1
0
mirror of https://github.com/hashcat/hashcat.git synced 2024-11-01 04:58:57 +00:00
hashcat/OpenCL/m10410_a1-optimized.cl
R. Yushaev 5de004103a Replace kernel parameter lists with macros
Substitute long parameter lists in ~2900 kernel function declarations
with macros. This cleans up the code, reduces probability of copy-paste
errors and highlights the differences between kernel functions. Also
reduces the size of the OpenCL folder by ~3 MB.
2018-11-16 11:44:33 +01:00

432 lines
9.0 KiB
Common Lisp

/**
* Author......: See docs/credits.txt
* License.....: MIT
*/
//too much register pressure
//#define NEW_SIMD_CODE
#include "inc_vendor.cl"
#include "inc_hash_constants.h"
#include "inc_hash_functions.cl"
#include "inc_types.cl"
#include "inc_common.cl"
#include "inc_simd.cl"
#include "inc_hash_md5.cl"
__constant u32a padding[8] =
{
0x5e4ebf28,
0x418a754e,
0x564e0064,
0x0801faff,
0xb6002e2e,
0x803e68d0,
0xfea90c2f,
0x7a695364
};
typedef struct
{
u8 S[256];
u32 wtf_its_faster;
} RC4_KEY;
DECLSPEC void swap (__local RC4_KEY *rc4_key, const u8 i, const u8 j)
{
u8 tmp;
tmp = rc4_key->S[i];
rc4_key->S[i] = rc4_key->S[j];
rc4_key->S[j] = tmp;
}
DECLSPEC void rc4_init_16 (__local RC4_KEY *rc4_key, const u32 *data)
{
u32 v = 0x03020100;
u32 a = 0x04040404;
__local u32 *ptr = (__local u32 *) rc4_key->S;
#ifdef _unroll
#pragma unroll
#endif
for (u32 i = 0; i < 64; i++)
{
ptr[i] = v; v += a;
}
const u32 d0 = data[0] >> 0;
const u32 d1 = data[0] >> 8;
const u32 d2 = data[0] >> 16;
const u32 d3 = data[0] >> 24;
const u32 d4 = data[1] >> 0;
u32 j = 0;
#ifdef _unroll
#pragma unroll
#endif
for (u32 i = 0; i < 255; i += 5)
{
j += rc4_key->S[i + 0] + d0; swap (rc4_key, i + 0, j);
j += rc4_key->S[i + 1] + d1; swap (rc4_key, i + 1, j);
j += rc4_key->S[i + 2] + d2; swap (rc4_key, i + 2, j);
j += rc4_key->S[i + 3] + d3; swap (rc4_key, i + 3, j);
j += rc4_key->S[i + 4] + d4; swap (rc4_key, i + 4, j);
}
j += rc4_key->S[255] + d0; swap (rc4_key, 255, j);
}
DECLSPEC u8 rc4_next_16 (__local RC4_KEY *rc4_key, u8 i, u8 j, __constant u32 *in, u32 *out)
{
#ifdef _unroll
#pragma unroll
#endif
for (u32 k = 0; k < 4; k++)
{
u32 xor4 = 0;
u8 idx;
i += 1;
j += rc4_key->S[i];
swap (rc4_key, i, j);
idx = rc4_key->S[i] + rc4_key->S[j];
xor4 |= rc4_key->S[idx] << 0;
i += 1;
j += rc4_key->S[i];
swap (rc4_key, i, j);
idx = rc4_key->S[i] + rc4_key->S[j];
xor4 |= rc4_key->S[idx] << 8;
i += 1;
j += rc4_key->S[i];
swap (rc4_key, i, j);
idx = rc4_key->S[i] + rc4_key->S[j];
xor4 |= rc4_key->S[idx] << 16;
i += 1;
j += rc4_key->S[i];
swap (rc4_key, i, j);
idx = rc4_key->S[i] + rc4_key->S[j];
xor4 |= rc4_key->S[idx] << 24;
out[k] = in[k] ^ xor4;
}
return j;
}
__kernel void __attribute__((reqd_work_group_size(64, 1, 1))) m10410_m04 (KERN_ATTR_ESALT (pdf_t))
{
/**
* modifier
*/
const u64 lid = get_local_id (0);
/**
* base
*/
const u64 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 pw_buf0[4];
u32 pw_buf1[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];
const u32 pw_l_len = pws[gid].pw_len;
/**
* shared
*/
__local RC4_KEY rc4_keys[64];
__local RC4_KEY *rc4_key = &rc4_keys[lid];
/**
* U_buf
*/
u32 o_buf[8];
o_buf[0] = esalt_bufs[digests_offset].o_buf[0];
o_buf[1] = esalt_bufs[digests_offset].o_buf[1];
