LXR linux/net/core/filter.c

   1/*
   2 * Linux Socket Filter - Kernel level socket filtering
   3 *
   4 * Author:
   5 *     Jay Schulist <jschlst@samba.org>
   6 *
   7 * Based on the design of:
   8 *     - The Berkeley Packet Filter
   9 *
  10 * This program is free software; you can redistribute it and/or
  11 * modify it under the terms of the GNU General Public License
  12 * as published by the Free Software Foundation; either version
  13 * 2 of the License, or (at your option) any later version.
  14 *
  15 * Andi Kleen - Fix a few bad bugs and races.
  16 * Kris Katterjohn - Added many additional checks in sk_chk_filter()
  17 */
  18
  19#include <linux/module.h>
  20#include <linux/types.h>
  21#include <linux/mm.h>
  22#include <linux/fcntl.h>
  23#include <linux/socket.h>
  24#include <linux/in.h>
  25#include <linux/inet.h>
  26#include <linux/netdevice.h>
  27#include <linux/if_packet.h>
  28#include <net/ip.h>
  29#include <net/protocol.h>
  30#include <linux/skbuff.h>
  31#include <net/sock.h>
  32#include <linux/errno.h>
  33#include <linux/timer.h>
  34#include <asm/system.h>
  35#include <asm/uaccess.h>
  36#include <asm/unaligned.h>
  37#include <linux/filter.h>
  38
  39/* No hurry in this branch */
  40static void *__load_pointer(struct sk_buff *skb, int k)
  41{
  42        u8 *ptr = NULL;
  43
  44        if (k >= SKF_NET_OFF)
  45                ptr = skb_network_header(skb) + k - SKF_NET_OFF;
  46        else if (k >= SKF_LL_OFF)
  47                ptr = skb_mac_header(skb) + k - SKF_LL_OFF;
  48
  49        if (ptr >= skb->head && ptr < skb_tail_pointer(skb))
  50                return ptr;
  51        return NULL;
  52}
  53
  54static inline void *load_pointer(struct sk_buff *skb, int k,
  55                                 unsigned int size, void *buffer)
  56{
  57        if (k >= 0)
  58                return skb_header_pointer(skb, k, size, buffer);
  59        else {
  60                if (k >= SKF_AD_OFF)
  61                        return NULL;
  62                return __load_pointer(skb, k);
  63        }
  64}
  65
  66/**
  67 *      sk_run_filter - run a filter on a socket
  68 *      @skb: buffer to run the filter on
  69 *      @filter: filter to apply
  70 *      @flen: length of filter
  71 *
  72 * Decode and apply filter instructions to the skb->data.
  73 * Return length to keep, 0 for none. skb is the data we are
  74 * filtering, filter is the array of filter instructions, and
  75 * len is the number of filter blocks in the array.
  76 */
  77unsigned int sk_run_filter(struct sk_buff *skb, struct sock_filter *filter, int flen)
  78{
  79        struct sock_filter *fentry;     /* We walk down these */
  80        void *ptr;
  81        u32 A = 0;                      /* Accumulator */
  82        u32 X = 0;                      /* Index Register */
  83        u32 mem[BPF_MEMWORDS];          /* Scratch Memory Store */
  84        u32 tmp;
  85        int k;
  86        int pc;
  87
  88        /*
  89         * Process array of filter instructions.
