]>
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> | |
5a0e3ad6 | 23 | #include <linux/slab.h> |
bc3b2d7f | 24 | #include <linux/export.h> |
1da177e4 | 25 | |
457c4cbc | 26 | #include <net/net_namespace.h> |
1da177e4 LT |
27 | #include <net/sock.h> |
28 | ||
29 | #include <net/ipv6.h> | |
30 | #include <net/ndisc.h> | |
31 | #include <net/protocol.h> | |
32 | #include <net/ip6_route.h> | |
33 | #include <net/addrconf.h> | |
34 | #include <net/rawv6.h> | |
35 | #include <net/icmp.h> | |
36 | #include <net/transp_v6.h> | |
37 | ||
38 | #include <asm/uaccess.h> | |
39 | ||
40 | #define FL_MIN_LINGER 6 /* Minimal linger. It is set to 6sec specified | |
41 | in old IPv6 RFC. Well, it was reasonable value. | |
42 | */ | |
43 | #define FL_MAX_LINGER 60 /* Maximal linger timeout */ | |
44 | ||
45 | /* FL hash table */ | |
46 | ||
47 | #define FL_MAX_PER_SOCK 32 | |
48 | #define FL_MAX_SIZE 4096 | |
49 | #define FL_HASH_MASK 255 | |
50 | #define FL_HASH(l) (ntohl(l)&FL_HASH_MASK) | |
51 | ||
52 | static atomic_t fl_size = ATOMIC_INIT(0); | |
53 | static struct ip6_flowlabel *fl_ht[FL_HASH_MASK+1]; | |
54 | ||
55 | static void ip6_fl_gc(unsigned long dummy); | |
8d06afab | 56 | static DEFINE_TIMER(ip6_fl_gc_timer, ip6_fl_gc, 0, 0); |
1da177e4 LT |
57 | |
58 | /* FL hash table lock: it protects only of GC */ | |
59 | ||
60 | static DEFINE_RWLOCK(ip6_fl_lock); | |
61 | ||
62 | /* Big socket sock */ | |
63 | ||
64 | static DEFINE_RWLOCK(ip6_sk_fl_lock); | |
65 | ||
66 | ||
60e8fbc4 | 67 | static inline struct ip6_flowlabel *__fl_lookup(struct net *net, __be32 label) |
1da177e4 LT |
68 | { |
69 | struct ip6_flowlabel *fl; | |
70 | ||
71 | for (fl=fl_ht[FL_HASH(label)]; fl; fl = fl->next) { | |
09ad9bc7 | 72 | if (fl->label == label && net_eq(fl->fl_net, net)) |
1da177e4 LT |
73 | return fl; |
74 | } | |
75 | return NULL; | |
76 | } | |
77 | ||
60e8fbc4 | 78 | static struct ip6_flowlabel *fl_lookup(struct net *net, __be32 label) |
1da177e4 LT |
79 | { |
80 | struct ip6_flowlabel *fl; | |
81 | ||
82 | read_lock_bh(&ip6_fl_lock); | |
60e8fbc4 | 83 | fl = __fl_lookup(net, label); |
1da177e4 LT |
84 | if (fl) |
85 | atomic_inc(&fl->users); | |
86 | read_unlock_bh(&ip6_fl_lock); | |
87 | return fl; | |
88 | } | |
89 | ||
90 | ||
91 | static void fl_free(struct ip6_flowlabel *fl) | |
92 | { | |
60e8fbc4 BT |
93 | if (fl) { |
94 | release_net(fl->fl_net); | |
1da177e4 | 95 | kfree(fl->opt); |
60e8fbc4 | 96 | } |
1da177e4 LT |
97 | kfree(fl); |
98 | } | |
99 | ||
100 | static void fl_release(struct ip6_flowlabel *fl) | |
101 | { | |
102 | write_lock_bh(&ip6_fl_lock); | |
103 | ||
104 | fl->lastuse = jiffies; | |
105 | if (atomic_dec_and_test(&fl->users)) { | |
106 | unsigned long ttd = fl->lastuse + fl->linger; | |
107 | if (time_after(ttd, fl->expires)) | |
108 | fl->expires = ttd; | |
109 | ttd = fl->expires; | |
110 | if (fl->opt && fl->share == IPV6_FL_S_EXCL) { | |
111 | struct ipv6_txoptions *opt = fl->opt; | |
112 | fl->opt = NULL; | |
113 | kfree(opt); | |
114 | } | |
115 | if (!timer_pending(&ip6_fl_gc_timer) || | |
116 | time_after(ip6_fl_gc_timer.expires, ttd)) | |
117 | mod_timer(&ip6_fl_gc_timer, ttd); | |
118 | } | |
1da177e4 LT |
