552 543 .tos = tos } },
553 .proto = i 544 .proto = inet->hdrincl ? IPPROTO_RAW :
554 545 sk->sk_protocol,
555 }; 546 };
556 if (!inet->hdrincl) { 547 if (!inet->hdrincl) {
557 err = raw_probe_proto_ 548 err = raw_probe_proto_opt(&fl, msg);
558 if (err) 549 if (err)
559 goto done; 550 goto done;
560 } 551 }
561 552
562 security_sk_classify_flow(sk, 553 security_sk_classify_flow(sk, &fl);
563 err = ip_route_output_flow(&in !! 554 err = ip_route_output_flow(sock_net(sk), &rt, &fl, sk, 1);
564 } 555 }
565 if (err) 556 if (err)
566 goto done; 557 goto done;
567 558
568 err = -EACCES; 559 err = -EACCES;
569 if (rt->rt_flags & RTCF_BROADCAST && ! 560 if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST))
570 goto done; 561 goto done;
571 562
572 if (msg->msg_flags & MSG_CONFIRM) 563 if (msg->msg_flags & MSG_CONFIRM)
573 goto do_confirm; 564 goto do_confirm;
574 back_from_confirm: 565 back_from_confirm:
575 566
576 if (inet->hdrincl) 567 if (inet->hdrincl)
577 err = raw_send_hdrinc(sk, msg- 568 err = raw_send_hdrinc(sk, msg->msg_iov, len,
578 rt, ms 569 rt, msg->msg_flags);
579 570
580 else { 571 else {
581 if (!ipc.addr) 572 if (!ipc.addr)
582 ipc.addr = rt->rt_dst; 573 ipc.addr = rt->rt_dst;
583 lock_sock(sk); 574 lock_sock(sk);
584 err = ip_append_data(sk, ip_ge 575 err = ip_append_data(sk, ip_generic_getfrag, msg->msg_iov, len, 0,
585 &ipc, !! 576 &ipc, &rt, msg->msg_flags);
586 if (err) 577 if (err)
587 ip_flush_pending_frame 578 ip_flush_pending_frames(sk);
588 else if (!(msg->msg_flags & MS 579 else if (!(msg->msg_flags & MSG_MORE))
589 err = ip_push_pending_ 580 err = ip_push_pending_frames(sk);
590 release_sock(sk); 581 release_sock(sk);
591 } 582 }
592 done: 583 done:
593 if (free) 584 if (free)
594 kfree(ipc.opt); 585 kfree(ipc.opt);
595 ip_rt_put(rt); 586 ip_rt_put(rt);
596 587
597 out: 588 out:
598 if (err < 0) 589 if (err < 0)
599 return err; 590 return err;
600 return len; 591 return len;
601 592
602 do_confirm: 593 do_confirm:
603 dst_confirm(&rt->u.dst); 594 dst_confirm(&rt->u.dst);
604 if (!(msg->msg_flags & MSG_PROBE) || l 595 if (!(msg->msg_flags & MSG_PROBE) || len)
605 goto back_from_confirm; 596 goto back_from_confirm;
606 err = 0; 597 err = 0;
607 goto done; 598 goto done;
608 } 599 }
609 600
610 static void raw_close(struct sock *sk, long ti 601 static void raw_close(struct sock *sk, long timeout)
611 { 602 {
612 /* 603 /*
613 * Raw sockets may have direct kernel 604 * Raw sockets may have direct kernel refereneces. Kill them.