o_buf[2] = esalt_bufs[digests_offset].o_buf[2];
o_buf[3] = esalt_bufs[digests_offset].o_buf[3];
o_buf[4] = esalt_bufs[digests_offset].o_buf[4];
o_buf[5] = esalt_bufs[digests_offset].o_buf[5];
o_buf[6] = esalt_bufs[digests_offset].o_buf[6];
o_buf[7] = esalt_bufs[digests_offset].o_buf[7];
u32 P = esalt_bufs[digests_offset].P;
u32 id_buf[4];
id_buf[0] = esalt_bufs[digests_offset].id_buf[0];
id_buf[1] = esalt_bufs[digests_offset].id_buf[1];
id_buf[2] = esalt_bufs[digests_offset].id_buf[2];
id_buf[3] = esalt_bufs[digests_offset].id_buf[3];
/**
* loop
*/
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_le_VV (wordr0, wordr1, wordr2, wordr3, pw_l_len);
}
else
{
switch_buffer_by_offset_le_VV (wordl0, wordl1, wordl2, wordl3, pw_r_len);
}
u32x w0[4];
w0[0] = wordl0[0] | wordr0[0];
w0[1] = wordl0[1] | wordr0[1];
w0[2] = 0;
w0[3] = 0;
/**
* pdf
*/
rc4_init_16 (rc4_key, w0);
u32 out[4];
rc4_next_16 (rc4_key, 0, 0, padding, out);
COMPARE_M_SIMD (out[0], out[1], out[2], out[3]);
}
}
__kernel void __attribute__((reqd_work_group_size(64, 1, 1))) m10410_m08 (KERN_ATTR_ESALT (pdf_t))
{
}
__kernel void __attribute__((reqd_work_group_size(64, 1, 1))) m10410_m16 (KERN_ATTR_ESALT (pdf_t))
{
}
__kernel void __attribute__((reqd_work_group_size(64, 1, 1))) m10410_s04 (KERN_ATTR_ESALT (pdf_t))
{
/**
* modifier
*/
const u64 lid = get_local_id (0);
/**
* base
*/
const u64 gid = get_global_id (0);
if (gid >= gid_max) return;
u32 pw_buf0[4];
u32 pw_buf1[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];
const u32 pw_l_len = pws[gid].pw_len;
/**
* shared
*/
__local RC4_KEY rc4_keys[64];
__local RC4_KEY *rc4_key = &rc4_keys[lid];
/**
* U_buf
*/
u32 o_buf[8];
o_buf[0] = esalt_bufs[digests_offset].o_buf[0];
o_buf[1] = esalt_bufs[digests_offset].o_buf[1];
o_buf[2] = esalt_bufs[digests_offset].o_buf[2];
o_buf[3] = esalt_bufs[digests_offset].o_buf[3];
o_buf[4] = esalt_bufs[digests_offset].o_buf[4];
o_buf[5] = esalt_bufs[digests_offset].o_buf[5];
o_buf[6] = esalt_bufs[digests_offset].o_buf[6];
o_buf[7] = esalt_bufs[digests_offset].o_buf[7];
u32 P = esalt_bufs[digests_offset].P;
u32 id_buf[4];
id_buf[0] = esalt_bufs[digests_offset].id_buf[0];
id_buf[1] = esalt_bufs[digests_offset].id_buf[1];
id_buf[2] = esalt_bufs[digests_offset].id_buf[2];
id_buf[3] = esalt_bufs[digests_offset].id_buf[3];
/**
* digest
*/
const u32 search[4] =
{
digests_buf[digests_offset].digest_buf[DGST_R0],
digests_buf[digests_offset].digest_buf[DGST_R1],
digests_buf[digests_offset].digest_buf[DGST_R2],
digests_buf[digests_offset].digest_buf[DGST_R3]
};
/**
* loop
*/
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_le_VV (wordr0, wordr1, wordr2, wordr3, pw_l_len);
}
else
{
switch_buffer_by_offset_le_VV (wordl0, wordl1, wordl2, wordl3, pw_r_len);
}
u32x w0[4];
w0[0] = wordl0[0] | wordr0[0];
w0[1] = wordl0[1] | wordr0[1];
w0[2] = 0;
w0[3] = 0;
/**
* pdf
*/
rc4_init_16 (rc4_key, w0);
u32 out[4];
rc4_next_16 (rc4_key, 0, 0, padding, out);
COMPARE_S_SIMD (out[0], out[1], out[2], out[3]);
}
}
__kernel void __attribute__((reqd_work_group_size(64, 1, 1))) m10410_s08 (KERN_ATTR_ESALT (pdf_t))
{
}
__kernel void __attribute__((reqd_work_group_size(64, 1, 1))) m10410_s16 (KERN_ATTR_ESALT (pdf_t))
{
}