1
mirror of https://github.com/hashcat/hashcat synced 2024-11-20 23:27:31 +01:00
hashcat/OpenCL/m11100_a0-pure.cl

334 lines
8.4 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_rp.h)
#include M2S(INCLUDE_PATH/inc_rp.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_lower8(i) make_u32x (l_bin2asc[(i)])
#elif VECT_SIZE == 2
#define uint_to_hex_lower8(i) make_u32x (l_bin2asc[(i).s0], l_bin2asc[(i).s1])
#elif VECT_SIZE == 4
#define uint_to_hex_lower8(i) make_u32x (l_bin2asc[(i).s0], l_bin2asc[(i).s1], l_bin2asc[(i).s2], l_bin2asc[(i).s3])
#elif VECT_SIZE == 8
#define uint_to_hex_lower8(i) make_u32x (l_bin2asc[(i).s0], l_bin2asc[(i).s1], l_bin2asc[(i).s2], l_bin2asc[(i).s3], l_bin2asc[(i).s4], l_bin2asc[(i).s5], l_bin2asc[(i).s6], l_bin2asc[(i).s7])
#elif VECT_SIZE == 16
#define uint_to_hex_lower8(i) make_u32x (l_bin2asc[(i).s0], l_bin2asc[(i).s1], l_bin2asc[(i).s2], l_bin2asc[(i).s3], l_bin2asc[(i).s4], l_bin2asc[(i).s5], l_bin2asc[(i).s6], l_bin2asc[(i).s7], l_bin2asc[(i).s8], l_bin2asc[(i).s9], l_bin2asc[(i).sa], l_bin2asc[(i).sb], l_bin2asc[(i).sc], l_bin2asc[(i).sd], l_bin2asc[(i).se], l_bin2asc[(i).sf])
#endif
KERNEL_FQ void m11100_mxx (KERN_ATTR_RULES ())
{
/**
* modifier
*/
const u64 gid = get_global_id (0);
const u64 lid = get_local_id (0);
const u64 lsz = get_local_size (0);
/**
* bin2asc table
*/
LOCAL_VK u32 l_bin2asc[256];
for (u32 i = lid; i < 256; i += lsz)
{
const u32 i0 = (i >> 0) & 15;
const u32 i1 = (i >> 4) & 15;
l_bin2asc[i] = ((i0 < 10) ? '0' + i0 : 'a' - 10 + i0) << 8
| ((i1 < 10) ? '0' + i1 : 'a' - 10 + i1) << 0;
}
SYNC_THREADS ();
if (gid >= GID_CNT) return;
/**
* salt
*/
u32 challenge;
challenge = salt_bufs[SALT_POS_HOST].salt_buf[0];
u32 salt_buf0[4];
u32 salt_buf1[4];
salt_buf0[0] = salt_bufs[SALT_POS_HOST].salt_buf[1]; // not a bug, see challenge
salt_buf0[1] = salt_bufs[SALT_POS_HOST].salt_buf[2];
salt_buf0[2] = salt_bufs[SALT_POS_HOST].salt_buf[3];
salt_buf0[3] = salt_bufs[SALT_POS_HOST].salt_buf[4];
salt_buf1[0] = salt_bufs[SALT_POS_HOST].salt_buf[5];
salt_buf1[1] = salt_bufs[SALT_POS_HOST].salt_buf[6];
salt_buf1[2] = salt_bufs[SALT_POS_HOST].salt_buf[7];
salt_buf1[3] = salt_bufs[SALT_POS_HOST].salt_buf[8];
const u32 salt_len = salt_bufs[SALT_POS_HOST].salt_len - 4;
/**
* base
*/
COPY_PW (pws[gid]);
/**
* loop
*/
for (u32 il_pos = 0; il_pos < IL_CNT; il_pos++)
{
pw_t tmp = PASTE_PW;
tmp.pw_len = apply_rules (rules_buf[il_pos].cmds, tmp.i, tmp.pw_len);
md5_ctx_t ctx1;
md5_init (&ctx1);
md5_update (&ctx1, tmp.i, tmp.pw_len);
u32 s0[4];
u32 s1[4];
u32 s2[4];
u32 s3[4];
s0[0] = salt_buf0[0];
s0[1] = salt_buf0[1];
s0[2] = salt_buf0[2];
s0[3] = salt_buf0[3];
s1[0] = salt_buf1[0];
s1[1] = salt_buf1[1];
s1[2] = salt_buf1[2];
s1[3] = salt_buf1[3];
s2[0] = 0;
s2[1] = 0;
s2[2] = 0;
s2[3] = 0;
s3[0] = 0;
s3[1] = 0;
s3[2] = 0;
s3[3] = 0;
md5_update_64 (&ctx1, s0, s1, s2, s3, salt_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];
md5_ctx_t ctx;
md5_init (&ctx);
ctx.w0[0] = uint_to_hex_lower8 ((a >> 0) & 255) << 0
| uint_to_hex_lower8 ((a >> 8) & 255) << 16;
ctx.w0[1] = uint_to_hex_lower8 ((a >> 16) & 255) << 0
| uint_to_hex_lower8 ((a >> 24) & 255) << 16;
ctx.w0[2] = uint_to_hex_lower8 ((b >> 0) & 255) << 0
| uint_to_hex_lower8 ((b >> 8) & 255) << 16;
ctx.w0[3] = uint_to_hex_lower8 ((b >> 16) & 255) << 0
| uint_to_hex_lower8 ((b >> 24) & 255) << 16;
