/**
- * Author......: Jens Steube <jens.steube@gmail.com>
+ * Authors.....: Jens Steube <jens.steube@gmail.com>
+ * Gabriele Gristina <matrix@hashcat.net>
+ *
* License.....: MIT
*/
#define _GOST_
+#define NEW_SIMD_CODE
+
#include "include/constants.h"
#include "include/kernel_vendor.h"
#define DGST_R3 3
#include "include/kernel_functions.c"
-#include "types_ocl.c"
-#include "common.c"
-
-#define COMPARE_S "check_single_comp4.c"
-#define COMPARE_M "check_multi_comp4.c"
+#include "OpenCL/types_ocl.c"
+#include "OpenCL/common.c"
+#include "OpenCL/simd.c"
__constant u32 c_tables[4][256] =
{
}
};
+#if VECT_SIZE == 1
#define BOX(i,n,S) (S)[(n)][(i)]
-
-#define round(k1,k2,tbl) \
+#elif VECT_SIZE == 2
+#define BOX(i,n,S) (u32x) ((S)[(n)][(i).s0], (S)[(n)][(i).s1])
+#elif VECT_SIZE == 4
+#define BOX(i,n,S) (u32x) ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3])
+#elif VECT_SIZE == 8
+#define BOX(i,n,S) (u32x) ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7])
+#elif VECT_SIZE == 16
+#define BOX(i,n,S) (u32x) ((S)[(n)][(i).s0], (S)[(n)][(i).s1], (S)[(n)][(i).s2], (S)[(n)][(i).s3], (S)[(n)][(i).s4], (S)[(n)][(i).s5], (S)[(n)][(i).s6], (S)[(n)][(i).s7], (S)[(n)][(i).s8], (S)[(n)][(i).s9], (S)[(n)][(i).sa], (S)[(n)][(i).sb], (S)[(n)][(i).sc], (S)[(n)][(i).sd], (S)[(n)][(i).se], (S)[(n)][(i).sf])
+#endif
+
+#define _round(k1,k2,tbl) \
{ \
- u32 t; \
+ u32x t; \
t = (k1) + r; \
- l ^= BOX (amd_bfe (t, 0, 8), 0, tbl) ^ \
- BOX (amd_bfe (t, 8, 8), 1, tbl) ^ \
- BOX (amd_bfe (t, 16, 8), 2, tbl) ^ \
- BOX (amd_bfe (t, 24, 8), 3, tbl); \
+ l ^= BOX (((t >> 0) & 0xff), 0, tbl) ^ \
+ BOX (((t >> 8) & 0xff), 1, tbl) ^ \
+ BOX (((t >> 16) & 0xff), 2, tbl) ^ \
+ BOX (((t >> 24) & 0xff), 3, tbl); \
t = (k2) + l; \
- r ^= BOX (amd_bfe (t, 0, 8), 0, tbl) ^ \
- BOX (amd_bfe (t, 8, 8), 1, tbl) ^ \
- BOX (amd_bfe (t, 16, 8), 2, tbl) ^ \
- BOX (amd_bfe (t, 24, 8), 3, tbl); \
+ r ^= BOX (((t >> 0) & 0xff), 0, tbl) ^ \
+ BOX (((t >> 8) & 0xff), 1, tbl) ^ \
+ BOX (((t >> 16) & 0xff), 2, tbl) ^ \
+ BOX (((t >> 24) & 0xff), 3, tbl); \
}
#define R(k,h,s,i,t) \
{ \
- u32 r; \
- u32 l; \
+ u32x r; \
+ u32x l; \
r = h[i + 0]; \
l = h[i + 1]; \
- round (k[0], k[1], t); \
- round (k[2], k[3], t); \
- round (k[4], k[5], t); \
- round (k[6], k[7], t); \
- round (k[0], k[1], t); \
- round (k[2], k[3], t); \
- round (k[4], k[5], t); \
- round (k[6], k[7], t); \
- round (k[0], k[1], t); \
- round (k[2], k[3], t); \
- round (k[4], k[5], t); \
- round (k[6], k[7], t); \
- round (k[7], k[6], t); \
- round (k[5], k[4], t); \
