filter.c 48 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3
/*
 * Linux Socket Filter - Kernel level socket filtering
 *
4 5
 * Based on the design of the Berkeley Packet Filter. The new
 * internal format has been designed by PLUMgrid:
Linus Torvalds's avatar
Linus Torvalds committed
6
 *
7 8 9 10 11 12 13
 *	Copyright (c) 2011 - 2014 PLUMgrid, http://plumgrid.com
 *
 * Authors:
 *
 *	Jay Schulist <jschlst@samba.org>
 *	Alexei Starovoitov <ast@plumgrid.com>
 *	Daniel Borkmann <dborkman@redhat.com>
Linus Torvalds's avatar
Linus Torvalds committed
14 15 16 17 18 19 20
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 *
 * Andi Kleen - Fix a few bad bugs and races.
21
 * Kris Katterjohn - Added many additional checks in bpf_check_classic()
Linus Torvalds's avatar
Linus Torvalds committed
22 23 24 25 26 27 28 29 30 31 32
 */

#include <linux/module.h>
#include <linux/types.h>
#include <linux/mm.h>
#include <linux/fcntl.h>
#include <linux/socket.h>
#include <linux/in.h>
#include <linux/inet.h>
#include <linux/netdevice.h>
#include <linux/if_packet.h>
33
#include <linux/gfp.h>
Linus Torvalds's avatar
Linus Torvalds committed
34 35
#include <net/ip.h>
#include <net/protocol.h>
36
#include <net/netlink.h>
Linus Torvalds's avatar
Linus Torvalds committed
37 38
#include <linux/skbuff.h>
#include <net/sock.h>
39
#include <net/flow_dissector.h>
Linus Torvalds's avatar
Linus Torvalds committed
40 41 42
#include <linux/errno.h>
#include <linux/timer.h>
#include <asm/uaccess.h>
43
#include <asm/unaligned.h>
Linus Torvalds's avatar
Linus Torvalds committed
44
#include <linux/filter.h>
45
#include <linux/ratelimit.h>
46
#include <linux/seccomp.h>
47
#include <linux/if_vlan.h>
48
#include <linux/bpf.h>
49
#include <net/sch_generic.h>
50
#include <net/cls_cgroup.h>
51
#include <net/dst_metadata.h>
Linus Torvalds's avatar
Linus Torvalds committed
52

53 54 55 56 57 58
/**
 *	sk_filter - run a packet through a socket filter
 *	@sk: sock associated with &sk_buff
 *	@skb: buffer to filter
 *
 * Run the filter code and then cut skb->data to correct size returned by
Li RongQing's avatar
Li RongQing committed
59
 * SK_RUN_FILTER. If pkt_len is 0 we toss packet. If skb->len is smaller
60
 * than pkt_len we keep whole skb->data. This is the socket level
Li RongQing's avatar
Li RongQing committed
61
 * wrapper to SK_RUN_FILTER. It returns 0 if the packet should
62 63 64 65 66 67 68 69
 * be accepted or -EPERM if the packet should be tossed.
 *
 */
int sk_filter(struct sock *sk, struct sk_buff *skb)
{
	int err;
	struct sk_filter *filter;

70 71 72 73 74 75 76 77
	/*
	 * If the skb was allocated from pfmemalloc reserves, only
	 * allow SOCK_MEMALLOC sockets to use it as this socket is
	 * helping free memory
	 */
	if (skb_pfmemalloc(skb) && !sock_flag(sk, SOCK_MEMALLOC))
		return -ENOMEM;

78 79 80 81
	err = security_sock_rcv_skb(sk, skb);
	if (err)
		return err;

82 83
	rcu_read_lock();
	filter = rcu_dereference(sk->sk_filter);
84
	if (filter) {
85
		unsigned int pkt_len = SK_RUN_FILTER(filter, skb);
86

87 88
		err = pkt_len ? pskb_trim(skb, pkt_len) : -EPERM;
	}
89
	rcu_read_unlock();
90 91 92 93 94

	return err;
}
EXPORT_SYMBOL(sk_filter);

95
static u64 __skb_get_pay_offset(u64 ctx, u64 a, u64 x, u64 r4, u64 r5)
96
{
97
	return skb_get_poff((struct sk_buff *)(unsigned long) ctx);
98 99
}

100
static u64 __skb_get_nlattr(u64 ctx, u64 a, u64 x, u64 r4, u64 r5)
101
{
102
	struct sk_buff *skb = (struct sk_buff *)(unsigned long) ctx;
103 104 105 106 107
	struct nlattr *nla;

	if (skb_is_nonlinear(skb))
		return 0;

108 109 110
	if (skb->len < sizeof(struct nlattr))
		return 0;

111
	if (a > skb->len - sizeof(struct nlattr))
112 113
		return 0;

114
	nla = nla_find((struct nlattr *) &skb->data[a], skb->len - a, x);
115 116 117 118 119 120
	if (nla)
		return (void *) nla - (void *) skb->data;

	return 0;
}

121
static u64 __skb_get_nlattr_nest(u64 ctx, u64 a, u64 x, u64 r4, u64 r5)
122
{
123
	struct sk_buff *skb = (struct sk_buff *)(unsigned long) ctx;
124 125 126 127 128
	struct nlattr *nla;

	if (skb_is_nonlinear(skb))
		return 0;

129 130 131
	if (skb->len < sizeof(struct nlattr))
		return 0;

132
	if (a > skb->len - sizeof(struct nlattr))
133 134
		return 0;

135 136
	nla = (struct nlattr *) &skb->data[a];
	if (nla->nla_len > skb->len - a)
137 138
		return 0;

139
	nla = nla_find_nested(nla, x);
140 141 142 143 144 145
	if (nla)
		return (void *) nla - (void *) skb->data;

	return 0;
}

146
static u64 __get_raw_cpu_id(u64 ctx, u64 a, u64 x, u64 r4, u64 r5)
147 148 149 150
{
	return raw_smp_processor_id();
}

151
/* note that this only generates 32-bit random numbers */
152
static u64 __get_random_u32(u64 ctx, u64 a, u64 x, u64 r4, u64 r5)
153
{
154
	return prandom_u32();
155 156
}

157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183
static u32 convert_skb_access(int skb_field, int dst_reg, int src_reg,
			      struct bpf_insn *insn_buf)
{
	struct bpf_insn *insn = insn_buf;

	switch (skb_field) {
	case SKF_AD_MARK:
		BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, mark) != 4);

