diff --git a/OpenCL/inc_hash_md5.cl b/OpenCL/inc_hash_md5.cl index fe31dc099..3c52c1f40 100644 --- a/OpenCL/inc_hash_md5.cl +++ b/OpenCL/inc_hash_md5.cl @@ -771,262 +771,6 @@ DECLSPEC void md5_update_global_utf16le_swap (md5_ctx_t *ctx, GLOBAL_AS const u3 md5_update_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2); } -DECLSPEC void md5_update_local (md5_ctx_t *ctx, LOCAL_AS const u32 *w, const int len) -{ - u32 w0[4]; - u32 w1[4]; - u32 w2[4]; - u32 w3[4]; - - int pos1; - int pos4; - - for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16) - { - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - w2[0] = w[pos4 + 8]; - w2[1] = w[pos4 + 9]; - w2[2] = w[pos4 + 10]; - w2[3] = w[pos4 + 11]; - w3[0] = w[pos4 + 12]; - w3[1] = w[pos4 + 13]; - w3[2] = w[pos4 + 14]; - w3[3] = w[pos4 + 15]; - - md5_update_64 (ctx, w0, w1, w2, w3, 64); - } - - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - w2[0] = w[pos4 + 8]; - w2[1] = w[pos4 + 9]; - w2[2] = w[pos4 + 10]; - w2[3] = w[pos4 + 11]; - w3[0] = w[pos4 + 12]; - w3[1] = w[pos4 + 13]; - w3[2] = w[pos4 + 14]; - w3[3] = w[pos4 + 15]; - - md5_update_64 (ctx, w0, w1, w2, w3, len - pos1); -} - -DECLSPEC void md5_update_local_swap (md5_ctx_t *ctx, LOCAL_AS const u32 *w, const int len) -{ - u32 w0[4]; - u32 w1[4]; - u32 w2[4]; - u32 w3[4]; - - int pos1; - int pos4; - - for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16) - { - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - w2[0] = w[pos4 + 8]; - w2[1] = w[pos4 + 9]; - w2[2] = w[pos4 + 10]; - w2[3] = w[pos4 + 11]; - w3[0] = w[pos4 + 12]; - w3[1] = w[pos4 + 13]; - w3[2] = w[pos4 + 14]; - w3[3] = w[pos4 + 15]; - - w0[0] = hc_swap32_S (w0[0]); - w0[1] = hc_swap32_S (w0[1]); - w0[2] = hc_swap32_S (w0[2]); - w0[3] = hc_swap32_S (w0[3]); - w1[0] = hc_swap32_S (w1[0]); - w1[1] = hc_swap32_S (w1[1]); - w1[2] = hc_swap32_S (w1[2]); - w1[3] = hc_swap32_S (w1[3]); - w2[0] = hc_swap32_S (w2[0]); - w2[1] = hc_swap32_S (w2[1]); - w2[2] = hc_swap32_S (w2[2]); - w2[3] = hc_swap32_S (w2[3]); - w3[0] = hc_swap32_S (w3[0]); - w3[1] = hc_swap32_S (w3[1]); - w3[2] = hc_swap32_S (w3[2]); - w3[3] = hc_swap32_S (w3[3]); - - md5_update_64 (ctx, w0, w1, w2, w3, 64); - } - - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - w2[0] = w[pos4 + 8]; - w2[1] = w[pos4 + 9]; - w2[2] = w[pos4 + 10]; - w2[3] = w[pos4 + 11]; - w3[0] = w[pos4 + 12]; - w3[1] = w[pos4 + 13]; - w3[2] = w[pos4 + 14]; - w3[3] = w[pos4 + 15]; - - w0[0] = hc_swap32_S (w0[0]); - w0[1] = hc_swap32_S (w0[1]); - w0[2] = hc_swap32_S (w0[2]); - w0[3] = hc_swap32_S (w0[3]); - w1[0] = hc_swap32_S (w1[0]); - w1[1] = hc_swap32_S (w1[1]); - w1[2] = hc_swap32_S (w1[2]); - w1[3] = hc_swap32_S (w1[3]); - w2[0] = hc_swap32_S (w2[0]); - w2[1] = hc_swap32_S (w2[1]); - w2[2] = hc_swap32_S (w2[2]); - w2[3] = hc_swap32_S (w2[3]); - w3[0] = hc_swap32_S (w3[0]); - w3[1] = hc_swap32_S (w3[1]); - w3[2] = hc_swap32_S (w3[2]); - w3[3] = hc_swap32_S (w3[3]); - - md5_update_64 (ctx, w0, w1, w2, w3, len - pos1); -} - -DECLSPEC void md5_update_local_utf16le (md5_ctx_t *ctx, LOCAL_AS const u32 *w, const int len) -{ - u32 w0[4]; - u32 w1[4]; - u32 w2[4]; - u32 w3[4]; - - int pos1; - int pos4; - - for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8) - { - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - - make_utf16le_S (w1, w2, w3); - make_utf16le_S (w0, w0, w1); - - md5_update_64 (ctx, w0, w1, w2, w3, 32 * 2); - } - - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - - make_utf16le_S (w1, w2, w3); - make_utf16le_S (w0, w0, w1); - - md5_update_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2); -} - -DECLSPEC void md5_update_local_utf16le_swap (md5_ctx_t *ctx, LOCAL_AS const u32 *w, const int len) -{ - u32 w0[4]; - u32 w1[4]; - u32 w2[4]; - u32 w3[4]; - - int pos1; - int pos4; - - for (pos1 = 0, pos4 = 0; pos1 < len - 32; pos1 += 32, pos4 += 8) - { - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - - make_utf16le_S (w1, w2, w3); - make_utf16le_S (w0, w0, w1); - - w0[0] = hc_swap32_S (w0[0]); - w0[1] = hc_swap32_S (w0[1]); - w0[2] = hc_swap32_S (w0[2]); - w0[3] = hc_swap32_S (w0[3]); - w1[0] = hc_swap32_S (w1[0]); - w1[1] = hc_swap32_S (w1[1]); - w1[2] = hc_swap32_S (w1[2]); - w1[3] = hc_swap32_S (w1[3]); - w2[0] = hc_swap32_S (w2[0]); - w2[1] = hc_swap32_S (w2[1]); - w2[2] = hc_swap32_S (w2[2]); - w2[3] = hc_swap32_S (w2[3]); - w3[0] = hc_swap32_S (w3[0]); - w3[1] = hc_swap32_S (w3[1]); - w3[2] = hc_swap32_S (w3[2]); - w3[3] = hc_swap32_S (w3[3]); - - md5_update_64 (ctx, w0, w1, w2, w3, 32 * 2); - } - - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - - make_utf16le_S (w1, w2, w3); - make_utf16le_S (w0, w0, w1); - - w0[0] = hc_swap32_S (w0[0]); - w0[1] = hc_swap32_S (w0[1]); - w0[2] = hc_swap32_S (w0[2]); - w0[3] = hc_swap32_S (w0[3]); - w1[0] = hc_swap32_S (w1[0]); - w1[1] = hc_swap32_S (w1[1]); - w1[2] = hc_swap32_S (w1[2]); - w1[3] = hc_swap32_S (w1[3]); - w2[0] = hc_swap32_S (w2[0]); - w2[1] = hc_swap32_S (w2[1]); - w2[2] = hc_swap32_S (w2[2]); - w2[3] = hc_swap32_S (w2[3]); - w3[0] = hc_swap32_S (w3[0]); - w3[1] = hc_swap32_S (w3[1]); - w3[2] = hc_swap32_S (w3[2]); - w3[3] = hc_swap32_S (w3[3]); - - md5_update_64 (ctx, w0, w1, w2, w3, (len - pos1) * 2); -} - DECLSPEC void md5_final (md5_ctx_t *ctx) { MAYBE_VOLATILE const int pos = ctx->len & 63; @@ -1742,58 +1486,6 @@ DECLSPEC void md5_update_vector (md5_ctx_vector_t *ctx, const u32x *w, const int md5_update_vector_64 (ctx, w0, w1, w2, w3, len - pos1); } -DECLSPEC void md5_update_vector_from_scalar (md5_ctx_vector_t *ctx, const u32 *w, const int len) -{ - u32x w0[4]; - u32x w1[4]; - u32x w2[4]; - u32x w3[4]; - - int pos1; - int pos4; - - for (pos1 = 0, pos4 = 0; pos1 < len - 64; pos1 += 64, pos4 += 16) - { - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - w2[0] = w[pos4 + 8]; - w2[1] = w[pos4 + 9]; - w2[2] = w[pos4 + 10]; - w2[3] = w[pos4 + 11]; - w3[0] = w[pos4 + 12]; - w3[1] = w[pos4 + 13]; - w3[2] = w[pos4 + 14]; - w3[3] = w[pos4 + 15]; - - md5_update_vector_64 (ctx, w0, w1, w2, w3, 64); - } - - w0[0] = w[pos4 + 0]; - w0[1] = w[pos4 + 1]; - w0[2] = w[pos4 + 2]; - w0[3] = w[pos4 + 3]; - w1[0] = w[pos4 + 4]; - w1[1] = w[pos4 + 5]; - w1[2] = w[pos4 + 6]; - w1[3] = w[pos4 + 7]; - w2[0] = w[pos4 + 8]; - w2[1] = w[pos4 + 9]; - w2[2] = w[pos4 + 10]; - w2[3] = w[pos4 + 11]; - w3[0] = w[pos4 + 12]; - w3[1] = w[pos4 + 13]; - w3[2] = w[pos4 + 14]; - w3[3] = w[pos4 + 15]; - - md5_update_vector_64 (ctx, w0, w1, w2, w3, len - pos1); -} - DECLSPEC void md5_update_vector_swap (md5_ctx_vector_t *ctx, const u32x *w, const int len) { u32x w0[4]; diff --git a/OpenCL/inc_hash_md5.h b/OpenCL/inc_hash_md5.h index e16dda780..1e6eaaf93 100644 --- a/OpenCL/inc_hash_md5.h +++ b/OpenCL/inc_hash_md5.h @@ -100,10 +100,6 @@ DECLSPEC void md5_update_global (md5_ctx_t *ctx, GLOBAL_AS const u32 *w, const i DECLSPEC void md5_update_global_swap (md5_ctx_t *ctx, GLOBAL_AS const u32 *w, const int len); DECLSPEC void md5_update_global_utf16le (md5_ctx_t *ctx, GLOBAL_AS const u32 *w, const int len); DECLSPEC void md5_update_global_utf16le_swap (md5_ctx_t *ctx, GLOBAL_AS const u32 *w, const int len); -DECLSPEC void md5_update_local (md5_ctx_t *ctx, LOCAL_AS const u32 *w, const int len); -DECLSPEC void md5_update_local_swap (md5_ctx_t *ctx, LOCAL_AS const u32 *w, const int len); -DECLSPEC void md5_update_local_utf16le (md5_ctx_t *ctx, LOCAL_AS const u32 *w, const int len); -DECLSPEC void md5_update_local_utf16le_swap (md5_ctx_t *ctx, LOCAL_AS const u32 *w, const int len); DECLSPEC void md5_final (md5_ctx_t *ctx); DECLSPEC void md5_hmac_init_64 (md5_hmac_ctx_t *ctx, const u32 *w0, const u32 *w1, const u32 *w2, const u32 *w3); DECLSPEC void md5_hmac_init (md5_hmac_ctx_t *ctx, const u32 *w, const int len); @@ -125,7 +121,6 @@ DECLSPEC void md5_init_vector (md5_ctx_vector_t *ctx); DECLSPEC void md5_init_vector_from_scalar (md5_ctx_vector_t *ctx, md5_ctx_t *ctx0); DECLSPEC void md5_update_vector_64 (md5_ctx_vector_t *ctx, u32x *w0, u32x *w1, u32x *w2, u32x *w3, const int len); DECLSPEC void md5_update_vector (md5_ctx_vector_t *ctx, const u32x *w, const int len); -DECLSPEC void md5_update_vector_from_scalar (md5_ctx_vector_t *ctx, const u32 *w, const int len); DECLSPEC void md5_update_vector_swap (md5_ctx_vector_t *ctx, const u32x *w, const int len); DECLSPEC void md5_update_vector_utf16le (md5_ctx_vector_t *ctx, const u32x *w, const int len); DECLSPEC void md5_update_vector_utf16le_swap (md5_ctx_vector_t *ctx, const u32x *w, const int len); diff --git a/OpenCL/inc_pem_common.cl b/OpenCL/inc_pem_common.cl deleted file mode 100644 index 9266b62cf..000000000 --- a/OpenCL/inc_pem_common.cl +++ /dev/null @@ -1,190 +0,0 @@ -/** - * Author......: See docs/credits.txt - * License.....: MIT - */ - -#include "inc_types.h" -#include "inc_vendor.h" -#include "inc_pem_common.h" - -#ifdef KERNEL_STATIC -#include "inc_hash_md5.cl" -#endif - -DECLSPEC void generate_key (u32 *salt_buf, u32 *pw, size_t pw_len, u32 *key) -{ - #ifdef DEBUG - printf("salt_buf:"); - for (u32 i = 0; i < 16; i++) printf(" 0x%08x", salt_buf[i]); - printf("\n"); - printf("pw:"); - for (u32 i = 0; i < 16; i++) printf(" 0x%08x", pw[i]); - printf("\n"); - printf("pw_len: %lu\n", pw_len); - #endif - - u32 md_buf[16] = { 0 }; - md5_ctx_t md_ctx; - - md5_init (&md_ctx); - md5_update (&md_ctx, pw, pw_len); - md5_update (&md_ctx, salt_buf, HC_PEM_SALT_LENGTH); - md5_final (&md_ctx); - - key[0] = md_ctx.h[0]; - - #if KEY_LENGTH > 4 - key[1] = md_ctx.h[1]; - #endif - - #if KEY_LENGTH > 8 - key[2] = md_ctx.h[2]; - #endif - - #if KEY_LENGTH > 12 - key[3] = md_ctx.h[3]; - #endif - - #if KEY_LENGTH > 16 - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < HC_PEM_MD_LENGTH / 4; i++) - { - md_buf[i] = md_ctx.h[i]; - } - - md5_init (&md_ctx); - md5_update (&md_ctx, md_buf, HC_PEM_MD_LENGTH); - md5_update (&md_ctx, pw, pw_len); - md5_update (&md_ctx, salt_buf, HC_PEM_SALT_LENGTH); - md5_final (&md_ctx); - - key[4] = md_ctx.h[0]; - #endif // KEY_LENGTH > 16 - - #if KEY_LENGTH > 20 - key[5] = md_ctx.h[1]; - #endif - - #if KEY_LENGTH > 24 - key[6] = md_ctx.h[2]; - #endif - - #if KEY_LENGTH > 28 - key[7] = md_ctx.h[3]; - #endif - - #if KEY_LENGTH > 32 - #error Only supports up to KEY_LENGTH == 32 at present. Extend generate_key! - #endif - - #ifdef DEBUG - printf("key:"); - for (u32 i = 0; i < KEY_LENGTH / 4; i++) printf(" 0x%08x", key[i]); - printf("\n"); - #endif // DEBUG -} - -DECLSPEC void generate_key_vector (u32 *salt_buf, u32x *pw, size_t pw_len, u32x *key) -{ - #ifdef DEBUG - printf("salt_buf:"); - for (u32 i = 0; i < 16; i++) printf(" 0x%08x", salt_buf[i]); - printf("\n"); - for (u32 v = 0; v < VECT_SIZE; v++) - { - printf("pw[%u]:", v); - for (u32 i = 0; i < 16; i++) printf(" 0x%08x", VECTOR_ELEMENT(pw[i], v)); - printf("\n"); - } - printf("pw_len: %lu\n", pw_len); - #endif - - u32x md_buf[16] = { 0 }; - md5_ctx_vector_t md_ctx; - - md5_init_vector (&md_ctx); - md5_update_vector (&md_ctx, pw, pw_len); - md5_update_vector_from_scalar (&md_ctx, salt_buf, HC_PEM_SALT_LENGTH); - md5_final_vector (&md_ctx); - - key[0] = md_ctx.h[0]; - - #if KEY_LENGTH > 4 - key[1] = md_ctx.h[1]; - #endif - - #if KEY_LENGTH > 8 - key[2] = md_ctx.h[2]; - #endif - - #if KEY_LENGTH > 12 - key[3] = md_ctx.h[3]; - #endif - - #if KEY_LENGTH > 16 - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < HC_PEM_MD_LENGTH / 4; i++) - { - md_buf[i] = md_ctx.h[i]; - } - - md5_init_vector (&md_ctx); - md5_update_vector (&md_ctx, md_buf, HC_PEM_MD_LENGTH); - md5_update_vector (&md_ctx, pw, pw_len); - md5_update_vector_from_scalar (&md_ctx, salt_buf, HC_PEM_SALT_LENGTH); - md5_final_vector (&md_ctx); - - key[4] = md_ctx.h[0]; - #endif // KEY_LENGTH > 16 - - #if KEY_LENGTH > 20 - key[5] = md_ctx.h[1]; - #endif - - #if KEY_LENGTH > 24 - key[6] = md_ctx.h[2]; - #endif - - #if KEY_LENGTH > 28 - key[7] = md_ctx.h[3]; - #endif - - #if KEY_LENGTH > 32 - #error Only supports up to KEY_LENGTH == 32 at present. Extend generate_key! - #endif - - #ifdef DEBUG - for (u32 v = 0; v < VECT_SIZE; v++) - { - printf("key[%u]:", v); - for (u32 i = 0; i < KEY_LENGTH / 4; i++) printf(" 0x%08x", VECTOR_ELEMENT(key[i], v)); - printf("\n"); - } - #endif // DEBUG -} - -DECLSPEC void prep_buffers(u32 *salt_buf, u32 *salt_iv, u32 *first_block, PSEUDO_SHM_TYPE u32 *data, GLOBAL_AS const pem_t *esalt) -{ - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < HC_PEM_SALT_LENGTH / 4; i++) - { - salt_buf[i] = esalt->salt_iv[i]; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - salt_iv[i] = esalt->salt_iv[i]; - first_block[i] = data[i]; - } -} diff --git a/OpenCL/inc_pem_common.h b/OpenCL/inc_pem_common.h deleted file mode 100644 index ddd189412..000000000 --- a/OpenCL/inc_pem_common.h +++ /dev/null @@ -1,35 +0,0 @@ -/** - * Author......: See docs/credits.txt - * License.....: MIT - */ - -#ifndef _INC_PEM_COMMON_H -#define _INC_PEM_COMMON_H - -#define HC_PEM_SALT_LENGTH 8 -#define HC_PEM_MD_LENGTH 16 - -#define HC_PEM_MAX_BLOCK_SIZE 16 -#define HC_PEM_MAX_KEY_LENGTH 32 -#define HC_PEM_MAX_DATA_LENGTH 12288 - -typedef struct pem -{ - void *chosen_cipher; - u32 salt_iv[HC_PEM_MAX_BLOCK_SIZE / 4]; - - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; - size_t data_len; -} pem_t; - -#ifdef REAL_SHM -#define PSEUDO_SHM_TYPE LOCAL_AS -#else -#define PSEUDO_SHM_TYPE -#endif - -DECLSPEC void generate_key (u32 *salt_buf, u32 *pw, size_t pw_len, u32 *key); -DECLSPEC void generate_key_vector (u32 *salt_buf, u32x *pw, size_t pw_len, u32x *key); -DECLSPEC void prep_buffers (u32 *salt_buf, u32 *salt_iv, u32 *first_block, PSEUDO_SHM_TYPE u32 *data, GLOBAL_AS const pem_t *esalt); - -#endif // _INC_PEM_COMMON_H diff --git a/OpenCL/m22911_a0-pure.cl b/OpenCL/m22911_a0-pure.cl index d084296e4..b5bbaf507 100644 --- a/OpenCL/m22911_a0-pure.cl +++ b/OpenCL/m22911_a0-pure.cl @@ -3,42 +3,242 @@ * License.....: MIT */ -#define BLOCK_SIZE 8 -#define KEY_LENGTH 24 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" +#include "inc_rp.h" #include "inc_rp.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_des.