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"
29 #define VECT_COMPARE_M "check_multi_vect1_comp4.c"
33 #define VECT_COMPARE_M "check_multi_vect2_comp4.c"
36 __device__ __constant__ u32 padding[8] =
56 __device__ static void swap (RC4_KEY *rc4_key, const u8 i, const u8 j)
61 rc4_key->S[i] = rc4_key->S[j];
65 __device__ static void rc4_init_16 (RC4_KEY *rc4_key, const u32 data[4])
70 u32 *ptr = (u32 *) rc4_key->S;
73 for (u32 i = 0; i < 64; i++)
80 for (u32 i = 0; i < 16; i++)
88 j += rc4_key->S[idx] + (v >> 0); swap (rc4_key, idx, j); idx++;
89 j += rc4_key->S[idx] + (v >> 8); swap (rc4_key, idx, j); idx++;
90 j += rc4_key->S[idx] + (v >> 16); swap (rc4_key, idx, j); idx++;
91 j += rc4_key->S[idx] + (v >> 24); swap (rc4_key, idx, j); idx++;
95 j += rc4_key->S[idx] + (v >> 0); swap (rc4_key, idx, j); idx++;
96 j += rc4_key->S[idx] + (v >> 8); swap (rc4_key, idx, j); idx++;
97 j += rc4_key->S[idx] + (v >> 16); swap (rc4_key, idx, j); idx++;
98 j += rc4_key->S[idx] + (v >> 24); swap (rc4_key, idx, j); idx++;
102 j += rc4_key->S[idx] + (v >> 0); swap (rc4_key, idx, j); idx++;
103 j += rc4_key->S[idx] + (v >> 8); swap (rc4_key, idx, j); idx++;
104 j += rc4_key->S[idx] + (v >> 16); swap (rc4_key, idx, j); idx++;
105 j += rc4_key->S[idx] + (v >> 24); swap (rc4_key, idx, j); idx++;
109 j += rc4_key->S[idx] + (v >> 0); swap (rc4_key, idx, j); idx++;
110 j += rc4_key->S[idx] + (v >> 8); swap (rc4_key, idx, j); idx++;
111 j += rc4_key->S[idx] + (v >> 16); swap (rc4_key, idx, j); idx++;
112 j += rc4_key->S[idx] + (v >> 24); swap (rc4_key, idx, j); idx++;
116 __device__ static u8 rc4_next_16 (RC4_KEY *rc4_key, u8 i, u8 j, const u32 in[4], u32 out[4])
118 for (u32 k = 0; k < 4; k++)
127 swap (rc4_key, i, j);
129 idx = rc4_key->S[i] + rc4_key->S[j];
131 xor4 |= rc4_key->S[idx] << 0;
136 swap (rc4_key, i, j);
138 idx = rc4_key->S[i] + rc4_key->S[j];
140 xor4 |= rc4_key->S[idx] << 8;
145 swap (rc4_key, i, j);
147 idx = rc4_key->S[i] + rc4_key->S[j];
149 xor4 |= rc4_key->S[idx] << 16;
154 swap (rc4_key, i, j);
156 idx = rc4_key->S[i] + rc4_key->S[j];
158 xor4 |= rc4_key->S[idx] << 24;
160 out[k] = in[k] ^ xor4;
166 __device__ static void md5_transform (const u32x w0[4], const u32x w1[4], const u32x w2[4], const u32x w3[4], u32x digest[4])
192 MD5_STEP (MD5_Fo, a, b, c, d, w0_t, MD5C00, MD5S00);
193 MD5_STEP (MD5_Fo, d, a, b, c, w1_t, MD5C01, MD5S01);
194 MD5_STEP (MD5_Fo, c, d, a, b, w2_t, MD5C02, MD5S02);
195 MD5_STEP (MD5_Fo, b, c, d, a, w3_t, MD5C03, MD5S03);
196 MD5_STEP (MD5_Fo, a, b, c, d, w4_t, MD5C04, MD5S00);
197 MD5_STEP (MD5_Fo, d, a, b, c, w5_t, MD5C05, MD5S01);
198 MD5_STEP (MD5_Fo, c, d, a, b, w6_t, MD5C06, MD5S02);