		*insn++ = BPF_LDX_MEM(BPF_W, dst_reg, src_reg,
				      offsetof(struct sk_buff, mark));
		break;

	case SKF_AD_PKTTYPE:
		*insn++ = BPF_LDX_MEM(BPF_B, dst_reg, src_reg, PKT_TYPE_OFFSET());
		*insn++ = BPF_ALU32_IMM(BPF_AND, dst_reg, PKT_TYPE_MAX);
#ifdef __BIG_ENDIAN_BITFIELD
		*insn++ = BPF_ALU32_IMM(BPF_RSH, dst_reg, 5);
#endif
		break;

	case SKF_AD_QUEUE:
		BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, queue_mapping) != 2);

		*insn++ = BPF_LDX_MEM(BPF_H, dst_reg, src_reg,
				      offsetof(struct sk_buff, queue_mapping));
		break;
184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202

	case SKF_AD_VLAN_TAG:
	case SKF_AD_VLAN_TAG_PRESENT:
		BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, vlan_tci) != 2);
		BUILD_BUG_ON(VLAN_TAG_PRESENT != 0x1000);

		/* dst_reg = *(u16 *) (src_reg + offsetof(vlan_tci)) */
		*insn++ = BPF_LDX_MEM(BPF_H, dst_reg, src_reg,
				      offsetof(struct sk_buff, vlan_tci));
		if (skb_field == SKF_AD_VLAN_TAG) {
			*insn++ = BPF_ALU32_IMM(BPF_AND, dst_reg,
						~VLAN_TAG_PRESENT);
		} else {
			/* dst_reg >>= 12 */
			*insn++ = BPF_ALU32_IMM(BPF_RSH, dst_reg, 12);
			/* dst_reg &= 1 */
			*insn++ = BPF_ALU32_IMM(BPF_AND, dst_reg, 1);
		}
		break;
203 204 205 206 207
	}

	return insn - insn_buf;
}

208
static bool convert_bpf_extensions(struct sock_filter *fp,
209
				   struct bpf_insn **insnp)
210
{
211
	struct bpf_insn *insn = *insnp;
212
	u32 cnt;
213 214 215

	switch (fp->k) {
	case SKF_AD_OFF + SKF_AD_PROTOCOL:
216 217 218 219 220 221 222
		BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, protocol) != 2);

		/* A = *(u16 *) (CTX + offsetof(protocol)) */
		*insn++ = BPF_LDX_MEM(BPF_H, BPF_REG_A, BPF_REG_CTX,
				      offsetof(struct sk_buff, protocol));
		/* A = ntohs(A) [emitting a nop or swap16] */
		*insn = BPF_ENDIAN(BPF_FROM_BE, BPF_REG_A, 16);
223 224 225
		break;

	case SKF_AD_OFF + SKF_AD_PKTTYPE:
226 227
		cnt = convert_skb_access(SKF_AD_PKTTYPE, BPF_REG_A, BPF_REG_CTX, insn);
		insn += cnt - 1;
228 229 230 231 232 233
		break;

	case SKF_AD_OFF + SKF_AD_IFINDEX:
	case SKF_AD_OFF + SKF_AD_HATYPE:
		BUILD_BUG_ON(FIELD_SIZEOF(struct net_device, ifindex) != 4);
		BUILD_BUG_ON(FIELD_SIZEOF(struct net_device, type) != 2);
234 235 236 237 238 239 240 241 242 243 244 245 246 247
		BUILD_BUG_ON(bytes_to_bpf_size(FIELD_SIZEOF(struct sk_buff, dev)) < 0);

		*insn++ = BPF_LDX_MEM(bytes_to_bpf_size(FIELD_SIZEOF(struct sk_buff, dev)),
				      BPF_REG_TMP, BPF_REG_CTX,
				      offsetof(struct sk_buff, dev));
		/* if (tmp != 0) goto pc + 1 */
		*insn++ = BPF_JMP_IMM(BPF_JNE, BPF_REG_TMP, 0, 1);
		*insn++ = BPF_EXIT_INSN();
		if (fp->k == SKF_AD_OFF + SKF_AD_IFINDEX)
			*insn = BPF_LDX_MEM(BPF_W, BPF_REG_A, BPF_REG_TMP,
					    offsetof(struct net_device, ifindex));
		else
			*insn = BPF_LDX_MEM(BPF_H, BPF_REG_A, BPF_REG_TMP,
					    offsetof(struct net_device, type));
248 249 250
		break;

	case SKF_AD_OFF + SKF_AD_MARK:
251 252
		cnt = convert_skb_access(SKF_AD_MARK, BPF_REG_A, BPF_REG_CTX, insn);
		insn += cnt - 1;
253 254 255 256 257
		break;

	case SKF_AD_OFF + SKF_AD_RXHASH:
		BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, hash) != 4);

258 259
		*insn = BPF_LDX_MEM(BPF_W, BPF_REG_A, BPF_REG_CTX,
				    offsetof(struct sk_buff, hash));
260 261 262
		break;

	case SKF_AD_OFF + SKF_AD_QUEUE:
263 264
		cnt = convert_skb_access(SKF_AD_QUEUE, BPF_REG_A, BPF_REG_CTX, insn);
		insn += cnt - 1;
265 266 267
		break;

	case SKF_AD_OFF + SKF_AD_VLAN_TAG:
268 269 270 271
		cnt = convert_skb_access(SKF_AD_VLAN_TAG,
					 BPF_REG_A, BPF_REG_CTX, insn);
		insn += cnt - 1;
		break;
272

273 274 275 276
	case SKF_AD_OFF + SKF_AD_VLAN_TAG_PRESENT:
		cnt = convert_skb_access(SKF_AD_VLAN_TAG_PRESENT,
					 BPF_REG_A, BPF_REG_CTX, insn);
		insn += cnt - 1;
277 278
		break;

279 280 281 282 283 284 285 286 287 288
	case SKF_AD_OFF + SKF_AD_VLAN_TPID:
		BUILD_BUG_ON(FIELD_SIZEOF(struct sk_buff, vlan_proto) != 2);

		/* A = *(u16 *) (CTX + offsetof(vlan_proto)) */
		*insn++ = BPF_LDX_MEM(BPF_H, BPF_REG_A, BPF_REG_CTX,
				      offsetof(struct sk_buff, vlan_proto));
		/* A = ntohs(A) [emitting a nop or swap16] */
		*insn = BPF_ENDIAN(BPF_FROM_BE, BPF_REG_A, 16);
		break;

