2 * Author......: Jens Steube <jens.steube@gmail.com>
6 #define _PBKDF2_SHA512_
8 #include "include/constants.h"
9 #include "include/kernel_vendor.h"
16 #include "include/kernel_functions.c"
17 #include "types_ocl.c"
21 #define COMPARE_M "check_multi_vect1_comp4.c"
24 __constant u64 k_sha512[80] =
26 SHA512C00, SHA512C01, SHA512C02, SHA512C03,
27 SHA512C04, SHA512C05, SHA512C06, SHA512C07,
28 SHA512C08, SHA512C09, SHA512C0a, SHA512C0b,
29 SHA512C0c, SHA512C0d, SHA512C0e, SHA512C0f,
30 SHA512C10, SHA512C11, SHA512C12, SHA512C13,
31 SHA512C14, SHA512C15, SHA512C16, SHA512C17,
32 SHA512C18, SHA512C19, SHA512C1a, SHA512C1b,
33 SHA512C1c, SHA512C1d, SHA512C1e, SHA512C1f,
34 SHA512C20, SHA512C21, SHA512C22, SHA512C23,
35 SHA512C24, SHA512C25, SHA512C26, SHA512C27,
36 SHA512C28, SHA512C29, SHA512C2a, SHA512C2b,
37 SHA512C2c, SHA512C2d, SHA512C2e, SHA512C2f,
38 SHA512C30, SHA512C31, SHA512C32, SHA512C33,
39 SHA512C34, SHA512C35, SHA512C36, SHA512C37,
40 SHA512C38, SHA512C39, SHA512C3a, SHA512C3b,
41 SHA512C3c, SHA512C3d, SHA512C3e, SHA512C3f,
42 SHA512C40, SHA512C41, SHA512C42, SHA512C43,
43 SHA512C44, SHA512C45, SHA512C46, SHA512C47,
44 SHA512C48, SHA512C49, SHA512C4a, SHA512C4b,
45 SHA512C4c, SHA512C4d, SHA512C4e, SHA512C4f,
48 static void sha512_transform (const u64 w[16], u64 dgst[8])
76 #define ROUND_EXPAND() \
78 w0_t = SHA512_EXPAND (we_t, w9_t, w1_t, w0_t); \
79 w1_t = SHA512_EXPAND (wf_t, wa_t, w2_t, w1_t); \
80 w2_t = SHA512_EXPAND (w0_t, wb_t, w3_t, w2_t); \
81 w3_t = SHA512_EXPAND (w1_t, wc_t, w4_t, w3_t); \
82 w4_t = SHA512_EXPAND (w2_t, wd_t, w5_t, w4_t); \
83 w5_t = SHA512_EXPAND (w3_t, we_t, w6_t, w5_t); \
84 w6_t = SHA512_EXPAND (w4_t, wf_t, w7_t, w6_t); \
85 w7_t = SHA512_EXPAND (w5_t, w0_t, w8_t, w7_t); \
86 w8_t = SHA512_EXPAND (w6_t, w1_t, w9_t, w8_t); \
87 w9_t = SHA512_EXPAND (w7_t, w2_t, wa_t, w9_t); \
88 wa_t = SHA512_EXPAND (w8_t, w3_t, wb_t, wa_t); \
89 wb_t = SHA512_EXPAND (w9_t, w4_t, wc_t, wb_t); \
90 wc_t = SHA512_EXPAND (wa_t, w5_t, wd_t, wc_t); \
91 wd_t = SHA512_EXPAND (wb_t, w6_t, we_t, wd_t); \
92 we_t = SHA512_EXPAND (wc_t, w7_t, wf_t, we_t); \
93 wf_t = SHA512_EXPAND (wd_t, w8_t, w0_t, wf_t); \
96 #define ROUND_STEP(i) \
98 SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha512[i + 0]); \
99 SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha512[i + 1]); \
100 SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha512[i + 2]); \
101 SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha512[i + 3]); \
102 SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha512[i + 4]); \
103 SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha512[i + 5]); \
104 SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha512[i + 6]); \
105 SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha512[i + 7]); \
106 SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha512[i + 8]); \
107 SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha512[i + 9]); \
