blob: 37a4e777e3476e0d59060e2694e374ac7408dcc1 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * ip6_flowlabel.c IPv6 flowlabel manager.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License
6 * as published by the Free Software Foundation; either version
7 * 2 of the License, or (at your option) any later version.
8 *
9 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
10 */
11
Randy Dunlap4fc268d2006-01-11 12:17:47 -080012#include <linux/capability.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070013#include <linux/errno.h>
14#include <linux/types.h>
15#include <linux/socket.h>
16#include <linux/net.h>
17#include <linux/netdevice.h>
18#include <linux/if_arp.h>
19#include <linux/in6.h>
20#include <linux/route.h>
21#include <linux/proc_fs.h>
22#include <linux/seq_file.h>
23
Eric W. Biederman457c4cb2007-09-12 12:01:34 +020024#include <net/net_namespace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070025#include <net/sock.h>
26
27#include <net/ipv6.h>
28#include <net/ndisc.h>
29#include <net/protocol.h>
30#include <net/ip6_route.h>
31#include <net/addrconf.h>
32#include <net/rawv6.h>
33#include <net/icmp.h>
34#include <net/transp_v6.h>
35
36#include <asm/uaccess.h>
37
38#define FL_MIN_LINGER 6 /* Minimal linger. It is set to 6sec specified
39 in old IPv6 RFC. Well, it was reasonable value.
40 */
41#define FL_MAX_LINGER 60 /* Maximal linger timeout */
42
43/* FL hash table */
44
45#define FL_MAX_PER_SOCK 32
46#define FL_MAX_SIZE 4096
47#define FL_HASH_MASK 255
48#define FL_HASH(l) (ntohl(l)&FL_HASH_MASK)
49
50static atomic_t fl_size = ATOMIC_INIT(0);
51static struct ip6_flowlabel *fl_ht[FL_HASH_MASK+1];
52
53static void ip6_fl_gc(unsigned long dummy);
Ingo Molnar8d06afa2005-09-09 13:10:40 -070054static DEFINE_TIMER(ip6_fl_gc_timer, ip6_fl_gc, 0, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -070055
56/* FL hash table lock: it protects only of GC */
57
58static DEFINE_RWLOCK(ip6_fl_lock);
59
60/* Big socket sock */
61
62static DEFINE_RWLOCK(ip6_sk_fl_lock);
63
64
Benjamin Thery60e8fbc2008-03-26 16:53:08 -070065static inline struct ip6_flowlabel *__fl_lookup(struct net *net, __be32 label)
Linus Torvalds1da177e2005-04-16 15:20:36 -070066{
67 struct ip6_flowlabel *fl;
68
69 for (fl=fl_ht[FL_HASH(label)]; fl; fl = fl->next) {
Benjamin Thery60e8fbc2008-03-26 16:53:08 -070070 if (fl->label == label && fl->fl_net == net)
Linus Torvalds1da177e2005-04-16 15:20:36 -070071 return fl;
72 }
73 return NULL;
74}
75
Benjamin Thery60e8fbc2008-03-26 16:53:08 -070076static struct ip6_flowlabel *fl_lookup(struct net *net, __be32 label)
Linus Torvalds1da177e2005-04-16 15:20:36 -070077{
78 struct ip6_flowlabel *fl;
79
80 read_lock_bh(&ip6_fl_lock);
Benjamin Thery60e8fbc2008-03-26 16:53:08 -070081 fl = __fl_lookup(net, label);
Linus Torvalds1da177e2005-04-16 15:20:36 -070082 if (fl)
83 atomic_inc(&fl->users);
84 read_unlock_bh(&ip6_fl_lock);
85 return fl;
86}
87
88
89static void fl_free(struct ip6_flowlabel *fl)
90{
Benjamin Thery60e8fbc2008-03-26 16:53:08 -070091 if (fl) {
92 release_net(fl->fl_net);
Linus Torvalds1da177e2005-04-16 15:20:36 -070093 kfree(fl->opt);