199 MD5_STEP (MD5_Fo, b, c, d, a, w7_t, MD5C07, MD5S03);
200 MD5_STEP (MD5_Fo, a, b, c, d, w8_t, MD5C08, MD5S00);
201 MD5_STEP (MD5_Fo, d, a, b, c, w9_t, MD5C09, MD5S01);
202 MD5_STEP (MD5_Fo, c, d, a, b, wa_t, MD5C0a, MD5S02);
203 MD5_STEP (MD5_Fo, b, c, d, a, wb_t, MD5C0b, MD5S03);
204 MD5_STEP (MD5_Fo, a, b, c, d, wc_t, MD5C0c, MD5S00);
205 MD5_STEP (MD5_Fo, d, a, b, c, wd_t, MD5C0d, MD5S01);
206 MD5_STEP (MD5_Fo, c, d, a, b, we_t, MD5C0e, MD5S02);
207 MD5_STEP (MD5_Fo, b, c, d, a, wf_t, MD5C0f, MD5S03);
209 MD5_STEP (MD5_Go, a, b, c, d, w1_t, MD5C10, MD5S10);
210 MD5_STEP (MD5_Go, d, a, b, c, w6_t, MD5C11, MD5S11);
211 MD5_STEP (MD5_Go, c, d, a, b, wb_t, MD5C12, MD5S12);
212 MD5_STEP (MD5_Go, b, c, d, a, w0_t, MD5C13, MD5S13);
213 MD5_STEP (MD5_Go, a, b, c, d, w5_t, MD5C14, MD5S10);
214 MD5_STEP (MD5_Go, d, a, b, c, wa_t, MD5C15, MD5S11);
215 MD5_STEP (MD5_Go, c, d, a, b, wf_t, MD5C16, MD5S12);
216 MD5_STEP (MD5_Go, b, c, d, a, w4_t, MD5C17, MD5S13);
217 MD5_STEP (MD5_Go, a, b, c, d, w9_t, MD5C18, MD5S10);
218 MD5_STEP (MD5_Go, d, a, b, c, we_t, MD5C19, MD5S11);
219 MD5_STEP (MD5_Go, c, d, a, b, w3_t, MD5C1a, MD5S12);
220 MD5_STEP (MD5_Go, b, c, d, a, w8_t, MD5C1b, MD5S13);
221 MD5_STEP (MD5_Go, a, b, c, d, wd_t, MD5C1c, MD5S10);
222 MD5_STEP (MD5_Go, d, a, b, c, w2_t, MD5C1d, MD5S11);
223 MD5_STEP (MD5_Go, c, d, a, b, w7_t, MD5C1e, MD5S12);
224 MD5_STEP (MD5_Go, b, c, d, a, wc_t, MD5C1f, MD5S13);
226 MD5_STEP (MD5_H1, a, b, c, d, w5_t, MD5C20, MD5S20);
227 MD5_STEP (MD5_H2, d, a, b, c, w8_t, MD5C21, MD5S21);
228 MD5_STEP (MD5_H1, c, d, a, b, wb_t, MD5C22, MD5S22);
229 MD5_STEP (MD5_H2, b, c, d, a, we_t, MD5C23, MD5S23);
230 MD5_STEP (MD5_H1, a, b, c, d, w1_t, MD5C24, MD5S20);
231 MD5_STEP (MD5_H2, d, a, b, c, w4_t, MD5C25, MD5S21);
232 MD5_STEP (MD5_H1, c, d, a, b, w7_t, MD5C26, MD5S22);
233 MD5_STEP (MD5_H2, b, c, d, a, wa_t, MD5C27, MD5S23);
234 MD5_STEP (MD5_H1, a, b, c, d, wd_t, MD5C28, MD5S20);
235 MD5_STEP (MD5_H2, d, a, b, c, w0_t, MD5C29, MD5S21);
236 MD5_STEP (MD5_H1, c, d, a, b, w3_t, MD5C2a, MD5S22);
237 MD5_STEP (MD5_H2, b, c, d, a, w6_t, MD5C2b, MD5S23);
238 MD5_STEP (MD5_H1, a, b, c, d, w9_t, MD5C2c, MD5S20);
239 MD5_STEP (MD5_H2, d, a, b, c, wc_t, MD5C2d, MD5S21);
240 MD5_STEP (MD5_H1, c, d, a, b, wf_t, MD5C2e, MD5S22);
241 MD5_STEP (MD5_H2, b, c, d, a, w2_t, MD5C2f, MD5S23);
243 MD5_STEP (MD5_I , a, b, c, d, w0_t, MD5C30, MD5S30);
244 MD5_STEP (MD5_I , d, a, b, c, w7_t, MD5C31, MD5S31);
245 MD5_STEP (MD5_I , c, d, a, b, we_t, MD5C32, MD5S32);
246 MD5_STEP (MD5_I , b, c, d, a, w5_t, MD5C33, MD5S33);
247 MD5_STEP (MD5_I , a, b, c, d, wc_t, MD5C34, MD5S30);
248 MD5_STEP (MD5_I , d, a, b, c, w3_t, MD5C35, MD5S31);
