Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
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 | ||
4fc268d2 | 12 | #include <linux/capability.h> |
1da177e4 LT |
13 | #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 | ||
457c4cbc | 24 | #include <net/net_namespace.h> |
1da177e4 LT |
25 | #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 | ||
50 | static atomic_t fl_size = ATOMIC_INIT(0); | |
51 | static struct ip6_flowlabel *fl_ht[FL_HASH_MASK+1]; | |
52 | ||
53 | static void ip6_fl_gc(unsigned long dummy); | |
8d06afab | 54 | static DEFINE_TIMER(ip6_fl_gc_timer, ip6_fl_gc, 0, 0); |
1da177e4 LT |
55 | |
56 | /* FL hash table lock: it protects only of GC */ | |
57 | ||
58 | static DEFINE_RWLOCK(ip6_fl_lock); | |
59 | ||
60 | /* Big socket sock */ | |
61 | ||
62 | static DEFINE_RWLOCK(ip6_sk_fl_lock); | |
63 | ||
64 | ||
60e8fbc4 | 65 | static inline struct ip6_flowlabel *__fl_lookup(struct net *net, __be32 label) |
1da177e4 LT |
66 | { |
67 | struct ip6_flowlabel *fl; | |
68 | ||
69 | for (fl=fl_ht[FL_HASH(label)]; fl; fl = fl->next) { | |
60e8fbc4 | 70 | if (fl->label == label && fl->fl_net == net) |
1da177e4 LT |
71 | return fl; |
72 | } | |
73 | return NULL; | |
74 | } | |
75 | ||
60e8fbc4 | 76 | static struct ip6_flowlabel *fl_lookup(struct net *net, __be32 label) |
1da177e4 LT |
77 | { |
78 | struct ip6_flowlabel *fl; | |
79 | ||
80 | read_lock_bh(&ip6_fl_lock); | |
60e8fbc4 | 81 | fl = __fl_lookup(net, label); |
1da177e4 LT |
82 | if (fl) |
83 | atomic_inc(&fl->users); | |
84 | read_unlock_bh(&ip6_fl_lock); | |
85 | return fl; | |
86 | } | |
87 | ||
88 | ||
89 | static void fl_free(struct ip6_flowlabel *fl) | |
90 | { | |
60e8fbc4 BT |
91 | if (fl) { |
92 | release_net(fl->fl_net); | |
1da177e4 | 93 | kfree(fl->opt); |
60e8fbc4 | 94 | } |
1da177e4 LT |
95 | kfree(fl); |
96 | } | |
97 | ||
98 | static 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 | } | |
1da177e4 LT |
117 | write_unlock_bh(&ip6_fl_lock); |
118 | } | |
119 | ||
120 | static 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) { | |
60e8fbc4 | 152 | mod_timer(&ip6_fl_gc_timer, sched); |
1da177e4 LT |
153 | } |
154 | write_unlock(&ip6_fl_lock); | |
155 | } | |
156 | ||
60e8fbc4 BT |
157 | static 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 | ||
178 | static struct ip6_flowlabel *fl_intern(struct net *net, | |
179 | struct ip6_flowlabel *fl, __be32 label) | |
1da177e4 | 180 | { |
78c2e502 PE |
181 | struct ip6_flowlabel *lfl; |
182 | ||
1da177e4 LT |
183 | 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) { | |
60e8fbc4 | 190 | lfl = __fl_lookup(net, fl->label); |
1da177e4 LT |
191 | if (lfl == NULL) |
192 | break; | |
193 | } | |
194 | } | |
78c2e502 PE |
195 | } 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 | */ | |
60e8fbc4 | 204 | lfl = __fl_lookup(net, fl->label); |