Benjamin Thery60e8fbc2008-03-26 16:53:08 -070094 }
Linus Torvalds1da177e2005-04-16 15:20:36 -070095 kfree(fl);
96}
97
98static void fl_release(struct ip6_flowlabel *fl)
99{
100 write_lock_bh(&ip6_fl_lock);
101
102 fl->lastuse = jiffies;
103 if (atomic_dec_and_test(&fl->users)) {
104 unsigned long ttd = fl->lastuse + fl->linger;
105 if (time_after(ttd, fl->expires))
106 fl->expires = ttd;
107 ttd = fl->expires;
108 if (fl->opt && fl->share == IPV6_FL_S_EXCL) {
109 struct ipv6_txoptions *opt = fl->opt;
110 fl->opt = NULL;
111 kfree(opt);
112 }
113 if (!timer_pending(&ip6_fl_gc_timer) ||
114 time_after(ip6_fl_gc_timer.expires, ttd))
115 mod_timer(&ip6_fl_gc_timer, ttd);
116 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700117 write_unlock_bh(&ip6_fl_lock);
118}
119
120static void ip6_fl_gc(unsigned long dummy)
121{
122 int i;
123 unsigned long now = jiffies;
124 unsigned long sched = 0;
125
126 write_lock(&ip6_fl_lock);
127
128 for (i=0; i<=FL_HASH_MASK; i++) {
129 struct ip6_flowlabel *fl, **flp;
130 flp = &fl_ht[i];
131 while ((fl=*flp) != NULL) {
132 if (atomic_read(&fl->users) == 0) {
133 unsigned long ttd = fl->lastuse + fl->linger;
134 if (time_after(ttd, fl->expires))
135 fl->expires = ttd;
136 ttd = fl->expires;
137 if (time_after_eq(now, ttd)) {
138 *flp = fl->next;
139 fl_free(fl);
140 atomic_dec(&fl_size);
141 continue;
142 }
143 if (!sched || time_before(ttd, sched))
144 sched = ttd;
145 }
146 flp = &fl->next;
147 }
148 }
149 if (!sched && atomic_read(&fl_size))
150 sched = now + FL_MAX_LINGER;
151 if (sched) {
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700152 mod_timer(&ip6_fl_gc_timer, sched);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700153 }
154 write_unlock(&ip6_fl_lock);
155}
156
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700157static void ip6_fl_purge(struct net *net)
158{
159 int i;
160
161 write_lock(&ip6_fl_lock);
162 for (i = 0; i <= FL_HASH_MASK; i++) {
163 struct ip6_flowlabel *fl, **flp;
164 flp = &fl_ht[i];
165 while ((fl = *flp) != NULL) {
166 if (fl->fl_net == net && atomic_read(&fl->users) == 0) {
167 *flp = fl->next;
168 fl_free(fl);
169 atomic_dec(&fl_size);
170 continue;
171 }
172 flp = &fl->next;
173 }
174 }
175 write_unlock(&ip6_fl_lock);
176}
177
178static struct ip6_flowlabel *fl_intern(struct net *net,
179 struct ip6_flowlabel *fl, __be32 label)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700180{
Pavel Emelyanov78c2e502007-10-18 05:18:56 -0700181 struct ip6_flowlabel *lfl;
182
Linus Torvalds1da177e2005-04-16 15:20:36 -0700183 fl->label = label & IPV6_FLOWLABEL_MASK;
184
185 write_lock_bh(&ip6_fl_lock);
186 if (label == 0) {
187 for (;;) {
188 fl->label = htonl(net_random())&IPV6_FLOWLABEL_MASK;
189 if (fl->label) {
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700190 lfl = __fl_lookup(net, fl->label);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700191 if (lfl == NULL)
192 break;
193 }
194 }
Pavel Emelyanov78c2e502007-10-18 05:18:56 -0700195 } else {
196 /*
197 * we dropper the ip6_fl_lock, so this entry could reappear
198 * and we need to recheck with it.
