2 * Author......: Jens Steube <jens.steube@gmail.com>
6 #define _PBKDF2_SHA512_
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"
36 __constant u64 k_sha512[80] =
38 SHA512C00, SHA512C01, SHA512C02, SHA512C03,
39 SHA512C04, SHA512C05, SHA512C06, SHA512C07,
40 SHA512C08, SHA512C09, SHA512C0a, SHA512C0b,
41 SHA512C0c, SHA512C0d, SHA512C0e, SHA512C0f,
42 SHA512C10, SHA512C11, SHA512C12, SHA512C13,
43 SHA512C14, SHA512C15, SHA512C16, SHA512C17,
44 SHA512C18, SHA512C19, SHA512C1a, SHA512C1b,
45 SHA512C1c, SHA512C1d, SHA512C1e, SHA512C1f,
46 SHA512C20, SHA512C21, SHA512C22, SHA512C23,
47 SHA512C24, SHA512C25, SHA512C26, SHA512C27,
48 SHA512C28, SHA512C29, SHA512C2a, SHA512C2b,
49 SHA512C2c, SHA512C2d, SHA512C2e, SHA512C2f,
50 SHA512C30, SHA512C31, SHA512C32, SHA512C33,
51 SHA512C34, SHA512C35, SHA512C36, SHA512C37,
52 SHA512C38, SHA512C39, SHA512C3a, SHA512C3b,
53 SHA512C3c, SHA512C3d, SHA512C3e, SHA512C3f,
54 SHA512C40, SHA512C41, SHA512C42, SHA512C43,
55 SHA512C44, SHA512C45, SHA512C46, SHA512C47,
56 SHA512C48, SHA512C49, SHA512C4a, SHA512C4b,
57 SHA512C4c, SHA512C4d, SHA512C4e, SHA512C4f,
60 static void sha512_transform (const u64 w[16], u64 dgst[8])
88 #define ROUND_EXPAND() \
90 w0_t = SHA512_EXPAND (we_t, w9_t, w1_t, w0_t); \
91 w1_t = SHA512_EXPAND (wf_t, wa_t, w2_t, w1_t); \
92 w2_t = SHA512_EXPAND (w0_t, wb_t, w3_t, w2_t); \
93 w3_t = SHA512_EXPAND (w1_t, wc_t, w4_t, w3_t); \
94 w4_t = SHA512_EXPAND (w2_t, wd_t, w5_t, w4_t); \
95 w5_t = SHA512_EXPAND (w3_t, we_t, w6_t, w5_t); \
96 w6_t = SHA512_EXPAND (w4_t, wf_t, w7_t, w6_t); \
97 w7_t = SHA512_EXPAND (w5_t, w0_t, w8_t, w7_t); \
98 w8_t = SHA512_EXPAND (w6_t, w1_t, w9_t, w8_t); \
99 w9_t = SHA512_EXPAND (w7_t, w2_t, wa_t, w9_t); \
100 wa_t = SHA512_EXPAND (w8_t, w3_t, wb_t, wa_t); \
101 wb_t = SHA512_EXPAND (w9_t, w4_t, wc_t, wb_t); \
102 wc_t = SHA512_EXPAND (wa_t, w5_t, wd_t, wc_t); \
103 wd_t = SHA512_EXPAND (wb_t, w6_t, we_t, wd_t); \
104 we_t = SHA512_EXPAND (wc_t, w7_t, wf_t, we_t); \
105 wf_t = SHA512_EXPAND (wd_t, w8_t, w0_t, wf_t); \
108 #define ROUND_STEP(i) \
110 SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha512[i + 0]); \
111 SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha512[i + 1]); \
112 SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha512[i + 2]); \
113 SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha512[i + 3]); \
114 SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha512[i + 4]); \
115 SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha512[i + 5]); \
116 SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha512[i + 6]); \
117 SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha512[i + 7]); \
118 SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha512[i + 8]); \
119 SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha512[i + 9]); \
120 SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha512[i + 10]); \
121 SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha512[i + 11]); \
122 SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha512[i + 12]); \
