bpf_filter.c revision 1.65 1 /* $NetBSD: bpf_filter.c,v 1.65 2014/06/25 09:51:34 alnsn Exp $ */
2
3 /*-
4 * Copyright (c) 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997
5 * The Regents of the University of California. All rights reserved.
6 *
7 * This code is derived from the Stanford/CMU enet packet filter,
8 * (net/enet.c) distributed as part of 4.3BSD, and code contributed
9 * to Berkeley by Steven McCanne and Van Jacobson both of Lawrence
10 * Berkeley Laboratory.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * SUCH DAMAGE.
35 *
36 * @(#)bpf_filter.c 8.1 (Berkeley) 6/10/93
37 */
38
39 #include <sys/cdefs.h>
40 __KERNEL_RCSID(0, "$NetBSD: bpf_filter.c,v 1.65 2014/06/25 09:51:34 alnsn Exp $");
41
42 #if 0
43 #if !(defined(lint) || defined(KERNEL))
44 static const char rcsid[] =
45 "@(#) Header: bpf_filter.c,v 1.33 97/04/26 13:37:18 leres Exp (LBL)";
46 #endif
47 #endif
48
49 #include <sys/param.h>
50 #include <sys/time.h>
51 #include <sys/kmem.h>
52 #include <sys/endian.h>
53
54 #define __BPF_PRIVATE
55 #include <net/bpf.h>
56
57 #ifdef _KERNEL
58
59 bpf_ctx_t *
60 bpf_create(void)
61 {
62 return kmem_zalloc(sizeof(bpf_ctx_t), KM_SLEEP);
63 }
64
65 void
66 bpf_destroy(bpf_ctx_t *bc)
67 {
68 kmem_free(bc, sizeof(bpf_ctx_t));
69 }
70
71 int
72 bpf_set_cop(bpf_ctx_t *bc, const bpf_copfunc_t *funcs, size_t n)
73 {
74 bc->copfuncs = funcs;
75 bc->nfuncs = n;
76 return 0;
77 }
78
79 int
80 bpf_set_extmem(bpf_ctx_t *bc, size_t nwords, bpf_memword_init_t preinited)
81 {
82 if (nwords > BPF_MAX_MEMWORDS || (preinited >> nwords) != 0) {
83 return EINVAL;
84 }
85 bc->extwords = nwords;
86 bc->preinited = preinited;
87 return 0;
88 }
89
90 #endif
91
92 #define EXTRACT_SHORT(p) be16dec(p)
93 #define EXTRACT_LONG(p) be32dec(p)
94
95 #ifdef _KERNEL
96 #include <sys/mbuf.h>
97 #define MINDEX(len, m, k) \
98 { \
99 len = m->m_len; \
100 while (k >= len) { \
101 k -= len; \
102 m = m->m_next; \
103 if (m == 0) \
104 return 0; \
105 len = m->m_len; \
106 } \
107 }
108
109 uint32_t m_xword(const struct mbuf *, uint32_t, int *);
110 uint32_t m_xhalf(const struct mbuf *, uint32_t, int *);
111 uint32_t m_xbyte(const struct mbuf *, uint32_t, int *);
112
113 #define xword(p, k, err) m_xword((const struct mbuf *)(p), (k), (err))
114 #define xhalf(p, k, err) m_xhalf((const struct mbuf *)(p), (k), (err))
115 #define xbyte(p, k, err) m_xbyte((const struct mbuf *)(p), (k), (err))
116
117 uint32_t
118 m_xword(const struct mbuf *m, uint32_t k, int *err)
119 {
120 int len;
121 u_char *cp, *np;
122 struct mbuf *m0;
123
124 *err = 1;
125 MINDEX(len, m, k);
126 cp = mtod(m, u_char *) + k;
127 if (len >= k + 4) {
128 *err = 0;
129 return EXTRACT_LONG(cp);
130 }
131 m0 = m->m_next;
132 if (m0 == 0 || m0->m_len + len - k < 4)
133 return 0;
134 *err = 0;
135 np = mtod(m0, u_char *);
136
137 switch (len - k) {
138 case 1:
139 return (cp[0] << 24) | (np[0] << 16) | (np[1] << 8) | np[2];
140 case 2:
141 return (cp[0] << 24) | (cp[1] << 16) | (np[0] << 8) | np[1];
142 default:
143 return (cp[0] << 24) | (cp[1] << 16) | (cp[2] << 8) | np[0];
144 }
145 }
146
147 uint32_t
148 m_xhalf(const struct mbuf *m, uint32_t k, int *err)
149 {
150 int len;
151 u_char *cp;