289 290 291 292
	case SKF_AD_OFF + SKF_AD_PAY_OFFSET:
	case SKF_AD_OFF + SKF_AD_NLATTR:
	case SKF_AD_OFF + SKF_AD_NLATTR_NEST:
	case SKF_AD_OFF + SKF_AD_CPU:
293
	case SKF_AD_OFF + SKF_AD_RANDOM:
294
		/* arg1 = CTX */
295
		*insn++ = BPF_MOV64_REG(BPF_REG_ARG1, BPF_REG_CTX);
296
		/* arg2 = A */
297
		*insn++ = BPF_MOV64_REG(BPF_REG_ARG2, BPF_REG_A);
298
		/* arg3 = X */
299
		*insn++ = BPF_MOV64_REG(BPF_REG_ARG3, BPF_REG_X);
300
		/* Emit call(arg1=CTX, arg2=A, arg3=X) */
301 302
		switch (fp->k) {
		case SKF_AD_OFF + SKF_AD_PAY_OFFSET:
303
			*insn = BPF_EMIT_CALL(__skb_get_pay_offset);
304 305
			break;
		case SKF_AD_OFF + SKF_AD_NLATTR:
306
			*insn = BPF_EMIT_CALL(__skb_get_nlattr);
307 308
			break;
		case SKF_AD_OFF + SKF_AD_NLATTR_NEST:
309
			*insn = BPF_EMIT_CALL(__skb_get_nlattr_nest);
310 311
			break;
		case SKF_AD_OFF + SKF_AD_CPU:
312
			*insn = BPF_EMIT_CALL(__get_raw_cpu_id);
313
			break;
314
		case SKF_AD_OFF + SKF_AD_RANDOM:
315
			*insn = BPF_EMIT_CALL(__get_random_u32);
316
			break;
317 318 319 320
		}
		break;

	case SKF_AD_OFF + SKF_AD_ALU_XOR_X:
321 322
		/* A ^= X */
		*insn = BPF_ALU32_REG(BPF_XOR, BPF_REG_A, BPF_REG_X);
323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338
		break;

	default:
		/* This is just a dummy call to avoid letting the compiler
		 * evict __bpf_call_base() as an optimization. Placed here
		 * where no-one bothers.
		 */
		BUG_ON(__bpf_call_base(0, 0, 0, 0, 0) != 0);
		return false;
	}

	*insnp = insn;
	return true;
}

/**
339
 *	bpf_convert_filter - convert filter program
340 341 342 343 344 345 346 347 348
 *	@prog: the user passed filter program
 *	@len: the length of the user passed filter program
 *	@new_prog: buffer where converted program will be stored
 *	@new_len: pointer to store length of converted program
 *
 * Remap 'sock_filter' style BPF instruction set to 'sock_filter_ext' style.
 * Conversion workflow:
 *
 * 1) First pass for calculating the new program length:
349
 *   bpf_convert_filter(old_prog, old_len, NULL, &new_len)
350 351 352
 *
 * 2) 2nd pass to remap in two passes: 1st pass finds new
 *    jump offsets, 2nd pass remapping:
353
 *   new_prog = kmalloc(sizeof(struct bpf_insn) * new_len);
354
 *   bpf_convert_filter(old_prog, old_len, new_prog, &new_len);
355 356 357 358 359 360 361
 *
 * User BPF's register A is mapped to our BPF register 6, user BPF
 * register X is mapped to BPF register 7; frame pointer is always
 * register 10; Context 'void *ctx' is stored in register 1, that is,
 * for socket filters: ctx == 'struct sk_buff *', for seccomp:
 * ctx == 'struct seccomp_data *'.
 */
362 363
static int bpf_convert_filter(struct sock_filter *prog, int len,
			      struct bpf_insn *new_prog, int *new_len)
364 365
{
	int new_flen = 0, pass = 0, target, i;
366
	struct bpf_insn *new_insn;
367 368 369 370 371
	struct sock_filter *fp;
	int *addrs = NULL;
	u8 bpf_src;

	BUILD_BUG_ON(BPF_MEMWORDS * sizeof(u32) > MAX_BPF_STACK);
372
	BUILD_BUG_ON(BPF_REG_FP + 1 != MAX_BPF_REG);
373

374
	if (len <= 0 || len > BPF_MAXINSNS)
375 376 377
		return -EINVAL;

	if (new_prog) {
378 379
		addrs = kcalloc(len, sizeof(*addrs),
				GFP_KERNEL | __GFP_NOWARN);
380 381 382 383 384 385 386 387
		if (!addrs)
			return -ENOMEM;
	}

do_pass:
	new_insn = new_prog;
	fp = prog;

388 389
	if (new_insn)
		*new_insn = BPF_MOV64_REG(BPF_REG_CTX, BPF_REG_ARG1);
390 391 392
	new_insn++;

	for (i = 0; i < len; fp++, i++) {
393 394
		struct bpf_insn tmp_insns[6] = { };
		struct bpf_insn *insn = tmp_insns;
395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436

		if (addrs)
			addrs[i] = new_insn - new_prog;

		switch (fp->code) {
		/* All arithmetic insns and skb loads map as-is. */
		case BPF_ALU | BPF_ADD | BPF_X:
		case BPF_ALU | BPF_ADD | BPF_K:
		case BPF_ALU | BPF_SUB | BPF_X:
		case BPF_ALU | BPF_SUB | BPF_K:
		case BPF_ALU | BPF_AND | BPF_X:
		case BPF_ALU | BPF_AND | BPF_K:
		case BPF_ALU | BPF_OR | BPF_X:
		case BPF_ALU | BPF_OR | BPF_K:
		case BPF_ALU | BPF_LSH | BPF_X:
		case BPF_ALU | BPF_LSH | BPF_K:
		case BPF_ALU | BPF_RSH | BPF_X:
		case BPF_ALU | BPF_RSH | BPF_K:
		case BPF_ALU | BPF_XOR | BPF_X:
		case BPF_ALU | BPF_XOR | BPF_K:
		case BPF_ALU | BPF_MUL | BPF_X:
		case BPF_ALU | BPF_MUL | BPF_K:
		case BPF_ALU | BPF_DIV | BPF_X:
		case BPF_ALU | BPF_DIV | BPF_K:
		case BPF_ALU | BPF_MOD | BPF_X:
		case BPF_ALU | BPF_MOD | BPF_K:
		case BPF_ALU | BPF_NEG:
		case BPF_LD | BPF_ABS | BPF_W:
		case BPF_LD | BPF_ABS | BPF_H:
		case BPF_LD | BPF_ABS | BPF_B:
		case BPF_LD | BPF_IND | BPF_W:
		case BPF_LD | BPF_IND | BPF_H:
		case BPF_LD | BPF_IND | BPF_B:
			/* Check for overloaded BPF extension and
			 * directly convert it if found, otherwise
			 * just move on with mapping.
			 */
			if (BPF_CLASS(fp->code) == BPF_LD &&
			    BPF_MODE(fp->code) == BPF_ABS &&
			    convert_bpf_extensions(fp, &insn))
				break;

