]>
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 | } | |
297 | ||
298 | static unsigned long check_linger(unsigned long ttl) | |
299 | { | |
300 | if (ttl < FL_MIN_LINGER) | |
301 | return FL_MIN_LINGER*HZ; | |
302 | if (ttl > FL_MAX_LINGER && !capable(CAP_NET_ADMIN)) | |
303 | return 0; | |
304 | return ttl*HZ; | |
305 | } | |
306 | ||
307 | static int fl6_renew(struct ip6_flowlabel *fl, unsigned long linger, unsigned long expires) | |
308 | { | |
309 | linger = check_linger(linger); | |
310 | if (!linger) | |
311 | return -EPERM; | |
312 | expires = check_linger(expires); | |
313 | if (!expires) | |
314 | return -EPERM; | |
315 | fl->lastuse = jiffies; | |
316 | if (time_before(fl->linger, linger)) | |
317 | fl->linger = linger; | |
318 | if (time_before(expires, fl->linger)) | |
319 | expires = fl->linger; | |
320 | if (time_before(fl->expires, fl->lastuse + expires)) | |
321 | fl->expires = fl->lastuse + expires; | |
322 | return 0; | |
323 | } | |
324 | ||
325 | static struct ip6_flowlabel * | |
ec0506db MÅ» |
326 | fl_create(struct net *net, struct sock *sk, struct in6_flowlabel_req *freq, |
327 | char __user *optval, int optlen, int *err_p) | |
1da177e4 | 328 | { |
684de409 | 329 | struct ip6_flowlabel *fl = NULL; |
1da177e4 LT |
330 | int olen; |
331 | int addr_type; | |
332 | int err; | |
333 | ||
684de409 DM |
334 | olen = optlen - CMSG_ALIGN(sizeof(*freq)); |
335 | err = -EINVAL; | |
336 | if (olen > 64 * 1024) | |
337 | goto done; | |
338 | ||
1da177e4 | 339 | err = -ENOMEM; |
0c600eda | 340 | fl = kzalloc(sizeof(*fl), GFP_KERNEL); |
1da177e4 LT |
341 | if (fl == NULL) |
342 | goto done; | |
1da177e4 | 343 | |
1da177e4 LT |
344 | if (olen > 0) { |
345 | struct msghdr msg; | |
4c9483b2 | 346 | struct flowi6 flowi6; |
1da177e4 LT |
347 | int junk; |
348 | ||
349 | err = -ENOMEM; | |
350 | fl->opt = kmalloc(sizeof(*fl->opt) + olen, GFP_KERNEL); | |
351 | if (fl->opt == NULL) | |
352 | goto done; | |
353 | ||
354 | memset(fl->opt, 0, sizeof(*fl->opt)); | |
355 | fl->opt->tot_len = sizeof(*fl->opt) + olen; | |
356 | err = -EFAULT; | |
357 | if (copy_from_user(fl->opt+1, optval+CMSG_ALIGN(sizeof(*freq)), olen)) | |
358 | goto done; | |
359 | ||
360 | msg.msg_controllen = olen; | |
361 | msg.msg_control = (void*)(fl->opt+1); | |
4c9483b2 | 362 | memset(&flowi6, 0, sizeof(flowi6)); |
1da177e4 | 363 | |
ec0506db | 364 | err = datagram_send_ctl(net, sk, &msg, &flowi6, fl->opt, &junk, |
13b52cd4 | 365 | &junk, &junk); |
1da177e4 LT |
366 | if (err) |
367 | goto done; | |
368 | err = -EINVAL; | |
369 | if (fl->opt->opt_flen) | |
370 | goto done; | |
371 | if (fl->opt->opt_nflen == 0) { | |
372 | kfree(fl->opt); | |
373 | fl->opt = NULL; | |
374 | } | |
375 | } | |
