1
0
mirror of https://github.com/hashcat/hashcat.git synced 2025-07-23 15:08:37 +00:00
hashcat/OpenCL/m28800-pure.cl
Jens Steube 15ada5124e Further simplified the use of inc_hash_scrypt.cl without any speed regression, and updated all affected plugin kernels. Use m08900-pure.cl as a template.
Updated kernel declarations from "KERNEL_FQ void HC_ATTR_SEQ" to "KERNEL_FQ KERNEL_FA void". Please update your custom plugin kernels accordingly.
Added spilling size as a factor in calculating usable memory per device. This is based on undocumented variables and may not be 100% accurate, but it works well in practice.
Added a compiler hint to scrypt-based kernels indicating the guaranteed maximum thread count per kernel invocation.
Removed redundant kernel code 29800, as it is identical to 27700, and updated the plugin.
2025-06-21 17:41:26 +02:00

400 lines
9.4 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_simd.cl)
#include M2S(INCLUDE_PATH/inc_hash_sha1.cl)
#include M2S(INCLUDE_PATH/inc_cipher_aes.cl)
#endif
#define COMPARE_S M2S(INCLUDE_PATH/inc_comp_single.cl)
#define COMPARE_M M2S(INCLUDE_PATH/inc_comp_multi.cl)
typedef struct krb5db_17
{
u32 user[128];
u32 domain[128];
u32 account_info[512];
u32 account_info_len;
} krb5db_17_t;
typedef struct krb5db_17_tmp
{
u32 ipad[5];
u32 opad[5];
u32 dgst[10];
u32 out[10];
} krb5db_17_tmp_t;
DECLSPEC void aes128_encrypt_cbc (PRIVATE_AS const u32 *aes_ks, PRIVATE_AS u32 *aes_iv, PRIVATE_AS const u32 *in, PRIVATE_AS u32 *out, SHM_TYPE u32 *s_te0, SHM_TYPE u32 *s_te1, SHM_TYPE u32 *s_te2, SHM_TYPE u32 *s_te3, SHM_TYPE u32 *s_te4)
{
u32 data[4];
data[0] = hc_swap32_S (in[0]);
data[1] = hc_swap32_S (in[1]);
data[2] = hc_swap32_S (in[2]);
data[3] = hc_swap32_S (in[3]);
data[0] ^= aes_iv[0];
data[1] ^= aes_iv[1];
data[2] ^= aes_iv[2];
data[3] ^= aes_iv[3];
aes128_encrypt (aes_ks, data, out, s_te0, s_te1, s_te2, s_te3, s_te4);
aes_iv[0] = out[0];
aes_iv[1] = out[1];
aes_iv[2] = out[2];
aes_iv[3] = out[3];
out[0] = hc_swap32_S (out[0]);
out[1] = hc_swap32_S (out[1]);
out[2] = hc_swap32_S (out[2]);
out[3] = hc_swap32_S (out[3]);
}
DECLSPEC void aes128_decrypt_cbc (PRIVATE_AS const u32 *ks1, PRIVATE_AS const u32 *in, PRIVATE_AS u32 *out, PRIVATE_AS u32 *essiv, SHM_TYPE u32 *s_td0, SHM_TYPE u32 *s_td1, SHM_TYPE u32 *s_td2, SHM_TYPE u32 *s_td3, SHM_TYPE u32 *s_td4)
{
aes128_decrypt (ks1, in, out, s_td0, s_td1, s_td2, s_td3, s_td4);
out[0] ^= essiv[0];
out[1] ^= essiv[1];
out[2] ^= essiv[2];
out[3] ^= essiv[3];
essiv[0] = in[0];
essiv[1] = in[1];
essiv[2] = in[2];
essiv[3] = in[3];
}
DECLSPEC void hmac_sha1_run_V (PRIVATE_AS u32x *w0, PRIVATE_AS u32x *w1, PRIVATE_AS u32x *w2, PRIVATE_AS u32x *w3, PRIVATE_AS u32x *ipad, PRIVATE_AS u32x *opad, PRIVATE_AS u32x *digest)
{
digest[0] = ipad[0];
digest[1] = ipad[1];
digest[2] = ipad[2];
digest[3] = ipad[3];
digest[4] = ipad[4];
sha1_transform_vector (w0, w1, w2, w3, digest);
w0[0] = digest[0];
w0[1] = digest[1];
w0[2] = digest[2];
w0[3] = digest[3];
w1[0] = digest[4];
w1[1] = 0x80000000;
w1[2] = 0;
w1[3] = 0;
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = (64 + 20) * 8;
