1 /*
2 * net/sched/cls_u32.c Ugly (or Universal) 32bit key Packet Classifier.
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 * The filters are packed to hash tables of key nodes
12 * with a set of 32bit key/mask pairs at every node.
13 * Nodes reference next level hash tables etc.
14 *
15 * This scheme is the best universal classifier I managed to
16 * invent; it is not super-fast, but it is not slow (provided you
17 * program it correctly), and general enough. And its relative
18 * speed grows as the number of rules becomes larger.
19 *
20 * It seems that it represents the best middle point between
21 * speed and manageability both by human and by machine.
22 *
23 * It is especially useful for link sharing combined with QoS;
24 * pure RSVP doesn't need such a general approach and can use
25 * much simpler (and faster) schemes, sort of cls_rsvp.c.
26 */
27
28 #include <asm/uaccess.h>
29 #include <asm/system.h>
30 #include <asm/bitops.h>
31 #include <linux/config.h>
32 #include <linux/module.h>
33 #include <linux/types.h>
34 #include <linux/kernel.h>
35 #include <linux/sched.h>
36 #include <linux/string.h>
37 #include <linux/mm.h>
38 #include <linux/socket.h>
39 #include <linux/sockios.h>
40 #include <linux/in.h>
41 #include <linux/errno.h>
42 #include <linux/interrupt.h>
43 #include <linux/if_ether.h>
44 #include <linux/inet.h>
45 #include <linux/netdevice.h>
46 #include <linux/etherdevice.h>
47 #include <linux/notifier.h>
48 #include <linux/rtnetlink.h>
49 #include <net/ip.h>
50 #include <net/route.h>
51 #include <linux/skbuff.h>
52 #include <net/sock.h>
53 #include <net/pkt_sched.h>
54
55
56 struct tc_u_knode
57 {
58 struct tc_u_knode *next;
59 u32 handle;
60 struct tc_u_hnode *ht_up;
61 #ifdef CONFIG_NET_CLS_POLICE
62 struct tcf_police *police;
63 #endif
64 struct tcf_result res;
65 struct tc_u_hnode *ht_down;
66 struct tc_u32_sel sel;
67 };
68
69 struct tc_u_hnode
70 {
71 struct tc_u_hnode *next;
72 u32 handle;
73 struct tc_u_common *tp_c;
74 int refcnt;
75 unsigned divisor;
76 u32 hgenerator;
77 struct tc_u_knode *ht[1];
78 };
79
80 struct tc_u_common
81 {
82 struct tc_u_common *next;
83 struct tc_u_hnode *hlist;
84 struct Qdisc *q;
85 int refcnt;
86 u32 hgenerator;
87 };
88
89 static struct tc_u_common *u32_list;
90
91 static __inline__ unsigned u32_hash_fold(u32 key, struct tc_u32_sel *sel)
92 {
93 unsigned h = key & sel->hmask;
94
95 h ^= h>>16;
96 h ^= h>>8;
97 return h;
98 }
99
100 static int u32_classify(struct sk_buff *skb, struct tcf_proto *tp, struct tcf_result *res)
101 {
102 struct {
103 struct tc_u_knode *knode;
104 u8 *ptr;
105 } stack[TC_U32_MAXDEPTH];
106
107 struct tc_u_hnode *ht = (struct tc_u_hnode*)tp->root;
108 u8 *ptr = skb->nh.raw;
109 struct tc_u_knode *n;
110 int sdepth = 0;
111 int off2 = 0;
