#define _KRB5PA_
+//shared mem too small
+//#define NEW_SIMD_CODE
+
#include "include/constants.h"
#include "include/kernel_vendor.h"
#include "OpenCL/common.c"
#include "include/rp_kernel.h"
#include "OpenCL/rp.c"
+#include "OpenCL/simd.c"
typedef struct
{
} RC4_KEY;
-static void swap (__local RC4_KEY *rc4_key, const u8 i, const u8 j)
+void swap (__local RC4_KEY *rc4_key, const u8 i, const u8 j)
{
u8 tmp;
rc4_key->S[j] = tmp;
}
-static void rc4_init_16 (__local RC4_KEY *rc4_key, const u32 data[4])
+void rc4_init_16 (__local RC4_KEY *rc4_key, const u32 data[4])
{
u32 v = 0x03020100;
u32 a = 0x04040404;
}
}
-static u8 rc4_next_16 (__local RC4_KEY *rc4_key, u8 i, u8 j, const u32 in[4], u32 out[4])
+u8 rc4_next_16 (__local RC4_KEY *rc4_key, u8 i, u8 j, const u32 in[4], u32 out[4])
{
#pragma unroll
for (u32 k = 0; k < 4; k++)
return j;
}
-static int decrypt_and_check (__local RC4_KEY *rc4_key, u32 data[4], u32 timestamp_ct[8])
+int decrypt_and_check (__local RC4_KEY *rc4_key, u32 data[4], u32 timestamp_ct[8])
{
rc4_init_16 (rc4_key, data);
return 1;
}
-static void md4_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[4])
+void md4_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[4])
{
u32 a = digest[0];
u32 b = digest[1];
digest[3] += d;
}
-static void md5_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[4])
+void md5_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[4])
{
u32 a = digest[0];
u32 b = digest[1];
digest[3] += d;
}
-static void hmac_md5_pad (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[4], u32 opad[4])
+void hmac_md5_pad (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[4], u32 opad[4])
{
w0[0] = w0[0] ^ 0x36363636;
w0[1] = w0[1] ^ 0x36363636;
md5_transform (w0, w1, w2, w3, opad);
}
-static void hmac_md5_run (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[4], u32 opad[4], u32 digest[4])
+void hmac_md5_run (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[4], u32 opad[4], u32 digest[4])
{
digest[0] = ipad[0];
digest[1] = ipad[1];
md5_transform (w0, w1, w2, w3, digest);
}
-static void kerb_prepare (const u32 w0[4], const u32 w1[4], const u32 pw_len, const u32 checksum[4], u32 digest[4])
+void kerb_prepare (const u32 w0[4], const u32 w1[4], const u32 pw_len, const u32 checksum[4], u32 digest[4])
{
/**
* pads
* modifier
*/
- __local RC4_KEY rc4_keys[64];
-
const u32 lid = get_local_id (0);
- const u32 gid = get_global_id (0);
-
- if (gid >= gid_max) return;
/**
* base
*/
- u32 pw_buf0[4];
+ const u32 gid = get_global_id (0);
- 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];
+ if (gid >= gid_max) return;
+ u32 pw_buf0[4];
u32 pw_buf1[4];
- 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];
+ 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_len = pws[gid].pw_len;
timestamp_ct[7] = krb5pa_bufs[salt_pos].timestamp[7];
/**
- * loop
+ * shared
*/
- for (u32 il_pos = 0; il_pos < il_cnt; il_pos++)
- {
- u32 w0[4];
-
- w0[0] = pw_buf0[0];
- w0[1] = pw_buf0[1];
- w0[2] = pw_buf0[2];
- w0[3] = pw_buf0[3];
-
- u32 w1[4];
-
- w1[0] = pw_buf1[0];
- w1[1] = pw_buf1[1];
- w1[2] = pw_buf1[2];
- w1[3] = pw_buf1[3];
-
- u32 w2[4];
-
- w2[0] = 0;
- w2[1] = 0;
- w2[2] = 0;
- w2[3] = 0;
+ __local RC4_KEY rc4_keys[64];
- u32 w3[4];
+ /**
+ * loop
+ */
- w3[0] = 0;
- w3[1] = 0;
- w3[2] = 0;
- w3[3] = 0;
+ for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
+ {
+ u32x w0[4] = { 0 };
+ u32x w1[4] = { 0 };
+ u32x w2[4] = { 0 };
+ u32x w3[4] = { 0 };
- const u32 out_len = apply_rules (rules_buf[il_pos].cmds, w0, w1, pw_len);
+ const u32x out_len = apply_rules_vect (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1);
/**
* kerberos
* modifier
*/
- __local RC4_KEY rc4_keys[64];
-
const u32 lid = get_local_id (0);
- const u32 gid = get_global_id (0);
-
- if (gid >= gid_max) return;
/**
* base
*/
- u32 pw_buf0[4];
+ const u32 gid = get_global_id (0);
- 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];
+ if (gid >= gid_max) return;
+ u32 pw_buf0[4];
u32 pw_buf1[4];
- 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];
+ 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_len = pws[gid].pw_len;
timestamp_ct[7] = krb5pa_bufs[salt_pos].timestamp[7];
/**
- * loop
+ * shared
*/
- for (u32 il_pos = 0; il_pos < il_cnt; il_pos++)
- {
- u32 w0[4];
-
- w0[0] = pw_buf0[0];
- w0[1] = pw_buf0[1];
- w0[2] = pw_buf0[2];
- w0[3] = pw_buf0[3];
-
- u32 w1[4];
-
- w1[0] = pw_buf1[0];
- w1[1] = pw_buf1[1];
- w1[2] = pw_buf1[2];
- w1[3] = pw_buf1[3];
-
- u32 w2[4];
-
- w2[0] = 0;
- w2[1] = 0;
- w2[2] = 0;
- w2[3] = 0;
+ __local RC4_KEY rc4_keys[64];
- u32 w3[4];
+ /**
+ * loop
+ */
- w3[0] = 0;
- w3[1] = 0;
- w3[2] = 0;
- w3[3] = 0;
+ for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE)
+ {
+ u32x w0[4] = { 0 };
+ u32x w1[4] = { 0 };
+ u32x w2[4] = { 0 };
+ u32x w3[4] = { 0 };
- const u32 out_len = apply_rules (rules_buf[il_pos].cmds, w0, w1, pw_len);
+ const u32x out_len = apply_rules_vect (pw_buf0, pw_buf1, pw_len, rules_buf, il_pos, w0, w1);
/**
* kerberos
projects / hashcat.git / blobdiff