437
			*insn = BPF_RAW_INSN(fp->code, BPF_REG_A, BPF_REG_X, 0, fp->k);
438 439
			break;

440 441 442 443 444 445 446
		/* Jump transformation cannot use BPF block macros
		 * everywhere as offset calculation and target updates
		 * require a bit more work than the rest, i.e. jump
		 * opcodes map as-is, but offsets need adjustment.
		 */

#define BPF_EMIT_JMP							\
447 448 449 450 451 452 453 454
	do {								\
		if (target >= len || target < 0)			\
			goto err;					\
		insn->off = addrs ? addrs[target] - addrs[i] - 1 : 0;	\
		/* Adjust pc relative offset for 2nd or 3rd insn. */	\
		insn->off -= insn - tmp_insns;				\
	} while (0)

455 456 457 458
		case BPF_JMP | BPF_JA:
			target = i + fp->k + 1;
			insn->code = fp->code;
			BPF_EMIT_JMP;
459 460 461 462 463 464 465 466 467 468 469 470 471 472 473
			break;

		case BPF_JMP | BPF_JEQ | BPF_K:
		case BPF_JMP | BPF_JEQ | BPF_X:
		case BPF_JMP | BPF_JSET | BPF_K:
		case BPF_JMP | BPF_JSET | BPF_X:
		case BPF_JMP | BPF_JGT | BPF_K:
		case BPF_JMP | BPF_JGT | BPF_X:
		case BPF_JMP | BPF_JGE | BPF_K:
		case BPF_JMP | BPF_JGE | BPF_X:
			if (BPF_SRC(fp->code) == BPF_K && (int) fp->k < 0) {
				/* BPF immediates are signed, zero extend
				 * immediate into tmp register and use it
				 * in compare insn.
				 */
474
				*insn++ = BPF_MOV32_IMM(BPF_REG_TMP, fp->k);
475

476 477
				insn->dst_reg = BPF_REG_A;
				insn->src_reg = BPF_REG_TMP;
478 479
				bpf_src = BPF_X;
			} else {
480 481
				insn->dst_reg = BPF_REG_A;
				insn->src_reg = BPF_REG_X;
482 483
				insn->imm = fp->k;
				bpf_src = BPF_SRC(fp->code);
Linus Torvalds's avatar
Linus Torvalds committed
484
			}
485 486 487 488 489

			/* Common case where 'jump_false' is next insn. */
			if (fp->jf == 0) {
				insn->code = BPF_JMP | BPF_OP(fp->code) | bpf_src;
				target = i + fp->jt + 1;
490
				BPF_EMIT_JMP;
491
				break;
Linus Torvalds's avatar
Linus Torvalds committed
492
			}
493 494 495 496 497

			/* Convert JEQ into JNE when 'jump_true' is next insn. */
			if (fp->jt == 0 && BPF_OP(fp->code) == BPF_JEQ) {
				insn->code = BPF_JMP | BPF_JNE | bpf_src;
				target = i + fp->jf + 1;
498
				BPF_EMIT_JMP;
499
				break;
500
			}
501 502 503 504

			/* Other jumps are mapped into two insns: Jxx and JA. */
			target = i + fp->jt + 1;
			insn->code = BPF_JMP | BPF_OP(fp->code) | bpf_src;
505
			BPF_EMIT_JMP;
506 507 508 509
			insn++;

			insn->code = BPF_JMP | BPF_JA;
			target = i + fp->jf + 1;
510
			BPF_EMIT_JMP;
511 512 513 514
			break;

		/* ldxb 4 * ([14] & 0xf) is remaped into 6 insns. */
		case BPF_LDX | BPF_MSH | BPF_B:
515
			/* tmp = A */
516
			*insn++ = BPF_MOV64_REG(BPF_REG_TMP, BPF_REG_A);
517
			/* A = BPF_R0 = *(u8 *) (skb->data + K) */
518
			*insn++ = BPF_LD_ABS(BPF_B, fp->k);
519
			/* A &= 0xf */
520
			*insn++ = BPF_ALU32_IMM(BPF_AND, BPF_REG_A, 0xf);
521
			/* A <<= 2 */
522
			*insn++ = BPF_ALU32_IMM(BPF_LSH, BPF_REG_A, 2);
523
			/* X = A */
524
			*insn++ = BPF_MOV64_REG(BPF_REG_X, BPF_REG_A);
525
			/* A = tmp */
526
			*insn = BPF_MOV64_REG(BPF_REG_A, BPF_REG_TMP);
527 528 529 530 531
			break;

		/* RET_K, RET_A are remaped into 2 insns. */
		case BPF_RET | BPF_A:
		case BPF_RET | BPF_K:
532 533 534
			*insn++ = BPF_MOV32_RAW(BPF_RVAL(fp->code) == BPF_K ?
						BPF_K : BPF_X, BPF_REG_0,
						BPF_REG_A, fp->k);
535
			*insn = BPF_EXIT_INSN();
536 537 538 539 540
			break;

		/* Store to stack. */
		case BPF_ST:
		case BPF_STX:
541 542 543
			*insn = BPF_STX_MEM(BPF_W, BPF_REG_FP, BPF_CLASS(fp->code) ==
					    BPF_ST ? BPF_REG_A : BPF_REG_X,
					    -(BPF_MEMWORDS - fp->k) * 4);
544 545 546 547 548
			break;

		/* Load from stack. */
		case BPF_LD | BPF_MEM:
		case BPF_LDX | BPF_MEM:
549 550 551
			*insn = BPF_LDX_MEM(BPF_W, BPF_CLASS(fp->code) == BPF_LD  ?
					    BPF_REG_A : BPF_REG_X, BPF_REG_FP,
					    -(BPF_MEMWORDS - fp->k) * 4);
552 553 554 555 556
			break;

		/* A = K or X = K */
		case BPF_LD | BPF_IMM:
		case BPF_LDX | BPF_IMM:
557 558
			*insn = BPF_MOV32_IMM(BPF_CLASS(fp->code) == BPF_LD ?
					      BPF_REG_A : BPF_REG_X, fp->k);
559 560 561 562
			break;

		/* X = A */
		case BPF_MISC | BPF_TAX:
563
			*insn = BPF_MOV64_REG(BPF_REG_X, BPF_REG_A);
564 565 566 567
			break;