112 int sel = 0;
113 int i;
114
115 next_ht:
116 n = ht->ht[sel];
117
118 next_knode:
119 if (n) {
120 struct tc_u32_key *key = n->sel.keys;
121
122 for (i = n->sel.nkeys; i>0; i--, key++) {
123 if ((*(u32*)(ptr+key->off+(off2&key->offmask))^key->val)&key->mask) {
124 n = n->next;
125 goto next_knode;
126 }
127 }
128 if (n->ht_down == NULL) {
129 check_terminal:
130 if (n->sel.flags&TC_U32_TERMINAL) {
131 *res = n->res;
132 #ifdef CONFIG_NET_CLS_POLICE
133 if (n->police) {
134 int pol_res = tcf_police(skb, n->police);
135 if (pol_res >= 0)
136 return pol_res;
137 } else
138 #endif
139 return 0;
140 }
141 n = n->next;
142 goto next_knode;
143 }
144
145 /* PUSH */
146 if (sdepth >= TC_U32_MAXDEPTH)
147 goto deadloop;
148 stack[sdepth].knode = n;
149 stack[sdepth].ptr = ptr;
150 sdepth++;
151
152 ht = n->ht_down;
153 sel = 0;
154 if (ht->divisor)
155 sel = ht->divisor&u32_hash_fold(*(u32*)(ptr+n->sel.hoff), &n->sel);
156
157 if (!(n->sel.flags&(TC_U32_VAROFFSET|TC_U32_OFFSET|TC_U32_EAT)))
158 goto next_ht;
159
160 if (n->sel.flags&(TC_U32_OFFSET|TC_U32_VAROFFSET)) {
161 off2 = n->sel.off + 3;
162 if (n->sel.flags&TC_U32_VAROFFSET)
163 off2 += ntohs(n->sel.offmask & *(u16*)(ptr+n->sel.offoff)) >>n->sel.offshift;
164 off2 &= ~3;
165 }
166 if (n->sel.flags&TC_U32_EAT) {
167 ptr += off2;
168 off2 = 0;
169 }
170
171 if (ptr < skb->tail)
172 goto next_ht;
173 }
174
175 /* POP */
176 if (sdepth--) {
177 n = stack[sdepth].knode;
178 ht = n->ht_up;
179 ptr = stack[sdepth].ptr;
180 goto check_terminal;
181 }
182 return -1;
183
184 deadloop:
185 if (net_ratelimit())
186 printk("cls_u32: dead loop\n");
187 return -1;
188 }
189
190 static __inline__ struct tc_u_hnode *
191 u32_lookup_ht(struct tc_u_common *tp_c, u32 handle)
192 {
193 struct tc_u_hnode *ht;
194
195 for (ht = tp_c->hlist; ht; ht = ht->next)
196 if (ht->handle == handle)
197 break;
198
199 return ht;
200 }
201
202 static __inline__ struct tc_u_knode *
203 u32_lookup_key(struct tc_u_hnode *ht, u32 handle)
204 {
205 unsigned sel;
206 struct tc_u_knode *n = NULL;
207
208 sel = TC_U32_HASH(handle);
209 if (sel > ht->divisor)
210 goto out;
211
212 for (n = ht->ht[sel]; n; n = n->next)
213 if (n->handle == handle)
214 break;
215 out:
216 return n;
217 }
218
219
220 static unsigned long u32_get(struct tcf_proto *tp, u32 handle)
221 {
222 struct tc_u_hnode *ht;
223 struct tc_u_common *tp_c = tp->data;
224
225 if (TC_U32_HTID(handle) == TC_U32_ROOT)
226 ht = tp->root;
227 else
228 ht = u32_lookup_ht(tp_c, TC_U32_HTID(handle));
229
230 if (!ht)
231 return 0;
232
233 if (TC_U32_KEY(handle) == 0)
234 return (unsigned long)ht;
235
236 return (unsigned long)u32_lookup_key(ht, handle);
237 }
238
239 static void u32_put(struct tcf_proto *tp, unsigned long f)
240 {
241 }
242
243 static u32 gen_new_htid(struct tc_u_common *tp_c)