119 | write_unlock_bh(&ip6_fl_lock); |
120 | } | |
121 | ||
122 | static void ip6_fl_gc(unsigned long dummy) | |
123 | { | |
124 | int i; | |
125 | unsigned long now = jiffies; | |
126 | unsigned long sched = 0; | |
127 | ||
128 | write_lock(&ip6_fl_lock); | |
129 | ||
130 | for (i=0; i<=FL_HASH_MASK; i++) { | |
131 | struct ip6_flowlabel *fl, **flp; | |
132 | flp = &fl_ht[i]; | |
133 | while ((fl=*flp) != NULL) { | |
134 | if (atomic_read(&fl->users) == 0) { | |
135 | unsigned long ttd = fl->lastuse + fl->linger; | |
136 | if (time_after(ttd, fl->expires)) | |
137 | fl->expires = ttd; | |
138 | ttd = fl->expires; | |
139 | if (time_after_eq(now, ttd)) { | |
140 | *flp = fl->next; | |
141 | fl_free(fl); | |
142 | atomic_dec(&fl_size); | |
143 | continue; | |
144 | } | |
145 | if (!sched || time_before(ttd, sched)) | |
146 | sched = ttd; | |
147 | } | |
148 | flp = &fl->next; | |
149 | } | |
150 | } | |
151 | if (!sched && atomic_read(&fl_size)) | |
152 | sched = now + FL_MAX_LINGER; | |
153 | if (sched) { | |
60e8fbc4 | 154 | mod_timer(&ip6_fl_gc_timer, sched); |
1da177e4 LT |
155 | } |
156 | write_unlock(&ip6_fl_lock); | |
157 | } | |
158 | ||
2c8c1e72 | 159 | static void __net_exit ip6_fl_purge(struct net *net) |
60e8fbc4 BT |
160 | { |
161 | int i; | |
162 | ||
163 | write_lock(&ip6_fl_lock); | |
164 | for (i = 0; i <= FL_HASH_MASK; i++) { | |
165 | struct ip6_flowlabel *fl, **flp; | |
166 | flp = &fl_ht[i]; | |
167 | while ((fl = *flp) != NULL) { | |
09ad9bc7 OP |
168 | if (net_eq(fl->fl_net, net) && |
169 | atomic_read(&fl->users) == 0) { | |
60e8fbc4 BT |
170 | *flp = fl->next; |
171 | fl_free(fl); | |
172 | atomic_dec(&fl_size); | |
173 | continue; | |
174 | } | |
175 | flp = &fl->next; | |
176 | } | |
177 | } | |
178 | write_unlock(&ip6_fl_lock); | |
179 | } | |
180 | ||
181 | static struct ip6_flowlabel *fl_intern(struct net *net, | |
182 | struct ip6_flowlabel *fl, __be32 label) | |
1da177e4 | 183 | { |
78c2e502 PE |
184 | struct ip6_flowlabel *lfl; |
185 | ||
1da177e4 LT |
186 | fl->label = label & IPV6_FLOWLABEL_MASK; |
187 | ||
188 | write_lock_bh(&ip6_fl_lock); | |
189 | if (label == 0) { | |
190 | for (;;) { | |
191 | fl->label = htonl(net_random())&IPV6_FLOWLABEL_MASK; | |
192 | if (fl->label) { | |
60e8fbc4 | 193 | lfl = __fl_lookup(net, fl->label); |
1da177e4 LT |
194 | if (lfl == NULL) |
195 | break; | |
196 | } | |
197 | } | |
78c2e502 PE |
198 | } else { |
199 | /* | |
200 | * we dropper the ip6_fl_lock, so this entry could reappear | |
201 | * and we need to recheck with it. | |
202 | * | |
203 | * OTOH no need to search the active socket first, like it is | |
204 | * done in ipv6_flowlabel_opt - sock is locked, so new entry | |
205 | * with the same label can only appear on another sock | |
206 | */ | |
60e8fbc4 | 207 | lfl = __fl_lookup(net, fl->label); |
78c2e502 PE |
208 | if (lfl != NULL) { |
209 | atomic_inc(&lfl->users); | |
210 | write_unlock_bh(&ip6_fl_lock); | |
211 | return lfl; | |
212 | } | |
1da177e4 LT |
213 | } |
214 | ||
215 | fl->lastuse = jiffies; | |
216 | fl->next = fl_ht[FL_HASH(fl->label)]; | |
217 | fl_ht[FL_HASH(fl->label)] = fl; | |
218 | atomic_inc(&fl_size); | |
219 | write_unlock_bh(&ip6_fl_lock); | |