78c2e502 PE |
205 | if (lfl != NULL) { |
206 | atomic_inc(&lfl->users); | |
207 | write_unlock_bh(&ip6_fl_lock); | |
208 | return lfl; | |
209 | } | |
1da177e4 LT |
210 | } |
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); | |
78c2e502 | 217 | return NULL; |
1da177e4 LT |
218 | } |
219 | ||
220 | ||
221 | ||
222 | /* Socket flowlabel lists */ | |
223 | ||
90bcaf7b | 224 | struct ip6_flowlabel * fl6_sock_lookup(struct sock *sk, __be32 label) |
1da177e4 LT |
225 | { |
226 | struct ipv6_fl_socklist *sfl; | |
227 | struct ipv6_pinfo *np = inet6_sk(sk); | |
228 | ||
229 | label &= IPV6_FLOWLABEL_MASK; | |
230 | ||
bd0bf577 | 231 | read_lock_bh(&ip6_sk_fl_lock); |
1da177e4 LT |
232 | 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); | |
52f095ee | 237 | read_unlock_bh(&ip6_sk_fl_lock); |
1da177e4 LT |
238 | return fl; |
239 | } | |
240 | } | |
bd0bf577 | 241 | read_unlock_bh(&ip6_sk_fl_lock); |
1da177e4 LT |
242 | return NULL; |
243 | } | |
244 | ||
3cf3dc6c ACM |
245 | EXPORT_SYMBOL_GPL(fl6_sock_lookup); |
246 | ||
1da177e4 LT |
247 | void 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 | ||
268 | struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions * opt_space, | |
269 | struct ip6_flowlabel * fl, | |
270 | struct ipv6_txoptions * fopt) | |
271 | { | |
df9890c3 | 272 | struct ipv6_txoptions * fl_opt = fl->opt; |
1ab1457c | 273 | |
df9890c3 YH |
274 | if (fopt == NULL || fopt->opt_flen == 0) |
275 | return fl_opt; | |
1ab1457c | 276 | |
1da177e4 LT |
277 | if (fl_opt != NULL) { |
278 | opt_space->hopopt = fl_opt->hopopt; | |
df9890c3 | 279 | opt_space->dst0opt = fl_opt->dst0opt; |
1da177e4 LT |
280 | 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; | |
1da177e4 LT |
291 | opt_space->opt_flen = fopt->opt_flen; |
292 | return opt_space; | |
293 | } | |
294 | ||
295 | static 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 | ||
304 | static 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 | ||
322 | static struct ip6_flowlabel * | |
60e8fbc4 BT |
323 | fl_create(struct net *net, struct in6_flowlabel_req *freq, char __user *optval, |
324 | int optlen, int *err_p) | |
1da177e4 | 325 | { |
684de409 | 326 | struct ip6_flowlabel *fl = NULL; |
1da177e4 LT |
327 | int olen; |
328 | int addr_type; | |
329 | int err; | |
330 | ||
684de409 DM |
331 | olen = optlen - CMSG_ALIGN(sizeof(*freq)); |
332 | err = -EINVAL; | |
333 | if (olen > 64 * 1024) | |
334 | goto done; | |
335 | ||
1da177e4 | 336 | err = -ENOMEM; |
0c600eda | 337 | fl = kzalloc(sizeof(*fl), GFP_KERNEL); |
1da177e4 LT |
338 | if (fl == NULL) |
339 | goto done; | |
1da177e4 | 340 | |
1da177e4 LT |
341 | if (olen > 0) { |
342 | struct msghdr msg; | |
343 | struct flowi flowi; | |
344 | int junk; | |
345 | ||
346 | err = -ENOMEM; | |
347 | fl->opt = kmalloc(sizeof(*fl->opt) + olen, GFP_KERNEL); | |
