]> bbs.cooldavid.org Git - net-next-2.6.git/blame - net/ipv4/udp.c
[NETFILTER]: Fix nf_conntrack_ftp.c build.
[net-next-2.6.git] / net / ipv4 / udp.c
CommitLineData
1da177e4
LT
1/*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
5 *
6 * The User Datagram Protocol (UDP).
7 *
8 * Version: $Id: udp.c,v 1.102 2002/02/01 22:01:04 davem Exp $
9 *
02c30a84 10 * Authors: Ross Biro
1da177e4
LT
11 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12 * Arnt Gulbrandsen, <agulbra@nvg.unit.no>
13 * Alan Cox, <Alan.Cox@linux.org>
14 * Hirokazu Takahashi, <taka@valinux.co.jp>
15 *
16 * Fixes:
17 * Alan Cox : verify_area() calls
18 * Alan Cox : stopped close while in use off icmp
19 * messages. Not a fix but a botch that
20 * for udp at least is 'valid'.
21 * Alan Cox : Fixed icmp handling properly
22 * Alan Cox : Correct error for oversized datagrams
23 * Alan Cox : Tidied select() semantics.
24 * Alan Cox : udp_err() fixed properly, also now
25 * select and read wake correctly on errors
26 * Alan Cox : udp_send verify_area moved to avoid mem leak
27 * Alan Cox : UDP can count its memory
28 * Alan Cox : send to an unknown connection causes
29 * an ECONNREFUSED off the icmp, but
30 * does NOT close.
31 * Alan Cox : Switched to new sk_buff handlers. No more backlog!
32 * Alan Cox : Using generic datagram code. Even smaller and the PEEK
33 * bug no longer crashes it.
34 * Fred Van Kempen : Net2e support for sk->broadcast.
35 * Alan Cox : Uses skb_free_datagram
36 * Alan Cox : Added get/set sockopt support.
37 * Alan Cox : Broadcasting without option set returns EACCES.
38 * Alan Cox : No wakeup calls. Instead we now use the callbacks.
39 * Alan Cox : Use ip_tos and ip_ttl
40 * Alan Cox : SNMP Mibs
41 * Alan Cox : MSG_DONTROUTE, and 0.0.0.0 support.
42 * Matt Dillon : UDP length checks.
43 * Alan Cox : Smarter af_inet used properly.
44 * Alan Cox : Use new kernel side addressing.
45 * Alan Cox : Incorrect return on truncated datagram receive.
46 * Arnt Gulbrandsen : New udp_send and stuff
47 * Alan Cox : Cache last socket
48 * Alan Cox : Route cache
49 * Jon Peatfield : Minor efficiency fix to sendto().
50 * Mike Shaver : RFC1122 checks.
51 * Alan Cox : Nonblocking error fix.
52 * Willy Konynenberg : Transparent proxying support.
53 * Mike McLagan : Routing by source
54 * David S. Miller : New socket lookup architecture.
55 * Last socket cache retained as it
56 * does have a high hit rate.
57 * Olaf Kirch : Don't linearise iovec on sendmsg.
58 * Andi Kleen : Some cleanups, cache destination entry
59 * for connect.
60 * Vitaly E. Lavrov : Transparent proxy revived after year coma.
61 * Melvin Smith : Check msg_name not msg_namelen in sendto(),
62 * return ENOTCONN for unconnected sockets (POSIX)
63 * Janos Farkas : don't deliver multi/broadcasts to a different
64 * bound-to-device socket
65 * Hirokazu Takahashi : HW checksumming for outgoing UDP
66 * datagrams.
67 * Hirokazu Takahashi : sendfile() on UDP works now.
68 * Arnaldo C. Melo : convert /proc/net/udp to seq_file
69 * YOSHIFUJI Hideaki @USAGI and: Support IPV6_V6ONLY socket option, which
70 * Alexey Kuznetsov: allow both IPv4 and IPv6 sockets to bind
71 * a single port at the same time.
72 * Derek Atkins <derek@ihtfp.com>: Add Encapulation Support
73 *
74 *
75 * This program is free software; you can redistribute it and/or
76 * modify it under the terms of the GNU General Public License
77 * as published by the Free Software Foundation; either version
78 * 2 of the License, or (at your option) any later version.
79 */
80
81#include <asm/system.h>
82#include <asm/uaccess.h>
83#include <asm/ioctls.h>
84#include <linux/types.h>
85#include <linux/fcntl.h>
86#include <linux/module.h>
87#include <linux/socket.h>
88#include <linux/sockios.h>
14c85021 89#include <linux/igmp.h>
1da177e4
LT
90#include <linux/in.h>
91#include <linux/errno.h>
92#include <linux/timer.h>
93#include <linux/mm.h>
1da177e4
LT
94#include <linux/inet.h>
95#include <linux/ipv6.h>
96#include <linux/netdevice.h>
97#include <net/snmp.h>
c752f073
ACM
98#include <net/ip.h>
99#include <net/tcp_states.h>
1da177e4
LT
100#include <net/protocol.h>
101#include <linux/skbuff.h>
102#include <linux/proc_fs.h>
103#include <linux/seq_file.h>
104#include <net/sock.h>
105#include <net/udp.h>
106#include <net/icmp.h>
107#include <net/route.h>
108#include <net/inet_common.h>
109#include <net/checksum.h>
110#include <net/xfrm.h>
111
112/*
113 * Snmp MIB for the UDP layer
114 */
115
ba89966c 116DEFINE_SNMP_STAT(struct udp_mib, udp_statistics) __read_mostly;
1da177e4
LT
117
118struct hlist_head udp_hash[UDP_HTABLE_SIZE];
119DEFINE_RWLOCK(udp_hash_lock);
120
121/* Shared by v4/v6 udp. */
122int udp_port_rover;
123
124static int udp_v4_get_port(struct sock *sk, unsigned short snum)
125{
126 struct hlist_node *node;
127 struct sock *sk2;
128 struct inet_sock *inet = inet_sk(sk);
129
130 write_lock_bh(&udp_hash_lock);
131 if (snum == 0) {
132 int best_size_so_far, best, result, i;
133
134 if (udp_port_rover > sysctl_local_port_range[1] ||
135 udp_port_rover < sysctl_local_port_range[0])
136 udp_port_rover = sysctl_local_port_range[0];
137 best_size_so_far = 32767;
138 best = result = udp_port_rover;
139 for (i = 0; i < UDP_HTABLE_SIZE; i++, result++) {
140 struct hlist_head *list;
141 int size;
142
143 list = &udp_hash[result & (UDP_HTABLE_SIZE - 1)];
144 if (hlist_empty(list)) {
145 if (result > sysctl_local_port_range[1])
146 result = sysctl_local_port_range[0] +
147 ((result - sysctl_local_port_range[0]) &
148 (UDP_HTABLE_SIZE - 1));
149 goto gotit;
150 }
151 size = 0;