  90         */
  91        for (pc = 0; pc < flen; pc++) {
  92                fentry = &filter[pc];
  93
  94                switch (fentry->code) {
  95                case BPF_ALU|BPF_ADD|BPF_X:
  96                        A += X;
  97                        continue;
  98                case BPF_ALU|BPF_ADD|BPF_K:
  99                        A += fentry->k;
 100                        continue;
 101                case BPF_ALU|BPF_SUB|BPF_X:
 102                        A -= X;
 103                        continue;
 104                case BPF_ALU|BPF_SUB|BPF_K:
 105                        A -= fentry->k;
 106                        continue;
 107                case BPF_ALU|BPF_MUL|BPF_X:
 108                        A *= X;
 109                        continue;
 110                case BPF_ALU|BPF_MUL|BPF_K:
 111                        A *= fentry->k;
 112                        continue;
 113                case BPF_ALU|BPF_DIV|BPF_X:
 114                        if (X == 0)
 115                                return 0;
 116                        A /= X;
 117                        continue;
 118                case BPF_ALU|BPF_DIV|BPF_K:
 119                        A /= fentry->k;
 120                        continue;
 121                case BPF_ALU|BPF_AND|BPF_X:
 122                        A &= X;
 123                        continue;
 124                case BPF_ALU|BPF_AND|BPF_K:
 125                        A &= fentry->k;
 126                        continue;
 127                case BPF_ALU|BPF_OR|BPF_X:
 128                        A |= X;
 129                        continue;
 130                case BPF_ALU|BPF_OR|BPF_K:
 131                        A |= fentry->k;
 132                        continue;
 133                case BPF_ALU|BPF_LSH|BPF_X:
 134                        A <<= X;
 135                        continue;
 136                case BPF_ALU|BPF_LSH|BPF_K:
 137                        A <<= fentry->k;
 138                        continue;
 139                case BPF_ALU|BPF_RSH|BPF_X:
 140                        A >>= X;
 141                        continue;
 142                case BPF_ALU|BPF_RSH|BPF_K:
 143                        A >>= fentry->k;
 144                        continue;
 145                case BPF_ALU|BPF_NEG:
 146                        A = -A;
 147                        continue;
 148                case BPF_JMP|BPF_JA:
 149                        pc += fentry->k;
 150                        continue;
 151                case BPF_JMP|BPF_JGT|BPF_K:
 152                        pc += (A > fentry->k) ? fentry->jt : fentry->jf;
 153                        continue;
 154                case BPF_JMP|BPF_JGE|BPF_K:
 155                        pc += (A >= fentry->k) ? fentry->jt : fentry->jf;
 156                        continue;
 157                case BPF_JMP|BPF_JEQ|BPF_K:
 158                        pc += (A == fentry->k) ? fentry->jt : fentry->jf;
 159                        continue;
 160                case BPF_JMP|BPF_JSET|BPF_K:
 161                        pc += (A & fentry->k) ? fentry->jt : fentry->jf;
 162                        continue;
 163                case BPF_JMP|BPF_JGT|BPF_X:
 164                        pc += (A > X) ? fentry->jt : fentry->jf;
 165                        continue;
 166                case BPF_JMP|BPF_JGE|BPF_X:
 167                        pc += (A >= X) ? fentry->jt : fentry->jf;
 168                        continue;
 169                case BPF_JMP|BPF_JEQ|BPF_X:
 170                        pc += (A == X) ? fentry->jt : fentry->jf;
 171                        continue;
 172                case BPF_JMP|BPF_JSET|BPF_X:
 173                        pc += (A & X) ? fentry->jt : fentry->jf;
 174                        continue;
 175                case BPF_LD|BPF_W|BPF_ABS:
 176                        k = fentry->k;
 177load_w:
 178                        ptr = load_pointer(skb, k, 4, &tmp);
 179                        if (ptr != NULL) {
 180                                A = ntohl(get_unaligned((__be32 *)ptr));
 181                                continue;
 182                        }
 183                        break;
 184                case BPF_LD|BPF_H|BPF_ABS:
 185                        k = fentry->k;
 186load_h:
 187                        ptr = load_pointer(skb, k, 2, &tmp);
 188                        if (ptr != NULL) {
 189                                A = ntohs(get_unaligned((__be16 *)ptr));
 190                                continue;
 191                        }
 192                        break;
 193                case BPF_LD|BPF_B|BPF_ABS:
 194                        k = fentry->k;
 195load_b:
 196                        ptr = load_pointer(skb, k, 1, &tmp);
 197                        if (ptr != NULL) {
 198                                A = *(u8 *)ptr;
 199                                continue;
 200                        }
 201                        break;
 202                case BPF_LD|BPF_W|BPF_LEN:
 203                        A = skb->len;
 204                        continue;
 205                case BPF_LDX|BPF_W|BPF_LEN:
 206                        X = skb->len;
 207                        continue;
 208                case BPF_LD|BPF_W|BPF_IND:
 209                        k = X + fentry->k;
 210                        goto load_w;
 211                case BPF_LD|BPF_H|BPF_IND:
 212                        k = X + fentry->k;
 213                        goto load_h;
 214                case BPF_LD|BPF_B|BPF_IND:
 215                        k = X + fentry->k;
 216                        goto load_b;
 217                case BPF_LDX|BPF_B|BPF_MSH:
 218                        ptr = load_pointer(skb, fentry->k, 1, &tmp);
 219                        if (ptr != NULL) {
 220                                X = (*(u8 *)ptr & 0xf) << 2;
 221                                continue;
 222                        }
 223                        return 0;
 224                case BPF_LD|BPF_IMM:
 225                        A = fentry->k;
 226                        continue;
 227                case BPF_LDX|BPF_IMM:
 228                        X = fentry->k;
 229                        continue;
 230                case BPF_LD|BPF_MEM:
 231                        A = mem[fentry->k];
 232                        continue;
 233                case BPF_LDX|BPF_MEM:
 234                        X = mem[fentry->k];
 235                        continue;
 236                case BPF_MISC|BPF_TAX:
 237                        X = A;
 238                        continue;
 239                case BPF_MISC|BPF_TXA:
 240                        A = X;
 241                        continue;
 242                case BPF_RET|BPF_K:
 243                        return fentry->k;
 244                case BPF_RET|BPF_A:
 245                        return A;
 246                case BPF_ST:
 247                        mem[fentry->k] = A;
 248                        continue;
 249                case BPF_STX:
 250                        mem[fentry->k] = X;
 251                        continue;
 252                default:
 253                        WARN_ON(1);
 254                        return 0;
 255                }
 256
 257                /*
 258                 * Handle ancillary data, which are impossible
 259                 * (or very difficult) to get parsing packet contents.