348 | if (fl->opt == NULL) | |
349 | goto done; | |
350 | ||
351 | memset(fl->opt, 0, sizeof(*fl->opt)); | |
352 | fl->opt->tot_len = sizeof(*fl->opt) + olen; | |
353 | err = -EFAULT; | |
354 | if (copy_from_user(fl->opt+1, optval+CMSG_ALIGN(sizeof(*freq)), olen)) | |
355 | goto done; | |
356 | ||
357 | msg.msg_controllen = olen; | |
358 | msg.msg_control = (void*)(fl->opt+1); | |
359 | flowi.oif = 0; | |
360 | ||
91e1908f | 361 | err = datagram_send_ctl(net, &msg, &flowi, fl->opt, &junk, &junk); |
1da177e4 LT |
362 | if (err) |
363 | goto done; | |
364 | err = -EINVAL; | |
365 | if (fl->opt->opt_flen) | |
366 | goto done; | |
367 | if (fl->opt->opt_nflen == 0) { | |
368 | kfree(fl->opt); | |
369 | fl->opt = NULL; | |
370 | } | |
371 | } | |
372 | ||
60e8fbc4 | 373 | fl->fl_net = hold_net(net); |
1da177e4 LT |
374 | fl->expires = jiffies; |
375 | err = fl6_renew(fl, freq->flr_linger, freq->flr_expires); | |
376 | if (err) | |
377 | goto done; | |
378 | fl->share = freq->flr_share; | |
379 | addr_type = ipv6_addr_type(&freq->flr_dst); | |
380 | if ((addr_type&IPV6_ADDR_MAPPED) | |
c6817e4c JM |
381 | || addr_type == IPV6_ADDR_ANY) { |
382 | err = -EINVAL; | |
1da177e4 | 383 | goto done; |
c6817e4c | 384 | } |
1da177e4 LT |
385 | ipv6_addr_copy(&fl->dst, &freq->flr_dst); |
386 | atomic_set(&fl->users, 1); | |
387 | switch (fl->share) { | |
388 | case IPV6_FL_S_EXCL: | |
389 | case IPV6_FL_S_ANY: | |
390 | break; | |
391 | case IPV6_FL_S_PROCESS: | |
392 | fl->owner = current->pid; | |
393 | break; | |
394 | case IPV6_FL_S_USER: | |
f82b3590 | 395 | fl->owner = current_euid(); |
1da177e4 LT |
396 | break; |
397 | default: | |
398 | err = -EINVAL; | |
399 | goto done; | |
400 | } | |
401 | return fl; | |
402 | ||
403 | done: | |
404 | fl_free(fl); | |
405 | *err_p = err; | |
406 | return NULL; | |
407 | } | |
408 | ||
409 | static int mem_check(struct sock *sk) | |
410 | { | |
411 | struct ipv6_pinfo *np = inet6_sk(sk); | |
412 | struct ipv6_fl_socklist *sfl; | |
413 | int room = FL_MAX_SIZE - atomic_read(&fl_size); | |
414 | int count = 0; | |
415 | ||
416 | if (room > FL_MAX_SIZE - FL_MAX_PER_SOCK) | |
417 | return 0; | |
418 | ||
419 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) | |
420 | count++; | |
421 | ||
422 | if (room <= 0 || | |
423 | ((count >= FL_MAX_PER_SOCK || | |
424 | (count > 0 && room < FL_MAX_SIZE/2) || room < FL_MAX_SIZE/4) | |
425 | && !capable(CAP_NET_ADMIN))) | |
426 | return -ENOBUFS; | |
427 | ||
428 | return 0; | |
429 | } | |
430 | ||
431 | static int ipv6_hdr_cmp(struct ipv6_opt_hdr *h1, struct ipv6_opt_hdr *h2) | |
432 | { | |
433 | if (h1 == h2) | |
434 | return 0; | |
435 | if (h1 == NULL || h2 == NULL) | |
436 | return 1; | |
437 | if (h1->hdrlen != h2->hdrlen) | |
438 | return 1; | |
439 | return memcmp(h1+1, h2+1, ((h1->hdrlen+1)<<3) - sizeof(*h1)); | |
440 | } | |