152 sk_for_each(sk2, node, list)
153 if (++size >= best_size_so_far)
154 goto next;
155 best_size_so_far = size;
156 best = result;
157 next:;
158 }
159 result = best;
160 for(i = 0; i < (1 << 16) / UDP_HTABLE_SIZE; i++, result += UDP_HTABLE_SIZE) {
161 if (result > sysctl_local_port_range[1])
162 result = sysctl_local_port_range[0]
163 + ((result - sysctl_local_port_range[0]) &
164 (UDP_HTABLE_SIZE - 1));
165 if (!udp_lport_inuse(result))
166 break;
167 }
168 if (i >= (1 << 16) / UDP_HTABLE_SIZE)
169 goto fail;
170gotit:
171 udp_port_rover = snum = result;
172 } else {
173 sk_for_each(sk2, node,
174 &udp_hash[snum & (UDP_HTABLE_SIZE - 1)]) {
175 struct inet_sock *inet2 = inet_sk(sk2);
176
177 if (inet2->num == snum &&
178 sk2 != sk &&
179 !ipv6_only_sock(sk2) &&
180 (!sk2->sk_bound_dev_if ||
181 !sk->sk_bound_dev_if ||
182 sk2->sk_bound_dev_if == sk->sk_bound_dev_if) &&
183 (!inet2->rcv_saddr ||
184 !inet->rcv_saddr ||
185 inet2->rcv_saddr == inet->rcv_saddr) &&
186 (!sk2->sk_reuse || !sk->sk_reuse))
187 goto fail;
188 }
189 }
190 inet->num = snum;
191 if (sk_unhashed(sk)) {
192 struct hlist_head *h = &udp_hash[snum & (UDP_HTABLE_SIZE - 1)];
193
194 sk_add_node(sk, h);
195 sock_prot_inc_use(sk->sk_prot);
196 }
197 write_unlock_bh(&udp_hash_lock);
198 return 0;
199
200fail:
201 write_unlock_bh(&udp_hash_lock);
202 return 1;
203}
204
205static void udp_v4_hash(struct sock *sk)
206{
207 BUG();
208}
209
210static void udp_v4_unhash(struct sock *sk)
211{
212 write_lock_bh(&udp_hash_lock);
213 if (sk_del_node_init(sk)) {
214 inet_sk(sk)->num = 0;
215 sock_prot_dec_use(sk->sk_prot);
216 }
217 write_unlock_bh(&udp_hash_lock);
218}
219
220/* UDP is nearly always wildcards out the wazoo, it makes no sense to try
221 * harder than this. -DaveM
222 */
223static struct sock *udp_v4_lookup_longway(u32 saddr, u16 sport,
224 u32 daddr, u16 dport, int dif)
225{
226 struct sock *sk, *result = NULL;
227 struct hlist_node *node;
228 unsigned short hnum = ntohs(dport);
229 int badness = -1;
230
231 sk_for_each(sk, node, &udp_hash[hnum & (UDP_HTABLE_SIZE - 1)]) {
232 struct inet_sock *inet = inet_sk(sk);
233
234 if (inet->num == hnum && !ipv6_only_sock(sk)) {
235 int score = (sk->sk_family == PF_INET ? 1 : 0);
236 if (inet->rcv_saddr) {
237 if (inet->rcv_saddr != daddr)
238 continue;
239 score+=2;
240 }
241 if (inet->daddr) {
242 if (inet->daddr != saddr)
243 continue;
244 score+=2;
245 }
246 if (inet->dport) {
247 if (inet->dport != sport)
248 continue;
249 score+=2;
250 }
251 if (sk->sk_bound_dev_if) {
252 if (sk->sk_bound_dev_if != dif)
253 continue;
254 score+=2;
255 }
256 if(score == 9) {
257 result = sk;
258 break;
259 } else if(score > badness) {
260 result = sk;
261 badness = score;
262 }
263 }
264 }
265 return result;
266}
267
268static __inline__ struct sock *udp_v4_lookup(u32 saddr, u16 sport,
269 u32 daddr, u16 dport, int dif)
270{
271 struct sock *sk;
272
273 read_lock(&udp_hash_lock);
274 sk = udp_v4_lookup_longway(saddr, sport, daddr, dport, dif);
275 if (sk)
276 sock_hold(sk);
277 read_unlock(&udp_hash_lock);
278 return sk;
279}
280
281static inline struct sock *udp_v4_mcast_next(struct sock *sk,
282 u16 loc_port, u32 loc_addr,
283 u16 rmt_port, u32 rmt_addr,
284 int dif)
285{
286 struct hlist_node *node;
287 struct sock *s = sk;
288 unsigned short hnum = ntohs(loc_port);
289
290 sk_for_each_from(s, node) {
291 struct inet_sock *inet = inet_sk(s);
292
293 if (inet->num != hnum ||
294 (inet->daddr && inet->daddr != rmt_addr) ||
295 (inet->dport != rmt_port && inet->dport) ||
296 (inet->rcv_saddr && inet->rcv_saddr != loc_addr) ||
297 ipv6_only_sock(s) ||
298 (s->sk_bound_dev_if && s->sk_bound_dev_if != dif))
299 continue;
300 if (!ip_mc_sf_allow(s, loc_addr, rmt_addr, dif))
301 continue;
302 goto found;
303 }
304 s = NULL;
305found:
306 return s;
307}
308
309/*
310 * This routine is called by the ICMP module when it gets some
311 * sort of error condition. If err < 0 then the socket should
312 * be closed and the error returned to the user. If err > 0
313 * it's just the icmp type << 8 | icmp code.
314 * Header points to the ip header of the error packet. We move
315 * on past this. Then (as it used to claim before adjustment)
316 * header points to the first 8 bytes of the udp header. We need
317 * to find the appropriate port.
318 */
319
320void udp_err(struct sk_buff *skb, u32 info)
321{
322 struct inet_sock *inet;
323 struct iphdr *iph = (struct iphdr*)skb->data;
324 struct udphdr *uh = (struct udphdr*)(skb->data+(iph->ihl<<2));
325 int type = skb->h.icmph->type;
326 int code = skb->h.icmph->code;
327 struct sock *sk;
328 int harderr;
329 int err;
330
331 sk = udp_v4_lookup(iph->daddr, uh->dest, iph->saddr, uh->source, skb->dev->ifindex);
332 if (sk == NULL) {
333 ICMP_INC_STATS_BH(ICMP_MIB_INERRORS);
334 return; /* No socket for error */
335 }
336
337 err = 0;
338 harderr = 0;
339 inet = inet_sk(sk);
340
341 switch (type) {
342 default:
343 case ICMP_TIME_EXCEEDED:
344 err = EHOSTUNREACH;
345 break;
346 case ICMP_SOURCE_QUENCH:
347 goto out;
348 case ICMP_PARAMETERPROB:
349 err = EPROTO;
350 harderr = 1;
351 break;
352 case ICMP_DEST_UNREACH:
353 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
354 if (inet->pmtudisc != IP_PMTUDISC_DONT) {
355 err = EMSGSIZE;
356 harderr = 1;
357 break;
358 }
359 goto out;
360 }
361 err = EHOSTUNREACH;
362 if (code <= NR_ICMP_UNREACH) {
363 harderr = icmp_err_convert[code].fatal;
364 err = icmp_err_convert[code].errno;
365 }
366 break;
367 }
368
369 /*
370 * RFC1122: OK. Passes ICMP errors back to application, as per
371 * 4.1.3.3.
