sha1.c 6.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202
  1. /*
  2. * SHA1 routine optimized to do word accesses rather than byte accesses,
  3. * and to avoid unnecessary copies into the context array.
  4. *
  5. * This was based on the git SHA1 implementation.
  6. */
  7. #include <linux/kernel.h>
  8. #include <linux/export.h>
  9. #include <linux/bitops.h>
  10. #include <linux/cryptohash.h>
  11. #include <asm/unaligned.h>
  12. /*
  13. * If you have 32 registers or more, the compiler can (and should)
  14. * try to change the array[] accesses into registers. However, on
  15. * machines with less than ~25 registers, that won't really work,
  16. * and at least gcc will make an unholy mess of it.
  17. *
  18. * So to avoid that mess which just slows things down, we force
  19. * the stores to memory to actually happen (we might be better off
  20. * with a 'W(t)=(val);asm("":"+m" (W(t))' there instead, as
  21. * suggested by Artur Skawina - that will also make gcc unable to
  22. * try to do the silly "optimize away loads" part because it won't
  23. * see what the value will be).
  24. *
  25. * Ben Herrenschmidt reports that on PPC, the C version comes close
  26. * to the optimized asm with this (ie on PPC you don't want that
  27. * 'volatile', since there are lots of registers).
  28. *
  29. * On ARM we get the best code generation by forcing a full memory barrier
  30. * between each SHA_ROUND, otherwise gcc happily get wild with spilling and
  31. * the stack frame size simply explode and performance goes down the drain.
  32. */
  33. #ifdef CONFIG_X86
  34. #define setW(x, val) (*(volatile __u32 *)&W(x) = (val))
  35. #elif defined(CONFIG_ARM)
  36. #define setW(x, val) do { W(x) = (val); __asm__("":::"memory"); } while (0)
  37. #else
  38. #define setW(x, val) (W(x) = (val))
  39. #endif
  40. /* This "rolls" over the 512-bit array */
  41. #define W(x) (array[(x)&15])
  42. /*
  43. * Where do we get the source from? The first 16 iterations get it from
  44. * the input data, the next mix it from the 512-bit array.
  45. */
  46. #define SHA_SRC(t) get_unaligned_be32((__u32 *)data + t)
  47. #define SHA_MIX(t) rol32(W(t+13) ^ W(t+8) ^ W(t+2) ^ W(t), 1)
  48. #define SHA_ROUND(t, input, fn, constant, A, B, C, D, E) do { \
  49. __u32 TEMP = input(t); setW(t, TEMP); \
  50. E += TEMP + rol32(A,5) + (fn) + (constant); \
  51. B = ror32(B, 2); } while (0)
  52. #define T_0_15(t, A, B, C, D, E) SHA_ROUND(t, SHA_SRC, (((C^D)&B)^D) , 0x5a827999, A, B, C, D, E )
  53. #define T_16_19(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (((C^D)&B)^D) , 0x5a827999, A, B, C, D, E )
  54. #define T_20_39(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0x6ed9eba1, A, B, C, D, E )
  55. #define T_40_59(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, ((B&C)+(D&(B^C))) , 0x8f1bbcdc, A, B, C, D, E )
  56. #define T_60_79(t, A, B, C, D, E) SHA_ROUND(t, SHA_MIX, (B^C^D) , 0xca62c1d6, A, B, C, D, E )
  57. /**
  58. * sha_transform - single block SHA1 transform
  59. *
  60. * @digest: 160 bit digest to update
  61. * @data: 512 bits of data to hash
  62. * @array: 16 words of workspace (see note)
  63. *
  64. * This function generates a SHA1 digest for a single 512-bit block.
  65. * Be warned, it does not handle padding and message digest, do not
  66. * confuse it with the full FIPS 180-1 digest algorithm for variable
  67. * length messages.
  68. *
  69. * Note: If the hash is security sensitive, the caller should be sure
  70. * to clear the workspace. This is left to the caller to avoid
  71. * unnecessary clears between chained hashing operations.