108 SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha512[i + 10]); \
109 SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha512[i + 11]); \
110 SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha512[i + 12]); \
111 SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha512[i + 13]); \
112 SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, we_t, k_sha512[i + 14]); \
113 SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha512[i + 15]); \
118 //#pragma unroll // kernel fails if used
119 for (int i = 16; i < 80; i += 16)
121 ROUND_EXPAND (); ROUND_STEP (i);
134 static void hmac_run (const u64 w1[16], const u64 ipad[8], const u64 opad[8], u64 dgst[8])
145 sha512_transform (w1, dgst);
157 w[ 8] = 0x8000000000000000;
164 w[15] = (128 + 64) * 8;
175 sha512_transform (w, dgst);
178 static void hmac_init (u64 w[16], u64 ipad[8], u64 opad[8])
180 w[ 0] ^= 0x3636363636363636;
181 w[ 1] ^= 0x3636363636363636;
182 w[ 2] ^= 0x3636363636363636;
183 w[ 3] ^= 0x3636363636363636;
184 w[ 4] ^= 0x3636363636363636;
185 w[ 5] ^= 0x3636363636363636;
186 w[ 6] ^= 0x3636363636363636;
187 w[ 7] ^= 0x3636363636363636;
188 w[ 8] ^= 0x3636363636363636;
189 w[ 9] ^= 0x3636363636363636;
190 w[10] ^= 0x3636363636363636;
191 w[11] ^= 0x3636363636363636;
192 w[12] ^= 0x3636363636363636;
193 w[13] ^= 0x3636363636363636;
194 w[14] ^= 0x3636363636363636;
195 w[15] ^= 0x3636363636363636;
206 sha512_transform (w, ipad);
208 w[ 0] ^= 0x6a6a6a6a6a6a6a6a;
209 w[ 1] ^= 0x6a6a6a6a6a6a6a6a;
210 w[ 2] ^= 0x6a6a6a6a6a6a6a6a;
211 w[ 3] ^= 0x6a6a6a6a6a6a6a6a;
212 w[ 4] ^= 0x6a6a6a6a6a6a6a6a;
213 w[ 5] ^= 0x6a6a6a6a6a6a6a6a;
214 w[ 6] ^= 0x6a6a6a6a6a6a6a6a;
215 w[ 7] ^= 0x6a6a6a6a6a6a6a6a;
216 w[ 8] ^= 0x6a6a6a6a6a6a6a6a;
217 w[ 9] ^= 0x6a6a6a6a6a6a6a6a;
218 w[10] ^= 0x6a6a6a6a6a6a6a6a;
219 w[11] ^= 0x6a6a6a6a6a6a6a6a;
220 w[12] ^= 0x6a6a6a6a6a6a6a6a;
221 w[13] ^= 0x6a6a6a6a6a6a6a6a;
222 w[14] ^= 0x6a6a6a6a6a6a6a6a;
223 w[15] ^= 0x6a6a6a6a6a6a6a6a;
234 sha512_transform (w, opad);
237 __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)
243 const u32 gid = get_global_id (0);
245 if (gid >= gid_max) return;
249 w0[0] = swap_workaround (pws[gid].i[ 0]);
250 w0[1] = swap_workaround (pws[gid].i[ 1]);
251 w0[2] = swap_workaround (pws[gid].i[ 2]);
252 w0[3] = swap_workaround (pws[gid].i[ 3]);
256 w1[0] = swap_workaround (pws[gid].i[ 4]);
257 w1[1] = swap_workaround (pws[gid].i[ 5]);
258 w1[2] = swap_workaround (pws[gid].i[ 6]);
259 w1[3] = swap_workaround (pws[gid].i[ 7]);
263 w2[0] = swap_workaround (pws[gid].i[ 8]);
264 w2[1] = swap_workaround (pws[gid].i[ 9]);
265 w2[2] = swap_workaround (pws[gid].i[10]);
266 w2[3] = swap_workaround (pws[gid].i[11]);
270 w3[0] = swap_workaround (pws[gid].i[12]);
271 w3[1] = swap_workaround (pws[gid].i[13]);
272 w3[2] = swap_workaround (pws[gid].i[14]);
273 w3[3] = swap_workaround (pws[gid].i[15]);
281 u32 salt_len = salt_bufs[salt_pos].salt_len;