376 | ||
60e8fbc4 | 377 | fl->fl_net = hold_net(net); |
1da177e4 LT |
378 | fl->expires = jiffies; |
379 | err = fl6_renew(fl, freq->flr_linger, freq->flr_expires); | |
380 | if (err) | |
381 | goto done; | |
382 | fl->share = freq->flr_share; | |
383 | addr_type = ipv6_addr_type(&freq->flr_dst); | |
35700212 JP |
384 | if ((addr_type & IPV6_ADDR_MAPPED) || |
385 | addr_type == IPV6_ADDR_ANY) { | |
c6817e4c | 386 | err = -EINVAL; |
1da177e4 | 387 | goto done; |
c6817e4c | 388 | } |
4e3fd7a0 | 389 | fl->dst = freq->flr_dst; |
1da177e4 LT |
390 | atomic_set(&fl->users, 1); |
391 | switch (fl->share) { | |
392 | case IPV6_FL_S_EXCL: | |
393 | case IPV6_FL_S_ANY: | |
394 | break; | |
395 | case IPV6_FL_S_PROCESS: | |
396 | fl->owner = current->pid; | |
397 | break; | |
398 | case IPV6_FL_S_USER: | |
f82b3590 | 399 | fl->owner = current_euid(); |
1da177e4 LT |
400 | break; |
401 | default: | |
402 | err = -EINVAL; | |
403 | goto done; | |
404 | } | |
405 | return fl; | |
406 | ||
407 | done: | |
408 | fl_free(fl); | |
409 | *err_p = err; | |
410 | return NULL; | |
411 | } | |
412 | ||
413 | static int mem_check(struct sock *sk) | |
414 | { | |
415 | struct ipv6_pinfo *np = inet6_sk(sk); | |
416 | struct ipv6_fl_socklist *sfl; | |
417 | int room = FL_MAX_SIZE - atomic_read(&fl_size); | |
418 | int count = 0; | |
419 | ||
420 | if (room > FL_MAX_SIZE - FL_MAX_PER_SOCK) | |
421 | return 0; | |
422 | ||
423 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) | |
424 | count++; | |
425 | ||
426 | if (room <= 0 || | |
427 | ((count >= FL_MAX_PER_SOCK || | |
35700212 JP |
428 | (count > 0 && room < FL_MAX_SIZE/2) || room < FL_MAX_SIZE/4) && |
429 | !capable(CAP_NET_ADMIN))) | |
1da177e4 LT |
430 | return -ENOBUFS; |
431 | ||
432 | return 0; | |
433 | } | |
434 | ||
435 | static int ipv6_hdr_cmp(struct ipv6_opt_hdr *h1, struct ipv6_opt_hdr *h2) | |
436 | { | |
437 | if (h1 == h2) | |
438 | return 0; | |
439 | if (h1 == NULL || h2 == NULL) | |
440 | return 1; | |
441 | if (h1->hdrlen != h2->hdrlen) | |
442 | return 1; | |
443 | return memcmp(h1+1, h2+1, ((h1->hdrlen+1)<<3) - sizeof(*h1)); | |
444 | } | |
445 | ||
446 | static int ipv6_opt_cmp(struct ipv6_txoptions *o1, struct ipv6_txoptions *o2) | |
447 | { | |
448 | if (o1 == o2) | |
449 | return 0; | |
450 | if (o1 == NULL || o2 == NULL) | |
451 | return 1; | |
452 | if (o1->opt_nflen != o2->opt_nflen) | |
453 | return 1; | |
454 | if (ipv6_hdr_cmp(o1->hopopt, o2->hopopt)) | |
455 | return 1; | |
456 | if (ipv6_hdr_cmp(o1->dst0opt, o2->dst0opt)) | |
457 | return 1; | |
458 | if (ipv6_hdr_cmp((struct ipv6_opt_hdr *)o1->srcrt, (struct ipv6_opt_hdr *)o2->srcrt)) | |
459 | return 1; | |
460 | return 0; | |
461 | } | |
462 | ||
04028045 PE |