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22911_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22911_mxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); + #ifdef REAL_SHM + + LOCAL_VK u32 s_SPtrans[8][64]; + LOCAL_VK u32 s_skb[8][64]; + + for (u32 i = lid; i < 64; i += lsz) + { + s_SPtrans[0][i] = c_SPtrans[0][i]; + s_SPtrans[1][i] = c_SPtrans[1][i]; + s_SPtrans[2][i] = c_SPtrans[2][i]; + s_SPtrans[3][i] = c_SPtrans[3][i]; + s_SPtrans[4][i] = c_SPtrans[4][i]; + s_SPtrans[5][i] = c_SPtrans[5][i]; + s_SPtrans[6][i] = c_SPtrans[6][i]; + s_SPtrans[7][i] = c_SPtrans[7][i]; + + s_skb[0][i] = c_skb[0][i]; + s_skb[1][i] = c_skb[1][i]; + s_skb[2][i] = c_skb[2][i]; + s_skb[3][i] = c_skb[3][i]; + s_skb[4][i] = c_skb[4][i]; + s_skb[5][i] = c_skb[5][i]; + s_skb[6][i] = c_skb[6][i]; + s_skb[7][i] = c_skb[7][i]; + } + + SYNC_THREADS (); + + #else + + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; + + #endif + if (gid >= gid_max) return; - #ifdef REAL_SHM + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + COPY_PW (pws[gid]); + + u32 s[2]; - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + u32 first_data[2]; - for (u32 i = lid; i <= data_len / 4; i += lsz) + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - data[i] = esalt_bufs[digests_offset].data[i]; + pw_t tmp = PASTE_PW; + + tmp.pw_len = apply_rules (rules_buf[il_pos].cmds, tmp.i, tmp.pw_len); + + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, tmp.i, tmp.pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + u32 K2[16]; + u32 K3[16]; + u32 K4[16]; + u32 K5[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + _des_crypt_keysetup (ukey[2], ukey[3], K2, K3, s_skb); + _des_crypt_keysetup (ukey[4], ukey[5], K4, K5, s_skb); + + u32 dec[2]; + + // first check the padding + + u32 p1[2]; + u32 p2[2]; + + _des_crypt_decrypt (p1, enc, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (p1, first_data, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); } +} + +KERNEL_FQ void m22911_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ + const u64 gid = get_global_id (0); + const u64 lid = get_local_id (0); + const u64 lsz = get_local_size (0); + + #ifdef REAL_SHM LOCAL_VK u32 s_SPtrans[8][64]; LOCAL_VK u32 s_skb[8][64]; @@ -68,171 +268,172 @@ KERNEL_FQ void m22911_sxx (KERN_ATTR_RULES_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) - { - data[i] = esalt_bufs[digests_offset].data[i]; - } - CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; - CONSTANT_AS u32a (*s_skb)[64] = c_skb; + if (gid >= gid_max) return; - #endif // REAL_SHM + /** + * digest + */ - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + /** + * base + */ COPY_PW (pws[gid]); + u32 s[2]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + + u32 first_data[2]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + /** * loop */ for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; - pw_t tmp = PASTE_PW; tmp.pw_len = apply_rules (rules_buf[il_pos].cmds, tmp.i, tmp.pw_len); - generate_key (salt_buf, tmp.i, tmp.pw_len, key); - - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; - u32 K0[16], K1[16], K2[16], K3[16], K4[16], K5[16]; - - _des_crypt_keysetup (key[0], key[1], K0, K1, s_skb); - _des_crypt_keysetup (key[2], key[3], K2, K3, s_skb); - _des_crypt_keysetup (key[4], key[5], K4, K5, s_skb); - - u32 p1[BLOCK_SIZE / 4], p2[BLOCK_SIZE / 4]; - - _des_crypt_decrypt (p1, first_block, K4, K5, s_SPtrans); - _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); - _des_crypt_decrypt (plaintext, p2, K0, K1, s_SPtrans); - - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - _des_crypt_decrypt (p1, ciphertext, K4, K5, s_SPtrans); - _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); - _des_crypt_decrypt (plaintext, p2, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos, 0, 0); - } - } + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, tmp.i, tmp.pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + u32 K2[16]; + u32 K3[16]; + u32 K4[16]; + u32 K5[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + _des_crypt_keysetup (ukey[2], ukey[3], K2, K3, s_skb); + _des_crypt_keysetup (ukey[4], ukey[5], K4, K5, s_skb); + + u32 dec[2]; + + // first check the padding + + u32 p1[2]; + u32 p2[2]; + + _des_crypt_decrypt (p1, enc, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (p1, first_data, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22911_a1-pure.cl b/OpenCL/m22911_a1-pure.cl index e611b1ebc..2ff5468b0 100644 --- a/OpenCL/m22911_a1-pure.cl +++ b/OpenCL/m22911_a1-pure.cl @@ -3,41 +3,238 @@ * License.....: MIT */ -#define BLOCK_SIZE 8 -#define KEY_LENGTH 24 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_des.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22911_sxx (KERN_ATTR_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22911_mxx (KERN_ATTR_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); + #ifdef REAL_SHM + + LOCAL_VK u32 s_SPtrans[8][64]; + LOCAL_VK u32 s_skb[8][64]; + + for (u32 i = lid; i < 64; i += lsz) + { + s_SPtrans[0][i] = c_SPtrans[0][i]; + s_SPtrans[1][i] = c_SPtrans[1][i]; + s_SPtrans[2][i] = c_SPtrans[2][i]; + s_SPtrans[3][i] = c_SPtrans[3][i]; + s_SPtrans[4][i] = c_SPtrans[4][i]; + s_SPtrans[5][i] = c_SPtrans[5][i]; + s_SPtrans[6][i] = c_SPtrans[6][i]; + s_SPtrans[7][i] = c_SPtrans[7][i]; + + s_skb[0][i] = c_skb[0][i]; + s_skb[1][i] = c_skb[1][i]; + s_skb[2][i] = c_skb[2][i]; + s_skb[3][i] = c_skb[3][i]; + s_skb[4][i] = c_skb[4][i]; + s_skb[5][i] = c_skb[5][i]; + s_skb[6][i] = c_skb[6][i]; + s_skb[7][i] = c_skb[7][i]; + } + + SYNC_THREADS (); + + #else + + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; + + #endif + if (gid >= gid_max) return; - #ifdef REAL_SHM + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[2]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; + u32 first_data[2]; - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; - for (u32 i = lid; i <= data_len / 4; i += lsz) + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - data[i] = esalt_bufs[digests_offset].data[i]; + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + u32 K2[16]; + u32 K3[16]; + u32 K4[16]; + u32 K5[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + _des_crypt_keysetup (ukey[2], ukey[3], K2, K3, s_skb); + _des_crypt_keysetup (ukey[4], ukey[5], K4, K5, s_skb); + + u32 dec[2]; + + // first check the padding + + u32 p1[2]; + u32 p2[2]; + + _des_crypt_decrypt (p1, enc, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (p1, first_data, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); } +} + +KERNEL_FQ void m22911_sxx (KERN_ATTR_ESALT (pem_t)) +{ + const u64 gid = get_global_id (0); + const u64 lid = get_local_id (0); + const u64 lsz = get_local_size (0); + + #ifdef REAL_SHM LOCAL_VK u32 s_SPtrans[8][64]; LOCAL_VK u32 s_skb[8][64]; @@ -67,35 +264,54 @@ KERNEL_FQ void m22911_sxx (KERN_ATTR_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = { - data[i] = esalt_bufs[digests_offset].data[i]; - } + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; - CONSTANT_AS u32a (*s_skb)[64] = c_skb; + /** + * base + */ - #endif // REAL_SHM + u32 s[2]; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + u32 first_data[2]; - const u32 pw_len = pws[gid].pw_len; + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; - u32 w[16] = { 0 }; + const int data_len = esalt_bufs[digests_offset].data_len; - for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) - { - w[idx] = pws[gid].i[idx]; - } + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; /** * loop @@ -103,159 +319,115 @@ KERNEL_FQ void m22911_sxx (KERN_ATTR_ESALT (pem_t)) for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - const u32 comb_len = combs_buf[il_pos].pw_len; - u32 c[64]; - - #ifdef _unroll - #pragma unroll - #endif - for (int i = 0; i < 16; i++) - { - c[i] = combs_buf[il_pos].i[i]; - } - - switch_buffer_by_offset_1x64_be_S (c, pw_len); - - #ifdef _unroll - #pragma unroll - #endif - for (int i = 0; i < 16; i++) - { - c[i] |= w[i]; - } - - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; - - generate_key (salt_buf, c, pw_len + comb_len, key); - - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; - u32 K0[16], K1[16], K2[16], K3[16], K4[16], K5[16]; - - _des_crypt_keysetup (key[0], key[1], K0, K1, s_skb); - _des_crypt_keysetup (key[2], key[3], K2, K3, s_skb); - _des_crypt_keysetup (key[4], key[5], K4, K5, s_skb); - - u32 p1[BLOCK_SIZE / 4], p2[BLOCK_SIZE / 4]; - - _des_crypt_decrypt (p1, first_block, K4, K5, s_SPtrans); - _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); - _des_crypt_decrypt (plaintext, p2, K0, K1, s_SPtrans); - - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - _des_crypt_decrypt (p1, ciphertext, K4, K5, s_SPtrans); - _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); - _des_crypt_decrypt (plaintext, p2, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos, 0, 0); - } - } + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + u32 K2[16]; + u32 K3[16]; + u32 K4[16]; + u32 K5[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + _des_crypt_keysetup (ukey[2], ukey[3], K2, K3, s_skb); + _des_crypt_keysetup (ukey[4], ukey[5], K4, K5, s_skb); + + u32 dec[2]; + + // first check the padding + + u32 p1[2]; + u32 p2[2]; + + _des_crypt_decrypt (p1, enc, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (p1, first_data, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22911_a3-pure.cl b/OpenCL/m22911_a3-pure.cl index 585c96b15..594e9805f 100644 --- a/OpenCL/m22911_a3-pure.cl +++ b/OpenCL/m22911_a3-pure.cl @@ -3,44 +3,255 @@ * License.....: MIT */ -#define NEW_SIMD_CODE - -#define BLOCK_SIZE 8 -#define KEY_LENGTH 24 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" -#include "inc_simd.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_des.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22911_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22911_mxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); + #ifdef REAL_SHM + + LOCAL_VK u32 s_SPtrans[8][64]; + LOCAL_VK u32 s_skb[8][64]; + + for (u32 i = lid; i < 64; i += lsz) + { + s_SPtrans[0][i] = c_SPtrans[0][i]; + s_SPtrans[1][i] = c_SPtrans[1][i]; + s_SPtrans[2][i] = c_SPtrans[2][i]; + s_SPtrans[3][i] = c_SPtrans[3][i]; + s_SPtrans[4][i] = c_SPtrans[4][i]; + s_SPtrans[5][i] = c_SPtrans[5][i]; + s_SPtrans[6][i] = c_SPtrans[6][i]; + s_SPtrans[7][i] = c_SPtrans[7][i]; + + s_skb[0][i] = c_skb[0][i]; + s_skb[1][i] = c_skb[1][i]; + s_skb[2][i] = c_skb[2][i]; + s_skb[3][i] = c_skb[3][i]; + s_skb[4][i] = c_skb[4][i]; + s_skb[5][i] = c_skb[5][i]; + s_skb[6][i] = c_skb[6][i]; + s_skb[7][i] = c_skb[7][i]; + } + + SYNC_THREADS (); + + #else + + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; + + #endif + if (gid >= gid_max) return; - #ifdef REAL_SHM + /** + * digest + */ - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + /** + * base + */ + + u32 s[2]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + + u32 first_data[2]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; + + for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) + { + w[idx] = pws[gid].i[idx]; + } - for (u32 i = lid; i <= data_len / 4; i += lsz) + /** + * loop + */ + + u32x w0l = w[0]; + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { - data[i] = esalt_bufs[digests_offset].data[i]; + const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + + const u32x w0 = w0l | w0r; + + w[0] = w0; + + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, w, pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + u32 K2[16]; + u32 K3[16]; + u32 K4[16]; + u32 K5[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + _des_crypt_keysetup (ukey[2], ukey[3], K2, K3, s_skb); + _des_crypt_keysetup (ukey[4], ukey[5], K4, K5, s_skb); + + u32 dec[2]; + + // first check the padding + + u32 p1[2]; + u32 p2[2]; + + _des_crypt_decrypt (p1, enc, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (p1, first_data, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); } +} + +KERNEL_FQ void m22911_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ + const u64 gid = get_global_id (0); + const u64 lid = get_local_id (0); + const u64 lsz = get_local_size (0); + + #ifdef REAL_SHM LOCAL_VK u32 s_SPtrans[8][64]; LOCAL_VK u32 s_skb[8][64]; @@ -70,184 +281,187 @@ KERNEL_FQ void m22911_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = { - data[i] = esalt_bufs[digests_offset].data[i]; - } + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; - CONSTANT_AS u32a (*s_skb)[64] = c_skb; + /** + * base + */ - #endif // REAL_SHM + u32 s[2]; - const u32 pw_len = pws[gid].pw_len; + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + + u32 first_data[2]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + const int last_pad_pos = data_len - 1; - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + const int last_pad_elem = last_pad_pos / 4; - u32x w[16] = { 0 }; + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) { w[idx] = pws[gid].i[idx]; } - u32x w0l = w[0]; - /** * loop */ + u32x w0l = w[0]; + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + const u32x w0 = w0l | w0r; w[0] = w0; - u32x key[HC_PEM_MAX_KEY_LENGTH / 4]; - - generate_key_vector (salt_buf, w, pw_len, key); - - for (u32 v_pos = 0; v_pos < VECT_SIZE; v_pos++) - { - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; - u32 K0[16], K1[16], K2[16], K3[16], K4[16], K5[16]; - - _des_crypt_keysetup (VECTOR_ELEMENT(key[0], v_pos), VECTOR_ELEMENT(key[1], v_pos), K0, K1, s_skb); - _des_crypt_keysetup (VECTOR_ELEMENT(key[2], v_pos), VECTOR_ELEMENT(key[3], v_pos), K2, K3, s_skb); - _des_crypt_keysetup (VECTOR_ELEMENT(key[4], v_pos), VECTOR_ELEMENT(key[5], v_pos), K4, K5, s_skb); - - u32 p1[BLOCK_SIZE / 4], p2[BLOCK_SIZE / 4]; - - _des_crypt_decrypt (p1, first_block, K4, K5, s_SPtrans); - _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); - _des_crypt_decrypt (plaintext, p2, K0, K1, s_SPtrans); - - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - _des_crypt_decrypt (p1, ciphertext, K4, K5, s_SPtrans); - _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); - _des_crypt_decrypt (plaintext, p2, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos + v_pos, 0, 0); - } - } - } + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, w, pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + u32 K2[16]; + u32 K3[16]; + u32 K4[16]; + u32 K5[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + _des_crypt_keysetup (ukey[2], ukey[3], K2, K3, s_skb); + _des_crypt_keysetup (ukey[4], ukey[5], K4, K5, s_skb); + + u32 dec[2]; + + // first check the padding + + u32 p1[2]; + u32 p2[2]; + + _des_crypt_decrypt (p1, enc, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (p1, first_data, K4, K5, s_SPtrans); + _des_crypt_encrypt (p2, p1, K2, K3, s_SPtrans); + _des_crypt_decrypt (dec, p2, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22921_a0-pure.cl b/OpenCL/m22921_a0-pure.cl index 3ded312d8..e3de223ae 100644 --- a/OpenCL/m22921_a0-pure.cl +++ b/OpenCL/m22921_a0-pure.cl @@ -3,42 +3,209 @@ * License.....: MIT */ -#define BLOCK_SIZE 8 -#define KEY_LENGTH 8 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" +#include "inc_rp.h" #include "inc_rp.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_des.