283 esalt_buf[ 0] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 0]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 1]));
284 esalt_buf[ 1] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 2]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 3]));
285 esalt_buf[ 2] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 4]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 5]));
286 esalt_buf[ 3] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 6]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 7]));
287 esalt_buf[ 4] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[ 8]), swap_workaround (esalt_bufs[salt_pos].salt_buf[ 9]));
288 esalt_buf[ 5] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[10]), swap_workaround (esalt_bufs[salt_pos].salt_buf[11]));
289 esalt_buf[ 6] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[12]), swap_workaround (esalt_bufs[salt_pos].salt_buf[13]));
290 esalt_buf[ 7] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[14]), swap_workaround (esalt_bufs[salt_pos].salt_buf[15]));
291 esalt_buf[ 8] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[16]), swap_workaround (esalt_bufs[salt_pos].salt_buf[17]));
292 esalt_buf[ 9] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[18]), swap_workaround (esalt_bufs[salt_pos].salt_buf[19]));
293 esalt_buf[10] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[20]), swap_workaround (esalt_bufs[salt_pos].salt_buf[21]));
294 esalt_buf[11] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[22]), swap_workaround (esalt_bufs[salt_pos].salt_buf[23]));
295 esalt_buf[12] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[24]), swap_workaround (esalt_bufs[salt_pos].salt_buf[25]));
296 esalt_buf[13] = hl32_to_64 (swap_workaround (esalt_bufs[salt_pos].salt_buf[26]), swap_workaround (esalt_bufs[salt_pos].salt_buf[27]));
298 esalt_buf[15] = (128 + salt_len + 4) * 8;
302 w[ 0] = hl32_to_64 (w0[0], w0[1]);
303 w[ 1] = hl32_to_64 (w0[2], w0[3]);
304 w[ 2] = hl32_to_64 (w1[0], w1[1]);
305 w[ 3] = hl32_to_64 (w1[2], w1[3]);
306 w[ 4] = hl32_to_64 (w2[0], w2[1]);
307 w[ 5] = hl32_to_64 (w2[2], w2[3]);
308 w[ 6] = hl32_to_64 (w3[0], w3[1]);
309 w[ 7] = hl32_to_64 (w3[2], w3[3]);
322 hmac_init (w, ipad, opad);
324 tmps[gid].ipad[0] = ipad[0];
325 tmps[gid].ipad[1] = ipad[1];
326 tmps[gid].ipad[2] = ipad[2];
327 tmps[gid].ipad[3] = ipad[3];
328 tmps[gid].ipad[4] = ipad[4];
329 tmps[gid].ipad[5] = ipad[5];
330 tmps[gid].ipad[6] = ipad[6];
331 tmps[gid].ipad[7] = ipad[7];
333 tmps[gid].opad[0] = opad[0];
334 tmps[gid].opad[1] = opad[1];
335 tmps[gid].opad[2] = opad[2];
336 tmps[gid].opad[3] = opad[3];
337 tmps[gid].opad[4] = opad[4];
338 tmps[gid].opad[5] = opad[5];
339 tmps[gid].opad[6] = opad[6];
340 tmps[gid].opad[7] = opad[7];
342 for (u32 i = 0, j = 1; i < 8; i += 8, j += 1)
346 hmac_run (esalt_buf, ipad, opad, dgst);
348 tmps[gid].dgst[i + 0] = dgst[0];