78c2e502 | 220 | return NULL; |
1da177e4 LT |
221 | } |
222 | ||
223 | ||
224 | ||
225 | /* Socket flowlabel lists */ | |
226 | ||
90bcaf7b | 227 | struct ip6_flowlabel * fl6_sock_lookup(struct sock *sk, __be32 label) |
1da177e4 LT |
228 | { |
229 | struct ipv6_fl_socklist *sfl; | |
230 | struct ipv6_pinfo *np = inet6_sk(sk); | |
231 | ||
232 | label &= IPV6_FLOWLABEL_MASK; | |
233 | ||
bd0bf577 | 234 | read_lock_bh(&ip6_sk_fl_lock); |
1da177e4 LT |
235 | for (sfl=np->ipv6_fl_list; sfl; sfl = sfl->next) { |
236 | struct ip6_flowlabel *fl = sfl->fl; | |
237 | if (fl->label == label) { | |
238 | fl->lastuse = jiffies; | |
239 | atomic_inc(&fl->users); | |
52f095ee | 240 | read_unlock_bh(&ip6_sk_fl_lock); |
1da177e4 LT |
241 | return fl; |
242 | } | |
243 | } | |
bd0bf577 | 244 | read_unlock_bh(&ip6_sk_fl_lock); |
1da177e4 LT |
245 | return NULL; |
246 | } | |
247 | ||
3cf3dc6c ACM |
248 | EXPORT_SYMBOL_GPL(fl6_sock_lookup); |
249 | ||
1da177e4 LT |
250 | void fl6_free_socklist(struct sock *sk) |
251 | { | |
252 | struct ipv6_pinfo *np = inet6_sk(sk); | |
253 | struct ipv6_fl_socklist *sfl; | |
254 | ||
255 | while ((sfl = np->ipv6_fl_list) != NULL) { | |
256 | np->ipv6_fl_list = sfl->next; | |
257 | fl_release(sfl->fl); | |
258 | kfree(sfl); | |
259 | } | |
260 | } | |
261 | ||
262 | /* Service routines */ | |
263 | ||
264 | ||
265 | /* | |
266 | It is the only difficult place. flowlabel enforces equal headers | |
267 | before and including routing header, however user may supply options | |
268 | following rthdr. | |
269 | */ | |
270 | ||
271 | struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions * opt_space, | |
272 | struct ip6_flowlabel * fl, | |
273 | struct ipv6_txoptions * fopt) | |
274 | { | |
df9890c3 | 275 | struct ipv6_txoptions * fl_opt = fl->opt; |
1ab1457c | 276 | |
df9890c3 YH |
277 | if (fopt == NULL || fopt->opt_flen == 0) |
278 | return fl_opt; | |
1ab1457c | 279 | |
1da177e4 LT |
280 | if (fl_opt != NULL) { |
281 | opt_space->hopopt = fl_opt->hopopt; | |
df9890c3 | 282 | opt_space->dst0opt = fl_opt->dst0opt; |
1da177e4 LT |
283 | opt_space->srcrt = fl_opt->srcrt; |
284 | opt_space->opt_nflen = fl_opt->opt_nflen; | |
285 | } else { | |
286 | if (fopt->opt_nflen == 0) | |
287 | return fopt; | |
288 | opt_space->hopopt = NULL; | |
289 | opt_space->dst0opt = NULL; | |
290 | opt_space->srcrt = NULL; | |
291 | opt_space->opt_nflen = 0; | |
292 | } | |
293 | opt_space->dst1opt = fopt->dst1opt; | |
1da177e4 LT |
294 | opt_space->opt_flen = fopt->opt_flen; |
295 | return opt_space; | |
296 | } | |
a495f836 | 297 | EXPORT_SYMBOL_GPL(fl6_merge_options); |
1da177e4 LT |
298 | |
299 | static unsigned long check_linger(unsigned long ttl) | |
300 | { | |
301 | if (ttl < FL_MIN_LINGER) | |
302 | return FL_MIN_LINGER*HZ; | |
303 | if (ttl > FL_MAX_LINGER && !capable(CAP_NET_ADMIN)) | |
304 | return 0; | |
305 | return ttl*HZ; | |
306 | } | |
307 | ||
308 | static int fl6_renew(struct ip6_flowlabel *fl, unsigned long linger, unsigned long expires) | |
309 | { | |
310 | linger = check_linger(linger); | |
311 | if (!linger) | |
312 | return -EPERM; | |
313 | expires = check_linger(expires); | |
314 | if (!expires) | |