199 *
200 * OTOH no need to search the active socket first, like it is
201 * done in ipv6_flowlabel_opt - sock is locked, so new entry
202 * with the same label can only appear on another sock
203 */
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700204 lfl = __fl_lookup(net, fl->label);
Pavel Emelyanov78c2e502007-10-18 05:18:56 -0700205 if (lfl != NULL) {
206 atomic_inc(&lfl->users);
207 write_unlock_bh(&ip6_fl_lock);
208 return lfl;
209 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700210 }
211
212 fl->lastuse = jiffies;
213 fl->next = fl_ht[FL_HASH(fl->label)];
214 fl_ht[FL_HASH(fl->label)] = fl;
215 atomic_inc(&fl_size);
216 write_unlock_bh(&ip6_fl_lock);
Pavel Emelyanov78c2e502007-10-18 05:18:56 -0700217 return NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700218}
219
220
221
222/* Socket flowlabel lists */
223
Al Viro90bcaf72006-11-08 00:25:17 -0800224struct ip6_flowlabel * fl6_sock_lookup(struct sock *sk, __be32 label)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225{
226 struct ipv6_fl_socklist *sfl;
227 struct ipv6_pinfo *np = inet6_sk(sk);
228
229 label &= IPV6_FLOWLABEL_MASK;
230
Pavel Emelyanovbd0bf572007-10-18 05:15:57 -0700231 read_lock_bh(&ip6_sk_fl_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232 for (sfl=np->ipv6_fl_list; sfl; sfl = sfl->next) {
233 struct ip6_flowlabel *fl = sfl->fl;
234 if (fl->label == label) {
235 fl->lastuse = jiffies;
236 atomic_inc(&fl->users);
Pavel Emelyanov52f095e2007-10-18 05:38:48 -0700237 read_unlock_bh(&ip6_sk_fl_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700238 return fl;
239 }
240 }
Pavel Emelyanovbd0bf572007-10-18 05:15:57 -0700241 read_unlock_bh(&ip6_sk_fl_lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700242 return NULL;
243}
244
Arnaldo Carvalho de Melo3cf3dc62005-12-13 23:23:20 -0800245EXPORT_SYMBOL_GPL(fl6_sock_lookup);
246
Linus Torvalds1da177e2005-04-16 15:20:36 -0700247void fl6_free_socklist(struct sock *sk)
248{
249 struct ipv6_pinfo *np = inet6_sk(sk);
250 struct ipv6_fl_socklist *sfl;
251
252 while ((sfl = np->ipv6_fl_list) != NULL) {
253 np->ipv6_fl_list = sfl->next;
254 fl_release(sfl->fl);
255 kfree(sfl);
256 }
257}
258
259/* Service routines */
260
261
262/*
263 It is the only difficult place. flowlabel enforces equal headers
264 before and including routing header, however user may supply options
265 following rthdr.