249 MD5_STEP (MD5_I , c, d, a, b, wa_t, MD5C36, MD5S32);
250 MD5_STEP (MD5_I , b, c, d, a, w1_t, MD5C37, MD5S33);
251 MD5_STEP (MD5_I , a, b, c, d, w8_t, MD5C38, MD5S30);
252 MD5_STEP (MD5_I , d, a, b, c, wf_t, MD5C39, MD5S31);
253 MD5_STEP (MD5_I , c, d, a, b, w6_t, MD5C3a, MD5S32);
254 MD5_STEP (MD5_I , b, c, d, a, wd_t, MD5C3b, MD5S33);
255 MD5_STEP (MD5_I , a, b, c, d, w4_t, MD5C3c, MD5S30);
256 MD5_STEP (MD5_I , d, a, b, c, wb_t, MD5C3d, MD5S31);
257 MD5_STEP (MD5_I , c, d, a, b, w2_t, MD5C3e, MD5S32);
258 MD5_STEP (MD5_I , b, c, d, a, w9_t, MD5C3f, MD5S33);
266 extern "C" __global__ void __launch_bounds__ (64, 1) m10500_init (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, pdf14_tmp_t *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 pdf_t *pdf_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)
272 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
274 if (gid >= gid_max) return;
278 w0[0] = pws[gid].i[ 0];
279 w0[1] = pws[gid].i[ 1];
280 w0[2] = pws[gid].i[ 2];
281 w0[3] = pws[gid].i[ 3];
285 w1[0] = pws[gid].i[ 4];
286 w1[1] = pws[gid].i[ 5];
287 w1[2] = pws[gid].i[ 6];
288 w1[3] = pws[gid].i[ 7];
292 w2[0] = pws[gid].i[ 8];
293 w2[1] = pws[gid].i[ 9];
297 const u32 pw_len = pws[gid].pw_len;
305 o_buf[0] = pdf_bufs[salt_pos].o_buf[0];
306 o_buf[1] = pdf_bufs[salt_pos].o_buf[1];
307 o_buf[2] = pdf_bufs[salt_pos].o_buf[2];
308 o_buf[3] = pdf_bufs[salt_pos].o_buf[3];
309 o_buf[4] = pdf_bufs[salt_pos].o_buf[4];
310 o_buf[5] = pdf_bufs[salt_pos].o_buf[5];
311 o_buf[6] = pdf_bufs[salt_pos].o_buf[6];
312 o_buf[7] = pdf_bufs[salt_pos].o_buf[7];
314 u32 P = pdf_bufs[salt_pos].P;
318 id_buf[ 0] = pdf_bufs[salt_pos].id_buf[0];
319 id_buf[ 1] = pdf_bufs[salt_pos].id_buf[1];
320 id_buf[ 2] = pdf_bufs[salt_pos].id_buf[2];
321 id_buf[ 3] = pdf_bufs[salt_pos].id_buf[3];
323 id_buf[ 4] = pdf_bufs[salt_pos].id_buf[4];
324 id_buf[ 5] = pdf_bufs[salt_pos].id_buf[5];
325 id_buf[ 6] = pdf_bufs[salt_pos].id_buf[6];
326 id_buf[ 7] = pdf_bufs[salt_pos].id_buf[7];
333 u32 id_len = pdf_bufs[salt_pos].id_len;
334 u32 id_len4 = id_len / 4;
338 rc4data[0] = pdf_bufs[salt_pos].rc4data[0];
339 rc4data[1] = pdf_bufs[salt_pos].rc4data[1];
341 u32 final_length = 68 + id_len;
346 if (pdf_bufs[salt_pos].enc_md != 1)
354 id_buf[id_len4 + 0] = w11;
355 id_buf[id_len4 + 1] = w12;
366 // max length supported by pdf11 is 32
368 w0_t[0] = padding[0];
369 w0_t[1] = padding[1];
370 w0_t[2] = padding[2];
371 w0_t[3] = padding[3];
372 w1_t[0] = padding[4];
373 w1_t[1] = padding[5];
374 w1_t[2] = padding[6];
375 w1_t[3] = padding[7];
385 switch_buffer_by_offset (w0_t, w1_t, w2_t, w3_t, pw_len);
388 // truncate at 32 is wanted, not a bug!
