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"
27 #include "include/rp_gpu.h"
31 #define VECT_COMPARE_S "check_single_vect1_comp4.c"
32 #define VECT_COMPARE_M "check_multi_vect1_comp4.c"
36 #define VECT_COMPARE_S "check_single_vect2_comp4.c"
37 #define VECT_COMPARE_M "check_multi_vect2_comp4.c"
40 __device__ __constant__ u32 lotus_magic_table[256] =
42 0xbd, 0x56, 0xea, 0xf2, 0xa2, 0xf1, 0xac, 0x2a,
43 0xb0, 0x93, 0xd1, 0x9c, 0x1b, 0x33, 0xfd, 0xd0,
44 0x30, 0x04, 0xb6, 0xdc, 0x7d, 0xdf, 0x32, 0x4b,
45 0xf7, 0xcb, 0x45, 0x9b, 0x31, 0xbb, 0x21, 0x5a,
46 0x41, 0x9f, 0xe1, 0xd9, 0x4a, 0x4d, 0x9e, 0xda,
47 0xa0, 0x68, 0x2c, 0xc3, 0x27, 0x5f, 0x80, 0x36,
48 0x3e, 0xee, 0xfb, 0x95, 0x1a, 0xfe, 0xce, 0xa8,
49 0x34, 0xa9, 0x13, 0xf0, 0xa6, 0x3f, 0xd8, 0x0c,
50 0x78, 0x24, 0xaf, 0x23, 0x52, 0xc1, 0x67, 0x17,
51 0xf5, 0x66, 0x90, 0xe7, 0xe8, 0x07, 0xb8, 0x60,
52 0x48, 0xe6, 0x1e, 0x53, 0xf3, 0x92, 0xa4, 0x72,
53 0x8c, 0x08, 0x15, 0x6e, 0x86, 0x00, 0x84, 0xfa,
54 0xf4, 0x7f, 0x8a, 0x42, 0x19, 0xf6, 0xdb, 0xcd,
55 0x14, 0x8d, 0x50, 0x12, 0xba, 0x3c, 0x06, 0x4e,
56 0xec, 0xb3, 0x35, 0x11, 0xa1, 0x88, 0x8e, 0x2b,
57 0x94, 0x99, 0xb7, 0x71, 0x74, 0xd3, 0xe4, 0xbf,
58 0x3a, 0xde, 0x96, 0x0e, 0xbc, 0x0a, 0xed, 0x77,
59 0xfc, 0x37, 0x6b, 0x03, 0x79, 0x89, 0x62, 0xc6,
60 0xd7, 0xc0, 0xd2, 0x7c, 0x6a, 0x8b, 0x22, 0xa3,
61 0x5b, 0x05, 0x5d, 0x02, 0x75, 0xd5, 0x61, 0xe3,
62 0x18, 0x8f, 0x55, 0x51, 0xad, 0x1f, 0x0b, 0x5e,
63 0x85, 0xe5, 0xc2, 0x57, 0x63, 0xca, 0x3d, 0x6c,
64 0xb4, 0xc5, 0xcc, 0x70, 0xb2, 0x91, 0x59, 0x0d,
65 0x47, 0x20, 0xc8, 0x4f, 0x58, 0xe0, 0x01, 0xe2,
66 0x16, 0x38, 0xc4, 0x6f, 0x3b, 0x0f, 0x65, 0x46,
67 0xbe, 0x7e, 0x2d, 0x7b, 0x82, 0xf9, 0x40, 0xb5,
68 0x1d, 0x73, 0xf8, 0xeb, 0x26, 0xc7, 0x87, 0x97,
69 0x25, 0x54, 0xb1, 0x28, 0xaa, 0x98, 0x9d, 0xa5,
70 0x64, 0x6d, 0x7a, 0xd4, 0x10, 0x81, 0x44, 0xef,
71 0x49, 0xd6, 0xae, 0x2e, 0xdd, 0x76, 0x5c, 0x2f,
72 0xa7, 0x1c, 0xc9, 0x09, 0x69, 0x9a, 0x83, 0xcf,
73 0x29, 0x39, 0xb9, 0xe9, 0x4c, 0xff, 0x43, 0xab,
77 #define BOX(S,i) u32x ((S)[(i)])
81 #define BOX(S,i) u32x ((S)[(i).x], (S)[(i).y])
84 __device__ static void lotus_mix (u32x *in, u32 s_lotus_magic_table[256])
88 for (int i = 0; i < 18; i++)
93 for (int j = 0; j < 12; j++)
98 p = (p + s--) & 0xff; p = ((tmp_in >> 0) & 0xff) ^ BOX (s_lotus_magic_table, p); tmp_out |= p << 0;
99 p = (p + s--) & 0xff; p = ((tmp_in >> 8) & 0xff) ^ BOX (s_lotus_magic_table, p); tmp_out |= p << 8;
100 p = (p + s--) & 0xff; p = ((tmp_in >> 16) & 0xff) ^ BOX (s_lotus_magic_table, p); tmp_out |= p << 16;
101 p = (p + s--) & 0xff; p = ((tmp_in >> 24) & 0xff) ^ BOX (s_lotus_magic_table, p); tmp_out |= p << 24;
108 __device__ static void lotus_transform_password (u32x *in, u32x *out, u32 s_lotus_magic_table[256])
110 u32x t = out[3] >> 24;