ctx.w1[0] = uint_to_hex_lower8 ((c >> 0) & 255) << 0
| uint_to_hex_lower8 ((c >> 8) & 255) << 16;
ctx.w1[1] = uint_to_hex_lower8 ((c >> 16) & 255) << 0
| uint_to_hex_lower8 ((c >> 24) & 255) << 16;
ctx.w1[2] = uint_to_hex_lower8 ((d >> 0) & 255) << 0
| uint_to_hex_lower8 ((d >> 8) & 255) << 16;
ctx.w1[3] = uint_to_hex_lower8 ((d >> 16) & 255) << 0
| uint_to_hex_lower8 ((d >> 24) & 255) << 16;
ctx.w2[0] = challenge;
ctx.w2[1] = 0;
ctx.w2[2] = 0;
ctx.w2[3] = 0;
ctx.w3[0] = 0;
ctx.w3[1] = 0;
ctx.w3[2] = 0;
ctx.w3[3] = 0;
ctx.len = 32 + 4;
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 m11100_sxx (KERN_ATTR_RULES ())
{
/**
* modifier
*/
const u64 gid = get_global_id (0);
const u64 lid = get_local_id (0);
const u64 lsz = get_local_size (0);
/**
* bin2asc table
*/
LOCAL_VK u32 l_bin2asc[256];
for (u32 i = lid; i < 256; i += lsz)
{
const u32 i0 = (i >> 0) & 15;
const u32 i1 = (i >> 4) & 15;
l_bin2asc[i] = ((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]
};
/**
* salt
*/
u32 challenge;
challenge = salt_bufs[SALT_POS_HOST].salt_buf[0];
u32 salt_buf0[4];
u32 salt_buf1[4];
salt_buf0[0] = salt_bufs[SALT_POS_HOST].salt_buf[1]; // not a bug, see challenge
salt_buf0[1] = salt_bufs[SALT_POS_HOST].salt_buf[2];
salt_buf0[2] = salt_bufs[SALT_POS_HOST].salt_buf[3];
salt_buf0[3] = salt_bufs[SALT_POS_HOST].salt_buf[4];
salt_buf1[0] = salt_bufs[SALT_POS_HOST].salt_buf[5];
salt_buf1[1] = salt_bufs[SALT_POS_HOST].salt_buf[6];
salt_buf1[2] = salt_bufs[SALT_POS_HOST].salt_buf[7];
salt_buf1[3] = salt_bufs[SALT_POS_HOST].salt_buf[8];
const u32 salt_len = salt_bufs[SALT_POS_HOST].salt_len - 4;
/**
* base
*/
COPY_PW (pws[gid]);
/**
* loop
*/
for (u32 il_pos = 0; il_pos < IL_CNT; il_pos++)
{
pw_t tmp = PASTE_PW;
tmp.pw_len = apply_rules (rules_buf[il_pos].cmds, tmp.i, tmp.pw_len);
md5_ctx_t ctx1;
md5_init (&ctx1);
md5_update (&ctx1, tmp.i, tmp.pw_len);
u32 s0[4];
u32 s1[4];
u32 s2[4];
u32 s3[4];
s0[0] = salt_buf0[0];
s0[1] = salt_buf0[1];
s0[2] = salt_buf0[2];
s0[3] = salt_buf0[3];
s1[0] = salt_buf1[0];
s1[1] = salt_buf1[1];
s1[2] = salt_buf1[2];
s1[3] = salt_buf1[3];
s2[0] = 0;
s2[1] = 0;
s2[2] = 0;
s2[3] = 0;
s3[0] = 0;
s3[1] = 0;
s3[2] = 0;
s3[3] = 0;
md5_update_64 (&ctx1, s0, s1, s2, s3, salt_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];
md5_ctx_t ctx;
md5_init (&ctx);
ctx.w0[0] = uint_to_hex_lower8 ((a >> 0) & 255) << 0
| uint_to_hex_lower8 ((a >> 8) & 255) << 16;
ctx.w0[1] = uint_to_hex_lower8 ((a >> 16) & 255) << 0
| uint_to_hex_lower8 ((a >> 24) & 255) << 16;
ctx.w0[2] = uint_to_hex_lower8 ((b >> 0) & 255) << 0
| uint_to_hex_lower8 ((b >> 8) & 255) << 16;
ctx.w0[3] = uint_to_hex_lower8 ((b >> 16) & 255) << 0
| uint_to_hex_lower8 ((b >> 24) & 255) << 16;
ctx.w1[0] = uint_to_hex_lower8 ((c >> 0) & 255) << 0
| uint_to_hex_lower8 ((c >> 8) & 255) << 16;
ctx.w1[1] = uint_to_hex_lower8 ((c >> 16) & 255) << 0
| uint_to_hex_lower8 ((c >> 24) & 255) << 16;
ctx.w1[2] = uint_to_hex_lower8 ((d >> 0) & 255) << 0
| uint_to_hex_lower8 ((d >> 8) & 255) << 16;
ctx.w1[3] = uint_to_hex_lower8 ((d >> 16) & 255) << 0
| uint_to_hex_lower8 ((d >> 24) & 255) << 16;
ctx.w2[0] = challenge;
ctx.w2[1] = 0;
ctx.w2[2] = 0;
ctx.w2[3] = 0;
ctx.w3[0] = 0;
ctx.w3[1] = 0;
ctx.w3[2] = 0;
ctx.w3[3] = 0;
ctx.len = 32 + 4;
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);
}
}