463 | static inline void fl_link(struct ipv6_pinfo *np, struct ipv6_fl_socklist *sfl, |
464 | struct ip6_flowlabel *fl) | |
465 | { | |
466 | write_lock_bh(&ip6_sk_fl_lock); | |
467 | sfl->fl = fl; | |
468 | sfl->next = np->ipv6_fl_list; | |
469 | np->ipv6_fl_list = sfl; | |
470 | write_unlock_bh(&ip6_sk_fl_lock); | |
471 | } | |
472 | ||
1da177e4 LT |
473 | int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen) |
474 | { | |
55205d40 | 475 | int uninitialized_var(err); |
60e8fbc4 | 476 | struct net *net = sock_net(sk); |
1da177e4 LT |
477 | struct ipv6_pinfo *np = inet6_sk(sk); |
478 | struct in6_flowlabel_req freq; | |
479 | struct ipv6_fl_socklist *sfl1=NULL; | |
480 | struct ipv6_fl_socklist *sfl, **sflp; | |
78c2e502 PE |
481 | struct ip6_flowlabel *fl, *fl1 = NULL; |
482 | ||
1da177e4 LT |
483 | |
484 | if (optlen < sizeof(freq)) | |
485 | return -EINVAL; | |
486 | ||
487 | if (copy_from_user(&freq, optval, sizeof(freq))) | |
488 | return -EFAULT; | |
489 | ||
490 | switch (freq.flr_action) { | |
491 | case IPV6_FL_A_PUT: | |
492 | write_lock_bh(&ip6_sk_fl_lock); | |
493 | for (sflp = &np->ipv6_fl_list; (sfl=*sflp)!=NULL; sflp = &sfl->next) { | |
494 | if (sfl->fl->label == freq.flr_label) { | |
495 | if (freq.flr_label == (np->flow_label&IPV6_FLOWLABEL_MASK)) | |
496 | np->flow_label &= ~IPV6_FLOWLABEL_MASK; | |
497 | *sflp = sfl->next; | |
498 | write_unlock_bh(&ip6_sk_fl_lock); | |
499 | fl_release(sfl->fl); | |
500 | kfree(sfl); | |
501 | return 0; | |
502 | } | |
503 | } | |
504 | write_unlock_bh(&ip6_sk_fl_lock); | |
505 | return -ESRCH; | |
506 | ||
507 | case IPV6_FL_A_RENEW: | |
508 | read_lock_bh(&ip6_sk_fl_lock); | |
509 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) { | |
510 | if (sfl->fl->label == freq.flr_label) { | |
511 | err = fl6_renew(sfl->fl, freq.flr_linger, freq.flr_expires); | |
512 | read_unlock_bh(&ip6_sk_fl_lock); | |
513 | return err; | |
514 | } | |
515 | } | |
516 | read_unlock_bh(&ip6_sk_fl_lock); | |
517 | ||
518 | if (freq.flr_share == IPV6_FL_S_NONE && capable(CAP_NET_ADMIN)) { | |
60e8fbc4 | 519 | fl = fl_lookup(net, freq.flr_label); |
1da177e4 LT |
520 | if (fl) { |
521 | err = fl6_renew(fl, freq.flr_linger, freq.flr_expires); | |
522 | fl_release(fl); | |
523 | return err; | |
524 | } | |
525 | } | |
526 | return -ESRCH; | |
527 | ||
528 | case IPV6_FL_A_GET: | |
529 | if (freq.flr_label & ~IPV6_FLOWLABEL_MASK) | |
530 | return -EINVAL; | |
531 | ||
ec0506db | 532 | fl = fl_create(net, sk, &freq, optval, optlen, &err); |
1da177e4 LT |
533 | if (fl == NULL) |
534 | return err; | |
535 | sfl1 = kmalloc(sizeof(*sfl1), GFP_KERNEL); | |
536 | ||
537 | if (freq.flr_label) { | |
1da177e4 LT |
538 | err = -EEXIST; |