372 */
373 if (!inet->recverr) {
374 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
375 goto out;
376 } else {
377 ip_icmp_error(sk, skb, err, uh->dest, info, (u8*)(uh+1));
378 }
379 sk->sk_err = err;
380 sk->sk_error_report(sk);
381out:
382 sock_put(sk);
383}
384
385/*
386 * Throw away all pending data and cancel the corking. Socket is locked.
387 */
388static void udp_flush_pending_frames(struct sock *sk)
389{
390 struct udp_sock *up = udp_sk(sk);
391
392 if (up->pending) {
393 up->len = 0;
394 up->pending = 0;
395 ip_flush_pending_frames(sk);
396 }
397}
398
399/*
400 * Push out all pending data as one UDP datagram. Socket is locked.
401 */
402static int udp_push_pending_frames(struct sock *sk, struct udp_sock *up)
403{
404 struct inet_sock *inet = inet_sk(sk);
405 struct flowi *fl = &inet->cork.fl;
406 struct sk_buff *skb;
407 struct udphdr *uh;
408 int err = 0;
409
410 /* Grab the skbuff where UDP header space exists. */
411 if ((skb = skb_peek(&sk->sk_write_queue)) == NULL)
412 goto out;
413
414 /*
415 * Create a UDP header
416 */
417 uh = skb->h.uh;
418 uh->source = fl->fl_ip_sport;
419 uh->dest = fl->fl_ip_dport;
420 uh->len = htons(up->len);
421 uh->check = 0;
422
423 if (sk->sk_no_check == UDP_CSUM_NOXMIT) {
424 skb->ip_summed = CHECKSUM_NONE;
425 goto send;
426 }
427
428 if (skb_queue_len(&sk->sk_write_queue) == 1) {
429 /*
430 * Only one fragment on the socket.
431 */
84fa7933 432 if (skb->ip_summed == CHECKSUM_PARTIAL) {
1da177e4
LT
433 skb->csum = offsetof(struct udphdr, check);
434 uh->check = ~csum_tcpudp_magic(fl->fl4_src, fl->fl4_dst,
435 up->len, IPPROTO_UDP, 0);
436 } else {
437 skb->csum = csum_partial((char *)uh,
438 sizeof(struct udphdr), skb->csum);
439 uh->check = csum_tcpudp_magic(fl->fl4_src, fl->fl4_dst,
440 up->len, IPPROTO_UDP, skb->csum);
441 if (uh->check == 0)
442 uh->check = -1;
443 }
444 } else {
445 unsigned int csum = 0;
446 /*
447 * HW-checksum won't work as there are two or more
448 * fragments on the socket so that all csums of sk_buffs
449 * should be together.
450 */
84fa7933 451 if (skb->ip_summed == CHECKSUM_PARTIAL) {
1da177e4
LT
452 int offset = (unsigned char *)uh - skb->data;
453 skb->csum = skb_checksum(skb, offset, skb->len - offset, 0);
454
455 skb->ip_summed = CHECKSUM_NONE;
456 } else {
457 skb->csum = csum_partial((char *)uh,
458 sizeof(struct udphdr), skb->csum);
459 }
460
461 skb_queue_walk(&sk->sk_write_queue, skb) {
462 csum = csum_add(csum, skb->csum);
463 }
464 uh->check = csum_tcpudp_magic(fl->fl4_src, fl->fl4_dst,
465 up->len, IPPROTO_UDP, csum);
466 if (uh->check == 0)
467 uh->check = -1;
468 }
469send:
470 err = ip_push_pending_frames(sk);
471out:
472 up->len = 0;
473 up->pending = 0;
474 return err;
475}
476
477
478static unsigned short udp_check(struct udphdr *uh, int len, unsigned long saddr, unsigned long daddr, unsigned long base)
479{
480 return(csum_tcpudp_magic(saddr, daddr, len, IPPROTO_UDP, base));
481}
482
483int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
484 size_t len)
485{
486 struct inet_sock *inet = inet_sk(sk);
487 struct udp_sock *up = udp_sk(sk);
488 int ulen = len;
489 struct ipcm_cookie ipc;
490 struct rtable *rt = NULL;
491 int free = 0;
492 int connected = 0;
493 u32 daddr, faddr, saddr;
494 u16 dport;
495 u8 tos;
496 int err;
497 int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
498
499 if (len > 0xFFFF)
500 return -EMSGSIZE;
501
502 /*
503 * Check the flags.
504 */
505
506 if (msg->msg_flags&MSG_OOB) /* Mirror BSD error message compatibility */
507 return -EOPNOTSUPP;
508
509 ipc.opt = NULL;
510
511 if (up->pending) {
512 /*
513 * There are pending frames.
514 * The socket lock must be held while it's corked.
515 */
516 lock_sock(sk);
517 if (likely(up->pending)) {
518 if (unlikely(up->pending != AF_INET)) {
519 release_sock(sk);
520 return -EINVAL;
521 }
522 goto do_append_data;
523 }
524 release_sock(sk);
525 }
526 ulen += sizeof(struct udphdr);
527
528 /*
529 * Get and verify the address.
530 */
531 if (msg->msg_name) {
532 struct sockaddr_in * usin = (struct sockaddr_in*)msg->msg_name;
533 if (msg->msg_namelen < sizeof(*usin))
534 return -EINVAL;
535 if (usin->sin_family != AF_INET) {
536 if (usin->sin_family != AF_UNSPEC)
537 return -EAFNOSUPPORT;
538 }
539
540 daddr = usin->sin_addr.s_addr;
541 dport = usin->sin_port;
542 if (dport == 0)
543 return -EINVAL;
544 } else {
545 if (sk->sk_state != TCP_ESTABLISHED)
546 return -EDESTADDRREQ;
547 daddr = inet->daddr;
548 dport = inet->dport;
549 /* Open fast path for connected socket.
550 Route will not be used, if at least one option is set.
551 */
552 connected = 1;
553 }
554 ipc.addr = inet->saddr;
555
556 ipc.oif = sk->sk_bound_dev_if;
557 if (msg->msg_controllen) {
558 err = ip_cmsg_send(msg, &ipc);
559 if (err)
560 return err;
561 if (ipc.opt)
562 free = 1;
563 connected = 0;
564 }
565 if (!ipc.opt)
566 ipc.opt = inet->opt;
567
568 saddr = ipc.addr;
569 ipc.addr = faddr = daddr;
570
571 if (ipc.opt && ipc.opt->srr) {
572 if (!daddr)
573 return -EINVAL;
574 faddr = ipc.opt->faddr;
575 connected = 0;
576 }
577 tos = RT_TOS(inet->tos);
578 if (sock_flag(sk, SOCK_LOCALROUTE) ||
579 (msg->msg_flags & MSG_DONTROUTE) ||
580 (ipc.opt && ipc.opt->is_strictroute)) {
581 tos |= RTO_ONLINK;
582 connected = 0;
583 }
584
585 if (MULTICAST(daddr)) {
586 if (!ipc.oif)
587 ipc.oif = inet->mc_index;
588 if (!saddr)
589 saddr = inet->mc_addr;
590 connected = 0;
591 }
592
593 if (connected)
594 rt = (struct rtable*)sk_dst_check(sk, 0);
595
596 if (rt == NULL) {
597 struct flowi fl = { .oif = ipc.oif,
598 .nl_u = { .ip4_u =
599 { .daddr = faddr,
600 .saddr = saddr,
601 .tos = tos } },
602 .proto = IPPROTO_UDP,
603 .uli_u = { .ports =
604 { .sport = inet->sport,
605 .dport = dport } } };
beb8d13b 606 security_sk_classify_flow(sk, &fl);
1da177e4
LT
607 err = ip_route_output_flow(&rt, &fl, sk, !(msg->msg_flags&MSG_DONTWAIT));
608 if (err)
609 goto out;
610
611 err = -EACCES;
612 if ((rt->rt_flags & RTCF_BROADCAST) &&
613 !sock_flag(sk, SOCK_BROADCAST))
614 goto out;
615 if (connected)
616 sk_dst_set(sk, dst_clone(&rt->u.dst));
617 }
618
619 if (msg->msg_flags&MSG_CONFIRM)
620 goto do_confirm;
621back_from_confirm:
622
623 saddr = rt->rt_src;
624 if (!ipc.addr)
625 daddr = ipc.addr = rt->rt_dst;
626
627 lock_sock(sk);
628 if (unlikely(up->pending)) {
629 /* The socket is already corked while preparing it. */
630 /* ... which is an evident application bug. --ANK */
631 release_sock(sk);
632
64ce2073 633 LIMIT_NETDEBUG(KERN_DEBUG "udp cork app bug 2\n");
1da177e4
LT
634 err = -EINVAL;
635 goto out;
636 }
637 /*
638 * Now cork the socket to pend data.
