From 907b065e00913a0f4f55a37ed47ef7efa381e025 Mon Sep 17 00:00:00 2001 From: jsteube Date: Tue, 4 Jul 2017 12:19:40 +0200 Subject: [PATCH] Vectorized TrueCrypt PBKDF2-HMAC-SHA512 and added support for long passwords --- OpenCL/inc_types.cl | 2 +- OpenCL/m06221.cl | 735 +++++++++++++++++++----------------------- OpenCL/m06222.cl | 703 ++++++++++++++++++---------------------- OpenCL/m06223.cl | 756 ++++++++++++++++++-------------------------- include/interface.h | 2 +- src/interface.c | 21 +- 6 files changed, 941 insertions(+), 1278 deletions(-) diff --git a/OpenCL/inc_types.cl b/OpenCL/inc_types.cl index 848e73b86..69208e795 100644 --- a/OpenCL/inc_types.cl +++ b/OpenCL/inc_types.cl @@ -885,7 +885,7 @@ typedef struct krb5tgs typedef struct tc { - u32 salt_buf[16]; + u32 salt_buf[32]; u32 data_buf[112]; u32 keyfile_buf[16]; u32 signature; diff --git a/OpenCL/m06221.cl b/OpenCL/m06221.cl index 4de488b9e..bc2f23b10 100644 --- a/OpenCL/m06221.cl +++ b/OpenCL/m06221.cl @@ -3,11 +3,15 @@ * License.....: MIT */ +#define NEW_SIMD_CODE + #include "inc_vendor.cl" #include "inc_hash_constants.h" #include "inc_hash_functions.cl" #include "inc_types.cl" #include "inc_common.cl" +#include "inc_simd.cl" +#include "inc_hash_sha512.cl" #include "inc_cipher_aes.cl" #include "inc_cipher_twofish.cl" @@ -16,221 +20,6 @@ #include "inc_truecrypt_crc32.cl" #include "inc_truecrypt_xts.cl" -__constant u64a k_sha512[80] = -{ - SHA512C00, SHA512C01, SHA512C02, SHA512C03, - SHA512C04, SHA512C05, SHA512C06, SHA512C07, - SHA512C08, SHA512C09, SHA512C0a, SHA512C0b, - SHA512C0c, SHA512C0d, SHA512C0e, SHA512C0f, - SHA512C10, SHA512C11, SHA512C12, SHA512C13, - SHA512C14, SHA512C15, SHA512C16, SHA512C17, - SHA512C18, SHA512C19, SHA512C1a, SHA512C1b, - SHA512C1c, SHA512C1d, SHA512C1e, SHA512C1f, - SHA512C20, SHA512C21, SHA512C22, SHA512C23, - SHA512C24, SHA512C25, SHA512C26, SHA512C27, - SHA512C28, SHA512C29, SHA512C2a, SHA512C2b, - SHA512C2c, SHA512C2d, SHA512C2e, SHA512C2f, - SHA512C30, SHA512C31, SHA512C32, SHA512C33, - SHA512C34, SHA512C35, SHA512C36, SHA512C37, - SHA512C38, SHA512C39, SHA512C3a, SHA512C3b, - SHA512C3c, SHA512C3d, SHA512C3e, SHA512C3f, - SHA512C40, SHA512C41, SHA512C42, SHA512C43, - SHA512C44, SHA512C45, SHA512C46, SHA512C47, - SHA512C48, SHA512C49, SHA512C4a, SHA512C4b, - SHA512C4c, SHA512C4d, SHA512C4e, SHA512C4f, -}; - -void sha512_transform (const u64 w[16], u64 dgst[8]) -{ - u64 a = dgst[0]; - u64 b = dgst[1]; - u64 c = dgst[2]; - u64 d = dgst[3]; - u64 e = dgst[4]; - u64 f = dgst[5]; - u64 g = dgst[6]; - u64 h = dgst[7]; - - u64 w0_t = w[ 0]; - u64 w1_t = w[ 1]; - u64 w2_t = w[ 2]; - u64 w3_t = w[ 3]; - u64 w4_t = w[ 4]; - u64 w5_t = w[ 5]; - u64 w6_t = w[ 6]; - u64 w7_t = w[ 7]; - u64 w8_t = w[ 8]; - u64 w9_t = w[ 9]; - u64 wa_t = w[10]; - u64 wb_t = w[11]; - u64 wc_t = w[12]; - u64 wd_t = w[13]; - u64 we_t = w[14]; - u64 wf_t = w[15]; - - #define ROUND_EXPAND() \ - { \ - w0_t = SHA512_EXPAND (we_t, w9_t, w1_t, w0_t); \ - w1_t = SHA512_EXPAND (wf_t, wa_t, w2_t, w1_t); \ - w2_t = SHA512_EXPAND (w0_t, wb_t, w3_t, w2_t); \ - w3_t = SHA512_EXPAND (w1_t, wc_t, w4_t, w3_t); \ - w4_t = SHA512_EXPAND (w2_t, wd_t, w5_t, w4_t); \ - w5_t = SHA512_EXPAND (w3_t, we_t, w6_t, w5_t); \ - w6_t = SHA512_EXPAND (w4_t, wf_t, w7_t, w6_t); \ - w7_t = SHA512_EXPAND (w5_t, w0_t, w8_t, w7_t); \ - w8_t = SHA512_EXPAND (w6_t, w1_t, w9_t, w8_t); \ - w9_t = SHA512_EXPAND (w7_t, w2_t, wa_t, w9_t); \ - wa_t = SHA512_EXPAND (w8_t, w3_t, wb_t, wa_t); \ - wb_t = SHA512_EXPAND (w9_t, w4_t, wc_t, wb_t); \ - wc_t = SHA512_EXPAND (wa_t, w5_t, wd_t, wc_t); \ - wd_t = SHA512_EXPAND (wb_t, w6_t, we_t, wd_t); \ - we_t = SHA512_EXPAND (wc_t, w7_t, wf_t, we_t); \ - wf_t = SHA512_EXPAND (wd_t, w8_t, w0_t, wf_t); \ - } - - #define ROUND_STEP(i) \ - { \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha512[i + 0]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha512[i + 1]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha512[i + 2]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha512[i + 3]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha512[i + 4]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha512[i + 5]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha512[i + 6]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha512[i + 7]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha512[i + 8]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha512[i + 9]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha512[i + 10]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha512[i + 11]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha512[i + 12]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha512[i + 13]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, we_t, k_sha512[i + 14]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha512[i + 15]); \ - } - - ROUND_STEP (0); - - #ifdef _unroll - #pragma unroll - #endif - for (int i = 16; i < 80; i += 16) - { - ROUND_EXPAND (); ROUND_STEP (i); - } - - dgst[0] += a; - dgst[1] += b; - dgst[2] += c; - dgst[3] += d; - dgst[4] += e; - dgst[5] += f; - dgst[6] += g; - dgst[7] += h; -} - -void hmac_run (const u64 w1[16], const u64 ipad[8], const u64 opad[8], u64 dgst[8]) -{ - dgst[0] = ipad[0]; - dgst[1] = ipad[1]; - dgst[2] = ipad[2]; - dgst[3] = ipad[3]; - dgst[4] = ipad[4]; - dgst[5] = ipad[5]; - dgst[6] = ipad[6]; - dgst[7] = ipad[7]; - - sha512_transform (w1, dgst); - - u64 w[16]; - - w[ 0] = dgst[0]; - w[ 1] = dgst[1]; - w[ 2] = dgst[2]; - w[ 3] = dgst[3]; - w[ 4] = dgst[4]; - w[ 5] = dgst[5]; - w[ 6] = dgst[6]; - w[ 7] = dgst[7]; - w[ 8] = 0x8000000000000000; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = (128 + 64) * 8; - - dgst[0] = opad[0]; - dgst[1] = opad[1]; - dgst[2] = opad[2]; - dgst[3] = opad[3]; - dgst[4] = opad[4]; - dgst[5] = opad[5]; - dgst[6] = opad[6]; - dgst[7] = opad[7]; - - sha512_transform (w, dgst); -} - -void hmac_init (u64 w[16], u64 ipad[8], u64 opad[8]) -{ - w[ 0] ^= 0x3636363636363636; - w[ 1] ^= 0x3636363636363636; - w[ 2] ^= 0x3636363636363636; - w[ 3] ^= 0x3636363636363636; - w[ 4] ^= 0x3636363636363636; - w[ 5] ^= 0x3636363636363636; - w[ 6] ^= 0x3636363636363636; - w[ 7] ^= 0x3636363636363636; - w[ 8] ^= 0x3636363636363636; - w[ 9] ^= 0x3636363636363636; - w[10] ^= 0x3636363636363636; - w[11] ^= 0x3636363636363636; - w[12] ^= 0x3636363636363636; - w[13] ^= 0x3636363636363636; - w[14] ^= 0x3636363636363636; - w[15] ^= 0x3636363636363636; - - ipad[0] = SHA512M_A; - ipad[1] = SHA512M_B; - ipad[2] = SHA512M_C; - ipad[3] = SHA512M_D; - ipad[4] = SHA512M_E; - ipad[5] = SHA512M_F; - ipad[6] = SHA512M_G; - ipad[7] = SHA512M_H; - - sha512_transform (w, ipad); - - w[ 0] ^= 0x6a6a6a6a6a6a6a6a; - w[ 1] ^= 0x6a6a6a6a6a6a6a6a; - w[ 2] ^= 0x6a6a6a6a6a6a6a6a; - w[ 3] ^= 0x6a6a6a6a6a6a6a6a; - w[ 4] ^= 0x6a6a6a6a6a6a6a6a; - w[ 5] ^= 0x6a6a6a6a6a6a6a6a; - w[ 6] ^= 0x6a6a6a6a6a6a6a6a; - w[ 7] ^= 0x6a6a6a6a6a6a6a6a; - w[ 8] ^= 0x6a6a6a6a6a6a6a6a; - w[ 9] ^= 0x6a6a6a6a6a6a6a6a; - w[10] ^= 0x6a6a6a6a6a6a6a6a; - w[11] ^= 0x6a6a6a6a6a6a6a6a; - w[12] ^= 0x6a6a6a6a6a6a6a6a; - w[13] ^= 0x6a6a6a6a6a6a6a6a; - w[14] ^= 