244 {
245 int i = 0x800;
246
247 do {
248 if (++tp_c->hgenerator == 0x7FF)
249 tp_c->hgenerator = 1;
250 } while (--i>0 && u32_lookup_ht(tp_c, (tp_c->hgenerator|0x800)<<20));
251
252 return i > 0 ? (tp_c->hgenerator|0x800)<<20 : 0;
253 }
254
255 static int u32_init(struct tcf_proto *tp)
256 {
257 struct tc_u_hnode *root_ht;
258 struct tc_u_common *tp_c;
259
260 for (tp_c = u32_list; tp_c; tp_c = tp_c->next)
261 if (tp_c->q == tp->q)
262 break;
263
264 root_ht = kmalloc(sizeof(*root_ht), GFP_KERNEL);
265 if (root_ht == NULL)
266 return -ENOBUFS;
267
268 memset(root_ht, 0, sizeof(*root_ht));
269 root_ht->divisor = 0;
270 root_ht->refcnt++;
271 root_ht->handle = tp_c ? gen_new_htid(tp_c) : 0x80000000;
272
273 if (tp_c == NULL) {
274 tp_c = kmalloc(sizeof(*tp_c), GFP_KERNEL);
275 if (tp_c == NULL) {
276 kfree(root_ht);
277 return -ENOBUFS;
278 }
279 memset(tp_c, 0, sizeof(*tp_c));
280 tp_c->q = tp->q;
281 tp_c->next = u32_list;
282 u32_list = tp_c;
283 }
284
285 tp_c->refcnt++;
286 root_ht->next = tp_c->hlist;
287 tp_c->hlist = root_ht;
288 root_ht->tp_c = tp_c;
289
290 tp->root = root_ht;
291 tp->data = tp_c;
292 return 0;
293 }
294
295 static int u32_destroy_key(struct tcf_proto *tp, struct tc_u_knode *n)
296 {
297 unsigned long cl;
298
299 if ((cl = __cls_set_class(&n->res.class, 0)) != 0)
300 tp->q->ops->cl_ops->unbind_tcf(tp->q, cl);
301 #ifdef CONFIG_NET_CLS_POLICE
302 tcf_police_release(n->police);
303 #endif
304 if (n->ht_down)
305 n->ht_down->refcnt--;
306 kfree(n);
307 return 0;
308 }
309
310 static int u32_delete_key(struct tcf_proto *tp, struct tc_u_knode* key)
311 {
312 struct tc_u_knode **kp;
313 struct tc_u_hnode *ht = key->ht_up;
314
315 if (ht) {
316 for (kp = &ht->ht[TC_U32_HASH(key->handle)]; *kp; kp = &(*kp)->next) {
317 if (*kp == key) {
318 tcf_tree_lock(tp);
319 *kp = key->next;
320 tcf_tree_unlock(tp);
321
322 u32_destroy_key(tp, key);
323 return 0;
324 }
325 }
326 }
327 BUG_TRAP(0);
328 return 0;
329 }
330
331 static void u32_clear_hnode(struct tcf_proto *tp, struct tc_u_hnode *ht)
332 {
333 struct tc_u_knode *n;
334 unsigned h;
335
336 for (h=0; h<=ht->divisor; h++) {
337 while ((n = ht->ht[h]) != NULL) {
338 ht->ht[h] = n->next;
339
340 u32_destroy_key(tp, n);
341 }
342 }
343 }
344
345 static int u32_destroy_hnode(struct tcf_proto *tp, struct tc_u_hnode *ht)
346 {
347 struct tc_u_common *tp_c = tp->data;
348 struct tc_u_hnode **hn;
349
350 BUG_TRAP(!ht->refcnt);
351
352 u32_clear_hnode(tp, ht);
353
354 for (hn = &tp_c->hlist; *hn; hn = &(*hn)->next) {
355 if (*hn == ht) {
356 *hn = ht->next;
357 kfree(ht);
358 return 0;
359 }
360 }
361
362 BUG_TRAP(0);
363 return -ENOENT;
364 }
365
366 static void u32_destroy(struct tcf_proto *tp)
367 {
368 struct tc_u_common *tp_c = tp->data;