		/* A = X */
		case BPF_MISC | BPF_TXA:
568
			*insn = BPF_MOV64_REG(BPF_REG_A, BPF_REG_X);
569 570 571 572 573
			break;

		/* A = skb->len or X = skb->len */
		case BPF_LD | BPF_W | BPF_LEN:
		case BPF_LDX | BPF_W | BPF_LEN:
574 575 576
			*insn = BPF_LDX_MEM(BPF_W, BPF_CLASS(fp->code) == BPF_LD ?
					    BPF_REG_A : BPF_REG_X, BPF_REG_CTX,
					    offsetof(struct sk_buff, len));
577 578
			break;

579
		/* Access seccomp_data fields. */
580
		case BPF_LDX | BPF_ABS | BPF_W:
581 582
			/* A = *(u32 *) (ctx + K) */
			*insn = BPF_LDX_MEM(BPF_W, BPF_REG_A, BPF_REG_CTX, fp->k);
583 584
			break;

Stephen Hemminger's avatar
Stephen Hemminger committed
585
		/* Unknown instruction. */
Linus Torvalds's avatar
Linus Torvalds committed
586
		default:
587
			goto err;
Linus Torvalds's avatar
Linus Torvalds committed
588
		}
589 590 591 592 593 594

		insn++;
		if (new_prog)
			memcpy(new_insn, tmp_insns,
			       sizeof(*insn) * (insn - tmp_insns));
		new_insn += insn - tmp_insns;
Linus Torvalds's avatar
Linus Torvalds committed
595 596
	}

597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612
	if (!new_prog) {
		/* Only calculating new length. */
		*new_len = new_insn - new_prog;
		return 0;
	}

	pass++;
	if (new_flen != new_insn - new_prog) {
		new_flen = new_insn - new_prog;
		if (pass > 2)
			goto err;
		goto do_pass;
	}

	kfree(addrs);
	BUG_ON(*new_len != new_flen);
Linus Torvalds's avatar
Linus Torvalds committed
613
	return 0;
614 615 616
err:
	kfree(addrs);
	return -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
617 618
}

619 620
/* Security:
 *
621
 * As we dont want to clear mem[] array for each packet going through
Li RongQing's avatar
Li RongQing committed
622
 * __bpf_prog_run(), we check that filter loaded by user never try to read
623
 * a cell if not previously written, and we check all branches to be sure
Lucas De Marchi's avatar
Lucas De Marchi committed
624
 * a malicious user doesn't try to abuse us.
625
 */
626
static int check_load_and_stores(const struct sock_filter *filter, int flen)
627
{
628
	u16 *masks, memvalid = 0; /* One bit per cell, 16 cells */
629 630 631
	int pc, ret = 0;

	BUILD_BUG_ON(BPF_MEMWORDS > 16);
632

633
	masks = kmalloc_array(flen, sizeof(*masks), GFP_KERNEL);
634 635
	if (!masks)
		return -ENOMEM;
636

637 638 639 640 641 642
	memset(masks, 0xff, flen * sizeof(*masks));

	for (pc = 0; pc < flen; pc++) {
		memvalid &= masks[pc];

		switch (filter[pc].code) {
643 644
		case BPF_ST:
		case BPF_STX:
645 646
			memvalid |= (1 << filter[pc].k);
			break;
647 648
		case BPF_LD | BPF_MEM:
		case BPF_LDX | BPF_MEM:
649 650 651 652 653
			if (!(memvalid & (1 << filter[pc].k))) {
				ret = -EINVAL;
				goto error;
			}
			break;
654 655
		case BPF_JMP | BPF_JA:
			/* A jump must set masks on target */
656 657 658
			masks[pc + 1 + filter[pc].k] &= memvalid;
			memvalid = ~0;
			break;
659 660 661 662 663 664 665 666 667
		case BPF_JMP | BPF_JEQ | BPF_K:
		case BPF_JMP | BPF_JEQ | BPF_X:
		case BPF_JMP | BPF_JGE | BPF_K:
		case BPF_JMP | BPF_JGE | BPF_X:
		case BPF_JMP | BPF_JGT | BPF_K:
		case BPF_JMP | BPF_JGT | BPF_X:
		case BPF_JMP | BPF_JSET | BPF_K:
		case BPF_JMP | BPF_JSET | BPF_X:
			/* A jump must set masks on targets */
668 669 670 671 672 673 674 675 676 677 678
			masks[pc + 1 + filter[pc].jt] &= memvalid;
			masks[pc + 1 + filter[pc].jf] &= memvalid;
			memvalid = ~0;
			break;
		}
	}
error:
	kfree(masks);
	return ret;
}

679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
static bool chk_code_allowed(u16 code_to_probe)
{
	static const bool codes[] = {
		/* 32 bit ALU operations */
		[BPF_ALU | BPF_ADD | BPF_K] = true,
		[BPF_ALU | BPF_ADD | BPF_X] = true,
		[BPF_ALU | BPF_SUB | BPF_K] = true,
		[BPF_ALU | BPF_SUB | BPF_X] = true,
		[BPF_ALU | BPF_MUL | BPF_K] = true,
		[BPF_ALU | BPF_MUL | BPF_X] = true,
		[BPF_ALU | BPF_DIV | BPF_K] = true,
		[BPF_ALU | BPF_DIV | BPF_X] = true,
		[BPF_ALU | BPF_MOD | BPF_K] = true,
		[BPF_ALU | BPF_MOD | BPF_X] = true,
		[BPF_ALU | BPF_AND | BPF_K] = true,
		[BPF_ALU | BPF_AND | BPF_X] = true,
		[BPF_ALU | BPF_OR | BPF_K] = true,
		[BPF_ALU | BPF_OR | BPF_X] = true,
		[BPF_ALU | BPF_XOR | BPF_K] = true,
		[BPF_ALU | BPF_XOR | BPF_X] = true,
		[BPF_ALU | BPF_LSH | BPF_K] = true,
		[BPF_ALU | BPF_LSH | BPF_X] = true,
		[BPF_ALU | BPF_RSH | BPF_K] = true,
		[BPF_ALU | BPF_RSH | BPF_X] = true,
		[BPF_ALU | BPF_NEG] = true,
		/* Load instructions */
		[BPF_LD | BPF_W | BPF_ABS] = true,
		[BPF_LD | BPF_H | BPF_ABS] = true,
		[BPF_LD | BPF_B | BPF_ABS] = true,
		[BPF_LD | BPF_W | BPF_LEN] = true,
		[BPF_LD | BPF_W | BPF_IND] = true,
		[BPF_LD | BPF_H | BPF_IND] = true,
		[BPF_LD | BPF_B | BPF_IND] = true,
		[BPF_LD | BPF_IMM] = true,
		[BPF_LD | BPF_MEM] = true,
		[BPF_LDX | BPF_W | BPF_LEN] = true,
		[BPF_LDX | BPF_B | BPF_MSH] = true,
		[BPF_LDX | BPF_IMM] = true,
		[BPF_LDX | BPF_MEM] = true,
		/* Store instructions */
		[BPF_ST] = true,
		[BPF_STX] = true,
		/* Misc instructions */
		[BPF_MISC | BPF_TAX] = true,
		[BPF_MISC | BPF_TXA] = true,
		/* Return instructions */
		[BPF_RET | BPF_K] = true,
		[BPF_RET | BPF_A] = true,
		/* Jump instructions */
		[BPF_JMP | BPF_JA] = true,
		[BPF_JMP | BPF_JEQ | BPF_K] = true,
		[BPF_JMP | BPF_JEQ | BPF_X] = true,
		[BPF_JMP | BPF_JGE | BPF_K] = true,
		[BPF_JMP | BPF_JGE | BPF_X] = true,
		[BPF_JMP | BPF_JGT | BPF_K] = true,
		[BPF_JMP | BPF_JGT | BPF_X] = true,
		[BPF_JMP | BPF_JSET | BPF_K] = true,
		[BPF_JMP | BPF_JSET | BPF_X] = true,
	};