123 SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha512[i + 13]); \
124 SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, we_t, k_sha512[i + 14]); \
125 SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha512[i + 15]); \
130 //#pragma unroll // kernel fails if used
131 for (int i = 16; i < 80; i += 16)
133 ROUND_EXPAND (); ROUND_STEP (i);
146 static void hmac_run (const u64 w1[16], const u64 ipad[8], const u64 opad[8], u64 dgst[8])
157 sha512_transform (w1, dgst);
169 w[ 8] = 0x8000000000000000;
176 w[15] = (128 + 64) * 8;
187 sha512_transform (w, dgst);
190 static void hmac_init (u64 w[16], u64 ipad[8], u64 opad[8])
192 w[ 0] ^= 0x3636363636363636;
193 w[ 1] ^= 0x3636363636363636;
194 w[ 2] ^= 0x3636363636363636;
195 w[ 3] ^= 0x3636363636363636;
196 w[ 4] ^= 0x3636363636363636;
197 w[ 5] ^= 0x3636363636363636;
198 w[ 6] ^= 0x3636363636363636;
199 w[ 7] ^= 0x3636363636363636;
200 w[ 8] ^= 0x3636363636363636;
201 w[ 9] ^= 0x3636363636363636;
202 w[10] ^= 0x3636363636363636;
203 w[11] ^= 0x3636363636363636;
204 w[12] ^= 0x3636363636363636;
205 w[13] ^= 0x3636363636363636;
206 w[14] ^= 0x3636363636363636;
207 w[15] ^= 0x3636363636363636;
218 sha512_transform (w, ipad);
220 w[ 0] ^= 0x6a6a6a6a6a6a6a6a;
221 w[ 1] ^= 0x6a6a6a6a6a6a6a6a;
222 w[ 2] ^= 0x6a6a6a6a6a6a6a6a;
223 w[ 3] ^= 0x6a6a6a6a6a6a6a6a;
224 w[ 4] ^= 0x6a6a6a6a6a6a6a6a;
225 w[ 5] ^= 0x6a6a6a6a6a6a6a6a;
226 w[ 6] ^= 0x6a6a6a6a6a6a6a6a;
227 w[ 7] ^= 0x6a6a6a6a6a6a6a6a;
228 w[ 8] ^= 0x6a6a6a6a6a6a6a6a;
229 w[ 9] ^= 0x6a6a6a6a6a6a6a6a;
230 w[10] ^= 0x6a6a6a6a6a6a6a6a;
231 w[11] ^= 0x6a6a6a6a6a6a6a6a;
232 w[12] ^= 0x6a6a6a6a6a6a6a6a;
233 w[13] ^= 0x6a6a6a6a6a6a6a6a;
234 w[14] ^= 0x6a6a6a6a6a6a6a6a;
235 w[15] ^= 0x6a6a6a6a6a6a6a6a;
246 sha512_transform (w, opad);
249 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m07100_init (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha512_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 pbkdf2_sha512_t *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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
255 const u32 gid = get_global_id (0);
257 if (gid >= gid_max) return;
261 w0[0] = swap_workaround (pws[gid].i[ 0]);
262 w0[1] = swap_workaround (pws[gid].i[ 1]);
263 w0[2] = swap_workaround (pws[gid].i[ 2]);
264 w0[3] = swap_workaround (pws[gid].i[ 3]);
268 w1[0] = swap_workaround (pws[gid].i[ 4]);
269 w1[1] = swap_workaround (pws[gid].i[ 5]);
270 w1[2] = swap_workaround (pws[gid].i[ 6]);
271 w1[3] = swap_workaround (pws[gid].i[ 7]);
275 w2[0] = swap_workaround (pws[gid].i[ 8]);
276 w2[1] = swap_workaround (pws[gid].i[ 9]);
277 w2[2] = swap_workaround (pws[gid].i[10]);
278 w2[3] = swap_workaround (pws[gid].i[11]);
282 w3[0] = swap_workaround (pws[gid].i[12]);
283 w3[1] = swap_workaround (pws[gid].i[13]);
284 w3[2] = swap_workaround (pws[gid].i[14]);
285 w3[3] = swap_workaround (pws[gid].i[15]);
293 u32 salt_len = salt_bufs[salt_pos].salt_len;
295 esalt_buf[ 0] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 0]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 1]));