266 */
267
268struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions * opt_space,
269 struct ip6_flowlabel * fl,
270 struct ipv6_txoptions * fopt)
271{
YOSHIFUJI Hideakidf9890c2005-11-20 12:23:18 +0900272 struct ipv6_txoptions * fl_opt = fl->opt;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900273
YOSHIFUJI Hideakidf9890c2005-11-20 12:23:18 +0900274 if (fopt == NULL || fopt->opt_flen == 0)
275 return fl_opt;
YOSHIFUJI Hideaki1ab14572007-02-09 23:24:49 +0900276
Linus Torvalds1da177e2005-04-16 15:20:36 -0700277 if (fl_opt != NULL) {
278 opt_space->hopopt = fl_opt->hopopt;
YOSHIFUJI Hideakidf9890c2005-11-20 12:23:18 +0900279 opt_space->dst0opt = fl_opt->dst0opt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280 opt_space->srcrt = fl_opt->srcrt;
281 opt_space->opt_nflen = fl_opt->opt_nflen;
282 } else {
283 if (fopt->opt_nflen == 0)
284 return fopt;
285 opt_space->hopopt = NULL;
286 opt_space->dst0opt = NULL;
287 opt_space->srcrt = NULL;
288 opt_space->opt_nflen = 0;
289 }
290 opt_space->dst1opt = fopt->dst1opt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700291 opt_space->opt_flen = fopt->opt_flen;
292 return opt_space;
293}
294
295static unsigned long check_linger(unsigned long ttl)
296{
297 if (ttl < FL_MIN_LINGER)
298 return FL_MIN_LINGER*HZ;
299 if (ttl > FL_MAX_LINGER && !capable(CAP_NET_ADMIN))
300 return 0;
301 return ttl*HZ;
302}
303
304static int fl6_renew(struct ip6_flowlabel *fl, unsigned long linger, unsigned long expires)
305{
306 linger = check_linger(linger);
307 if (!linger)
308 return -EPERM;
309 expires = check_linger(expires);
310 if (!expires)
311 return -EPERM;
312 fl->lastuse = jiffies;
313 if (time_before(fl->linger, linger))
314 fl->linger = linger;
315 if (time_before(expires, fl->linger))
316 expires = fl->linger;
317 if (time_before(fl->expires, fl->lastuse + expires))
318 fl->expires = fl->lastuse + expires;
319 return 0;
320}
321
322static struct ip6_flowlabel *
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700323fl_create(struct net *net, struct in6_flowlabel_req *freq, char __user *optval,
324 int optlen, int *err_p)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700325{
326 struct ip6_flowlabel *fl;
327 int olen;
328 int addr_type;
329 int err;
330
331 err = -ENOMEM;
Ingo Oeser0c600ed2006-03-20 23:01:32 -0800332 fl = kzalloc(sizeof(*fl), GFP_KERNEL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700333 if (fl == NULL)
334 goto done;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700335
336 olen = optlen - CMSG_ALIGN(sizeof(*freq));
337 if (olen > 0) {
338 struct msghdr msg;
339 struct flowi flowi;
340 int junk;
341
342 err = -ENOMEM;
343 fl->opt = kmalloc(sizeof(*fl->opt) + olen, GFP_KERNEL);
344 if (fl->opt == NULL)
345 goto done;
346
347 memset(fl->opt, 0, sizeof(*fl->opt));
348 fl->opt->tot_len = sizeof(*fl->opt) + olen;
349 err = -EFAULT;
350 if (copy_from_user(fl->opt+1, optval+CMSG_ALIGN(sizeof(*freq)), olen))
351 goto done;
352
353 msg.msg_controllen = olen;
354 msg.msg_control = (void*)(fl->opt+1);
355 flowi.oif = 0;
356
YOSHIFUJI Hideaki91e19082008-06-04 13:02:49 +0900357 err = datagram_send_ctl(net, &msg, &flowi, fl->opt, &junk, &junk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700358 if (err)
359 goto done;
360 err = -EINVAL;
361 if (fl->opt->opt_flen)
362 goto done;
363 if (fl->opt->opt_nflen == 0) {
364 kfree(fl->opt);
365 fl->opt = NULL;
366 }
367 }
368
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700369 fl->fl_net = hold_net(net);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700370 fl->expires = jiffies;
