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"
27 #include "include/rp_gpu.h"
31 #define VECT_COMPARE_S "check_single_vect1_comp4.c"
32 #define VECT_COMPARE_M "check_multi_vect1_comp4.c"
36 #define VECT_COMPARE_S "check_single_vect2_comp4.c"
37 #define VECT_COMPARE_M "check_multi_vect2_comp4.c"
41 #define VECT_COMPARE_S "check_single_vect4_comp4.c"
42 #define VECT_COMPARE_M "check_multi_vect4_comp4.c"
45 __device__ __constant__ gpu_rule_t c_rules[1024];
47 __device__ __constant__ u64 k_sha512[80] =
49 SHA512C00, SHA512C01, SHA512C02, SHA512C03,
50 SHA512C04, SHA512C05, SHA512C06, SHA512C07,
51 SHA512C08, SHA512C09, SHA512C0a, SHA512C0b,
52 SHA512C0c, SHA512C0d, SHA512C0e, SHA512C0f,
53 SHA512C10, SHA512C11, SHA512C12, SHA512C13,
54 SHA512C14, SHA512C15, SHA512C16, SHA512C17,
55 SHA512C18, SHA512C19, SHA512C1a, SHA512C1b,
56 SHA512C1c, SHA512C1d, SHA512C1e, SHA512C1f,
57 SHA512C20, SHA512C21, SHA512C22, SHA512C23,
58 SHA512C24, SHA512C25, SHA512C26, SHA512C27,
59 SHA512C28, SHA512C29, SHA512C2a, SHA512C2b,
60 SHA512C2c, SHA512C2d, SHA512C2e, SHA512C2f,
61 SHA512C30, SHA512C31, SHA512C32, SHA512C33,
62 SHA512C34, SHA512C35, SHA512C36, SHA512C37,
63 SHA512C38, SHA512C39, SHA512C3a, SHA512C3b,
64 SHA512C3c, SHA512C3d, SHA512C3e, SHA512C3f,
65 SHA512C40, SHA512C41, SHA512C42, SHA512C43,
66 SHA512C44, SHA512C45, SHA512C46, SHA512C47,
67 SHA512C48, SHA512C49, SHA512C4a, SHA512C4b,
68 SHA512C4c, SHA512C4d, SHA512C4e, SHA512C4f,
71 __device__ static void sha512_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 = SHA512_EXPAND (we_t, w9_t, w1_t, w0_t); \
102 w1_t = SHA512_EXPAND (wf_t, wa_t, w2_t, w1_t); \
103 w2_t = SHA512_EXPAND (w0_t, wb_t, w3_t, w2_t); \
104 w3_t = SHA512_EXPAND (w1_t, wc_t, w4_t, w3_t); \
105 w4_t = SHA512_EXPAND (w2_t, wd_t, w5_t, w4_t); \
106 w5_t = SHA512_EXPAND (w3_t, we_t, w6_t, w5_t); \
107 w6_t = SHA512_EXPAND (w4_t, wf_t, w7_t, w6_t); \
108 w7_t = SHA512_EXPAND (w5_t, w0_t, w8_t, w7_t); \
109 w8_t = SHA512_EXPAND (w6_t, w1_t, w9_t, w8_t); \
110 w9_t = SHA512_EXPAND (w7_t, w2_t, wa_t, w9_t); \
111 wa_t = SHA512_EXPAND (w8_t, w3_t, wb_t, wa_t); \
112 wb_t = SHA512_EXPAND (w9_t, w4_t, wc_t, wb_t); \
113 wc_t = SHA512_EXPAND (wa_t, w5_t, wd_t, wc_t); \
114 wd_t = SHA512_EXPAND (wb_t, w6_t, we_t, wd_t); \
115 we_t = SHA512_EXPAND (wc_t, w7_t, wf_t, we_t); \
116 wf_t = SHA512_EXPAND (wd_t, w8_t, w0_t, wf_t); \
119 #define ROUND_STEP(i) \
121 SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha512[i + 0]); \
122 SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha512[i + 1]); \
123 SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha512[i + 2]); \
124 SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha512[i + 3]); \
125 SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha512[i + 4]); \
126 SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha512[i + 5]); \
127 SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha512[i + 6]); \
128 SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha512[i + 7]); \
129 SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha512[i + 8]); \
130 SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha512[i + 9]); \
131 SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha512[i + 10]); \
132 SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha512[i + 11]); \
133 SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha512[i + 12]); \
134 SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha512[i + 13]); \
135 SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, we_t, k_sha512[i + 14]); \
