2 * Author......: Jens Steube <jens.steube@gmail.com>
8 #include "include/constants.h"
9 #include "include/kernel_vendor.h"
24 #include "include/kernel_functions.c"
26 #include "common_nv.c"
29 #define VECT_COMPARE_M "check_multi_vect1_comp4.c"
32 __device__ __constant__ u32 k_sha256[64] =
34 SHA256C00, SHA256C01, SHA256C02, SHA256C03,
35 SHA256C04, SHA256C05, SHA256C06, SHA256C07,
36 SHA256C08, SHA256C09, SHA256C0a, SHA256C0b,
37 SHA256C0c, SHA256C0d, SHA256C0e, SHA256C0f,
38 SHA256C10, SHA256C11, SHA256C12, SHA256C13,
39 SHA256C14, SHA256C15, SHA256C16, SHA256C17,
40 SHA256C18, SHA256C19, SHA256C1a, SHA256C1b,
41 SHA256C1c, SHA256C1d, SHA256C1e, SHA256C1f,
42 SHA256C20, SHA256C21, SHA256C22, SHA256C23,
43 SHA256C24, SHA256C25, SHA256C26, SHA256C27,
44 SHA256C28, SHA256C29, SHA256C2a, SHA256C2b,
45 SHA256C2c, SHA256C2d, SHA256C2e, SHA256C2f,
46 SHA256C30, SHA256C31, SHA256C32, SHA256C33,
47 SHA256C34, SHA256C35, SHA256C36, SHA256C37,
48 SHA256C38, SHA256C39, SHA256C3a, SHA256C3b,
49 SHA256C3c, SHA256C3d, SHA256C3e, SHA256C3f,
52 __device__ static void sha256_transform (const u32x w[16], u32x digest[8])
63 u32x w0_t = swap_workaround (w[ 0]);
64 u32x w1_t = swap_workaround (w[ 1]);
65 u32x w2_t = swap_workaround (w[ 2]);
66 u32x w3_t = swap_workaround (w[ 3]);
67 u32x w4_t = swap_workaround (w[ 4]);
68 u32x w5_t = swap_workaround (w[ 5]);
69 u32x w6_t = swap_workaround (w[ 6]);
70 u32x w7_t = swap_workaround (w[ 7]);
71 u32x w8_t = swap_workaround (w[ 8]);
72 u32x w9_t = swap_workaround (w[ 9]);
73 u32x wa_t = swap_workaround (w[10]);
74 u32x wb_t = swap_workaround (w[11]);
75 u32x wc_t = swap_workaround (w[12]);
76 u32x wd_t = swap_workaround (w[13]);
77 u32x we_t = swap_workaround (w[14]);
78 u32x wf_t = swap_workaround (w[15]);
80 #define ROUND_EXPAND() \
82 w0_t = SHA256_EXPAND (we_t, w9_t, w1_t, w0_t); \
83 w1_t = SHA256_EXPAND (wf_t, wa_t, w2_t, w1_t); \
84 w2_t = SHA256_EXPAND (w0_t, wb_t, w3_t, w2_t); \
85 w3_t = SHA256_EXPAND (w1_t, wc_t, w4_t, w3_t); \
86 w4_t = SHA256_EXPAND (w2_t, wd_t, w5_t, w4_t); \
87 w5_t = SHA256_EXPAND (w3_t, we_t, w6_t, w5_t); \
88 w6_t = SHA256_EXPAND (w4_t, wf_t, w7_t, w6_t); \
89 w7_t = SHA256_EXPAND (w5_t, w0_t, w8_t, w7_t); \
90 w8_t = SHA256_EXPAND (w6_t, w1_t, w9_t, w8_t); \
91 w9_t = SHA256_EXPAND (w7_t, w2_t, wa_t, w9_t); \
92 wa_t = SHA256_EXPAND (w8_t, w3_t, wb_t, wa_t); \
93 wb_t = SHA256_EXPAND (w9_t, w4_t, wc_t, wb_t); \
94 wc_t = SHA256_EXPAND (wa_t, w5_t, wd_t, wc_t); \
95 wd_t = SHA256_EXPAND (wb_t, w6_t, we_t, wd_t); \
96 we_t = SHA256_EXPAND (wc_t, w7_t, wf_t, we_t); \
97 wf_t = SHA256_EXPAND (wd_t, w8_t, w0_t, wf_t); \
100 #define ROUND_STEP(i) \
102 SHA256_STEP (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha256[i + 0]); \
103 SHA256_STEP (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha256[i + 1]); \
104 SHA256_STEP (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha256[i + 2]); \
105 SHA256_STEP (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha256[i + 3]); \
106 SHA256_STEP (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha256[i + 4]); \