315 | return -EPERM; | |
316 | fl->lastuse = jiffies; | |
317 | if (time_before(fl->linger, linger)) | |
318 | fl->linger = linger; | |
319 | if (time_before(expires, fl->linger)) | |
320 | expires = fl->linger; | |
321 | if (time_before(fl->expires, fl->lastuse + expires)) | |
322 | fl->expires = fl->lastuse + expires; | |
323 | return 0; | |
324 | } | |
325 | ||
326 | static struct ip6_flowlabel * | |
ec0506db MÅ» |
327 | fl_create(struct net *net, struct sock *sk, struct in6_flowlabel_req *freq, |
328 | char __user *optval, int optlen, int *err_p) | |
1da177e4 | 329 | { |
684de409 | 330 | struct ip6_flowlabel *fl = NULL; |
1da177e4 LT |
331 | int olen; |
332 | int addr_type; | |
333 | int err; | |
334 | ||
684de409 DM |
335 | olen = optlen - CMSG_ALIGN(sizeof(*freq)); |
336 | err = -EINVAL; | |
337 | if (olen > 64 * 1024) | |
338 | goto done; | |
339 | ||
1da177e4 | 340 | err = -ENOMEM; |
0c600eda | 341 | fl = kzalloc(sizeof(*fl), GFP_KERNEL); |
1da177e4 LT |
342 | if (fl == NULL) |
343 | goto done; | |
1da177e4 | 344 | |
1da177e4 LT |
345 | if (olen > 0) { |
346 | struct msghdr msg; | |
4c9483b2 | 347 | struct flowi6 flowi6; |
1da177e4 LT |
348 | int junk; |
349 | ||
350 | err = -ENOMEM; | |
351 | fl->opt = kmalloc(sizeof(*fl->opt) + olen, GFP_KERNEL); | |
352 | if (fl->opt == NULL) | |
353 | goto done; | |
354 | ||
355 | memset(fl->opt, 0, sizeof(*fl->opt)); | |
356 | fl->opt->tot_len = sizeof(*fl->opt) + olen; | |
357 | err = -EFAULT; | |
358 | if (copy_from_user(fl->opt+1, optval+CMSG_ALIGN(sizeof(*freq)), olen)) | |
359 | goto done; | |
360 | ||
361 | msg.msg_controllen = olen; | |
362 | msg.msg_control = (void*)(fl->opt+1); | |
4c9483b2 | 363 | memset(&flowi6, 0, sizeof(flowi6)); |
1da177e4 | 364 | |
ec0506db | 365 | err = datagram_send_ctl(net, sk, &msg, &flowi6, fl->opt, &junk, |
13b52cd4 | 366 | &junk, &junk); |
1da177e4 LT |
367 | if (err) |
368 | goto done; | |
369 | err = -EINVAL; | |
370 | if (fl->opt->opt_flen) | |
371 | goto done; | |
372 | if (fl->opt->opt_nflen == 0) { | |
373 | kfree(fl->opt); | |
374 | fl->opt = NULL; | |
375 | } | |
376 | } | |
377 | ||
60e8fbc4 | 378 | fl->fl_net = hold_net(net); |
1da177e4 LT |
379 | fl->expires = jiffies; |
380 | err = fl6_renew(fl, freq->flr_linger, freq->flr_expires); | |
381 | if (err) | |
382 | goto done; | |
383 | fl->share = freq->flr_share; | |
384 | addr_type = ipv6_addr_type(&freq->flr_dst); | |
35700212 JP |
385 | if ((addr_type & IPV6_ADDR_MAPPED) || |
386 | addr_type == IPV6_ADDR_ANY) { | |
c6817e4c | 387 | err = -EINVAL; |
1da177e4 | 388 | goto done; |
c6817e4c | 389 | } |
4e3fd7a0 | 390 | fl->dst = freq->flr_dst; |
1da177e4 LT |
391 | atomic_set(&fl->users, 1); |
392 | switch (fl->share) { | |
393 | case IPV6_FL_S_EXCL: | |
394 | case IPV6_FL_S_ANY: | |
395 | break; | |
396 | case IPV6_FL_S_PROCESS: | |
397 | fl->owner = current->pid; | |
398 | break; | |
399 | case IPV6_FL_S_USER: | |
f82b3590 | 400 | fl->owner = current_euid(); |
1da177e4 LT |
401 | break; |
402 | default: | |
403 | err = -EINVAL; | |
404 | goto done; | |
405 | } | |
406 | return fl; | |
407 | ||
408 | done: | |
409 | fl_free(fl); | |
410 | *err_p = err; | |
411 | return NULL; | |
412 | } | |
413 | ||