639 */
640 inet->cork.fl.fl4_dst = daddr;
641 inet->cork.fl.fl_ip_dport = dport;
642 inet->cork.fl.fl4_src = saddr;
643 inet->cork.fl.fl_ip_sport = inet->sport;
644 up->pending = AF_INET;
645
646do_append_data:
647 up->len += ulen;
648 err = ip_append_data(sk, ip_generic_getfrag, msg->msg_iov, ulen,
649 sizeof(struct udphdr), &ipc, rt,
650 corkreq ? msg->msg_flags|MSG_MORE : msg->msg_flags);
651 if (err)
652 udp_flush_pending_frames(sk);
653 else if (!corkreq)
654 err = udp_push_pending_frames(sk, up);
655 release_sock(sk);
656
657out:
658 ip_rt_put(rt);
659 if (free)
660 kfree(ipc.opt);
661 if (!err) {
662 UDP_INC_STATS_USER(UDP_MIB_OUTDATAGRAMS);
663 return len;
664 }
81aa646c
MB
665 /*
666 * ENOBUFS = no kernel mem, SOCK_NOSPACE = no sndbuf space. Reporting
667 * ENOBUFS might not be good (it's not tunable per se), but otherwise
668 * we don't have a good statistic (IpOutDiscards but it can be too many
669 * things). We could add another new stat but at least for now that
670 * seems like overkill.
671 */
672 if (err == -ENOBUFS || test_bit(SOCK_NOSPACE, &sk->sk_socket->flags)) {
673 UDP_INC_STATS_USER(UDP_MIB_SNDBUFERRORS);
674 }
1da177e4
LT
675 return err;
676
677do_confirm:
678 dst_confirm(&rt->u.dst);
679 if (!(msg->msg_flags&MSG_PROBE) || len)
680 goto back_from_confirm;
681 err = 0;
682 goto out;
683}
684
685static int udp_sendpage(struct sock *sk, struct page *page, int offset,
686 size_t size, int flags)
687{
688 struct udp_sock *up = udp_sk(sk);
689 int ret;
690
691 if (!up->pending) {
692 struct msghdr msg = { .msg_flags = flags|MSG_MORE };
693
694 /* Call udp_sendmsg to specify destination address which
695 * sendpage interface can't pass.
696 * This will succeed only when the socket is connected.
697 */
698 ret = udp_sendmsg(NULL, sk, &msg, 0);
699 if (ret < 0)
700 return ret;
701 }
702
703 lock_sock(sk);
704
705 if (unlikely(!up->pending)) {
706 release_sock(sk);
707
64ce2073 708 LIMIT_NETDEBUG(KERN_DEBUG "udp cork app bug 3\n");
1da177e4
LT
709 return -EINVAL;
710 }
711
712 ret = ip_append_page(sk, page, offset, size, flags);
713 if (ret == -EOPNOTSUPP) {
714 release_sock(sk);
715 return sock_no_sendpage(sk->sk_socket, page, offset,
716 size, flags);
717 }
718 if (ret < 0) {
719 udp_flush_pending_frames(sk);
720 goto out;
721 }
722
723 up->len += size;
724 if (!(up->corkflag || (flags&MSG_MORE)))
725 ret = udp_push_pending_frames(sk, up);
726 if (!ret)
727 ret = size;
728out:
729 release_sock(sk);
730 return ret;
731}
732
733/*
734 * IOCTL requests applicable to the UDP protocol
735 */
736
737int udp_ioctl(struct sock *sk, int cmd, unsigned long arg)
738{
739 switch(cmd)
740 {
741 case SIOCOUTQ:
742 {
743 int amount = atomic_read(&sk->sk_wmem_alloc);
744 return put_user(amount, (int __user *)arg);
745 }
746
747 case SIOCINQ:
748 {
749 struct sk_buff *skb;
750 unsigned long amount;
751
752 amount = 0;
208d8984 753 spin_lock_bh(&sk->sk_receive_queue.lock);
1da177e4
LT
754 skb = skb_peek(&sk->sk_receive_queue);
755 if (skb != NULL) {
756 /*
757 * We will only return the amount
758 * of this packet since that is all
759 * that will be read.
760 */
761 amount = skb->len - sizeof(struct udphdr);
762 }
208d8984 763 spin_unlock_bh(&sk->sk_receive_queue.lock);
1da177e4
LT
764 return put_user(amount, (int __user *)arg);
765 }
766
767 default:
768 return -ENOIOCTLCMD;
769 }
770 return(0);
771}
772
773static __inline__ int __udp_checksum_complete(struct sk_buff *skb)
774{
fb286bb2 775 return __skb_checksum_complete(skb);
1da177e4
LT
776}
777
778static __inline__ int udp_checksum_complete(struct sk_buff *skb)
779{
780 return skb->ip_summed != CHECKSUM_UNNECESSARY &&
781 __udp_checksum_complete(skb);
782}
783
784/*
785 * This should be easy, if there is something there we
786 * return it, otherwise we block.