107 SHA256_STEP (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha256[i + 5]); \
108 SHA256_STEP (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha256[i + 6]); \
109 SHA256_STEP (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha256[i + 7]); \
110 SHA256_STEP (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha256[i + 8]); \
111 SHA256_STEP (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha256[i + 9]); \
112 SHA256_STEP (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha256[i + 10]); \
113 SHA256_STEP (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha256[i + 11]); \
114 SHA256_STEP (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha256[i + 12]); \
115 SHA256_STEP (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha256[i + 13]); \
116 SHA256_STEP (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, we_t, k_sha256[i + 14]); \
117 SHA256_STEP (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha256[i + 15]); \
122 for (int i = 16; i < 64; i += 16)
124 ROUND_EXPAND (); ROUND_STEP (i);
137 __device__ static void sha256_transform_no14 (const u32x w[16], u32x digest[8])
158 sha256_transform (w_t, digest);
161 __device__ static void init_ctx (u32x digest[8])
163 digest[0] = SHA256M_A;
164 digest[1] = SHA256M_B;
165 digest[2] = SHA256M_C;
166 digest[3] = SHA256M_D;
167 digest[4] = SHA256M_E;
168 digest[5] = SHA256M_F;
169 digest[6] = SHA256M_G;
170 digest[7] = SHA256M_H;
173 __device__ static void bzero16 (u32x block[16])
193 __device__ static void bswap8 (u32x block[16])
195 block[ 0] = swap_workaround (block[ 0]);
196 block[ 1] = swap_workaround (block[ 1]);
197 block[ 2] = swap_workaround (block[ 2]);
198 block[ 3] = swap_workaround (block[ 3]);
199 block[ 4] = swap_workaround (block[ 4]);
200 block[ 5] = swap_workaround (block[ 5]);
201 block[ 6] = swap_workaround (block[ 6]);
202 block[ 7] = swap_workaround (block[ 7]);
205 __device__ static u32 memcat16 (u32x block[16], const u32 block_len, const u32x append[4], const u32 append_len)
207 const u32 div = block_len / 4;
215 #if __CUDA_ARCH__ >= 200
217 const int offset_minus_4 = 4 - (block_len & 3);
219 const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff;
221 tmp0 = __byte_perm ( 0, append[0], selector);
222 tmp1 = __byte_perm (append[0], append[1], selector);
223 tmp2 = __byte_perm (append[1], append[2], selector);
224 tmp3 = __byte_perm (append[2], append[3], selector);
225 tmp4 = __byte_perm (append[3], 0, selector);
229 const u32 mod = block_len & 3;
233 case 0: tmp0 = append[0];
239 case 1: tmp0 = append[0] << 8;
240 tmp1 = append[0] >> 24 | append[1] << 8;
241 tmp2 = append[1] >> 24 | append[2] << 8;
242 tmp3 = append[2] >> 24 | append[3] << 8;
243 tmp4 = append[3] >> 24;
245 case 2: tmp0 = append[0] << 16;
246 tmp1 = append[0] >> 16 | append[1] << 16;
247 tmp2 = append[1] >> 16 | append[2] << 16;
248 tmp3 = append[2] >> 16 | append[3] << 16;
249 tmp4 = append[3] >> 16;
251 case 3: tmp0 = append[0] << 24;
252 tmp1 = append[0] >> 8 | append[1] << 24;
253 tmp2 = append[1] >> 8 | append[2] << 24;
254 tmp3 = append[2] >> 8 | append[3] << 24;
255 tmp4 = append[3] >> 8;