414 | static int mem_check(struct sock *sk) | |
415 | { | |
416 | struct ipv6_pinfo *np = inet6_sk(sk); | |
417 | struct ipv6_fl_socklist *sfl; | |
418 | int room = FL_MAX_SIZE - atomic_read(&fl_size); | |
419 | int count = 0; | |
420 | ||
421 | if (room > FL_MAX_SIZE - FL_MAX_PER_SOCK) | |
422 | return 0; | |
423 | ||
424 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) | |
425 | count++; | |
426 | ||
427 | if (room <= 0 || | |
428 | ((count >= FL_MAX_PER_SOCK || | |
35700212 JP |
429 | (count > 0 && room < FL_MAX_SIZE/2) || room < FL_MAX_SIZE/4) && |
430 | !capable(CAP_NET_ADMIN))) | |
1da177e4 LT |
431 | return -ENOBUFS; |
432 | ||
433 | return 0; | |
434 | } | |
435 | ||
a50feda5 | 436 | static bool ipv6_hdr_cmp(struct ipv6_opt_hdr *h1, struct ipv6_opt_hdr *h2) |
1da177e4 LT |
437 | { |
438 | if (h1 == h2) | |
a50feda5 | 439 | return false; |
1da177e4 | 440 | if (h1 == NULL || h2 == NULL) |
a50feda5 | 441 | return true; |
1da177e4 | 442 | if (h1->hdrlen != h2->hdrlen) |
a50feda5 | 443 | return true; |
1da177e4 LT |
444 | return memcmp(h1+1, h2+1, ((h1->hdrlen+1)<<3) - sizeof(*h1)); |
445 | } | |
446 | ||
a50feda5 | 447 | static bool ipv6_opt_cmp(struct ipv6_txoptions *o1, struct ipv6_txoptions *o2) |
1da177e4 LT |
448 | { |
449 | if (o1 == o2) | |
a50feda5 | 450 | return false; |
1da177e4 | 451 | if (o1 == NULL || o2 == NULL) |
a50feda5 | 452 | return true; |
1da177e4 | 453 | if (o1->opt_nflen != o2->opt_nflen) |
a50feda5 | 454 | return true; |
1da177e4 | 455 | if (ipv6_hdr_cmp(o1->hopopt, o2->hopopt)) |
a50feda5 | 456 | return true; |
1da177e4 | 457 | if (ipv6_hdr_cmp(o1->dst0opt, o2->dst0opt)) |
a50feda5 | 458 | return true; |
1da177e4 | 459 | if (ipv6_hdr_cmp((struct ipv6_opt_hdr *)o1->srcrt, (struct ipv6_opt_hdr *)o2->srcrt)) |
a50feda5 ED |
460 | return true; |
461 | return false; | |
1da177e4 LT |
462 | } |
463 | ||
04028045 PE |
464 | static inline void fl_link(struct ipv6_pinfo *np, struct ipv6_fl_socklist *sfl, |
465 | struct ip6_flowlabel *fl) | |
466 | { | |
467 | write_lock_bh(&ip6_sk_fl_lock); | |
468 | sfl->fl = fl; | |
469 | sfl->next = np->ipv6_fl_list; | |
470 | np->ipv6_fl_list = sfl; | |
471 | write_unlock_bh(&ip6_sk_fl_lock); | |
472 | } | |
473 | ||
1da177e4 LT |
474 | int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen) |
475 | { | |
55205d40 | 476 | int uninitialized_var(err); |
60e8fbc4 | 477 | struct net *net = sock_net(sk); |
1da177e4 LT |
478 | struct ipv6_pinfo *np = inet6_sk(sk); |
479 | struct in6_flowlabel_req freq; | |
480 | struct ipv6_fl_socklist *sfl1=NULL; | |
481 | struct ipv6_fl_socklist *sfl, **sflp; | |
78c2e502 PE |
482 | struct ip6_flowlabel *fl, *fl1 = NULL; |
483 | ||
1da177e4 LT |
484 | |
485 | if (optlen < sizeof(freq)) | |
486 | return -EINVAL; | |
487 | ||
488 | if (copy_from_user(&freq, optval, sizeof(freq))) | |
489 | return -EFAULT; | |
490 | ||
491 | switch (freq.flr_action) { | |
492 | case IPV6_FL_A_PUT: | |
493 | write_lock_bh(&ip6_sk_fl_lock); | |
494 | for (sflp = &np->ipv6_fl_list; (sfl=*sflp)!=NULL; sflp = &sfl->next) { | |
495 | if (sfl->fl->label == freq.flr_label) { | |
496 | if (freq.flr_label == (np->flow_label&IPV6_FLOWLABEL_MASK)) | |
