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_M "check_multi_vect1_comp4.c"
37 #define VECT_COMPARE_M "check_multi_vect2_comp4.c"
41 #define VECT_COMPARE_M "check_multi_vect4_comp4.c"
44 __constant u64 k_sha512[80] =
46 SHA512C00, SHA512C01, SHA512C02, SHA512C03,
47 SHA512C04, SHA512C05, SHA512C06, SHA512C07,
48 SHA512C08, SHA512C09, SHA512C0a, SHA512C0b,
49 SHA512C0c, SHA512C0d, SHA512C0e, SHA512C0f,
50 SHA512C10, SHA512C11, SHA512C12, SHA512C13,
51 SHA512C14, SHA512C15, SHA512C16, SHA512C17,
52 SHA512C18, SHA512C19, SHA512C1a, SHA512C1b,
53 SHA512C1c, SHA512C1d, SHA512C1e, SHA512C1f,
54 SHA512C20, SHA512C21, SHA512C22, SHA512C23,
55 SHA512C24, SHA512C25, SHA512C26, SHA512C27,
56 SHA512C28, SHA512C29, SHA512C2a, SHA512C2b,
57 SHA512C2c, SHA512C2d, SHA512C2e, SHA512C2f,
58 SHA512C30, SHA512C31, SHA512C32, SHA512C33,
59 SHA512C34, SHA512C35, SHA512C36, SHA512C37,
60 SHA512C38, SHA512C39, SHA512C3a, SHA512C3b,
61 SHA512C3c, SHA512C3d, SHA512C3e, SHA512C3f,
62 SHA512C40, SHA512C41, SHA512C42, SHA512C43,
63 SHA512C44, SHA512C45, SHA512C46, SHA512C47,
64 SHA512C48, SHA512C49, SHA512C4a, SHA512C4b,
65 SHA512C4c, SHA512C4d, SHA512C4e, SHA512C4f,
68 static void sha512_transform (const u64x w0[4], const u64x w1[4], const u64x w2[4], const u64x w3[4], u64x digest[8])
96 #define ROUND_EXPAND() \
98 w0_t = SHA512_EXPAND (we_t, w9_t, w1_t, w0_t); \
99 w1_t = SHA512_EXPAND (wf_t, wa_t, w2_t, w1_t); \
100 w2_t = SHA512_EXPAND (w0_t, wb_t, w3_t, w2_t); \
101 w3_t = SHA512_EXPAND (w1_t, wc_t, w4_t, w3_t); \
102 w4_t = SHA512_EXPAND (w2_t, wd_t, w5_t, w4_t); \
103 w5_t = SHA512_EXPAND (w3_t, we_t, w6_t, w5_t); \
104 w6_t = SHA512_EXPAND (w4_t, wf_t, w7_t, w6_t); \
105 w7_t = SHA512_EXPAND (w5_t, w0_t, w8_t, w7_t); \
106 w8_t = SHA512_EXPAND (w6_t, w1_t, w9_t, w8_t); \
107 w9_t = SHA512_EXPAND (w7_t, w2_t, wa_t, w9_t); \
108 wa_t = SHA512_EXPAND (w8_t, w3_t, wb_t, wa_t); \
109 wb_t = SHA512_EXPAND (w9_t, w4_t, wc_t, wb_t); \
110 wc_t = SHA512_EXPAND (wa_t, w5_t, wd_t, wc_t); \
111 wd_t = SHA512_EXPAND (wb_t, w6_t, we_t, wd_t); \
112 we_t = SHA512_EXPAND (wc_t, w7_t, wf_t, we_t); \
113 wf_t = SHA512_EXPAND (wd_t, w8_t, w0_t, wf_t); \
116 #define ROUND_STEP(i) \
118 SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha512[i + 0]); \
119 SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha512[i + 1]); \
120 SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha512[i + 2]); \
121 SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha512[i + 3]); \
122 SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha512[i + 4]); \
123 SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha512[i + 5]); \
124 SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha512[i + 6]); \
125 SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha512[i + 7]); \
126 SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha512[i + 8]); \
127 SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha512[i + 9]); \
128 SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha512[i + 10]); \
129 SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha512[i + 11]); \
130 SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha512[i + 12]); \
131 SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha512[i + 13]); \