- round (k[3], k[2], t); \
- round (k[1], k[0], t); \
+ _round (k[0], k[1], t); \
+ _round (k[2], k[3], t); \
+ _round (k[4], k[5], t); \
+ _round (k[6], k[7], t); \
+ _round (k[0], k[1], t); \
+ _round (k[2], k[3], t); \
+ _round (k[4], k[5], t); \
+ _round (k[6], k[7], t); \
+ _round (k[0], k[1], t); \
+ _round (k[2], k[3], t); \
+ _round (k[4], k[5], t); \
+ _round (k[6], k[7], t); \
+ _round (k[7], k[6], t); \
+ _round (k[5], k[4], t); \
+ _round (k[3], k[2], t); \
+ _round (k[1], k[0], t); \
s[i + 0] = l; \
s[i + 1] = r; \
}
#define A(x) \
{ \
- u32 l; \
- u32 r; \
+ u32x l; \
+ u32x r; \
l = x[0] ^ x[2]; \
r = x[1] ^ x[3]; \
x[0] = x[2]; \
#define AA(x) \
{ \
- u32 l; \
- u32 r; \
+ u32x l; \
+ u32x r; \
l = x[0]; \
r = x[2]; \
x[0] = x[4]; \
#define PASS0(h,s,u,v,t) \
{ \
- u32 k[8]; \
- u32 w[8]; \
+ u32x k[8]; \
+ u32x w[8]; \
X (w, u, v); \
P (k, w); \
R (k, h, s, 0, t); \
#define PASS2(h,s,u,v,t) \
{ \
- u32 k[8]; \
- u32 w[8]; \
+ u32x k[8]; \
+ u32x w[8]; \
X (w, u, v); \
P (k, w); \
R (k, h, s, 2, t); \
#define PASS4(h,s,u,v,t) \
{ \
- u32 k[8]; \
- u32 w[8]; \
+ u32x k[8]; \
+ u32x w[8]; \
X (w, u, v); \
P (k, w); \
R (k, h, s, 4, t); \
#define PASS6(h,s,u,v,t) \
{ \
- u32 k[8]; \
- u32 w[8]; \
+ u32x k[8]; \
+ u32x w[8]; \
X (w, u, v); \
P (k, w); \
R (k, h, s, 6, t); \
}
-__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06900_m04 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
+__kernel void m06900_m04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
- /**
- * modifier
- */
-
- const u32 lid = get_local_id (0);
-
/**
* base
*/
const u32 gid = get_global_id (0);
-
- u32 wordl0[4];
-
- wordl0[0] = pws[gid].i[ 0];
- wordl0[1] = pws[gid].i[ 1];
- wordl0[2] = pws[gid].i[ 2];
- wordl0[3] = pws[gid].i[ 3];
-
- u32 wordl1[4];
-
- wordl1[0] = pws[gid].i[ 4];
- wordl1[1] = pws[gid].i[ 5];
- wordl1[2] = pws[gid].i[ 6];
- wordl1[3] = pws[gid].i[ 7];
-
- u32 wordl2[4];
-
- wordl2[0] = 0;
- wordl2[1] = 0;
- wordl2[2] = 0;
- wordl2[3] = 0;
-
- u32 wordl3[4];
-
- wordl3[0] = 0;
- wordl3[1] = 0;
- wordl3[2] = 0;
- wordl3[3] = 0;
-
- const u32 pw_l_len = pws[gid].pw_len;
-
- if (combs_mode == COMBINATOR_MODE_BASE_RIGHT)
- {
- switch_buffer_by_offset (wordl0, wordl1, wordl2, wordl3, combs_buf[0].pw_len);
- }
+ const u32 lid = get_local_id (0);
+ const u32 lsz = get_local_size (0);
/**
* sbox
__local u32 s_tables[4][256];
- const u32 lid4 = lid * 4;
-
- s_tables[0][lid4 + 0] = c_tables[0][lid4 + 0];
- s_tables[0][lid4 + 1] = c_tables[0][lid4 + 1];
- s_tables[0][lid4 + 2] = c_tables[0][lid4 + 2];