614 */ 605 */
615 ip_ra_control(sk, 0, NULL); 606 ip_ra_control(sk, 0, NULL);
616 607
617 sk_common_release(sk); 608 sk_common_release(sk);
618 } 609 }
619 610
>> 611 static void raw_destroy(struct sock *sk)
>> 612 {
>> 613 lock_sock(sk);
>> 614 ip_flush_pending_frames(sk);
>> 615 release_sock(sk);
>> 616 }
>> 617
620 /* This gets rid of all the nasties in af_inet 618 /* This gets rid of all the nasties in af_inet. -DaveM */
621 static int raw_bind(struct sock *sk, struct so 619 static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
622 { 620 {
623 struct inet_sock *inet = inet_sk(sk); 621 struct inet_sock *inet = inet_sk(sk);
624 struct sockaddr_in *addr = (struct soc 622 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
625 int ret = -EINVAL; 623 int ret = -EINVAL;
626 int chk_addr_ret; 624 int chk_addr_ret;
627 625
628 if (sk->sk_state != TCP_CLOSE || addr_ 626 if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
629 goto out; 627 goto out;
630 chk_addr_ret = inet_addr_type(sk->sk_n !! 628 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
631 ret = -EADDRNOTAVAIL; 629 ret = -EADDRNOTAVAIL;
632 if (addr->sin_addr.s_addr && chk_addr_ 630 if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
633 chk_addr_ret != RTN_MULTICAST && c 631 chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
634 goto out; 632 goto out;
635 inet->rcv_saddr = inet->saddr = addr-> 633 inet->rcv_saddr = inet->saddr = addr->sin_addr.s_addr;
636 if (chk_addr_ret == RTN_MULTICAST || c 634 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
637 inet->saddr = 0; /* Use devic 635 inet->saddr = 0; /* Use device */
638 sk_dst_reset(sk); 636 sk_dst_reset(sk);
639 ret = 0; 637 ret = 0;
640 out: return ret; 638 out: return ret;
641 } 639 }
642 640
643 /* 641 /*
644 * This should be easy, if there is somet 642 * This should be easy, if there is something there
645 * we return it, otherwise we block. 643 * we return it, otherwise we block.
646 */ 644 */
647 645
648 static int raw_recvmsg(struct kiocb *iocb, str 646 static int raw_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
649 size_t len, int noblock 647 size_t len, int noblock, int flags, int *addr_len)
650 { 648 {
651 struct inet_sock *inet = inet_sk(sk); 649 struct inet_sock *inet = inet_sk(sk);
652 size_t copied = 0; 650 size_t copied = 0;
653 int err = -EOPNOTSUPP; 651 int err = -EOPNOTSUPP;
654 struct sockaddr_in *sin = (struct sock 652 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
655 struct sk_buff *skb; 653 struct sk_buff *skb;
656 654
657 if (flags & MSG_OOB) 655 if (flags & MSG_OOB)
658 goto out; 656 goto out;
659 657
660 if (addr_len) 658 if (addr_len)
661 *addr_len = sizeof(*sin); 659 *addr_len = sizeof(*sin);
662 660
663 if (flags & MSG_ERRQUEUE) { 661 if (flags & MSG_ERRQUEUE) {
664 err = ip_recv_error(sk, msg, l 662 err = ip_recv_error(sk, msg, len);
665 goto out; 663 goto out;
666 } 664 }
667 665
668 skb = skb_recv_datagram(sk, flags, nob 666 skb = skb_recv_datagram(sk, flags, noblock, &err);
669 if (!skb) 667 if (!skb)
670 goto out; 668 goto out;
671 669
672 copied = skb->len; 670 copied = skb->len;
673 if (len < copied) { 671 if (len < copied) {
674 msg->msg_flags |= MSG_TRUNC; 672 msg->msg_flags |= MSG_TRUNC;
675 copied = len; 673 copied = len;
676 } 674 }
677 675