263 case 0: block[ 0] |= tmp0;
269 case 1: block[ 1] |= tmp0;
275 case 2: block[ 2] |= tmp0;
281 case 3: block[ 3] |= tmp0;
287 case 4: block[ 4] |= tmp0;
293 case 5: block[ 5] |= tmp0;
299 case 6: block[ 6] |= tmp0;
305 case 7: block[ 7] |= tmp0;
311 case 8: block[ 8] |= tmp0;
317 case 9: block[ 9] |= tmp0;
323 case 10: block[10] |= tmp0;
329 case 11: block[11] |= tmp0;
335 case 12: block[12] |= tmp0;
340 case 13: block[13] |= tmp0;
344 case 14: block[14] |= tmp0;
347 case 15: block[15] |= tmp0;
351 u32 new_len = block_len + append_len;
356 __device__ static u32 memcat16c (u32x block[16], const u32 block_len, const u32x append[4], const u32 append_len, u32x digest[8])
358 const u32 div = block_len / 4;
366 #if __CUDA_ARCH__ >= 200
368 const int offset_minus_4 = 4 - (block_len & 3);
370 const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff;
372 tmp0 = __byte_perm ( 0, append[0], selector);
373 tmp1 = __byte_perm (append[0], append[1], selector);
374 tmp2 = __byte_perm (append[1], append[2], selector);
375 tmp3 = __byte_perm (append[2], append[3], selector);
376 tmp4 = __byte_perm (append[3], 0, selector);
380 const u32 mod = block_len & 3;
384 case 0: tmp0 = append[0];
390 case 1: tmp0 = append[0] << 8;
391 tmp1 = append[0] >> 24 | append[1] << 8;
392 tmp2 = append[1] >> 24 | append[2] << 8;
393 tmp3 = append[2] >> 24 | append[3] << 8;
394 tmp4 = append[3] >> 24;
396 case 2: tmp0 = append[0] << 16;
397 tmp1 = append[0] >> 16 | append[1] << 16;
398 tmp2 = append[1] >> 16 | append[2] << 16;
399 tmp3 = append[2] >> 16 | append[3] << 16;
400 tmp4 = append[3] >> 16;
402 case 3: tmp0 = append[0] << 24;
403 tmp1 = append[0] >> 8 | append[1] << 24;
404 tmp2 = append[1] >> 8 | append[2] << 24;
405 tmp3 = append[2] >> 8 | append[3] << 24;
406 tmp4 = append[3] >> 8;
412 u32x carry[4] = { 0, 0, 0, 0 };
416 case 0: block[ 0] |= tmp0;
422 case 1: block[ 1] |= tmp0;
428 case 2: block[ 2] |= tmp0;
434 case 3: block[ 3] |= tmp0;
440 case 4: block[ 4] |= tmp0;
446 case 5: block[ 5] |= tmp0;
452 case 6: block[ 6] |= tmp0;
458 case 7: block[ 7] |= tmp0;
464 case 8: block[ 8] |= tmp0;
470 case 9: block[ 9] |= tmp0;
476 case 10: block[10] |= tmp0;
482 case 11: block[11] |= tmp0;
488 case 12: block[12] |= tmp0;
494 case 13: block[13] |= tmp0;
500 case 14: block[14] |= tmp0;
506 case 15: block[15] |= tmp0;
514 u32 new_len = block_len + append_len;
520 sha256_transform (block, digest);
533 __device__ static u32 memcat20 (u32x block[20], const u32 block_len, const u32x append[4], const u32 append_len)
535 const u32 div = block_len / 4;
543 #if __CUDA_ARCH__ >= 200
545 const int offset_minus_4 = 4 - (block_len & 3);
547 const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff;
549 tmp0 = __byte_perm ( 0, append[0], selector);
550 tmp1 = __byte_perm (append[0], append[1], selector);
551 tmp2 = __byte_perm (append[1], append[2], selector);
552 tmp3 = __byte_perm (append[2], append[3], selector);
553 tmp4 = __byte_perm (append[3], 0, selector);
557 const u32 mod = block_len & 3;
561 case 0: tmp0 = append[0];
567 case 1: tmp0 = append[0] << 8;
568 tmp1 = append[0] >> 24 | append[1] << 8;
