2 * Author......: Jens Steube <jens.steube@gmail.com>
8 #include "include/constants.h"
9 #include "include/kernel_vendor.h"
28 #include "include/kernel_functions.c"
29 #include "types_amd.c"
30 #include "common_amd.c"
33 #define VECT_COMPARE_S "check_single_vect1_comp4.c"
34 #define VECT_COMPARE_M "check_multi_vect1_comp4.c"
38 #define VECT_COMPARE_S "check_single_vect2_comp4.c"
39 #define VECT_COMPARE_M "check_multi_vect2_comp4.c"
43 #define VECT_COMPARE_S "check_single_vect4_comp4.c"
44 #define VECT_COMPARE_M "check_multi_vect4_comp4.c"
47 __constant u64 k_sha384[80] =
49 SHA384C00, SHA384C01, SHA384C02, SHA384C03,
50 SHA384C04, SHA384C05, SHA384C06, SHA384C07,
51 SHA384C08, SHA384C09, SHA384C0a, SHA384C0b,
52 SHA384C0c, SHA384C0d, SHA384C0e, SHA384C0f,
53 SHA384C10, SHA384C11, SHA384C12, SHA384C13,
54 SHA384C14, SHA384C15, SHA384C16, SHA384C17,
55 SHA384C18, SHA384C19, SHA384C1a, SHA384C1b,
56 SHA384C1c, SHA384C1d, SHA384C1e, SHA384C1f,
57 SHA384C20, SHA384C21, SHA384C22, SHA384C23,
58 SHA384C24, SHA384C25, SHA384C26, SHA384C27,
59 SHA384C28, SHA384C29, SHA384C2a, SHA384C2b,
60 SHA384C2c, SHA384C2d, SHA384C2e, SHA384C2f,
61 SHA384C30, SHA384C31, SHA384C32, SHA384C33,
62 SHA384C34, SHA384C35, SHA384C36, SHA384C37,
63 SHA384C38, SHA384C39, SHA384C3a, SHA384C3b,
64 SHA384C3c, SHA384C3d, SHA384C3e, SHA384C3f,
65 SHA384C40, SHA384C41, SHA384C42, SHA384C43,
66 SHA384C44, SHA384C45, SHA384C46, SHA384C47,
67 SHA384C48, SHA384C49, SHA384C4a, SHA384C4b,
68 SHA384C4c, SHA384C4d, SHA384C4e, SHA384C4f,
71 static void sha384_transform (const u32x w0[4], const u32x w1[4], const u32x w2[4], const u32x w3[4], u64x digest[8])
73 u64x w0_t = hl32_to_64 (w0[0], w0[1]);
74 u64x w1_t = hl32_to_64 (w0[2], w0[3]);
75 u64x w2_t = hl32_to_64 (w1[0], w1[1]);
76 u64x w3_t = hl32_to_64 (w1[2], w1[3]);
77 u64x w4_t = hl32_to_64 (w2[0], w2[1]);
78 u64x w5_t = hl32_to_64 (w2[2], w2[3]);
79 u64x w6_t = hl32_to_64 (w3[0], w3[1]);
88 u64x wf_t = hl32_to_64 (w3[2], w3[3]);
99 #define ROUND_EXPAND() \
101 w0_t = SHA384_EXPAND (we_t, w9_t, w1_t, w0_t); \
102 w1_t = SHA384_EXPAND (wf_t, wa_t, w2_t, w1_t); \
103 w2_t = SHA384_EXPAND (w0_t, wb_t, w3_t, w2_t); \
104 w3_t = SHA384_EXPAND (w1_t, wc_t, w4_t, w3_t); \
105 w4_t = SHA384_EXPAND (w2_t, wd_t, w5_t, w4_t); \
106 w5_t = SHA384_EXPAND (w3_t, we_t, w6_t, w5_t); \
107 w6_t = SHA384_EXPAND (w4_t, wf_t, w7_t, w6_t); \
108 w7_t = SHA384_EXPAND (w5_t, w0_t, w8_t, w7_t); \
109 w8_t = SHA384_EXPAND (w6_t, w1_t, w9_t, w8_t); \
110 w9_t = SHA384_EXPAND (w7_t, w2_t, wa_t, w9_t); \
111 wa_t = SHA384_EXPAND (w8_t, w3_t, wb_t, wa_t); \
112 wb_t = SHA384_EXPAND (w9_t, w4_t, wc_t, wb_t); \
113 wc_t = SHA384_EXPAND (wa_t, w5_t, wd_t, wc_t); \
114 wd_t = SHA384_EXPAND (wb_t, w6_t, we_t, wd_t); \
115 we_t = SHA384_EXPAND (wc_t, w7_t, wf_t, we_t); \
116 wf_t = SHA384_EXPAND (wd_t, w8_t, w0_t, wf_t); \
119 #define ROUND_STEP(i) \
121 SHA384_STEP (SHA384_F0o, SHA384_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha384[i + 0]); \