- s_tables[0][lid4 + 3] = c_tables[0][lid4 + 3];
-
- s_tables[1][lid4 + 0] = c_tables[1][lid4 + 0];
- s_tables[1][lid4 + 1] = c_tables[1][lid4 + 1];
- s_tables[1][lid4 + 2] = c_tables[1][lid4 + 2];
- s_tables[1][lid4 + 3] = c_tables[1][lid4 + 3];
-
- s_tables[2][lid4 + 0] = c_tables[2][lid4 + 0];
- s_tables[2][lid4 + 1] = c_tables[2][lid4 + 1];
- s_tables[2][lid4 + 2] = c_tables[2][lid4 + 2];
- s_tables[2][lid4 + 3] = c_tables[2][lid4 + 3];
-
- s_tables[3][lid4 + 0] = c_tables[3][lid4 + 0];
- s_tables[3][lid4 + 1] = c_tables[3][lid4 + 1];
- s_tables[3][lid4 + 2] = c_tables[3][lid4 + 2];
- s_tables[3][lid4 + 3] = c_tables[3][lid4 + 3];
+ for (u32 i = lid; i < 256; i += lsz)
+ {
+ s_tables[0][i] = c_tables[0][i];
+ s_tables[1][i] = c_tables[1][i];
+ s_tables[2][i] = c_tables[2][i];
+ s_tables[3][i] = c_tables[3][i];
+ }
barrier (CLK_LOCAL_MEM_FENCE);
if (gid >= gid_max) return;
/**
- * loop
+ * base
*/
- for (u32 il_pos = 0; il_pos < combs_cnt; il_pos++)
- {
- const u32 pw_r_len = combs_buf[il_pos].pw_len;
-
- const u32 pw_len = pw_l_len + pw_r_len;
-
- u32 wordr0[4];
-
- wordr0[0] = combs_buf[il_pos].i[0];
- wordr0[1] = combs_buf[il_pos].i[1];
- wordr0[2] = combs_buf[il_pos].i[2];
- wordr0[3] = combs_buf[il_pos].i[3];
+ u32 pw_buf0[4];
+ u32 pw_buf1[4];
- u32 wordr1[4];
+ pw_buf0[0] = pws[gid].i[0];
+ pw_buf0[1] = pws[gid].i[1];
+ pw_buf0[2] = pws[gid].i[2];
+ pw_buf0[3] = pws[gid].i[3];
+ pw_buf1[0] = pws[gid].i[4];
+ pw_buf1[1] = pws[gid].i[5];
+ pw_buf1[2] = pws[gid].i[6];
+ pw_buf1[3] = pws[gid].i[7];
- wordr1[0] = combs_buf[il_pos].i[4];
- wordr1[1] = combs_buf[il_pos].i[5];
- wordr1[2] = combs_buf[il_pos].i[6];
- wordr1[3] = combs_buf[il_pos].i[7];
-
- u32 wordr2[4];
-
- wordr2[0] = 0;
- wordr2[1] = 0;
- wordr2[2] = 0;
- wordr2[3] = 0;
+ const u32 pw_l_len = pws[gid].pw_len;
- u32 wordr3[4];
+ /**
+ * loop
+ */
- wordr3[0] = 0;
- wordr3[1] = 0;
- wordr3[2] = 0;
- wordr3[3] = 0;
+ for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
+ {
+ const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos);
+
+ const u32x pw_len = pw_l_len + pw_r_len;
+
+ /**
+ * concat password candidate
+ */
+
+ u32x wordl0[4] = { 0 };
+ u32x wordl1[4] = { 0 };
+ u32x wordl2[4] = { 0 };
+ u32x wordl3[4] = { 0 };
+
+ wordl0[0] = pw_buf0[0];
+ wordl0[1] = pw_buf0[1];
+ wordl0[2] = pw_buf0[2];
+ wordl0[3] = pw_buf0[3];
+ wordl1[0] = pw_buf1[0];
+ wordl1[1] = pw_buf1[1];
+ wordl1[2] = pw_buf1[2];
+ wordl1[3] = pw_buf1[3];
+
+ u32x wordr0[4] = { 0 };
+ u32x wordr1[4] = { 0 };
+ u32x wordr2[4] = { 0 };
+ u32x wordr3[4] = { 0 };