678 err = skb_copy_datagram_iovec(skb, 0, 676 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
679 if (err) 677 if (err)
680 goto done; 678 goto done;
681 679
682 sock_recv_timestamp(msg, sk, skb); 680 sock_recv_timestamp(msg, sk, skb);
683 681
684 /* Copy the address. */ 682 /* Copy the address. */
685 if (sin) { 683 if (sin) {
686 sin->sin_family = AF_INET; 684 sin->sin_family = AF_INET;
687 sin->sin_addr.s_addr = ip_hdr( 685 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
688 sin->sin_port = 0; 686 sin->sin_port = 0;
689 memset(&sin->sin_zero, 0, size 687 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
690 } 688 }
691 if (inet->cmsg_flags) 689 if (inet->cmsg_flags)
692 ip_cmsg_recv(msg, skb); 690 ip_cmsg_recv(msg, skb);
693 if (flags & MSG_TRUNC) 691 if (flags & MSG_TRUNC)
694 copied = skb->len; 692 copied = skb->len;
695 done: 693 done:
696 skb_free_datagram(sk, skb); 694 skb_free_datagram(sk, skb);
697 out: 695 out:
698 if (err) 696 if (err)
699 return err; 697 return err;
700 return copied; 698 return copied;
701 } 699 }
702 700
703 static int raw_init(struct sock *sk) 701 static int raw_init(struct sock *sk)
704 { 702 {
705 struct raw_sock *rp = raw_sk(sk); 703 struct raw_sock *rp = raw_sk(sk);
706 704
707 if (inet_sk(sk)->num == IPPROTO_ICMP) 705 if (inet_sk(sk)->num == IPPROTO_ICMP)
708 memset(&rp->filter, 0, sizeof( 706 memset(&rp->filter, 0, sizeof(rp->filter));
709 return 0; 707 return 0;
710 } 708 }
711 709
712 static int raw_seticmpfilter(struct sock *sk, 710 static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen)
713 { 711 {
714 if (optlen > sizeof(struct icmp_filter 712 if (optlen > sizeof(struct icmp_filter))
715 optlen = sizeof(struct icmp_fi 713 optlen = sizeof(struct icmp_filter);
716 if (copy_from_user(&raw_sk(sk)->filter 714 if (copy_from_user(&raw_sk(sk)->filter, optval, optlen))
717 return -EFAULT; 715 return -EFAULT;
718 return 0; 716 return 0;
719 } 717 }
720 718
721 static int raw_geticmpfilter(struct sock *sk, 719 static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen)
722 { 720 {
723 int len, ret = -EFAULT; 721 int len, ret = -EFAULT;
724 722
725 if (get_user(len, optlen)) 723 if (get_user(len, optlen))
726 goto out; 724 goto out;
727 ret = -EINVAL; 725 ret = -EINVAL;
728 if (len < 0) 726 if (len < 0)
729 goto out; 727 goto out;
730 if (len > sizeof(struct icmp_filter)) 728 if (len > sizeof(struct icmp_filter))
731 len = sizeof(struct icmp_filte 729 len = sizeof(struct icmp_filter);
732 ret = -EFAULT; 730 ret = -EFAULT;
733 if (put_user(len, optlen) || 731 if (put_user(len, optlen) ||
734 copy_to_user(optval, &raw_sk(sk)-> 732 copy_to_user(optval, &raw_sk(sk)->filter, len))
735 goto out; 733 goto out;
736 ret = 0; 734 ret = 0;
737 out: return ret; 735 out: return ret;
738 } 736 }
739 737
740 static int do_raw_setsockopt(struct sock *sk, 738 static int do_raw_setsockopt(struct sock *sk, int level, int optname,
741 char __user *optval, 739 char __user *optval, int optlen)
742 { 740 {
743 if (optname == ICMP_FILTER) { 741 if (optname == ICMP_FILTER) {
744 if (inet_sk(sk)->num != IPPROT 742 if (inet_sk(sk)->num != IPPROTO_ICMP)
745 return -EOPNOTSUPP; 743 return -EOPNOTSUPP;
746 else 744 else
747 return raw_seticmpfilt 745 return raw_seticmpfilter(sk, optval, optlen);
748 } 746 }
749 return -ENOPROTOOPT; 747 return -ENOPROTOOPT;
750 } 748 }
751 749