539 | read_lock_bh(&ip6_sk_fl_lock); | |
540 | for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) { | |
541 | if (sfl->fl->label == freq.flr_label) { | |
542 | if (freq.flr_flags&IPV6_FL_F_EXCL) { | |
543 | read_unlock_bh(&ip6_sk_fl_lock); | |
544 | goto done; | |
545 | } | |
546 | fl1 = sfl->fl; | |
4ea6a804 | 547 | atomic_inc(&fl1->users); |
1da177e4 LT |
548 | break; |
549 | } | |
550 | } | |
551 | read_unlock_bh(&ip6_sk_fl_lock); | |
552 | ||
553 | if (fl1 == NULL) | |
60e8fbc4 | 554 | fl1 = fl_lookup(net, freq.flr_label); |
1da177e4 | 555 | if (fl1) { |
78c2e502 | 556 | recheck: |
1da177e4 LT |
557 | err = -EEXIST; |
558 | if (freq.flr_flags&IPV6_FL_F_EXCL) | |
559 | goto release; | |
560 | err = -EPERM; | |
561 | if (fl1->share == IPV6_FL_S_EXCL || | |
562 | fl1->share != fl->share || | |
563 | fl1->owner != fl->owner) | |
564 | goto release; | |
565 | ||
566 | err = -EINVAL; | |
567 | if (!ipv6_addr_equal(&fl1->dst, &fl->dst) || | |
568 | ipv6_opt_cmp(fl1->opt, fl->opt)) | |
569 | goto release; | |
570 | ||
571 | err = -ENOMEM; | |
572 | if (sfl1 == NULL) | |
573 | goto release; | |
574 | if (fl->linger > fl1->linger) | |
575 | fl1->linger = fl->linger; | |
576 | if ((long)(fl->expires - fl1->expires) > 0) | |
577 | fl1->expires = fl->expires; | |
04028045 | 578 | fl_link(np, sfl1, fl1); |
1da177e4 LT |
579 | fl_free(fl); |
580 | return 0; | |
581 | ||
582 | release: | |
583 | fl_release(fl1); | |
584 | goto done; | |
585 | } | |
586 | } | |
587 | err = -ENOENT; | |
588 | if (!(freq.flr_flags&IPV6_FL_F_CREATE)) | |
589 | goto done; | |
590 | ||
591 | err = -ENOMEM; | |
592 | if (sfl1 == NULL || (err = mem_check(sk)) != 0) | |
593 | goto done; | |
594 | ||
60e8fbc4 | 595 | fl1 = fl_intern(net, fl, freq.flr_label); |
78c2e502 PE |
596 | if (fl1 != NULL) |
597 | goto recheck; | |
1da177e4 | 598 | |
6c94d361 DM |
599 | if (!freq.flr_label) { |
600 | if (copy_to_user(&((struct in6_flowlabel_req __user *) optval)->flr_label, | |
601 | &fl->label, sizeof(fl->label))) { | |
602 | /* Intentionally ignore fault. */ | |
603 | } | |
604 | } | |
1da177e4 | 605 | |
04028045 | 606 | fl_link(np, sfl1, fl); |
1da177e4 LT |
607 | return 0; |
608 | ||
609 | default: | |
610 | return -EINVAL; | |
611 | } | |
612 | ||
613 | done: | |
614 | fl_free(fl); | |
615 | kfree(sfl1); | |
616 | return err; | |
617 | } | |
618 | ||
619 | #ifdef CONFIG_PROC_FS | |
620 | ||
621 | struct ip6fl_iter_state { | |
5983a3df | 622 | struct seq_net_private p; |
1da177e4 LT |
623 | int bucket; |
624 | }; | |
625 | ||
626 | #define ip6fl_seq_private(seq) ((struct ip6fl_iter_state *)(seq)->private) | |
627 | ||
628 | static struct ip6_flowlabel *ip6fl_get_first(struct seq_file *seq) | |
629 | { | |