+
+ wordr0[0] = ix_create_combt (combs_buf, il_pos, 0);
+ wordr0[1] = ix_create_combt (combs_buf, il_pos, 1);
+ wordr0[2] = ix_create_combt (combs_buf, il_pos, 2);
+ wordr0[3] = ix_create_combt (combs_buf, il_pos, 3);
+ wordr1[0] = ix_create_combt (combs_buf, il_pos, 4);
+ wordr1[1] = ix_create_combt (combs_buf, il_pos, 5);
+ wordr1[2] = ix_create_combt (combs_buf, il_pos, 6);
+ wordr1[3] = ix_create_combt (combs_buf, il_pos, 7);
if (combs_mode == COMBINATOR_MODE_BASE_LEFT)
{
- switch_buffer_by_offset (wordr0, wordr1, wordr2, wordr3, pw_l_len);
+ switch_buffer_by_offset_le_VV (wordr0, wordr1, wordr2, wordr3, pw_l_len);
+ }
+ else
+ {
+ switch_buffer_by_offset_le_VV (wordl0, wordl1, wordl2, wordl3, pw_r_len);
}
- u32 w0[4];
+ u32x w0[4];
+ u32x w1[4];
w0[0] = wordl0[0] | wordr0[0];
w0[1] = wordl0[1] | wordr0[1];
w0[2] = wordl0[2] | wordr0[2];
w0[3] = wordl0[3] | wordr0[3];
-
- u32 w1[4];
-
w1[0] = wordl1[0] | wordr1[0];
w1[1] = wordl1[1] | wordr1[1];
w1[2] = wordl1[2] | wordr1[2];
w1[3] = wordl1[3] | wordr1[3];
- u32 w2[4];
+ /**
+ * GOST
+ */
- w2[0] = wordl2[0] | wordr2[0];
- w2[1] = wordl2[1] | wordr2[1];
- w2[2] = wordl2[2] | wordr2[2];
- w2[3] = wordl2[3] | wordr2[3];
-
- u32 w3[4];
-
- w3[0] = wordl3[0] | wordr3[0];
- w3[1] = wordl3[1] | wordr3[1];
- w3[2] = pw_len * 8;
- w3[3] = 0;
-
- const u32 w14 = pw_len * 8;
-
- u32 data[8];
+ u32x data[8];
data[0] = w0[0];
data[1] = w0[1];
data[6] = w1[2];
data[7] = w1[3];
- u32 state[16];
+ u32x state[16];
state[ 0] = 0;
state[ 1] = 0;
state[14] = data[6];
state[15] = data[7];
- u32 state_m[8];
- u32 data_m[8];
+ u32x state_m[8];
+ u32x data_m[8];
/* gost1 */
data_m[6] = data[6];
data_m[7] = data[7];
- u32 tmp[8];
-
- PASS0 (state, tmp, state_m, data_m, s_tables);
- PASS2 (state, tmp, state_m, data_m, s_tables);
- PASS4 (state, tmp, state_m, data_m, s_tables);
- PASS6 (state, tmp, state_m, data_m, s_tables);
+ u32x tmp[8];
- SHIFT12 (state_m, data, tmp);
- SHIFT16 (state, data_m, state_m);
- SHIFT61 (state, data_m);
+ //if (pw_len > 0) // not really SIMD compatible
+ {
+ PASS0 (state, tmp, state_m, data_m, s_tables);
+ PASS2 (state, tmp, state_m, data_m, s_tables);
+ PASS4 (state, tmp, state_m, data_m, s_tables);
+ PASS6 (state, tmp, state_m, data_m, s_tables);
+
+ SHIFT12 (state_m, data, tmp);
+ SHIFT16 (state, data_m, state_m);
+ SHIFT61 (state, data_m);
+ }
- data[0] = w14;
+ data[0] = pw_len * 8;
data[1] = 0;
data[2] = 0;
data[3] = 0;
/* store */
- const u32 r0 = state[0];
- const u32 r1 = state[1];
- const u32 r2 = state[2];
- const u32 r3 = state[3];