132 SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, we_t, k_sha512[i + 14]); \
133 SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha512[i + 15]); \
138 //#pragma unroll // kernel fails if used
139 for (int i = 16; i < 80; i += 16)
141 ROUND_EXPAND (); ROUND_STEP (i);
154 static void hmac_sha512_pad (u64x w0[4], u64x w1[4], u64x w2[4], u64x w3[4], u64x ipad[8], u64x opad[8])
156 w0[0] = w0[0] ^ 0x3636363636363636;
157 w0[1] = w0[1] ^ 0x3636363636363636;
158 w0[2] = w0[2] ^ 0x3636363636363636;
159 w0[3] = w0[3] ^ 0x3636363636363636;
160 w1[0] = w1[0] ^ 0x3636363636363636;
161 w1[1] = w1[1] ^ 0x3636363636363636;
162 w1[2] = w1[2] ^ 0x3636363636363636;
163 w1[3] = w1[3] ^ 0x3636363636363636;
164 w2[0] = w2[0] ^ 0x3636363636363636;
165 w2[1] = w2[1] ^ 0x3636363636363636;
166 w2[2] = w2[2] ^ 0x3636363636363636;
167 w2[3] = w2[3] ^ 0x3636363636363636;
168 w3[0] = w3[0] ^ 0x3636363636363636;
169 w3[1] = w3[1] ^ 0x3636363636363636;
170 w3[2] = w3[2] ^ 0x3636363636363636;
171 w3[3] = w3[3] ^ 0x3636363636363636;
182 sha512_transform (w0, w1, w2, w3, ipad);
184 w0[0] = w0[0] ^ 0x6a6a6a6a6a6a6a6a;
185 w0[1] = w0[1] ^ 0x6a6a6a6a6a6a6a6a;
186 w0[2] = w0[2] ^ 0x6a6a6a6a6a6a6a6a;
187 w0[3] = w0[3] ^ 0x6a6a6a6a6a6a6a6a;
188 w1[0] = w1[0] ^ 0x6a6a6a6a6a6a6a6a;
189 w1[1] = w1[1] ^ 0x6a6a6a6a6a6a6a6a;
190 w1[2] = w1[2] ^ 0x6a6a6a6a6a6a6a6a;
191 w1[3] = w1[3] ^ 0x6a6a6a6a6a6a6a6a;
192 w2[0] = w2[0] ^ 0x6a6a6a6a6a6a6a6a;
193 w2[1] = w2[1] ^ 0x6a6a6a6a6a6a6a6a;
194 w2[2] = w2[2] ^ 0x6a6a6a6a6a6a6a6a;
195 w2[3] = w2[3] ^ 0x6a6a6a6a6a6a6a6a;
196 w3[0] = w3[0] ^ 0x6a6a6a6a6a6a6a6a;
197 w3[1] = w3[1] ^ 0x6a6a6a6a6a6a6a6a;
198 w3[2] = w3[2] ^ 0x6a6a6a6a6a6a6a6a;
199 w3[3] = w3[3] ^ 0x6a6a6a6a6a6a6a6a;
210 sha512_transform (w0, w1, w2, w3, opad);
213 static void hmac_sha512_run (u64x w0[4], u64x w1[4], u64x w2[4], u64x w3[4], u64x ipad[8], u64x opad[8], u64x digest[8])
224 sha512_transform (w0, w1, w2, w3, digest);
234 w2[0] = 0x8000000000000000;
241 w3[3] = (128 + 64) * 8;
252 sha512_transform (w0, w1, w2, w3, digest);
255 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06500_init (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global sha512aix_tmp_t *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 wpa_t *wpa_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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
261 const u32 gid = get_global_id (0);
263 if (gid >= gid_max) return;
267 w0[0] = pws[gid].i[ 0];
268 w0[1] = pws[gid].i[ 1];
269 w0[2] = pws[gid].i[ 2];
270 w0[3] = pws[gid].i[ 3];
274 w1[0] = pws[gid].i[ 4];
275 w1[1] = pws[gid].i[ 5];
276 w1[2] = pws[gid].i[ 6];
277 w1[3] = pws[gid].i[ 7];
281 w2[0] = pws[gid].i[ 8];
282 w2[1] = pws[gid].i[ 9];
283 w2[2] = pws[gid].i[10];
284 w2[3] = pws[gid].i[11];
288 w3[0] = pws[gid].i[12];
289 w3[1] = pws[gid].i[13];
290 w3[2] = pws[gid].i[14];
291 w3[3] = pws[gid].i[15];
297 u32 salt_len = salt_bufs[salt_pos].salt_len;
301 salt_buf0[0] = salt_bufs[salt_pos].salt_buf[ 0];
302 salt_buf0[1] = salt_bufs[salt_pos].salt_buf[ 1];
303 salt_buf0[2] = salt_bufs[salt_pos].salt_buf[ 2];
304 salt_buf0[3] = salt_bufs[salt_pos].salt_buf[ 3];
308 salt_buf1[0] = salt_bufs[salt_pos].salt_buf[ 4];
309 salt_buf1[1] = salt_bufs[salt_pos].salt_buf[ 5];