152 struct mbuf *m0;
153
154 *err = 1;
155 MINDEX(len, m, k);
156 cp = mtod(m, u_char *) + k;
157 if (len >= k + 2) {
158 *err = 0;
159 return EXTRACT_SHORT(cp);
160 }
161 m0 = m->m_next;
162 if (m0 == 0)
163 return 0;
164 *err = 0;
165 return (cp[0] << 8) | mtod(m0, u_char *)[0];
166 }
167
168 uint32_t
169 m_xbyte(const struct mbuf *m, uint32_t k, int *err)
170 {
171 int len;
172
173 *err = 0;
174 MINDEX(len, m, k);
175 return mtod(m, u_char *)[k];
176 }
177 #else /* _KERNEL */
178 #include <stdlib.h>
179 #endif /* !_KERNEL */
180
181 #include <net/bpf.h>
182
183 /*
184 * Execute the filter program starting at pc on the packet p
185 * wirelen is the length of the original packet
186 * buflen is the amount of data present
187 */
188 #ifdef _KERNEL
189
190 u_int
191 bpf_filter(const struct bpf_insn *pc, const u_char *p, u_int wirelen,
192 u_int buflen)
193 {
194 uint32_t mem[BPF_MEMWORDS];
195 bpf_args_t args = {
196 .pkt = p,
197 .wirelen = wirelen,
198 .buflen = buflen,
199 .mem = mem,
200 .arg = NULL
201 };
202
203 return bpf_filter_ext(NULL, pc, &args);
204 }
205
206 u_int
207 bpf_filter_ext(const bpf_ctx_t *bc, const struct bpf_insn *pc, bpf_args_t *args)
208 #else
209 u_int
210 bpf_filter(const struct bpf_insn *pc, const u_char *p, u_int wirelen,
211 u_int buflen)
212 #endif
213 {
214 uint32_t A, X, k;
215 #ifndef _KERNEL
216 uint32_t mem[BPF_MEMWORDS];
217 bpf_args_t args_store = {
218 .pkt = p,
219 .wirelen = wirelen,
220 .buflen = buflen,
221 .mem = mem,
222 .arg = NULL
223 };
224 bpf_args_t * const args = &args_store;
225 #else
226 const uint8_t * const p = args->pkt;
227 #endif
228 if (pc == 0) {
229 /*
230 * No filter means accept all.
231 */
232 return (u_int)-1;
233 }
234
235 /*
236 * Note: safe to leave memwords uninitialised, as the validation
237 * step ensures that it will not be read, if it was not written.
238 */
239 A = 0;
240 X = 0;
241 --pc;
242
243 for (;;) {
244 ++pc;
245 switch (pc->code) {
246
247 default:
248 #ifdef _KERNEL
249 return 0;
250 #else
251 abort();
252 /*NOTREACHED*/
253 #endif
254 case BPF_RET|BPF_K:
255 return (u_int)pc->k;
256
257 case BPF_RET|BPF_A:
258 return (u_int)A;
259
260 case BPF_LD|BPF_W|BPF_ABS:
261 k = pc->k;
262 if (k > args->buflen ||
263 sizeof(int32_t) > args->buflen - k) {
264 #ifdef _KERNEL
265 int merr;
266
267 if (args->buflen != 0)
268 return 0;
269 A = xword(args->pkt, k, &merr);
270 if (merr != 0)
271 return 0;
272 continue;
273 #else
274 return 0;
275 #endif
276 }
277 A = EXTRACT_LONG(&p[k]);
278 continue;
279
280 case BPF_LD|BPF_H|BPF_ABS:
281 k = pc->k;
282 if (k > args->buflen ||
283 sizeof(int16_t) > args->buflen - k) {
284 #ifdef _KERNEL
285 int merr;
286
287 if (args->buflen != 0)
288 return 0;
289 A = xhalf(args->pkt, k, &merr);
290 if (merr != 0)
291 return 0;
292 continue;
293 #else
294 return 0;
295 #endif
296 }
297 A = EXTRACT_SHORT(&p[k]);
298 continue;
299
300 case BPF_LD|BPF_B|BPF_ABS:
301 k = pc->k;
302 if (k >= args->buflen) {
303 #ifdef _KERNEL
304 int merr;
305
306 if (args->buflen != 0)
307 return 0;
308 A = xbyte(args->pkt, k, &merr);
309 continue;
310 #else
311 return 0;
312 #endif
313 }
314 A = p[k];
315 continue;
316
317 case BPF_LD|BPF_W|BPF_LEN:
318 A = args->wirelen;
319 continue;
320
321 case BPF_LDX|BPF_W|BPF_LEN:
322 X = args->wirelen;
323 continue;
324
325 case BPF_LD|BPF_W|BPF_IND:
326 k = X + pc->k;
327 if (pc->k > args->buflen ||
328 X > args->buflen - pc->k ||
329 sizeof(int32_t) > args->buflen - k) {
330 #ifdef _KERNEL
331 int merr;
332
333 if (args->buflen != 0)
334 return 0;
335 A = xword(args->pkt, k, &merr);
336 if (merr != 0)
337 return 0;
338 continue;
339 #else
340 return 0;
341 #endif
342 }
343 A = EXTRACT_LONG(&p[k]);
344 continue;
345
346 case BPF_LD|BPF_H|BPF_IND:
347 k = X + pc->k;
348 if (pc->k > args->buflen ||
349 X > args->buflen - pc->k ||
350 sizeof(int16_t) > args->buflen - k) {
351 #ifdef _KERNEL
352 int merr;
353
354 if (args->buflen != 0)
355 return 0;
356 A = xhalf(args->pkt, k, &merr);
357 if (merr != 0)
358 return 0;
359 continue;
360 #else
361 return 0;
362 #endif
363 }
364 A = EXTRACT_SHORT(&p[k]);
365 continue;
366
367 case BPF_LD|BPF_B|BPF_IND:
368 k = X + pc->k;
369 if (pc->k >= args->buflen ||
370 X >= args->buflen - pc->k) {
371 #ifdef _KERNEL
372 int merr;
373
374 if (args->buflen != 0)
375 return 0;
376 A = xbyte(args->pkt, k, &merr);
377 continue;
378 #else
379 return 0;
380 #endif
381 }
382 A = p[k];
383 continue;
384
385 case BPF_LDX|BPF_MSH|BPF_B:
386 k = pc->k;
387 if (k >= args->buflen) {
388 #ifdef _KERNEL
389 int merr;
390
391 if (args->buflen != 0)
392 return 0;
393 X = (xbyte(args->pkt, k, &merr) & 0xf) << 2;
394 continue;
395 #else
396 return 0;
397 #endif
398 }
399 X = (p[pc->k] & 0xf) << 2;
400 continue;
401
402 case BPF_LD|BPF_IMM:
403 A = pc->k;
404 continue;
405
406 case BPF_LDX|BPF_IMM:
407 X = pc->k;
408 continue;
409
410 case BPF_LD|BPF_MEM:
411 A = args->mem[pc->k];
412 continue;
413
414 case BPF_LDX|BPF_MEM:
415 X = args->mem[pc->k];
416 continue;
417
418 case BPF_ST:
419 args->mem[pc->k] = A;
420 continue;
421
422 case BPF_STX:
423 args->mem[pc->k] = X;
424 continue;
425
426 case BPF_JMP|BPF_JA:
427 pc += pc->k;
428 continue;
429
430 case BPF_JMP|BPF_JGT|BPF_K:
431 pc += (A > pc->k) ? pc->jt : pc->jf;
432 continue;
433
434 case BPF_JMP|BPF_JGE|BPF_K:
435 pc += (A >= pc->k) ? pc->jt : pc->jf;
436 continue;
437
438 case BPF_JMP|BPF_JEQ|BPF_K:
439 pc += (A == pc->k) ? pc->jt : pc->jf;
440 continue;
441
442 case BPF_JMP|BPF_JSET|BPF_K:
443 pc += (A & pc->k) ? pc->jt : pc->jf;
444 continue;
445
446 case BPF_JMP|BPF_JGT|BPF_X:
447 pc += (A > X) ? pc->jt : pc->jf;
448 continue;
449
450 case BPF_JMP|BPF_JGE|BPF_X:
451 pc += (A >= X) ? pc->jt : pc->jf;
452 continue;
453
454 case BPF_JMP|BPF_JEQ|BPF_X:
455 pc += (A == X) ? pc->jt : pc->jf;
456 continue;
457
458 case BPF_JMP|BPF_JSET|BPF_X:
459 pc += (A & X) ? pc->jt : pc->jf;
460 continue;
461
462 case BPF_ALU|BPF_ADD|BPF_X:
463 A += X;
464 continue;
465
466 case BPF_ALU|BPF_SUB|BPF_X:
467 A -= X;
468 continue;
469
470 case BPF_ALU|BPF_MUL|BPF_X:
471 A *= X;
472 continue;
473
474 case BPF_ALU|BPF_DIV|BPF_X:
475 if (X == 0)
476 return 0;
477 A /= X;
478 continue;
479
480 case BPF_ALU|BPF_AND|BPF_X:
481 A &= X;
482 continue;
483
484 case BPF_ALU|BPF_OR|BPF_X:
485 A |= X;
486 continue;
487
488 case BPF_ALU|BPF_LSH|BPF_X:
489 A <<= X;
490 continue;
491