114 for (int i = 0; i < 4; i++)
116 t ^= (in[i] >> 0) & 0xff; c = BOX (s_lotus_magic_table, t); out[i] ^= c << 0; t = ((out[i] >> 0) & 0xff);
117 t ^= (in[i] >> 8) & 0xff; c = BOX (s_lotus_magic_table, t); out[i] ^= c << 8; t = ((out[i] >> 8) & 0xff);
118 t ^= (in[i] >> 16) & 0xff; c = BOX (s_lotus_magic_table, t); out[i] ^= c << 16; t = ((out[i] >> 16) & 0xff);
119 t ^= (in[i] >> 24) & 0xff; c = BOX (s_lotus_magic_table, t); out[i] ^= c << 24; t = ((out[i] >> 24) & 0xff);
123 __device__ static void pad (u32x w[4], const u32 len)
125 const u32 val = 16 - len;
127 const u32 mask1 = val << 24;
129 const u32 mask2 = val << 16
132 const u32 mask3 = val << 8
136 const u32 mask4 = val << 0
143 case 0: w[0] = mask4;
148 case 1: w[0] |= mask3;
153 case 2: w[0] |= mask2;
158 case 3: w[0] |= mask1;
163 case 4: w[1] = mask4;
167 case 5: w[1] |= mask3;
171 case 6: w[1] |= mask2;
175 case 7: w[1] |= mask1;
179 case 8: w[2] = mask4;
182 case 9: w[2] |= mask3;
185 case 10: w[2] |= mask2;
188 case 11: w[2] |= mask1;
191 case 12: w[3] = mask4;
193 case 13: w[3] |= mask3;
195 case 14: w[3] |= mask2;
197 case 15: w[3] |= mask1;
202 __device__ static void mdtransform_norecalc (u32x state[4], u32x block[4], u32 s_lotus_magic_table[256])
214 x[ 8] = state[0] ^ block[0];
215 x[ 9] = state[1] ^ block[1];
216 x[10] = state[2] ^ block[2];
217 x[11] = state[3] ^ block[3];
219 lotus_mix (x, s_lotus_magic_table);
227 __device__ static void mdtransform (u32x state[4], u32x checksum[4], u32x block[4], u32 s_lotus_magic_table[256])
229 mdtransform_norecalc (state, block, s_lotus_magic_table);
231 lotus_transform_password (block, checksum, s_lotus_magic_table);
234 __device__ static void domino_big_md (const u32x saved_key[16], const u32 size, u32x state[4], u32 s_lotus_magic_table[256])
245 block[0] = saved_key[0];
246 block[1] = saved_key[1];
247 block[2] = saved_key[2];
248 block[3] = saved_key[3];
250 mdtransform (state, checksum, block, s_lotus_magic_table);
252 mdtransform_norecalc (state, checksum, s_lotus_magic_table);
255 __device__ __constant__ gpu_rule_t c_rules[1024];
257 extern "C" __global__ void __launch_bounds__ (256, 1) m08600_m04 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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)
263 const u32 lid = threadIdx.x;
269 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
275 __shared__ u32 s_lotus_magic_table[256];
277 s_lotus_magic_table[lid] = lotus_magic_table[lid];
281 if (gid >= gid_max) return;
285 pw_buf0[0] = pws[gid].i[ 0];
286 pw_buf0[1] = pws[gid].i[ 1];
287 pw_buf0[2] = pws[gid].i[ 2];
288 pw_buf0[3] = pws[gid].i[ 3];
292 pw_buf1[0] = pws[gid].i[ 4];
293 pw_buf1[1] = pws[gid].i[ 5];
294 pw_buf1[2] = pws[gid].i[ 6];
295 pw_buf1[3] = pws[gid].i[ 7];