569 tmp2 = append[1] >> 24 | append[2] << 8;
570 tmp3 = append[2] >> 24 | append[3] << 8;
571 tmp4 = append[3] >> 24;
573 case 2: tmp0 = append[0] << 16;
574 tmp1 = append[0] >> 16 | append[1] << 16;
575 tmp2 = append[1] >> 16 | append[2] << 16;
576 tmp3 = append[2] >> 16 | append[3] << 16;
577 tmp4 = append[3] >> 16;
579 case 3: tmp0 = append[0] << 24;
580 tmp1 = append[0] >> 8 | append[1] << 24;
581 tmp2 = append[1] >> 8 | append[2] << 24;
582 tmp3 = append[2] >> 8 | append[3] << 24;
583 tmp4 = append[3] >> 8;
591 case 0: block[ 0] |= tmp0;
597 case 1: block[ 1] |= tmp0;
603 case 2: block[ 2] |= tmp0;
609 case 3: block[ 3] |= tmp0;
615 case 4: block[ 4] |= tmp0;
621 case 5: block[ 5] |= tmp0;
627 case 6: block[ 6] |= tmp0;
633 case 7: block[ 7] |= tmp0;
639 case 8: block[ 8] |= tmp0;
645 case 9: block[ 9] |= tmp0;
651 case 10: block[10] |= tmp0;
657 case 11: block[11] |= tmp0;
663 case 12: block[12] |= tmp0;
669 case 13: block[13] |= tmp0;
675 case 14: block[14] |= tmp0;
681 case 15: block[15] |= tmp0;
689 return block_len + append_len;
692 __device__ static u32 memcat20_x80 (u32x block[20], const u32 block_len, const u32x append[4], const u32 append_len)
694 const u32 div = block_len / 4;
702 #if __CUDA_ARCH__ >= 200
704 const int offset_minus_4 = 4 - (block_len & 3);
706 const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff;
708 tmp0 = __byte_perm ( 0, append[0], selector);
709 tmp1 = __byte_perm (append[0], append[1], selector);
710 tmp2 = __byte_perm (append[1], append[2], selector);
711 tmp3 = __byte_perm (append[2], append[3], selector);
712 tmp4 = __byte_perm (append[3], 0x80, selector);
716 const u32 mod = block_len & 3;
720 case 0: tmp0 = append[0];
726 case 1: tmp0 = append[0] << 8;
727 tmp1 = append[0] >> 24 | append[1] << 8;
728 tmp2 = append[1] >> 24 | append[2] << 8;
729 tmp3 = append[2] >> 24 | append[3] << 8;
730 tmp4 = append[3] >> 24;
732 case 2: tmp0 = append[0] << 16;
733 tmp1 = append[0] >> 16 | append[1] << 16;
734 tmp2 = append[1] >> 16 | append[2] << 16;
735 tmp3 = append[2] >> 16 | append[3] << 16;
736 tmp4 = append[3] >> 16;
738 case 3: tmp0 = append[0] << 24;
739 tmp1 = append[0] >> 8 | append[1] << 24;
740 tmp2 = append[1] >> 8 | append[2] << 24;
741 tmp3 = append[2] >> 8 | append[3] << 24;
742 tmp4 = append[3] >> 8;
750 case 0: block[ 0] |= tmp0;
756 case 1: block[ 1] |= tmp0;
762 case 2: block[ 2] |= tmp0;
768 case 3: block[ 3] |= tmp0;
774 case 4: block[ 4] |= tmp0;
780 case 5: block[ 5] |= tmp0;
786 case 6: block[ 6] |= tmp0;
792 case 7: block[ 7] |= tmp0;
798 case 8: block[ 8] |= tmp0;
804 case 9: block[ 9] |= tmp0;
810 case 10: block[10] |= tmp0;
816 case 11: block[11] |= tmp0;
822 case 12: block[12] |= tmp0;
828 case 13: block[13] |= tmp0;
834 case 14: block[14] |= tmp0;
840 case 15: block[15] |= tmp0;
848 return block_len + append_len;