digest[0] = opad[0];
digest[1] = opad[1];
digest[2] = opad[2];
digest[3] = opad[3];
digest[4] = opad[4];
sha1_transform_vector (w0, w1, w2, w3, digest);
}
KERNEL_FQ KERNEL_FA void m28800_init (KERN_ATTR_TMPS_ESALT (krb5db_17_tmp_t, krb5db_17_t))
{
/**
* base
*/
const u64 gid = get_global_id (0);
if (gid >= GID_CNT) return;
/**
* main
*/
/* initialize hmac-sha1 for pbkdf2(password, account, 4096, account_len) */
sha1_hmac_ctx_t sha1_hmac_ctx;
sha1_hmac_init_global_swap (&sha1_hmac_ctx, pws[gid].i, pws[gid].pw_len);
tmps[gid].ipad[0] = sha1_hmac_ctx.ipad.h[0];
tmps[gid].ipad[1] = sha1_hmac_ctx.ipad.h[1];
tmps[gid].ipad[2] = sha1_hmac_ctx.ipad.h[2];
tmps[gid].ipad[3] = sha1_hmac_ctx.ipad.h[3];
tmps[gid].ipad[4] = sha1_hmac_ctx.ipad.h[4];
tmps[gid].opad[0] = sha1_hmac_ctx.opad.h[0];
tmps[gid].opad[1] = sha1_hmac_ctx.opad.h[1];
tmps[gid].opad[2] = sha1_hmac_ctx.opad.h[2];
tmps[gid].opad[3] = sha1_hmac_ctx.opad.h[3];
tmps[gid].opad[4] = sha1_hmac_ctx.opad.h[4];
sha1_hmac_update_global_swap (&sha1_hmac_ctx, esalt_bufs[DIGESTS_OFFSET_HOST].account_info, esalt_bufs[DIGESTS_OFFSET_HOST].account_info_len);
for (u32 i = 0, j = 1; i < 4; i += 5, j += 1)
{
sha1_hmac_ctx_t sha1_hmac_ctx2 = sha1_hmac_ctx;
u32 w0[4];
u32 w1[4];
u32 w2[4];
u32 w3[4];
w0[0] = j;
w0[1] = 0;
w0[2] = 0;
w0[3] = 0;
w1[0] = 0;
w1[1] = 0;
w1[2] = 0;
w1[3] = 0;
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = 0;
sha1_hmac_update_64 (&sha1_hmac_ctx2, w0, w1, w2, w3, 4);
sha1_hmac_final (&sha1_hmac_ctx2);
tmps[gid].dgst[i + 0] = sha1_hmac_ctx2.opad.h[0];
tmps[gid].dgst[i + 1] = sha1_hmac_ctx2.opad.h[1];
tmps[gid].dgst[i + 2] = sha1_hmac_ctx2.opad.h[2];
tmps[gid].dgst[i + 3] = sha1_hmac_ctx2.opad.h[3];
tmps[gid].dgst[i + 4] = sha1_hmac_ctx2.opad.h[4];
tmps[gid].out[i + 0] = tmps[gid].dgst[i + 0];
tmps[gid].out[i + 1] = tmps[gid].dgst[i + 1];
tmps[gid].out[i + 2] = tmps[gid].dgst[i + 2];
tmps[gid].out[i + 3] = tmps[gid].dgst[i + 3];
tmps[gid].out[i + 4] = tmps[gid].dgst[i + 4];
}
}
KERNEL_FQ KERNEL_FA void m28800_loop (KERN_ATTR_TMPS_ESALT (krb5db_17_tmp_t, krb5db_17_t))
{
/**
* base
*/
const u64 gid = get_global_id (0);
if ((gid * VECT_SIZE) >= GID_CNT) return;
u32x ipad[5];
u32x opad[5];
ipad[0] = packv (tmps, ipad, gid, 0);
ipad[1] = packv (tmps, ipad, gid, 1);
ipad[2] = packv (tmps, ipad, gid, 2);
ipad[3] = packv (tmps, ipad, gid, 3);
ipad[4] = packv (tmps, ipad, gid, 4);
opad[0] = packv (tmps, opad, gid, 0);
opad[1] = packv (tmps, opad, gid, 1);
opad[2] = packv (tmps, opad, gid, 2);
opad[3] = packv (tmps, opad, gid, 3);
opad[4] = packv (tmps, opad, gid, 4);
for (u32 i = 0; i < 4; i += 5)
{
u32x dgst[5];
u32x out[5];
dgst[0] = packv (tmps, dgst, gid, i + 0);
dgst[1] = packv (tmps, dgst, gid, i + 1);
dgst[2] = packv (tmps, dgst, gid, i + 2);
dgst[3] = packv (tmps, dgst, gid, i + 3);
dgst[4] = packv (tmps, dgst, gid, i + 4);
out[0] = packv (tmps, out, gid, i + 0);
out[1] = packv (tmps, out, gid, i + 1);
out[2] = packv (tmps, out, gid, i + 2);
out[3] = packv (tmps, out, gid, i + 3);
out[4] = packv (tmps, out, gid, i + 4);
for (u32 j = 0; j < LOOP_CNT; j++)
{
u32x w0[4];
u32x w1[4];