371 err = fl6_renew(fl, freq->flr_linger, freq->flr_expires);
372 if (err)
373 goto done;
374 fl->share = freq->flr_share;
375 addr_type = ipv6_addr_type(&freq->flr_dst);
376 if ((addr_type&IPV6_ADDR_MAPPED)
James Morrisc6817e42006-10-30 18:56:06 -0800377 || addr_type == IPV6_ADDR_ANY) {
378 err = -EINVAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700379 goto done;
James Morrisc6817e42006-10-30 18:56:06 -0800380 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700381 ipv6_addr_copy(&fl->dst, &freq->flr_dst);
382 atomic_set(&fl->users, 1);
383 switch (fl->share) {
384 case IPV6_FL_S_EXCL:
385 case IPV6_FL_S_ANY:
386 break;
387 case IPV6_FL_S_PROCESS:
388 fl->owner = current->pid;
389 break;
390 case IPV6_FL_S_USER:
391 fl->owner = current->euid;
392 break;
393 default:
394 err = -EINVAL;
395 goto done;
396 }
397 return fl;
398
399done:
400 fl_free(fl);
401 *err_p = err;
402 return NULL;
403}
404
405static int mem_check(struct sock *sk)
406{
407 struct ipv6_pinfo *np = inet6_sk(sk);
408 struct ipv6_fl_socklist *sfl;
409 int room = FL_MAX_SIZE - atomic_read(&fl_size);
410 int count = 0;
411
412 if (room > FL_MAX_SIZE - FL_MAX_PER_SOCK)
413 return 0;
414
415 for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next)
416 count++;
417
418 if (room <= 0 ||
419 ((count >= FL_MAX_PER_SOCK ||
420 (count > 0 && room < FL_MAX_SIZE/2) || room < FL_MAX_SIZE/4)
421 && !capable(CAP_NET_ADMIN)))
422 return -ENOBUFS;
423
424 return 0;
425}
426
427static int ipv6_hdr_cmp(struct ipv6_opt_hdr *h1, struct ipv6_opt_hdr *h2)
428{
429 if (h1 == h2)
430 return 0;
431 if (h1 == NULL || h2 == NULL)
432 return 1;
433 if (h1->hdrlen != h2->hdrlen)
434 return 1;
435 return memcmp(h1+1, h2+1, ((h1->hdrlen+1)<<3) - sizeof(*h1));
436}
437
438static int ipv6_opt_cmp(struct ipv6_txoptions *o1, struct ipv6_txoptions *o2)
439{
440 if (o1 == o2)
441 return 0;
442 if (o1 == NULL || o2 == NULL)
443 return 1;
444 if (o1->opt_nflen != o2->opt_nflen)
445 return 1;
446 if (ipv6_hdr_cmp(o1->hopopt, o2->hopopt))
447 return 1;
448 if (ipv6_hdr_cmp(o1->dst0opt, o2->dst0opt))
449 return 1;
450 if (ipv6_hdr_cmp((struct ipv6_opt_hdr *)o1->srcrt, (struct ipv6_opt_hdr *)o2->srcrt))
451 return 1;
452 return 0;
453}
454
Pavel Emelyanov04028042007-10-18 05:14:58 -0700455static inline void fl_link(struct ipv6_pinfo *np, struct ipv6_fl_socklist *sfl,
456 struct ip6_flowlabel *fl)
457{
458 write_lock_bh(&ip6_sk_fl_lock);
459 sfl->fl = fl;
460 sfl->next = np->ipv6_fl_list;
461 np->ipv6_fl_list = sfl;
462 write_unlock_bh(&ip6_sk_fl_lock);
463}
464
Linus Torvalds1da177e2005-04-16 15:20:36 -0700465int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen)
466{
467 int err;
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700468 struct net *net = sock_net(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700469 struct ipv6_pinfo *np = inet6_sk(sk);
470 struct in6_flowlabel_req freq;
471 struct ipv6_fl_socklist *sfl1=NULL;
472 struct ipv6_fl_socklist *sfl, **sflp;
Pavel Emelyanov78c2e502007-10-18 05:18:56 -0700473 struct ip6_flowlabel *fl, *fl1 = NULL;
474
Linus Torvalds1da177e2005-04-16 15:20:36 -0700475
476 if (optlen < sizeof(freq))
477 return -EINVAL;
478
479 if (copy_from_user(&freq, optval, sizeof(freq)))
480 return -EFAULT;
481
482 switch (freq.flr_action) {
483 case IPV6_FL_A_PUT:
484 write_lock_bh(&ip6_sk_fl_lock);
485 for (sflp = &np->ipv6_fl_list; (sfl=*sflp)!=NULL; sflp = &sfl->next) {
486 if (sfl->fl->label == freq.flr_label) {
487 if (freq.flr_label == (np->flow_label&IPV6_FLOWLABEL_MASK))