	if (code_to_probe >= ARRAY_SIZE(codes))
		return false;

	return codes[code_to_probe];
}

Linus Torvalds's avatar
Linus Torvalds committed
745
/**
746
 *	bpf_check_classic - verify socket filter code
Linus Torvalds's avatar
Linus Torvalds committed
747 748 749 750 751
 *	@filter: filter to verify
 *	@flen: length of filter
 *
 * Check the user's filter code. If we let some ugly
 * filter code slip through kaboom! The filter must contain
752 753
 * no references or jumps that are out of range, no illegal
 * instructions, and must end with a RET instruction.
Linus Torvalds's avatar
Linus Torvalds committed
754
 *
755 756 757
 * All jumps are forward as they are not signed.
 *
 * Returns 0 if the rule set is legal or -EINVAL if not.
Linus Torvalds's avatar
Linus Torvalds committed
758
 */
759 760
static int bpf_check_classic(const struct sock_filter *filter,
			     unsigned int flen)
Linus Torvalds's avatar
Linus Torvalds committed
761
{
762
	bool anc_found;
763
	int pc;
Linus Torvalds's avatar
Linus Torvalds committed
764

765
	if (flen == 0 || flen > BPF_MAXINSNS)
Linus Torvalds's avatar
Linus Torvalds committed
766 767
		return -EINVAL;

768
	/* Check the filter code now */
Linus Torvalds's avatar
Linus Torvalds committed
769
	for (pc = 0; pc < flen; pc++) {
770
		const struct sock_filter *ftest = &filter[pc];
771

772 773
		/* May we actually operate on this code? */
		if (!chk_code_allowed(ftest->code))
774
			return -EINVAL;
775

776
		/* Some instructions need special checks */
777 778 779 780
		switch (ftest->code) {
		case BPF_ALU | BPF_DIV | BPF_K:
		case BPF_ALU | BPF_MOD | BPF_K:
			/* Check for division by zero */
Eric Dumazet's avatar
Eric Dumazet committed
781 782 783
			if (ftest->k == 0)
				return -EINVAL;
			break;
784 785 786 787 788
		case BPF_LD | BPF_MEM:
		case BPF_LDX | BPF_MEM:
		case BPF_ST:
		case BPF_STX:
			/* Check for invalid memory addresses */
789 790 791
			if (ftest->k >= BPF_MEMWORDS)
				return -EINVAL;
			break;
792 793
		case BPF_JMP | BPF_JA:
			/* Note, the large ftest->k might cause loops.
794 795 796
			 * Compare this with conditional jumps below,
			 * where offsets are limited. --ANK (981016)
			 */
797
			if (ftest->k >= (unsigned int)(flen - pc - 1))
798
				return -EINVAL;
799
			break;
800 801 802 803 804 805 806 807 808
		case BPF_JMP | BPF_JEQ | BPF_K:
		case BPF_JMP | BPF_JEQ | BPF_X:
		case BPF_JMP | BPF_JGE | BPF_K:
		case BPF_JMP | BPF_JGE | BPF_X:
		case BPF_JMP | BPF_JGT | BPF_K:
		case BPF_JMP | BPF_JGT | BPF_X:
		case BPF_JMP | BPF_JSET | BPF_K:
		case BPF_JMP | BPF_JSET | BPF_X:
			/* Both conditionals must be safe */
809
			if (pc + ftest->jt + 1 >= flen ||
810 811
			    pc + ftest->jf + 1 >= flen)
				return -EINVAL;
812
			break;
813 814 815
		case BPF_LD | BPF_W | BPF_ABS:
		case BPF_LD | BPF_H | BPF_ABS:
		case BPF_LD | BPF_B | BPF_ABS:
816
			anc_found = false;
817 818 819
			if (bpf_anc_helper(ftest) & BPF_ANC)
				anc_found = true;
			/* Ancillary operation unknown or unsupported */
820 821
			if (anc_found == false && ftest->k >= SKF_AD_OFF)
				return -EINVAL;
822 823
		}
	}
824

825
	/* Last instruction must be a RET code */
826
	switch (filter[flen - 1].code) {
827 828
	case BPF_RET | BPF_K:
	case BPF_RET | BPF_A:
829
		return check_load_and_stores(filter, flen);
830
	}
831

832
	return -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
833 834
}

835 836
static int bpf_prog_store_orig_filter(struct bpf_prog *fp,
				      const struct sock_fprog *fprog)
837
{
838
	unsigned int fsize = bpf_classic_proglen(fprog);
839 840 841 842 843 844 845 846
	struct sock_fprog_kern *fkprog;

	fp->orig_prog = kmalloc(sizeof(*fkprog), GFP_KERNEL);
	if (!fp->orig_prog)
		return -ENOMEM;

	fkprog = fp->orig_prog;
	fkprog->len = fprog->len;
847 848 849

	fkprog->filter = kmemdup(fp->insns, fsize,
				 GFP_KERNEL | __GFP_NOWARN);
850 851 852 853 854 855 856 857
	if (!fkprog->filter) {
		kfree(fp->orig_prog);
		return -ENOMEM;
	}

	return 0;
}

858
static void bpf_release_orig_filter(struct bpf_prog *fp)
859 860 861 862 863 864 865 866 867
{
	struct sock_fprog_kern *fprog = fp->orig_prog;

	if (fprog) {
		kfree(fprog->filter);
		kfree(fprog);
	}
}

868 869
static void __bpf_prog_release(struct bpf_prog *prog)
{
870
	if (prog->type == BPF_PROG_TYPE_SOCKET_FILTER) {
871 872 873 874 875
		bpf_prog_put(prog);
	} else {
		bpf_release_orig_filter(prog);
		bpf_prog_free(prog);
	}
876 877
}