851 extern "C" __global__ void __launch_bounds__ (256, 1) m07400_init (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, sha256crypt_tmp_t *tmps, void *hooks, const u32 *bitmaps_buf_s1_a, const u32 *bitmaps_buf_s1_b, const u32 *bitmaps_buf_s1_c, const u32 *bitmaps_buf_s1_d, const u32 *bitmaps_buf_s2_a, const u32 *bitmaps_buf_s2_b, const u32 *bitmaps_buf_s2_c, const u32 *bitmaps_buf_s2_d, plain_t *plains_buf, const digest_t *digests_buf, u32 *hashes_shown, const salt_t *salt_bufs, const void *esalt_bufs, u32 *d_return_buf, u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
857 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
859 if (gid >= gid_max) return;
863 w0[0] = pws[gid].i[0];
864 w0[1] = pws[gid].i[1];
865 w0[2] = pws[gid].i[2];
866 w0[3] = pws[gid].i[3];
868 const u32 pw_len = pws[gid].pw_len;
876 salt_buf[0] = salt_bufs[salt_pos].salt_buf[0];
877 salt_buf[1] = salt_bufs[salt_pos].salt_buf[1];
878 salt_buf[2] = salt_bufs[salt_pos].salt_buf[2];
879 salt_buf[3] = salt_bufs[salt_pos].salt_buf[3];
881 u32 salt_len = salt_bufs[salt_pos].salt_len;
887 u32 block_len; // never reaches > 64
888 u32 transform_len; // required for w[15] = len * 8
896 /* Prepare for the real work. */
904 block_len = memcat16 (block, block_len, w0, pw_len);
908 block_len = memcat16 (block, block_len, salt_buf, salt_len);
912 block_len = memcat16 (block, block_len, w0, pw_len);
914 append_0x80_4 (block, block_len);
916 block[15] = swap_workaround (block_len * 8);
918 init_ctx (alt_result);
920 sha256_transform (block, alt_result);
928 u32x alt_result_tmp[8];
930 alt_result_tmp[0] = alt_result[0];
931 alt_result_tmp[1] = alt_result[1];
932 alt_result_tmp[2] = alt_result[2];
933 alt_result_tmp[3] = alt_result[3];
934 alt_result_tmp[4] = 0;
935 alt_result_tmp[5] = 0;
936 alt_result_tmp[6] = 0;
937 alt_result_tmp[7] = 0;
939 truncate_block (alt_result_tmp, pw_len);
941 /* Add the key string. */
943 block_len = memcat16 (block, block_len, w0, pw_len);
945 /* The last part is the salt string. This must be at most 8
946 characters and it ends at the first `$' character (for
947 compatibility with existing implementations). */
949 block_len = memcat16 (block, block_len, salt_buf, salt_len);
951 /* Now get result of this (32 bytes) and add it to the other
954 block_len = memcat16 (block, block_len, alt_result_tmp, pw_len);
956 transform_len = block_len;
958 /* Take the binary representation of the length of the key and for every
959 1 add the alternate sum, for every 0 the key. */
961 alt_result_tmp[0] = alt_result[0];
962 alt_result_tmp[1] = alt_result[1];
963 alt_result_tmp[2] = alt_result[2];
964 alt_result_tmp[3] = alt_result[3];
965 alt_result_tmp[4] = alt_result[4];
966 alt_result_tmp[5] = alt_result[5];
967 alt_result_tmp[6] = alt_result[6];
968 alt_result_tmp[7] = alt_result[7];
970 init_ctx (alt_result);
972 for (u32 j = pw_len; j; j >>= 1)
976 block_len = memcat16c (block, block_len, &alt_result_tmp[0], 16, alt_result);
977 block_len = memcat16c (block, block_len, &alt_result_tmp[4], 16, alt_result);
983 block_len = memcat16c (block, block_len, w0, pw_len, alt_result);
985 transform_len += pw_len;
989 append_0x80_4 (block, block_len);
993 sha256_transform (block, alt_result);
998 block[15] = swap_workaround (transform_len * 8);
1000 sha256_transform (block, alt_result);
1002 bswap8 (alt_result);
1004 tmps[gid].alt_result[0] = alt_result[0];