296 esalt_buf[ 1] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 2]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 3]));
297 esalt_buf[ 2] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 4]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 5]));
298 esalt_buf[ 3] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 6]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 7]));
299 esalt_buf[ 4] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 8]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 9]));
300 esalt_buf[ 5] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[10]), swap_workaround (esalt_bufs[salt_pos].salt_buf[11]));
301 esalt_buf[ 6] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[12]), swap_workaround (esalt_bufs[salt_pos].salt_buf[13]));
302 esalt_buf[ 7] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[14]), swap_workaround (esalt_bufs[salt_pos].salt_buf[15]));
303 esalt_buf[ 8] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[16]), swap_workaround (esalt_bufs[salt_pos].salt_buf[17]));
304 esalt_buf[ 9] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[18]), swap_workaround (esalt_bufs[salt_pos].salt_buf[19]));
305 esalt_buf[10] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[20]), swap_workaround (esalt_bufs[salt_pos].salt_buf[21]));
306 esalt_buf[11] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[22]), swap_workaround (esalt_bufs[salt_pos].salt_buf[23]));
307 esalt_buf[12] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[24]), swap_workaround (esalt_bufs[salt_pos].salt_buf[25]));
308 esalt_buf[13] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[26]), swap_workaround (esalt_bufs[salt_pos].salt_buf[27]));
310 esalt_buf[15] = (128 + salt_len + 4) * 8;
314 w[ 0] = hl32_to_64 (w0[0], w0[1]);
315 w[ 1] = hl32_to_64 (w0[2], w0[3]);
316 w[ 2] = hl32_to_64 (w1[0], w1[1]);
317 w[ 3] = hl32_to_64 (w1[2], w1[3]);
318 w[ 4] = hl32_to_64 (w2[0], w2[1]);
319 w[ 5] = hl32_to_64 (w2[2], w2[3]);
320 w[ 6] = hl32_to_64 (w3[0], w3[1]);
321 w[ 7] = hl32_to_64 (w3[2], w3[3]);
334 hmac_init (w, ipad, opad);
336 tmps[gid].ipad[0] = ipad[0];
337 tmps[gid].ipad[1] = ipad[1];
338 tmps[gid].ipad[2] = ipad[2];
339 tmps[gid].ipad[3] = ipad[3];
340 tmps[gid].ipad[4] = ipad[4];
341 tmps[gid].ipad[5] = ipad[5];
342 tmps[gid].ipad[6] = ipad[6];
343 tmps[gid].ipad[7] = ipad[7];
345 tmps[gid].opad[0] = opad[0];
346 tmps[gid].opad[1] = opad[1];
347 tmps[gid].opad[2] = opad[2];
348 tmps[gid].opad[3] = opad[3];
349 tmps[gid].opad[4] = opad[4];
350 tmps[gid].opad[5] = opad[5];
351 tmps[gid].opad[6] = opad[6];
352 tmps[gid].opad[7] = opad[7];
354 for (u32 i = 0, j = 1; i < 8; i += 8, j += 1)
358 hmac_run (esalt_buf, ipad, opad, dgst);
360 tmps[gid].dgst[i + 0] = dgst[0];
361 tmps[gid].dgst[i + 1] = dgst[1];
362 tmps[gid].dgst[i + 2] = dgst[2];
363 tmps[gid].dgst[i + 3] = dgst[3];
364 tmps[gid].dgst[i + 4] = dgst[4];
365 tmps[gid].dgst[i + 5] = dgst[5];
366 tmps[gid].dgst[i + 6] = dgst[6];
367 tmps[gid].dgst[i + 7] = dgst[7];
369 tmps[gid].out[i + 0] = dgst[0];
370 tmps[gid].out[i + 1] = dgst[1];
371 tmps[gid].out[i + 2] = dgst[2];