878 879
static void __sk_filter_release(struct sk_filter *fp)
{
880 881
	__bpf_prog_release(fp->prog);
	kfree(fp);
882 883
}

884
/**
885
 * 	sk_filter_release_rcu - Release a socket filter by rcu_head
886 887
 *	@rcu: rcu_head that contains the sk_filter to free
 */
888
static void sk_filter_release_rcu(struct rcu_head *rcu)
889 890 891
{
	struct sk_filter *fp = container_of(rcu, struct sk_filter, rcu);

892
	__sk_filter_release(fp);
893
}
894 895 896 897 898 899 900 901 902 903 904 905 906 907 908

/**
 *	sk_filter_release - release a socket filter
 *	@fp: filter to remove
 *
 *	Remove a filter from a socket and release its resources.
 */
static void sk_filter_release(struct sk_filter *fp)
{
	if (atomic_dec_and_test(&fp->refcnt))
		call_rcu(&fp->rcu, sk_filter_release_rcu);
}

void sk_filter_uncharge(struct sock *sk, struct sk_filter *fp)
{
909
	u32 filter_size = bpf_prog_size(fp->prog->len);
910

911 912
	atomic_sub(filter_size, &sk->sk_omem_alloc);
	sk_filter_release(fp);
913
}
914

915 916 917 918
/* try to charge the socket memory if there is space available
 * return true on success
 */
bool sk_filter_charge(struct sock *sk, struct sk_filter *fp)
919
{
920
	u32 filter_size = bpf_prog_size(fp->prog->len);
921 922 923 924 925 926 927

	/* same check as in sock_kmalloc() */
	if (filter_size <= sysctl_optmem_max &&
	    atomic_read(&sk->sk_omem_alloc) + filter_size < sysctl_optmem_max) {
		atomic_inc(&fp->refcnt);
		atomic_add(filter_size, &sk->sk_omem_alloc);
		return true;
928
	}
929
	return false;
930 931
}

932
static struct bpf_prog *bpf_migrate_filter(struct bpf_prog *fp)
933 934
{
	struct sock_filter *old_prog;
935
	struct bpf_prog *old_fp;
936
	int err, new_len, old_len = fp->len;
937 938 939 940 941 942 943

	/* We are free to overwrite insns et al right here as it
	 * won't be used at this point in time anymore internally
	 * after the migration to the internal BPF instruction
	 * representation.
	 */
	BUILD_BUG_ON(sizeof(struct sock_filter) !=
944
		     sizeof(struct bpf_insn));
945 946 947 948 949 950

	/* Conversion cannot happen on overlapping memory areas,
	 * so we need to keep the user BPF around until the 2nd
	 * pass. At this time, the user BPF is stored in fp->insns.
	 */
	old_prog = kmemdup(fp->insns, old_len * sizeof(struct sock_filter),
951
			   GFP_KERNEL | __GFP_NOWARN);
952 953 954 955 956 957
	if (!old_prog) {
		err = -ENOMEM;
		goto out_err;
	}

	/* 1st pass: calculate the new program length. */
958
	err = bpf_convert_filter(old_prog, old_len, NULL, &new_len);
959 960 961 962 963
	if (err)
		goto out_err_free;

	/* Expand fp for appending the new filter representation. */
	old_fp = fp;
964
	fp = bpf_prog_realloc(old_fp, bpf_prog_size(new_len), 0);
965 966 967 968 969 970 971 972 973 974 975
	if (!fp) {
		/* The old_fp is still around in case we couldn't
		 * allocate new memory, so uncharge on that one.
		 */
		fp = old_fp;
		err = -ENOMEM;
		goto out_err_free;
	}

	fp->len = new_len;

976
	/* 2nd pass: remap sock_filter insns into bpf_insn insns. */
977
	err = bpf_convert_filter(old_prog, old_len, fp->insnsi, &new_len);
978
	if (err)
979
		/* 2nd bpf_convert_filter() can fail only if it fails
980 981
		 * to allocate memory, remapping must succeed. Note,
		 * that at this time old_fp has already been released
982
		 * by krealloc().
983 984 985
		 */
		goto out_err_free;

986
	bpf_prog_select_runtime(fp);
987

988 989 990 991 992 993
	kfree(old_prog);
	return fp;

out_err_free:
	kfree(old_prog);
out_err:
994
	__bpf_prog_release(fp);
995 996 997
	return ERR_PTR(err);
}

998 999
static struct bpf_prog *bpf_prepare_filter(struct bpf_prog *fp,
					   bpf_aux_classic_check_t trans)
1000 1001 1002
{
	int err;

1003
	fp->bpf_func = NULL;
1004
	fp->jited = false;
1005

1006
	err = bpf_check_classic(fp->insns, fp->len);
1007
	if (err) {
1008
		__bpf_prog_release(fp);
1009
		return ERR_PTR(err);
1010
	}
1011

1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
	/* There might be additional checks and transformations
	 * needed on classic filters, f.e. in case of seccomp.
	 */
	if (trans) {
		err = trans(fp->insns, fp->len);
		if (err) {
			__bpf_prog_release(fp);
			return ERR_PTR(err);
		}
	}

1023 1024 1025
	/* Probe if we can JIT compile the filter and if so, do
	 * the compilation of the filter.
	 */
1026
	bpf_jit_compile(fp);
1027 1028 1029 1030

	/* JIT compiler couldn't process this filter, so do the
	 * internal BPF translation for the optimized interpreter.
	 */
1031
	if (!fp->jited)
1032
		fp = bpf_migrate_filter(fp);
1033 1034

	return fp;
1035 1036 1037
}

/**
1038
 *	bpf_prog_create - create an unattached filter
1039
 *	@pfp: the unattached filter that is created
1040
 *	@fprog: the filter program
1041
 *
1042
 * Create a filter independent of any socket. We first run some
1043 1044 1045 1046
 * sanity checks on it to make sure it does not explode on us later.
 * If an error occurs or there is insufficient memory for the filter
 * a negative errno code is returned. On success the return is zero.
 */
1047
int bpf_prog_create(struct bpf_prog **pfp, struct sock_fprog_kern *fprog)
1048
{
1049
	unsigned int fsize = bpf_classic_proglen(fprog);
1050
	struct bpf_prog *fp;
1051 1052 1053 1054 1055