0x6a6a6a6a6a6a6a6a; - w[15] ^= 0x6a6a6a6a6a6a6a6a; - - opad[0] = SHA512M_A; - opad[1] = SHA512M_B; - opad[2] = SHA512M_C; - opad[3] = SHA512M_D; - opad[4] = SHA512M_E; - opad[5] = SHA512M_F; - opad[6] = SHA512M_G; - opad[7] = SHA512M_H; - - sha512_transform (w, opad); -} - u32 u8add (const u32 a, const u32 b) { const u32 a1 = (a >> 0) & 0xff; @@ -249,13 +38,71 @@ u32 u8add (const u32 a, const u32 b) const u32 r4 = (a4 + b4) & 0xff; const u32 r = r1 << 0 - | r2 << 8 - | r3 << 16 - | r4 << 24; + | r2 << 8 + | r3 << 16 + | r4 << 24; return r; } +void hmac_sha512_run_V (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], u64x ipad[8], u64x opad[8], u64x digest[8]) +{ + digest[0] = ipad[0]; + digest[1] = ipad[1]; + digest[2] = ipad[2]; + digest[3] = ipad[3]; + digest[4] = ipad[4]; + digest[5] = ipad[5]; + digest[6] = ipad[6]; + digest[7] = ipad[7]; + + sha512_transform_vector (w0, w1, w2, w3, w4, w5, w6, w7, digest); + + w0[0] = h32_from_64 (digest[0]); + w0[1] = l32_from_64 (digest[0]); + w0[2] = h32_from_64 (digest[1]); + w0[3] = l32_from_64 (digest[1]); + w1[0] = h32_from_64 (digest[2]); + w1[1] = l32_from_64 (digest[2]); + w1[2] = h32_from_64 (digest[3]); + w1[3] = l32_from_64 (digest[3]); + w2[0] = h32_from_64 (digest[4]); + w2[1] = l32_from_64 (digest[4]); + w2[2] = h32_from_64 (digest[5]); + w2[3] = l32_from_64 (digest[5]); + w3[0] = h32_from_64 (digest[6]); + w3[1] = l32_from_64 (digest[6]); + w3[2] = h32_from_64 (digest[7]); + w3[3] = l32_from_64 (digest[7]); + w4[0] = 0x80000000; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = (128 + 64) * 8; + + digest[0] = opad[0]; + digest[1] = opad[1]; + digest[2] = opad[2]; + digest[3] = opad[3]; + digest[4] = opad[4]; + digest[5] = opad[5]; + digest[6] = opad[6]; + digest[7] = opad[7]; + + sha512_transform_vector (w0, w1, w2, w3, w4, w5, w6, w7, digest); +} + __kernel void m06221_init (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global tc64_tmp_t *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global tc_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max) { /** @@ -267,36 +114,46 @@ __kernel void m06221_init (__global pw_t *pws, __global const kernel_rule_t *rul if (gid >= gid_max) return; u32 w0[4]; + u32 w1[4]; + u32 w2[4]; + u32 w3[4]; + u32 w4[4]; + u32 w5[4]; + u32 w6[4]; + u32 w7[4]; w0[0] = pws[gid].i[ 0]; w0[1] = pws[gid].i[ 1]; w0[2] = pws[gid].i[ 2]; w0[3] = pws[gid].i[ 3]; - - u32 w1[4]; - w1[0] = pws[gid].i[ 4]; w1[1] = pws[gid].i[ 5]; w1[2] = pws[gid].i[ 6]; w1[3] = pws[gid].i[ 7]; - - u32 w2[4]; - w2[0] = pws[gid].i[ 8]; w2[1] = pws[gid].i[ 9]; w2[2] = pws[gid].i[10]; w2[3] = pws[gid].i[11]; - - u32 w3[4]; - w3[0] = pws[gid].i[12]; w3[1] = pws[gid].i[13]; w3[2] = pws[gid].i[14]; w3[3] = pws[gid].i[15]; - - /** - * keyfile - */ + w4[0] = pws[gid].i[16]; + w4[1] = pws[gid].i[17]; + w4[2] = pws[gid].i[18]; + w4[3] = pws[gid].i[19]; + w5[0] = pws[gid].i[20]; + w5[1] = pws[gid].i[21]; + w5[2] = pws[gid].i[22]; + w5[3] = pws[gid].i[23]; + w6[0] = pws[gid].i[24]; + w6[1] = pws[gid].i[25]; + w6[2] = pws[gid].i[26]; + w6[3] = pws[gid].i[27]; + w7[0] = pws[gid].i[28]; + w7[1] = pws[gid].i[29]; + w7[2] = pws[gid].i[30]; + w7[3] = pws[gid].i[31]; w0[0] = u8add (w0[0], esalt_bufs[digests_offset].keyfile_buf[ 0]); w0[1] = u8add (w0[1], esalt_bufs[digests_offset].keyfile_buf[ 1]); @@ -315,98 +172,121 @@ __kernel void m06221_init (__global pw_t *pws, __global const kernel_rule_t *rul w3[2] = u8add (w3[2], esalt_bufs[digests_offset].keyfile_buf[14]); w3[3] = u8add (w3[3], esalt_bufs[digests_offset].keyfile_buf[15]); - /** - * salt - */ - - u64 salt_buf[16]; - - // swap fehlt - - salt_buf[ 0] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 0])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 1]); - salt_buf[ 1] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 2])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 3]); - salt_buf[ 2] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 4])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 5]); - salt_buf[ 3] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 6])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 7]); - salt_buf[ 4] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 8])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 9]); - salt_buf[ 5] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[10])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[11]); - salt_buf[ 6] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[12])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[13]); - salt_buf[ 7] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[14])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[15]); - salt_buf[ 8] = 0; - salt_buf[ 9] = 0; - salt_buf[10] = 0; - salt_buf[11] = 0; - salt_buf[12] = 0; - salt_buf[13] = 0; - salt_buf[14] = 0; - salt_buf[15] = (128 + 64 + 4) * 8; - - u64 w[16]; - - w[ 0] = ((u64) swap32 (w0[0])) << 32 | (u64) swap32 (w0[1]); - w[ 1] = ((u64) swap32 (w0[2])) << 32 | (u64) swap32 (w0[3]); - w[ 2] = ((u64) swap32 (w1[0])) << 32 | (u64) swap32 (w1[1]); - w[ 3] = ((u64) swap32 (w1[2])) << 32 | (u64) swap32 (w1[3]); - w[ 4] = ((u64) swap32 (w2[0])) << 32 | (u64) swap32 (w2[1]); - w[ 5] = ((u64) swap32 (w2[2])) << 32 | (u64) swap32 (w2[3]); - w[ 6] = ((u64) swap32 (w3[0])) << 32 | (u64) swap32 (w3[1]); - w[ 7] = ((u64) swap32 (w3[2])) << 32 | (u64) swap32 (w3[3]); - w[ 8] = 0; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = 0; - - u64 ipad[8]; - u64 opad[8]; - - hmac_init (w, ipad, opad); - - tmps[gid].ipad[0] = ipad[0]; - tmps[gid].ipad[1] = ipad[1]; - tmps[gid].ipad[2] = ipad[2]; - tmps[gid].ipad[3] = ipad[3]; - tmps[gid].ipad[4] = ipad[4]; - tmps[gid].ipad[5] = ipad[5]; - tmps[gid].ipad[6] = ipad[6]; - tmps[gid].ipad[7] = ipad[7]; - - tmps[gid].opad[0] = opad[0]; - tmps[gid].opad[1] = opad[1]; - tmps[gid].opad[2] = opad[2]; - tmps[gid].opad[3] = opad[3]; - tmps[gid].opad[4] = opad[4]; - tmps[gid].opad[5] = opad[5]; - tmps[gid].opad[6] = opad[6]; - tmps[gid].opad[7] = opad[7]; + w0[0] = swap32_S (w0[0]); + w0[1] = swap32_S (w0[1]); + w0[2] = swap32_S (w0[2]); + w0[3] = swap32_S (w0[3]); + w1[0] = swap32_S (w1[0]); + w1[1] = swap32_S (w1[1]); + w1[2] = swap32_S (w1[2]); + w1[3] = swap32_S (w1[3]); + w2[0] = swap32_S (w2[0]); + w2[1] = swap32_S (w2[1]); + w2[2] = swap32_S (w2[2]); + w2[3] = swap32_S (w2[3]); + w3[0] = swap32_S (w3[0]); + w3[1] = swap32_S (w3[1]); + w3[2] = swap32_S (w3[2]); + w3[3] = swap32_S (w3[3]); + w4[0] = swap32_S (w4[0]); + w4[1] = swap32_S (w4[1]); + w4[2] = swap32_S (w4[2]); + w4[3] = swap32_S (w4[3]); + w5[0] = swap32_S (w5[0]); + w5[1] = swap32_S (w5[1]); + w5[2] = swap32_S (w5[2]); + w5[3] = swap32_S (w5[3]); + w6[0] = swap32_S (w6[0]); + w6[1] = swap32_S (w6[1]); + w6[2] = swap32_S (w6[2]); + w6[3] = swap32_S (w6[3]); + w7[0] = swap32_S (w7[0]); + w7[1] = swap32_S (w7[1]); + w7[2] = swap32_S (w7[2]); + w7[3] = swap32_S (w7[3]); + + sha512_hmac_ctx_t sha512_hmac_ctx; + + sha512_hmac_init (&sha512_hmac_ctx, w0, w1, w2, w3, w5, w5, w6, w7); + + tmps[gid].ipad[0] = sha512_hmac_ctx.ipad.h[0]; + tmps[gid].ipad[1] = sha512_hmac_ctx.ipad.h[1]; + tmps[gid].ipad[2] = sha512_hmac_ctx.ipad.h[2]; + tmps[gid].ipad[3] = sha512_hmac_ctx.ipad.h[3]; + tmps[gid].ipad[4] = sha512_hmac_ctx.ipad.h[4]; + tmps[gid].ipad[5] = sha512_hmac_ctx.ipad.h[5]; + tmps[gid].ipad[6] = sha512_hmac_ctx.ipad.h[6]; + tmps[gid].ipad[7] = sha512_hmac_ctx.ipad.h[7]; + + tmps[gid].opad[0] = sha512_hmac_ctx.opad.h[0]; + tmps[gid].opad[1] = sha512_hmac_ctx.opad.h[1]; + tmps[gid].opad[2] = sha512_hmac_ctx.opad.h[2]; + tmps[gid].opad[3] = sha512_hmac_ctx.opad.h[3]; + tmps[gid].opad[4] = sha512_hmac_ctx.opad.h[4]; + tmps[gid].opad[5] = sha512_hmac_ctx.opad.h[5]; + tmps[gid].opad[6] = sha512_hmac_ctx.opad.h[6]; + tmps[gid].opad[7] = sha512_hmac_ctx.opad.h[7]; + + sha512_hmac_update_global_swap (&sha512_hmac_ctx, esalt_bufs[digests_offset].salt_buf, 64); for (u32 i = 0, j = 1; i < 8; i += 8, j += 1) { - salt_buf[8] = (u64) j << 32 | (u64) 0x80000000; - - u64 dgst[8]; - - hmac_run (salt_buf, ipad, opad, dgst); - - tmps[gid].dgst[i + 0] = dgst[0]; - tmps[gid].dgst[i + 1] = dgst[1]; - tmps[gid].dgst[i + 2] = dgst[2]; - tmps[gid].dgst[i + 3] = dgst[3]; - tmps[gid].dgst[i + 4] = dgst[4]; - tmps[gid].dgst[i + 5] = dgst[5]; - tmps[gid].dgst[i + 6] = dgst[6]; - tmps[gid].dgst[i + 7] = dgst[7]; - - tmps[gid].out[i + 0] = dgst[0]; - tmps[gid].out[i + 1] = dgst[1]; - tmps[gid].out[i + 2] = dgst[2]; - tmps[gid].out[i + 3] = dgst[3]; - tmps[gid].out[i + 4] = dgst[4]; - tmps[gid].out[i + 5] = dgst[5]; - tmps[gid].out[i + 6] = dgst[6]; - tmps[gid].out[i + 7] = dgst[7]; + sha512_hmac_ctx_t sha512_hmac_ctx2 = sha512_hmac_ctx; + + w0[0] = j; + w0[1] = 0; + w0[2] = 0; + w0[3] = 0; + w1[0] = 0; + w1[1] = 0; + w1[2] = 0; + w1[3] = 0; + w2[0] = 0; + w2[1] = 0; + w2[2] = 0; + w2[3] = 0; + w3[0] = 0; + w3[1] = 0; + w3[2] = 0; + w3[3] = 0; + w4[0] = 0; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = 0; + + sha512_hmac_update_128 (&sha512_hmac_ctx2, w0, w1, w2, w3, w4, w5, w6, w7, 4); + + sha512_hmac_final (&sha512_hmac_ctx2); + + tmps[gid].dgst[i + 0] = sha512_hmac_ctx2.opad.h[0]; + tmps[gid].dgst[i + 1] = sha512_hmac_ctx2.opad.h[1]; + tmps[gid].dgst[i + 2] = sha512_hmac_ctx2.opad.h[2]; + tmps[gid].dgst[i + 3] = sha512_hmac_ctx2.opad.h[3]; + tmps[gid].dgst[i + 4] = sha512_hmac_ctx2.opad.h[4]; + tmps[gid].dgst[i + 5] = sha512_hmac_ctx2.opad.h[5]; + tmps[gid].dgst[i + 6] = sha512_hmac_ctx2.opad.h[6]; + tmps[gid].dgst[i + 7] = sha512_hmac_ctx2.opad.h[7]; + + tmps[gid].out[i + 0] = tmps[gid].dgst[i + 0]; + tmps[gid].out[i + 1] = tmps[gid].dgst[i + 1]; + tmps[gid].out[i + 2] = tmps[gid].dgst[i + 2]; + tmps[gid].out[i + 3] = tmps[gid].dgst[i + 3]; + tmps[gid].out[i + 4] = tmps[gid].dgst[i + 4]; + tmps[gid].out[i + 5] = tmps[gid].dgst[i + 5]; + tmps[gid].out[i + 6] = tmps[gid].dgst[i + 6]; + tmps[gid].out[i + 7] = tmps[gid].dgst[i + 7]; } } @@ -414,76 +294,97 @@ __kernel void m06221_loop (__global pw_t *pws, __global const kernel_rule_t *rul { const u32 gid = get_global_id (0); - if (gid >= gid_max) return; - - u64 ipad[8]; - - ipad[0] = tmps[gid].ipad[0]; - ipad[1] = tmps[gid].ipad[1]; - ipad[2] = tmps[gid].ipad[2]; - ipad[3] = tmps[gid].ipad[3]; - ipad[4] = tmps[gid].ipad[4]; - ipad[5] = tmps[gid].ipad[5]; - ipad[6] = tmps[gid].ipad[6]; - ipad[7] = tmps[gid].ipad[7]; - - u64 opad[8]; - - opad[0] = tmps[gid].opad[0]; - opad[1] = tmps[gid].opad[1]; - opad[2] = tmps[gid].opad[2]; - opad[3] = tmps[gid].opad[3]; - opad[4] = tmps[gid].opad[4]; - opad[5] = tmps[gid].opad[5]; - opad[6] = tmps[gid].opad[6]; - opad[7] = tmps[gid].opad[7]; + if ((gid * VECT_SIZE) >= gid_max) return; + + u64x ipad[8]; + u64x opad[8]; + + ipad[0] = pack64v (tmps, ipad, gid, 0); + ipad[1] = pack64v (tmps, ipad, gid, 1); + ipad[2] = pack64v (tmps, ipad, gid, 2); + ipad[3] = pack64v (tmps, ipad, gid, 3); + ipad[4] = pack64v (tmps, ipad, gid, 4); + ipad[5] = pack64v (tmps, ipad, gid, 5); + ipad[6] = pack64v (tmps, ipad, gid, 6); + ipad[7] = pack64v (tmps, ipad, gid, 7); + + opad[0] = pack64v (tmps, opad, gid, 0); + opad[1] = pack64v (tmps, opad, gid, 1); + opad[2] = pack64v (tmps, opad, gid, 2); + opad[3] = pack64v (tmps, opad, gid, 3); + opad[4] = pack64v (tmps, opad, gid, 4); + opad[5] = pack64v (tmps, opad, gid, 5); + opad[6] = pack64v (tmps, opad, gid, 6); + opad[7] = pack64v (tmps, opad, gid, 7); for (u32 i = 0; i < 8; i += 8) { - u64 dgst[8]; - - dgst[0] = tmps[gid].dgst[i + 0]; - dgst[1] = tmps[gid].dgst[i + 1]; - dgst[2] = tmps[gid].dgst[i + 2]; - dgst[3] = tmps[gid].dgst[i + 3]; - dgst[4] = tmps[gid].dgst[i + 4]; - dgst[5] = tmps[gid].dgst[i + 5]; - dgst[6] = tmps[gid].dgst[i + 6]; - dgst[7] = tmps[gid].dgst[i + 7]; - - u64 out[8]; - - out[0] = tmps[gid].out[i + 0]; - out[1] = tmps[gid].out[i + 1]; - out[2] = tmps[gid].out[i + 2]; - out[3] = tmps[gid].out[i + 3]; - out[4] = tmps[gid].out[i + 4]; - out[5] = tmps[gid].out[i + 5]; - out[6] = tmps[gid].out[i + 6]; - out[7] = tmps[gid].out[i + 7]; + u64x dgst[8]; + u64x out[8]; + + dgst[0] = pack64v (tmps, dgst, gid, i + 0); + dgst[1] = pack64v (tmps, dgst, gid, i + 1); + dgst[2] = pack64v (tmps, dgst, gid, i + 2); + dgst[3] = pack64v (tmps, dgst, gid, i + 3); + dgst[4] = pack64v (tmps, dgst, gid, i + 4); + dgst[5] = pack64v (tmps, dgst, gid, i + 5); + dgst[6] = pack64v (tmps, dgst, gid, i + 6); + dgst[7] = pack64v (tmps, dgst, gid, i + 7); + + out[0] = pack64v (tmps, out, gid, i + 0); + out[1] = pack64v (tmps, out, gid, i + 1); + out[2] = pack64v (tmps, out, gid, i + 2); + out[3] = pack64v (tmps, out, gid, i + 3); + out[4] = pack64v (tmps, out, gid, i + 4); + out[5] = pack64v (tmps, out, gid, i + 5); + out[6] = pack64v (tmps, out, gid, i + 6); + out[7] = pack64v (tmps, out, gid, i + 7); for (u32 j = 0; j < loop_cnt; j++) { - u64 w[16]; - - w[ 0] = dgst[0]; - w[ 1] = dgst[1]; - w[ 2] = dgst[2]; - w[ 3] = dgst[3]; - w[ 4] = dgst[4]; - w[ 5] = dgst[5]; - w[ 6] = dgst[6]; - w[ 7] = dgst[7]; - w[ 8] = 0x8000000000000000; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = (128 + 64) * 8; - - hmac_run (w, ipad, opad, dgst); + u32x w0[4]; + u32x w1[4]; + u32x w2[4]; + u32x w3[4]; + u32x w4[4]; + u32x w5[4]; + u32x w6[4]; + u32x w7[4]; + + w0[0] = h32_from_64 (dgst[0]); + w0[1] = l32_from_64 (dgst[0]); + w0[2] = h32_from_64 (dgst[1]); + w0[3] = l32_from_64 (dgst[1]); + w1[0] = h32_from_64 (dgst[2]); + w1[1] = l32_from_64 (dgst[2]); + w1[2] = h32_from_64 (dgst[3]); + w1[3] = l32_from_64 (dgst[3]); + w2[0] = h32_from_64 (dgst[4]); + w2[1] = l32_from_64 (dgst[4]); + w2[2] = h32_from_64 (dgst[5]); + w2[3] = l32_from_64 (dgst[5]); + w3[0] = h32_from_64 (dgst[6]); + w3[1] = l32_from_64 (dgst[6]); + w3[2] = h32_from_64 (dgst[7]); + w3[3] = l32_from_64 (dgst[7]); + w4[0] = 0x80000000; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = (128 + 64) * 8; + + hmac_sha512_run_V (w0, w1, w2, w3, w4, w5, w6, w7, ipad, opad, dgst); out[0] ^= dgst[0]; out[1] ^= dgst[1]; @@ -495,23 +396,23 @@ __kernel void m06221_loop (__global pw_t *pws, __global const kernel_rule_t *rul out[7] ^= dgst[7]; } - tmps[gid].dgst[i + 0] = dgst[0]; - tmps[gid].dgst[i + 1] = dgst[1]; - tmps[gid].dgst[i + 2] = dgst[2]; - tmps[gid].dgst[i + 3] = dgst[3]; - tmps[gid].dgst[i + 4] = dgst[4]; - tmps[gid].dgst[i + 5] = dgst[5]; - tmps[gid].dgst[i + 6] = dgst[6]; - tmps[gid].dgst[i + 7] = dgst[7]; - - tmps[gid].out[i + 0] = out[0]; - tmps[gid].out[i + 1] = out[1]; - tmps[gid].out[i + 2] = out[2]; - tmps[gid].out[i + 3] = out[3]; - tmps[gid].out[i + 4] = out[4]; - tmps[gid].out[i + 5] = out[5]; - tmps[gid].out[i + 6] = out[6]; - tmps[gid].out[i + 7] = out[7]; + unpack64v (tmps, dgst, gid, i + 0, dgst[0]); + unpack64v (tmps, dgst, gid, i + 1, dgst[1]); + unpack64v (tmps, dgst, gid, i + 2, dgst[2]); + unpack64v (tmps, dgst, gid, i + 3, dgst[3]); + unpack64v (tmps, dgst, gid, i + 4, dgst[4]); + unpack64v (tmps, dgst, gid, i + 5, dgst[5]); + unpack64v (tmps, dgst, gid, i + 6, dgst[6]); + unpack64v (tmps, dgst, gid, i + 7, dgst[7]); + + unpack64v (tmps, out, gid, i + 0, out[0]); + unpack64v (tmps, out, gid, i + 1, out[1]); + unpack64v (tmps, out, gid, i + 2, out[2]); + unpack64v (tmps, out, gid, i + 3, out[3]); + unpack64v (tmps, out, gid, i + 4, out[4]); + unpack64v (tmps, out, gid, i + 5, out[5]); + unpack64v (tmps, out, gid, i + 6, out[6]); + unpack64v (tmps, out, gid, i + 7, out[7]); } } @@ -576,25 +477,25 @@ __kernel void m06221_comp (__global pw_t *pws, __global const kernel_rule_t *rul u32 ukey1[8]; - ukey1[0] = swap32 (h32_from_64 (tmps[gid].out[ 0])); - ukey1[1] = swap32 (l32_from_64 (tmps[gid].out[ 0])); - ukey1[2] = swap32 (h32_from_64 (tmps[gid].out[ 1])); - ukey1[3] = swap32 (l32_from_64 (tmps[gid].out[ 1])); - ukey1[4] = swap32 (h32_from_64 (tmps[gid].out[ 2])); - ukey1[5] = swap32 (l32_from_64 (tmps[gid].out[ 2])); - ukey1[6] = swap32 (h32_from_64 (tmps[gid].out[ 3])); - ukey1[7] = swap32 (l32_from_64 (tmps[gid].out[ 3])); + ukey1[0] = swap32_S (h32_from_64 (tmps[gid].out[ 0])); + ukey1[1] = swap32_S (l32_from_64 (tmps[gid].out[ 0])); + ukey1[2] = swap32_S (h32_from_64 (tmps[gid].out[ 1])); + ukey1[3] = swap32_S (l32_from_64 (tmps[gid].out[ 1])); + ukey1[4] = swap32_S (h32_from_64 (tmps[gid].out[ 2])); + ukey1[5] = swap32_S (l32_from_64 (tmps[gid].out[ 2])); + ukey1[6] = swap32_S (h32_from_64 (tmps[gid].out[ 3])); + ukey1[7] = swap32_S (l32_from_64 (tmps[gid].out[ 3])); u32 ukey2[8]; - ukey2[0] = swap32 (h32_from_64 (tmps[gid].out[ 4])); - ukey2[1] = swap32 (l32_from_64 (tmps[gid].out[ 4])); - ukey2[2] = swap32 (h32_from_64 (tmps[gid].out[ 5])); - ukey2[3] = swap32 (l32_from_64 (tmps[gid].out[ 5])); - ukey2[4] = swap32 (h32_from_64 (tmps[gid].out[ 6])); - ukey2[5] = swap32 (l32_from_64 (tmps[gid].out[ 6])); - ukey2[6] = swap32 (h32_from_64 (tmps[gid].out[ 7])); - ukey2[7] = swap32 (l32_from_64 (tmps[gid].out[ 7])); + ukey2[0] = swap32_S (h32_from_64 (tmps[gid].out[ 4])); + ukey2[1] = swap32_S (l32_from_64 (tmps[gid].out[ 4])); + ukey2[2] = swap32_S (h32_from_64 (tmps[gid].out[ 5])); + ukey2[3] = swap32_S (l32_from_64 (tmps[gid].out[ 5])); + ukey2[4] = swap32_S (h32_from_64 (tmps[gid].out[ 6])); + ukey2[5] = swap32_S (l32_from_64 (tmps[gid].out[ 6])); + ukey2[6] = swap32_S (h32_from_64 (tmps[gid].out[ 7])); + ukey2[7] = swap32_S (l32_from_64 (tmps[gid].out[ 7])); if (verify_header_aes (esalt_bufs, ukey1, ukey2, s_te0, s_te1, s_te2, s_te3, s_te4, s_td0, s_td1, s_td2, s_td3, s_td4) == 1) { diff --git a/OpenCL/m06222.cl b/OpenCL/m06222.cl index ae79ec5f8..7dda8f6cc 100644 --- a/OpenCL/m06222.cl +++ b/OpenCL/m06222.cl @@ -3,11 +3,15 @@ * License.....: MIT */ +#define NEW_SIMD_CODE + #include "inc_vendor.cl" #include "inc_hash_constants.h" #include "inc_hash_functions.cl" #include "inc_types.cl" #include "inc_common.cl" +#include "inc_simd.cl" +#include "inc_hash_sha512.cl" #include "inc_cipher_aes.cl" #include "inc_cipher_twofish.cl" @@ -16,221 +20,6 @@ #include "inc_truecrypt_crc32.cl" #include "inc_truecrypt_xts.cl" -__constant u64a k_sha512[80] = -{ - SHA512C00, SHA512C01, SHA512C02, SHA512C03, - SHA512C04, SHA512C05, SHA512C06, SHA512C07, - SHA512C08, SHA512C09, SHA512C0a, SHA512C0b, - SHA512C0c, SHA512C0d, SHA512C0e, SHA512C0f, - SHA512C10, SHA512C11, SHA512C12, SHA512C13, - SHA512C14, SHA512C15, SHA512C16, SHA512C17, - SHA512C18, SHA512C19, SHA512C1a, SHA512C1b, - SHA512C1c, SHA512C1d, SHA512C1e, SHA512C1f, - SHA512C20, SHA512C21, SHA512C22, SHA512C23, - SHA512C24, SHA512C25, SHA512C26, SHA512C27, - SHA512C28, SHA512C29, SHA512C2a, SHA512C2b, - SHA512C2c, SHA512C2d, SHA512C2e, SHA512C2f, - SHA512C30, SHA512C31, SHA512C32, SHA512C33, - SHA512C34, SHA512C35, SHA512C36, SHA512C37, - SHA512C38, SHA512C39, SHA512C3a, SHA512C3b, - SHA512C3c, SHA512C3d, SHA512C3e, SHA512C3f, - SHA512C40, SHA512C41, SHA512C42, SHA512C43, - SHA512C44, SHA512C45, SHA512C46, SHA512C47, - SHA512C48, SHA512C49, SHA512C4a, SHA512C4b, - SHA512C4c, SHA512C4d, SHA512C4e, SHA512C4f, -}; - -void sha512_transform (const u64 w[16], u64 dgst[8]) -{ - u64 a = dgst[0]; - u64 b = dgst[1]; - u64 c = dgst[2]; - u64 d = dgst[3]; - u64 e = dgst[4]; - u64 f = dgst[5]; - u64 g = dgst[6]; - u64 h = dgst[7]; - - u64 w0_t = w[ 0]; - u64 w1_t = w[ 1]; - u64 w2_t = w[ 2]; - u64 w3_t = w[ 3]; - u64 w4_t = w[ 4]; - u64 w5_t = w[ 5]; - u64 w6_t = w[ 6]; - u64 w7_t = w[ 7]; - u64 w8_t = w[ 8]; - u64 w9_t = w[ 9]; - u64 wa_t = w[10]; - u64 wb_t = w[11]; - u64 wc_t = w[12]; - u64 wd_t = w[13]; - u64 we_t = w[14]; - u64 wf_t = w[15]; - - #define ROUND_EXPAND() \ - { \ - w0_t = SHA512_EXPAND (we_t, w9_t, w1_t, w0_t); \ - w1_t = SHA512_EXPAND (wf_t, wa_t, w2_t, w1_t); \ - w2_t = SHA512_EXPAND (w0_t, wb_t, w3_t, w2_t); \ - w3_t = SHA512_EXPAND (w1_t, wc_t, w4_t, w3_t); \ - w4_t = SHA512_EXPAND (w2_t, wd_t, w5_t, w4_t); \ - w5_t = SHA512_EXPAND (w3_t, we_t, w6_t, w5_t); \ - w6_t = SHA512_EXPAND (w4_t, wf_t, w7_t, w6_t); \ - w7_t = SHA512_EXPAND (w5_t, w0_t, w8_t, w7_t); \ - w8_t = SHA512_EXPAND (w6_t, w1_t, w9_t, w8_t); \ - w9_t = SHA512_EXPAND (w7_t, w2_t, wa_t, w9_t); \ - wa_t = SHA512_EXPAND (w8_t, w3_t, wb_t, wa_t); \ - wb_t = SHA512_EXPAND (w9_t, w4_t, wc_t, wb_t); \ - wc_t = SHA512_EXPAND (wa_t, w5_t, wd_t, wc_t); \ - wd_t = SHA512_EXPAND (wb_t, w6_t, we_t, wd_t); \ - we_t = SHA512_EXPAND (wc_t, w7_t, wf_t, we_t); \ - wf_t = SHA512_EXPAND (wd_t, w8_t, w0_t, wf_t); \ - } - - #define ROUND_STEP(i) \ - { \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha512[i + 0]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha512[i + 1]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha512[i + 2]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha512[i + 3]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha512[i + 4]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha512[i + 5]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha512[i + 6]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha512[i + 7]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha512[i + 8]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha512[i + 9]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha512[i + 10]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha512[i + 11]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha512[i + 12]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha512[i + 13]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, we_t, k_sha512[i + 14]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha512[i + 15]); \ - } - - ROUND_STEP (0); - - #ifdef _unroll - #pragma unroll - #endif - for (int i = 16; i < 80; i += 16) - { - ROUND_EXPAND (); ROUND_STEP (i); - } - - dgst[0] += a; - dgst[1] += b; - dgst[2] += c; - dgst[3] += d; - dgst[4] += e; - dgst[5] += f; - dgst[6] += g; - dgst[7] += h; -} - -void hmac_run (const u64 w1[16], const u64 ipad[8], const u64 opad[8], u64 dgst[8]) -{ - dgst[0] = ipad[0]; - dgst[1] = ipad[1]; - dgst[2] = ipad[2]; - dgst[3] = ipad[3]; - dgst[4] = ipad[4]; - dgst[5] = ipad[5]; - dgst[6] = ipad[6]; - dgst[7] = ipad[7]; - - sha512_transform (w1, dgst); - - u64 w[16]; - - w[ 0] = dgst[0]; - w[ 1] = dgst[1]; - w[ 2] = dgst[2]; - w[ 3] = dgst[3]; - w[ 4] = dgst[4]; - w[ 5] = dgst[5]; - w[ 6] = dgst[6]; - w[ 7] = dgst[7]; - w[ 8] = 0x8000000000000000; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = (128 + 64) * 8; - - dgst[0] = opad[0]; - dgst[1] = opad[1]; - dgst[2] = opad[2]; - dgst[3] = opad[3]; - dgst[4] = opad[4]; - dgst[5] = opad[5]; - dgst[6] = opad[6]; - dgst[7] = opad[7]; - - sha512_transform (w, dgst); -} - -void hmac_init (u64 w[16], u64 ipad[8], u64 opad[8]) -{ - w[ 0] ^= 0x3636363636363636; - w[ 1] ^= 0x3636363636363636; - w[ 2] ^= 0x3636363636363636; - w[ 3] ^= 0x3636363636363636; - w[ 4] ^= 0x3636363636363636; - w[ 5] ^= 0x3636363636363636; - w[ 6] ^= 0x3636363636363636; - w[ 7] ^= 0x3636363636363636; - w[ 8] ^= 0x3636363636363636; - w[ 9] ^= 0x3636363636363636; - w[10] ^= 0x3636363636363636; - w[11] ^= 0x3636363636363636; - w[12] ^= 0x3636363636363636; - w[13] ^= 0x3636363636363636; - w[14] ^= 0x3636363636363636; - w[15] ^= 0x3636363636363636; - - ipad[0] = SHA512M_A; - ipad[1] = SHA512M_B; - ipad[2] = SHA512M_C; - ipad[3] = SHA512M_D; - ipad[4] = SHA512M_E; - ipad[5] = SHA512M_F; - ipad[6] = SHA512M_G; - ipad[7] = SHA512M_H; - - sha512_transform (w, ipad); - - w[ 0] ^= 0x6a6a6a6a6a6a6a6a; - w[ 1] ^= 0x6a6a6a6a6a6a6a6a; - w[ 2] ^= 0x6a6a6a6a6a6a6a6a; - w[ 3] ^= 0x6a6a6a6a6a6a6a6a; - w[ 4] ^= 0x6a6a6a6a6a6a6a6a; - w[ 5] ^= 0x6a6a6a6a6a6a6a6a; - w[ 6] ^= 0x6a6a6a6a6a6a6a6a; - w[ 7] ^= 0x6a6a6a6a6a6a6a6a; - w[ 8] ^= 0x6a6a6a6a6a6a6a6a; - w[ 9] ^= 0x6a6a6a6a6a6a6a6a; - w[10] ^= 0x6a6a6a6a6a6a6a6a; - w[11] ^= 0x6a6a6a6a6a6a6a6a; - w[12] ^= 0x6a6a6a6a6a6a6a6a; - w[13] ^= 0x6a6a6a6a6a6a6a6a; - w[14] ^= 0x6a6a6a6a6a6a6a6a; - w[15] ^= 0x6a6a6a6a6a6a6a6a; - - opad[0] = SHA512M_A; - opad[1] = SHA512M_B; - opad[2] = SHA512M_C; - opad[3] = SHA512M_D; - opad[4] = SHA512M_E; - opad[5] = SHA512M_F; - opad[6] = SHA512M_G; - opad[7] = SHA512M_H; - - sha512_transform (w, opad); -} - u32 u8add (const u32 a, const u32 b) { const u32 a1 = (a >> 0) & 0xff; @@ -249,13 +38,71 @@ u32 u8add (const u32 a, const u32 b) const u32 r4 = (a4 + b4) & 0xff; const u32 r = r1 << 0 - | r2 << 8 - | r3 << 16 - | r4 << 24; + | r2 << 8 + | r3 << 16 + | r4 << 24; return r; } +void hmac_sha512_run_V (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], u64x ipad[8], u64x opad[8], u64x digest[8]) +{ + digest[0] = ipad[0]; + digest[1] = ipad[1]; + digest[2] = ipad[2]; + digest[3] = ipad[3]; + digest[4] = ipad[4]; + digest[5] = ipad[5]; + digest[6] = ipad[6]; + digest[7] = ipad[7]; + + sha512_transform_vector (w0, w1, w2, w3, w4, w5, w6, w7, digest); + + w0[0] = h32_from_64 (digest[0]); + w0[1] = l32_from_64 (digest[0]); + w0[2] = h32_from_64 (digest[1]); + w0[3] = l32_from_64 (digest[1]); + w1[0] = h32_from_64 (digest[2]); + w1[1] = l32_from_64 (digest[2]); + w1[2] = h32_from_64 (digest[3]); + w1[3] = l32_from_64 (digest[3]); + w2[0] = h32_from_64 (digest[4]); + w2[1] = l32_from_64 (digest[4]); + w2[2] = h32_from_64 (digest[5]); + w2[3] = l32_from_64 (digest[5]); + w3[0] = h32_from_64 (digest[6]); + w3[1] = l32_from_64 (digest[6]); + w3[2] = h32_from_64 (digest[7]); + w3[3] = l32_from_64 (digest[7]); + w4[0] = 0x80000000; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = (128 + 64) * 8; + + digest[0] = opad[0]; + digest[1] = opad[1]; + digest[2] = opad[2]; + digest[3] = opad[3]; + digest[4] = opad[4]; + digest[5] = opad[5]; + digest[6] = opad[6]; + digest[7] = opad[7]; + + sha512_transform_vector (w0, w1, w2, w3, w4, w5, w6, w7, digest); +} + __kernel void m06222_init (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global tc64_tmp_t *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global tc_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max) { /** @@ -267,36 +114,46 @@ __kernel void m06222_init (__global pw_t *pws, __global const kernel_rule_t *rul if (gid >= gid_max) return; u32 w0[4]; + u32 w1[4]; + u32 w2[4]; + u32 w3[4]; + u32 w4[4]; + u32 w5[4]; + u32 w6[4]; + u32 w7[4]; w0[0] = pws[gid].i[ 0]; w0[1] = pws[gid].i[ 1]; w0[2] = pws[gid].i[ 2]; w0[3] = pws[gid].i[ 3]; - - u32 w1[4]; - w1[0] = pws[gid].i[ 4]; w1[1] = pws[gid].i[ 5]; w1[2] = pws[gid].i[ 6]; w1[3] = pws[gid].i[ 7]; - - u32 w2[4]; - w2[0] = pws[gid].i[ 8]; w2[1] = pws[gid].i[ 9]; w2[2] = pws[gid].i[10]; w2[3] = pws[gid].i[11]; - - u32 w3[4]; - w3[0] = pws[gid].i[12]; w3[1] = pws[gid].i[13]; w3[2] = pws[gid].i[14]; w3[3] = pws[gid].i[15]; - - /** - * keyfile - */ + w4[0] = pws[gid].i[16]; + w4[1] = pws[gid].i[17]; + w4[2] = pws[gid].i[18]; + w4[3] = pws[gid].i[19]; + w5[0] = pws[gid].i[20]; + w5[1] = pws[gid].i[21]; + w5[2] = pws[gid].i[22]; + w5[3] = pws[gid].i[23]; + w6[0] = pws[gid].i[24]; + w6[1] = pws[gid].i[25]; + w6[2] = pws[gid].i[26]; + w6[3] = pws[gid].i[27]; + w7[0] = pws[gid].i[28]; + w7[1] = pws[gid].i[29]; + w7[2] = pws[gid].i[30]; + w7[3] = pws[gid].i[31]; w0[0] = u8add (w0[0], esalt_bufs[digests_offset].keyfile_buf[ 0]); w0[1] = u8add (w0[1], esalt_bufs[digests_offset].keyfile_buf[ 1]); @@ -315,98 +172,121 @@ __kernel void m06222_init (__global pw_t *pws, __global const kernel_rule_t *rul w3[2] = u8add (w3[2], esalt_bufs[digests_offset].keyfile_buf[14]); w3[3] = u8add (w3[3], esalt_bufs[digests_offset].keyfile_buf[15]); - /** - * salt - */ - - u64 salt_buf[16]; - - // swap fehlt - - salt_buf[ 0] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 0])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 1]); - salt_buf[ 1] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 2])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 3]); - salt_buf[ 2] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 4])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 5]); - salt_buf[ 3] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 6])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 7]); - salt_buf[ 4] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 8])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 9]); - salt_buf[ 5] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[10])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[11]); - salt_buf[ 6] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[12])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[13]); - salt_buf[ 7] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[14])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[15]); - salt_buf[ 8] = 0; - salt_buf[ 9] = 0; - salt_buf[10] = 0; - salt_buf[11] = 0; - salt_buf[12] = 0; - salt_buf[13] = 0; - salt_buf[14] = 