492 case BPF_ALU|BPF_RSH|BPF_X:
493 A >>= X;
494 continue;
495
496 case BPF_ALU|BPF_ADD|BPF_K:
497 A += pc->k;
498 continue;
499
500 case BPF_ALU|BPF_SUB|BPF_K:
501 A -= pc->k;
502 continue;
503
504 case BPF_ALU|BPF_MUL|BPF_K:
505 A *= pc->k;
506 continue;
507
508 case BPF_ALU|BPF_DIV|BPF_K:
509 A /= pc->k;
510 continue;
511
512 case BPF_ALU|BPF_AND|BPF_K:
513 A &= pc->k;
514 continue;
515
516 case BPF_ALU|BPF_OR|BPF_K:
517 A |= pc->k;
518 continue;
519
520 case BPF_ALU|BPF_LSH|BPF_K:
521 A <<= pc->k;
522 continue;
523
524 case BPF_ALU|BPF_RSH|BPF_K:
525 A >>= pc->k;
526 continue;
527
528 case BPF_ALU|BPF_NEG:
529 A = -A;
530 continue;
531
532 case BPF_MISC|BPF_TAX:
533 X = A;
534 continue;
535
536 case BPF_MISC|BPF_TXA:
537 A = X;
538 continue;
539
540 case BPF_MISC|BPF_COP:
541 #ifdef _KERNEL
542 if (pc->k < bc->nfuncs) {
543 const bpf_copfunc_t fn = bc->copfuncs[pc->k];
544 A = fn(bc, args, A);
545 continue;
546 }
547 #endif
548 return 0;
549
550 case BPF_MISC|BPF_COPX:
551 #ifdef _KERNEL
552 if (X < bc->nfuncs) {
553 const bpf_copfunc_t fn = bc->copfuncs[X];
554 A = fn(bc, args, A);
555 continue;
556 }
557 #endif
558 return 0;
559 }
560 }
561 }
562
563 /*
564 * Return true if the 'fcode' is a valid filter program.
565 * The constraints are that each jump be forward and to a valid
566 * code, that memory accesses are within valid ranges (to the
567 * extent that this can be checked statically; loads of packet
568 * data have to be, and are, also checked at run time), and that
569 * the code terminates with either an accept or reject.
570 *
571 * The kernel needs to be able to verify an application's filter code.
572 * Otherwise, a bogus program could easily crash the system.
573 */
574
575 #if defined(KERNEL) || defined(_KERNEL)
576
577 int
578 bpf_validate(const struct bpf_insn *f, int signed_len)
579 {
580 return bpf_validate_ext(NULL, f, signed_len);
581 }
582
583 int
584 bpf_validate_ext(const bpf_ctx_t *bc, const struct bpf_insn *f, int signed_len)
585 #else
586 int
587 bpf_validate(const struct bpf_insn *f, int signed_len)
588 #endif
589 {
590 u_int i, from, len, ok = 0;
591 const struct bpf_insn *p;
592 #if defined(KERNEL) || defined(_KERNEL)
593 bpf_memword_init_t *mem, invalid;
594 size_t size;
595 const size_t extwords = bc ? bc->extwords : 0;
596 const size_t memwords = extwords ? extwords : BPF_MEMWORDS;
597 const bpf_memword_init_t preinited = extwords ? bc->preinited : 0;
598 #else
599 const size_t memwords = BPF_MEMWORDS;
600 #endif
601
602 len = (u_int)signed_len;
603 if (len < 1)
604 return 0;
605 #if defined(KERNEL) || defined(_KERNEL)
606 if (len > BPF_MAXINSNS)
607 return 0;
608 #endif
609 if (BPF_CLASS(f[len - 1].code) != BPF_RET)
610 return 0;
611
612 #if defined(KERNEL) || defined(_KERNEL)
613 /* Note: only the pre-initialised is valid on startup */
614 mem = kmem_zalloc(size = sizeof(*mem) * len, KM_SLEEP);
615 invalid = ~preinited;
616 #endif
617
618 for (i = 0; i < len; ++i) {
619 #if defined(KERNEL) || defined(_KERNEL)
620 /* blend in any invalid bits for current pc */
621 invalid |= mem[i];
622 #endif
623 p = &f[i];
624 switch (BPF_CLASS(p->code)) {
625 /*
626 * Check that memory operations use valid addresses.