122 SHA384_STEP (SHA384_F0o, SHA384_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha384[i + 1]); \
123 SHA384_STEP (SHA384_F0o, SHA384_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha384[i + 2]); \
124 SHA384_STEP (SHA384_F0o, SHA384_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha384[i + 3]); \
125 SHA384_STEP (SHA384_F0o, SHA384_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha384[i + 4]); \
126 SHA384_STEP (SHA384_F0o, SHA384_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha384[i + 5]); \
127 SHA384_STEP (SHA384_F0o, SHA384_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha384[i + 6]); \
128 SHA384_STEP (SHA384_F0o, SHA384_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha384[i + 7]); \
129 SHA384_STEP (SHA384_F0o, SHA384_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha384[i + 8]); \
130 SHA384_STEP (SHA384_F0o, SHA384_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha384[i + 9]); \
131 SHA384_STEP (SHA384_F0o, SHA384_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha384[i + 10]); \
132 SHA384_STEP (SHA384_F0o, SHA384_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha384[i + 11]); \
133 SHA384_STEP (SHA384_F0o, SHA384_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha384[i + 12]); \
134 SHA384_STEP (SHA384_F0o, SHA384_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha384[i + 13]); \
135 SHA384_STEP (SHA384_F0o, SHA384_F1o, c, d, e, f, g, h, a, b, we_t, k_sha384[i + 14]); \
136 SHA384_STEP (SHA384_F0o, SHA384_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha384[i + 15]); \
141 ROUND_EXPAND (); ROUND_STEP (16);
142 ROUND_EXPAND (); ROUND_STEP (32);
143 ROUND_EXPAND (); ROUND_STEP (48);
144 ROUND_EXPAND (); ROUND_STEP (64);
167 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10800_m04 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global 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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
173 const u32 lid = get_local_id (0);
179 const u32 gid = get_global_id (0);
181 if (gid >= gid_max) return;
185 wordl0[0] = pws[gid].i[ 0];
186 wordl0[1] = pws[gid].i[ 1];
187 wordl0[2] = pws[gid].i[ 2];
188 wordl0[3] = pws[gid].i[ 3];
192 wordl1[0] = pws[gid].i[ 4];
193 wordl1[1] = pws[gid].i[ 5];
194 wordl1[2] = pws[gid].i[ 6];
195 wordl1[3] = pws[gid].i[ 7];
211 const u32 pw_l_len = pws[gid].pw_len;
213 if (combs_mode == COMBINATOR_MODE_BASE_RIGHT)
215 append_0x80_2 (wordl0, wordl1, pw_l_len);
217 switch_buffer_by_offset (wordl0, wordl1, wordl2, wordl3, combs_buf[0].pw_len);
224 for (u32 il_pos = 0; il_pos < combs_cnt; il_pos++)
226 const u32 pw_r_len = combs_buf[il_pos].pw_len;
228 const u32 pw_len = pw_l_len + pw_r_len;
232 wordr0[0] = combs_buf[il_pos].i[0];
233 wordr0[1] = combs_buf[il_pos].i[1];
234 wordr0[2] = combs_buf[il_pos].i[2];
235 wordr0[3] = combs_buf[il_pos].i[3];
239 wordr1[0] = combs_buf[il_pos].i[4];
240 wordr1[1] = combs_buf[il_pos].i[5];
241 wordr1[2] = combs_buf[il_pos].i[6];
242 wordr1[3] = combs_buf[il_pos].i[7];
258 if (combs_mode == COMBINATOR_MODE_BASE_LEFT)