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22921_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; +} pem_t; + +KERNEL_FQ void m22921_mxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); + #ifdef REAL_SHM + + LOCAL_VK u32 s_SPtrans[8][64]; + LOCAL_VK u32 s_skb[8][64]; + + for (u32 i = lid; i < 64; i += lsz) + { + s_SPtrans[0][i] = c_SPtrans[0][i]; + s_SPtrans[1][i] = c_SPtrans[1][i]; + s_SPtrans[2][i] = c_SPtrans[2][i]; + s_SPtrans[3][i] = c_SPtrans[3][i]; + s_SPtrans[4][i] = c_SPtrans[4][i]; + s_SPtrans[5][i] = c_SPtrans[5][i]; + s_SPtrans[6][i] = c_SPtrans[6][i]; + s_SPtrans[7][i] = c_SPtrans[7][i]; + + s_skb[0][i] = c_skb[0][i]; + s_skb[1][i] = c_skb[1][i]; + s_skb[2][i] = c_skb[2][i]; + s_skb[3][i] = c_skb[3][i]; + s_skb[4][i] = c_skb[4][i]; + s_skb[5][i] = c_skb[5][i]; + s_skb[6][i] = c_skb[6][i]; + s_skb[7][i] = c_skb[7][i]; + } + + SYNC_THREADS (); + + #else + + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; + + #endif + if (gid >= gid_max) return; - #ifdef REAL_SHM + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + COPY_PW (pws[gid]); + + u32 s[2]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + + u32 first_data[2]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; + const int last_pad_pos = data_len - 1; - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + const int last_pad_elem = last_pad_pos / 4; - for (u32 i = lid; i <= data_len / 4; i += lsz) + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - data[i] = esalt_bufs[digests_offset].data[i]; + pw_t tmp = PASTE_PW; + + tmp.pw_len = apply_rules (rules_buf[il_pos].cmds, tmp.i, tmp.pw_len); + + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[2]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + + u32 dec[2]; + + // first check the padding + + _des_crypt_decrypt (dec, enc, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (dec, first_data, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); } +} + +KERNEL_FQ void m22921_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ + const u64 gid = get_global_id (0); + const u64 lid = get_local_id (0); + const u64 lsz = get_local_size (0); + + #ifdef REAL_SHM LOCAL_VK u32 s_SPtrans[8][64]; LOCAL_VK u32 s_skb[8][64]; @@ -68,28 +235,56 @@ KERNEL_FQ void m22921_sxx (KERN_ATTR_RULES_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = { - data[i] = esalt_bufs[digests_offset].data[i]; - } + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; - CONSTANT_AS u32a (*s_skb)[64] = c_skb; + /** + * base + */ + + COPY_PW (pws[gid]); + + u32 s[2]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + + u32 first_data[2]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; - #endif // REAL_SHM + const int last_pad_pos = data_len - 1; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + const int last_pad_elem = last_pad_pos / 4; - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + u32 iv[2]; - COPY_PW(pws[gid]); + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; /** * loop @@ -97,133 +292,82 @@ KERNEL_FQ void m22921_sxx (KERN_ATTR_RULES_ESALT (pem_t)) for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; - pw_t tmp = PASTE_PW; tmp.pw_len = apply_rules (rules_buf[il_pos].cmds, tmp.i, tmp.pw_len); - generate_key (salt_buf, tmp.i, tmp.pw_len, key); - - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; - u32 K0[16], K1[16]; - - _des_crypt_keysetup (key[0], key[1], K0, K1, s_skb); - - _des_crypt_decrypt (plaintext, first_block, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - _des_crypt_decrypt (plaintext, ciphertext, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos, 0, 0); - } - } + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[2]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + + u32 dec[2]; + + // first check the padding + + _des_crypt_decrypt (dec, enc, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (dec, first_data, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22921_a1-pure.cl b/OpenCL/m22921_a1-pure.cl index ca5358595..d91767287 100644 --- a/OpenCL/m22921_a1-pure.cl +++ b/OpenCL/m22921_a1-pure.cl @@ -3,41 +3,203 @@ * License.....: MIT */ -#define BLOCK_SIZE 8 -#define KEY_LENGTH 8 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_des.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22921_sxx (KERN_ATTR_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; +} pem_t; + +KERNEL_FQ void m22921_mxx (KERN_ATTR_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); + #ifdef REAL_SHM + + LOCAL_VK u32 s_SPtrans[8][64]; + LOCAL_VK u32 s_skb[8][64]; + + for (u32 i = lid; i < 64; i += lsz) + { + s_SPtrans[0][i] = c_SPtrans[0][i]; + s_SPtrans[1][i] = c_SPtrans[1][i]; + s_SPtrans[2][i] = c_SPtrans[2][i]; + s_SPtrans[3][i] = c_SPtrans[3][i]; + s_SPtrans[4][i] = c_SPtrans[4][i]; + s_SPtrans[5][i] = c_SPtrans[5][i]; + s_SPtrans[6][i] = c_SPtrans[6][i]; + s_SPtrans[7][i] = c_SPtrans[7][i]; + + s_skb[0][i] = c_skb[0][i]; + s_skb[1][i] = c_skb[1][i]; + s_skb[2][i] = c_skb[2][i]; + s_skb[3][i] = c_skb[3][i]; + s_skb[4][i] = c_skb[4][i]; + s_skb[5][i] = c_skb[5][i]; + s_skb[6][i] = c_skb[6][i]; + s_skb[7][i] = c_skb[7][i]; + } + + SYNC_THREADS (); + + #else + + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; + + #endif + if (gid >= gid_max) return; - #ifdef REAL_SHM + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[2]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + + u32 first_data[2]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; + const int last_pad_pos = data_len - 1; - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + const int last_pad_elem = last_pad_pos / 4; - for (u32 i = lid; i <= data_len / 4; i += lsz) + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - data[i] = esalt_bufs[digests_offset].data[i]; + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[2]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + + u32 dec[2]; + + // first check the padding + + _des_crypt_decrypt (dec, enc, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (dec, first_data, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); } +} + +KERNEL_FQ void m22921_sxx (KERN_ATTR_ESALT (pem_t)) +{ + const u64 gid = get_global_id (0); + const u64 lid = get_local_id (0); + const u64 lsz = get_local_size (0); + + #ifdef REAL_SHM LOCAL_VK u32 s_SPtrans[8][64]; LOCAL_VK u32 s_skb[8][64]; @@ -67,35 +229,54 @@ KERNEL_FQ void m22921_sxx (KERN_ATTR_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = { - data[i] = esalt_bufs[digests_offset].data[i]; - } + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; - CONSTANT_AS u32a (*s_skb)[64] = c_skb; + /** + * base + */ - #endif // REAL_SHM + u32 s[2]; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + u32 first_data[2]; - const u32 pw_len = pws[gid].pw_len; + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; - u32 w[16] = { 0 }; + const int data_len = esalt_bufs[digests_offset].data_len; - for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) - { - w[idx] = pws[gid].i[idx]; - } + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; /** * loop @@ -103,150 +284,80 @@ KERNEL_FQ void m22921_sxx (KERN_ATTR_ESALT (pem_t)) for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - const u32 comb_len = combs_buf[il_pos].pw_len; - u32 c[64]; - - #ifdef _unroll - #pragma unroll - #endif - for (int i = 0; i < 16; i++) - { - c[i] = combs_buf[il_pos].i[i]; - } - - switch_buffer_by_offset_1x64_be_S (c, pw_len); - - #ifdef _unroll - #pragma unroll - #endif - for (int i = 0; i < 16; i++) - { - c[i] |= w[i]; - } - - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; - - generate_key (salt_buf, c, pw_len + comb_len, key); - - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; - u32 K0[16], K1[16]; - - _des_crypt_keysetup (key[0], key[1], K0, K1, s_skb); - - _des_crypt_decrypt (plaintext, first_block, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - _des_crypt_decrypt (plaintext, ciphertext, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos, 0, 0); - } - } + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[2]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + + u32 dec[2]; + + // first check the padding + + _des_crypt_decrypt (dec, enc, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (dec, first_data, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22921_a3-pure.cl b/OpenCL/m22921_a3-pure.cl index 85b81d7e9..8eb15d0a1 100644 --- a/OpenCL/m22921_a3-pure.cl +++ b/OpenCL/m22921_a3-pure.cl @@ -3,45 +3,224 @@ * License.....: MIT */ -#define NEW_SIMD_CODE - -#define BLOCK_SIZE 8 -#define KEY_LENGTH 8 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" -#include "inc_simd.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_des.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22921_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; +} pem_t; + +KERNEL_FQ void m22921_mxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); + #ifdef REAL_SHM + + LOCAL_VK u32 s_SPtrans[8][64]; + LOCAL_VK u32 s_skb[8][64]; + + for (u32 i = lid; i < 64; i += lsz) + { + s_SPtrans[0][i] = c_SPtrans[0][i]; + s_SPtrans[1][i] = c_SPtrans[1][i]; + s_SPtrans[2][i] = c_SPtrans[2][i]; + s_SPtrans[3][i] = c_SPtrans[3][i]; + s_SPtrans[4][i] = c_SPtrans[4][i]; + s_SPtrans[5][i] = c_SPtrans[5][i]; + s_SPtrans[6][i] = c_SPtrans[6][i]; + s_SPtrans[7][i] = c_SPtrans[7][i]; + + s_skb[0][i] = c_skb[0][i]; + s_skb[1][i] = c_skb[1][i]; + s_skb[2][i] = c_skb[2][i]; + s_skb[3][i] = c_skb[3][i]; + s_skb[4][i] = c_skb[4][i]; + s_skb[5][i] = c_skb[5][i]; + s_skb[6][i] = c_skb[6][i]; + s_skb[7][i] = c_skb[7][i]; + } + + SYNC_THREADS (); + + #else + + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; + + #endif + if (gid >= gid_max) return; - #ifdef REAL_SHM + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[2]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + + u32 first_data[2]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + u32x w[64] = { 0 }; - for (u32 i = lid; i <= data_len / 4; i += lsz) + for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) { - data[i] = esalt_bufs[digests_offset].data[i]; + w[idx] = pws[gid].i[idx]; } + /** + * loop + */ + + u32x w0l = w[0]; + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) + { + const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + + const u32x w0 = w0l | w0r; + + w[0] = w0; + + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[2]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + + u32 dec[2]; + + // first check the padding + + _des_crypt_decrypt (dec, enc, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (dec, first_data, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + + +KERNEL_FQ void m22921_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ + const u64 gid = get_global_id (0); + const u64 lid = get_local_id (0); + const u64 lsz = get_local_size (0); + + #ifdef REAL_SHM + LOCAL_VK u32 s_SPtrans[8][64]; LOCAL_VK u32 s_skb[8][64]; @@ -70,175 +249,154 @@ KERNEL_FQ void m22921_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; + CONSTANT_AS u32a (*s_skb)[64] = c_skb; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = { - data[i] = esalt_bufs[digests_offset].data[i]; - } + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - CONSTANT_AS u32a (*s_SPtrans)[64] = c_SPtrans; - CONSTANT_AS u32a (*s_skb)[64] = c_skb; + /** + * base + */ - #endif // REAL_SHM + u32 s[2]; - const u32 pw_len = pws[gid].pw_len; + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + + u32 first_data[2]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + u32 iv[2]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + + u32 enc[2]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + const u32 pw_len = pws[gid].pw_len; - u32x w[16] = { 0 }; + u32x w[64] = { 0 }; for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) { w[idx] = pws[gid].i[idx]; } - u32x w0l = w[0]; - /** * loop */ + u32x w0l = w[0]; + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + const u32x w0 = w0l | w0r; w[0] = w0; - u32x key[HC_PEM_MAX_KEY_LENGTH / 4]; - - generate_key_vector (salt_buf, w, pw_len, key); - - for (u32 v_pos = 0; v_pos < VECT_SIZE; v_pos++) - { - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; - u32 K0[16], K1[16]; - - _des_crypt_keysetup (VECTOR_ELEMENT(key[0], v_pos), VECTOR_ELEMENT(key[1], v_pos), K0, K1, s_skb); - - _des_crypt_decrypt (plaintext, first_block, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - _des_crypt_decrypt (plaintext, ciphertext, K0, K1, s_SPtrans); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos + v_pos, 0, 0); - } - } - } + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[2]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + + // DES + + u32 K0[16]; + u32 K1[16]; + + _des_crypt_keysetup (ukey[0], ukey[1], K0, K1, s_skb); + + u32 dec[2]; + + // first check the padding + + _des_crypt_decrypt (dec, enc, K0, K1, s_SPtrans); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + + const int paddingv = pkcs_padding_bs8 (dec, 8); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + _des_crypt_decrypt (dec, first_data, K0, K1, s_SPtrans); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + + const int real_len = (data_len - 8) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22931_a0-pure.cl b/OpenCL/m22931_a0-pure.cl index 816e82d67..62aaf84d4 100644 --- a/OpenCL/m22931_a0-pure.cl +++ b/OpenCL/m22931_a0-pure.cl @@ -3,43 +3,41 @@ * License.....: MIT */ -#define BLOCK_SIZE 16 -#define KEY_LENGTH 16 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" +#include "inc_rp.h" #include "inc_rp.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_aes.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22931_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22931_mxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); - if (gid >= gid_max) return; + /** + * aes shared + */ #ifdef REAL_SHM - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; - - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; - - for (u32 i = lid; i <= data_len / 4; i += lsz) - { - data[i] = esalt_bufs[digests_offset].data[i]; - } - LOCAL_VK u32 s_td0[256]; LOCAL_VK u32 s_td1[256]; LOCAL_VK u32 s_td2[256]; @@ -71,17 +69,213 @@ KERNEL_FQ void m22931_sxx (KERN_ATTR_RULES_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + 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; + + 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; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + COPY_PW (pws[gid]); + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * 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 ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[4]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + + u32 ks[44]; + + AES128_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes128_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes128_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m22931_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) { - data[i] = esalt_bufs[digests_offset].data[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]; + + 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]; } + SYNC_THREADS (); + + #else + CONSTANT_AS u32a *s_td0 = td0; CONSTANT_AS u32a *s_td1 = td1; CONSTANT_AS u32a *s_td2 = td2; @@ -94,148 +288,154 @@ KERNEL_FQ void m22931_sxx (KERN_ATTR_RULES_ESALT (pem_t)) CONSTANT_AS u32a *s_te3 = te3; CONSTANT_AS u32a *s_te4 = te4; - #endif // REAL_SHM + #endif + + if (gid >= gid_max) return; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + /** + * digest + */ - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ COPY_PW (pws[gid]); + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + /** * loop */ for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; - pw_t tmp = PASTE_PW; tmp.pw_len = apply_rules (rules_buf[il_pos].cmds, tmp.i, tmp.pw_len); - generate_key (salt_buf, tmp.i, tmp.pw_len, key); + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[4]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; u32 ks[44]; - aes128_set_decrypt_key (ks, key, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); - - aes128_decrypt (ks, first_block, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - aes128_decrypt (ks, ciphertext, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos, 0, 0); - } - } + AES128_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes128_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes128_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22931_a1-pure.cl b/OpenCL/m22931_a1-pure.cl index 9d0072da3..24810cd5f 100644 --- a/OpenCL/m22931_a1-pure.cl +++ b/OpenCL/m22931_a1-pure.cl @@ -3,42 +3,39 @@ * License.....: MIT */ -#define BLOCK_SIZE 16 -#define KEY_LENGTH 16 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_aes.