441 | ||
442 | static int ipv6_opt_cmp(struct ipv6_txoptions *o1, struct ipv6_txoptions *o2) | |
443 | { | |
444 | if (o1 == o2) | |
445 | return 0; | |
446 | if (o1 == NULL || o2 == NULL) | |
447 | return 1; | |
448 | if (o1->opt_nflen != o2->opt_nflen) | |
449 | return 1; | |
450 | if (ipv6_hdr_cmp(o1->hopopt, o2->hopopt)) | |
451 | return 1; | |
452 | if (ipv6_hdr_cmp(o1->dst0opt, o2->dst0opt)) | |
453 | return 1; | |
454 | if (ipv6_hdr_cmp((struct ipv6_opt_hdr *)o1->srcrt, (struct ipv6_opt_hdr *)o2->srcrt)) | |
455 | return 1; | |
456 | return 0; | |
457 | } | |
458 | ||
04028045 PE |
459 | static inline void fl_link(struct ipv6_pinfo *np, struct ipv6_fl_socklist *sfl, |
460 | struct ip6_flowlabel *fl) | |
461 | { | |
462 | write_lock_bh(&ip6_sk_fl_lock); | |
463 | sfl->fl = fl; | |
464 | sfl->next = np->ipv6_fl_list; | |
465 | np->ipv6_fl_list = sfl; | |
466 | write_unlock_bh(&ip6_sk_fl_lock); | |
467 | } | |
468 | ||
1da177e4 LT |
469 | int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen) |
470 | { | |
55205d40 | 471 | int uninitialized_var(err); |
60e8fbc4 | 472 | struct net *net = sock_net(sk); |
1da177e4 LT |
473 | struct ipv6_pinfo *np = inet6_sk(sk); |
474 | struct in6_flowlabel_req freq; | |
475 | struct ipv6_fl_socklist *sfl1=NULL; | |
476 | struct ipv6_fl_socklist *sfl, **sflp; | |
78c2e502 PE |
477 | struct ip6_flowlabel *fl, *fl1 = NULL; |
478 | ||
1da177e4 LT |
479 | |
480 | if (optlen < sizeof(freq)) | |
481 | return -EINVAL; | |
482 | ||
483 | if (copy_from_user(&freq, optval, sizeof(freq))) | |
484 | return -EFAULT; | |
485 | ||
486 | switch (freq.flr_action) { | |
487 | case IPV6_FL_A_PUT: | |
488 | write_lock_bh(&ip6_sk_fl_lock); | |
489 | for (sflp = &np->ipv6_fl_list; (sfl=*sflp)!=NULL; sflp = &sfl->next) { | |
490 | if (sfl->fl->label == freq.flr_label) { | |
491 | if (freq.flr_label == (np->flow_label&IPV6_FLOWLABEL_MASK)) | |
492 | np->flow_label &= ~IPV6_FLOWLABEL_MASK; | |
493 | *sflp = sfl->next; | |
494 | write_unlock_bh(&ip6_sk_fl_lock); | |
495 | fl_release(sfl->fl); | |
496 | kfree(sfl); | |
497 | return 0; | |
498 | } | |
499 | } | |
500 | write_unlock_bh(&ip6_sk_fl_lock); | |
501 | return -ESRCH; | |
502 | ||
503 | case IPV6_FL_A_RENEW: | |
504 | read_lock_bh(&ip6_sk_fl_lock); | |
505 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) { | |
506 | if (sfl->fl->label == freq.flr_label) { | |
507 | err = fl6_renew(sfl->fl, freq.flr_linger, freq.flr_expires); | |
508 | read_unlock_bh(&ip6_sk_fl_lock); | |
509 | return err; | |
510 | } | |
511 | } | |
512 | read_unlock_bh(&ip6_sk_fl_lock); | |
513 | ||
514 | if (freq.flr_share == IPV6_FL_S_NONE && capable(CAP_NET_ADMIN)) { | |
60e8fbc4 | 515 | fl = fl_lookup(net, freq.flr_label); |
1da177e4 LT |
516 | if (fl) { |
517 | err = fl6_renew(fl, freq.flr_linger, freq.flr_expires); | |
518 | fl_release(fl); | |