0; - salt_buf[15] = (128 + 64 + 4) * 8; - - u64 w[16]; - - w[ 0] = ((u64) swap32 (w0[0])) << 32 | (u64) swap32 (w0[1]); - w[ 1] = ((u64) swap32 (w0[2])) << 32 | (u64) swap32 (w0[3]); - w[ 2] = ((u64) swap32 (w1[0])) << 32 | (u64) swap32 (w1[1]); - w[ 3] = ((u64) swap32 (w1[2])) << 32 | (u64) swap32 (w1[3]); - w[ 4] = ((u64) swap32 (w2[0])) << 32 | (u64) swap32 (w2[1]); - w[ 5] = ((u64) swap32 (w2[2])) << 32 | (u64) swap32 (w2[3]); - w[ 6] = ((u64) swap32 (w3[0])) << 32 | (u64) swap32 (w3[1]); - w[ 7] = ((u64) swap32 (w3[2])) << 32 | (u64) swap32 (w3[3]); - w[ 8] = 0; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = 0; - - u64 ipad[8]; - u64 opad[8]; - - hmac_init (w, ipad, opad); - - tmps[gid].ipad[0] = ipad[0]; - tmps[gid].ipad[1] = ipad[1]; - tmps[gid].ipad[2] = ipad[2]; - tmps[gid].ipad[3] = ipad[3]; - tmps[gid].ipad[4] = ipad[4]; - tmps[gid].ipad[5] = ipad[5]; - tmps[gid].ipad[6] = ipad[6]; - tmps[gid].ipad[7] = ipad[7]; - - tmps[gid].opad[0] = opad[0]; - tmps[gid].opad[1] = opad[1]; - tmps[gid].opad[2] = opad[2]; - tmps[gid].opad[3] = opad[3]; - tmps[gid].opad[4] = opad[4]; - tmps[gid].opad[5] = opad[5]; - tmps[gid].opad[6] = opad[6]; - tmps[gid].opad[7] = opad[7]; + w0[0] = swap32_S (w0[0]); + w0[1] = swap32_S (w0[1]); + w0[2] = swap32_S (w0[2]); + w0[3] = swap32_S (w0[3]); + w1[0] = swap32_S (w1[0]); + w1[1] = swap32_S (w1[1]); + w1[2] = swap32_S (w1[2]); + w1[3] = swap32_S (w1[3]); + w2[0] = swap32_S (w2[0]); + w2[1] = swap32_S (w2[1]); + w2[2] = swap32_S (w2[2]); + w2[3] = swap32_S (w2[3]); + w3[0] = swap32_S (w3[0]); + w3[1] = swap32_S (w3[1]); + w3[2] = swap32_S (w3[2]); + w3[3] = swap32_S (w3[3]); + w4[0] = swap32_S (w4[0]); + w4[1] = swap32_S (w4[1]); + w4[2] = swap32_S (w4[2]); + w4[3] = swap32_S (w4[3]); + w5[0] = swap32_S (w5[0]); + w5[1] = swap32_S (w5[1]); + w5[2] = swap32_S (w5[2]); + w5[3] = swap32_S (w5[3]); + w6[0] = swap32_S (w6[0]); + w6[1] = swap32_S (w6[1]); + w6[2] = swap32_S (w6[2]); + w6[3] = swap32_S (w6[3]); + w7[0] = swap32_S (w7[0]); + w7[1] = swap32_S (w7[1]); + w7[2] = swap32_S (w7[2]); + w7[3] = swap32_S (w7[3]); + + sha512_hmac_ctx_t sha512_hmac_ctx; + + sha512_hmac_init (&sha512_hmac_ctx, w0, w1, w2, w3, w5, w5, w6, w7); + + tmps[gid].ipad[0] = sha512_hmac_ctx.ipad.h[0]; + tmps[gid].ipad[1] = sha512_hmac_ctx.ipad.h[1]; + tmps[gid].ipad[2] = sha512_hmac_ctx.ipad.h[2]; + tmps[gid].ipad[3] = sha512_hmac_ctx.ipad.h[3]; + tmps[gid].ipad[4] = sha512_hmac_ctx.ipad.h[4]; + tmps[gid].ipad[5] = sha512_hmac_ctx.ipad.h[5]; + tmps[gid].ipad[6] = sha512_hmac_ctx.ipad.h[6]; + tmps[gid].ipad[7] = sha512_hmac_ctx.ipad.h[7]; + + tmps[gid].opad[0] = sha512_hmac_ctx.opad.h[0]; + tmps[gid].opad[1] = sha512_hmac_ctx.opad.h[1]; + tmps[gid].opad[2] = sha512_hmac_ctx.opad.h[2]; + tmps[gid].opad[3] = sha512_hmac_ctx.opad.h[3]; + tmps[gid].opad[4] = sha512_hmac_ctx.opad.h[4]; + tmps[gid].opad[5] = sha512_hmac_ctx.opad.h[5]; + tmps[gid].opad[6] = sha512_hmac_ctx.opad.h[6]; + tmps[gid].opad[7] = sha512_hmac_ctx.opad.h[7]; + + sha512_hmac_update_global_swap (&sha512_hmac_ctx, esalt_bufs[digests_offset].salt_buf, 64); for (u32 i = 0, j = 1; i < 16; i += 8, j += 1) { - salt_buf[8] = (u64) j << 32 | (u64) 0x80000000; - - u64 dgst[8]; - - hmac_run (salt_buf, ipad, opad, dgst); - - tmps[gid].dgst[i + 0] = dgst[0]; - tmps[gid].dgst[i + 1] = dgst[1]; - tmps[gid].dgst[i + 2] = dgst[2]; - tmps[gid].dgst[i + 3] = dgst[3]; - tmps[gid].dgst[i + 4] = dgst[4]; - tmps[gid].dgst[i + 5] = dgst[5]; - tmps[gid].dgst[i + 6] = dgst[6]; - tmps[gid].dgst[i + 7] = dgst[7]; - - tmps[gid].out[i + 0] = dgst[0]; - tmps[gid].out[i + 1] = dgst[1]; - tmps[gid].out[i + 2] = dgst[2]; - tmps[gid].out[i + 3] = dgst[3]; - tmps[gid].out[i + 4] = dgst[4]; - tmps[gid].out[i + 5] = dgst[5]; - tmps[gid].out[i + 6] = dgst[6]; - tmps[gid].out[i + 7] = dgst[7]; + sha512_hmac_ctx_t sha512_hmac_ctx2 = sha512_hmac_ctx; + + w0[0] = j; + w0[1] = 0; + w0[2] = 0; + w0[3] = 0; + w1[0] = 0; + w1[1] = 0; + w1[2] = 0; + w1[3] = 0; + w2[0] = 0; + w2[1] = 0; + w2[2] = 0; + w2[3] = 0; + w3[0] = 0; + w3[1] = 0; + w3[2] = 0; + w3[3] = 0; + w4[0] = 0; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = 0; + + sha512_hmac_update_128 (&sha512_hmac_ctx2, w0, w1, w2, w3, w4, w5, w6, w7, 4); + + sha512_hmac_final (&sha512_hmac_ctx2); + + tmps[gid].dgst[i + 0] = sha512_hmac_ctx2.opad.h[0]; + tmps[gid].dgst[i + 1] = sha512_hmac_ctx2.opad.h[1]; + tmps[gid].dgst[i + 2] = sha512_hmac_ctx2.opad.h[2]; + tmps[gid].dgst[i + 3] = sha512_hmac_ctx2.opad.h[3]; + tmps[gid].dgst[i + 4] = sha512_hmac_ctx2.opad.h[4]; + tmps[gid].dgst[i + 5] = sha512_hmac_ctx2.opad.h[5]; + tmps[gid].dgst[i + 6] = sha512_hmac_ctx2.opad.h[6]; + tmps[gid].dgst[i + 7] = sha512_hmac_ctx2.opad.h[7]; + + tmps[gid].out[i + 0] = tmps[gid].dgst[i + 0]; + tmps[gid].out[i + 1] = tmps[gid].dgst[i + 1]; + tmps[gid].out[i + 2] = tmps[gid].dgst[i + 2]; + tmps[gid].out[i + 3] = tmps[gid].dgst[i + 3]; + tmps[gid].out[i + 4] = tmps[gid].dgst[i + 4]; + tmps[gid].out[i + 5] = tmps[gid].dgst[i + 5]; + tmps[gid].out[i + 6] = tmps[gid].dgst[i + 6]; + tmps[gid].out[i + 7] = tmps[gid].dgst[i + 7]; } } @@ -414,76 +294,97 @@ __kernel void m06222_loop (__global pw_t *pws, __global const kernel_rule_t *rul { const u32 gid = get_global_id (0); - if (gid >= gid_max) return; - - u64 ipad[8]; - - ipad[0] = tmps[gid].ipad[0]; - ipad[1] = tmps[gid].ipad[1]; - ipad[2] = tmps[gid].ipad[2]; - ipad[3] = tmps[gid].ipad[3]; - ipad[4] = tmps[gid].ipad[4]; - ipad[5] = tmps[gid].ipad[5]; - ipad[6] = tmps[gid].ipad[6]; - ipad[7] = tmps[gid].ipad[7]; - - u64 opad[8]; - - opad[0] = tmps[gid].opad[0]; - opad[1] = tmps[gid].opad[1]; - opad[2] = tmps[gid].opad[2]; - opad[3] = tmps[gid].opad[3]; - opad[4] = tmps[gid].opad[4]; - opad[5] = tmps[gid].opad[5]; - opad[6] = tmps[gid].opad[6]; - opad[7] = tmps[gid].opad[7]; + if ((gid * VECT_SIZE) >= gid_max) return; + + u64x ipad[8]; + u64x opad[8]; + + ipad[0] = pack64v (tmps, ipad, gid, 0); + ipad[1] = pack64v (tmps, ipad, gid, 1); + ipad[2] = pack64v (tmps, ipad, gid, 2); + ipad[3] = pack64v (tmps, ipad, gid, 3); + ipad[4] = pack64v (tmps, ipad, gid, 4); + ipad[5] = pack64v (tmps, ipad, gid, 5); + ipad[6] = pack64v (tmps, ipad, gid, 6); + ipad[7] = pack64v (tmps, ipad, gid, 7); + + opad[0] = pack64v (tmps, opad, gid, 0); + opad[1] = pack64v (tmps, opad, gid, 1); + opad[2] = pack64v (tmps, opad, gid, 2); + opad[3] = pack64v (tmps, opad, gid, 3); + opad[4] = pack64v (tmps, opad, gid, 4); + opad[5] = pack64v (tmps, opad, gid, 5); + opad[6] = pack64v (tmps, opad, gid, 6); + opad[7] = pack64v (tmps, opad, gid, 7); for (u32 i = 0; i < 16; i += 8) { - u64 dgst[8]; - - dgst[0] = tmps[gid].dgst[i + 0]; - dgst[1] = tmps[gid].dgst[i + 1]; - dgst[2] = tmps[gid].dgst[i + 2]; - dgst[3] = tmps[gid].dgst[i + 3]; - dgst[4] = tmps[gid].dgst[i + 4]; - dgst[5] = tmps[gid].dgst[i + 5]; - dgst[6] = tmps[gid].dgst[i + 6]; - dgst[7] = tmps[gid].dgst[i + 7]; - - u64 out[8]; - - out[0] = tmps[gid].out[i + 0]; - out[1] = tmps[gid].out[i + 1]; - out[2] = tmps[gid].out[i + 2]; - out[3] = tmps[gid].out[i + 3]; - out[4] = tmps[gid].out[i + 4]; - out[5] = tmps[gid].out[i + 5]; - out[6] = tmps[gid].out[i + 6]; - out[7] = tmps[gid].out[i + 7]; + u64x dgst[8]; + u64x out[8]; + + dgst[0] = pack64v (tmps, dgst, gid, i + 0); + dgst[1] = pack64v (tmps, dgst, gid, i + 1); + dgst[2] = pack64v (tmps, dgst, gid, i + 2); + dgst[3] = pack64v (tmps, dgst, gid, i + 3); + dgst[4] = pack64v (tmps, dgst, gid, i + 4); + dgst[5] = pack64v (tmps, dgst, gid, i + 5); + dgst[6] = pack64v (tmps, dgst, gid, i + 6); + dgst[7] = pack64v (tmps, dgst, gid, i + 7); + + out[0] = pack64v (tmps, out, gid, i + 0); + out[1] = pack64v (tmps, out, gid, i + 1); + out[2] = pack64v (tmps, out, gid, i + 2); + out[3] = pack64v (tmps, out, gid, i + 3); + out[4] = pack64v (tmps, out, gid, i + 4); + out[5] = pack64v (tmps, out, gid, i + 5); + out[6] = pack64v (tmps, out, gid, i + 6); + out[7] = pack64v (tmps, out, gid, i + 7); for (u32 j = 0; j < loop_cnt; j++) { - u64 w[16]; - - w[ 0] = dgst[0]; - w[ 1] = dgst[1]; - w[ 2] = dgst[2]; - w[ 3] = dgst[3]; - w[ 4] = dgst[4]; - w[ 5] = dgst[5]; - w[ 6] = dgst[6]; - w[ 7] = dgst[7]; - w[ 8] = 0x8000000000000000; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = (128 + 64) * 8; - - hmac_run (w, ipad, opad, dgst); + u32x w0[4]; + u32x w1[4]; + u32x w2[4]; + u32x w3[4]; + u32x w4[4]; + u32x w5[4]; + u32x w6[4]; + u32x w7[4]; + + w0[0] = h32_from_64 (dgst[0]); + w0[1] = l32_from_64 (dgst[0]); + w0[2] = h32_from_64 (dgst[1]); + w0[3] = l32_from_64 (dgst[1]); + w1[0] = h32_from_64 (dgst[2]); + w1[1] = l32_from_64 (dgst[2]); + w1[2] = h32_from_64 (dgst[3]); + w1[3] = l32_from_64 (dgst[3]); + w2[0] = h32_from_64 (dgst[4]); + w2[1] = l32_from_64 (dgst[4]); + w2[2] = h32_from_64 (dgst[5]); + w2[3] = l32_from_64 (dgst[5]); + w3[0] = h32_from_64 (dgst[6]); + w3[1] = l32_from_64 (dgst[6]); + w3[2] = h32_from_64 (dgst[7]); + w3[3] = l32_from_64 (dgst[7]); + w4[0] = 0x80000000; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = (128 + 64) * 8; + + hmac_sha512_run_V (w0, w1, w2, w3, w4, w5, w6, w7, ipad, opad, dgst); out[0] ^= dgst[0]; out[1] ^= dgst[1]; @@ -495,23 +396,23 @@ __kernel void m06222_loop (__global pw_t *pws, __global const kernel_rule_t *rul out[7] ^= dgst[7]; } - tmps[gid].dgst[i + 0] = dgst[0]; - tmps[gid].dgst[i + 1] = dgst[1]; - tmps[gid].dgst[i + 2] = dgst[2]; - tmps[gid].dgst[i + 3] = dgst[3]; - tmps[gid].dgst[i + 4] = dgst[4]; - tmps[gid].dgst[i + 5] = dgst[5]; - tmps[gid].dgst[i + 6] = dgst[6]; - tmps[gid].dgst[i + 7] = dgst[7]; - - tmps[gid].out[i + 0] = out[0]; - tmps[gid].out[i + 1] = out[1]; - tmps[gid].out[i + 2] = out[2]; - tmps[gid].out[i + 3] = out[3]; - tmps[gid].out[i + 4] = out[4]; - tmps[gid].out[i + 5] = out[5]; - tmps[gid].out[i + 6] = out[6]; - tmps[gid].out[i + 7] = out[7]; + unpack64v (tmps, dgst, gid, i + 0, dgst[0]); + unpack64v (tmps, dgst, gid, i + 1, dgst[1]); + unpack64v (tmps, dgst, gid, i + 2, dgst[2]); + unpack64v (tmps, dgst, gid, i + 3, dgst[3]); + unpack64v (tmps, dgst, gid, i + 4, dgst[4]); + unpack64v (tmps, dgst, gid, i + 5, dgst[5]); + unpack64v (tmps, dgst, gid, i + 6, dgst[6]); + unpack64v (tmps, dgst, gid, i + 7, dgst[7]); + + unpack64v (tmps, out, gid, i + 0, out[0]); + unpack64v (tmps, out, gid, i + 1, out[1]); + unpack64v (tmps, out, gid, i + 2, out[2]); + unpack64v (tmps, out, gid, i + 3, out[3]); + unpack64v (tmps, out, gid, i + 4, out[4]); + unpack64v (tmps, out, gid, i + 5, out[5]); + unpack64v (tmps, out, gid, i + 6, out[6]); + unpack64v (tmps, out, gid, i + 7, out[7]); } } diff --git a/OpenCL/m06223.cl b/OpenCL/m06223.cl index 667c83fc7..67977161f 100644 --- a/OpenCL/m06223.cl +++ b/OpenCL/m06223.cl @@ -3,11 +3,15 @@ * License.....: MIT */ +#define NEW_SIMD_CODE + #include "inc_vendor.cl" #include "inc_hash_constants.h" #include "inc_hash_functions.cl" #include "inc_types.cl" #include "inc_common.cl" +#include "inc_simd.cl" +#include "inc_hash_sha512.cl" #include "inc_cipher_aes.cl" #include "inc_cipher_twofish.cl" @@ -16,221 +20,6 @@ #include "inc_truecrypt_crc32.cl" #include "inc_truecrypt_xts.cl" -__constant u64a k_sha512[80] = -{ - SHA512C00, SHA512C01, SHA512C02, SHA512C03, - SHA512C04, SHA512C05, SHA512C06, SHA512C07, - SHA512C08, SHA512C09, SHA512C0a, SHA512C0b, - SHA512C0c, SHA512C0d, SHA512C0e, SHA512C0f, - SHA512C10, SHA512C11, SHA512C12, SHA512C13, - SHA512C14, SHA512C15, SHA512C16, SHA512C17, - SHA512C18, SHA512C19, SHA512C1a, SHA512C1b, - SHA512C1c, SHA512C1d, SHA512C1e, SHA512C1f, - SHA512C20, SHA512C21, SHA512C22, SHA512C23, - SHA512C24, SHA512C25, SHA512C26, SHA512C27, - SHA512C28, SHA512C29, SHA512C2a, SHA512C2b, - SHA512C2c, SHA512C2d, SHA512C2e, SHA512C2f, - SHA512C30, SHA512C31, SHA512C32, SHA512C33, - SHA512C34, SHA512C35, SHA512C36, SHA512C37, - SHA512C38, SHA512C39, SHA512C3a, SHA512C3b, - SHA512C3c, SHA512C3d, SHA512C3e, SHA512C3f, - SHA512C40, SHA512C41, SHA512C42, SHA512C43, - SHA512C44, SHA512C45, SHA512C46, SHA512C47, - SHA512C48, SHA512C49, SHA512C4a, SHA512C4b, - SHA512C4c, SHA512C4d, SHA512C4e, SHA512C4f, -}; - -void sha512_transform (const u64 w[16], u64 dgst[8]) -{ - u64 a = dgst[0]; - u64 b = dgst[1]; - u64 c = dgst[2]; - u64 d = dgst[3]; - u64 e = dgst[4]; - u64 f = dgst[5]; - u64 g = dgst[6]; - u64 h = dgst[7]; - - u64 w0_t = w[ 0]; - u64 w1_t = w[ 1]; - u64 w2_t = w[ 2]; - u64 w3_t = w[ 3]; - u64 w4_t = w[ 4]; - u64 w5_t = w[ 5]; - u64 w6_t = w[ 6]; - u64 w7_t = w[ 7]; - u64 w8_t = w[ 8]; - u64 w9_t = w[ 9]; - u64 wa_t = w[10]; - u64 wb_t = w[11]; - u64 wc_t = w[12]; - u64 wd_t = w[13]; - u64 we_t = w[14]; - u64 wf_t = w[15]; - - #define ROUND_EXPAND() \ - { \ - w0_t = SHA512_EXPAND (we_t, w9_t, w1_t, w0_t); \ - w1_t = SHA512_EXPAND (wf_t, wa_t, w2_t, w1_t); \ - w2_t = SHA512_EXPAND (w0_t, wb_t, w3_t, w2_t); \ - w3_t = SHA512_EXPAND (w1_t, wc_t, w4_t, w3_t); \ - w4_t = SHA512_EXPAND (w2_t, wd_t, w5_t, w4_t); \ - w5_t = SHA512_EXPAND (w3_t, we_t, w6_t, w5_t); \ - w6_t = SHA512_EXPAND (w4_t, wf_t, w7_t, w6_t); \ - w7_t = SHA512_EXPAND (w5_t, w0_t, w8_t, w7_t); \ - w8_t = SHA512_EXPAND (w6_t, w1_t, w9_t, w8_t); \ - w9_t = SHA512_EXPAND (w7_t, w2_t, wa_t, w9_t); \ - wa_t = SHA512_EXPAND (w8_t, w3_t, wb_t, wa_t); \ - wb_t = SHA512_EXPAND (w9_t, w4_t, wc_t, wb_t); \ - wc_t = SHA512_EXPAND (wa_t, w5_t, wd_t, wc_t); \ - wd_t = SHA512_EXPAND (wb_t, w6_t, we_t, wd_t); \ - we_t = SHA512_EXPAND (wc_t, w7_t, wf_t, we_t); \ - wf_t = SHA512_EXPAND (wd_t, w8_t, w0_t, wf_t); \ - } - - #define ROUND_STEP(i) \ - { \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha512[i + 0]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha512[i + 1]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha512[i + 2]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha512[i + 3]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha512[i + 4]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha512[i + 5]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha512[i + 6]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha512[i + 7]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha512[i + 8]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha512[i + 9]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha512[i + 10]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha512[i + 