310 salt_buf1[2] = salt_bufs[salt_pos].salt_buf[ 6];
311 salt_buf1[3] = salt_bufs[salt_pos].salt_buf[ 7];
315 salt_buf2[0] = salt_bufs[salt_pos].salt_buf[ 8];
316 salt_buf2[1] = salt_bufs[salt_pos].salt_buf[ 9];
317 salt_buf2[2] = salt_bufs[salt_pos].salt_buf[10];
318 salt_buf2[3] = salt_bufs[salt_pos].salt_buf[11];
327 append_0x01_4 (salt_buf0, salt_buf1, salt_buf2, salt_buf3, salt_len + 3);
329 append_0x80_4 (salt_buf0, salt_buf1, salt_buf2, salt_buf3, salt_len + 4);
335 w0[0] = swap_workaround (w0[0]);
336 w0[1] = swap_workaround (w0[1]);
337 w0[2] = swap_workaround (w0[2]);
338 w0[3] = swap_workaround (w0[3]);
339 w1[0] = swap_workaround (w1[0]);
340 w1[1] = swap_workaround (w1[1]);
341 w1[2] = swap_workaround (w1[2]);
342 w1[3] = swap_workaround (w1[3]);
343 w2[0] = swap_workaround (w2[0]);
344 w2[1] = swap_workaround (w2[1]);
345 w2[2] = swap_workaround (w2[2]);
346 w2[3] = swap_workaround (w2[3]);
347 w3[0] = swap_workaround (w3[0]);
348 w3[1] = swap_workaround (w3[1]);
349 w3[2] = swap_workaround (w3[2]);
350 w3[3] = swap_workaround (w3[3]);
357 w0l[0] = (u64x) (w0[0]) << 32 | (u64x) (w0[1]);
358 w0l[1] = (u64x) (w0[2]) << 32 | (u64x) (w0[3]);
359 w0l[2] = (u64x) (w1[0]) << 32 | (u64x) (w1[1]);
360 w0l[3] = (u64x) (w1[2]) << 32 | (u64x) (w1[3]);
361 w1l[0] = (u64x) (w2[0]) << 32 | (u64x) (w2[1]);
362 w1l[1] = (u64x) (w2[2]) << 32 | (u64x) (w2[3]);
363 w1l[2] = (u64x) (w3[0]) << 32 | (u64x) (w3[1]);
364 w1l[3] = (u64x) (w3[2]) << 32 | (u64x) (w3[3]);
377 hmac_sha512_pad (w0l, w1l, w2l, w3l, ipad, opad);
379 tmps[gid].ipad[0] = ipad[0];
380 tmps[gid].ipad[1] = ipad[1];
381 tmps[gid].ipad[2] = ipad[2];
382 tmps[gid].ipad[3] = ipad[3];
383 tmps[gid].ipad[4] = ipad[4];
384 tmps[gid].ipad[5] = ipad[5];
385 tmps[gid].ipad[6] = ipad[6];
386 tmps[gid].ipad[7] = ipad[7];
388 tmps[gid].opad[0] = opad[0];
389 tmps[gid].opad[1] = opad[1];
390 tmps[gid].opad[2] = opad[2];
391 tmps[gid].opad[3] = opad[3];
392 tmps[gid].opad[4] = opad[4];
393 tmps[gid].opad[5] = opad[5];
394 tmps[gid].opad[6] = opad[6];
395 tmps[gid].opad[7] = opad[7];
397 w0l[0] = (u64x) salt_buf0[1] << 32 | (u64x) salt_buf0[0];
398 w0l[1] = (u64x) salt_buf0[3] << 32 | (u64x) salt_buf0[2];
399 w0l[2] = (u64x) salt_buf1[1] << 32 | (u64x) salt_buf1[0];
400 w0l[3] = (u64x) salt_buf1[3] << 32 | (u64x) salt_buf1[2];
401 w1l[0] = (u64x) salt_buf2[1] << 32 | (u64x) salt_buf2[0];
402 w1l[1] = (u64x) salt_buf2[3] << 32 | (u64x) salt_buf2[2];
403 w1l[2] = (u64x) salt_buf3[1] << 32 | (u64x) salt_buf3[0];
404 w1l[3] = (u64x) salt_buf3[3] << 32 | (u64x) salt_buf3[2];
414 w0l[0] = swap_workaround (w0l[0]);
415 w0l[1] = swap_workaround (w0l[1]);
416 w0l[2] = swap_workaround (w0l[2]);
417 w0l[3] = swap_workaround (w0l[3]);
418 w1l[0] = swap_workaround (w1l[0]);
419 w1l[1] = swap_workaround (w1l[1]);
420 w1l[2] = swap_workaround (w1l[2]);
421 w1l[3] = swap_workaround (w1l[3]);
429 w3l[3] = (128 + salt_len + 4) * 8;
433 hmac_sha512_run (w0l, w1l, w2l, w3l, ipad, opad, dgst);
435 tmps[gid].dgst[0] = dgst[0];
436 tmps[gid].dgst[1] = dgst[1];
437 tmps[gid].dgst[2] = dgst[2];
438 tmps[gid].dgst[3] = dgst[3];
439 tmps[gid].dgst[4] = dgst[4];
440 tmps[gid].dgst[5] = dgst[5];
441 tmps[gid].dgst[6] = dgst[6];
442 tmps[gid].dgst[7] = dgst[7];
444 tmps[gid].out[0] = dgst[0];
