ts_bm.c 4.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193
  1. /*
  2. * lib/ts_bm.c Boyer-Moore text search implementation
  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: Pablo Neira Ayuso <pablo@eurodev.net>
  10. *
  11. * ==========================================================================
  12. *
  13. * Implements Boyer-Moore string matching algorithm:
  14. *
  15. * [1] A Fast String Searching Algorithm, R.S. Boyer and Moore.
  16. * Communications of the Association for Computing Machinery,
  17. * 20(10), 1977, pp. 762-772.
  18. * http://www.cs.utexas.edu/users/moore/publications/fstrpos.pdf
  19. *
  20. * [2] Handbook of Exact String Matching Algorithms, Thierry Lecroq, 2004
  21. * http://www-igm.univ-mlv.fr/~lecroq/string/string.pdf
  22. *
  23. * Note: Since Boyer-Moore (BM) performs searches for matchings from right
  24. * to left, it's still possible that a matching could be spread over
  25. * multiple blocks, in that case this algorithm won't find any coincidence.
  26. *
  27. * If you're willing to ensure that such thing won't ever happen, use the
  28. * Knuth-Pratt-Morris (KMP) implementation instead. In conclusion, choose
  29. * the proper string search algorithm depending on your setting.
  30. *
  31. * Say you're using the textsearch infrastructure for filtering, NIDS or
  32. * any similar security focused purpose, then go KMP. Otherwise, if you
  33. * really care about performance, say you're classifying packets to apply
  34. * Quality of Service (QoS) policies, and you don't mind about possible
  35. * matchings spread over multiple fragments, then go BM.
  36. */
  37. #include <linux/kernel.h>
  38. #include <linux/module.h>
  39. #include <linux/types.h>
  40. #include <linux/string.h>
  41. #include <linux/textsearch.h>
  42. /* Alphabet size, use ASCII */
  43. #define ASIZE 256
  44. #if 0
  45. #define DEBUGP printk
  46. #else
  47. #define DEBUGP(args, format...)
  48. #endif
  49. struct ts_bm
  50. {
  51. u8 * pattern;
  52. unsigned int patlen;
  53. unsigned int bad_shift[ASIZE];
  54. unsigned int good_shift[0];
  55. };
  56. static unsigned int bm_find(struct ts_config *conf, struct ts_state *state)
  57. {
  58. struct ts_bm *bm = ts_config_priv(conf);
  59. unsigned int i, text_len, consumed = state->offset;
  60. const u8 *text;
  61. int shift = bm->patlen, bs;
  62. for (;;) {
  63. text_len = conf->get_next_block(consumed, &text, conf, state);
  64. if (unlikely(text_len == 0))
  65. break;
  66. while (shift < text_len) {
  67. DEBUGP("Searching in position %d (%c)\n",
  68. shift, text[shift]);
  69. for (i = 0; i < bm->patlen; i++)
  70. if (text[shift-i] != bm->pattern[bm->patlen-1-i])
  71. goto next;
  72. /* London calling... */
  73. DEBUGP("found!\n");
  74. return consumed += (shift-(bm->patlen-1));
  75. next: bs = bm->bad_shift[text[shift-i]];
  76. /* Now jumping to... */
  77. shift = max_t(int, shift-i+bs, shift+bm->good_shift[i]);
  78. }
  79. consumed += text_len;
  80. }
  81. return UINT_MAX;
  82. }
  83. static int subpattern(u8 *pattern, int i, int j, int g)
  84. {
  85. int x = i+g-1, y = j+g-1, ret = 0;
  86. while(pattern[x--] == pattern[y--]) {
  87. if (y < 0) {
  88. ret = 1;
  89. break;
  90. }
  91. if (--g == 0) {
  92. ret = pattern[i-1] != pattern[j-1];
  93. break;
  94. }
  95. }
  96. return ret;
  97. }
  98. static void compute_prefix_tbl(struct ts_bm *bm)
  99. {
  100. int i, j, g;
  101. for (i = 0; i < ASIZE; i++)
  102. bm->bad_shift[i] = bm->patlen;
  103. for (i = 0; i < bm->patlen - 1; i++)
  104. bm->bad_shift[bm->pattern[i]] = bm->patlen - 1 - i;
  105. /* Compute the good shift array, used to match reocurrences
  106. * of a subpattern */
  107. bm->good_shift[0] = 1;
  108. for (i = 1; i < bm->patlen; i++)
  109. bm->good_shift[i] = bm->patlen;
  110. for (i = bm->patlen-1, g = 1; i > 0; g++, i--) {
  111. for (j = i-1; j >= 1-g ; j--)
  112. if (subpattern(bm->pattern, i, j, g)) {
  113. bm->good_shift[g] = bm->patlen-j-g;
  114. break;
  115. }
  116. }
  117. }
  118. static struct ts_config *bm_init(const void *pattern, unsigned int len,
  119. gfp_t gfp_mask)
  120. {
  121. struct ts_config *conf;
  122. struct ts_bm *bm;
  123. unsigned int prefix_tbl_len = len * sizeof(unsigned int);
  124. size_t priv_size = sizeof(*bm) + len + prefix_tbl_len;
  125. conf = alloc_ts_config(priv_size, gfp_mask);
  126. if (IS_ERR(conf))
  127. return conf;
  128. bm = ts_config_priv(conf);
  129. bm->patlen = len;
  130. bm->pattern = (u8 *) bm->good_shift + prefix_tbl_len;
  131. memcpy(bm->pattern, pattern, len);
  132. compute_prefix_tbl(bm);
  133. return conf;
  134. }
  135. static void *bm_get_pattern(struct ts_config *conf)
  136. {
  137. struct ts_bm *bm = ts_config_priv(conf);
  138. return bm->pattern;
  139. }
  140. static unsigned int bm_get_pattern_len(struct ts_config *conf)
  141. {
  142. struct ts_bm *bm = ts_config_priv(conf);
  143. return bm->patlen;
  144. }
  145. static struct ts_ops bm_ops = {
  146. .name = "bm",
  147. .find = bm_find,
  148. .init = bm_init,
  149. .get_pattern = bm_get_pattern,
  150. .get_pattern_len = bm_get_pattern_len,
  151. .owner = THIS_MODULE,
  152. .list = LIST_HEAD_INIT(bm_ops.list)
  153. };
  154. static int __init init_bm(void)
  155. {
  156. return textsearch_register(&bm_ops);
  157. }
  158. static void __exit exit_bm(void)
  159. {
  160. textsearch_unregister(&bm_ops);
  161. }
  162. MODULE_LICENSE("GPL");
  163. module_init(init_bm);
  164. module_exit(exit_bm);