1005 tmps[gid].alt_result[1] = alt_result[1];
1006 tmps[gid].alt_result[2] = alt_result[2];
1007 tmps[gid].alt_result[3] = alt_result[3];
1008 tmps[gid].alt_result[4] = alt_result[4];
1009 tmps[gid].alt_result[5] = alt_result[5];
1010 tmps[gid].alt_result[6] = alt_result[6];
1011 tmps[gid].alt_result[7] = alt_result[7];
1013 /* Start computation of P byte sequence. */
1021 /* For every character in the password add the entire password. */
1025 for (u32 j = 0; j < pw_len; j++)
1027 block_len = memcat16c (block, block_len, w0, pw_len, p_bytes);
1029 transform_len += pw_len;
1032 /* Finish the digest. */
1034 append_0x80_4 (block, block_len);
1036 if (block_len >= 56)
1038 sha256_transform (block, p_bytes);
1043 block[15] = swap_workaround (transform_len * 8);
1045 sha256_transform (block, p_bytes);
1049 truncate_block (p_bytes, pw_len);
1051 tmps[gid].p_bytes[0] = p_bytes[0];
1052 tmps[gid].p_bytes[1] = p_bytes[1];
1053 tmps[gid].p_bytes[2] = p_bytes[2];
1054 tmps[gid].p_bytes[3] = p_bytes[3];
1056 /* Start computation of S byte sequence. */
1064 /* For every character in the password add the entire password. */
1068 for (u32 j = 0; j < 16 + (alt_result[0] & 0xff); j++)
1070 block_len = memcat16c (block, block_len, salt_buf, salt_len, s_bytes);
1072 transform_len += salt_len;
1075 /* Finish the digest. */
1077 append_0x80_4 (block, block_len);
1079 if (block_len >= 56)
1081 sha256_transform (block, s_bytes);
1086 block[15] = swap_workaround (transform_len * 8);
1088 sha256_transform (block, s_bytes);
1092 truncate_block (s_bytes, salt_len);
1094 tmps[gid].s_bytes[0] = s_bytes[0];
1095 tmps[gid].s_bytes[1] = s_bytes[1];
1096 tmps[gid].s_bytes[2] = s_bytes[2];
1097 tmps[gid].s_bytes[3] = s_bytes[3];
1100 extern "C" __global__ void __launch_bounds__ (256, 1) m07400_loop (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, sha256crypt_tmp_t *tmps, void *hooks, const u32 *bitmaps_buf_s1_a, const u32 *bitmaps_buf_s1_b, const u32 *bitmaps_buf_s1_c, const u32 *bitmaps_buf_s1_d, const u32 *bitmaps_buf_s2_a, const u32 *bitmaps_buf_s2_b, const u32 *bitmaps_buf_s2_c, const u32 *bitmaps_buf_s2_d, plain_t *plains_buf, const digest_t *digests_buf, u32 *hashes_shown, const salt_t *salt_bufs, const void *esalt_bufs, u32 *d_return_buf, u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
1106 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
1108 if (gid >= gid_max) return;
1110 const u32 pw_len = pws[gid].pw_len;
1118 p_bytes[0] = tmps[gid].p_bytes[0];
1119 p_bytes[1] = tmps[gid].p_bytes[1];
1120 p_bytes[2] = tmps[gid].p_bytes[2];
1121 p_bytes[3] = tmps[gid].p_bytes[3];
1123 u32x p_bytes_x80[4];
1125 p_bytes_x80[0] = tmps[gid].p_bytes[0];
1126 p_bytes_x80[1] = tmps[gid].p_bytes[1];
1127 p_bytes_x80[2] = tmps[gid].p_bytes[2];
1128 p_bytes_x80[3] = tmps[gid].p_bytes[3];
1130 append_0x80_1 (p_bytes_x80, pw_len);
1134 s_bytes[0] = tmps[gid].s_bytes[0];
1135 s_bytes[1] = tmps[gid].s_bytes[1];
1136 s_bytes[2] = tmps[gid].s_bytes[2];
1137 s_bytes[3] = tmps[gid].s_bytes[3];
1141 alt_result[0] = tmps[gid].alt_result[0];
1142 alt_result[1] = tmps[gid].alt_result[1];