136 SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha512[i + 15]); \
141 for (int i = 16; i < 80; i += 16)
143 ROUND_EXPAND (); ROUND_STEP (i);
167 extern "C" __global__ void __launch_bounds__ (256, 1) m01740_m04 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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 void *esalt_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)
173 const u32 lid = threadIdx.x;
179 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
181 if (gid >= gid_max) return;
185 pw_buf0[0] = pws[gid].i[ 0];
186 pw_buf0[1] = pws[gid].i[ 1];
187 pw_buf0[2] = pws[gid].i[ 2];
188 pw_buf0[3] = pws[gid].i[ 3];
192 pw_buf1[0] = pws[gid].i[ 4];
193 pw_buf1[1] = pws[gid].i[ 5];
194 pw_buf1[2] = pws[gid].i[ 6];
195 pw_buf1[3] = pws[gid].i[ 7];
197 const u32 pw_len = pws[gid].pw_len;
205 salt_buf0[0] = salt_bufs[salt_pos].salt_buf[0];
206 salt_buf0[1] = salt_bufs[salt_pos].salt_buf[1];
207 salt_buf0[2] = salt_bufs[salt_pos].salt_buf[2];
208 salt_buf0[3] = salt_bufs[salt_pos].salt_buf[3];
212 salt_buf1[0] = salt_bufs[salt_pos].salt_buf[4];
213 salt_buf1[1] = salt_bufs[salt_pos].salt_buf[5];
214 salt_buf1[2] = salt_bufs[salt_pos].salt_buf[6];
215 salt_buf1[3] = salt_bufs[salt_pos].salt_buf[7];
217 const u32 salt_len = salt_bufs[salt_pos].salt_len;
223 for (u32 il_pos = 0; il_pos < rules_cnt; il_pos++)
247 const u32 out_len = apply_rules (c_rules[il_pos].cmds, w0, w1, pw_len);
253 const u32 out_salt_len = (out_len * 2) + salt_len;
260 make_unicode (w0, w0_t, w1_t);
261 make_unicode (w1, w2_t, w3_t);
263 switch_buffer_by_offset (w0_t, w1_t, w2_t, w3_t, salt_len);
265 w0_t[0] |= salt_buf0[0];
266 w0_t[1] |= salt_buf0[1];
267 w0_t[2] |= salt_buf0[2];
268 w0_t[3] |= salt_buf0[3];
269 w1_t[0] |= salt_buf1[0];
270 w1_t[1] |= salt_buf1[1];
271 w1_t[2] |= salt_buf1[2];
272 w1_t[3] |= salt_buf1[3];
274 append_0x80_4 (w0_t, w1_t, w2_t, w3_t, out_salt_len);
280 w0_t[0] = swap_workaround (w0_t[0]);
281 w0_t[1] = swap_workaround (w0_t[1]);
282 w0_t[2] = swap_workaround (w0_t[2]);
283 w0_t[3] = swap_workaround (w0_t[3]);
284 w1_t[0] = swap_workaround (w1_t[0]);
285 w1_t[1] = swap_workaround (w1_t[1]);
286 w1_t[2] = swap_workaround (w1_t[2]);
287 w1_t[3] = swap_workaround (w1_t[3]);
288 w2_t[0] = swap_workaround (w2_t[0]);
289 w2_t[1] = swap_workaround (w2_t[1]);
290 w2_t[2] = swap_workaround (w2_t[2]);
291 w2_t[3] = swap_workaround (w2_t[3]);
292 w3_t[0] = swap_workaround (w3_t[0]);
293 w3_t[1] = swap_workaround (w3_t[1]);
295 w3_t[3] = out_salt_len * 8;
299 digest[0] = SHA512M_A;
300 digest[1] = SHA512M_B;
301 digest[2] = SHA512M_C;
302 digest[3] = SHA512M_D;
303 digest[4] = SHA512M_E;
304 digest[5] = SHA512M_F;
305 digest[6] = SHA512M_G;
306 digest[7] = SHA512M_H;
308 sha512_transform (w0_t, w1_t, w2_t, w3_t, digest);
311 const u32x r0 = l32_from_64 (digest[7]);
312 const u32x r1 = h32_from_64 (digest[7]);
313 const u32x r2 = l32_from_64 (digest[3]);
314 const u32x r3 = h32_from_64 (digest[3]);
316 #include VECT_COMPARE_M
320 extern "C" __global__ void __launch_bounds__ (256, 1) m01740_m08 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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 void *esalt_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)
324 extern "C" __global__ void __launch_bounds__ (256, 1) m01740_m16 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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 void *esalt_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)