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22931_sxx (KERN_ATTR_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22931_mxx (KERN_ATTR_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); - if (gid >= gid_max) return; + /** + * aes shared + */ #ifdef REAL_SHM - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; - - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; - - for (u32 i = lid; i <= data_len / 4; i += lsz) - { - data[i] = esalt_bufs[digests_offset].data[i]; - } - LOCAL_VK u32 s_td0[256]; LOCAL_VK u32 s_td1[256]; LOCAL_VK u32 s_td2[256]; @@ -70,16 +67,208 @@ KERNEL_FQ void m22931_sxx (KERN_ATTR_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + 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; + + 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; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = { - data[i] = esalt_bufs[digests_offset].data[i]; + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) + { + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[4]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + + u32 ks[44]; + + AES128_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes128_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes128_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); } +} + +KERNEL_FQ void m22931_sxx (KERN_ATTR_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) + { + 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]; + + 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]; + } + + SYNC_THREADS (); + + #else CONSTANT_AS u32a *s_td0 = td0; CONSTANT_AS u32a *s_td1 = td1; @@ -93,21 +282,59 @@ KERNEL_FQ void m22931_sxx (KERN_ATTR_ESALT (pem_t)) CONSTANT_AS u32a *s_te3 = te3; CONSTANT_AS u32a *s_te4 = te4; - #endif // REAL_SHM + #endif - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + if (gid >= gid_max) return; - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + /** + * digest + */ - const u32 pw_len = pws[gid].pw_len; + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - u32 w[16] = { 0 }; + /** + * base + */ - for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) - { - w[idx] = pws[gid].i[idx]; - } + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; /** * loop @@ -115,150 +342,90 @@ KERNEL_FQ void m22931_sxx (KERN_ATTR_ESALT (pem_t)) for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - const u32 comb_len = combs_buf[il_pos].pw_len; - u32 c[64]; + md5_ctx_t ctx; - #ifdef _unroll - #pragma unroll - #endif - for (int i = 0; i < 16; i++) - { - c[i] = combs_buf[il_pos].i[i]; - } + md5_init (&ctx); - switch_buffer_by_offset_1x64_be_S (c, pw_len); + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); - #ifdef _unroll - #pragma unroll - #endif - for (int i = 0; i < 16; i++) - { - c[i] |= w[i]; - } + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; + u32 t[16]; - generate_key (salt_buf, c, pw_len + comb_len, key); + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[4]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; u32 ks[44]; - aes128_set_decrypt_key (ks, key, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); - - aes128_decrypt (ks, first_block, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - aes128_decrypt (ks, ciphertext, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos, 0, 0); - } - } + AES128_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes128_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes128_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22931_a3-pure.cl b/OpenCL/m22931_a3-pure.cl index fd9e4c438..1ce87a47b 100644 --- a/OpenCL/m22931_a3-pure.cl +++ b/OpenCL/m22931_a3-pure.cl @@ -3,45 +3,39 @@ * License.....: MIT */ -#define NEW_SIMD_CODE - -#define BLOCK_SIZE 16 -#define KEY_LENGTH 16 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" -#include "inc_simd.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_aes.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22931_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22931_mxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); - if (gid >= gid_max) return; + /** + * aes shared + */ #ifdef REAL_SHM - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; - - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; - - for (u32 i = lid; i <= data_len / 4; i += lsz) - { - data[i] = esalt_bufs[digests_offset].data[i]; - } - LOCAL_VK u32 s_td0[256]; LOCAL_VK u32 s_td1[256]; LOCAL_VK u32 s_td2[256]; @@ -73,17 +67,228 @@ KERNEL_FQ void m22931_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + 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; + + 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; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; + + for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) + { + w[idx] = pws[gid].i[idx]; + } + + /** + * loop + */ + + u32x w0l = w[0]; + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) + { + const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + + const u32x w0 = w0l | w0r; + + w[0] = w0; + + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[4]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + + u32 ks[44]; + + AES128_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes128_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes128_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m22931_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) { - data[i] = esalt_bufs[digests_offset].data[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]; + + 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]; } + SYNC_THREADS (); + + #else + CONSTANT_AS u32a *s_td0 = td0; CONSTANT_AS u32a *s_td1 = td1; CONSTANT_AS u32a *s_td2 = td2; @@ -96,170 +301,169 @@ KERNEL_FQ void m22931_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) CONSTANT_AS u32a *s_te3 = te3; CONSTANT_AS u32a *s_te4 = te4; - #endif // REAL_SHM + #endif - const u32 pw_len = pws[gid].pw_len; + if (gid >= gid_max) return; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + /** + * digest + */ - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; - u32x w[16] = { 0 }; + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) { w[idx] = pws[gid].i[idx]; } - u32x w0l = w[0]; - /** * loop */ + u32x w0l = w[0]; + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + const u32x w0 = w0l | w0r; w[0] = w0; - u32x keys[KEY_LENGTH / 4]; - - generate_key_vector (salt_buf, w, pw_len, keys); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 v_pos = 0; v_pos < VECT_SIZE; v_pos++) - { - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; - u32 ks[44]; - u32 key[KEY_LENGTH / 4]; - - for (u32 i = 0; i < KEY_LENGTH; i++) - { - key[i] = VECTOR_ELEMENT(keys[i], v_pos); - } - - aes128_set_decrypt_key (ks, key, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); - - aes128_decrypt (ks, first_block, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - aes128_decrypt (ks, ciphertext, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos + v_pos, 0, 0); - } - } - } + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[4]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + + u32 ks[44]; + + AES128_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes128_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes128_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22941_a0-pure.cl b/OpenCL/m22941_a0-pure.cl new file mode 100644 index 000000000..50e2d7265 --- /dev/null +++ b/OpenCL/m22941_a0-pure.cl @@ -0,0 +1,481 @@ +/** + * Author......: See docs/credits.txt + * License.....: MIT + */ + +//#define NEW_SIMD_CODE + +#ifdef KERNEL_STATIC +#include "inc_vendor.h" +#include "inc_types.h" +#include "inc_platform.cl" +#include "inc_common.cl" +#include "inc_rp.h" +#include "inc_rp.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" +#include "inc_cipher_aes.cl" +#endif + +typedef struct pem +{ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; + +KERNEL_FQ void m22941_mxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) + { + 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]; + + 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]; + } + + SYNC_THREADS (); + + #else + + 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; + + 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; + + #endif + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + COPY_PW (pws[gid]); + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * 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 ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, tmp.i, tmp.pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + + u32 ks[52]; + + AES192_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes192_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes192_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m22941_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) + { + 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]; + + 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]; + } + + SYNC_THREADS (); + + #else + + 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; + + 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; + + #endif + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + COPY_PW (pws[gid]); + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * 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 ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, tmp.i, tmp.pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + + u32 ks[52]; + + AES192_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes192_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes192_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); + } +} diff --git a/OpenCL/m22941_a1-pure.cl b/OpenCL/m22941_a1-pure.cl new file mode 100644 index 000000000..a4e4ab7d1 --- /dev/null +++ b/OpenCL/m22941_a1-pure.cl @@ -0,0 +1,475 @@ +/** + * Author......: See docs/credits.txt + * License.....: MIT + */ + +//#define NEW_SIMD_CODE + +#ifdef KERNEL_STATIC +#include "inc_vendor.h" +#include "inc_types.h" +#include "inc_platform.cl" +#include "inc_common.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" +#include "inc_cipher_aes.cl" +#endif + +typedef struct pem +{ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; + +KERNEL_FQ void m22941_mxx (KERN_ATTR_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) + { + 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]; + + 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]; + } + + SYNC_THREADS (); + + #else + + 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; + + 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; + + #endif + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) + { + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + + u32 ks[52]; + + AES192_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes192_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes192_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m22941_sxx (KERN_ATTR_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) + { + 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]; + + 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]; + } + + SYNC_THREADS (); + + #else + + 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; + + 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; + + #endif + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) + { + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + + u32 ks[52]; + + AES192_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes192_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes192_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); + } +} diff --git a/OpenCL/m22941_a3-pure.cl b/OpenCL/m22941_a3-pure.cl new file mode 100644 index 000000000..08d14a35d --- /dev/null +++ b/OpenCL/m22941_a3-pure.cl @@ -0,0 +1,509 @@ +/** + * Author......: See docs/credits.txt + * License.....: MIT + */ + +//#define NEW_SIMD_CODE + +#ifdef KERNEL_STATIC +#include "inc_vendor.h" +#include "inc_types.h" +#include "inc_platform.cl" +#include "inc_common.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" +#include "inc_cipher_aes.cl" +#endif + +typedef struct pem +{ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; + +KERNEL_FQ void m22941_mxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) + { + 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]; + + 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]; + } + + SYNC_THREADS (); + + #else + + 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; + + 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; + + #endif + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; + + for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) + { + w[idx] = pws[gid].i[idx]; + } + + /** + * loop + */ + + u32x w0l = w[0]; + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) + { + const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + + const u32x w0 = w0l | w0r; + + w[0] = w0; + + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, w, pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + + u32 ks[52]; + + AES192_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes192_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes192_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m22941_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) + { + 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]; + + 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]; + } + + SYNC_THREADS (); + + #else + + 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; + + 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; + + #endif + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; + + for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) + { + w[idx] = pws[gid].i[idx]; + } + + /** + * loop + */ + + u32x w0l = w[0]; + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) + { + const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + + const u32x w0 = w0l | w0r; + + w[0] = w0; + + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[6]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, w, pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + + u32 ks[52]; + + AES192_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes192_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes192_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); + } +} diff --git a/OpenCL/m22951_a0-pure.cl b/OpenCL/m22951_a0-pure.cl index 3f53e4edb..38427e991 100644 --- a/OpenCL/m22951_a0-pure.cl +++ b/OpenCL/m22951_a0-pure.cl @@ -3,43 +3,41 @@ * License.....: MIT */ -#define BLOCK_SIZE 16 -#define KEY_LENGTH 32 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" +#include "inc_rp.h" #include "inc_rp.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_aes.