-
- #include COMPARE_M
+ COMPARE_M_SIMD (state[0], state[1], state[2], state[3]);
}
}
-__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06900_m08 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
+__kernel void m06900_m08 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
}
-__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06900_m16 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
+__kernel void m06900_m16 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
}
-__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06900_s04 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
+__kernel void m06900_s04 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
- /**
- * modifier
- */
-
- const u32 lid = get_local_id (0);
-
/**
* base
*/
const u32 gid = get_global_id (0);
-
- u32 wordl0[4];
-
- wordl0[0] = pws[gid].i[ 0];
- wordl0[1] = pws[gid].i[ 1];
- wordl0[2] = pws[gid].i[ 2];
- wordl0[3] = pws[gid].i[ 3];
-
- u32 wordl1[4];
-
- wordl1[0] = pws[gid].i[ 4];
- wordl1[1] = pws[gid].i[ 5];
- wordl1[2] = pws[gid].i[ 6];
- wordl1[3] = pws[gid].i[ 7];
-
- u32 wordl2[4];
-
- wordl2[0] = 0;
- wordl2[1] = 0;
- wordl2[2] = 0;
- wordl2[3] = 0;
-
- u32 wordl3[4];
-
- wordl3[0] = 0;
- wordl3[1] = 0;
- wordl3[2] = 0;
- wordl3[3] = 0;
-
- const u32 pw_l_len = pws[gid].pw_len;
-
- if (combs_mode == COMBINATOR_MODE_BASE_RIGHT)
- {
- switch_buffer_by_offset (wordl0, wordl1, wordl2, wordl3, combs_buf[0].pw_len);
- }
+ const u32 lid = get_local_id (0);
+ const u32 lsz = get_local_size (0);
/**
* sbox
__local u32 s_tables[4][256];
- const u32 lid4 = lid * 4;
+ for (u32 i = lid; i < 256; i += lsz)
+ {
+ s_tables[0][i] = c_tables[0][i];
+ s_tables[1][i] = c_tables[1][i];
+ s_tables[2][i] = c_tables[2][i];
+ s_tables[3][i] = c_tables[3][i];
+ }
- s_tables[0][lid4 + 0] = c_tables[0][lid4 + 0];
- s_tables[0][lid4 + 1] = c_tables[0][lid4 + 1];
- s_tables[0][lid4 + 2] = c_tables[0][lid4 + 2];
- s_tables[0][lid4 + 3] = c_tables[0][lid4 + 3];
+ barrier (CLK_LOCAL_MEM_FENCE);
- s_tables[1][lid4 + 0] = c_tables[1][lid4 + 0];
- s_tables[1][lid4 + 1] = c_tables[1][lid4 + 1];
- s_tables[1][lid4 + 2] = c_tables[1][lid4 + 2];
- s_tables[1][lid4 + 3] = c_tables[1][lid4 + 3];
+ if (gid >= gid_max) return;
- s_tables[2][lid4 + 0] = c_tables[2][lid4 + 0];
- s_tables[2][lid4 + 1] = c_tables[2][lid4 + 1];
- s_tables[2][lid4 + 2] = c_tables[2][lid4 + 2];
- s_tables[2][lid4 + 3] = c_tables[2][lid4 + 3];
+ /**
+ * base
+ */
- s_tables[3][lid4 + 0] = c_tables[3][lid4 + 0];
- s_tables[3][lid4 + 1] = c_tables[3][lid4 + 1];