627 */
628 case BPF_LD:
629 case BPF_LDX:
630 switch (BPF_MODE(p->code)) {
631 case BPF_MEM:
632 /*
633 * There's no maximum packet data size
634 * in userland. The runtime packet length
635 * check suffices.
636 */
637 #if defined(KERNEL) || defined(_KERNEL)
638 /*
639 * More strict check with actual packet length
640 * is done runtime.
641 */
642 if (p->k >= memwords)
643 goto out;
644 /* check for current memory invalid */
645 if (invalid & BPF_MEMWORD_INIT(p->k))
646 goto out;
647 #endif
648 break;
649 case BPF_ABS:
650 case BPF_IND:
651 case BPF_MSH:
652 case BPF_IMM:
653 case BPF_LEN:
654 break;
655 default:
656 goto out;
657 }
658 break;
659 case BPF_ST:
660 case BPF_STX:
661 if (p->k >= memwords)
662 goto out;
663 #if defined(KERNEL) || defined(_KERNEL)
664 /* validate the memory word */
665 invalid &= ~BPF_MEMWORD_INIT(p->k);
666 #endif
667 break;
668 case BPF_ALU:
669 switch (BPF_OP(p->code)) {
670 case BPF_ADD:
671 case BPF_SUB:
672 case BPF_MUL:
673 case BPF_OR:
674 case BPF_AND:
675 case BPF_LSH:
676 case BPF_RSH:
677 case BPF_NEG:
678 break;
679 case BPF_DIV:
680 /*
681 * Check for constant division by 0.
682 */
683 if (BPF_SRC(p->code) == BPF_K && p->k == 0)
684 goto out;
685 break;
686 default:
687 goto out;
688 }
689 break;
690 case BPF_JMP:
691 /*
692 * Check that jumps are within the code block,
693 * and that unconditional branches don't go
694 * backwards as a result of an overflow.
695 * Unconditional branches have a 32-bit offset,
696 * so they could overflow; we check to make
697 * sure they don't. Conditional branches have
698 * an 8-bit offset, and the from address is <=
699 * BPF_MAXINSNS, and we assume that BPF_MAXINSNS
700 * is sufficiently small that adding 255 to it
701 * won't overflow.
702 *
703 * We know that len is <= BPF_MAXINSNS, and we
704 * assume that BPF_MAXINSNS is < the maximum size
705 * of a u_int, so that i + 1 doesn't overflow.
706 *
707 * For userland, we don't know that the from
708 * or len are <= BPF_MAXINSNS, but we know that
709 * from <= len, and, except on a 64-bit system,
710 * it's unlikely that len, if it truly reflects
711 * the size of the program we've been handed,
712 * will be anywhere near the maximum size of
713 * a u_int. We also don't check for backward
714 * branches, as we currently support them in
715 * userland for the protochain operation.
716 */
717 from = i + 1;
718 switch (BPF_OP(p->code)) {
719 case BPF_JA:
720 if (from + p->k >= len)
721 goto out;
722 #if defined(KERNEL) || defined(_KERNEL)
723 if (from + p->k < from)
724 goto out;
725 /*
726 * mark the currently invalid bits for the
727 * destination
728 */
729 mem[from + p->k] |= invalid;
730 invalid = 0;
731 #endif
732 break;
733 case BPF_JEQ:
734 case BPF_JGT:
735 case BPF_JGE:
736 case BPF_JSET:
737 if (from + p->jt >= len || from + p->jf >= len)
738 goto out;
739 #if defined(KERNEL) || defined(_KERNEL)
740 /*
741 * mark the currently invalid bits for both
742 * possible jump destinations
743 */
744 mem[from + p->jt] |= invalid;
745 mem[from + p->jf] |= invalid;
746 invalid = 0;
747 #endif
748 break;
749 default:
750 goto out;
751 }
752 break;
753 case BPF_RET:
754 break;
755 case BPF_MISC:
756 switch (BPF_MISCOP(p->code)) {
757 case BPF_COP:
758 case BPF_COPX:
759 /* In-kernel COP use only. */
760 #if defined(KERNEL) || defined(_KERNEL)
761 if (bc == NULL || bc->copfuncs == NULL)
762 goto out;
763 if (BPF_MISCOP(p->code) == BPF_COP &&
764 p->k >= bc->nfuncs) {
765 goto out;
766 }
767 break;
768 #else
769 goto out;
770 #endif
771 default:
772 break;
773 }
774 break;
775 default:
776 goto out;
777 }
778 }
779 ok = 1;
780 out:
781 #if defined(KERNEL) || defined(_KERNEL)
782 kmem_free(mem, size);
783 #endif
784 return ok;
785 }
786