260 append_0x80_2 (wordr0, wordr1, pw_r_len);
262 switch_buffer_by_offset (wordr0, wordr1, wordr2, wordr3, pw_l_len);
270 w0[0] = wordl0[0] | wordr0[0];
271 w0[1] = wordl0[1] | wordr0[1];
272 w0[2] = wordl0[2] | wordr0[2];
273 w0[3] = wordl0[3] | wordr0[3];
274 w1[0] = wordl1[0] | wordr1[0];
275 w1[1] = wordl1[1] | wordr1[1];
276 w1[2] = wordl1[2] | wordr1[2];
277 w1[3] = wordl1[3] | wordr1[3];
278 w2[0] = wordl2[0] | wordr2[0];
279 w2[1] = wordl2[1] | wordr2[1];
280 w2[2] = wordl2[2] | wordr2[2];
281 w2[3] = wordl2[3] | wordr2[3];
282 w3[0] = wordl3[0] | wordr3[0];
283 w3[1] = wordl3[1] | wordr3[1];
284 w3[2] = wordl3[2] | wordr3[2];
285 w3[3] = wordl3[3] | wordr3[3];
296 w0_t[0] = swap_workaround (w0[0]);
297 w0_t[1] = swap_workaround (w0[1]);
298 w0_t[2] = swap_workaround (w0[2]);
299 w0_t[3] = swap_workaround (w0[3]);
300 w1_t[0] = swap_workaround (w1[0]);
301 w1_t[1] = swap_workaround (w1[1]);
302 w1_t[2] = swap_workaround (w1[2]);
303 w1_t[3] = swap_workaround (w1[3]);
304 w2_t[0] = swap_workaround (w2[0]);
305 w2_t[1] = swap_workaround (w2[1]);
306 w2_t[2] = swap_workaround (w2[2]);
307 w2_t[3] = swap_workaround (w2[3]);
308 w3_t[0] = swap_workaround (w3[0]);
309 w3_t[1] = swap_workaround (w3[1]);
311 w3_t[3] = pw_len * 8;
315 digest[0] = SHA384M_A;
316 digest[1] = SHA384M_B;
317 digest[2] = SHA384M_C;
318 digest[3] = SHA384M_D;
319 digest[4] = SHA384M_E;
320 digest[5] = SHA384M_F;
321 digest[6] = SHA384M_G;
322 digest[7] = SHA384M_H;
324 sha384_transform (w0_t, w1_t, w2_t, w3_t, digest);
326 const u32x r0 = l32_from_64 (digest[3]);
327 const u32x r1 = h32_from_64 (digest[3]);
328 const u32x r2 = l32_from_64 (digest[2]);
329 const u32x r3 = h32_from_64 (digest[2]);
331 #include VECT_COMPARE_M
335 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10800_m08 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global 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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
339 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10800_m16 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global 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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
343 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10800_s04 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global 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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
349 const u32 lid = get_local_id (0);
355 const u32 gid = get_global_id (0);
357 if (gid >= gid_max) return;
361 wordl0[0] = pws[gid].i[ 0];
362 wordl0[1] = pws[gid].i[ 1];
363 wordl0[2] = pws[gid].i[ 2];
364 wordl0[3] = pws[gid].i[ 3];
368 wordl1[0] = pws[gid].i[ 4];
369 wordl1[1] = pws[gid].i[ 5];
370 wordl1[2] = pws[gid].i[ 6];
371 wordl1[3] = pws[gid].i[ 7];
387 const u32 pw_l_len = pws[gid].pw_len;
389 if (combs_mode == COMBINATOR_MODE_BASE_RIGHT)
391 append_0x80_2 (wordl0, wordl1, pw_l_len);
393 switch_buffer_by_offset (wordl0, wordl1, wordl2, wordl3, combs_buf[0].pw_len);