 260                 */
 261                switch (k-SKF_AD_OFF) {
 262                case SKF_AD_PROTOCOL:
 263                        A = ntohs(skb->protocol);
 264                        continue;
 265                case SKF_AD_PKTTYPE:
 266                        A = skb->pkt_type;
 267                        continue;
 268                case SKF_AD_IFINDEX:
 269                        A = skb->dev->ifindex;
 270                        continue;
 271                default:
 272                        return 0;
 273                }
 274        }
 275
 276        return 0;
 277}
 278
 279/**
 280 *      sk_chk_filter - verify socket filter code
 281 *      @filter: filter to verify
 282 *      @flen: length of filter
 283 *
 284 * Check the user's filter code. If we let some ugly
 285 * filter code slip through kaboom! The filter must contain
 286 * no references or jumps that are out of range, no illegal
 287 * instructions, and must end with a RET instruction.
 288 *
 289 * All jumps are forward as they are not signed.
 290 *
 291 * Returns 0 if the rule set is legal or -EINVAL if not.
 292 */
 293int sk_chk_filter(struct sock_filter *filter, int flen)
 294{
 295        struct sock_filter *ftest;
 296        int pc;
 297
 298        if (flen == 0 || flen > BPF_MAXINSNS)
 299                return -EINVAL;
 300
 301        /* check the filter code now */
 302        for (pc = 0; pc < flen; pc++) {
 303                ftest = &filter[pc];
 304
 305                /* Only allow valid instructions */
 306                switch (ftest->code) {
 307                case BPF_ALU|BPF_ADD|BPF_K:
 308                case BPF_ALU|BPF_ADD|BPF_X:
 309                case BPF_ALU|BPF_SUB|BPF_K:
 310                case BPF_ALU|BPF_SUB|BPF_X:
 311                case BPF_ALU|BPF_MUL|BPF_K:
 312                case BPF_ALU|BPF_MUL|BPF_X:
 313                case BPF_ALU|BPF_DIV|BPF_X:
 314                case BPF_ALU|BPF_AND|BPF_K:
 315                case BPF_ALU|BPF_AND|BPF_X:
 316                case BPF_ALU|BPF_OR|BPF_K:
 317                case BPF_ALU|BPF_OR|BPF_X:
 318                case BPF_ALU|BPF_LSH|BPF_K:
 319                case BPF_ALU|BPF_LSH|BPF_X:
 320                case BPF_ALU|BPF_RSH|BPF_K:
 321                case BPF_ALU|BPF_RSH|BPF_X:
 322                case BPF_ALU|BPF_NEG:
 323                case BPF_LD|BPF_W|BPF_ABS:
 324                case BPF_LD|BPF_H|BPF_ABS:
 325                case BPF_LD|BPF_B|BPF_ABS:
 326                case BPF_LD|BPF_W|BPF_LEN:
 327                case BPF_LD|BPF_W|BPF_IND:
 328                case BPF_LD|BPF_H|BPF_IND:
 329                case BPF_LD|BPF_B|BPF_IND:
 330                case BPF_LD|BPF_IMM:
 331                case BPF_LDX|BPF_W|BPF_LEN:
 332                case BPF_LDX|BPF_B|BPF_MSH:
 333                case BPF_LDX|BPF_IMM:
 334                case BPF_MISC|BPF_TAX:
 335                case BPF_MISC|BPF_TXA:
 336                case BPF_RET|BPF_K:
 337                case BPF_RET|BPF_A:
 338                        break;
 339
 340                /* Some instructions need special checks */
 341
 342                case BPF_ALU|BPF_DIV|BPF_K:
 343                        /* check for division by zero */
 344                        if (ftest->k == 0)
 345                                return -EINVAL;
 346                        break;
 347
 348                case BPF_LD|BPF_MEM:
 349                case BPF_LDX|BPF_MEM:
 350                case BPF_ST:
 351                case BPF_STX:
 352                        /* check for invalid memory addresses */
 353                        if (ftest->k >= BPF_MEMWORDS)
 354                                return -EINVAL;
 355                        break;
 356
 357                case BPF_JMP|BPF_JA:
 358                        /*
 359                         * Note, the large ftest->k might cause loops.