752 static int raw_setsockopt(struct sock *sk, int 750 static int raw_setsockopt(struct sock *sk, int level, int optname,
753 char __user *optval, 751 char __user *optval, int optlen)
754 { 752 {
755 if (level != SOL_RAW) 753 if (level != SOL_RAW)
756 return ip_setsockopt(sk, level 754 return ip_setsockopt(sk, level, optname, optval, optlen);
757 return do_raw_setsockopt(sk, level, op 755 return do_raw_setsockopt(sk, level, optname, optval, optlen);
758 } 756 }
759 757
760 #ifdef CONFIG_COMPAT 758 #ifdef CONFIG_COMPAT
761 static int compat_raw_setsockopt(struct sock * 759 static int compat_raw_setsockopt(struct sock *sk, int level, int optname,
762 char __user * 760 char __user *optval, int optlen)
763 { 761 {
764 if (level != SOL_RAW) 762 if (level != SOL_RAW)
765 return compat_ip_setsockopt(sk 763 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
766 return do_raw_setsockopt(sk, level, op 764 return do_raw_setsockopt(sk, level, optname, optval, optlen);
767 } 765 }
768 #endif 766 #endif
769 767
770 static int do_raw_getsockopt(struct sock *sk, 768 static int do_raw_getsockopt(struct sock *sk, int level, int optname,
771 char __user *optval, 769 char __user *optval, int __user *optlen)
772 { 770 {
773 if (optname == ICMP_FILTER) { 771 if (optname == ICMP_FILTER) {
774 if (inet_sk(sk)->num != IPPROT 772 if (inet_sk(sk)->num != IPPROTO_ICMP)
775 return -EOPNOTSUPP; 773 return -EOPNOTSUPP;
776 else 774 else
777 return raw_geticmpfilt 775 return raw_geticmpfilter(sk, optval, optlen);
778 } 776 }
779 return -ENOPROTOOPT; 777 return -ENOPROTOOPT;
780 } 778 }
781 779
782 static int raw_getsockopt(struct sock *sk, int 780 static int raw_getsockopt(struct sock *sk, int level, int optname,
783 char __user *optval, 781 char __user *optval, int __user *optlen)
784 { 782 {
785 if (level != SOL_RAW) 783 if (level != SOL_RAW)
786 return ip_getsockopt(sk, level 784 return ip_getsockopt(sk, level, optname, optval, optlen);
787 return do_raw_getsockopt(sk, level, op 785 return do_raw_getsockopt(sk, level, optname, optval, optlen);
788 } 786 }
789 787
790 #ifdef CONFIG_COMPAT 788 #ifdef CONFIG_COMPAT
791 static int compat_raw_getsockopt(struct sock * 789 static int compat_raw_getsockopt(struct sock *sk, int level, int optname,
792 char __user * 790 char __user *optval, int __user *optlen)
793 { 791 {
794 if (level != SOL_RAW) 792 if (level != SOL_RAW)
795 return compat_ip_getsockopt(sk 793 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
796 return do_raw_getsockopt(sk, level, op 794 return do_raw_getsockopt(sk, level, optname, optval, optlen);
797 } 795 }
798 #endif 796 #endif
799 797
800 static int raw_ioctl(struct sock *sk, int cmd, 798 static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg)
801 { 799 {
802 switch (cmd) { 800 switch (cmd) {
803 case SIOCOUTQ: { 801 case SIOCOUTQ: {
804 int amount = atomic_re !! 802 int amount = sk_wmem_alloc_get(sk);
>> 803
805 return put_user(amount 804 return put_user(amount, (int __user *)arg);
806 } 805 }
807 case SIOCINQ: { 806 case SIOCINQ: {
808 struct sk_buff *skb; 807 struct sk_buff *skb;
809 int amount = 0; 808 int amount = 0;
810 809
811 spin_lock_bh(&sk->sk_r 810 spin_lock_bh(&sk->sk_receive_queue.lock);
812 skb = skb_peek(&sk->sk 811 skb = skb_peek(&sk->sk_receive_queue);
813 if (skb != NULL) 812 if (skb != NULL)