519 | return err; | |
520 | } | |
521 | } | |
522 | return -ESRCH; | |
523 | ||
524 | case IPV6_FL_A_GET: | |
525 | if (freq.flr_label & ~IPV6_FLOWLABEL_MASK) | |
526 | return -EINVAL; | |
527 | ||
60e8fbc4 | 528 | fl = fl_create(net, &freq, optval, optlen, &err); |
1da177e4 LT |
529 | if (fl == NULL) |
530 | return err; | |
531 | sfl1 = kmalloc(sizeof(*sfl1), GFP_KERNEL); | |
532 | ||
533 | if (freq.flr_label) { | |
1da177e4 LT |
534 | err = -EEXIST; |
535 | read_lock_bh(&ip6_sk_fl_lock); | |
536 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) { | |
537 | if (sfl->fl->label == freq.flr_label) { | |
538 | if (freq.flr_flags&IPV6_FL_F_EXCL) { | |
539 | read_unlock_bh(&ip6_sk_fl_lock); | |
540 | goto done; | |
541 | } | |
542 | fl1 = sfl->fl; | |
4ea6a804 | 543 | atomic_inc(&fl1->users); |
1da177e4 LT |
544 | break; |
545 | } | |
546 | } | |
547 | read_unlock_bh(&ip6_sk_fl_lock); | |
548 | ||
549 | if (fl1 == NULL) | |
60e8fbc4 | 550 | fl1 = fl_lookup(net, freq.flr_label); |
1da177e4 | 551 | if (fl1) { |
78c2e502 | 552 | recheck: |
1da177e4 LT |
553 | err = -EEXIST; |
554 | if (freq.flr_flags&IPV6_FL_F_EXCL) | |
555 | goto release; | |
556 | err = -EPERM; | |
557 | if (fl1->share == IPV6_FL_S_EXCL || | |
558 | fl1->share != fl->share || | |
559 | fl1->owner != fl->owner) | |
560 | goto release; | |
561 | ||
562 | err = -EINVAL; | |
563 | if (!ipv6_addr_equal(&fl1->dst, &fl->dst) || | |
564 | ipv6_opt_cmp(fl1->opt, fl->opt)) | |
565 | goto release; | |
566 | ||
567 | err = -ENOMEM; | |
568 | if (sfl1 == NULL) | |
569 | goto release; | |
570 | if (fl->linger > fl1->linger) | |
571 | fl1->linger = fl->linger; | |
572 | if ((long)(fl->expires - fl1->expires) > 0) | |
573 | fl1->expires = fl->expires; | |
04028045 | 574 | fl_link(np, sfl1, fl1); |
1da177e4 LT |
575 | fl_free(fl); |
576 | return 0; | |
577 | ||
578 | release: | |
579 | fl_release(fl1); | |
580 | goto done; | |
581 | } | |
582 | } | |
583 | err = -ENOENT; | |
584 | if (!(freq.flr_flags&IPV6_FL_F_CREATE)) | |
585 | goto done; | |
586 | ||
587 | err = -ENOMEM; | |
588 | if (sfl1 == NULL || (err = mem_check(sk)) != 0) | |
589 | goto done; | |
590 | ||
60e8fbc4 | 591 | fl1 = fl_intern(net, fl, freq.flr_label); |
78c2e502 PE |
592 | if (fl1 != NULL) |
593 | goto recheck; | |
1da177e4 | 594 | |
6c94d361 DM |
595 | if (!freq.flr_label) { |
596 | if (copy_to_user(&((struct in6_flowlabel_req __user *) optval)->flr_label, | |
597 | &fl->label, sizeof(fl->label))) { | |
598 | /* Intentionally ignore fault. */ | |
599 | } | |
600 | } | |
1da177e4 | 601 | |
04028045 | 602 | fl_link(np, sfl1, fl); |
1da177e4 LT |
603 | return 0; |
604 | ||
605 | default: | |
606 | return -EINVAL; | |
607 | } | |
608 | ||
609 | done: | |
610 | fl_free(fl); | |
611 | kfree(sfl1); | |
612 | return err; | |
613 | } | |
614 | ||
615 | #ifdef CONFIG_PROC_FS | |