369 struct tc_u_hnode *root_ht = xchg(&tp->root, NULL);
370
371 BUG_TRAP(root_ht != NULL);
372
373 if (root_ht && --root_ht->refcnt == 0)
374 u32_destroy_hnode(tp, root_ht);
375
376 if (--tp_c->refcnt == 0) {
377 struct tc_u_hnode *ht;
378 struct tc_u_common **tp_cp;
379
380 for (tp_cp = &u32_list; *tp_cp; tp_cp = &(*tp_cp)->next) {
381 if (*tp_cp == tp_c) {
382 *tp_cp = tp_c->next;
383 break;
384 }
385 }
386
387 for (ht=tp_c->hlist; ht; ht = ht->next)
388 u32_clear_hnode(tp, ht);
389
390 while ((ht = tp_c->hlist) != NULL) {
391 tp_c->hlist = ht->next;
392
393 BUG_TRAP(ht->refcnt == 0);
394
395 kfree(ht);
396 };
397
398 kfree(tp_c);
399 }
400
401 tp->data = NULL;
402 }
403
404 static int u32_delete(struct tcf_proto *tp, unsigned long arg)
405 {
406 struct tc_u_hnode *ht = (struct tc_u_hnode*)arg;
407
408 if (ht == NULL)
409 return 0;
410
411 if (TC_U32_KEY(ht->handle))
412 return u32_delete_key(tp, (struct tc_u_knode*)ht);
413
414 if (tp->root == ht)
415 return -EINVAL;
416
417 if (--ht->refcnt == 0)
418 u32_destroy_hnode(tp, ht);
419
420 return 0;
421 }
422
423 static u32 gen_new_kid(struct tc_u_hnode *ht, u32 handle)
424 {
425 struct tc_u_knode *n;
426 unsigned i = 0x7FF;
427
428 for (n=ht->ht[TC_U32_HASH(handle)]; n; n = n->next)
429 if (i < TC_U32_NODE(n->handle))
430 i = TC_U32_NODE(n->handle);
431 i++;
432
433 return handle|(i>0xFFF ? 0xFFF : i);
434 }
435
436 static int u32_set_parms(struct Qdisc *q, unsigned long base,
437 struct tc_u_hnode *ht,
438 struct tc_u_knode *n, struct rtattr **tb,
439 struct rtattr *est)
440 {
441 if (tb[TCA_U32_LINK-1]) {
442 u32 handle = *(u32*)RTA_DATA(tb[TCA_U32_LINK-1]);
443 struct tc_u_hnode *ht_down = NULL;
444
445 if (TC_U32_KEY(handle))
446 return -EINVAL;
447
448 if (handle) {
449 ht_down = u32_lookup_ht(ht->tp_c, handle);
450
451 if (ht_down == NULL)
452 return -EINVAL;
453 ht_down->refcnt++;
454 }
455
456 sch_tree_lock(q);
457 ht_down = xchg(&n->ht_down, ht_down);
458 sch_tree_unlock(q);
459
460 if (ht_down)
461 ht_down->refcnt--;
462 }
463 if (tb[TCA_U32_CLASSID-1]) {
464 unsigned long cl;
465
466 n->res.classid = *(u32*)RTA_DATA(tb[TCA_U32_CLASSID-1]);
467 sch_tree_lock(q);
468 cl = __cls_set_class(&n->res.class, q->ops->cl_ops->bind_tcf(q, base, n->res.classid));
469 sch_tree_unlock(q);
470 if (cl)
471 q->ops->cl_ops->unbind_tcf(q, cl);
472 }
473 #ifdef CONFIG_NET_CLS_POLICE
474 if (tb[TCA_U32_POLICE-1]) {
475 struct tcf_police *police = tcf_police_locate(tb[TCA_U32_POLICE-1], est);
476
477 sch_tree_lock(q);
478 police = xchg(&n->police, police);
479 sch_tree_unlock(q);
480
481 tcf_police_release(police);
482 }
483 #endif
484 return 0;
485 }
486
487 static int u32_change(struct tcf_proto *tp, unsigned long base, u32 handle,
488 struct rtattr **tca,
489 unsigned long *arg)
490 {