630 | struct ip6_flowlabel *fl = NULL; | |
631 | struct ip6fl_iter_state *state = ip6fl_seq_private(seq); | |
5983a3df | 632 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
633 | |
634 | for (state->bucket = 0; state->bucket <= FL_HASH_MASK; ++state->bucket) { | |
5983a3df BT |
635 | fl = fl_ht[state->bucket]; |
636 | ||
09ad9bc7 | 637 | while (fl && !net_eq(fl->fl_net, net)) |
5983a3df BT |
638 | fl = fl->next; |
639 | if (fl) | |
1da177e4 | 640 | break; |
1da177e4 LT |
641 | } |
642 | return fl; | |
643 | } | |
644 | ||
645 | static struct ip6_flowlabel *ip6fl_get_next(struct seq_file *seq, struct ip6_flowlabel *fl) | |
646 | { | |
647 | struct ip6fl_iter_state *state = ip6fl_seq_private(seq); | |
5983a3df | 648 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
649 | |
650 | fl = fl->next; | |
5983a3df | 651 | try_again: |
09ad9bc7 | 652 | while (fl && !net_eq(fl->fl_net, net)) |
5983a3df BT |
653 | fl = fl->next; |
654 | ||
1da177e4 | 655 | while (!fl) { |
5983a3df | 656 | if (++state->bucket <= FL_HASH_MASK) { |
1da177e4 | 657 | fl = fl_ht[state->bucket]; |
5983a3df BT |
658 | goto try_again; |
659 | } else | |
bcd62075 | 660 | break; |
1da177e4 LT |
661 | } |
662 | return fl; | |
663 | } | |
664 | ||
665 | static struct ip6_flowlabel *ip6fl_get_idx(struct seq_file *seq, loff_t pos) | |
666 | { | |
667 | struct ip6_flowlabel *fl = ip6fl_get_first(seq); | |
668 | if (fl) | |
669 | while (pos && (fl = ip6fl_get_next(seq, fl)) != NULL) | |
670 | --pos; | |
671 | return pos ? NULL : fl; | |
672 | } | |
673 | ||
674 | static void *ip6fl_seq_start(struct seq_file *seq, loff_t *pos) | |
9a429c49 | 675 | __acquires(ip6_fl_lock) |
1da177e4 LT |
676 | { |
677 | read_lock_bh(&ip6_fl_lock); | |
678 | return *pos ? ip6fl_get_idx(seq, *pos - 1) : SEQ_START_TOKEN; | |
679 | } | |
680 | ||
681 | static void *ip6fl_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
682 | { | |
683 | struct ip6_flowlabel *fl; | |
684 | ||
685 | if (v == SEQ_START_TOKEN) | |
686 | fl = ip6fl_get_first(seq); | |
687 | else | |
688 | fl = ip6fl_get_next(seq, v); | |
689 | ++*pos; | |
690 | return fl; | |
691 | } | |
692 | ||
693 | static void ip6fl_seq_stop(struct seq_file *seq, void *v) | |
9a429c49 | 694 | __releases(ip6_fl_lock) |
1da177e4 LT |
695 | { |
696 | read_unlock_bh(&ip6_fl_lock); | |
697 | } | |
698 | ||
1b7c2dbc | 699 | static int ip6fl_seq_show(struct seq_file *seq, void *v) |
1da177e4 | 700 | { |
1b7c2dbc JM |
701 | if (v == SEQ_START_TOKEN) |
702 | seq_printf(seq, "%-5s %-1s %-6s %-6s %-6s %-8s %-32s %s\n", | |
703 | "Label", "S", "Owner", "Users", "Linger", "Expires", "Dst", "Opt"); | |
704 | else { | |
705 | struct ip6_flowlabel *fl = v; | |
1da177e4 | 706 | seq_printf(seq, |