616 | ||
617 | struct ip6fl_iter_state { | |
5983a3df | 618 | struct seq_net_private p; |
1da177e4 LT |
619 | int bucket; |
620 | }; | |
621 | ||
622 | #define ip6fl_seq_private(seq) ((struct ip6fl_iter_state *)(seq)->private) | |
623 | ||
624 | static struct ip6_flowlabel *ip6fl_get_first(struct seq_file *seq) | |
625 | { | |
626 | struct ip6_flowlabel *fl = NULL; | |
627 | struct ip6fl_iter_state *state = ip6fl_seq_private(seq); | |
5983a3df | 628 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
629 | |
630 | for (state->bucket = 0; state->bucket <= FL_HASH_MASK; ++state->bucket) { | |
5983a3df BT |
631 | fl = fl_ht[state->bucket]; |
632 | ||
633 | while (fl && fl->fl_net != net) | |
634 | fl = fl->next; | |
635 | if (fl) | |
1da177e4 | 636 | break; |
1da177e4 LT |
637 | } |
638 | return fl; | |
639 | } | |
640 | ||
641 | static struct ip6_flowlabel *ip6fl_get_next(struct seq_file *seq, struct ip6_flowlabel *fl) | |
642 | { | |
643 | struct ip6fl_iter_state *state = ip6fl_seq_private(seq); | |
5983a3df | 644 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
645 | |
646 | fl = fl->next; | |
5983a3df BT |
647 | try_again: |
648 | while (fl && fl->fl_net != net) | |
649 | fl = fl->next; | |
650 | ||
1da177e4 | 651 | while (!fl) { |
5983a3df | 652 | if (++state->bucket <= FL_HASH_MASK) { |
1da177e4 | 653 | fl = fl_ht[state->bucket]; |
5983a3df BT |
654 | goto try_again; |
655 | } else | |
bcd62075 | 656 | break; |
1da177e4 LT |
657 | } |
658 | return fl; | |
659 | } | |
660 | ||
661 | static struct ip6_flowlabel *ip6fl_get_idx(struct seq_file *seq, loff_t pos) | |
662 | { | |
663 | struct ip6_flowlabel *fl = ip6fl_get_first(seq); | |
664 | if (fl) | |
665 | while (pos && (fl = ip6fl_get_next(seq, fl)) != NULL) | |
666 | --pos; | |
667 | return pos ? NULL : fl; | |
668 | } | |
669 | ||
670 | static void *ip6fl_seq_start(struct seq_file *seq, loff_t *pos) | |
9a429c49 | 671 | __acquires(ip6_fl_lock) |
1da177e4 LT |
672 | { |
673 | read_lock_bh(&ip6_fl_lock); | |
674 | return *pos ? ip6fl_get_idx(seq, *pos - 1) : SEQ_START_TOKEN; | |
675 | } | |
676 | ||
677 | static void *ip6fl_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
678 | { | |
679 | struct ip6_flowlabel *fl; | |
680 | ||
681 | if (v == SEQ_START_TOKEN) | |
682 | fl = ip6fl_get_first(seq); | |
683 | else | |
684 | fl = ip6fl_get_next(seq, v); | |
685 | ++*pos; | |
686 | return fl; | |
687 | } | |
688 | ||
689 | static void ip6fl_seq_stop(struct seq_file *seq, void *v) | |
9a429c49 | 690 | __releases(ip6_fl_lock) |
1da177e4 LT |
691 | { |
692 | read_unlock_bh(&ip6_fl_lock); | |
693 | } | |
694 | ||
1b7c2dbc | 695 | static int ip6fl_seq_show(struct seq_file *seq, void *v) |
1da177e4 | 696 | { |
1b7c2dbc JM |
697 | if (v == SEQ_START_TOKEN) |
698 | seq_printf(seq, "%-5s %-1s %-6s %-6s %-6s %-8s %-32s %s\n", | |