	/* Make sure new filter is there and in the right amounts. */
	if (fprog->filter == NULL)
		return -EINVAL;

1056
	fp = bpf_prog_alloc(bpf_prog_size(fprog->len), 0);
1057 1058
	if (!fp)
		return -ENOMEM;
1059

1060 1061 1062
	memcpy(fp->insns, fprog->filter, fsize);

	fp->len = fprog->len;
1063 1064 1065 1066 1067
	/* Since unattached filters are not copied back to user
	 * space through sk_get_filter(), we do not need to hold
	 * a copy here, and can spare us the work.
	 */
	fp->orig_prog = NULL;
1068

1069
	/* bpf_prepare_filter() already takes care of freeing
1070 1071
	 * memory in case something goes wrong.
	 */
1072
	fp = bpf_prepare_filter(fp, NULL);
1073 1074
	if (IS_ERR(fp))
		return PTR_ERR(fp);
1075 1076 1077 1078

	*pfp = fp;
	return 0;
}
1079
EXPORT_SYMBOL_GPL(bpf_prog_create);
1080

1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127
/**
 *	bpf_prog_create_from_user - create an unattached filter from user buffer
 *	@pfp: the unattached filter that is created
 *	@fprog: the filter program
 *	@trans: post-classic verifier transformation handler
 *
 * This function effectively does the same as bpf_prog_create(), only
 * that it builds up its insns buffer from user space provided buffer.
 * It also allows for passing a bpf_aux_classic_check_t handler.
 */
int bpf_prog_create_from_user(struct bpf_prog **pfp, struct sock_fprog *fprog,
			      bpf_aux_classic_check_t trans)
{
	unsigned int fsize = bpf_classic_proglen(fprog);
	struct bpf_prog *fp;

	/* Make sure new filter is there and in the right amounts. */
	if (fprog->filter == NULL)
		return -EINVAL;

	fp = bpf_prog_alloc(bpf_prog_size(fprog->len), 0);
	if (!fp)
		return -ENOMEM;

	if (copy_from_user(fp->insns, fprog->filter, fsize)) {
		__bpf_prog_free(fp);
		return -EFAULT;
	}

	fp->len = fprog->len;
	/* Since unattached filters are not copied back to user
	 * space through sk_get_filter(), we do not need to hold
	 * a copy here, and can spare us the work.
	 */
	fp->orig_prog = NULL;

	/* bpf_prepare_filter() already takes care of freeing
	 * memory in case something goes wrong.
	 */
	fp = bpf_prepare_filter(fp, trans);
	if (IS_ERR(fp))
		return PTR_ERR(fp);

	*pfp = fp;
	return 0;
}

1128
void bpf_prog_destroy(struct bpf_prog *fp)
1129
{
1130
	__bpf_prog_release(fp);
1131
}
1132
EXPORT_SYMBOL_GPL(bpf_prog_destroy);
1133

1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159
static int __sk_attach_prog(struct bpf_prog *prog, struct sock *sk)
{
	struct sk_filter *fp, *old_fp;

	fp = kmalloc(sizeof(*fp), GFP_KERNEL);
	if (!fp)
		return -ENOMEM;

	fp->prog = prog;
	atomic_set(&fp->refcnt, 0);

	if (!sk_filter_charge(sk, fp)) {
		kfree(fp);
		return -ENOMEM;
	}

	old_fp = rcu_dereference_protected(sk->sk_filter,
					   sock_owned_by_user(sk));
	rcu_assign_pointer(sk->sk_filter, fp);

	if (old_fp)
		sk_filter_uncharge(sk, old_fp);

	return 0;
}

Linus Torvalds's avatar
Linus Torvalds committed
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171
/**
 *	sk_attach_filter - attach a socket filter
 *	@fprog: the filter program
 *	@sk: the socket to use
 *
 * Attach the user's filter code. We first run some sanity checks on
 * it to make sure it does not explode on us later. If an error
 * occurs or there is insufficient memory for the filter a negative
 * errno code is returned. On success the return is zero.
 */
int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
{
1172
	unsigned int fsize = bpf_classic_proglen(fprog);
1173 1174
	unsigned int bpf_fsize = bpf_prog_size(fprog->len);
	struct bpf_prog *prog;
Linus Torvalds's avatar
Linus Torvalds committed
1175 1176
	int err;

1177 1178 1179
	if (sock_flag(sk, SOCK_FILTER_LOCKED))
		return -EPERM;

Linus Torvalds's avatar
Linus Torvalds committed
1180
	/* Make sure new filter is there and in the right amounts. */
1181 1182
	if (fprog->filter == NULL)
		return -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
1183

1184
	prog = bpf_prog_alloc(bpf_fsize, 0);
1185
	if (!prog)
Linus Torvalds's avatar
Linus Torvalds committed
1186
		return -ENOMEM;
1187

1188
	if (copy_from_user(prog->insns, fprog->filter, fsize)) {
1189
		__bpf_prog_free(prog);
Linus Torvalds's avatar
Linus Torvalds committed
1190 1191 1192
		return -EFAULT;
	}

1193
	prog->len = fprog->len;
Linus Torvalds's avatar
Linus Torvalds committed
1194

1195
	err = bpf_prog_store_orig_filter(prog, fprog);
1196
	if (err) {
1197
		__bpf_prog_free(prog);
1198 1199 1200
		return -ENOMEM;
	}

1201
	/* bpf_prepare_filter() already takes care of freeing
1202 1203
	 * memory in case something goes wrong.
	 */
1204
	prog = bpf_prepare_filter(prog, NULL);
1205 1206 1207
	if (IS_ERR(prog))
		return PTR_ERR(prog);

1208 1209
	err = __sk_attach_prog(prog, sk);
	if (err < 0) {
1210
		__bpf_prog_release(prog);
1211
		return err;
1212 1213
	}

1214
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
1215
}
1216
EXPORT_SYMBOL_GPL(sk_attach_filter);
Linus Torvalds's avatar
Linus Torvalds committed
1217

1218 1219 1220
int sk_attach_bpf(u32 ufd, struct sock *sk)
{
	struct bpf_prog *prog;
1221
	int err;
1222 1223 1224 1225 1226

	if (sock_flag(sk, SOCK_FILTER_LOCKED))
		return -EPERM;

	prog = bpf_prog_get(ufd);
1227 1228
	if (IS_ERR(prog))
		return PTR_ERR(prog);
1229