297 const u32 pw_len = pws[gid].pw_len;
303 for (u32 il_pos = 0; il_pos < rules_cnt; il_pos++)
333 const u32 out_len = apply_rules (c_rules[il_pos].cmds, w0, w1, pw_len);
367 pad (&w[ 0], pw_len & 0xf);
369 else if (pw_len < 32)
371 pad (&w[ 4], pw_len & 0xf);
373 else if (pw_len < 48)
375 pad (&w[ 8], pw_len & 0xf);
377 else if (pw_len < 64)
379 pad (&w[12], pw_len & 0xf);
382 domino_big_md (w, pw_len, state, s_lotus_magic_table);
384 const u32x r0 = state[0];
385 const u32x r1 = state[1];
386 const u32x r2 = state[2];
387 const u32x r3 = state[3];
389 #include VECT_COMPARE_M
393 extern "C" __global__ void __launch_bounds__ (256, 1) m08600_m08 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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)
397 extern "C" __global__ void __launch_bounds__ (256, 1) m08600_m16 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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)
401 extern "C" __global__ void __launch_bounds__ (256, 1) m08600_s04 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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)
407 const u32 lid = threadIdx.x;
413 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
419 __shared__ u32 s_lotus_magic_table[256];
421 s_lotus_magic_table[lid] = lotus_magic_table[lid];
425 if (gid >= gid_max) return;
429 pw_buf0[0] = pws[gid].i[ 0];
430 pw_buf0[1] = pws[gid].i[ 1];
431 pw_buf0[2] = pws[gid].i[ 2];
432 pw_buf0[3] = pws[gid].i[ 3];
436 pw_buf1[0] = pws[gid].i[ 4];
437 pw_buf1[1] = pws[gid].i[ 5];
438 pw_buf1[2] = pws[gid].i[ 6];
439 pw_buf1[3] = pws[gid].i[ 7];
441 const u32 pw_len = pws[gid].pw_len;
447 const u32 search[4] =
449 digests_buf[digests_offset].digest_buf[DGST_R0],
450 digests_buf[digests_offset].digest_buf[DGST_R1],
451 digests_buf[digests_offset].digest_buf[DGST_R2],
452 digests_buf[digests_offset].digest_buf[DGST_R3]
459 for (u32 il_pos = 0; il_pos < rules_cnt; il_pos++)
489 const u32 out_len = apply_rules (c_rules[il_pos].cmds, w0, w1, pw_len);
523 pad (&w[ 0], pw_len & 0xf);
525 else if (pw_len < 32)
527 pad (&w[ 4], pw_len & 0xf);
529 else if (pw_len < 48)
531 pad (&w[ 8], pw_len & 0xf);
533 else if (pw_len < 64)
535 pad (&w[12], pw_len & 0xf);
538 domino_big_md (w, pw_len, state, s_lotus_magic_table);
540 const u32x r0 = state[0];
541 const u32x r1 = state[1];
542 const u32x r2 = state[2];
543 const u32x r3 = state[3];
545 #include VECT_COMPARE_S
549 extern "C" __global__ void __launch_bounds__ (256, 1) m08600_s08 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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)
553 extern "C" __global__ void __launch_bounds__ (256, 1) m08600_s16 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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)