787 */
788
789static int udp_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
790 size_t len, int noblock, int flags, int *addr_len)
791{
792 struct inet_sock *inet = inet_sk(sk);
793 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
794 struct sk_buff *skb;
795 int copied, err;
796
797 /*
798 * Check any passed addresses
799 */
800 if (addr_len)
801 *addr_len=sizeof(*sin);
802
803 if (flags & MSG_ERRQUEUE)
804 return ip_recv_error(sk, msg, len);
805
806try_again:
807 skb = skb_recv_datagram(sk, flags, noblock, &err);
808 if (!skb)
809 goto out;
810
811 copied = skb->len - sizeof(struct udphdr);
812 if (copied > len) {
813 copied = len;
814 msg->msg_flags |= MSG_TRUNC;
815 }
816
817 if (skb->ip_summed==CHECKSUM_UNNECESSARY) {
818 err = skb_copy_datagram_iovec(skb, sizeof(struct udphdr), msg->msg_iov,
819 copied);
820 } else if (msg->msg_flags&MSG_TRUNC) {
821 if (__udp_checksum_complete(skb))
822 goto csum_copy_err;
823 err = skb_copy_datagram_iovec(skb, sizeof(struct udphdr), msg->msg_iov,
824 copied);
825 } else {
826 err = skb_copy_and_csum_datagram_iovec(skb, sizeof(struct udphdr), msg->msg_iov);
827
828 if (err == -EINVAL)
829 goto csum_copy_err;
830 }
831
832 if (err)
833 goto out_free;
834
835 sock_recv_timestamp(msg, sk, skb);
836
837 /* Copy the address. */
838 if (sin)
839 {
840 sin->sin_family = AF_INET;
841 sin->sin_port = skb->h.uh->source;
842 sin->sin_addr.s_addr = skb->nh.iph->saddr;
843 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
844 }
845 if (inet->cmsg_flags)
846 ip_cmsg_recv(msg, skb);
847
848 err = copied;
849 if (flags & MSG_TRUNC)
850 err = skb->len - sizeof(struct udphdr);
851
852out_free:
853 skb_free_datagram(sk, skb);
854out:
855 return err;
856
857csum_copy_err:
858 UDP_INC_STATS_BH(UDP_MIB_INERRORS);
859
3305b80c 860 skb_kill_datagram(sk, skb, flags);
1da177e4
LT
861
862 if (noblock)
863 return -EAGAIN;
864 goto try_again;
865}
866
867
868int udp_disconnect(struct sock *sk, int flags)
869{
870 struct inet_sock *inet = inet_sk(sk);
871 /*
872 * 1003.1g - break association.
873 */
874
875 sk->sk_state = TCP_CLOSE;
876 inet->daddr = 0;
877 inet->dport = 0;
878 sk->sk_bound_dev_if = 0;
879 if (!(sk->sk_userlocks & SOCK_BINDADDR_LOCK))
880 inet_reset_saddr(sk);
881
882 if (!(sk->sk_userlocks & SOCK_BINDPORT_LOCK)) {
883 sk->sk_prot->unhash(sk);
884 inet->sport = 0;
885 }
886 sk_dst_reset(sk);
887 return 0;
888}
889
890static void udp_close(struct sock *sk, long timeout)
891{
892 sk_common_release(sk);
893}
894
895/* return:
896 * 1 if the the UDP system should process it
897 * 0 if we should drop this packet
898 * -1 if it should get processed by xfrm4_rcv_encap
899 */
900static int udp_encap_rcv(struct sock * sk, struct sk_buff *skb)
901{
902#ifndef CONFIG_XFRM
903 return 1;
904#else
905 struct udp_sock *up = udp_sk(sk);
906 struct udphdr *uh = skb->h.uh;
907 struct iphdr *iph;
908 int iphlen, len;
909
910 __u8 *udpdata = (__u8 *)uh + sizeof(struct udphdr);
911 __u32 *udpdata32 = (__u32 *)udpdata;
912 __u16 encap_type = up->encap_type;
913
914 /* if we're overly short, let UDP handle it */
915 if (udpdata > skb->tail)
916 return 1;
917
918 /* if this is not encapsulated socket, then just return now */
919 if (!encap_type)
920 return 1;
921
922 len = skb->tail - udpdata;
923
924 switch (encap_type) {
925 default:
926 case UDP_ENCAP_ESPINUDP:
927 /* Check if this is a keepalive packet. If so, eat it. */
928 if (len == 1 && udpdata[0] == 0xff) {
929 return 0;
930 } else if (len > sizeof(struct ip_esp_hdr) && udpdata32[0] != 0 ) {
931 /* ESP Packet without Non-ESP header */
932 len = sizeof(struct udphdr);
933 } else
934 /* Must be an IKE packet.. pass it through */
935 return 1;
936 break;
937 case UDP_ENCAP_ESPINUDP_NON_IKE:
938 /* Check if this is a keepalive packet. If so, eat it. */
939 if (len == 1 && udpdata[0] == 0xff) {
940 return 0;
941 } else if (len > 2 * sizeof(u32) + sizeof(struct ip_esp_hdr) &&
942 udpdata32[0] == 0 && udpdata32[1] == 0) {
943
944 /* ESP Packet with Non-IKE marker */
945 len = sizeof(struct udphdr) + 2 * sizeof(u32);
946 } else
947 /* Must be an IKE packet.. pass it through */
948 return 1;
949 break;
950 }
951
952 /* At this point we are sure that this is an ESPinUDP packet,
953 * so we need to remove 'len' bytes from the packet (the UDP
954 * header and optional ESP marker bytes) and then modify the
955 * protocol to ESP, and then call into the transform receiver.
956 */
4d78b6c7
HX
957 if (skb_cloned(skb) && pskb_expand_head(skb, 0, 0, GFP_ATOMIC))
958 return 0;
1da177e4
LT
959
960 /* Now we can update and verify the packet length... */
961 iph = skb->nh.iph;
962 iphlen = iph->ihl << 2;
963 iph->tot_len = htons(ntohs(iph->tot_len) - len);
964 if (skb->len < iphlen + len) {
965 /* packet is too small!?! */
966 return 0;
967 }
968
969 /* pull the data buffer up to the ESP header and set the
970 * transport header to point to ESP. Keep UDP on the stack
971 * for later.
972 */
973 skb->h.raw = skb_pull(skb, len);
974
975 /* modify the protocol (it's ESP!) */
976 iph->protocol = IPPROTO_ESP;
977
978 /* and let the caller know to send this into the ESP processor... */
979 return -1;
980#endif
981}
982
983/* returns:
984 * -1: error
985 * 0: success
986 * >0: "udp encap" protocol resubmission
987 *
988 * Note that in the success and error cases, the skb is assumed to
989 * have either been requeued or freed.
990 */
991static int udp_queue_rcv_skb(struct sock * sk, struct sk_buff *skb)
992{
993 struct udp_sock *up = udp_sk(sk);
81aa646c 994 int rc;
1da177e4
LT
995
996 /*
997 * Charge it to the socket, dropping if the queue is full.
998 */
999 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
1000 kfree_skb(skb);
1001 return -1;
1002 }
b59c2701 1003 nf_reset(skb);
1da177e4
LT
1004
1005 if (up->encap_type) {
1006 /*
1007 * This is an encapsulation socket, so let's see if this is
1008 * an encapsulated packet.
1009 * If it's a keepalive packet, then just eat it.
1010 * If it's an encapsulateed packet, then pass it to the
1011 * IPsec xfrm input and return the response
1012 * appropriately. Otherwise, just fall through and
1013 * pass this up the UDP socket.