u32x w2[4];
u32x w3[4];
w0[0] = dgst[0];
w0[1] = dgst[1];
w0[2] = dgst[2];
w0[3] = dgst[3];
w1[0] = dgst[4];
w1[1] = 0x80000000;
w1[2] = 0;
w1[3] = 0;
w2[0] = 0;
w2[1] = 0;
w2[2] = 0;
w2[3] = 0;
w3[0] = 0;
w3[1] = 0;
w3[2] = 0;
w3[3] = (64 + 20) * 8;
hmac_sha1_run_V (w0, w1, w2, w3, ipad, opad, dgst);
out[0] ^= dgst[0];
out[1] ^= dgst[1];
out[2] ^= dgst[2];
out[3] ^= dgst[3];
out[4] ^= dgst[4];
}
unpackv (tmps, dgst, gid, i + 0, dgst[0]);
unpackv (tmps, dgst, gid, i + 1, dgst[1]);
unpackv (tmps, dgst, gid, i + 2, dgst[2]);
unpackv (tmps, dgst, gid, i + 3, dgst[3]);
unpackv (tmps, dgst, gid, i + 4, dgst[4]);
unpackv (tmps, out, gid, i + 0, out[0]);
unpackv (tmps, out, gid, i + 1, out[1]);
unpackv (tmps, out, gid, i + 2, out[2]);
unpackv (tmps, out, gid, i + 3, out[3]);
unpackv (tmps, out, gid, i + 4, out[4]);
}
}
KERNEL_FQ KERNEL_FA void m28800_comp (KERN_ATTR_TMPS_ESALT (krb5db_17_tmp_t, krb5db_17_t))
{
/**
* base
*/
const u64 gid = get_global_id (0);
const u64 lid = get_local_id (0);
const u64 lsz = get_local_size (0);
/**
* aes shared
*/
#ifdef REAL_SHM
LOCAL_VK u32 s_te0[256];
LOCAL_VK u32 s_te1[256];
LOCAL_VK u32 s_te2[256];
LOCAL_VK u32 s_te3[256];
LOCAL_VK u32 s_te4[256];
LOCAL_VK u32 s_td0[256];
LOCAL_VK u32 s_td1[256];
LOCAL_VK u32 s_td2[256];
LOCAL_VK u32 s_td3[256];
LOCAL_VK u32 s_td4[256];
for (u32 i = lid; i < 256; i += lsz)
{
s_te0[i] = te0[i];
s_te1[i] = te1[i];
s_te2[i] = te2[i];
s_te3[i] = te3[i];
s_te4[i] = te4[i];
s_td0[i] = td0[i];
s_td1[i] = td1[i];
s_td2[i] = td2[i];
s_td3[i] = td3[i];
s_td4[i] = td4[i];
}
SYNC_THREADS ();
#else
CONSTANT_AS u32a *s_te0 = te0;
CONSTANT_AS u32a *s_te1 = te1;
CONSTANT_AS u32a *s_te2 = te2;
CONSTANT_AS u32a *s_te3 = te3;
CONSTANT_AS u32a *s_te4 = te4;
CONSTANT_AS u32a *s_td0 = td0;
CONSTANT_AS u32a *s_td1 = td1;
CONSTANT_AS u32a *s_td2 = td2;
CONSTANT_AS u32a *s_td3 = td3;
CONSTANT_AS u32a *s_td4 = td4;
#endif
if (gid >= GID_CNT) return;
/*
at this point, the output ('seed') will be used to generate AES keys:
key_bytes = derive(seed, 'kerberos'.encode(), seedsize)
'key_bytes' will be the AES key used to generate 'ke' and 'ki'
'ke' will be the AES key to decrypt the ticket
'ki' will be the key to compute the final HMAC
*/
u32 nfolded[4];
// we can precompute _nfold('kerberos', 16)
nfolded[0] = 0x6b657262;
nfolded[1] = 0x65726f73;
nfolded[2] = 0x7b9b5b2b;
nfolded[3] = 0x93132b93;
// then aes_cbc encrypt this nfolded value with 'seed' as key along with a null IV
u32 aes_key[4];
aes_key[0] = hc_swap32_S (tmps[gid].out[0]);
aes_key[1] = hc_swap32_S (tmps[gid].out[1]);
aes_key[2] = hc_swap32_S (tmps[gid].out[2]);
aes_key[3] = hc_swap32_S (tmps[gid].out[3]);
u32 aes_iv[4];
aes_iv[0] = 0;
aes_iv[1] = 0;
aes_iv[2] = 0;
aes_iv[3] = 0;
u32 aes_ks[44];
aes128_set_encrypt_key (aes_ks, aes_key, s_te0, s_te1, s_te2, s_te3);
u32 key_bytes[4];
aes128_encrypt_cbc (aes_ks, aes_iv, nfolded, key_bytes, s_te0, s_te1, s_te2, s_te3, s_te4);
const u32 r0 = key_bytes[0];
const u32 r1 = key_bytes[1];
const u32 r2 = key_bytes[2];
const u32 r3 = key_bytes[3];
#define il_pos 0
#ifdef KERNEL_STATIC
#include COMPARE_M
#endif
}