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22951_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22951_mxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); - if (gid >= gid_max) return; + /** + * aes shared + */ #ifdef REAL_SHM - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; - - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; - - for (u32 i = lid; i <= data_len / 4; i += lsz) - { - data[i] = esalt_bufs[digests_offset].data[i]; - } - LOCAL_VK u32 s_td0[256]; LOCAL_VK u32 s_td1[256]; LOCAL_VK u32 s_td2[256]; @@ -71,17 +69,237 @@ KERNEL_FQ void m22951_sxx (KERN_ATTR_RULES_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + 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; + + 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; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + COPY_PW (pws[gid]); + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * 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 ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[8]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, tmp.i, tmp.pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + ukey[6] = ctx.h[2]; + ukey[7] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + ukey[6] = hc_swap32_S (ukey[6]); + ukey[7] = hc_swap32_S (ukey[7]); + + u32 ks[60]; + + AES256_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes256_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes256_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m22951_sxx (KERN_ATTR_RULES_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) { - data[i] = esalt_bufs[digests_offset].data[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]; + + 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]; } + SYNC_THREADS (); + + #else + CONSTANT_AS u32a *s_td0 = td0; CONSTANT_AS u32a *s_td1 = td1; CONSTANT_AS u32a *s_td2 = td2; @@ -94,148 +312,178 @@ KERNEL_FQ void m22951_sxx (KERN_ATTR_RULES_ESALT (pem_t)) CONSTANT_AS u32a *s_te3 = te3; CONSTANT_AS u32a *s_te4 = te4; - #endif // REAL_SHM + #endif + + if (gid >= gid_max) return; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + /** + * digest + */ - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ COPY_PW (pws[gid]); + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + /** * loop */ for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; - pw_t tmp = PASTE_PW; tmp.pw_len = apply_rules (rules_buf[il_pos].cmds, tmp.i, tmp.pw_len); - generate_key (salt_buf, tmp.i, tmp.pw_len, key); + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, tmp.i, tmp.pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[8]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, tmp.i, tmp.pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + ukey[6] = ctx.h[2]; + ukey[7] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + ukey[6] = hc_swap32_S (ukey[6]); + ukey[7] = hc_swap32_S (ukey[7]); - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; u32 ks[60]; - aes256_set_decrypt_key (ks, key, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); - - aes256_decrypt (ks, first_block, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - aes256_decrypt (ks, ciphertext, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos, 0, 0); - } - } + AES256_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes256_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes256_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22951_a1-pure.cl b/OpenCL/m22951_a1-pure.cl index b3ded8b40..aaf6af3be 100644 --- a/OpenCL/m22951_a1-pure.cl +++ b/OpenCL/m22951_a1-pure.cl @@ -3,42 +3,39 @@ * License.....: MIT */ -#define BLOCK_SIZE 16 -#define KEY_LENGTH 32 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_aes.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22951_sxx (KERN_ATTR_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22951_mxx (KERN_ATTR_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); - if (gid >= gid_max) return; + /** + * aes shared + */ #ifdef REAL_SHM - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; - - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; - - for (u32 i = lid; i <= data_len / 4; i += lsz) - { - data[i] = esalt_bufs[digests_offset].data[i]; - } - LOCAL_VK u32 s_td0[256]; LOCAL_VK u32 s_td1[256]; LOCAL_VK u32 s_td2[256]; @@ -70,16 +67,234 @@ KERNEL_FQ void m22951_sxx (KERN_ATTR_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + 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; + + 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; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = { - data[i] = esalt_bufs[digests_offset].data[i]; + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * loop + */ + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) + { + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[8]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + ukey[6] = ctx.h[2]; + ukey[7] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + ukey[6] = hc_swap32_S (ukey[6]); + ukey[7] = hc_swap32_S (ukey[7]); + + u32 ks[60]; + + AES256_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes256_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes256_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); } +} + +KERNEL_FQ void m22951_sxx (KERN_ATTR_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) + { + 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]; + + 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]; + } + + SYNC_THREADS (); + + #else CONSTANT_AS u32a *s_td0 = td0; CONSTANT_AS u32a *s_td1 = td1; @@ -93,21 +308,59 @@ KERNEL_FQ void m22951_sxx (KERN_ATTR_ESALT (pem_t)) CONSTANT_AS u32a *s_te3 = te3; CONSTANT_AS u32a *s_te4 = te4; - #endif // REAL_SHM + #endif + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + u32 s[4]; - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; - const u32 pw_len = pws[gid].pw_len; + u32 first_data[4]; - u32 w[16] = { 0 }; + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; - for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) - { - w[idx] = pws[gid].i[idx]; - } + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; /** * loop @@ -115,150 +368,116 @@ KERNEL_FQ void m22951_sxx (KERN_ATTR_ESALT (pem_t)) for (u32 il_pos = 0; il_pos < il_cnt; il_pos++) { - const u32 comb_len = combs_buf[il_pos].pw_len; - u32 c[64]; + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); - #ifdef _unroll - #pragma unroll - #endif - for (int i = 0; i < 16; i++) - { - c[i] = combs_buf[il_pos].i[i]; - } + md5_final (&ctx); - switch_buffer_by_offset_1x64_be_S (c, pw_len); + u32 ukey[8]; - #ifdef _unroll - #pragma unroll - #endif - for (int i = 0; i < 16; i++) - { - c[i] |= w[i]; - } + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; + md5_init (&ctx); - generate_key (salt_buf, c, pw_len + comb_len, key); + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update_global (&ctx, pws[gid].i, pws[gid].pw_len); + + md5_update_global (&ctx, combs_buf[il_pos].i, combs_buf[il_pos].pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + ukey[6] = ctx.h[2]; + ukey[7] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + ukey[6] = hc_swap32_S (ukey[6]); + ukey[7] = hc_swap32_S (ukey[7]); - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; u32 ks[60]; - aes256_set_decrypt_key (ks, key, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); - - aes256_decrypt (ks, first_block, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - aes256_decrypt (ks, ciphertext, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos, 0, 0); - } - } + AES256_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes256_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes256_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/OpenCL/m22951_a3-pure.cl b/OpenCL/m22951_a3-pure.cl index 875f20333..e2753d621 100644 --- a/OpenCL/m22951_a3-pure.cl +++ b/OpenCL/m22951_a3-pure.cl @@ -3,45 +3,39 @@ * License.....: MIT */ -#define NEW_SIMD_CODE - -#define BLOCK_SIZE 16 -#define KEY_LENGTH 32 +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" -#include "inc_simd.cl" +#include "inc_scalar.cl" +#include "inc_hash_md5.cl" #include "inc_cipher_aes.cl" -#include "inc_pem_common.cl" -#endif // KERNEL_STATIC +#endif -KERNEL_FQ void m22951_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +typedef struct pem { - /** - * base - */ + u32 data_buf[16384]; + int data_len; + + int cipher; + +} pem_t; +KERNEL_FQ void m22951_mxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ const u64 gid = get_global_id (0); const u64 lid = get_local_id (0); const u64 lsz = get_local_size (0); - if (gid >= gid_max) return; + /** + * aes shared + */ #ifdef REAL_SHM - LOCAL_VK u32 data_len; - data_len = esalt_bufs[digests_offset].data_len; - - LOCAL_VK u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; - - for (u32 i = lid; i <= data_len / 4; i += lsz) - { - data[i] = esalt_bufs[digests_offset].data[i]; - } - LOCAL_VK u32 s_td0[256]; LOCAL_VK u32 s_td1[256]; LOCAL_VK u32 s_td2[256]; @@ -73,17 +67,252 @@ KERNEL_FQ void m22951_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) #else - const size_t data_len = esalt_bufs[digests_offset].data_len; - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; + 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; + + 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; - #ifdef _unroll - #pragma unroll #endif - for (u32 i = 0; i < data_len / 4; i++) + + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; + + const int last_pad_elem = last_pad_pos / 4; + + u32 iv[4]; + + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; + + for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) + { + w[idx] = pws[gid].i[idx]; + } + + /** + * loop + */ + + u32x w0l = w[0]; + + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) + { + const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + + const u32x w0 = w0l | w0r; + + w[0] = w0; + + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[8]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, w, pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + ukey[6] = ctx.h[2]; + ukey[7] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + ukey[6] = hc_swap32_S (ukey[6]); + ukey[7] = hc_swap32_S (ukey[7]); + + u32 ks[60]; + + AES256_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes256_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes256_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_M_SCALAR (r0, r1, r2, r3); + } +} + +KERNEL_FQ void m22951_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) +{ + 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_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]; + + 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]; + + for (u32 i = lid; i < 256; i += lsz) { - data[i] = esalt_bufs[digests_offset].data[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]; + + 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]; } + SYNC_THREADS (); + + #else + CONSTANT_AS u32a *s_td0 = td0; CONSTANT_AS u32a *s_td1 = td1; CONSTANT_AS u32a *s_td2 = td2; @@ -96,167 +325,193 @@ KERNEL_FQ void m22951_sxx (KERN_ATTR_VECTOR_ESALT (pem_t)) CONSTANT_AS u32a *s_te3 = te3; CONSTANT_AS u32a *s_te4 = te4; - #endif // REAL_SHM + #endif - const u32 pw_len = pws[gid].pw_len; + if (gid >= gid_max) return; + + /** + * digest + */ + + const u32 search[4] = + { + digests_buf[digests_offset].digest_buf[0], + digests_buf[digests_offset].digest_buf[1], + digests_buf[digests_offset].digest_buf[2], + digests_buf[digests_offset].digest_buf[3] + }; + + /** + * base + */ + + u32 s[4]; + + s[0] = salt_bufs[salt_pos].salt_buf[0]; + s[1] = salt_bufs[salt_pos].salt_buf[1]; + s[2] = salt_bufs[salt_pos].salt_buf[2]; + s[3] = salt_bufs[salt_pos].salt_buf[3]; + + u32 first_data[4]; + + first_data[0] = esalt_bufs[digests_offset].data_buf[0]; + first_data[1] = esalt_bufs[digests_offset].data_buf[1]; + first_data[2] = esalt_bufs[digests_offset].data_buf[2]; + first_data[3] = esalt_bufs[digests_offset].data_buf[3]; + + const int data_len = esalt_bufs[digests_offset].data_len; + + const int last_pad_pos = data_len - 1; - u32 salt_buf[16] = { 0 }; - u32 salt_iv[BLOCK_SIZE / 4], first_block[BLOCK_SIZE / 4]; + const int last_pad_elem = last_pad_pos / 4; - prep_buffers(salt_buf, salt_iv, first_block, data, &esalt_bufs[digests_offset]); + u32 iv[4]; - u32x w[16] = { 0 }; + iv[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 7]; + iv[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 6]; + iv[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 5]; + iv[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 4]; + + u32 enc[4]; + + enc[0] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 3]; + enc[1] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 2]; + enc[2] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 1]; + enc[3] = esalt_bufs[digests_offset].data_buf[last_pad_elem - 0]; + + /** + * base + */ + + const u32 pw_len = pws[gid].pw_len; + + u32x w[64] = { 0 }; for (u32 i = 0, idx = 0; i < pw_len; i += 4, idx += 1) { w[idx] = pws[gid].i[idx]; } - u32x w0l = w[0]; - /** * loop */ + u32x w0l = w[0]; + for (u32 il_pos = 0; il_pos < il_cnt; il_pos += VECT_SIZE) { const u32x w0r = words_buf_r[il_pos / VECT_SIZE]; + const u32x w0 = w0l | w0r; w[0] = w0; - u32x keys[HC_PEM_MAX_KEY_LENGTH / 4]; - - generate_key_vector (salt_buf, w, pw_len, keys); - - for (u32 v_pos = 0; v_pos < VECT_SIZE; v_pos++) - { - u32 asn1_ok = 0, padding_ok = 0, plaintext_length, plaintext[BLOCK_SIZE / 4]; - u32 ciphertext[BLOCK_SIZE / 4], iv[BLOCK_SIZE / 4]; - u32 ks[60]; - u32 key[KEY_LENGTH / 4]; - - for (u32 i = 0; i < KEY_LENGTH; i++) - { - key[i] = VECTOR_ELEMENT(keys[i], v_pos); - } - - aes256_set_decrypt_key (ks, key, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); - - aes256_decrypt (ks, first_block, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - plaintext[i] ^= salt_iv[i]; - } - - #ifdef DEBUG - printf("First plaintext block:"); - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) printf(" 0x%08x", plaintext[i]); - printf("\n"); - #endif // DEBUG - - if (data_len < 128) - { - asn1_ok = (plaintext[0] & 0x00ff80ff) == 0x00020030; - plaintext_length = ((plaintext[0] & 0x00007f00) >> 8) + 2; - } - else if (data_len < 256) - { - asn1_ok = (plaintext[0] & 0xff00ffff) == 0x02008130; - plaintext_length = ((plaintext[0] & 0x00ff0000) >> 16) + 3; - } - else if (data_len < 65536) - { - asn1_ok = ((plaintext[0] & 0x0000ffff) == 0x00008230) && ((plaintext[1] & 0x000000ff) == 0x00000002); - plaintext_length = ((plaintext[0] & 0xff000000) >> 24) + ((plaintext[0] & 0x00ff0000) >> 8) + 4; - } - - #ifdef DEBUG - if (asn1_ok == 1) printf("Passed ASN.1 sanity check\n"); - #endif // DEBUG - - if (asn1_ok == 0) - { - continue; - } - - #ifdef _unroll - #pragma unroll - #endif - for (u32 i = 0; i < BLOCK_SIZE / 4; i++) - { - iv[i] = first_block[i]; - } - - for (u32 i = BLOCK_SIZE / 4; i < data_len / 4; i += BLOCK_SIZE / 4) - { - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - ciphertext[j] = data[i + j]; - } - - aes256_decrypt (ks, ciphertext, plaintext, s_td0, s_td1, s_td2, s_td3, s_td4); - - #ifdef _unroll - #pragma unroll - #endif - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) - { - plaintext[j] ^= iv[j]; - iv[j] = ciphertext[j]; - } - - #ifdef DEBUG - printf("Plaintext block %u:", i / (BLOCK_SIZE / 4)); - for (u32 j = 0; j < BLOCK_SIZE / 4; j++) printf(" 0x%08x", plaintext[j]); - printf("\n"); - #endif - } - - u32 padding_count = (plaintext[BLOCK_SIZE / 4 - 1] & 0xff000000) >> 24; - u8 *pt_bytes = (u8 *) plaintext; - - #ifdef DEBUG - printf("Padding byte: 0x%02x\n", padding_count); - #endif - - if (padding_count > BLOCK_SIZE || padding_count == 0) - { - // That *can't* be right - padding_ok = 0; - } else { - padding_ok = 1; - } - - for (u32 i = 0; i < padding_count; i++) - { - if (pt_bytes[BLOCK_SIZE - 1 - i] != padding_count) - { - padding_ok = 0; - break; - } - plaintext_length++; - } - - #ifdef DEBUG - if (padding_ok == 1) printf("Padding checks out\n"); - if (plaintext_length == data_len) printf("ASN.1 sequence length checks out\n"); - #endif - - if (asn1_ok == 1 && padding_ok == 1 && plaintext_length == data_len) - { - if (atomic_inc (&hashes_shown[digests_offset]) == 0) - { - mark_hash (plains_buf, d_return_buf, salt_pos, digests_cnt, 