1014 */
1015 int ret;
1016
1017 ret = udp_encap_rcv(sk, skb);
1018 if (ret == 0) {
1019 /* Eat the packet .. */
1020 kfree_skb(skb);
1021 return 0;
1022 }
1023 if (ret < 0) {
1024 /* process the ESP packet */
1025 ret = xfrm4_rcv_encap(skb, up->encap_type);
1026 UDP_INC_STATS_BH(UDP_MIB_INDATAGRAMS);
1027 return -ret;
1028 }
1029 /* FALLTHROUGH -- it's a UDP Packet */
1030 }
1031
1032 if (sk->sk_filter && skb->ip_summed != CHECKSUM_UNNECESSARY) {
1033 if (__udp_checksum_complete(skb)) {
1034 UDP_INC_STATS_BH(UDP_MIB_INERRORS);
1035 kfree_skb(skb);
1036 return -1;
1037 }
1038 skb->ip_summed = CHECKSUM_UNNECESSARY;
1039 }
1040
81aa646c
MB
1041 if ((rc = sock_queue_rcv_skb(sk,skb)) < 0) {
1042 /* Note that an ENOMEM error is charged twice */
1043 if (rc == -ENOMEM)
1044 UDP_INC_STATS_BH(UDP_MIB_RCVBUFERRORS);
1da177e4
LT
1045 UDP_INC_STATS_BH(UDP_MIB_INERRORS);
1046 kfree_skb(skb);
1047 return -1;
1048 }
1049 UDP_INC_STATS_BH(UDP_MIB_INDATAGRAMS);
1050 return 0;
1051}
1052
1053/*
1054 * Multicasts and broadcasts go to each listener.
1055 *
1056 * Note: called only from the BH handler context,
1057 * so we don't need to lock the hashes.
1058 */
1059static int udp_v4_mcast_deliver(struct sk_buff *skb, struct udphdr *uh,
1060 u32 saddr, u32 daddr)
1061{
1062 struct sock *sk;
1063 int dif;
1064
1065 read_lock(&udp_hash_lock);
1066 sk = sk_head(&udp_hash[ntohs(uh->dest) & (UDP_HTABLE_SIZE - 1)]);
1067 dif = skb->dev->ifindex;
1068 sk = udp_v4_mcast_next(sk, uh->dest, daddr, uh->source, saddr, dif);
1069 if (sk) {
1070 struct sock *sknext = NULL;
1071
1072 do {
1073 struct sk_buff *skb1 = skb;
1074
1075 sknext = udp_v4_mcast_next(sk_next(sk), uh->dest, daddr,
1076 uh->source, saddr, dif);
1077 if(sknext)
1078 skb1 = skb_clone(skb, GFP_ATOMIC);
1079
1080 if(skb1) {
1081 int ret = udp_queue_rcv_skb(sk, skb1);
1082 if (ret > 0)
1083 /* we should probably re-process instead
1084 * of dropping packets here. */
1085 kfree_skb(skb1);
1086 }
1087 sk = sknext;
1088 } while(sknext);
1089 } else
1090 kfree_skb(skb);
1091 read_unlock(&udp_hash_lock);
1092 return 0;
1093}
1094
1095/* Initialize UDP checksum. If exited with zero value (success),
1096 * CHECKSUM_UNNECESSARY means, that no more checks are required.
1097 * Otherwise, csum completion requires chacksumming packet body,
1098 * including udp header and folding it to skb->csum.
1099 */
65a45441 1100static void udp_checksum_init(struct sk_buff *skb, struct udphdr *uh,
1da177e4
LT
1101 unsigned short ulen, u32 saddr, u32 daddr)
1102{
1103 if (uh->check == 0) {
1104 skb->ip_summed = CHECKSUM_UNNECESSARY;
84fa7933 1105 } else if (skb->ip_summed == CHECKSUM_COMPLETE) {
1da177e4 1106 if (!udp_check(uh, ulen, saddr, daddr, skb->csum))
fb286bb2 1107 skb->ip_summed = CHECKSUM_UNNECESSARY;
1da177e4
LT
1108 }
1109 if (skb->ip_summed != CHECKSUM_UNNECESSARY)
1110 skb->csum = csum_tcpudp_nofold(saddr, daddr, ulen, IPPROTO_UDP, 0);
1111 /* Probably, we should checksum udp header (it should be in cache
1112 * in any case) and data in tiny packets (< rx copybreak).
1113 */
1da177e4
LT
1114}
1115
1116/*
1117 * All we need to do is get the socket, and then do a checksum.
1118 */
1119
1120int udp_rcv(struct sk_buff *skb)
1121{
1122 struct sock *sk;
1123 struct udphdr *uh;
1124 unsigned short ulen;
1125 struct rtable *rt = (struct rtable*)skb->dst;
1126 u32 saddr = skb->nh.iph->saddr;
1127 u32 daddr = skb->nh.iph->daddr;
1128 int len = skb->len;
1129
1130 /*
1131 * Validate the packet and the UDP length.
1132 */
1133 if (!pskb_may_pull(skb, sizeof(struct udphdr)))
1134 goto no_header;
1135
1136 uh = skb->h.uh;
1137
1138 ulen = ntohs(uh->len);
1139
1140 if (ulen > len || ulen < sizeof(*uh))
1141 goto short_packet;
1142
e308e25c 1143 if (pskb_trim_rcsum(skb, ulen))
1da177e4
LT
1144 goto short_packet;
1145
65a45441 1146 udp_checksum_init(skb, uh, ulen, saddr, daddr);
1da177e4
LT
1147
1148 if(rt->rt_flags & (RTCF_BROADCAST|RTCF_MULTICAST))
1149 return udp_v4_mcast_deliver(skb, uh, saddr, daddr);
1150
1151 sk = udp_v4_lookup(saddr, uh->source, daddr, uh->dest, skb->dev->ifindex);
1152
1153 if (sk != NULL) {
1154 int ret = udp_queue_rcv_skb(sk, skb);
1155 sock_put(sk);
1156
1157 /* a return value > 0 means to resubmit the input, but
1158 * it it wants the return to be -protocol, or 0
1159 */
1160 if (ret > 0)
1161 return -ret;
1162 return 0;
1163 }
1164
1165 if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
1166 goto drop;
b59c2701 1167 nf_reset(skb);
1da177e4
LT
1168
1169 /* No socket. Drop packet silently, if checksum is wrong */
1170 if (udp_checksum_complete(skb))
1171 goto csum_error;
1172
1173 UDP_INC_STATS_BH(UDP_MIB_NOPORTS);
1174 icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PORT_UNREACH, 0);
1175
1176 /*
1177 * Hmm. We got an UDP packet to a port to which we
1178 * don't wanna listen. Ignore it.
1179 */
1180 kfree_skb(skb);
1181 return(0);
1182
1183short_packet:
64ce2073
PM
1184 LIMIT_NETDEBUG(KERN_DEBUG "UDP: short packet: From %u.%u.%u.%u:%u %d/%d to %u.%u.%u.%u:%u\n",
1185 NIPQUAD(saddr),
1186 ntohs(uh->source),
1187 ulen,
1188 len,
1189 NIPQUAD(daddr),
1190 ntohs(uh->dest));
1da177e4
LT
1191no_header:
1192 UDP_INC_STATS_BH(UDP_MIB_INERRORS);
1193 kfree_skb(skb);
1194 return(0);
1195
1196csum_error:
1197 /*
1198 * RFC1122: OK. Discards the bad packet silently (as far as
1199 * the network is concerned, anyway) as per 4.1.3.4 (MUST).
