mirror of
https://github.com/hashcat/hashcat.git
synced 2024-11-23 00:28:11 +00:00
282 lines
7.0 KiB
Common Lisp
282 lines
7.0 KiB
Common Lisp
/**
|
|
* Author......: See docs/credits.txt
|
|
* License.....: MIT
|
|
*/
|
|
|
|
//#define NEW_SIMD_CODE
|
|
|
|
#ifdef KERNEL_STATIC
|
|
#include M2S(INCLUDE_PATH/inc_vendor.h)
|
|
#include M2S(INCLUDE_PATH/inc_types.h)
|
|
#include M2S(INCLUDE_PATH/inc_platform.cl)
|
|
#include M2S(INCLUDE_PATH/inc_common.cl)
|
|
#include M2S(INCLUDE_PATH/inc_scalar.cl)
|
|
#include M2S(INCLUDE_PATH/inc_hash_md5.cl)
|
|
#endif
|
|
|
|
#if VECT_SIZE == 1
|
|
#define uint_to_hex_upper8(i) make_u32x (u_bin2asc[(i)])
|
|
#elif VECT_SIZE == 2
|
|
#define uint_to_hex_upper8(i) make_u32x (u_bin2asc[(i).s0], u_bin2asc[(i).s1])
|
|
#elif VECT_SIZE == 4
|
|
#define uint_to_hex_upper8(i) make_u32x (u_bin2asc[(i).s0], u_bin2asc[(i).s1], u_bin2asc[(i).s2], u_bin2asc[(i).s3])
|
|
#elif VECT_SIZE == 8
|
|
#define uint_to_hex_upper8(i) make_u32x (u_bin2asc[(i).s0], u_bin2asc[(i).s1], u_bin2asc[(i).s2], u_bin2asc[(i).s3], u_bin2asc[(i).s4], u_bin2asc[(i).s5], u_bin2asc[(i).s6], u_bin2asc[(i).s7])
|
|
#elif VECT_SIZE == 16
|
|
#define uint_to_hex_upper8(i) make_u32x (u_bin2asc[(i).s0], u_bin2asc[(i).s1], u_bin2asc[(i).s2], u_bin2asc[(i).s3], u_bin2asc[(i).s4], u_bin2asc[(i).s5], u_bin2asc[(i).s6], u_bin2asc[(i).s7], u_bin2asc[(i).s8], u_bin2asc[(i).s9], u_bin2asc[(i).sa], u_bin2asc[(i).sb], u_bin2asc[(i).sc], u_bin2asc[(i).sd], u_bin2asc[(i).se], u_bin2asc[(i).sf])
|
|
#endif
|
|
|
|
typedef struct md5_double_salt
|
|
{
|
|
u32 salt1_buf[64];
|
|
int salt1_len;
|
|
|
|
u32 salt2_buf[64];
|
|
int salt2_len;
|
|
|
|
} md5_double_salt_t;
|
|
|
|
KERNEL_FQ void m03730_mxx (KERN_ATTR_ESALT (md5_double_salt_t))
|
|
{
|
|
/**
|
|
* modifier
|
|
*/
|
|
|
|
const u64 gid = get_global_id (0);
|
|
const u64 lid = get_local_id (0);
|
|
const u64 lsz = get_local_size (0);
|
|
|
|
/**
|
|
* bin2asc uppercase array
|
|
*/
|
|
|
|
LOCAL_VK u32 u_bin2asc[256];
|
|
|
|
for (u32 j = lid; j < 256; j += lsz)
|
|
{
|
|
const u32 i0 = (j >> 0) & 15;
|
|
const u32 i1 = (j >> 4) & 15;
|
|
|
|
u_bin2asc[j] = ((i0 < 10) ? '0' + i0 : 'A' - 10 + i0) << 8
|
|
| ((i1 < 10) ? '0' + i1 : 'A' - 10 + i1) << 0;
|
|
}
|
|
|
|
SYNC_THREADS ();
|
|
|
|
if (gid >= GID_CNT) return;
|
|
|
|
/**
|
|
* base
|
|
*/
|
|
|
|
const u32 salt_len = esalt_bufs[DIGESTS_OFFSET_HOST].salt1_len;
|
|
|
|
u32 s[64] = { 0 };
|
|
|
|
for (u32 i = 0, idx = 0; i < salt_len; i += 4, idx += 1)
|
|
{
|
|
s[idx] = esalt_bufs[DIGESTS_OFFSET_HOST].salt1_buf[idx];
|
|
}
|
|