4b7a4274 | 707 | "%05X %-1d %-6d %-6d %-6ld %-8ld %pi6 %-4d\n", |
1da177e4 LT |
708 | (unsigned)ntohl(fl->label), |
709 | fl->share, | |
710 | (unsigned)fl->owner, | |
711 | atomic_read(&fl->users), | |
712 | fl->linger/HZ, | |
713 | (long)(fl->expires - jiffies)/HZ, | |
b071195d | 714 | &fl->dst, |
1da177e4 | 715 | fl->opt ? fl->opt->opt_nflen : 0); |
1da177e4 | 716 | } |
1da177e4 LT |
717 | return 0; |
718 | } | |
719 | ||
56b3d975 | 720 | static const struct seq_operations ip6fl_seq_ops = { |
1da177e4 LT |
721 | .start = ip6fl_seq_start, |
722 | .next = ip6fl_seq_next, | |
723 | .stop = ip6fl_seq_stop, | |
724 | .show = ip6fl_seq_show, | |
725 | }; | |
726 | ||
727 | static int ip6fl_seq_open(struct inode *inode, struct file *file) | |
728 | { | |
5983a3df BT |
729 | return seq_open_net(inode, file, &ip6fl_seq_ops, |
730 | sizeof(struct ip6fl_iter_state)); | |
1da177e4 LT |
731 | } |
732 | ||
9a32144e | 733 | static const struct file_operations ip6fl_seq_fops = { |
1da177e4 LT |
734 | .owner = THIS_MODULE, |
735 | .open = ip6fl_seq_open, | |
736 | .read = seq_read, | |
737 | .llseek = seq_lseek, | |
5983a3df | 738 | .release = seq_release_net, |
1da177e4 | 739 | }; |
1da177e4 | 740 | |
2c8c1e72 | 741 | static int __net_init ip6_flowlabel_proc_init(struct net *net) |
0a3e78ac | 742 | { |
5983a3df BT |
743 | if (!proc_net_fops_create(net, "ip6_flowlabel", |
744 | S_IRUGO, &ip6fl_seq_fops)) | |
0a3e78ac DL |
745 | return -ENOMEM; |
746 | return 0; | |
747 | } | |
1da177e4 | 748 | |
2c8c1e72 | 749 | static void __net_exit ip6_flowlabel_proc_fini(struct net *net) |
1da177e4 | 750 | { |
0a3e78ac DL |
751 | proc_net_remove(net, "ip6_flowlabel"); |
752 | } | |
753 | #else | |
754 | static inline int ip6_flowlabel_proc_init(struct net *net) | |
755 | { | |
756 | return 0; | |
757 | } | |
758 | static inline void ip6_flowlabel_proc_fini(struct net *net) | |
759 | { | |
0a3e78ac | 760 | } |
1da177e4 | 761 | #endif |
0a3e78ac | 762 | |
2c8c1e72 | 763 | static void __net_exit ip6_flowlabel_net_exit(struct net *net) |
60e8fbc4 BT |
764 | { |
765 | ip6_fl_purge(net); | |
5983a3df | 766 | ip6_flowlabel_proc_fini(net); |
60e8fbc4 BT |
767 | } |
768 | ||
769 | static struct pernet_operations ip6_flowlabel_net_ops = { | |
5983a3df | 770 | .init = ip6_flowlabel_proc_init, |
60e8fbc4 BT |
771 | .exit = ip6_flowlabel_net_exit, |
772 | }; | |
773 | ||
0a3e78ac DL |
774 | int ip6_flowlabel_init(void) |
775 | { | |
5983a3df | 776 | return register_pernet_subsys(&ip6_flowlabel_net_ops); |
1da177e4 LT |
777 | } |
778 | ||
779 | void ip6_flowlabel_cleanup(void) | |
780 | { | |
781 | del_timer(&ip6_fl_gc_timer); | |
60e8fbc4 | 782 | unregister_pernet_subsys(&ip6_flowlabel_net_ops); |
1da177e4 | 783 | } |