1200 */
64ce2073
PM
1201 LIMIT_NETDEBUG(KERN_DEBUG "UDP: bad checksum. From %d.%d.%d.%d:%d to %d.%d.%d.%d:%d ulen %d\n",
1202 NIPQUAD(saddr),
1203 ntohs(uh->source),
1204 NIPQUAD(daddr),
1205 ntohs(uh->dest),
1206 ulen);
1da177e4
LT
1207drop:
1208 UDP_INC_STATS_BH(UDP_MIB_INERRORS);
1209 kfree_skb(skb);
1210 return(0);
1211}
1212
1213static int udp_destroy_sock(struct sock *sk)
1214{
1215 lock_sock(sk);
1216 udp_flush_pending_frames(sk);
1217 release_sock(sk);
1218 return 0;
1219}
1220
1221/*
1222 * Socket option code for UDP
1223 */
3fdadf7d 1224static int do_udp_setsockopt(struct sock *sk, int level, int optname,
1da177e4
LT
1225 char __user *optval, int optlen)
1226{
1227 struct udp_sock *up = udp_sk(sk);
1228 int val;
1229 int err = 0;
1230
1da177e4
LT
1231 if(optlen<sizeof(int))
1232 return -EINVAL;
1233
1234 if (get_user(val, (int __user *)optval))
1235 return -EFAULT;
1236
1237 switch(optname) {
1238 case UDP_CORK:
1239 if (val != 0) {
1240 up->corkflag = 1;
1241 } else {
1242 up->corkflag = 0;
1243 lock_sock(sk);
1244 udp_push_pending_frames(sk, up);
1245 release_sock(sk);
1246 }
1247 break;
1248
1249 case UDP_ENCAP:
1250 switch (val) {
1251 case 0:
1252 case UDP_ENCAP_ESPINUDP:
1253 case UDP_ENCAP_ESPINUDP_NON_IKE:
1254 up->encap_type = val;
1255 break;
1256 default:
1257 err = -ENOPROTOOPT;
1258 break;
1259 }
1260 break;
1261
1262 default:
1263 err = -ENOPROTOOPT;
1264 break;
1265 };
1266
1267 return err;
1268}
1269
3fdadf7d
DM
1270static int udp_setsockopt(struct sock *sk, int level, int optname,
1271 char __user *optval, int optlen)
1272{
1273 if (level != SOL_UDP)
1274 return ip_setsockopt(sk, level, optname, optval, optlen);
1275 return do_udp_setsockopt(sk, level, optname, optval, optlen);
1276}
1277
1278#ifdef CONFIG_COMPAT
1279static int compat_udp_setsockopt(struct sock *sk, int level, int optname,
543d9cfe 1280 char __user *optval, int optlen)
3fdadf7d
DM
1281{
1282 if (level != SOL_UDP)
543d9cfe 1283 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
3fdadf7d
DM
1284 return do_udp_setsockopt(sk, level, optname, optval, optlen);
1285}
1286#endif
1287
1288static int do_udp_getsockopt(struct sock *sk, int level, int optname,
1da177e4
LT
1289 char __user *optval, int __user *optlen)
1290{
1291 struct udp_sock *up = udp_sk(sk);
1292 int val, len;
1293
1da177e4
LT
1294 if(get_user(len,optlen))
1295 return -EFAULT;
1296
1297 len = min_t(unsigned int, len, sizeof(int));
1298
1299 if(len < 0)
1300 return -EINVAL;
1301
1302 switch(optname) {
1303 case UDP_CORK:
1304 val = up->corkflag;
1305 break;
1306
1307 case UDP_ENCAP:
1308 val = up->encap_type;
1309 break;
1310
1311 default:
1312 return -ENOPROTOOPT;
1313 };
1314
1315 if(put_user(len, optlen))
1316 return -EFAULT;
1317 if(copy_to_user(optval, &val,len))
1318 return -EFAULT;
1319 return 0;
1320}
1321
3fdadf7d
DM
1322static int udp_getsockopt(struct sock *sk, int level, int optname,
1323 char __user *optval, int __user *optlen)
1324{
1325 if (level != SOL_UDP)
1326 return ip_getsockopt(sk, level, optname, optval, optlen);
1327 return do_udp_getsockopt(sk, level, optname, optval, optlen);
1328}
1329
1330#ifdef CONFIG_COMPAT
1331static int compat_udp_getsockopt(struct sock *sk, int level, int optname,
543d9cfe 1332 char __user *optval, int __user *optlen)
3fdadf7d
DM
1333{
1334 if (level != SOL_UDP)
543d9cfe 1335 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
3fdadf7d
DM
1336 return do_udp_getsockopt(sk, level, optname, optval, optlen);
1337}
1338#endif
1da177e4
LT
1339/**
1340 * udp_poll - wait for a UDP event.
1341 * @file - file struct
1342 * @sock - socket
1343 * @wait - poll table
1344 *
1345 * This is same as datagram poll, except for the special case of
1346 * blocking sockets. If application is using a blocking fd
1347 * and a packet with checksum error is in the queue;
1348 * then it could get return from select indicating data available
1349 * but then block when reading it. Add special case code
1350 * to work around these arguably broken applications.
1351 */
1352unsigned int udp_poll(struct file *file, struct socket *sock, poll_table *wait)
1353{
1354 unsigned int mask = datagram_poll(file, sock, wait);
1355 struct sock *sk = sock->sk;
1356
1357 /* Check for false positives due to checksum errors */
1358 if ( (mask & POLLRDNORM) &&
1359 !(file->f_flags & O_NONBLOCK) &&
1360 !(sk->sk_shutdown & RCV_SHUTDOWN)){
1361 struct sk_buff_head *rcvq = &sk->sk_receive_queue;
1362 struct sk_buff *skb;
1363
208d8984 1364 spin_lock_bh(&rcvq->lock);
1da177e4
LT
1365 while ((skb = skb_peek(rcvq)) != NULL) {
1366 if (udp_checksum_complete(skb)) {
1367 UDP_INC_STATS_BH(UDP_MIB_INERRORS);
1368 __skb_unlink(skb, rcvq);
1369 kfree_skb(skb);
1370 } else {
1371 skb->ip_summed = CHECKSUM_UNNECESSARY;
1372 break;
1373 }
1374 }
208d8984 1375 spin_unlock_bh(&rcvq->lock);
1da177e4
LT
1376
1377 /* nothing to see, move along */
1378 if (skb == NULL)
1379 mask &= ~(POLLIN | POLLRDNORM);
1380 }
1381
1382 return mask;
1383
1384}
1385
1386struct proto udp_prot = {
543d9cfe
ACM
1387 .name = "UDP",
1388 .owner = THIS_MODULE,
1389 .close = udp_close,
1390 .connect = ip4_datagram_connect,
1391 .disconnect = udp_disconnect,
1392 .ioctl = udp_ioctl,
1393 .destroy = udp_destroy_sock,
1394 .setsockopt = udp_setsockopt,
1395 .getsockopt = udp_getsockopt,
1396 .sendmsg = udp_sendmsg,
1397 .recvmsg = udp_recvmsg,
1398 .sendpage = udp_sendpage,
1399 .backlog_rcv = udp_queue_rcv_skb,
1400 .hash = udp_v4_hash,
1401 .unhash = udp_v4_unhash,
1402 .get_port = udp_v4_get_port,
1403 .obj_size = sizeof(struct udp_sock),