372 tmps[gid].out[i + 3] = dgst[3];
373 tmps[gid].out[i + 4] = dgst[4];
374 tmps[gid].out[i + 5] = dgst[5];
375 tmps[gid].out[i + 6] = dgst[6];
376 tmps[gid].out[i + 7] = dgst[7];
380 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m07100_loop (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha512_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 pbkdf2_sha512_t *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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
382 const u32 gid = get_global_id (0);
384 if (gid >= gid_max) return;
388 ipad[0] = tmps[gid].ipad[0];
389 ipad[1] = tmps[gid].ipad[1];
390 ipad[2] = tmps[gid].ipad[2];
391 ipad[3] = tmps[gid].ipad[3];
392 ipad[4] = tmps[gid].ipad[4];
393 ipad[5] = tmps[gid].ipad[5];
394 ipad[6] = tmps[gid].ipad[6];
395 ipad[7] = tmps[gid].ipad[7];
399 opad[0] = tmps[gid].opad[0];
400 opad[1] = tmps[gid].opad[1];
401 opad[2] = tmps[gid].opad[2];
402 opad[3] = tmps[gid].opad[3];
403 opad[4] = tmps[gid].opad[4];
404 opad[5] = tmps[gid].opad[5];
405 opad[6] = tmps[gid].opad[6];
406 opad[7] = tmps[gid].opad[7];
408 for (u32 i = 0; i < 8; i += 8)
412 dgst[0] = tmps[gid].dgst[i + 0];
413 dgst[1] = tmps[gid].dgst[i + 1];
414 dgst[2] = tmps[gid].dgst[i + 2];
415 dgst[3] = tmps[gid].dgst[i + 3];
416 dgst[4] = tmps[gid].dgst[i + 4];
417 dgst[5] = tmps[gid].dgst[i + 5];
418 dgst[6] = tmps[gid].dgst[i + 6];
419 dgst[7] = tmps[gid].dgst[i + 7];
423 out[0] = tmps[gid].out[i + 0];
424 out[1] = tmps[gid].out[i + 1];
425 out[2] = tmps[gid].out[i + 2];
426 out[3] = tmps[gid].out[i + 3];
427 out[4] = tmps[gid].out[i + 4];
428 out[5] = tmps[gid].out[i + 5];
429 out[6] = tmps[gid].out[i + 6];
430 out[7] = tmps[gid].out[i + 7];
432 for (u32 j = 0; j < loop_cnt; j++)
444 w[ 8] = 0x8000000000000000;
451 w[15] = (128 + 64) * 8;
453 hmac_run (w, ipad, opad, dgst);
465 tmps[gid].dgst[i + 0] = dgst[0];
466 tmps[gid].dgst[i + 1] = dgst[1];
467 tmps[gid].dgst[i + 2] = dgst[2];
468 tmps[gid].dgst[i + 3] = dgst[3];
469 tmps[gid].dgst[i + 4] = dgst[4];
470 tmps[gid].dgst[i + 5] = dgst[5];
471 tmps[gid].dgst[i + 6] = dgst[6];
472 tmps[gid].dgst[i + 7] = dgst[7];
474 tmps[gid].out[i + 0] = out[0];
475 tmps[gid].out[i + 1] = out[1];
476 tmps[gid].out[i + 2] = out[2];
477 tmps[gid].out[i + 3] = out[3];
478 tmps[gid].out[i + 4] = out[4];
479 tmps[gid].out[i + 5] = out[5];
480 tmps[gid].out[i + 6] = out[6];
481 tmps[gid].out[i + 7] = out[7];
485 __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m07100_comp (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha512_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 pbkdf2_sha512_t *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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
491 const u32 gid = get_global_id (0);
493 if (gid >= gid_max) return;
495 const u32 lid = get_local_id (0);
497 const u64x a = tmps[gid].out[0];
498 const u64x b = tmps[gid].out[1];
500 const u32x r0 = l32_from_64 (a);
501 const u32x r1 = h32_from_64 (a);
502 const u32x r2 = l32_from_64 (b);
503 const u32x r3 = h32_from_64 (b);
507 #include VECT_COMPARE_M