1143 alt_result[2] = tmps[gid].alt_result[2];
1144 alt_result[3] = tmps[gid].alt_result[3];
1145 alt_result[4] = tmps[gid].alt_result[4];
1146 alt_result[5] = tmps[gid].alt_result[5];
1147 alt_result[6] = tmps[gid].alt_result[6];
1148 alt_result[7] = tmps[gid].alt_result[7];
1150 u32 salt_len = salt_bufs[salt_pos].salt_len;
1152 /* Repeatedly run the collected hash value through SHA256 to burn
1155 for (u32 i = 0, j = loop_pos; i < loop_cnt; i++, j++)
1163 bzero16 (&block[ 0]);
1164 bzero16 (&block[16]);
1168 const u32 j1 = (j & 1) ? 1 : 0;
1169 const u32 j3 = (j % 3) ? 1 : 0;
1170 const u32 j7 = (j % 7) ? 1 : 0;
1174 block[0] = p_bytes[0];
1175 block[1] = p_bytes[1];
1176 block[2] = p_bytes[2];
1177 block[3] = p_bytes[3];
1183 block[0] = alt_result[0];
1184 block[1] = alt_result[1];
1185 block[2] = alt_result[2];
1186 block[3] = alt_result[3];
1187 block[4] = alt_result[4];
1188 block[5] = alt_result[5];
1189 block[6] = alt_result[6];
1190 block[7] = alt_result[7];
1197 block_len = memcat20 (block, block_len, s_bytes, salt_len);
1202 block_len = memcat20 (block, block_len, p_bytes, pw_len);
1207 block_len = memcat20 (block, block_len, &alt_result[0], 16);
1208 block_len = memcat20_x80 (block, block_len, &alt_result[4], 16);
1212 block_len = memcat20 (block, block_len, p_bytes_x80, pw_len);
1215 if (block_len >= 56)
1217 sha256_transform (block, tmp);
1219 block[ 0] = block[16];
1220 block[ 1] = block[17];
1221 block[ 2] = block[18];
1222 block[ 3] = block[19];
1237 block[15] = swap_workaround (block_len * 8);
1239 sha256_transform_no14 (block, tmp);
1243 alt_result[0] = tmp[0];
1244 alt_result[1] = tmp[1];
1245 alt_result[2] = tmp[2];
1246 alt_result[3] = tmp[3];
1247 alt_result[4] = tmp[4];
1248 alt_result[5] = tmp[5];
1249 alt_result[6] = tmp[6];
1250 alt_result[7] = tmp[7];
1253 tmps[gid].alt_result[0] = alt_result[0];
1254 tmps[gid].alt_result[1] = alt_result[1];
1255 tmps[gid].alt_result[2] = alt_result[2];
1256 tmps[gid].alt_result[3] = alt_result[3];
1257 tmps[gid].alt_result[4] = alt_result[4];
1258 tmps[gid].alt_result[5] = alt_result[5];
1259 tmps[gid].alt_result[6] = alt_result[6];
1260 tmps[gid].alt_result[7] = alt_result[7];
1263 extern "C" __global__ void __launch_bounds__ (256, 1) m07400_comp (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, sha256crypt_tmp_t *tmps, void *hooks, const u32 *bitmaps_buf_s1_a, const u32 *bitmaps_buf_s1_b, const u32 *bitmaps_buf_s1_c, const u32 *bitmaps_buf_s1_d, const u32 *bitmaps_buf_s2_a, const u32 *bitmaps_buf_s2_b, const u32 *bitmaps_buf_s2_c, const u32 *bitmaps_buf_s2_d, plain_t *plains_buf, const digest_t *digests_buf, u32 *hashes_shown, const salt_t *salt_bufs, const void *esalt_bufs, u32 *d_return_buf, u32 *d_scryptV_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
1269 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
1271 if (gid >= gid_max) return;
1273 const u32 lid = threadIdx.x;
1275 const u32x r0 = tmps[gid].alt_result[0];
1276 const u32x r1 = tmps[gid].alt_result[1];
1277 const u32x r2 = tmps[gid].alt_result[2];
1278 const u32x r3 = tmps[gid].alt_result[3];
1282 #include VECT_COMPARE_M