400 const u32 search[4] =
402 digests_buf[digests_offset].digest_buf[DGST_R0],
403 digests_buf[digests_offset].digest_buf[DGST_R1],
404 digests_buf[digests_offset].digest_buf[DGST_R2],
405 digests_buf[digests_offset].digest_buf[DGST_R3]
412 for (u32 il_pos = 0; il_pos < combs_cnt; il_pos++)
414 const u32 pw_r_len = combs_buf[il_pos].pw_len;
416 const u32 pw_len = pw_l_len + pw_r_len;
420 wordr0[0] = combs_buf[il_pos].i[0];
421 wordr0[1] = combs_buf[il_pos].i[1];
422 wordr0[2] = combs_buf[il_pos].i[2];
423 wordr0[3] = combs_buf[il_pos].i[3];
427 wordr1[0] = combs_buf[il_pos].i[4];
428 wordr1[1] = combs_buf[il_pos].i[5];
429 wordr1[2] = combs_buf[il_pos].i[6];
430 wordr1[3] = combs_buf[il_pos].i[7];
446 if (combs_mode == COMBINATOR_MODE_BASE_LEFT)
448 append_0x80_2 (wordr0, wordr1, pw_r_len);
450 switch_buffer_by_offset (wordr0, wordr1, wordr2, wordr3, pw_l_len);
458 w0[0] = wordl0[0] | wordr0[0];
459 w0[1] = wordl0[1] | wordr0[1];
460 w0[2] = wordl0[2] | wordr0[2];
461 w0[3] = wordl0[3] | wordr0[3];
462 w1[0] = wordl1[0] | wordr1[0];
463 w1[1] = wordl1[1] | wordr1[1];
464 w1[2] = wordl1[2] | wordr1[2];
465 w1[3] = wordl1[3] | wordr1[3];
466 w2[0] = wordl2[0] | wordr2[0];
467 w2[1] = wordl2[1] | wordr2[1];
468 w2[2] = wordl2[2] | wordr2[2];
469 w2[3] = wordl2[3] | wordr2[3];
470 w3[0] = wordl3[0] | wordr3[0];
471 w3[1] = wordl3[1] | wordr3[1];
472 w3[2] = wordl3[2] | wordr3[2];
473 w3[3] = wordl3[3] | wordr3[3];
484 w0_t[0] = swap_workaround (w0[0]);
485 w0_t[1] = swap_workaround (w0[1]);
486 w0_t[2] = swap_workaround (w0[2]);
487 w0_t[3] = swap_workaround (w0[3]);
488 w1_t[0] = swap_workaround (w1[0]);
489 w1_t[1] = swap_workaround (w1[1]);
490 w1_t[2] = swap_workaround (w1[2]);
491 w1_t[3] = swap_workaround (w1[3]);
492 w2_t[0] = swap_workaround (w2[0]);
493 w2_t[1] = swap_workaround (w2[1]);
494 w2_t[2] = swap_workaround (w2[2]);
495 w2_t[3] = swap_workaround (w2[3]);
496 w3_t[0] = swap_workaround (w3[0]);
497 w3_t[1] = swap_workaround (w3[1]);
499 w3_t[3] = pw_len * 8;
503 digest[0] = SHA384M_A;
504 digest[1] = SHA384M_B;
505 digest[2] = SHA384M_C;
506 digest[3] = SHA384M_D;
507 digest[4] = SHA384M_E;
508 digest[5] = SHA384M_F;
509 digest[6] = SHA384M_G;
510 digest[7] = SHA384M_H;
512 sha384_transform (w0_t, w1_t, w2_t, w3_t, digest);
514 const u32x r0 = l32_from_64 (digest[3]);
515 const u32x r1 = h32_from_64 (digest[3]);
516 const u32x r2 = l32_from_64 (digest[2]);
517 const u32x r3 = h32_from_64 (digest[2]);
519 #include VECT_COMPARE_S
523 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10800_s08 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global 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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
527 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10800_s16 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global void *esalt_bufs, __global u32 *d_return_buf, __global 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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)