814 amount = skb-> 813 amount = skb->len;
815 spin_unlock_bh(&sk->sk 814 spin_unlock_bh(&sk->sk_receive_queue.lock);
816 return put_user(amount 815 return put_user(amount, (int __user *)arg);
817 } 816 }
818 817
819 default: 818 default:
820 #ifdef CONFIG_IP_MROUTE 819 #ifdef CONFIG_IP_MROUTE
821 return ipmr_ioctl(sk, 820 return ipmr_ioctl(sk, cmd, (void __user *)arg);
822 #else 821 #else
823 return -ENOIOCTLCMD; 822 return -ENOIOCTLCMD;
824 #endif 823 #endif
825 } 824 }
826 } 825 }
827 826
828 DEFINE_PROTO_INUSE(raw) <<
829 <<
830 struct proto raw_prot = { 827 struct proto raw_prot = {
831 .name = "RAW", 828 .name = "RAW",
832 .owner = THIS_MODULE, 829 .owner = THIS_MODULE,
833 .close = raw_close, 830 .close = raw_close,
>> 831 .destroy = raw_destroy,
834 .connect = ip4_datagram_conn 832 .connect = ip4_datagram_connect,
835 .disconnect = udp_disconnect, 833 .disconnect = udp_disconnect,
836 .ioctl = raw_ioctl, 834 .ioctl = raw_ioctl,
837 .init = raw_init, 835 .init = raw_init,
838 .setsockopt = raw_setsockopt, 836 .setsockopt = raw_setsockopt,
839 .getsockopt = raw_getsockopt, 837 .getsockopt = raw_getsockopt,
840 .sendmsg = raw_sendmsg, 838 .sendmsg = raw_sendmsg,
841 .recvmsg = raw_recvmsg, 839 .recvmsg = raw_recvmsg,
842 .bind = raw_bind, 840 .bind = raw_bind,
843 .backlog_rcv = raw_rcv_skb, 841 .backlog_rcv = raw_rcv_skb,
844 .hash = raw_v4_hash, !! 842 .hash = raw_hash_sk,
845 .unhash = raw_v4_unhash, !! 843 .unhash = raw_unhash_sk,
846 .obj_size = sizeof(struct raw 844 .obj_size = sizeof(struct raw_sock),
>> 845 .h.raw_hash = &raw_v4_hashinfo,
847 #ifdef CONFIG_COMPAT 846 #ifdef CONFIG_COMPAT
848 .compat_setsockopt = compat_raw_setsoc 847 .compat_setsockopt = compat_raw_setsockopt,
849 .compat_getsockopt = compat_raw_getsoc 848 .compat_getsockopt = compat_raw_getsockopt,
850 #endif 849 #endif
851 REF_PROTO_INUSE(raw) <<
852 }; 850 };
853 851
854 #ifdef CONFIG_PROC_FS 852 #ifdef CONFIG_PROC_FS
855 static struct sock *raw_get_first(struct seq_f 853 static struct sock *raw_get_first(struct seq_file *seq)
856 { 854 {
857 struct sock *sk; 855 struct sock *sk;
858 struct raw_iter_state* state = raw_seq !! 856 struct raw_iter_state *state = raw_seq_private(seq);
859 857
860 for (state->bucket = 0; state->bucket 858 for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE;
861 ++state->bucket) { 859 ++state->bucket) {
862 struct hlist_node *node; 860 struct hlist_node *node;
863 861
864 sk_for_each(sk, node, &state-> 862 sk_for_each(sk, node, &state->h->ht[state->bucket])
865 if (sk->sk_net == stat !! 863 if (sock_net(sk) == seq_file_net(seq))
866 goto found; 864 goto found;
867 } 865 }
868 sk = NULL; 866 sk = NULL;
869 found: 867 found:
870 return sk; 868 return sk;
871 } 869 }
872 870
873 static struct sock *raw_get_next(struct seq_fi 871 static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk)
874 { 872 {
875 struct raw_iter_state* state = raw_seq !! 873 struct raw_iter_state *state = raw_seq_private(seq);
876 874
877 do { 875 do {
878 sk = sk_next(sk); 876 sk = sk_next(sk);
879 try_again: 877 try_again:
880 ; 878 ;
881 } while (sk && sk->sk_net != state->p. !! 879 } while (sk && sock_net(sk) != seq_file_net(seq));
882 880
883 if (!sk && ++state->bucket < RAW_HTABL 881 if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
884 sk = sk_head(&state->h->ht[sta 882 sk = sk_head(&state->h->ht[state->bucket]);