497 | np->flow_label &= ~IPV6_FLOWLABEL_MASK; | |
498 | *sflp = sfl->next; | |
499 | write_unlock_bh(&ip6_sk_fl_lock); | |
500 | fl_release(sfl->fl); | |
501 | kfree(sfl); | |
502 | return 0; | |
503 | } | |
504 | } | |
505 | write_unlock_bh(&ip6_sk_fl_lock); | |
506 | return -ESRCH; | |
507 | ||
508 | case IPV6_FL_A_RENEW: | |
509 | read_lock_bh(&ip6_sk_fl_lock); | |
510 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) { | |
511 | if (sfl->fl->label == freq.flr_label) { | |
512 | err = fl6_renew(sfl->fl, freq.flr_linger, freq.flr_expires); | |
513 | read_unlock_bh(&ip6_sk_fl_lock); | |
514 | return err; | |
515 | } | |
516 | } | |
517 | read_unlock_bh(&ip6_sk_fl_lock); | |
518 | ||
519 | if (freq.flr_share == IPV6_FL_S_NONE && capable(CAP_NET_ADMIN)) { | |
60e8fbc4 | 520 | fl = fl_lookup(net, freq.flr_label); |
1da177e4 LT |
521 | if (fl) { |
522 | err = fl6_renew(fl, freq.flr_linger, freq.flr_expires); | |
523 | fl_release(fl); | |
524 | return err; | |
525 | } | |
526 | } | |
527 | return -ESRCH; | |
528 | ||
529 | case IPV6_FL_A_GET: | |
530 | if (freq.flr_label & ~IPV6_FLOWLABEL_MASK) | |
531 | return -EINVAL; | |
532 | ||
ec0506db | 533 | fl = fl_create(net, sk, &freq, optval, optlen, &err); |
1da177e4 LT |
534 | if (fl == NULL) |
535 | return err; | |
536 | sfl1 = kmalloc(sizeof(*sfl1), GFP_KERNEL); | |
537 | ||
538 | if (freq.flr_label) { | |
1da177e4 LT |
539 | err = -EEXIST; |
540 | read_lock_bh(&ip6_sk_fl_lock); | |
541 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) { | |
542 | if (sfl->fl->label == freq.flr_label) { | |
543 | if (freq.flr_flags&IPV6_FL_F_EXCL) { | |
544 | read_unlock_bh(&ip6_sk_fl_lock); | |
545 | goto done; | |
546 | } | |
547 | fl1 = sfl->fl; | |
4ea6a804 | 548 | atomic_inc(&fl1->users); |
1da177e4 LT |
549 | break; |
550 | } | |
551 | } | |
552 | read_unlock_bh(&ip6_sk_fl_lock); | |
553 | ||
554 | if (fl1 == NULL) | |
60e8fbc4 | 555 | fl1 = fl_lookup(net, freq.flr_label); |
1da177e4 | 556 | if (fl1) { |
78c2e502 | 557 | recheck: |
1da177e4 LT |
558 | err = -EEXIST; |
559 | if (freq.flr_flags&IPV6_FL_F_EXCL) | |
560 | goto release; | |
561 | err = -EPERM; | |
562 | if (fl1->share == IPV6_FL_S_EXCL || | |
563 | fl1->share != fl->share || | |
564 | fl1->owner != fl->owner) | |
565 | goto release; | |
566 | ||
567 | err = -EINVAL; | |
568 | if (!ipv6_addr_equal(&fl1->dst, &fl->dst) || | |
569 | ipv6_opt_cmp(fl1->opt, fl->opt)) | |
570 | goto release; | |
571 | ||
572 | err = -ENOMEM; | |
573 | if (sfl1 == NULL) | |
574 | goto release; | |
575 | if (fl->linger > fl1->linger) | |
576 | fl1->linger = fl->linger; | |
577 | if ((long)(fl->expires - fl1->expires) > 0) | |
578 | fl1->expires = fl->expires; | |
04028045 | 579 | fl_link(np, sfl1, fl1); |
1da177e4 LT |
580 | fl_free(fl); |
581 | return 0; | |
582 | ||
583 | release: | |
584 | fl_release(fl1); | |
585 | goto done; | |
586 | } | |
587 | } | |
588 | err = -ENOENT; | |
589 | if (!(freq.flr_flags&IPV6_FL_F_CREATE)) | |
590 | goto done; | |
591 | ||
592 | err = -ENOMEM; | |
593 | if (sfl1 == NULL || (err = mem_check(sk)) != 0) | |
594 | goto done; | |
595 | ||
60e8fbc4 | 596 | fl1 = fl_intern(net, fl, freq.flr_label); |
78c2e502 PE |
597 | if (fl1 != NULL) |