- s_tables[3][lid4 + 2] = c_tables[3][lid4 + 2];
- s_tables[3][lid4 + 3] = c_tables[3][lid4 + 3];
+ u32 pw_buf0[4];
+ u32 pw_buf1[4];
- barrier (CLK_LOCAL_MEM_FENCE);
+ pw_buf0[0] = pws[gid].i[0];
+ pw_buf0[1] = pws[gid].i[1];
+ pw_buf0[2] = pws[gid].i[2];
+ pw_buf0[3] = pws[gid].i[3];
+ pw_buf1[0] = pws[gid].i[4];
+ pw_buf1[1] = pws[gid].i[5];
+ pw_buf1[2] = pws[gid].i[6];
+ pw_buf1[3] = pws[gid].i[7];
- if (gid >= gid_max) return;
+ const u32 pw_l_len = pws[gid].pw_len;
/**
* digest
* loop
*/
- for (u32 il_pos = 0; il_pos < combs_cnt; il_pos++)
+ for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
{
- const u32 pw_r_len = combs_buf[il_pos].pw_len;
-
- const u32 pw_len = pw_l_len + pw_r_len;
-
- u32 wordr0[4];
-
- wordr0[0] = combs_buf[il_pos].i[0];
- wordr0[1] = combs_buf[il_pos].i[1];
- wordr0[2] = combs_buf[il_pos].i[2];
- wordr0[3] = combs_buf[il_pos].i[3];
-
- u32 wordr1[4];
-
- wordr1[0] = combs_buf[il_pos].i[4];
- wordr1[1] = combs_buf[il_pos].i[5];
- wordr1[2] = combs_buf[il_pos].i[6];
- wordr1[3] = combs_buf[il_pos].i[7];
-
- u32 wordr2[4];
-
- wordr2[0] = 0;
- wordr2[1] = 0;
- wordr2[2] = 0;
- wordr2[3] = 0;
-
- u32 wordr3[4];
-
- wordr3[0] = 0;
- wordr3[1] = 0;
- wordr3[2] = 0;
- wordr3[3] = 0;
+ const u32x pw_r_len = pwlenx_create_combt (combs_buf, il_pos);
+
+ const u32x pw_len = pw_l_len + pw_r_len;
+
+ /**
+ * concat password candidate
+ */
+
+ u32x wordl0[4] = { 0 };
+ u32x wordl1[4] = { 0 };
+ u32x wordl2[4] = { 0 };
+ u32x wordl3[4] = { 0 };
+
+ wordl0[0] = pw_buf0[0];
+ wordl0[1] = pw_buf0[1];
+ wordl0[2] = pw_buf0[2];
+ wordl0[3] = pw_buf0[3];
+ wordl1[0] = pw_buf1[0];
+ wordl1[1] = pw_buf1[1];
+ wordl1[2] = pw_buf1[2];
+ wordl1[3] = pw_buf1[3];
+
+ u32x wordr0[4] = { 0 };
+ u32x wordr1[4] = { 0 };
+ u32x wordr2[4] = { 0 };
+ u32x wordr3[4] = { 0 };
+
+ wordr0[0] = ix_create_combt (combs_buf, il_pos, 0);
+ wordr0[1] = ix_create_combt (combs_buf, il_pos, 1);
+ wordr0[2] = ix_create_combt (combs_buf, il_pos, 2);
+ wordr0[3] = ix_create_combt (combs_buf, il_pos, 3);
+ wordr1[0] = ix_create_combt (combs_buf, il_pos, 4);
+ wordr1[1] = ix_create_combt (combs_buf, il_pos, 5);
+ wordr1[2] = ix_create_combt (combs_buf, il_pos, 6);
+ wordr1[3] = ix_create_combt (combs_buf, il_pos, 7);
if (combs_mode == COMBINATOR_MODE_BASE_LEFT)
{
- switch_buffer_by_offset (wordr0, wordr1, wordr2, wordr3, pw_l_len);
+ switch_buffer_by_offset_le_VV (wordr0, wordr1, wordr2, wordr3, pw_l_len);
+ }
+ else
+ {
+ switch_buffer_by_offset_le_VV (wordl0, wordl1, wordl2, wordl3, pw_r_len);
}
- u32 w0[4];