885 goto try_again; 883 goto try_again;
886 } 884 }
887 return sk; 885 return sk;
888 } 886 }
889 887
890 static struct sock *raw_get_idx(struct seq_fil 888 static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos)
891 { 889 {
892 struct sock *sk = raw_get_first(seq); 890 struct sock *sk = raw_get_first(seq);
893 891
894 if (sk) 892 if (sk)
895 while (pos && (sk = raw_get_ne 893 while (pos && (sk = raw_get_next(seq, sk)) != NULL)
896 --pos; 894 --pos;
897 return pos ? NULL : sk; 895 return pos ? NULL : sk;
898 } 896 }
899 897
900 void *raw_seq_start(struct seq_file *seq, loff 898 void *raw_seq_start(struct seq_file *seq, loff_t *pos)
901 { 899 {
902 struct raw_iter_state *state = raw_seq 900 struct raw_iter_state *state = raw_seq_private(seq);
903 901
904 read_lock(&state->h->lock); 902 read_lock(&state->h->lock);
905 return *pos ? raw_get_idx(seq, *pos - 903 return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
906 } 904 }
907 EXPORT_SYMBOL_GPL(raw_seq_start); 905 EXPORT_SYMBOL_GPL(raw_seq_start);
908 906
909 void *raw_seq_next(struct seq_file *seq, void 907 void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos)
910 { 908 {
911 struct sock *sk; 909 struct sock *sk;
912 910
913 if (v == SEQ_START_TOKEN) 911 if (v == SEQ_START_TOKEN)
914 sk = raw_get_first(seq); 912 sk = raw_get_first(seq);
915 else 913 else
916 sk = raw_get_next(seq, v); 914 sk = raw_get_next(seq, v);
917 ++*pos; 915 ++*pos;
918 return sk; 916 return sk;
919 } 917 }
920 EXPORT_SYMBOL_GPL(raw_seq_next); 918 EXPORT_SYMBOL_GPL(raw_seq_next);
921 919
922 void raw_seq_stop(struct seq_file *seq, void * 920 void raw_seq_stop(struct seq_file *seq, void *v)
923 { 921 {
924 struct raw_iter_state *state = raw_seq 922 struct raw_iter_state *state = raw_seq_private(seq);
925 923
926 read_unlock(&state->h->lock); 924 read_unlock(&state->h->lock);
927 } 925 }
928 EXPORT_SYMBOL_GPL(raw_seq_stop); 926 EXPORT_SYMBOL_GPL(raw_seq_stop);
929 927
930 static void raw_sock_seq_show(struct seq_file 928 static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
931 { 929 {
932 struct inet_sock *inet = inet_sk(sp); 930 struct inet_sock *inet = inet_sk(sp);
933 __be32 dest = inet->daddr, 931 __be32 dest = inet->daddr,
934 src = inet->rcv_saddr; 932 src = inet->rcv_saddr;
935 __u16 destp = 0, 933 __u16 destp = 0,
936 srcp = inet->num; 934 srcp = inet->num;
937 935
938 seq_printf(seq, "%4d: %08X:%04X %08X:% 936 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
939 " %02X %08X:%08X %02X:%08lX %0 !! 937 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p %d\n",
940 i, src, srcp, dest, destp, sp- 938 i, src, srcp, dest, destp, sp->sk_state,
941 atomic_read(&sp->sk_wmem_alloc !! 939 sk_wmem_alloc_get(sp),
942 atomic_read(&sp->sk_rmem_alloc !! 940 sk_rmem_alloc_get(sp),
943 0, 0L, 0, sock_i_uid(sp), 0, s 941 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
944 atomic_read(&sp->sk_refcnt), s 942 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
945 } 943 }
946 944
947 static int raw_seq_show(struct seq_file *seq, 945 static int raw_seq_show(struct seq_file *seq, void *v)
948 { 946 {
949 if (v == SEQ_START_TOKEN) 947 if (v == SEQ_START_TOKEN)
950 seq_printf(seq, " sl local_a 948 seq_printf(seq, " sl local_address rem_address st tx_queue "
951 "rx_queue tr t 949 "rx_queue tr tm->when retrnsmt uid timeout "