488 np->flow_label &= ~IPV6_FLOWLABEL_MASK;
489 *sflp = sfl->next;
490 write_unlock_bh(&ip6_sk_fl_lock);
491 fl_release(sfl->fl);
492 kfree(sfl);
493 return 0;
494 }
495 }
496 write_unlock_bh(&ip6_sk_fl_lock);
497 return -ESRCH;
498
499 case IPV6_FL_A_RENEW:
500 read_lock_bh(&ip6_sk_fl_lock);
501 for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) {
502 if (sfl->fl->label == freq.flr_label) {
503 err = fl6_renew(sfl->fl, freq.flr_linger, freq.flr_expires);
504 read_unlock_bh(&ip6_sk_fl_lock);
505 return err;
506 }
507 }
508 read_unlock_bh(&ip6_sk_fl_lock);
509
510 if (freq.flr_share == IPV6_FL_S_NONE && capable(CAP_NET_ADMIN)) {
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700511 fl = fl_lookup(net, freq.flr_label);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700512 if (fl) {
513 err = fl6_renew(fl, freq.flr_linger, freq.flr_expires);
514 fl_release(fl);
515 return err;
516 }
517 }
518 return -ESRCH;
519
520 case IPV6_FL_A_GET:
521 if (freq.flr_label & ~IPV6_FLOWLABEL_MASK)
522 return -EINVAL;
523
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700524 fl = fl_create(net, &freq, optval, optlen, &err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700525 if (fl == NULL)
526 return err;
527 sfl1 = kmalloc(sizeof(*sfl1), GFP_KERNEL);
528
529 if (freq.flr_label) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700530 err = -EEXIST;
531 read_lock_bh(&ip6_sk_fl_lock);
532 for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) {
533 if (sfl->fl->label == freq.flr_label) {
534 if (freq.flr_flags&IPV6_FL_F_EXCL) {
535 read_unlock_bh(&ip6_sk_fl_lock);
536 goto done;
537 }
538 fl1 = sfl->fl;
Yan Zheng4ea6a802005-10-24 19:55:23 +0800539 atomic_inc(&fl1->users);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700540 break;
541 }
542 }
543 read_unlock_bh(&ip6_sk_fl_lock);
544
545 if (fl1 == NULL)
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700546 fl1 = fl_lookup(net, freq.flr_label);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700547 if (fl1) {
Pavel Emelyanov78c2e502007-10-18 05:18:56 -0700548recheck:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700549 err = -EEXIST;
550 if (freq.flr_flags&IPV6_FL_F_EXCL)
551 goto release;
552 err = -EPERM;
553 if (fl1->share == IPV6_FL_S_EXCL ||
554 fl1->share != fl->share ||
555 fl1->owner != fl->owner)
556 goto release;
557
558 err = -EINVAL;
559 if (!ipv6_addr_equal(&fl1->dst, &fl->dst) ||
560 ipv6_opt_cmp(fl1->opt, fl->opt))
561 goto release;
562
563 err = -ENOMEM;
564 if (sfl1 == NULL)
565 goto release;
566 if (fl->linger > fl1->linger)
567 fl1->linger = fl->linger;
568 if ((long)(fl->expires - fl1->expires) > 0)
569 fl1->expires = fl->expires;
Pavel Emelyanov04028042007-10-18 05:14:58 -0700570 fl_link(np, sfl1, fl1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700571 fl_free(fl);
572 return 0;
573
574release:
575 fl_release(fl1);
576 goto done;
577 }
578 }
579 err = -ENOENT;
580 if (!(freq.flr_flags&IPV6_FL_F_CREATE))
581 goto done;
582
583 err = -ENOMEM;
584 if (sfl1 == NULL || (err = mem_check(sk)) != 0)
585 goto done;
586
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700587 fl1 = fl_intern(net, fl, freq.flr_label);
Pavel Emelyanov78c2e502007-10-18 05:18:56 -0700588 if (fl1 != NULL)
589 goto recheck;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590
David S. Miller6c94d362005-05-29 20:28:01 -0700591 if (!freq.flr_label) {
592 if (copy_to_user(&((struct in6_flowlabel_req __user *) optval)->flr_label,
593 &fl->label, sizeof(fl->label))) {
594 /* Intentionally ignore fault. */
595 }