491 struct tc_u_common *tp_c = tp->data;
492 struct tc_u_hnode *ht;
493 struct tc_u_knode *n;
494 struct tc_u32_sel *s;
495 struct rtattr *opt = tca[TCA_OPTIONS-1];
496 struct rtattr *tb[TCA_U32_MAX];
497 u32 htid;
498 int err;
499
500 if (opt == NULL)
501 return handle ? -EINVAL : 0;
502
503 if (rtattr_parse(tb, TCA_U32_MAX, RTA_DATA(opt), RTA_PAYLOAD(opt)) < 0)
504 return -EINVAL;
505
506 if ((n = (struct tc_u_knode*)*arg) != NULL) {
507 if (TC_U32_KEY(n->handle) == 0)
508 return -EINVAL;
509
510 return u32_set_parms(tp->q, base, n->ht_up, n, tb, tca[TCA_RATE-1]);
511 }
512
513 if (tb[TCA_U32_DIVISOR-1]) {
514 unsigned divisor = *(unsigned*)RTA_DATA(tb[TCA_U32_DIVISOR-1]);
515
516 if (--divisor > 0x100)
517 return -EINVAL;
518 if (TC_U32_KEY(handle))
519 return -EINVAL;
520 if (handle == 0) {
521 handle = gen_new_htid(tp->data);
522 if (handle == 0)
523 return -ENOMEM;
524 }
525 ht = kmalloc(sizeof(*ht) + divisor*sizeof(void*), GFP_KERNEL);
526 if (ht == NULL)
527 return -ENOBUFS;
528 memset(ht, 0, sizeof(*ht) + divisor*sizeof(void*));
529 ht->tp_c = tp_c;
530 ht->refcnt = 0;
531 ht->divisor = divisor;
532 ht->handle = handle;
533 ht->next = tp_c->hlist;
534 tp_c->hlist = ht;
535 *arg = (unsigned long)ht;
536 return 0;
537 }
538
539 if (tb[TCA_U32_HASH-1]) {
540 htid = *(unsigned*)RTA_DATA(tb[TCA_U32_HASH-1]);
541 if (TC_U32_HTID(htid) == TC_U32_ROOT) {
542 ht = tp->root;
543 htid = ht->handle;
544 } else {
545 ht = u32_lookup_ht(tp->data, TC_U32_HTID(htid));
546 if (ht == NULL)
547 return -EINVAL;
548 }
549 } else {
550 ht = tp->root;
551 htid = ht->handle;
552 }
553
554 if (ht->divisor < TC_U32_HASH(htid))
555 return -EINVAL;
556
557 if (handle) {
558 if (TC_U32_HTID(handle) && TC_U32_HTID(handle^htid))
559 return -EINVAL;
560 handle = htid | TC_U32_NODE(handle);
561 } else
562 handle = gen_new_kid(ht, htid);
563
564 if (tb[TCA_U32_SEL-1] == 0 ||
565 RTA_PAYLOAD(tb[TCA_U32_SEL-1]) < sizeof(struct tc_u32_sel))
566 return -EINVAL;
567
568 s = RTA_DATA(tb[TCA_U32_SEL-1]);
569 n = kmalloc(sizeof(*n) + s->nkeys*sizeof(struct tc_u32_key), GFP_KERNEL);
570 if (n == NULL)
571 return -ENOBUFS;
572 memset(n, 0, sizeof(*n) + s->nkeys*sizeof(struct tc_u32_key));
573 memcpy(&n->sel, s, sizeof(*s) + s->nkeys*sizeof(struct tc_u32_key));
574 n->ht_up = ht;
575 n->handle = handle;
576 err = u32_set_parms(tp->q, base, ht, n, tb, tca[TCA_RATE-1]);
577 if (err == 0) {
578 struct tc_u_knode **ins;
579 for (ins = &ht->ht[TC_U32_HASH(handle)]; *ins; ins = &(*ins)->next)
580 if (TC_U32_NODE(handle) < TC_U32_NODE((*ins)->handle))
581 break;
582
583 n->next = *ins;
584 wmb();
585 *ins = n;
586
587 *arg = (unsigned long)n;
588 return 0;
589 }
590 kfree(n);
591 return err;
592 }
593
594 static void u32_walk(struct tcf_proto *tp, struct tcf_walker *arg)
595 {