+ u32x w0[4];
+ u32x w1[4];
w0[0] = wordl0[0] | wordr0[0];
w0[1] = wordl0[1] | wordr0[1];
w0[2] = wordl0[2] | wordr0[2];
w0[3] = wordl0[3] | wordr0[3];
-
- u32 w1[4];
-
w1[0] = wordl1[0] | wordr1[0];
w1[1] = wordl1[1] | wordr1[1];
w1[2] = wordl1[2] | wordr1[2];
w1[3] = wordl1[3] | wordr1[3];
- u32 w2[4];
-
- w2[0] = wordl2[0] | wordr2[0];
- w2[1] = wordl2[1] | wordr2[1];
- w2[2] = wordl2[2] | wordr2[2];
- w2[3] = wordl2[3] | wordr2[3];
+ /**
+ * GOST
+ */
- u32 w3[4];
-
- w3[0] = wordl3[0] | wordr3[0];
- w3[1] = wordl3[1] | wordr3[1];
- w3[2] = pw_len * 8;
- w3[3] = 0;
-
- const u32 w14 = pw_len * 8;
-
- u32 data[8];
+ u32x data[8];
data[0] = w0[0];
data[1] = w0[1];
data[6] = w1[2];
data[7] = w1[3];
- u32 state[16];
+ u32x state[16];
state[ 0] = 0;
state[ 1] = 0;
state[14] = data[6];
state[15] = data[7];
- u32 state_m[8];
- u32 data_m[8];
+ u32x state_m[8];
+ u32x data_m[8];
/* gost1 */
data_m[6] = data[6];
data_m[7] = data[7];
- u32 tmp[8];
-
- PASS0 (state, tmp, state_m, data_m, s_tables);
- PASS2 (state, tmp, state_m, data_m, s_tables);
- PASS4 (state, tmp, state_m, data_m, s_tables);
- PASS6 (state, tmp, state_m, data_m, s_tables);
+ u32x tmp[8];
- SHIFT12 (state_m, data, tmp);
- SHIFT16 (state, data_m, state_m);
- SHIFT61 (state, data_m);
+ //if (pw_len > 0) // not really SIMD compatible
+ {
+ PASS0 (state, tmp, state_m, data_m, s_tables);
+ PASS2 (state, tmp, state_m, data_m, s_tables);
+ PASS4 (state, tmp, state_m, data_m, s_tables);
+ PASS6 (state, tmp, state_m, data_m, s_tables);
+
+ SHIFT12 (state_m, data, tmp);
+ SHIFT16 (state, data_m, state_m);
+ SHIFT61 (state, data_m);
+ }
- data[0] = w14;
+ data[0] = pw_len * 8;
data[1] = 0;
data[2] = 0;
data[3] = 0;
/* store */
- const u32 r0 = state[0];
- const u32 r1 = state[1];
- const u32 r2 = state[2];
- const u32 r3 = state[3];
-
- #include COMPARE_S
+ COMPARE_S_SIMD (state[0], state[1], state[2], state[3]);
}
}
-__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06900_s08 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
+__kernel void m06900_s08 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
}
-__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m06900_s16 (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 combs_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
+__kernel void m06900_s16 (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global void *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 void *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 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max)
{
}