  72. */
  73. void sha_transform(__u32 *digest, const char *data, __u32 *array)
  74. {
  75. __u32 A, B, C, D, E;
  76. A = digest[0];
  77. B = digest[1];
  78. C = digest[2];
  79. D = digest[3];
  80. E = digest[4];
  81. /* Round 1 - iterations 0-16 take their input from 'data' */
  82. T_0_15( 0, A, B, C, D, E);
  83. T_0_15( 1, E, A, B, C, D);
  84. T_0_15( 2, D, E, A, B, C);
  85. T_0_15( 3, C, D, E, A, B);
  86. T_0_15( 4, B, C, D, E, A);
  87. T_0_15( 5, A, B, C, D, E);
  88. T_0_15( 6, E, A, B, C, D);
  89. T_0_15( 7, D, E, A, B, C);
  90. T_0_15( 8, C, D, E, A, B);
  91. T_0_15( 9, B, C, D, E, A);
  92. T_0_15(10, A, B, C, D, E);
  93. T_0_15(11, E, A, B, C, D);
  94. T_0_15(12, D, E, A, B, C);
  95. T_0_15(13, C, D, E, A, B);
  96. T_0_15(14, B, C, D, E, A);
  97. T_0_15(15, A, B, C, D, E);
  98. /* Round 1 - tail. Input from 512-bit mixing array */
  99. T_16_19(16, E, A, B, C, D);
  100. T_16_19(17, D, E, A, B, C);
  101. T_16_19(18, C, D, E, A, B);
  102. T_16_19(19, B, C, D, E, A);
  103. /* Round 2 */
  104. T_20_39(20, A, B, C, D, E);
  105. T_20_39(21, E, A, B, C, D);
  106. T_20_39(22, D, E, A, B, C);
  107. T_20_39(23, C, D, E, A, B);
  108. T_20_39(24, B, C, D, E, A);
  109. T_20_39(25, A, B, C, D, E);
  110. T_20_39(26, E, A, B, C, D);
  111. T_20_39(27, D, E, A, B, C);
  112. T_20_39(28, C, D, E, A, B);
  113. T_20_39(29, B, C, D, E, A);
  114. T_20_39(30, A, B, C, D, E);
  115. T_20_39(31, E, A, B, C, D);
  116. T_20_39(32, D, E, A, B, C);
  117. T_20_39(33, C, D, E, A, B);
  118. T_20_39(34, B, C, D, E, A);
  119. T_20_39(35, A, B, C, D, E);
  120. T_20_39(36, E, A, B, C, D);
  121. T_20_39(37, D, E, A, B, C);
  122. T_20_39(38, C, D, E, A, B);
  123. T_20_39(39, B, C, D, E, A);
  124. /* Round 3 */
  125. T_40_59(40, A, B, C, D, E);
  126. T_40_59(41, E, A, B, C, D);
  127. T_40_59(42, D, E, A, B, C);
  128. T_40_59(43, C, D, E, A, B);
  129. T_40_59(44, B, C, D, E, A);
  130. T_40_59(45, A, B, C, D, E);
  131. T_40_59(46, E, A, B, C, D);
  132. T_40_59(47, D, E, A, B, C);
  133. T_40_59(48, C, D, E, A, B);
  134. T_40_59(49, B, C, D, E, A);
  135. T_40_59(50, A, B, C, D, E);
  136. T_40_59(51, E, A, B, C, D);
  137. T_40_59(52, D, E, A, B, C);
  138. T_40_59(53, C, D, E, A, B);
  139. T_40_59(54, B, C, D, E, A);
  140. T_40_59(55, A, B, C, D, E);
  141. T_40_59(56, E, A, B, C, D);
  142. T_40_59(57, D, E, A, B, C);
  143. T_40_59(58, C, D, E, A, B);
  144. T_40_59(59, B, C, D, E, A);
  145. /* Round 4 */
  146. T_60_79(60, A, B, C, D, E);
  147. T_60_79(61, E, A, B, C, D);
  148. T_60_79(62, D, E, A, B, C);
  149. T_60_79(63, C, D, E, A, B);
  150. T_60_79(64, B, C, D, E, A);
  151. T_60_79(65, A, B, C, D, E);
  152. T_60_79(66, E, A, B, C, D);
  153. T_60_79(67, D, E, A, B, C);
  154. T_60_79(68, C, D, E, A, B);
  155. T_60_79(69, B, C, D, E, A);
  156. T_60_79(70, A, B, C, D, E);
  157. T_60_79(71, E, A, B, C, D);
  158. T_60_79(72, D, E, A, B, C);
  159. T_60_79(73, C, D, E, A, B);
  160. T_60_79(74, B, C, D, E, A);
  161. T_60_79(75, A, B, C, D, E);
  162. T_60_79(76, E, A, B, C, D);
  163. T_60_79(77, D, E, A, B, C);
  164. T_60_79(78, C, D, E, A, B);
  165. T_60_79(79, B, C, D, E, A);
  166. digest[0] += A;
  167. digest[1] += B;
  168. digest[2] += C;
  169. digest[3] += D;
  170. digest[4] += E;
  171. }
  172. EXPORT_SYMBOL(sha_transform);
  173. /**
  174. * sha_init - initialize the vectors for a SHA1 digest
  175. * @buf: vector to initialize
  176. */
  177. void sha_init(__u32 *buf)
  178. {
  179. buf[0] = 0x67452301;
  180. buf[1] = 0xefcdab89;
  181. buf[2] = 0x98badcfe;
  182. buf[3] = 0x10325476;
  183. buf[4] = 0xc3d2e1f0;
  184. }
  185. EXPORT_SYMBOL(sha_init);