952 "inode drops\ !! 950 "inode ref pointer drops\n");
953 else 951 else
954 raw_sock_seq_show(seq, v, raw_ 952 raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
955 return 0; 953 return 0;
956 } 954 }
957 955
958 static const struct seq_operations raw_seq_ops 956 static const struct seq_operations raw_seq_ops = {
959 .start = raw_seq_start, 957 .start = raw_seq_start,
960 .next = raw_seq_next, 958 .next = raw_seq_next,
961 .stop = raw_seq_stop, 959 .stop = raw_seq_stop,
962 .show = raw_seq_show, 960 .show = raw_seq_show,
963 }; 961 };
964 962
965 int raw_seq_open(struct inode *ino, struct fil 963 int raw_seq_open(struct inode *ino, struct file *file,
966 struct raw_hashinfo *h, const 964 struct raw_hashinfo *h, const struct seq_operations *ops)
967 { 965 {
968 int err; 966 int err;
969 struct raw_iter_state *i; 967 struct raw_iter_state *i;
970 968
971 err = seq_open_net(ino, file, ops, siz 969 err = seq_open_net(ino, file, ops, sizeof(struct raw_iter_state));
972 if (err < 0) 970 if (err < 0)
973 return err; 971 return err;
974 972
975 i = raw_seq_private((struct seq_file * 973 i = raw_seq_private((struct seq_file *)file->private_data);
976 i->h = h; 974 i->h = h;
977 return 0; 975 return 0;
978 } 976 }
979 EXPORT_SYMBOL_GPL(raw_seq_open); 977 EXPORT_SYMBOL_GPL(raw_seq_open);
980 978
981 static int raw_v4_seq_open(struct inode *inode 979 static int raw_v4_seq_open(struct inode *inode, struct file *file)
982 { 980 {
983 return raw_seq_open(inode, file, &raw_ 981 return raw_seq_open(inode, file, &raw_v4_hashinfo, &raw_seq_ops);
984 } 982 }
985 983
986 static const struct file_operations raw_seq_fo 984 static const struct file_operations raw_seq_fops = {
987 .owner = THIS_MODULE, 985 .owner = THIS_MODULE,
988 .open = raw_v4_seq_open, 986 .open = raw_v4_seq_open,
989 .read = seq_read, 987 .read = seq_read,
990 .llseek = seq_lseek, 988 .llseek = seq_lseek,
991 .release = seq_release_net, 989 .release = seq_release_net,
992 }; 990 };
993 991
994 static __net_init int raw_init_net(struct net 992 static __net_init int raw_init_net(struct net *net)
995 { 993 {
996 if (!proc_net_fops_create(net, "raw", 994 if (!proc_net_fops_create(net, "raw", S_IRUGO, &raw_seq_fops))
997 return -ENOMEM; 995 return -ENOMEM;
998 996
999 return 0; 997 return 0;
1000 } 998 }
1001 999
1002 static __net_exit void raw_exit_net(struct ne 1000 static __net_exit void raw_exit_net(struct net *net)
1003 { 1001 {
1004 proc_net_remove(net, "raw"); 1002 proc_net_remove(net, "raw");
1005 } 1003 }
1006 1004
1007 static __net_initdata struct pernet_operation 1005 static __net_initdata struct pernet_operations raw_net_ops = {
1008 .init = raw_init_net, 1006 .init = raw_init_net,
1009 .exit = raw_exit_net, 1007 .exit = raw_exit_net,
1010 }; 1008 };
1011 1009
1012 int __init raw_proc_init(void) 1010 int __init raw_proc_init(void)
1013 { 1011 {
1014 return register_pernet_subsys(&raw_ne 1012 return register_pernet_subsys(&raw_net_ops);
1015 } 1013 }
1016 1014
1017 void __init raw_proc_exit(void) 1015 void __init raw_proc_exit(void)
1018 { 1016 {
1019 unregister_pernet_subsys(&raw_net_ops 1017 unregister_pernet_subsys(&raw_net_ops);
1020 } 1018 }
1021 #endif /* CONFIG_PROC_FS */ 1019 #endif /* CONFIG_PROC_FS */
1022 1020
|
This page was automatically generated by the
LXR engine.
|