3fdadf7d 1404#ifdef CONFIG_COMPAT
543d9cfe
ACM
1405 .compat_setsockopt = compat_udp_setsockopt,
1406 .compat_getsockopt = compat_udp_getsockopt,
3fdadf7d 1407#endif
1da177e4
LT
1408};
1409
1410/* ------------------------------------------------------------------------ */
1411#ifdef CONFIG_PROC_FS
1412
1413static struct sock *udp_get_first(struct seq_file *seq)
1414{
1415 struct sock *sk;
1416 struct udp_iter_state *state = seq->private;
1417
1418 for (state->bucket = 0; state->bucket < UDP_HTABLE_SIZE; ++state->bucket) {
1419 struct hlist_node *node;
1420 sk_for_each(sk, node, &udp_hash[state->bucket]) {
1421 if (sk->sk_family == state->family)
1422 goto found;
1423 }
1424 }
1425 sk = NULL;
1426found:
1427 return sk;
1428}
1429
1430static struct sock *udp_get_next(struct seq_file *seq, struct sock *sk)
1431{
1432 struct udp_iter_state *state = seq->private;
1433
1434 do {
1435 sk = sk_next(sk);
1436try_again:
1437 ;
1438 } while (sk && sk->sk_family != state->family);
1439
1440 if (!sk && ++state->bucket < UDP_HTABLE_SIZE) {
1441 sk = sk_head(&udp_hash[state->bucket]);
1442 goto try_again;
1443 }
1444 return sk;
1445}
1446
1447static struct sock *udp_get_idx(struct seq_file *seq, loff_t pos)
1448{
1449 struct sock *sk = udp_get_first(seq);
1450
1451 if (sk)
1452 while(pos && (sk = udp_get_next(seq, sk)) != NULL)
1453 --pos;
1454 return pos ? NULL : sk;
1455}
1456
1457static void *udp_seq_start(struct seq_file *seq, loff_t *pos)
1458{
1459 read_lock(&udp_hash_lock);
1460 return *pos ? udp_get_idx(seq, *pos-1) : (void *)1;
1461}
1462
1463static void *udp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1464{
1465 struct sock *sk;
1466
1467 if (v == (void *)1)
1468 sk = udp_get_idx(seq, 0);
1469 else
1470 sk = udp_get_next(seq, v);
1471
1472 ++*pos;
1473 return sk;
1474}
1475
1476static void udp_seq_stop(struct seq_file *seq, void *v)
1477{
1478 read_unlock(&udp_hash_lock);
1479}
1480
1481static int udp_seq_open(struct inode *inode, struct file *file)
1482{
1483 struct udp_seq_afinfo *afinfo = PDE(inode)->data;
1484 struct seq_file *seq;
1485 int rc = -ENOMEM;
0da974f4 1486 struct udp_iter_state *s = kzalloc(sizeof(*s), GFP_KERNEL);
1da177e4
LT
1487
1488 if (!s)
1489 goto out;
1da177e4
LT
1490 s->family = afinfo->family;
1491 s->seq_ops.start = udp_seq_start;
1492 s->seq_ops.next = udp_seq_next;
1493 s->seq_ops.show = afinfo->seq_show;
1494 s->seq_ops.stop = udp_seq_stop;
1495
1496 rc = seq_open(file, &s->seq_ops);
1497 if (rc)
1498 goto out_kfree;
1499
1500 seq = file->private_data;
1501 seq->private = s;
1502out:
1503 return rc;
1504out_kfree:
1505 kfree(s);
1506 goto out;
1507}
1508
1509/* ------------------------------------------------------------------------ */
1510int udp_proc_register(struct udp_seq_afinfo *afinfo)
1511{
1512 struct proc_dir_entry *p;
1513 int rc = 0;
1514
1515 if (!afinfo)
1516 return -EINVAL;
1517 afinfo->seq_fops->owner = afinfo->owner;
1518 afinfo->seq_fops->open = udp_seq_open;
1519 afinfo->seq_fops->read = seq_read;
1520 afinfo->seq_fops->llseek = seq_lseek;
1521 afinfo->seq_fops->release = seq_release_private;
1522
1523 p = proc_net_fops_create(afinfo->name, S_IRUGO, afinfo->seq_fops);
1524 if (p)
1525 p->data = afinfo;
1526 else
1527 rc = -ENOMEM;
1528 return rc;
1529}
1530
1531void udp_proc_unregister(struct udp_seq_afinfo *afinfo)
1532{
1533 if (!afinfo)
1534 return;
1535 proc_net_remove(afinfo->name);
1536 memset(afinfo->seq_fops, 0, sizeof(*afinfo->seq_fops));
1537}
1538
1539/* ------------------------------------------------------------------------ */
1540static void udp4_format_sock(struct sock *sp, char *tmpbuf, int bucket)
1541{
1542 struct inet_sock *inet = inet_sk(sp);
1543 unsigned int dest = inet->daddr;
1544 unsigned int src = inet->rcv_saddr;
1545 __u16 destp = ntohs(inet->dport);
1546 __u16 srcp = ntohs(inet->sport);
1547
1548 sprintf(tmpbuf, "%4d: %08X:%04X %08X:%04X"
1549 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %p",
1550 bucket, src, srcp, dest, destp, sp->sk_state,
1551 atomic_read(&sp->sk_wmem_alloc),
1552 atomic_read(&sp->sk_rmem_alloc),
1553 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
1554 atomic_read(&sp->sk_refcnt), sp);
1555}
1556
1557static int udp4_seq_show(struct seq_file *seq, void *v)
1558{
1559 if (v == SEQ_START_TOKEN)
1560 seq_printf(seq, "%-127s\n",
1561 " sl local_address rem_address st tx_queue "
1562 "rx_queue tr tm->when retrnsmt uid timeout "
1563 "inode");
1564 else {
1565 char tmpbuf[129];
1566 struct udp_iter_state *state = seq->private;
1567
1568 udp4_format_sock(v, tmpbuf, state->bucket);
1569 seq_printf(seq, "%-127s\n", tmpbuf);
1570 }
1571 return 0;
1572}
1573
1574/* ------------------------------------------------------------------------ */
1575static struct file_operations udp4_seq_fops;
1576static struct udp_seq_afinfo udp4_seq_afinfo = {
1577 .owner = THIS_MODULE,
1578 .name = "udp",
1579 .family = AF_INET,
1580 .seq_show = udp4_seq_show,
1581 .seq_fops = &udp4_seq_fops,
1582};
1583
1584int __init udp4_proc_init(void)
1585{
1586 return udp_proc_register(&udp4_seq_afinfo);
1587}
1588
1589void udp4_proc_exit(void)
1590{
1591 udp_proc_unregister(&udp4_seq_afinfo);
1592}
1593#endif /* CONFIG_PROC_FS */
1594
1595EXPORT_SYMBOL(udp_disconnect);
1596EXPORT_SYMBOL(udp_hash);
1597EXPORT_SYMBOL(udp_hash_lock);
1598EXPORT_SYMBOL(udp_ioctl);
1599EXPORT_SYMBOL(udp_port_rover);
1600EXPORT_SYMBOL(udp_prot);
1601EXPORT_SYMBOL(udp_sendmsg);
1602EXPORT_SYMBOL(udp_poll);
1603
1604#ifdef CONFIG_PROC_FS
1605EXPORT_SYMBOL(udp_proc_register);
1606EXPORT_SYMBOL(udp_proc_unregister);
1607#endif