0, digests_offset, gid, il_pos + v_pos, 0, 0); - } - } - } + md5_ctx_t ctx; + + md5_init (&ctx); + + md5_update (&ctx, w, pw_len); + + u32 t[16]; + + t[ 0] = s[0]; + t[ 1] = s[1]; + t[ 2] = 0; + t[ 3] = 0; + t[ 4] = 0; + t[ 5] = 0; + t[ 6] = 0; + t[ 7] = 0; + t[ 8] = 0; + t[ 9] = 0; + t[10] = 0; + t[11] = 0; + t[12] = 0; + t[13] = 0; + t[14] = 0; + t[15] = 0; + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + u32 ukey[8]; + + ukey[0] = ctx.h[0]; + ukey[1] = ctx.h[1]; + ukey[2] = ctx.h[2]; + ukey[3] = ctx.h[3]; + + md5_init (&ctx); + + ctx.w0[0] = ukey[0]; + ctx.w0[1] = ukey[1]; + ctx.w0[2] = ukey[2]; + ctx.w0[3] = ukey[3]; + + ctx.len = 16; + + md5_update (&ctx, w, pw_len); + + md5_update (&ctx, t, 8); + + md5_final (&ctx); + + ukey[4] = ctx.h[0]; + ukey[5] = ctx.h[1]; + ukey[6] = ctx.h[2]; + ukey[7] = ctx.h[3]; + + // AES + + ukey[0] = hc_swap32_S (ukey[0]); + ukey[1] = hc_swap32_S (ukey[1]); + ukey[2] = hc_swap32_S (ukey[2]); + ukey[3] = hc_swap32_S (ukey[3]); + ukey[4] = hc_swap32_S (ukey[4]); + ukey[5] = hc_swap32_S (ukey[5]); + ukey[6] = hc_swap32_S (ukey[6]); + ukey[7] = hc_swap32_S (ukey[7]); + + u32 ks[60]; + + AES256_set_decrypt_key (ks, ukey, s_te0, s_te1, s_te2, s_te3, s_td0, s_td1, s_td2, s_td3); + + u32 dec[4]; + + // first check the padding + + aes256_decrypt (ks, enc, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= iv[0]; + dec[1] ^= iv[1]; + dec[2] ^= iv[2]; + dec[3] ^= iv[3]; + + const int paddingv = pkcs_padding_bs16 (dec, 16); + + if (paddingv == -1) continue; + + // second check (naive code) ASN.1 structure + + aes256_decrypt (ks, first_data, dec, s_td0, s_td1, s_td2, s_td3, s_td4); + + dec[0] ^= s[0]; + dec[1] ^= s[1]; + dec[2] ^= s[2]; + dec[3] ^= s[3]; + + const int real_len = (data_len - 16) + paddingv; + + const int asn1_ok = asn1_detect (dec, real_len); + + if (asn1_ok == 0) continue; + + const u32 r0 = search[0]; + const u32 r1 = search[1]; + const u32 r2 = search[2]; + const u32 r3 = search[3]; + + COMPARE_S_SCALAR (r0, r1, r2, r3); } } diff --git a/docs/changes.txt b/docs/changes.txt index 3c495a7c4..084385a8d 100644 --- a/docs/changes.txt +++ b/docs/changes.txt @@ -1,5 +1,11 @@ * changes v6.1.1 -> v6.x.x +## +## Algorithms +## + +- Added hash-mode: RSA/DSA/EC/OPENSSH Private Keys + * changes v6.1.0 -> v6.1.1 ## @@ -97,7 +103,6 @@ - Added hash-mode: Open Document Format (ODF) 1.1 (SHA-1, Blowfish) - Added hash-mode: Open Document Format (ODF) 1.2 (SHA-256, AES) - Added hash-mode: Oracle Transportation Management (SHA256) -- Added hash-mode: PKCS#1 key - Added hash-mode: PKZIP archive encryption - Added hash-mode: PKZIP Master Key - Added hash-mode: Python passlib pbkdf2-sha1 diff --git a/docs/readme.txt b/docs/readme.txt index e98e182fa..458b162c5 100644 --- a/docs/readme.txt +++ b/docs/readme.txt @@ -273,6 +273,7 @@ NVIDIA GPUs require "NVIDIA Driver" (440.64 or later) and "CUDA Toolkit" (9.0 or - Apple Keychain - Apple Secure Notes - JKS Java Key Store Private Keys (SHA1) +- RSA/DSA/EC/OPENSSH Private Keys - 1Password, agilekeychain - 1Password, cloudkeychain - Password Safe v2 @@ -327,7 +328,6 @@ NVIDIA GPUs require "NVIDIA Driver" (440.64 or later) and "CUDA Toolkit" (9.0 or - Django (SHA-1) - Web2py pbkdf2-sha512 - TOTP (HMAC-SHA1) -- PKCS#1 key ## ## Attack-Modes diff --git a/src/modules/module_22900.c b/src/modules/module_22900.c deleted file mode 100644 index 349e8d582..000000000 --- a/src/modules/module_22900.c +++ /dev/null @@ -1,524 +0,0 @@ -/** - * Author......: See docs/credits.txt - * License.....: MIT - */ - -#include "common.h" -#include "types.h" -#include "modules.h" -#include "bitops.h" -#include "convert.h" -#include "shared.h" -#include "memory.h" - -#define HC_PEM_SALT_LENGTH 8 -#define HC_PEM_MAX_BLOCK_SIZE 16 -#define HC_PEM_MAX_KEY_LENGTH 32 -#define HC_PEM_MAX_DATA_LENGTH 12288 - -// The longest OpenSSL cipher name I can find is 24 characters, so add on seven -// more characters for luck and one for the \0 gives us 32. -#define HC_PEM_MAX_CIPHER_NAME_LENGTH 32 - -static const u32 ATTACK_EXEC = ATTACK_EXEC_INSIDE_KERNEL; -static const u32 DGST_POS0 = 0; -static const u32 DGST_POS1 = 1; -static const u32 DGST_POS2 = 2; -static const u32 DGST_POS3 = 3; -static const u32 DGST_SIZE = DGST_SIZE_4_4; -static const u32 HASH_CATEGORY = HASH_CATEGORY_DOCUMENTS; -static const char *HASH_NAME = "PEM encrypted private key"; -static const u64 KERN_TYPE = 22911; // Kernel used for the benchmark esalt; will likely be overridden in production -static const u32 OPTI_TYPE = OPTI_TYPE_ZERO_BYTE; -static const u64 OPTS_TYPE = OPTS_TYPE_PT_GENERATE_LE - | OPTS_TYPE_BINARY_HASHFILE; -static const u32 SALT_TYPE = SALT_TYPE_EMBEDDED; -static const char *ST_PASS = "hashcat"; -static const char *ST_HASH = NULL; // Benchmark / self-test hash provided in module_benchmark_esalt - -u32 module_attack_exec (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return ATTACK_EXEC; } -u32 module_dgst_pos0 (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return DGST_POS0; } -u32 module_dgst_pos1 (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return DGST_POS1; } -u32 module_dgst_pos2 (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return DGST_POS2; } -u32 module_dgst_pos3 (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return DGST_POS3; } -u32 module_dgst_size (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return DGST_SIZE; } -u32 module_hash_category (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return HASH_CATEGORY; } -const char *module_hash_name (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return HASH_NAME; } -u64 module_kern_type (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return KERN_TYPE; } -u32 module_opti_type (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return OPTI_TYPE; } -u64 module_opts_type (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return OPTS_TYPE; } -u32 module_salt_type (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return SALT_TYPE; } -const char *module_st_hash (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return ST_HASH; } -const char *module_st_pass (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { return ST_PASS; } - -typedef enum hc_pem_cipher_type -{ - HC_PEM_CIPHER_TYPE_3DES = 1, - HC_PEM_CIPHER_TYPE_DES = 2, - HC_PEM_CIPHER_TYPE_AES128 = 3, - HC_PEM_CIPHER_TYPE_AES192 = 4, - HC_PEM_CIPHER_TYPE_AES256 = 5, -} hc_pem_cipher_type_t; - -typedef enum hc_pem_cipher_mode -{ - HC_PEM_CIPHER_MODE_CBC = 1, -} hc_pem_cipher_mode_t; - -typedef struct pem_cipher -{ - char *name; - - u32 block_size; - u32 key_length; - u32 cipher_type; - u32 cipher_mode; -} hc_pem_cipher_t; - -static hc_pem_cipher_t pem_ciphers[] = { - {"des-ede3-cbc", 8, 24, HC_PEM_CIPHER_TYPE_3DES, HC_PEM_CIPHER_MODE_CBC}, - {"des-cbc", 8, 8, HC_PEM_CIPHER_TYPE_DES, HC_PEM_CIPHER_MODE_CBC}, - {"aes-128-cbc", 16, 16, HC_PEM_CIPHER_TYPE_AES128, HC_PEM_CIPHER_MODE_CBC}, - {"aes-192-cbc", 16, 24, HC_PEM_CIPHER_TYPE_AES192, HC_PEM_CIPHER_MODE_CBC}, - {"aes-256-cbc", 16, 32, HC_PEM_CIPHER_TYPE_AES256, HC_PEM_CIPHER_MODE_CBC}, - {NULL, 0, 0, 0, 0} -}; - -typedef struct pem -{ - hc_pem_cipher_t *chosen_cipher; - - u32 salt_iv[HC_PEM_MAX_BLOCK_SIZE / 4]; - - u32 data[HC_PEM_MAX_DATA_LENGTH / 4]; - size_t data_len; -} pem_t; - -typedef struct pem_tmp -{ - u32 key[HC_PEM_MAX_KEY_LENGTH / 4]; -} pem_tmp_t; - - -u32 module_pw_max (MAYBE_UNUSED const hashconfig_t * hashconfig, MAYBE_UNUSED const user_options_t * user_options, MAYBE_UNUSED const user_options_extra_t * user_options_extra) -{ - const u32 pw_max = 64; - - return pw_max; -} - -bool module_outfile_check_disable (MAYBE_UNUSED const hashconfig_t * hashconfig, MAYBE_UNUSED const user_options_t * user_options, MAYBE_UNUSED const user_options_extra_t * user_options_extra) -{ - const bool outfile_check_disable = true; - - return outfile_check_disable; -} - -int module_hash_binary_count (MAYBE_UNUSED const hashes_t * hashes) -{ - return 1; -} - -int module_hash_binary_parse (MAYBE_UNUSED const hashconfig_t * hashconfig, MAYBE_UNUSED const user_options_t * user_options, MAYBE_UNUSED const user_options_extra_t * user_options_extra, hashes_t * hashes) -{ - hash_t *hashes_buf = hashes->hashes_buf; - hash_t *hash = &hashes_buf[0]; - - memset (hash->salt, 0, sizeof (salt_t)); - memset (hash->esalt, 0, sizeof (pem_t)); - - return module_hash_decode (hashconfig, hash->digest, hash->salt, hash->esalt, hash->hook_salt, hash->hash_info, hashes->hashfile, strlen (hashes->hashfile)); -} - -static int peminator (char *buf, char *type, char **start, size_t * len) -{ - char start_header[256], end_header[256]; - - snprintf (start_header, 256, "-----BEGIN %s-----", type); - snprintf (end_header, 256, "-----END %s-----", type); - - char *start_point = buf; - - while (start_point != NULL) - { - if ((start_point = strstr (start_point, start_header)) == NULL) - return -1; - - if (start_point != buf && start_point[-1] != '\n') - continue; - - if (start_point[strlen (start_header)] == '\n') - break; - } - - char *end_point = start_point; - - while (end_point != NULL) - { - if ((end_point = strstr (end_point, end_header)) == NULL) - return -1; - - if (end_point[-1] == '\n' && (end_point[strlen (end_header)] == '\n' || end_point[strlen (end_header)] == '\0')) - { - break; - } - else - { - end_point++; - } - } - - *start = start_point + strlen (start_header) + 1; - *len = end_point - *start; - - return 0; -} - -static int parse_dek_info (char *line, char *cipher_name, u8 * salt) -{ - line += strlen ("DEK-Info: "); - - u8 i = 0; - int salty = -1; - - for (; *line != '\0'; line++) - { - if (salty >= 0) - { - if (i++ % 2 == 0) - { - if (line[1] == '\0') - { - return PARSER_SALT_LENGTH; - } - - salt[salty++] = hex_to_u8 ((u8 *) line); - - if (salty > HC_PEM_MAX_BLOCK_SIZE) - { - return PARSER_SALT_LENGTH; - } - } - else if (line[1] == '\0') - { - if (salty < HC_PEM_SALT_LENGTH) - { - // Malformed salt, not long enough for PKCS5's liking - return PARSER_SALT_LENGTH; - } - else - { - return 0; - } - } - } - else if (*line == ',') - { - cipher_name[i] = '\0'; - salty = 0; - i = 0; - } - else - { - cipher_name[i++] = *line; - if (i >= HC_PEM_MAX_CIPHER_NAME_LENGTH) - { - return PARSER_CIPHER; - } - } - } - - return PARSER_SALT_VALUE; -} - -static int parse_pem_key_data (char *buf, char *cipher_name, u8 * salt, u8 * data, size_t * data_len) -{ - char *pemdata; - size_t pemdata_len; - - if (peminator (buf, "RSA PRIVATE KEY", &pemdata, &pemdata_len) < 0) - { - if (peminator (buf, "DSA PRIVATE KEY", &pemdata, &pemdata_len) < 0) - { - if (peminator (buf, "EC PRIVATE KEY", &pemdata, &pemdata_len) < 0) - { - if (peminator (buf, "PRIVATE KEY", &pemdata, &pemdata_len) < 0) - { - return PARSER_HASH_FILE; - } - } - } - } - - u8 in_header = 1, *b64data; - char line[256]; - size_t pd_idx = 0, l_idx = 0, b64_idx = 0; - - b64data = hcmalloc (pemdata_len); - - for (pd_idx = 0; pd_idx < pemdata_len; pd_idx++) - { - if (in_header) - { - if (pemdata[pd_idx] == '\n') - { - if (l_idx == 0) - { - // Empty line! - in_header = 0; - continue; - } - - line[l_idx] = '\0'; - - if (strstr (line, "DEK-Info: ") == line) - { - int err; - - if ((err = parse_dek_info (line, cipher_name, salt)) < 0) - { - return err; - } - } - - l_idx = 0; - } - else - { - line[l_idx++] = pemdata[pd_idx]; - } - } - else - { - if (pemdata[pd_idx] != '\n') - b64data[b64_idx++] = pemdata[pd_idx]; - } - } - - if (b64_idx * 6 / 8 > HC_PEM_MAX_DATA_LENGTH) - { - return PARSER_TOKEN_LENGTH; - } - - *data_len = base64_decode (base64_to_int, b64data, b64_idx, data); - - return 0; -} - -int module_hash_decode (MAYBE_UNUSED const hashconfig_t * hashconfig, MAYBE_UNUSED void *digest_buf, salt_t * salt, void *esalt_buf, MAYBE_UNUSED void *hook_salt_buf, MAYBE_UNUSED hashinfo_t * hash_info, const char *line_buf, MAYBE_UNUSED const int line_len) -{ - pem_t *pem = (pem_t *) esalt_buf; - - HCFILE fp; - struct stat fileinfo; - u8 *filebuf; - - if (stat (line_buf, &fileinfo) < 0) - return 0; - - if (hc_fopen (&fp, line_buf, "rb") == false) - return 0; - - filebuf = hcmalloc (fileinfo.st_size + 1); - - if (hc_fread (filebuf, 1, fileinfo.st_size, &fp) < (size_t) fileinfo.st_size) - { - hc_fclose (&fp); - hcfree (filebuf); - - return PARSER_FILE_SIZE; - } - - hc_fclose (&fp); - - filebuf[fileinfo.st_size] = '\0'; - - char cipher_name[HC_PEM_MAX_CIPHER_NAME_LENGTH] = { 0 }; - u8 saltbytes[MAX(HC_PEM_SALT_LENGTH, HC_PEM_MAX_BLOCK_SIZE)]; - int err; - - if ((err = parse_pem_key_data ((char *) filebuf, cipher_name, saltbytes, (u8 *) pem->data, &pem->data_len)) < 0) - { - hcfree (filebuf); - - return err; - } - - u32 *saltwords = (u32 *) saltbytes; - - for (u32 i = 0; i < HC_PEM_SALT_LENGTH / 4; i++) - { - pem->salt_iv[i] = saltwords[i]; - } - - hc_pem_cipher_t *candidate_cipher = pem_ciphers, *chosen_cipher = NULL; - - while (candidate_cipher->name) - { - if (strcasecmp (cipher_name, candidate_cipher->name) == 0) - { - chosen_cipher = candidate_cipher; - break; - } - else - { - candidate_cipher++; - } - } - - if (chosen_cipher == NULL) - { - hcfree (filebuf); - - return PARSER_CIPHER; - } - - if (chosen_cipher->block_size > HC_PEM_MAX_BLOCK_SIZE) - { - hcfree (filebuf); - - return PARSER_BLOCK_SIZE; - } - - if (pem->data_len % chosen_cipher->block_size) - { - hcfree (filebuf); - - return PARSER_HASH_LENGTH; - } - - if (chosen_cipher->key_length > HC_PEM_MAX_KEY_LENGTH) - { - // Nope nope nopety nope - return PARSER_KEY_SIZE; - } - - pem->chosen_cipher = chosen_cipher; - - memcpy (salt->salt_buf, pem->salt_iv, MIN (HC_PEM_SALT_LENGTH, 64 * 4)); - salt->salt_iter = 1; - - return 1; -} - -void *module_benchmark_esalt (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) -{ - pem_t *pem = (pem_t *) hcmalloc (sizeof (pem_t)); - - pem->chosen_cipher = &pem_ciphers[0]; - hex_decode ((u8 *) "7CC48DB27D461D30", 16, (u8 *) pem->salt_iv); - pem->data_len = base64_decode (base64_to_int, (u8 *) "ysVmp6tkcZXRqHyy3YMk5zd4bsT9D97kFcDIKkD2g5o/OBgc0pGQ/iSwJm/V+A2IkwgQlwvLW1OfKkAWdjcSFNKhmiWApVQB", 96, (u8 *) pem->data); - - return pem; -} - -salt_t *module_benchmark_salt (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) -{ - salt_t *salt = (salt_t *) hcmalloc (sizeof (salt_t)); - - salt->salt_iter = 1; - hex_decode ((u8 *) "7CC48DB27D461D30", 16, (u8 *) salt->salt_buf); - - return salt; -} - -u64 module_kern_type_dynamic (MAYBE_UNUSED const hashconfig_t * hashconfig, MAYBE_UNUSED const void *digest_buf, MAYBE_UNUSED const salt_t * salt, MAYBE_UNUSED const void *esalt_buf, MAYBE_UNUSED const void *hook_salt_buf, MAYBE_UNUSED const hashinfo_t * hash_info) -{ - const pem_t *pem = (const pem_t *) esalt_buf; - - u64 kern_type = 22900; - - kern_type += pem->chosen_cipher->cipher_type * 10; - kern_type += pem->chosen_cipher->cipher_mode; - - return kern_type; -} - -u64 module_esalt_size (MAYBE_UNUSED const hashconfig_t * hashconfig, MAYBE_UNUSED const user_options_t * user_options, MAYBE_UNUSED const user_options_extra_t * user_options_extra) -{ - const u64 esalt_size = (const u64) sizeof (pem_t); - - return esalt_size; -} - -char *module_jit_build_options (MAYBE_UNUSED const hashconfig_t * hashconfig, MAYBE_UNUSED const user_options_t * user_options, MAYBE_UNUSED const user_options_extra_t * user_options_extra, MAYBE_UNUSED const hashes_t * hashes, MAYBE_UNUSED const hc_device_param_t * device_param) -{ - char *jit_build_options = NULL; - - hc_asprintf (&jit_build_options, "-D _unroll"); - - return jit_build_options; -} - -void module_init (module_ctx_t * module_ctx) -{ - module_ctx->module_context_size = MODULE_CONTEXT_SIZE_CURRENT; - module_ctx->module_interface_version = MODULE_INTERFACE_VERSION_CURRENT; - - module_ctx->module_attack_exec = module_attack_exec; - module_ctx->module_benchmark_esalt = module_benchmark_esalt; - module_ctx->module_benchmark_hook_salt = MODULE_DEFAULT; - module_ctx->module_benchmark_mask = MODULE_DEFAULT; - module_ctx->module_benchmark_salt = module_benchmark_salt; - module_ctx->module_build_plain_postprocess = MODULE_DEFAULT; - module_ctx->module_deep_comp_kernel = MODULE_DEFAULT; - module_ctx->module_dgst_pos0 = module_dgst_pos0; - module_ctx->module_dgst_pos1 = module_dgst_pos1; - module_ctx->module_dgst_pos2 = module_dgst_pos2; - module_ctx->module_dgst_pos3 = module_dgst_pos3; - module_ctx->module_dgst_size = module_dgst_size; - module_ctx->module_dictstat_disable = MODULE_DEFAULT; - module_ctx->module_esalt_size = module_esalt_size; - module_ctx->module_extra_buffer_size = MODULE_DEFAULT; - module_ctx->module_extra_tmp_size = MODULE_DEFAULT; - module_ctx->module_forced_outfile_format = MODULE_DEFAULT; - module_ctx->module_hash_binary_count = module_hash_binary_count; - module_ctx->module_hash_binary_parse = module_hash_binary_parse; - module_ctx->module_hash_binary_save = MODULE_DEFAULT; - module_ctx->module_hash_decode_potfile = MODULE_DEFAULT; - module_ctx->module_hash_decode_zero_hash = MODULE_DEFAULT; - module_ctx->module_hash_decode = module_hash_decode; - module_ctx->module_hash_encode_status = MODULE_DEFAULT; - module_ctx->module_hash_encode_potfile = MODULE_DEFAULT; - module_ctx->module_hash_encode = MODULE_DEFAULT; - module_ctx->module_hash_init_selftest = MODULE_DEFAULT; - module_ctx->module_hash_mode = MODULE_DEFAULT; - module_ctx->module_hash_category = module_hash_category; - module_ctx->module_hash_name = module_hash_name; - module_ctx->module_hashes_count_min = MODULE_DEFAULT; - module_ctx->module_hashes_count_max = MODULE_DEFAULT; - module_ctx->module_hlfmt_disable = MODULE_DEFAULT; - module_ctx->module_hook12 = MODULE_DEFAULT; - module_ctx->module_hook23 = MODULE_DEFAULT; - module_ctx->module_hook_salt_size = MODULE_DEFAULT; - module_ctx->module_hook_size = MODULE_DEFAULT; - module_ctx->module_jit_build_options = module_jit_build_options; - module_ctx->module_jit_cache_disable = MODULE_DEFAULT; - module_ctx->module_kernel_accel_max = MODULE_DEFAULT; - module_ctx->module_kernel_accel_min = MODULE_DEFAULT; - module_ctx->module_kernel_loops_max = MODULE_DEFAULT; - module_ctx->module_kernel_loops_min = MODULE_DEFAULT; - module_ctx->module_kernel_threads_max = MODULE_DEFAULT; - module_ctx->module_kernel_threads_min = MODULE_DEFAULT; - module_ctx->module_kern_type = module_kern_type; - module_ctx->module_kern_type_dynamic = module_kern_type_dynamic; - module_ctx->module_opti_type = module_opti_type; - module_ctx->module_opts_type = module_opts_type; - module_ctx->module_outfile_check_disable = module_outfile_check_disable; - module_ctx->module_outfile_check_nocomp = MODULE_DEFAULT; - module_ctx->module_potfile_custom_check = MODULE_DEFAULT; - module_ctx->module_potfile_disable = MODULE_DEFAULT; - module_ctx->module_potfile_keep_all_hashes = MODULE_DEFAULT; - module_ctx->module_pwdump_column = MODULE_DEFAULT; - module_ctx->module_pw_max = module_pw_max; - module_ctx->module_pw_min = MODULE_DEFAULT; - module_ctx->module_salt_max = MODULE_DEFAULT; - module_ctx->module_salt_min = MODULE_DEFAULT; - module_ctx->module_salt_type = module_salt_type; - module_ctx->module_separator = MODULE_DEFAULT; - module_ctx->module_st_hash = module_st_hash; - module_ctx->module_st_pass = module_st_pass; - module_ctx->module_tmp_size = MODULE_DEFAULT; - module_ctx->module_unstable_warning = MODULE_DEFAULT; - module_ctx->module_warmup_disable = MODULE_DEFAULT; -} diff --git a/tools/test_modules/m22911.pm b/tools/test_modules/m22911.pm new file mode 100644 index 000000000..3bb086300 --- /dev/null +++ b/tools/test_modules/m22911.pm @@ -0,0 +1,159 @@ +#!/usr/bin/env perl + +## +## Author......: See docs/credits.txt +## License.....: MIT +## + +use strict; +use warnings; + +use Crypt::CBC; +use Digest::MD5 qw (md5); + +sub module_constraints { [[0, 256], [16, 16], [-1, -1], [-1, -1], [-1, -1]] } + +sub generate_key +{ + my $word = shift; + my $salt = shift; + my $bytes = shift; + + my $salt8 = substr ($salt, 0, 8); + + my $out = md5 ($word . $salt8); + + $out .= md5 ($out . $word . $salt8); + + return substr ($out, 0, 24); +} + +sub module_generate_hash +{ + my $word = shift; + my $salt = shift; + my $cid = shift // 0; + my $data = shift; + + my $salt_bin = pack ("H*", $salt); + my $data_bin; + + my $key = generate_key ($word, $salt_bin); + + my $is_decrypt = defined ($data); + + if ($is_decrypt == 1) + { + $data_bin = pack ("H*", $data); + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::DES_EDE3", + key => $key, + iv => $salt_bin, + keysize => 24, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $dec_bin = $aes->decrypt ($data_bin); + + ## This is a ridiculous check of successfull decryption + ## There are no useable asn1 parsers for perl available + ## We have to rely on a combination of padding check and pattern matching + ## The (minimum) 16 bit should be good enough for a unit test + + if (length ($dec_bin) < length ($data_bin)) + { + if (substr ($dec_bin, 0, 1) eq "\x30") + { + $data_bin = $dec_bin; + } + } + } + else + { + $data_bin = pack ("H*", + "308204bd020100300d06092a864886f70d0101010500048204a7308204a30201000282010100b7a2" . + "e4c254c8174219e60d9cce96737a906797b8edb86af8f055f60db7bd298b0d31d7ce97ebeae393d5" . + "0e6da5215b58dcd72f4d3cac9e79b6ccaed7da47d2bd04f6a767f5ab7dc0f58beb6298c1e2358ed6" . + "d3ef441f2326ac5db0027e08ae6c7724ff9a2a220a07e97319b6eff5cd653c7ab8b6ea9f9e89a40a" . + "b856f036acfd39b1e5926964a024de35052de6d3423fe763569f48869c834750b28f09cecdddb54a" . + "5526a2c5159d22d24606a2af6c6f47a5d9c04c454896192b8e7b82cf6f6934a23d3495059cb7e43b" . + "98a20bd5b5782e15d93c8b289838c0a1df82ee429f0708d97aa40d6e75ec57ff12a2714871f241a8" . + "6f6d8d3472b084aeb748da33e2d50203010001028201004afab8dadc1122e5fb7b225dbf4051005f" . + "4bdcf84620019589541ff633ea89b6dbf958fb62ae9226bfeac34c639b3e18077bd935792ba63d5e" . + "352ec2b5be93be57f37a21097f2f06857bceed601ff2041a417f2177b81afb246fd079040af96512" . + "34ca24a1456ac11641c7e319114cff23f59bcc1bfa769a0e9fcdeab98429973e10caf303f2bcb065" . + "f22c1cc259556de8377431237da7082cf03ce8da9530be398022f0171d468d92fcabbe776a5e9cf2" . + "045642868406fd03ab735a70bfec3a951bb3c7a1de0fb3ff63cef23897e4fc3f9c5edf62fd45d058" . + "fedc7d2fb22ec928984a061053a7138ce0417b5512579a92be0775104c0bc911f68a5e8ede298102" . + "818100e4e17e2c752dbf1ad1025a074dc5f9c3c5989d23c84594313373d3e4ed0c0ddd74429ab026" . + "535c5e77549d888835bc94f069ebc5e77fdbd2ddf4c8be6cf777799a6d8d18e2b8cecfc13ab26df8" . + "b71ca3d94c2193c294042fb1025fdab38ba7aaeafebd8dd1f9d78ee67100693e99255dad6b964ebf" . + "b7401a03b67d412fabb33502818100cd65067097e307643df1fc8214db1dd7d09342ef01417a2620" . + "adad87352a58b8fcf07521289da3851623d8d045935fab7ecccc52ba0b86adcb92da76255e00289b" . + "af9aacd936201861b0021249f4ab5e6020db823af7171aef0bbbd02dc94d2489fc0b68500bd1b7d2" . + "81ed69fe4a44384161fe906e49bc91e0362b446ec2c521028180497662d40c2c49b966ba758100a2" . + "799f2f8de369f7bef568b1560cfdde63cf13745c685fff7d2419a1fd83aeade1698cf87956d6a78e" . + "2f55482e683c4ea7432ec1b545e365e9e15f676ada98578b166334bcadce4a56cddd2cd85141d5fd" . + "0e2cdace36b30d613ea1bc2f2aed9cccf4e4536443d334cfb180680eabb73f80c1bd0281803f30fc" . + "b93951a4dd875d62e5968b0f746d7c51147d5b6abc3e4390e6cf4997005af993dfbec23923e1fae7" . + "62b47531f2ee510defc9c3700d1a5bb510b2506856160801db79fc78056850a16285145c80edac4e" . + "3c93ed9f532f067a2303633273b26c340a44ce4e1873107c3da6f9ac616e643ad0aecdcad14a9cff" . + "d4cf0ae76102818100b82528a3dfc595cf9c6a025998491e3b4849c71aa8d1222ddb14af7f82fbe5" . + "169ec3ba18ec28d5a9501e95bc9da72cea99e4cdfdf898f40bec6b28f838243d2f39d7226e0873ed" . + "ee752bcae07639a4bd0eb31be1718c456391630b83ad0e9bf3fa18a645007e64fe59af467ea021f9" . + "e9a0dd759b21cd0b93333a73116abcaa2a" + ); + } + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::DES_EDE3", + key => $key, + iv => $salt_bin, + keysize => 24, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $enc_bin = $aes->encrypt ($data_bin); + + my $hash = sprintf ('$sshng$%d$%d$%s$%d$%s', $cid, length ($salt_bin), unpack ("H*", $salt_bin), length ($enc_bin), unpack ("H*", $enc_bin)); + + return $hash; +} + +sub module_verify_hash +{ + my $line = shift; + + my $idx = index ($line, ':'); + + return unless $idx >= 0; + + my $hash = substr ($line, 0, $idx); + my $word = substr ($line, $idx + 1); + + return unless substr ($hash, 0, 7) eq '$sshng$'; + + my (undef, $signature, $cid, undef, $salt, undef, $data) = split '\$', $hash; + + return unless defined $signature; + return unless defined $cid; + return unless defined $salt; + return unless defined $data; + + return unless ($signature eq 'sshng'); + + return unless ($cid == 0); + + my $word_packed = pack_if_HEX_notation ($word); + + my $new_hash = module_generate_hash ($word_packed, $salt, $cid, $data); + + return ($new_hash, $word); +} + +1; diff --git a/tools/test_modules/m22921.pm b/tools/test_modules/m22921.pm new file mode 100644 index 000000000..63e9d4aad --- /dev/null +++ b/tools/test_modules/m22921.pm @@ -0,0 +1,157 @@ +#!/usr/bin/env perl + +## +## Author......: See docs/credits.txt +## License.....: MIT +## + +use strict; +use warnings; + +use Crypt::CBC; +use Digest::MD5 qw (md5); + +sub module_constraints { [[0, 256], [16, 16], [-1, -1], [-1, -1], [-1, -1]] } + +sub generate_key +{ + my $word = shift; + my $salt = shift; + my $bytes = shift; + + my $salt8 = substr ($salt, 0, 8); + + my $out = md5 ($word . $salt8); + + return substr ($out, 0, 8); +} + +sub module_generate_hash +{ + my $word = shift; + my $salt = shift; + my $cid = shift // 6; + my $data = shift; + + my $salt_bin = pack ("H*", $salt); + my $data_bin; + + my $key = generate_key ($word, $salt_bin); + + my $is_decrypt = defined ($data); + + if ($is_decrypt == 1) + { + $data_bin = pack ("H*", $data); + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::DES", + key => $key, + iv => $salt_bin, + keysize => 8, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $dec_bin = $aes->decrypt ($data_bin); + + ## This is a ridiculous check of successfull decryption + ## There are no useable asn1 parsers for perl available + ## We have to rely on a combination of padding check and pattern matching + ## The (minimum) 16 bit should be good enough for a unit test + + if (length ($dec_bin) < length ($data_bin)) + { + if (substr ($dec_bin, 0, 1) eq "\x30") + { + $data_bin = $dec_bin; + } + } + } + else + { + $data_bin = pack ("H*", + "308204bd020100300d06092a864886f70d0101010500048204a7308204a30201000282010100b7a2" . + "e4c254c8174219e60d9cce96737a906797b8edb86af8f055f60db7bd298b0d31d7ce97ebeae393d5" . + "0e6da5215b58dcd72f4d3cac9e79b6ccaed7da47d2bd04f6a767f5ab7dc0f58beb6298c1e2358ed6" . + "d3ef441f2326ac5db0027e08ae6c7724ff9a2a220a07e97319b6eff5cd653c7ab8b6ea9f9e89a40a" . + "b856f036acfd39b1e5926964a024de35052de6d3423fe763569f48869c834750b28f09cecdddb54a" . + "5526a2c5159d22d24606a2af6c6f47a5d9c04c454896192b8e7b82cf6f6934a23d3495059cb7e43b" . + "98a20bd5b5782e15d93c8b289838c0a1df82ee429f0708d97aa40d6e75ec57ff12a2714871f241a8" . + "6f6d8d3472b084aeb748da33e2d50203010001028201004afab8dadc1122e5fb7b225dbf4051005f" . + "4bdcf84620019589541ff633ea89b6dbf958fb62ae9226bfeac34c639b3e18077bd935792ba63d5e" . + "352ec2b5be93be57f37a21097f2f06857bceed601ff2041a417f2177b81afb246fd079040af96512" . + "34ca24a1456ac11641c7e319114cff23f59bcc1bfa769a0e9fcdeab98429973e10caf303f2bcb065" . + "f22c1cc259556de8377431237da7082cf03ce8da9530be398022f0171d468d92fcabbe776a5e9cf2" . + "045642868406fd03ab735a70bfec3a951bb3c7a1de0fb3ff63cef23897e4fc3f9c5edf62fd45d058" . + "fedc7d2fb22ec928984a061053a7138ce0417b5512579a92be0775104c0bc911f68a5e8ede298102" . + "818100e4e17e2c752dbf1ad1025a074dc5f9c3c5989d23c84594313373d3e4ed0c0ddd74429ab026" . + "535c5e77549d888835bc94f069ebc5e77fdbd2ddf4c8be6cf777799a6d8d18e2b8cecfc13ab26df8" . + "b71ca3d94c2193c294042fb1025fdab38ba7aaeafebd8dd1f9d78ee67100693e99255dad6b964ebf" . + "b7401a03b67d412fabb33502818100cd65067097e307643df1fc8214db1dd7d09342ef01417a2620" . + "adad87352a58b8fcf07521289da3851623d8d045935fab7ecccc52ba0b86adcb92da76255e00289b" . + "af9aacd936201861b0021249f4ab5e6020db823af7171aef0bbbd02dc94d2489fc0b68500bd1b7d2" . + "81ed69fe4a44384161fe906e49bc91e0362b446ec2c521028180497662d40c2c49b966ba758100a2" . + "799f2f8de369f7bef568b1560cfdde63cf13745c685fff7d2419a1fd83aeade1698cf87956d6a78e" . + "2f55482e683c4ea7432ec1b545e365e9e15f676ada98578b166334bcadce4a56cddd2cd85141d5fd" . + "0e2cdace36b30d613ea1bc2f2aed9cccf4e4536443d334cfb180680eabb73f80c1bd0281803f30fc" . + "b93951a4dd875d62e5968b0f746d7c51147d5b6abc3e4390e6cf4997005af993dfbec23923e1fae7" . + "62b47531f2ee510defc9c3700d1a5bb510b2506856160801db79fc78056850a16285145c80edac4e" . + "3c93ed9f532f067a2303633273b26c340a44ce4e1873107c3da6f9ac616e643ad0aecdcad14a9cff" . + "d4cf0ae76102818100b82528a3dfc595cf9c6a025998491e3b4849c71aa8d1222ddb14af7f82fbe5" . + "169ec3ba18ec28d5a9501e95bc9da72cea99e4cdfdf898f40bec6b28f838243d2f39d7226e0873ed" . + "ee752bcae07639a4bd0eb31be1718c456391630b83ad0e9bf3fa18a645007e64fe59af467ea021f9" . + "e9a0dd759b21cd0b93333a73116abcaa2a" + ); + } + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::DES", + key => $key, + iv => $salt_bin, + keysize => 8, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $enc_bin = $aes->encrypt ($data_bin); + + my $hash = sprintf ('$sshng$%d$%d$%s$%d$%s', $cid, length ($salt_bin), unpack ("H*", $salt_bin), length ($enc_bin), unpack ("H*", $enc_bin)); + + return $hash; +} + +sub module_verify_hash +{ + my $line = shift; + + my $idx = index ($line, ':'); + + return unless $idx >= 0; + + my $hash = substr ($line, 0, $idx); + my $word = substr ($line, $idx + 1); + + return unless substr ($hash, 0, 7) eq '$sshng$'; + + my (undef, $signature, $cid, undef, $salt, undef, $data) = split '\$', $hash; + + return unless defined $signature; + return unless defined $cid; + return unless defined $salt; + return unless defined $data; + + return unless ($signature eq 'sshng'); + + return unless ($cid == 6); + + my $word_packed = pack_if_HEX_notation ($word); + + my $new_hash = module_generate_hash ($word_packed, $salt, $cid, $data); + + return ($new_hash, $word); +} + +1; diff --git a/tools/test_modules/m22931.pm b/tools/test_modules/m22931.pm new file mode 100644 index 000000000..9e508b29e --- /dev/null +++ b/tools/test_modules/m22931.pm @@ -0,0 +1,157 @@ +#!