415 md5_transform (w0_t, w1_t, w2_t, w3_t, digest);
418 w0_t[1] = id_buf[ 0];
419 w0_t[2] = id_buf[ 1];
420 w0_t[3] = id_buf[ 2];
421 w1_t[0] = id_buf[ 3];
422 w1_t[1] = id_buf[ 4];
423 w1_t[2] = id_buf[ 5];
424 w1_t[3] = id_buf[ 6];
425 w2_t[0] = id_buf[ 7];
426 w2_t[1] = id_buf[ 8];
427 w2_t[2] = id_buf[ 9];
428 w2_t[3] = id_buf[10];
429 w3_t[0] = id_buf[11];
431 w3_t[2] = final_length * 8;
434 md5_transform (w0_t, w1_t, w2_t, w3_t, digest);
436 tmps[gid].digest[0] = digest[0];
437 tmps[gid].digest[1] = digest[1];
438 tmps[gid].digest[2] = digest[2];
439 tmps[gid].digest[3] = digest[3];
441 tmps[gid].out[0] = rc4data[0];
442 tmps[gid].out[1] = rc4data[1];
443 tmps[gid].out[2] = 0;
444 tmps[gid].out[3] = 0;
447 extern "C" __global__ void __launch_bounds__ (64, 1) m10500_loop (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, pdf14_tmp_t *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 pdf_t *pdf_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)
453 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
454 const u32 lid = threadIdx.x;
456 if (gid >= gid_max) return;
462 __shared__ RC4_KEY rc4_keys[64];
464 RC4_KEY *rc4_key = &rc4_keys[lid];
472 digest[0] = tmps[gid].digest[0];
473 digest[1] = tmps[gid].digest[1];
474 digest[2] = tmps[gid].digest[2];
475 digest[3] = tmps[gid].digest[3];
479 out[0] = tmps[gid].out[0];
480 out[1] = tmps[gid].out[1];
481 out[2] = tmps[gid].out[2];
482 out[3] = tmps[gid].out[3];
484 for (u32 i = 0, j = loop_pos; i < loop_cnt; i++, j++)
515 md5_transform (w0_t, w1_t, w2_t, w3_t, digest);
519 const u32 x = j - 50;
521 const u32 xv = x << 0
528 tmp[0] = digest[0] ^ xv;
529 tmp[1] = digest[1] ^ xv;
530 tmp[2] = digest[2] ^ xv;
531 tmp[3] = digest[3] ^ xv;
533 rc4_init_16 (rc4_key, tmp);
535 rc4_next_16 (rc4_key, 0, 0, out, out);
539 tmps[gid].digest[0] = digest[0];
540 tmps[gid].digest[1] = digest[1];
541 tmps[gid].digest[2] = digest[2];
542 tmps[gid].digest[3] = digest[3];
544 tmps[gid].out[0] = out[0];
545 tmps[gid].out[1] = out[1];
546 tmps[gid].out[2] = out[2];
547 tmps[gid].out[3] = out[3];
550 extern "C" __global__ void __launch_bounds__ (64, 1) m10500_comp (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, pdf14_tmp_t *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 pdf_t *pdf_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)
556 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
558 if (gid >= gid_max) return;
560 const u32 lid = threadIdx.x;
566 const u32x r0 = tmps[gid].out[0];
567 const u32x r1 = tmps[gid].out[1];
573 #include VECT_COMPARE_M