699 | "Label", "S", "Owner", "Users", "Linger", "Expires", "Dst", "Opt"); | |
700 | else { | |
701 | struct ip6_flowlabel *fl = v; | |
1da177e4 | 702 | seq_printf(seq, |
4b7a4274 | 703 | "%05X %-1d %-6d %-6d %-6ld %-8ld %pi6 %-4d\n", |
1da177e4 LT |
704 | (unsigned)ntohl(fl->label), |
705 | fl->share, | |
706 | (unsigned)fl->owner, | |
707 | atomic_read(&fl->users), | |
708 | fl->linger/HZ, | |
709 | (long)(fl->expires - jiffies)/HZ, | |
b071195d | 710 | &fl->dst, |
1da177e4 | 711 | fl->opt ? fl->opt->opt_nflen : 0); |
1da177e4 | 712 | } |
1da177e4 LT |
713 | return 0; |
714 | } | |
715 | ||
56b3d975 | 716 | static const struct seq_operations ip6fl_seq_ops = { |
1da177e4 LT |
717 | .start = ip6fl_seq_start, |
718 | .next = ip6fl_seq_next, | |
719 | .stop = ip6fl_seq_stop, | |
720 | .show = ip6fl_seq_show, | |
721 | }; | |
722 | ||
723 | static int ip6fl_seq_open(struct inode *inode, struct file *file) | |
724 | { | |
5983a3df BT |
725 | return seq_open_net(inode, file, &ip6fl_seq_ops, |
726 | sizeof(struct ip6fl_iter_state)); | |
1da177e4 LT |
727 | } |
728 | ||
9a32144e | 729 | static const struct file_operations ip6fl_seq_fops = { |
1da177e4 LT |
730 | .owner = THIS_MODULE, |
731 | .open = ip6fl_seq_open, | |
732 | .read = seq_read, | |
733 | .llseek = seq_lseek, | |
5983a3df | 734 | .release = seq_release_net, |
1da177e4 | 735 | }; |
1da177e4 | 736 | |
0a3e78ac DL |
737 | static int ip6_flowlabel_proc_init(struct net *net) |
738 | { | |
5983a3df BT |
739 | if (!proc_net_fops_create(net, "ip6_flowlabel", |
740 | S_IRUGO, &ip6fl_seq_fops)) | |
0a3e78ac DL |
741 | return -ENOMEM; |
742 | return 0; | |
743 | } | |
1da177e4 | 744 | |
0a3e78ac | 745 | static void ip6_flowlabel_proc_fini(struct net *net) |
1da177e4 | 746 | { |
0a3e78ac DL |
747 | proc_net_remove(net, "ip6_flowlabel"); |
748 | } | |
749 | #else | |
750 | static inline int ip6_flowlabel_proc_init(struct net *net) | |
751 | { | |
752 | return 0; | |
753 | } | |
754 | static inline void ip6_flowlabel_proc_fini(struct net *net) | |
755 | { | |
756 | return ; | |
757 | } | |
1da177e4 | 758 | #endif |
0a3e78ac | 759 | |
60e8fbc4 BT |
760 | static inline void ip6_flowlabel_net_exit(struct net *net) |
761 | { | |
762 | ip6_fl_purge(net); | |
5983a3df | 763 | ip6_flowlabel_proc_fini(net); |
60e8fbc4 BT |
764 | } |
765 | ||
766 | static struct pernet_operations ip6_flowlabel_net_ops = { | |
5983a3df | 767 | .init = ip6_flowlabel_proc_init, |
60e8fbc4 BT |
768 | .exit = ip6_flowlabel_net_exit, |
769 | }; | |
770 | ||
0a3e78ac DL |
771 | int ip6_flowlabel_init(void) |
772 | { | |
5983a3df | 773 | return register_pernet_subsys(&ip6_flowlabel_net_ops); |
1da177e4 LT |
774 | } |
775 | ||
776 | void ip6_flowlabel_cleanup(void) | |
777 | { | |
778 | del_timer(&ip6_fl_gc_timer); | |
60e8fbc4 | 779 | unregister_pernet_subsys(&ip6_flowlabel_net_ops); |
1da177e4 | 780 | } |