|
|
md5_ctx_t ctx0;
|
|
|
|
md5_init (&ctx0);
|
|
|
|
md5_update_global (&ctx0, esalt_bufs[DIGESTS_OFFSET_HOST].salt2_buf, esalt_bufs[DIGESTS_OFFSET_HOST].salt2_len);
|
|
|
|
md5_update_global (&ctx0, pws[gid].i, pws[gid].pw_len);
|
|
|
|
/**
|
|
* loop
|
|
*/
|
|
|
|
for (u32 il_pos = 0; il_pos < IL_CNT; il_pos++)
|
|
{
|
|
md5_ctx_t ctx1 = ctx0;
|
|
|
|
md5_update_global (&ctx1, combs_buf[il_pos].i, combs_buf[il_pos].pw_len);
|
|
|
|
md5_final (&ctx1);
|
|
|
|
const u32 a = ctx1.h[0];
|
|
const u32 b = ctx1.h[1];
|
|
const u32 c = ctx1.h[2];
|
|
const u32 d = ctx1.h[3];
|
|
|
|
u32 w0[4];
|
|
u32 w1[4];
|
|
u32 w2[4];
|
|
u32 w3[4];
|
|
|
|
w0[0] = uint_to_hex_upper8 ((a >> 0) & 255) << 0
|
|
| uint_to_hex_upper8 ((a >> 8) & 255) << 16;
|
|
w0[1] = uint_to_hex_upper8 ((a >> 16) & 255) << 0
|
|
| uint_to_hex_upper8 ((a >> 24) & 255) << 16;
|
|
w0[2] = uint_to_hex_upper8 ((b >> 0) & 255) << 0
|
|
| uint_to_hex_upper8 ((b >> 8) & 255) << 16;
|
|
w0[3] = uint_to_hex_upper8 ((b >> 16) & 255) << 0
|
|
| uint_to_hex_upper8 ((b >> 24) & 255) << 16;
|
|
w1[0] = uint_to_hex_upper8 ((c >> 0) & 255) << 0
|
|
| uint_to_hex_upper8 ((c >> 8) & 255) << 16;
|
|
w1[1] = uint_to_hex_upper8 ((c >> 16) & 255) << 0
|
|
| uint_to_hex_upper8 ((c >> 24) & 255) << 16;
|
|
w1[2] = uint_to_hex_upper8 ((d >> 0) & 255) << 0
|
|
| uint_to_hex_upper8 ((d >> 8) & 255) << 16;
|
|
w1[3] = uint_to_hex_upper8 ((d >> 16) & 255) << 0
|
|
| uint_to_hex_upper8 ((d >> 24) & 255) << 16;
|
|
w2[0] = 0;
|
|
w2[1] = 0;
|
|
w2[2] = 0;
|
|
w2[3] = 0;
|
|
w3[0] = 0;
|
|
w3[1] = 0;
|
|
w3[2] = 0;
|
|
w3[3] = 0;
|
|
|
|
md5_ctx_t ctx;
|
|
|
|
md5_init (&ctx);
|
|
|
|
md5_update (&ctx, s, salt_len);
|
|
|
|
md5_update_64 (&ctx, w0, w1, w2, w3, 32);
|
|
|
|
md5_final (&ctx);
|
|
|
|
const u32 r0 = ctx.h[DGST_R0];
|
|
const u32 r1 = ctx.h[DGST_R1];
|
|
const u32 r2 = ctx.h[DGST_R2];
|
|
const u32 r3 = ctx.h[DGST_R3];
|
|
|
|
COMPARE_M_SCALAR (r0, r1, r2, r3);
|
|
}
|
|
}
|
|
|
|
KERNEL_FQ void m03730_sxx (KERN_ATTR_ESALT (md5_double_salt_t))
|
|
{
|
|
/**
|
|
* modifier
|
|
*/
|
|
|
|
const u64 gid = get_global_id (0);
|
|
const u64 lid = get_local_id (0);
|
|
const u64 lsz = get_local_size (0);
|
|
|
|
/**
|
|
* bin2asc uppercase array
|
|
*/
|
|
|
|
LOCAL_VK u32 u_bin2asc[256];
|
|
|
|
for (u32 j = lid; j < 256; j += lsz)
|
|
{
|
|
const u32 i0 = (j >> 0) & 15;
|
|
const u32 i1 = (j >> 4) & 15;
|
|
|
|
u_bin2asc[j] = ((i0 < 10) ? '0' + i0 : 'A' - 10 + i0) << 8
|
|
| ((i1 < 10) ? '0' + i1 : 'A' - 10 + i1) << 0;
|
|
}
|
|
|
|
SYNC_THREADS ();
|
|
|
|
if (gid >= GID_CNT) return;