596 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700597
Pavel Emelyanov04028042007-10-18 05:14:58 -0700598 fl_link(np, sfl1, fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700599 return 0;
600
601 default:
602 return -EINVAL;
603 }
604
605done:
606 fl_free(fl);
607 kfree(sfl1);
608 return err;
609}
610
611#ifdef CONFIG_PROC_FS
612
613struct ip6fl_iter_state {
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700614 struct seq_net_private p;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700615 int bucket;
616};
617
618#define ip6fl_seq_private(seq) ((struct ip6fl_iter_state *)(seq)->private)
619
620static struct ip6_flowlabel *ip6fl_get_first(struct seq_file *seq)
621{
622 struct ip6_flowlabel *fl = NULL;
623 struct ip6fl_iter_state *state = ip6fl_seq_private(seq);
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700624 struct net *net = seq_file_net(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700625
626 for (state->bucket = 0; state->bucket <= FL_HASH_MASK; ++state->bucket) {
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700627 fl = fl_ht[state->bucket];
628
629 while (fl && fl->fl_net != net)
630 fl = fl->next;
631 if (fl)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700632 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700633 }
634 return fl;
635}
636
637static struct ip6_flowlabel *ip6fl_get_next(struct seq_file *seq, struct ip6_flowlabel *fl)
638{
639 struct ip6fl_iter_state *state = ip6fl_seq_private(seq);
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700640 struct net *net = seq_file_net(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700641
642 fl = fl->next;
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700643try_again:
644 while (fl && fl->fl_net != net)
645 fl = fl->next;
646
Linus Torvalds1da177e2005-04-16 15:20:36 -0700647 while (!fl) {
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700648 if (++state->bucket <= FL_HASH_MASK) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700649 fl = fl_ht[state->bucket];
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700650 goto try_again;
651 } else
James Morrisbcd62072006-10-30 15:08:42 -0800652 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700653 }
654 return fl;
655}
656
657static struct ip6_flowlabel *ip6fl_get_idx(struct seq_file *seq, loff_t pos)
658{
659 struct ip6_flowlabel *fl = ip6fl_get_first(seq);
660 if (fl)
661 while (pos && (fl = ip6fl_get_next(seq, fl)) != NULL)
662 --pos;
663 return pos ? NULL : fl;
664}
665
666static void *ip6fl_seq_start(struct seq_file *seq, loff_t *pos)
Eric Dumazet9a429c42008-01-01 21:58:02 -0800667 __acquires(ip6_fl_lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700668{
669 read_lock_bh(&ip6_fl_lock);
670 return *pos ? ip6fl_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
671}
672
673static void *ip6fl_seq_next(struct seq_file *seq, void *v, loff_t *pos)
674{
675 struct ip6_flowlabel *fl;
676
677 if (v == SEQ_START_TOKEN)
678 fl = ip6fl_get_first(seq);
679 else
680 fl = ip6fl_get_next(seq, v);
681 ++*pos;
682 return fl;
683}
684
685static void ip6fl_seq_stop(struct seq_file *seq, void *v)
Eric Dumazet9a429c42008-01-01 21:58:02 -0800686 __releases(ip6_fl_lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700687{
688 read_unlock_bh(&ip6_fl_lock);
689}
690
James Morris1b7c2db2006-10-31 00:43:44 -0800691static int ip6fl_seq_show(struct seq_file *seq, void *v)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700692{
James Morris1b7c2db2006-10-31 00:43:44 -0800693 if (v == SEQ_START_TOKEN)