/usr/bin/env perl + +## +## Author......: See docs/credits.txt +## License.....: MIT +## + +use strict; +use warnings; + +use Crypt::CBC; +use Digest::MD5 qw (md5); + +sub module_constraints { [[0, 256], [32, 32], [-1, -1], [-1, -1], [-1, -1]] } + +sub generate_key +{ + my $word = shift; + my $salt = shift; + my $bytes = shift; + + my $salt8 = substr ($salt, 0, 8); + + my $out = md5 ($word . $salt8); + + return substr ($out, 0, 16); +} + +sub module_generate_hash +{ + my $word = shift; + my $salt = shift; + my $cid = shift // 1; + my $data = shift; + + my $salt_bin = pack ("H*", $salt); + my $data_bin; + + my $key = generate_key ($word, $salt_bin); + + my $is_decrypt = defined ($data); + + if ($is_decrypt == 1) + { + $data_bin = pack ("H*", $data); + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::Rijndael", + key => $key, + iv => $salt_bin, + keysize => 16, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $dec_bin = $aes->decrypt ($data_bin); + + ## This is a ridiculous check of successfull decryption + ## There are no useable asn1 parsers for perl available + ## We have to rely on a combination of padding check and pattern matching + ## The (minimum) 16 bit should be good enough for a unit test + + if (length ($dec_bin) < length ($data_bin)) + { + if (substr ($dec_bin, 0, 1) eq "\x30") + { + $data_bin = $dec_bin; + } + } + } + else + { + $data_bin = pack ("H*", + "308204bd020100300d06092a864886f70d0101010500048204a7308204a30201000282010100b7a2" . + "e4c254c8174219e60d9cce96737a906797b8edb86af8f055f60db7bd298b0d31d7ce97ebeae393d5" . + "0e6da5215b58dcd72f4d3cac9e79b6ccaed7da47d2bd04f6a767f5ab7dc0f58beb6298c1e2358ed6" . + "d3ef441f2326ac5db0027e08ae6c7724ff9a2a220a07e97319b6eff5cd653c7ab8b6ea9f9e89a40a" . + "b856f036acfd39b1e5926964a024de35052de6d3423fe763569f48869c834750b28f09cecdddb54a" . + "5526a2c5159d22d24606a2af6c6f47a5d9c04c454896192b8e7b82cf6f6934a23d3495059cb7e43b" . + "98a20bd5b5782e15d93c8b289838c0a1df82ee429f0708d97aa40d6e75ec57ff12a2714871f241a8" . + "6f6d8d3472b084aeb748da33e2d50203010001028201004afab8dadc1122e5fb7b225dbf4051005f" . + "4bdcf84620019589541ff633ea89b6dbf958fb62ae9226bfeac34c639b3e18077bd935792ba63d5e" . + "352ec2b5be93be57f37a21097f2f06857bceed601ff2041a417f2177b81afb246fd079040af96512" . + "34ca24a1456ac11641c7e319114cff23f59bcc1bfa769a0e9fcdeab98429973e10caf303f2bcb065" . + "f22c1cc259556de8377431237da7082cf03ce8da9530be398022f0171d468d92fcabbe776a5e9cf2" . + "045642868406fd03ab735a70bfec3a951bb3c7a1de0fb3ff63cef23897e4fc3f9c5edf62fd45d058" . + "fedc7d2fb22ec928984a061053a7138ce0417b5512579a92be0775104c0bc911f68a5e8ede298102" . + "818100e4e17e2c752dbf1ad1025a074dc5f9c3c5989d23c84594313373d3e4ed0c0ddd74429ab026" . + "535c5e77549d888835bc94f069ebc5e77fdbd2ddf4c8be6cf777799a6d8d18e2b8cecfc13ab26df8" . + "b71ca3d94c2193c294042fb1025fdab38ba7aaeafebd8dd1f9d78ee67100693e99255dad6b964ebf" . + "b7401a03b67d412fabb33502818100cd65067097e307643df1fc8214db1dd7d09342ef01417a2620" . + "adad87352a58b8fcf07521289da3851623d8d045935fab7ecccc52ba0b86adcb92da76255e00289b" . + "af9aacd936201861b0021249f4ab5e6020db823af7171aef0bbbd02dc94d2489fc0b68500bd1b7d2" . + "81ed69fe4a44384161fe906e49bc91e0362b446ec2c521028180497662d40c2c49b966ba758100a2" . + "799f2f8de369f7bef568b1560cfdde63cf13745c685fff7d2419a1fd83aeade1698cf87956d6a78e" . + "2f55482e683c4ea7432ec1b545e365e9e15f676ada98578b166334bcadce4a56cddd2cd85141d5fd" . + "0e2cdace36b30d613ea1bc2f2aed9cccf4e4536443d334cfb180680eabb73f80c1bd0281803f30fc" . + "b93951a4dd875d62e5968b0f746d7c51147d5b6abc3e4390e6cf4997005af993dfbec23923e1fae7" . + "62b47531f2ee510defc9c3700d1a5bb510b2506856160801db79fc78056850a16285145c80edac4e" . + "3c93ed9f532f067a2303633273b26c340a44ce4e1873107c3da6f9ac616e643ad0aecdcad14a9cff" . + "d4cf0ae76102818100b82528a3dfc595cf9c6a025998491e3b4849c71aa8d1222ddb14af7f82fbe5" . + "169ec3ba18ec28d5a9501e95bc9da72cea99e4cdfdf898f40bec6b28f838243d2f39d7226e0873ed" . + "ee752bcae07639a4bd0eb31be1718c456391630b83ad0e9bf3fa18a645007e64fe59af467ea021f9" . + "e9a0dd759b21cd0b93333a73116abcaa2a" + ); + } + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::Rijndael", + key => $key, + iv => $salt_bin, + keysize => 16, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $enc_bin = $aes->encrypt ($data_bin); + + my $hash = sprintf ('$sshng$%d$%d$%s$%d$%s', $cid, length ($salt_bin), unpack ("H*", $salt_bin), length ($enc_bin), unpack ("H*", $enc_bin)); + + return $hash; +} + +sub module_verify_hash +{ + my $line = shift; + + my $idx = index ($line, ':'); + + return unless $idx >= 0; + + my $hash = substr ($line, 0, $idx); + my $word = substr ($line, $idx + 1); + + return unless substr ($hash, 0, 7) eq '$sshng$'; + + my (undef, $signature, $cid, undef, $salt, undef, $data) = split '\$', $hash; + + return unless defined $signature; + return unless defined $cid; + return unless defined $salt; + return unless defined $data; + + return unless ($signature eq 'sshng'); + + return unless (($cid == 1) || ($cid == 3)); + + my $word_packed = pack_if_HEX_notation ($word); + + my $new_hash = module_generate_hash ($word_packed, $salt, $cid, $data); + + return ($new_hash, $word); +} + +1; diff --git a/tools/test_modules/m22941.pm b/tools/test_modules/m22941.pm new file mode 100644 index 000000000..f4f6e0f91 --- /dev/null +++ b/tools/test_modules/m22941.pm @@ -0,0 +1,159 @@ +#!/usr/bin/env perl + +## +## Author......: See docs/credits.txt +## License.....: MIT +## + +use strict; +use warnings; + +use Crypt::CBC; +use Digest::MD5 qw (md5); + +sub module_constraints { [[0, 256], [32, 32], [-1, -1], [-1, -1], [-1, -1]] } + +sub generate_key +{ + my $word = shift; + my $salt = shift; + my $bytes = shift; + + my $salt8 = substr ($salt, 0, 8); + + my $out = md5 ($word . $salt8); + + $out .= md5 ($out . $word . $salt8); + + return substr ($out, 0, 24); +} + +sub module_generate_hash +{ + my $word = shift; + my $salt = shift; + my $cid = shift // 4; + my $data = shift; + + my $salt_bin = pack ("H*", $salt); + my $data_bin; + + my $key = generate_key ($word, $salt_bin); + + my $is_decrypt = defined ($data); + + if ($is_decrypt == 1) + { + $data_bin = pack ("H*", $data); + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::Rijndael", + key => $key, + iv => $salt_bin, + keysize => 24, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $dec_bin = $aes->decrypt ($data_bin); + + ## This is a ridiculous check of successfull decryption + ## There are no useable asn1 parsers for perl available + ## We have to rely on a combination of padding check and pattern matching + ## The (minimum) 16 bit should be good enough for a unit test + + if (length ($dec_bin) < length ($data_bin)) + { + if (substr ($dec_bin, 0, 1) eq "\x30") + { + $data_bin = $dec_bin; + } + } + } + else + { + $data_bin = pack ("H*", + "308204bd020100300d06092a864886f70d0101010500048204a7308204a30201000282010100b7a2" . + "e4c254c8174219e60d9cce96737a906797b8edb86af8f055f60db7bd298b0d31d7ce97ebeae393d5" . + "0e6da5215b58dcd72f4d3cac9e79b6ccaed7da47d2bd04f6a767f5ab7dc0f58beb6298c1e2358ed6" . + "d3ef441f2326ac5db0027e08ae6c7724ff9a2a220a07e97319b6eff5cd653c7ab8b6ea9f9e89a40a" . + "b856f036acfd39b1e5926964a024de35052de6d3423fe763569f48869c834750b28f09cecdddb54a" . + "5526a2c5159d22d24606a2af6c6f47a5d9c04c454896192b8e7b82cf6f6934a23d3495059cb7e43b" . + "98a20bd5b5782e15d93c8b289838c0a1df82ee429f0708d97aa40d6e75ec57ff12a2714871f241a8" . + "6f6d8d3472b084aeb748da33e2d50203010001028201004afab8dadc1122e5fb7b225dbf4051005f" . + "4bdcf84620019589541ff633ea89b6dbf958fb62ae9226bfeac34c639b3e18077bd935792ba63d5e" . + "352ec2b5be93be57f37a21097f2f06857bceed601ff2041a417f2177b81afb246fd079040af96512" . + "34ca24a1456ac11641c7e319114cff23f59bcc1bfa769a0e9fcdeab98429973e10caf303f2bcb065" . + "f22c1cc259556de8377431237da7082cf03ce8da9530be398022f0171d468d92fcabbe776a5e9cf2" . + "045642868406fd03ab735a70bfec3a951bb3c7a1de0fb3ff63cef23897e4fc3f9c5edf62fd45d058" . + "fedc7d2fb22ec928984a061053a7138ce0417b5512579a92be0775104c0bc911f68a5e8ede298102" . + "818100e4e17e2c752dbf1ad1025a074dc5f9c3c5989d23c84594313373d3e4ed0c0ddd74429ab026" . + "535c5e77549d888835bc94f069ebc5e77fdbd2ddf4c8be6cf777799a6d8d18e2b8cecfc13ab26df8" . + "b71ca3d94c2193c294042fb1025fdab38ba7aaeafebd8dd1f9d78ee67100693e99255dad6b964ebf" . + "b7401a03b67d412fabb33502818100cd65067097e307643df1fc8214db1dd7d09342ef01417a2620" . + "adad87352a58b8fcf07521289da3851623d8d045935fab7ecccc52ba0b86adcb92da76255e00289b" . + "af9aacd936201861b0021249f4ab5e6020db823af7171aef0bbbd02dc94d2489fc0b68500bd1b7d2" . + "81ed69fe4a44384161fe906e49bc91e0362b446ec2c521028180497662d40c2c49b966ba758100a2" . + "799f2f8de369f7bef568b1560cfdde63cf13745c685fff7d2419a1fd83aeade1698cf87956d6a78e" . + "2f55482e683c4ea7432ec1b545e365e9e15f676ada98578b166334bcadce4a56cddd2cd85141d5fd" . + "0e2cdace36b30d613ea1bc2f2aed9cccf4e4536443d334cfb180680eabb73f80c1bd0281803f30fc" . + "b93951a4dd875d62e5968b0f746d7c51147d5b6abc3e4390e6cf4997005af993dfbec23923e1fae7" . + "62b47531f2ee510defc9c3700d1a5bb510b2506856160801db79fc78056850a16285145c80edac4e" . + "3c93ed9f532f067a2303633273b26c340a44ce4e1873107c3da6f9ac616e643ad0aecdcad14a9cff" . + "d4cf0ae76102818100b82528a3dfc595cf9c6a025998491e3b4849c71aa8d1222ddb14af7f82fbe5" . + "169ec3ba18ec28d5a9501e95bc9da72cea99e4cdfdf898f40bec6b28f838243d2f39d7226e0873ed" . + "ee752bcae07639a4bd0eb31be1718c456391630b83ad0e9bf3fa18a645007e64fe59af467ea021f9" . + "e9a0dd759b21cd0b93333a73116abcaa2a" + ); + } + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::Rijndael", + key => $key, + iv => $salt_bin, + keysize => 24, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $enc_bin = $aes->encrypt ($data_bin); + + my $hash = sprintf ('$sshng$%d$%d$%s$%d$%s', $cid, length ($salt_bin), unpack ("H*", $salt_bin), length ($enc_bin), unpack ("H*", $enc_bin)); + + return $hash; +} + +sub module_verify_hash +{ + my $line = shift; + + my $idx = index ($line, ':'); + + return unless $idx >= 0; + + my $hash = substr ($line, 0, $idx); + my $word = substr ($line, $idx + 1); + + return unless substr ($hash, 0, 7) eq '$sshng$'; + + my (undef, $signature, $cid, undef, $salt, undef, $data) = split '\$', $hash; + + return unless defined $signature; + return unless defined $cid; + return unless defined $salt; + return unless defined $data; + + return unless ($signature eq 'sshng'); + + return unless ($cid == 4); + + my $word_packed = pack_if_HEX_notation ($word); + + my $new_hash = module_generate_hash ($word_packed, $salt, $cid, $data); + + return ($new_hash, $word); +} + +1; diff --git a/tools/test_modules/m22951.pm b/tools/test_modules/m22951.pm new file mode 100644 index 000000000..ad90582b0 --- /dev/null +++ b/tools/test_modules/m22951.pm @@ -0,0 +1,159 @@ +#!/usr/bin/env perl + +## +## Author......: See docs/credits.txt +## License.....: MIT +## + +use strict; +use warnings; + +use Crypt::CBC; +use Digest::MD5 qw (md5); + +sub module_constraints { [[0, 256], [32, 32], [-1, -1], [-1, -1], [-1, -1]] } + +sub generate_key +{ + my $word = shift; + my $salt = shift; + my $bytes = shift; + + my $salt8 = substr ($salt, 0, 8); + + my $out = md5 ($word . $salt8); + + $out .= md5 ($out . $word . $salt8); + + return substr ($out, 0, 32); +} + +sub module_generate_hash +{ + my $word = shift; + my $salt = shift; + my $cid = shift // 5; + my $data = shift; + + my $salt_bin = pack ("H*", $salt); + my $data_bin; + + my $key = generate_key ($word, $salt_bin); + + my $is_decrypt = defined ($data); + + if ($is_decrypt == 1) + { + $data_bin = pack ("H*", $data); + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::Rijndael", + key => $key, + iv => $salt_bin, + keysize => 32, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $dec_bin = $aes->decrypt ($data_bin); + + ## This is a ridiculous check of successfull decryption + ## There are no useable asn1 parsers for perl available + ## We have to rely on a combination of padding check and pattern matching + ## The (minimum) 16 bit should be good enough for a unit test + + if (length ($dec_bin) < length ($data_bin)) + { + if (substr ($dec_bin, 0, 1) eq "\x30") + { + $data_bin = $dec_bin; + } + } + } + else + { + $data_bin = pack ("H*", + "308204bd020100300d06092a864886f70d0101010500048204a7308204a30201000282010100b7a2" . + "e4c254c8174219e60d9cce96737a906797b8edb86af8f055f60db7bd298b0d31d7ce97ebeae393d5" . + "0e6da5215b58dcd72f4d3cac9e79b6ccaed7da47d2bd04f6a767f5ab7dc0f58beb6298c1e2358ed6" . + "d3ef441f2326ac5db0027e08ae6c7724ff9a2a220a07e97319b6eff5cd653c7ab8b6ea9f9e89a40a" . + "b856f036acfd39b1e5926964a024de35052de6d3423fe763569f48869c834750b28f09cecdddb54a" . + "5526a2c5159d22d24606a2af6c6f47a5d9c04c454896192b8e7b82cf6f6934a23d3495059cb7e43b" . + "98a20bd5b5782e15d93c8b289838c0a1df82ee429f0708d97aa40d6e75ec57ff12a2714871f241a8" . + "6f6d8d3472b084aeb748da33e2d50203010001028201004afab8dadc1122e5fb7b225dbf4051005f" . + "4bdcf84620019589541ff633ea89b6dbf958fb62ae9226bfeac34c639b3e18077bd935792ba63d5e" . + "352ec2b5be93be57f37a21097f2f06857bceed601ff2041a417f2177b81afb246fd079040af96512" . + "34ca24a1456ac11641c7e319114cff23f59bcc1bfa769a0e9fcdeab98429973e10caf303f2bcb065" . + "f22c1cc259556de8377431237da7082cf03ce8da9530be398022f0171d468d92fcabbe776a5e9cf2" . + "045642868406fd03ab735a70bfec3a951bb3c7a1de0fb3ff63cef23897e4fc3f9c5edf62fd45d058" . + "fedc7d2fb22ec928984a061053a7138ce0417b5512579a92be0775104c0bc911f68a5e8ede298102" . + "818100e4e17e2c752dbf1ad1025a074dc5f9c3c5989d23c84594313373d3e4ed0c0ddd74429ab026" . + "535c5e77549d888835bc94f069ebc5e77fdbd2ddf4c8be6cf777799a6d8d18e2b8cecfc13ab26df8" . + "b71ca3d94c2193c294042fb1025fdab38ba7aaeafebd8dd1f9d78ee67100693e99255dad6b964ebf" . + "b7401a03b67d412fabb33502818100cd65067097e307643df1fc8214db1dd7d09342ef01417a2620" . + "adad87352a58b8fcf07521289da3851623d8d045935fab7ecccc52ba0b86adcb92da76255e00289b" . + "af9aacd936201861b0021249f4ab5e6020db823af7171aef0bbbd02dc94d2489fc0b68500bd1b7d2" . + "81ed69fe4a44384161fe906e49bc91e0362b446ec2c521028180497662d40c2c49b966ba758100a2" . + "799f2f8de369f7bef568b1560cfdde63cf13745c685fff7d2419a1fd83aeade1698cf87956d6a78e" . + "2f55482e683c4ea7432ec1b545e365e9e15f676ada98578b166334bcadce4a56cddd2cd85141d5fd" . + "0e2cdace36b30d613ea1bc2f2aed9cccf4e4536443d334cfb180680eabb73f80c1bd0281803f30fc" . + "b93951a4dd875d62e5968b0f746d7c51147d5b6abc3e4390e6cf4997005af993dfbec23923e1fae7" . + "62b47531f2ee510defc9c3700d1a5bb510b2506856160801db79fc78056850a16285145c80edac4e" . + "3c93ed9f532f067a2303633273b26c340a44ce4e1873107c3da6f9ac616e643ad0aecdcad14a9cff" . + "d4cf0ae76102818100b82528a3dfc595cf9c6a025998491e3b4849c71aa8d1222ddb14af7f82fbe5" . + "169ec3ba18ec28d5a9501e95bc9da72cea99e4cdfdf898f40bec6b28f838243d2f39d7226e0873ed" . + "ee752bcae07639a4bd0eb31be1718c456391630b83ad0e9bf3fa18a645007e64fe59af467ea021f9" . + "e9a0dd759b21cd0b93333a73116abcaa2a" + ); + } + + my $aes = Crypt::CBC->new ({ + cipher => "Crypt::Rijndael", + key => $key, + iv => $salt_bin, + keysize => 32, + literal_key => 1, + header => "none", + padding => "standard", + }); + + my $enc_bin = $aes->encrypt ($data_bin); + + my $hash = sprintf ('$sshng$%d$%d$%s$%d$%s', $cid, length ($salt_bin), unpack ("H*", $salt_bin), length ($enc_bin), unpack ("H*", $enc_bin)); + + return $hash; +} + +sub module_verify_hash +{ + my $line = shift; + + my $idx = index ($line, ':'); + + return unless $idx >= 0; + + my $hash = substr ($line, 0, $idx); + my $word = substr ($line, $idx + 1); + + return unless substr ($hash, 0, 7) eq '$sshng$'; + + my (undef, $signature, $cid, undef, $salt, undef, $data) = split '\$', $hash; + + return unless defined $signature; + return unless defined $cid; + return unless defined $salt; + return unless defined $data; + + return unless ($signature eq 'sshng'); + + return unless ($cid == 5); + + my $word_packed = pack_if_HEX_notation ($word); + + my $new_hash = module_generate_hash ($word_packed, $salt, $cid, $data); + + return ($new_hash, $word); +} + +1;