598 | goto recheck; | |
1da177e4 | 599 | |
6c94d361 DM |
600 | if (!freq.flr_label) { |
601 | if (copy_to_user(&((struct in6_flowlabel_req __user *) optval)->flr_label, | |
602 | &fl->label, sizeof(fl->label))) { | |
603 | /* Intentionally ignore fault. */ | |
604 | } | |
605 | } | |
1da177e4 | 606 | |
04028045 | 607 | fl_link(np, sfl1, fl); |
1da177e4 LT |
608 | return 0; |
609 | ||
610 | default: | |
611 | return -EINVAL; | |
612 | } | |
613 | ||
614 | done: | |
615 | fl_free(fl); | |
616 | kfree(sfl1); | |
617 | return err; | |
618 | } | |
619 | ||
620 | #ifdef CONFIG_PROC_FS | |
621 | ||
622 | struct ip6fl_iter_state { | |
5983a3df | 623 | struct seq_net_private p; |
1da177e4 LT |
624 | int bucket; |
625 | }; | |
626 | ||
627 | #define ip6fl_seq_private(seq) ((struct ip6fl_iter_state *)(seq)->private) | |
628 | ||
629 | static struct ip6_flowlabel *ip6fl_get_first(struct seq_file *seq) | |
630 | { | |
631 | struct ip6_flowlabel *fl = NULL; | |
632 | struct ip6fl_iter_state *state = ip6fl_seq_private(seq); | |
5983a3df | 633 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
634 | |
635 | for (state->bucket = 0; state->bucket <= FL_HASH_MASK; ++state->bucket) { | |
5983a3df BT |
636 | fl = fl_ht[state->bucket]; |
637 | ||
09ad9bc7 | 638 | while (fl && !net_eq(fl->fl_net, net)) |
5983a3df BT |
639 | fl = fl->next; |
640 | if (fl) | |
1da177e4 | 641 | break; |
1da177e4 LT |
642 | } |
643 | return fl; | |
644 | } | |
645 | ||
646 | static struct ip6_flowlabel *ip6fl_get_next(struct seq_file *seq, struct ip6_flowlabel *fl) | |
647 | { | |
648 | struct ip6fl_iter_state *state = ip6fl_seq_private(seq); | |
5983a3df | 649 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
650 | |
651 | fl = fl->next; | |
5983a3df | 652 | try_again: |
09ad9bc7 | 653 | while (fl && !net_eq(fl->fl_net, net)) |
5983a3df BT |
654 | fl = fl->next; |
655 | ||
1da177e4 | 656 | while (!fl) { |
5983a3df | 657 | if (++state->bucket <= FL_HASH_MASK) { |
1da177e4 | 658 | fl = fl_ht[state->bucket]; |
5983a3df BT |
659 | goto try_again; |
660 | } else | |
bcd62075 | 661 | break; |
1da177e4 LT |
662 | } |
663 | return fl; | |
664 | } | |
665 | ||
666 | static struct ip6_flowlabel *ip6fl_get_idx(struct seq_file *seq, loff_t pos) | |
667 | { | |
668 | struct ip6_flowlabel *fl = ip6fl_get_first(seq); | |
669 | if (fl) | |
670 | while (pos && (fl = ip6fl_get_next(seq, fl)) != NULL) | |
671 | --pos; | |
672 | return pos ? NULL : fl; | |
673 | } | |
674 | ||
675 | static void *ip6fl_seq_start(struct seq_file *seq, loff_t *pos) | |
9a429c49 | 676 | __acquires(ip6_fl_lock) |
1da177e4 LT |
677 | { |
678 | read_lock_bh(&ip6_fl_lock); | |
679 | return *pos ? ip6fl_get_idx(seq, *pos - 1) : SEQ_START_TOKEN; | |
680 | } | |
681 | ||
682 | static void *ip6fl_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
683 | { | |
684 | struct ip6_flowlabel *fl; | |
685 | ||
686 | if (v == SEQ_START_TOKEN) | |
687 | fl = ip6fl_get_first(seq); | |
688 | else | |
689 | fl = ip6fl_get_next(seq, v); | |
690 | ++*pos; | |
691 | return fl; | |
692 | } | |
693 | ||
694 | static void ip6fl_seq_stop(struct seq_file *seq, void *v) | |
9a429c49 | 695 | __releases(ip6_fl_lock) |
1da177e4 LT |
696 | { |