694 seq_printf(seq, "%-5s %-1s %-6s %-6s %-6s %-8s %-32s %s\n",
695 "Label", "S", "Owner", "Users", "Linger", "Expires", "Dst", "Opt");
696 else {
697 struct ip6_flowlabel *fl = v;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698 seq_printf(seq,
YOSHIFUJI Hideaki9343e792006-01-17 02:10:53 -0800699 "%05X %-1d %-6d %-6d %-6ld %-8ld " NIP6_SEQFMT " %-4d\n",
Linus Torvalds1da177e2005-04-16 15:20:36 -0700700 (unsigned)ntohl(fl->label),
701 fl->share,
702 (unsigned)fl->owner,
703 atomic_read(&fl->users),
704 fl->linger/HZ,
705 (long)(fl->expires - jiffies)/HZ,
706 NIP6(fl->dst),
707 fl->opt ? fl->opt->opt_nflen : 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700708 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700709 return 0;
710}
711
Philippe De Muyter56b3d972007-07-10 23:07:31 -0700712static const struct seq_operations ip6fl_seq_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700713 .start = ip6fl_seq_start,
714 .next = ip6fl_seq_next,
715 .stop = ip6fl_seq_stop,
716 .show = ip6fl_seq_show,
717};
718
719static int ip6fl_seq_open(struct inode *inode, struct file *file)
720{
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700721 return seq_open_net(inode, file, &ip6fl_seq_ops,
722 sizeof(struct ip6fl_iter_state));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700723}
724
Arjan van de Ven9a321442007-02-12 00:55:35 -0800725static const struct file_operations ip6fl_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700726 .owner = THIS_MODULE,
727 .open = ip6fl_seq_open,
728 .read = seq_read,
729 .llseek = seq_lseek,
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700730 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700731};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700732
Daniel Lezcano0a3e78a2007-12-11 02:23:18 -0800733static int ip6_flowlabel_proc_init(struct net *net)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700734{
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700735 if (!proc_net_fops_create(net, "ip6_flowlabel",
736 S_IRUGO, &ip6fl_seq_fops))
Daniel Lezcano0a3e78a2007-12-11 02:23:18 -0800737 return -ENOMEM;
738 return 0;
739}
740
741static void ip6_flowlabel_proc_fini(struct net *net)
742{
743 proc_net_remove(net, "ip6_flowlabel");
744}
745#else
746static inline int ip6_flowlabel_proc_init(struct net *net)
747{
748 return 0;
749}
750static inline void ip6_flowlabel_proc_fini(struct net *net)
751{
752 return ;
753}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700754#endif
Daniel Lezcano0a3e78a2007-12-11 02:23:18 -0800755
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700756static inline void ip6_flowlabel_net_exit(struct net *net)
757{
758 ip6_fl_purge(net);
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700759 ip6_flowlabel_proc_fini(net);
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700760}
761
762static struct pernet_operations ip6_flowlabel_net_ops = {
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700763 .init = ip6_flowlabel_proc_init,
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700764 .exit = ip6_flowlabel_net_exit,
765};
766
Daniel Lezcano0a3e78a2007-12-11 02:23:18 -0800767int ip6_flowlabel_init(void)
768{
Benjamin Thery5983a3d2008-03-26 16:53:30 -0700769 return register_pernet_subsys(&ip6_flowlabel_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700770}
771
772void ip6_flowlabel_cleanup(void)
773{
774 del_timer(&ip6_fl_gc_timer);
Benjamin Thery60e8fbc2008-03-26 16:53:08 -0700775 unregister_pernet_subsys(&ip6_flowlabel_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700776}