445 tmps[gid].out[1] = dgst[1];
446 tmps[gid].out[2] = dgst[2];
447 tmps[gid].out[3] = dgst[3];
448 tmps[gid].out[4] = dgst[4];
449 tmps[gid].out[5] = dgst[5];
450 tmps[gid].out[6] = dgst[6];
451 tmps[gid].out[7] = dgst[7];
454 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06500_loop (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global sha512aix_tmp_t *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 wpa_t *wpa_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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
456 const u32 gid = get_global_id (0);
458 if (gid >= gid_max) return;
463 ipad[0] = tmps[gid].ipad[0];
464 ipad[1] = tmps[gid].ipad[1];
465 ipad[2] = tmps[gid].ipad[2];
466 ipad[3] = tmps[gid].ipad[3];
467 ipad[4] = tmps[gid].ipad[4];
468 ipad[5] = tmps[gid].ipad[5];
469 ipad[6] = tmps[gid].ipad[6];
470 ipad[7] = tmps[gid].ipad[7];
472 opad[0] = tmps[gid].opad[0];
473 opad[1] = tmps[gid].opad[1];
474 opad[2] = tmps[gid].opad[2];
475 opad[3] = tmps[gid].opad[3];
476 opad[4] = tmps[gid].opad[4];
477 opad[5] = tmps[gid].opad[5];
478 opad[6] = tmps[gid].opad[6];
479 opad[7] = tmps[gid].opad[7];
484 dgst[0] = tmps[gid].dgst[0];
485 dgst[1] = tmps[gid].dgst[1];
486 dgst[2] = tmps[gid].dgst[2];
487 dgst[3] = tmps[gid].dgst[3];
488 dgst[4] = tmps[gid].dgst[4];
489 dgst[5] = tmps[gid].dgst[5];
490 dgst[6] = tmps[gid].dgst[6];
491 dgst[7] = tmps[gid].dgst[7];
493 out[0] = tmps[gid].out[0];
494 out[1] = tmps[gid].out[1];
495 out[2] = tmps[gid].out[2];
496 out[3] = tmps[gid].out[3];
497 out[4] = tmps[gid].out[4];
498 out[5] = tmps[gid].out[5];
499 out[6] = tmps[gid].out[6];
500 out[7] = tmps[gid].out[7];
502 for (u32 j = 0; j < loop_cnt; j++)
517 w2[0] = 0x8000000000000000;
524 w3[3] = (128 + 64) * 8;
526 hmac_sha512_run (w0, w1, w2, w3, ipad, opad, dgst);
538 tmps[gid].dgst[0] = dgst[0];
539 tmps[gid].dgst[1] = dgst[1];
540 tmps[gid].dgst[2] = dgst[2];
541 tmps[gid].dgst[3] = dgst[3];
542 tmps[gid].dgst[4] = dgst[4];
543 tmps[gid].dgst[5] = dgst[5];
544 tmps[gid].dgst[6] = dgst[6];
545 tmps[gid].dgst[7] = dgst[7];
547 tmps[gid].out[0] = out[0];
548 tmps[gid].out[1] = out[1];
549 tmps[gid].out[2] = out[2];
550 tmps[gid].out[3] = out[3];
551 tmps[gid].out[4] = out[4];
552 tmps[gid].out[5] = out[5];
553 tmps[gid].out[6] = out[6];
554 tmps[gid].out[7] = out[7];
557 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06500_comp (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global sha512aix_tmp_t *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 wpa_t *wpa_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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
563 const u32 gid = get_global_id (0);
565 if (gid >= gid_max) return;
567 const u32 lid = get_local_id (0);
574 u64x a = tmps[gid].out[0];
575 u64x b = tmps[gid].out[1];
576 u64x c = tmps[gid].out[2];
577 u64x d = tmps[gid].out[3];
578 u64x e = tmps[gid].out[4];
579 u64x f = tmps[gid].out[5];
580 u64x g = tmps[gid].out[6];
581 u64x h = tmps[gid].out[7] & 0xffffffffffffff00;
584 const u32x r0 = l32_from_64 (tmps[gid].out[0]);
585 const u32x r1 = h32_from_64 (tmps[gid].out[0]);
586 const u32x r2 = l32_from_64 (tmps[gid].out[1]);
587 const u32x r3 = h32_from_64 (tmps[gid].out[1]);
591 #include VECT_COMPARE_M