|
|
|
|
/**
|
|
* digest
|
|
*/
|
|
|
|
const u32 search[4] =
|
|
{
|
|
digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R0],
|
|
digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R1],
|
|
digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R2],
|
|
digests_buf[DIGESTS_OFFSET_HOST].digest_buf[DGST_R3]
|
|
};
|
|
|
|
const u32 salt_len = esalt_bufs[DIGESTS_OFFSET_HOST].salt1_len;
|
|
|
|
u32 s[64] = { 0 };
|
|
|
|
for (u32 i = 0, idx = 0; i < salt_len; i += 4, idx += 1)
|
|
{
|
|
s[idx] = esalt_bufs[DIGESTS_OFFSET_HOST].salt1_buf[idx];
|
|
}
|
|
|
|
/**
|
|
* base
|
|
*/
|
|
|
|
md5_ctx_t ctx0;
|
|
|
|
md5_init (&ctx0);
|
|
|
|
md5_update_global (&ctx0, esalt_bufs[DIGESTS_OFFSET_HOST].salt2_buf, esalt_bufs[DIGESTS_OFFSET_HOST].salt2_len);
|
|
|
|
md5_update_global (&ctx0, pws[gid].i, pws[gid].pw_len);
|
|
|
|
/**
|
|
* loop
|
|
*/
|
|
|
|
for (u32 il_pos = 0; il_pos < IL_CNT; il_pos++)
|
|
{
|
|
md5_ctx_t ctx1 = ctx0;
|
|
|
|
md5_update_global (&ctx1, combs_buf[il_pos].i, combs_buf[il_pos].pw_len);
|
|
|
|
md5_final (&ctx1);
|
|
|
|
const u32 a = ctx1.h[0];
|
|
const u32 b = ctx1.h[1];
|
|
const u32 c = ctx1.h[2];
|
|
const u32 d = ctx1.h[3];
|
|
|
|
u32 w0[4];
|
|
u32 w1[4];
|
|
u32 w2[4];
|
|
u32 w3[4];
|
|
|
|
w0[0] = uint_to_hex_upper8 ((a >> 0) & 255) << 0
|
|
| uint_to_hex_upper8 ((a >> 8) & 255) << 16;
|
|
w0[1] = uint_to_hex_upper8 ((a >> 16) & 255) << 0
|
|
| uint_to_hex_upper8 ((a >> 24) & 255) << 16;
|
|
w0[2] = uint_to_hex_upper8 ((b >> 0) & 255) << 0
|
|
| uint_to_hex_upper8 ((b >> 8) & 255) << 16;
|
|
w0[3] = uint_to_hex_upper8 ((b >> 16) & 255) << 0
|
|
| uint_to_hex_upper8 ((b >> 24) & 255) << 16;
|
|
w1[0] = uint_to_hex_upper8 ((c >> 0) & 255) << 0
|
|
| uint_to_hex_upper8 ((c >> 8) & 255) << 16;
|
|
w1[1] = uint_to_hex_upper8 ((c >> 16) & 255) << 0
|
|
| uint_to_hex_upper8 ((c >> 24) & 255) << 16;
|
|
w1[2] = uint_to_hex_upper8 ((d >> 0) & 255) << 0
|
|
| uint_to_hex_upper8 ((d >> 8) & 255) << 16;
|
|
w1[3] = uint_to_hex_upper8 ((d >> 16) & 255) << 0
|
|
| uint_to_hex_upper8 ((d >> 24) & 255) << 16;
|
|
w2[0] = 0;
|
|
w2[1] = 0;
|
|
w2[2] = 0;
|
|
w2[3] = 0;
|
|
w3[0] = 0;
|
|
w3[1] = 0;
|
|
w3[2] = 0;
|
|
w3[3] = 0;
|
|
|
|
md5_ctx_t ctx;
|
|
|
|
md5_init (&ctx);
|
|
|
|
md5_update (&ctx, s, salt_len);
|
|
|
|
md5_update_64 (&ctx, w0, w1, w2, w3, 32);
|
|
|
|
md5_final (&ctx);
|
|
|
|
const u32 r0 = ctx.h[DGST_R0];
|
|
const u32 r1 = ctx.h[DGST_R1];
|
|
const u32 r2 = ctx.h[DGST_R2];
|
|
const u32 r3 = ctx.h[DGST_R3];
|
|
|
|
COMPARE_S_SCALAR (r0, r1, r2, r3);
|
|
}
|
|
}
|