328 extern "C" __global__ void __launch_bounds__ (256, 1) m01740_s04 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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 void *esalt_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)
334 const u32 lid = threadIdx.x;
340 const u32 gid = (blockIdx.x * blockDim.x) + threadIdx.x;
342 if (gid >= gid_max) return;
346 pw_buf0[0] = pws[gid].i[ 0];
347 pw_buf0[1] = pws[gid].i[ 1];
348 pw_buf0[2] = pws[gid].i[ 2];
349 pw_buf0[3] = pws[gid].i[ 3];
353 pw_buf1[0] = pws[gid].i[ 4];
354 pw_buf1[1] = pws[gid].i[ 5];
355 pw_buf1[2] = pws[gid].i[ 6];
356 pw_buf1[3] = pws[gid].i[ 7];
358 const u32 pw_len = pws[gid].pw_len;
366 salt_buf0[0] = salt_bufs[salt_pos].salt_buf[0];
367 salt_buf0[1] = salt_bufs[salt_pos].salt_buf[1];
368 salt_buf0[2] = salt_bufs[salt_pos].salt_buf[2];
369 salt_buf0[3] = salt_bufs[salt_pos].salt_buf[3];
373 salt_buf1[0] = salt_bufs[salt_pos].salt_buf[4];
374 salt_buf1[1] = salt_bufs[salt_pos].salt_buf[5];
375 salt_buf1[2] = salt_bufs[salt_pos].salt_buf[6];
376 salt_buf1[3] = salt_bufs[salt_pos].salt_buf[7];
378 const u32 salt_len = salt_bufs[salt_pos].salt_len;
384 const u32 search[4] =
386 digests_buf[digests_offset].digest_buf[DGST_R0],
387 digests_buf[digests_offset].digest_buf[DGST_R1],
388 digests_buf[digests_offset].digest_buf[DGST_R2],
389 digests_buf[digests_offset].digest_buf[DGST_R3]
396 for (u32 il_pos = 0; il_pos < rules_cnt; il_pos++)
420 const u32 out_len = apply_rules (c_rules[il_pos].cmds, w0, w1, pw_len);
426 const u32 out_salt_len = (out_len * 2) + salt_len;
433 make_unicode (w0, w0_t, w1_t);
434 make_unicode (w1, w2_t, w3_t);
436 switch_buffer_by_offset (w0_t, w1_t, w2_t, w3_t, salt_len);
438 w0_t[0] |= salt_buf0[0];
439 w0_t[1] |= salt_buf0[1];
440 w0_t[2] |= salt_buf0[2];
441 w0_t[3] |= salt_buf0[3];
442 w1_t[0] |= salt_buf1[0];
443 w1_t[1] |= salt_buf1[1];
444 w1_t[2] |= salt_buf1[2];
445 w1_t[3] |= salt_buf1[3];
447 append_0x80_4 (w0_t, w1_t, w2_t, w3_t, out_salt_len);
453 w0_t[0] = swap_workaround (w0_t[0]);
454 w0_t[1] = swap_workaround (w0_t[1]);
455 w0_t[2] = swap_workaround (w0_t[2]);
456 w0_t[3] = swap_workaround (w0_t[3]);
457 w1_t[0] = swap_workaround (w1_t[0]);
458 w1_t[1] = swap_workaround (w1_t[1]);
459 w1_t[2] = swap_workaround (w1_t[2]);
460 w1_t[3] = swap_workaround (w1_t[3]);
461 w2_t[0] = swap_workaround (w2_t[0]);
462 w2_t[1] = swap_workaround (w2_t[1]);
463 w2_t[2] = swap_workaround (w2_t[2]);
464 w2_t[3] = swap_workaround (w2_t[3]);
465 w3_t[0] = swap_workaround (w3_t[0]);
466 w3_t[1] = swap_workaround (w3_t[1]);
468 w3_t[3] = out_salt_len * 8;
472 digest[0] = SHA512M_A;
473 digest[1] = SHA512M_B;
474 digest[2] = SHA512M_C;
475 digest[3] = SHA512M_D;
476 digest[4] = SHA512M_E;
477 digest[5] = SHA512M_F;
478 digest[6] = SHA512M_G;
479 digest[7] = SHA512M_H;
481 sha512_transform (w0_t, w1_t, w2_t, w3_t, digest);
484 const u32x r0 = l32_from_64 (digest[7]);
485 const u32x r1 = h32_from_64 (digest[7]);
486 const u32x r2 = l32_from_64 (digest[3]);
487 const u32x r3 = h32_from_64 (digest[3]);
489 #include VECT_COMPARE_S
493 extern "C" __global__ void __launch_bounds__ (256, 1) m01740_s08 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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 void *esalt_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)
497 extern "C" __global__ void __launch_bounds__ (256, 1) m01740_s16 (const pw_t *pws, const gpu_rule_t *rules_buf, const comb_t *combs_buf, const bf_t *bfs_buf, const void *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 void *esalt_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)