349 tmps[gid].dgst[i + 1] = dgst[1];
350 tmps[gid].dgst[i + 2] = dgst[2];
351 tmps[gid].dgst[i + 3] = dgst[3];
352 tmps[gid].dgst[i + 4] = dgst[4];
353 tmps[gid].dgst[i + 5] = dgst[5];
354 tmps[gid].dgst[i + 6] = dgst[6];
355 tmps[gid].dgst[i + 7] = dgst[7];
357 tmps[gid].out[i + 0] = dgst[0];
358 tmps[gid].out[i + 1] = dgst[1];
359 tmps[gid].out[i + 2] = dgst[2];
360 tmps[gid].out[i + 3] = dgst[3];
361 tmps[gid].out[i + 4] = dgst[4];
362 tmps[gid].out[i + 5] = dgst[5];
363 tmps[gid].out[i + 6] = dgst[6];
364 tmps[gid].out[i + 7] = dgst[7];
368 __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)
370 const u32 gid = get_global_id (0);
372 if (gid >= gid_max) return;
376 ipad[0] = tmps[gid].ipad[0];
377 ipad[1] = tmps[gid].ipad[1];
378 ipad[2] = tmps[gid].ipad[2];
379 ipad[3] = tmps[gid].ipad[3];
380 ipad[4] = tmps[gid].ipad[4];
381 ipad[5] = tmps[gid].ipad[5];
382 ipad[6] = tmps[gid].ipad[6];
383 ipad[7] = tmps[gid].ipad[7];
387 opad[0] = tmps[gid].opad[0];
388 opad[1] = tmps[gid].opad[1];
389 opad[2] = tmps[gid].opad[2];
390 opad[3] = tmps[gid].opad[3];
391 opad[4] = tmps[gid].opad[4];
392 opad[5] = tmps[gid].opad[5];
393 opad[6] = tmps[gid].opad[6];
394 opad[7] = tmps[gid].opad[7];
396 for (u32 i = 0; i < 8; i += 8)
400 dgst[0] = tmps[gid].dgst[i + 0];
401 dgst[1] = tmps[gid].dgst[i + 1];
402 dgst[2] = tmps[gid].dgst[i + 2];
403 dgst[3] = tmps[gid].dgst[i + 3];
404 dgst[4] = tmps[gid].dgst[i + 4];
405 dgst[5] = tmps[gid].dgst[i + 5];
406 dgst[6] = tmps[gid].dgst[i + 6];
407 dgst[7] = tmps[gid].dgst[i + 7];
411 out[0] = tmps[gid].out[i + 0];
412 out[1] = tmps[gid].out[i + 1];
413 out[2] = tmps[gid].out[i + 2];
414 out[3] = tmps[gid].out[i + 3];
415 out[4] = tmps[gid].out[i + 4];
416 out[5] = tmps[gid].out[i + 5];
417 out[6] = tmps[gid].out[i + 6];
418 out[7] = tmps[gid].out[i + 7];
420 for (u32 j = 0; j < loop_cnt; j++)
432 w[ 8] = 0x8000000000000000;
439 w[15] = (128 + 64) * 8;
441 hmac_run (w, ipad, opad, dgst);
453 tmps[gid].dgst[i + 0] = dgst[0];
454 tmps[gid].dgst[i + 1] = dgst[1];
455 tmps[gid].dgst[i + 2] = dgst[2];
456 tmps[gid].dgst[i + 3] = dgst[3];
457 tmps[gid].dgst[i + 4] = dgst[4];
458 tmps[gid].dgst[i + 5] = dgst[5];
459 tmps[gid].dgst[i + 6] = dgst[6];
460 tmps[gid].dgst[i + 7] = dgst[7];
462 tmps[gid].out[i + 0] = out[0];
463 tmps[gid].out[i + 1] = out[1];
464 tmps[gid].out[i + 2] = out[2];
465 tmps[gid].out[i + 3] = out[3];
466 tmps[gid].out[i + 4] = out[4];
467 tmps[gid].out[i + 5] = out[5];
468 tmps[gid].out[i + 6] = out[6];
469 tmps[gid].out[i + 7] = out[7];
473 __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)
479 const u32 gid = get_global_id (0);
481 if (gid >= gid_max) return;
483 const u32 lid = get_local_id (0);
485 const u64 a = tmps[gid].out[0];
486 const u64 b = tmps[gid].out[1];
488 const u32 r0 = l32_from_64 (a);
489 const u32 r1 = h32_from_64 (a);
490 const u32 r2 = l32_from_64 (b);
491 const u32 r3 = h32_from_64 (b);