596 struct tc_u_common *tp_c = tp->data;
597 struct tc_u_hnode *ht;
598 struct tc_u_knode *n;
599 unsigned h;
600
601 if (arg->stop)
602 return;
603
604 for (ht = tp_c->hlist; ht; ht = ht->next) {
605 if (arg->count >= arg->skip) {
606 if (arg->fn(tp, (unsigned long)ht, arg) < 0) {
607 arg->stop = 1;
608 return;
609 }
610 }
611 arg->count++;
612 for (h = 0; h <= ht->divisor; h++) {
613 for (n = ht->ht[h]; n; n = n->next) {
614 if (arg->count < arg->skip) {
615 arg->count++;
616 continue;
617 }
618 if (arg->fn(tp, (unsigned long)n, arg) < 0) {
619 arg->stop = 1;
620 return;
621 }
622 arg->count++;
623 }
624 }
625 }
626 }
627
628 static int u32_dump(struct tcf_proto *tp, unsigned long fh,
629 struct sk_buff *skb, struct tcmsg *t)
630 {
631 struct tc_u_knode *n = (struct tc_u_knode*)fh;
632 unsigned char *b = skb->tail;
633 struct rtattr *rta;
634
635 if (n == NULL)
636 return skb->len;
637
638 t->tcm_handle = n->handle;
639
640 rta = (struct rtattr*)b;
641 RTA_PUT(skb, TCA_OPTIONS, 0, NULL);
642
643 if (TC_U32_KEY(n->handle) == 0) {
644 struct tc_u_hnode *ht = (struct tc_u_hnode*)fh;
645 u32 divisor = ht->divisor+1;
646 RTA_PUT(skb, TCA_U32_DIVISOR, 4, &divisor);
647 } else {
648 RTA_PUT(skb, TCA_U32_SEL,
649 sizeof(n->sel) + n->sel.nkeys*sizeof(struct tc_u32_key),
650 &n->sel);
651 if (n->ht_up) {
652 u32 htid = n->handle & 0xFFFFF000;
653 RTA_PUT(skb, TCA_U32_HASH, 4, &htid);
654 }
655 if (n->res.classid)
656 RTA_PUT(skb, TCA_U32_CLASSID, 4, &n->res.classid);
657 if (n->ht_down)
658 RTA_PUT(skb, TCA_U32_LINK, 4, &n->ht_down->handle);
659 #ifdef CONFIG_NET_CLS_POLICE
660 if (n->police) {
661 struct rtattr * p_rta = (struct rtattr*)skb->tail;
662
663 RTA_PUT(skb, TCA_U32_POLICE, 0, NULL);
664
665 if (tcf_police_dump(skb, n->police) < 0)
666 goto rtattr_failure;
667
668 p_rta->rta_len = skb->tail - (u8*)p_rta;
669 }
670 #endif
671 }
672
673 rta->rta_len = skb->tail - b;
674 #ifdef CONFIG_NET_CLS_POLICE
675 if (TC_U32_KEY(n->handle) && n->police) {
676 if (qdisc_copy_stats(skb, &n->police->stats))
677 goto rtattr_failure;
678 }
679 #endif
680 return skb->len;
681
682 rtattr_failure:
683 skb_trim(skb, b - skb->data);
684 return -1;
685 }
686
687 struct tcf_proto_ops cls_u32_ops = {
688 .next = NULL,
689 .kind = "u32",
690 .classify = u32_classify,
691 .init = u32_init,
692 .destroy = u32_destroy,
693 .get = u32_get,
694 .put = u32_put,
695 .change = u32_change,
696 .delete = u32_delete,
697 .walk = u32_walk,
698 .dump = u32_dump,
699 .owner = THIS_MODULE,
700 };
701
702 #ifdef MODULE
703 int init_module(void)
704 {
705 return register_tcf_proto_ops(&cls_u32_ops);
706 }
707
708 void cleanup_module(void)
709 {
710 unregister_tcf_proto_ops(&cls_u32_ops);
711 }
712 #endif
713 MODULE_LICENSE("GPL");
714
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.