 360                         * Compare this with conditional jumps below,
 361                         * where offsets are limited. --ANK (981016)
 362                         */
 363                        if (ftest->k >= (unsigned)(flen-pc-1))
 364                                return -EINVAL;
 365                        break;
 366
 367                case BPF_JMP|BPF_JEQ|BPF_K:
 368                case BPF_JMP|BPF_JEQ|BPF_X:
 369                case BPF_JMP|BPF_JGE|BPF_K:
 370                case BPF_JMP|BPF_JGE|BPF_X:
 371                case BPF_JMP|BPF_JGT|BPF_K:
 372                case BPF_JMP|BPF_JGT|BPF_X:
 373                case BPF_JMP|BPF_JSET|BPF_K:
 374                case BPF_JMP|BPF_JSET|BPF_X:
 375                        /* for conditionals both must be safe */
 376                        if (pc + ftest->jt + 1 >= flen ||
 377                            pc + ftest->jf + 1 >= flen)
 378                                return -EINVAL;
 379                        break;
 380
 381                default:
 382                        return -EINVAL;
 383                }
 384        }
 385
 386        return (BPF_CLASS(filter[flen - 1].code) == BPF_RET) ? 0 : -EINVAL;
 387}
 388
 389/**
 390 *      sk_filter_rcu_release: Release a socket filter by rcu_head
 391 *      @rcu: rcu_head that contains the sk_filter to free
 392 */
 393static void sk_filter_rcu_release(struct rcu_head *rcu)
 394{
 395        struct sk_filter *fp = container_of(rcu, struct sk_filter, rcu);
 396
 397        sk_filter_release(fp);
 398}
 399
 400static void sk_filter_delayed_uncharge(struct sock *sk, struct sk_filter *fp)
 401{
 402        unsigned int size = sk_filter_len(fp);
 403
 404        atomic_sub(size, &sk->sk_omem_alloc);
 405        call_rcu_bh(&fp->rcu, sk_filter_rcu_release);
 406}
 407
 408/**
 409 *      sk_attach_filter - attach a socket filter
 410 *      @fprog: the filter program
 411 *      @sk: the socket to use
 412 *
 413 * Attach the user's filter code. We first run some sanity checks on
 414 * it to make sure it does not explode on us later. If an error
 415 * occurs or there is insufficient memory for the filter a negative
 416 * errno code is returned. On success the return is zero.
 417 */
 418int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
 419{
 420        struct sk_filter *fp, *old_fp;
 421        unsigned int fsize = sizeof(struct sock_filter) * fprog->len;
 422        int err;
 423
 424        /* Make sure new filter is there and in the right amounts. */
 425        if (fprog->filter == NULL)
 426                return -EINVAL;
 427
 428        fp = sock_kmalloc(sk, fsize+sizeof(*fp), GFP_KERNEL);
 429        if (!fp)
 430                return -ENOMEM;
 431        if (copy_from_user(fp->insns, fprog->filter, fsize)) {
 432                sock_kfree_s(sk, fp, fsize+sizeof(*fp));
 433                return -EFAULT;
 434        }
 435
 436        atomic_set(&fp->refcnt, 1);
 437        fp->len = fprog->len;
 438
 439        err = sk_chk_filter(fp->insns, fp->len);
 440        if (err) {
 441                sk_filter_uncharge(sk, fp);
 442                return err;
 443        }
 444
 445        rcu_read_lock_bh();
 446        old_fp = rcu_dereference(sk->sk_filter);
 447        rcu_assign_pointer(sk->sk_filter, fp);
 448        rcu_read_unlock_bh();
 449
 450        if (old_fp)
 451                sk_filter_delayed_uncharge(sk, old_fp);
 452        return 0;
 453}
 454
 455int sk_detach_filter(struct sock *sk)
 456{
 457        int ret = -ENOENT;
 458        struct sk_filter *filter;
 459
 460        rcu_read_lock_bh();
 461        filter = rcu_dereference(sk->sk_filter);
 462        if (filter) {
 463                rcu_assign_pointer(sk->sk_filter, NULL);
 464                sk_filter_delayed_uncharge(sk, filter);
 465                ret = 0;
 466        }
 467        rcu_read_unlock_bh();
 468        return ret;
 469}
 470
 471EXPORT_SYMBOL(sk_chk_filter);
 472EXPORT_SYMBOL(sk_run_filter);
 473