697 | read_unlock_bh(&ip6_fl_lock); | |
698 | } | |
699 | ||
1b7c2dbc | 700 | static int ip6fl_seq_show(struct seq_file *seq, void *v) |
1da177e4 | 701 | { |
1b7c2dbc JM |
702 | if (v == SEQ_START_TOKEN) |
703 | seq_printf(seq, "%-5s %-1s %-6s %-6s %-6s %-8s %-32s %s\n", | |
704 | "Label", "S", "Owner", "Users", "Linger", "Expires", "Dst", "Opt"); | |
705 | else { | |
706 | struct ip6_flowlabel *fl = v; | |
1da177e4 | 707 | seq_printf(seq, |
4b7a4274 | 708 | "%05X %-1d %-6d %-6d %-6ld %-8ld %pi6 %-4d\n", |
95c96174 | 709 | (unsigned int)ntohl(fl->label), |
1da177e4 | 710 | fl->share, |
95c96174 | 711 | (int)fl->owner, |
1da177e4 LT |
712 | atomic_read(&fl->users), |
713 | fl->linger/HZ, | |
714 | (long)(fl->expires - jiffies)/HZ, | |
b071195d | 715 | &fl->dst, |
1da177e4 | 716 | fl->opt ? fl->opt->opt_nflen : 0); |
1da177e4 | 717 | } |
1da177e4 LT |
718 | return 0; |
719 | } | |
720 | ||
56b3d975 | 721 | static const struct seq_operations ip6fl_seq_ops = { |
1da177e4 LT |
722 | .start = ip6fl_seq_start, |
723 | .next = ip6fl_seq_next, | |
724 | .stop = ip6fl_seq_stop, | |
725 | .show = ip6fl_seq_show, | |
726 | }; | |
727 | ||
728 | static int ip6fl_seq_open(struct inode *inode, struct file *file) | |
729 | { | |
5983a3df BT |
730 | return seq_open_net(inode, file, &ip6fl_seq_ops, |
731 | sizeof(struct ip6fl_iter_state)); | |
1da177e4 LT |
732 | } |
733 | ||
9a32144e | 734 | static const struct file_operations ip6fl_seq_fops = { |
1da177e4 LT |
735 | .owner = THIS_MODULE, |
736 | .open = ip6fl_seq_open, | |
737 | .read = seq_read, | |
738 | .llseek = seq_lseek, | |
5983a3df | 739 | .release = seq_release_net, |
1da177e4 | 740 | }; |
1da177e4 | 741 | |
2c8c1e72 | 742 | static int __net_init ip6_flowlabel_proc_init(struct net *net) |
0a3e78ac | 743 | { |
5983a3df BT |
744 | if (!proc_net_fops_create(net, "ip6_flowlabel", |
745 | S_IRUGO, &ip6fl_seq_fops)) | |
0a3e78ac DL |
746 | return -ENOMEM; |
747 | return 0; | |
748 | } | |
1da177e4 | 749 | |
2c8c1e72 | 750 | static void __net_exit ip6_flowlabel_proc_fini(struct net *net) |
1da177e4 | 751 | { |
0a3e78ac DL |
752 | proc_net_remove(net, "ip6_flowlabel"); |
753 | } | |
754 | #else | |
755 | static inline int ip6_flowlabel_proc_init(struct net *net) | |
756 | { | |
757 | return 0; | |
758 | } | |
759 | static inline void ip6_flowlabel_proc_fini(struct net *net) | |
760 | { | |
0a3e78ac | 761 | } |
1da177e4 | 762 | #endif |
0a3e78ac | 763 | |
2c8c1e72 | 764 | static void __net_exit ip6_flowlabel_net_exit(struct net *net) |
60e8fbc4 BT |
765 | { |
766 | ip6_fl_purge(net); | |
5983a3df | 767 | ip6_flowlabel_proc_fini(net); |
60e8fbc4 BT |
768 | } |
769 | ||
770 | static struct pernet_operations ip6_flowlabel_net_ops = { | |
5983a3df | 771 | .init = ip6_flowlabel_proc_init, |
60e8fbc4 BT |
772 | .exit = ip6_flowlabel_net_exit, |
773 | }; | |
774 | ||
0a3e78ac DL |
775 | int ip6_flowlabel_init(void) |
776 | { | |
5983a3df | 777 | return register_pernet_subsys(&ip6_flowlabel_net_ops); |
1da177e4 LT |
778 | } |
779 | ||
780 | void ip6_flowlabel_cleanup(void) | |
781 | { | |
782 | del_timer(&ip6_fl_gc_timer); | |
60e8fbc4 | 783 | unregister_pernet_subsys(&ip6_flowlabel_net_ops); |
1da177e4 | 784 | } |