11]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha512[i + 12]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha512[i + 13]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, c, d, e, f, g, h, a, b, we_t, k_sha512[i + 14]); \ - SHA512_STEP (SHA512_F0o, SHA512_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha512[i + 15]); \ - } - - ROUND_STEP (0); - - #ifdef _unroll - #pragma unroll - #endif - for (int i = 16; i < 80; i += 16) - { - ROUND_EXPAND (); ROUND_STEP (i); - } - - dgst[0] += a; - dgst[1] += b; - dgst[2] += c; - dgst[3] += d; - dgst[4] += e; - dgst[5] += f; - dgst[6] += g; - dgst[7] += h; -} - -void hmac_run (const u64 w1[16], const u64 ipad[8], const u64 opad[8], u64 dgst[8]) -{ - dgst[0] = ipad[0]; - dgst[1] = ipad[1]; - dgst[2] = ipad[2]; - dgst[3] = ipad[3]; - dgst[4] = ipad[4]; - dgst[5] = ipad[5]; - dgst[6] = ipad[6]; - dgst[7] = ipad[7]; - - sha512_transform (w1, dgst); - - u64 w[16]; - - w[ 0] = dgst[0]; - w[ 1] = dgst[1]; - w[ 2] = dgst[2]; - w[ 3] = dgst[3]; - w[ 4] = dgst[4]; - w[ 5] = dgst[5]; - w[ 6] = dgst[6]; - w[ 7] = dgst[7]; - w[ 8] = 0x8000000000000000; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = (128 + 64) * 8; - - dgst[0] = opad[0]; - dgst[1] = opad[1]; - dgst[2] = opad[2]; - dgst[3] = opad[3]; - dgst[4] = opad[4]; - dgst[5] = opad[5]; - dgst[6] = opad[6]; - dgst[7] = opad[7]; - - sha512_transform (w, dgst); -} - -void hmac_init (u64 w[16], u64 ipad[8], u64 opad[8]) -{ - w[ 0] ^= 0x3636363636363636; - w[ 1] ^= 0x3636363636363636; - w[ 2] ^= 0x3636363636363636; - w[ 3] ^= 0x3636363636363636; - w[ 4] ^= 0x3636363636363636; - w[ 5] ^= 0x3636363636363636; - w[ 6] ^= 0x3636363636363636; - w[ 7] ^= 0x3636363636363636; - w[ 8] ^= 0x3636363636363636; - w[ 9] ^= 0x3636363636363636; - w[10] ^= 0x3636363636363636; - w[11] ^= 0x3636363636363636; - w[12] ^= 0x3636363636363636; - w[13] ^= 0x3636363636363636; - w[14] ^= 0x3636363636363636; - w[15] ^= 0x3636363636363636; - - ipad[0] = SHA512M_A; - ipad[1] = SHA512M_B; - ipad[2] = SHA512M_C; - ipad[3] = SHA512M_D; - ipad[4] = SHA512M_E; - ipad[5] = SHA512M_F; - ipad[6] = SHA512M_G; - ipad[7] = SHA512M_H; - - sha512_transform (w, ipad); - - w[ 0] ^= 0x6a6a6a6a6a6a6a6a; - w[ 1] ^= 0x6a6a6a6a6a6a6a6a; - w[ 2] ^= 0x6a6a6a6a6a6a6a6a; - w[ 3] ^= 0x6a6a6a6a6a6a6a6a; - w[ 4] ^= 0x6a6a6a6a6a6a6a6a; - w[ 5] ^= 0x6a6a6a6a6a6a6a6a; - w[ 6] ^= 0x6a6a6a6a6a6a6a6a; - w[ 7] ^= 0x6a6a6a6a6a6a6a6a; - w[ 8] ^= 0x6a6a6a6a6a6a6a6a; - w[ 9] ^= 0x6a6a6a6a6a6a6a6a; - w[10] ^= 0x6a6a6a6a6a6a6a6a; - w[11] ^= 0x6a6a6a6a6a6a6a6a; - w[12] ^= 0x6a6a6a6a6a6a6a6a; - w[13] ^= 0x6a6a6a6a6a6a6a6a; - w[14] ^= 0x6a6a6a6a6a6a6a6a; - w[15] ^= 0x6a6a6a6a6a6a6a6a; - - opad[0] = SHA512M_A; - opad[1] = SHA512M_B; - opad[2] = SHA512M_C; - opad[3] = SHA512M_D; - opad[4] = SHA512M_E; - opad[5] = SHA512M_F; - opad[6] = SHA512M_G; - opad[7] = SHA512M_H; - - sha512_transform (w, opad); -} - u32 u8add (const u32 a, const u32 b) { const u32 a1 = (a >> 0) & 0xff; @@ -249,103 +38,122 @@ u32 u8add (const u32 a, const u32 b) const u32 r4 = (a4 + b4) & 0xff; const u32 r = r1 << 0 - | r2 << 8 - | r3 << 16 - | r4 << 24; + | r2 << 8 + | r3 << 16 + | r4 << 24; return r; } -__kernel void m06223_init (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global tc64_tmp_t *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global tc_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max) +void hmac_sha512_run_V (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x w4[4], u32x w5[4], u32x w6[4], u32x w7[4], u64x ipad[8], u64x opad[8], u64x digest[8]) { - const u32 gid = get_global_id (0); - const u32 lid = get_local_id (0); - const u32 lsz = get_local_size (0); + digest[0] = ipad[0]; + digest[1] = ipad[1]; + digest[2] = ipad[2]; + digest[3] = ipad[3]; + digest[4] = ipad[4]; + digest[5] = ipad[5]; + digest[6] = ipad[6]; + digest[7] = ipad[7]; + + sha512_transform_vector (w0, w1, w2, w3, w4, w5, w6, w7, digest); + + w0[0] = h32_from_64 (digest[0]); + w0[1] = l32_from_64 (digest[0]); + w0[2] = h32_from_64 (digest[1]); + w0[3] = l32_from_64 (digest[1]); + w1[0] = h32_from_64 (digest[2]); + w1[1] = l32_from_64 (digest[2]); + w1[2] = h32_from_64 (digest[3]); + w1[3] = l32_from_64 (digest[3]); + w2[0] = h32_from_64 (digest[4]); + w2[1] = l32_from_64 (digest[4]); + w2[2] = h32_from_64 (digest[5]); + w2[3] = l32_from_64 (digest[5]); + w3[0] = h32_from_64 (digest[6]); + w3[1] = l32_from_64 (digest[6]); + w3[2] = h32_from_64 (digest[7]); + w3[3] = l32_from_64 (digest[7]); + w4[0] = 0x80000000; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = (128 + 64) * 8; + + digest[0] = opad[0]; + digest[1] = opad[1]; + digest[2] = opad[2]; + digest[3] = opad[3]; + digest[4] = opad[4]; + digest[5] = opad[5]; + digest[6] = opad[6]; + digest[7] = opad[7]; + + sha512_transform_vector (w0, w1, w2, w3, w4, w5, w6, w7, digest); +} +__kernel void m06223_init (__global pw_t *pws, __global const kernel_rule_t *rules_buf, __global const pw_t *combs_buf, __global const bf_t *bfs_buf, __global tc64_tmp_t *tmps, __global void *hooks, __global const u32 *bitmaps_buf_s1_a, __global const u32 *bitmaps_buf_s1_b, __global const u32 *bitmaps_buf_s1_c, __global const u32 *bitmaps_buf_s1_d, __global const u32 *bitmaps_buf_s2_a, __global const u32 *bitmaps_buf_s2_b, __global const u32 *bitmaps_buf_s2_c, __global const u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global const digest_t *digests_buf, __global u32 *hashes_shown, __global const salt_t *salt_bufs, __global tc_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV0_buf, __global u32 *d_scryptV1_buf, __global u32 *d_scryptV2_buf, __global u32 *d_scryptV3_buf, const u32 bitmap_mask, const u32 bitmap_shift1, const u32 bitmap_shift2, const u32 salt_pos, const u32 loop_pos, const u32 loop_cnt, const u32 il_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max) +{ /** - * aes shared + * base */ - #ifdef REAL_SHM - - __local u32 s_td0[256]; - __local u32 s_td1[256]; - __local u32 s_td2[256]; - __local u32 s_td3[256]; - __local u32 s_td4[256]; - - __local u32 s_te0[256]; - __local u32 s_te1[256]; - __local u32 s_te2[256]; - __local u32 s_te3[256]; - __local 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]; - } - - barrier (CLK_LOCAL_MEM_FENCE); - - #else - - __constant u32a *s_td0 = td0; - __constant u32a *s_td1 = td1; - __constant u32a *s_td2 = td2; - __constant u32a *s_td3 = td3; - __constant u32a *s_td4 = td4; - - __constant u32a *s_te0 = te0; - __constant u32a *s_te1 = te1; - __constant u32a *s_te2 = te2; - __constant u32a *s_te3 = te3; - __constant u32a *s_te4 = te4; - - #endif + const u32 gid = get_global_id (0); if (gid >= gid_max) return; u32 w0[4]; + u32 w1[4]; + u32 w2[4]; + u32 w3[4]; + u32 w4[4]; + u32 w5[4]; + u32 w6[4]; + u32 w7[4]; w0[0] = pws[gid].i[ 0]; w0[1] = pws[gid].i[ 1]; w0[2] = pws[gid].i[ 2]; w0[3] = pws[gid].i[ 3]; - - u32 w1[4]; - w1[0] = pws[gid].i[ 4]; w1[1] = pws[gid].i[ 5]; w1[2] = pws[gid].i[ 6]; w1[3] = pws[gid].i[ 7]; - - u32 w2[4]; - w2[0] = pws[gid].i[ 8]; w2[1] = pws[gid].i[ 9]; w2[2] = pws[gid].i[10]; w2[3] = pws[gid].i[11]; - - u32 w3[4]; - w3[0] = pws[gid].i[12]; w3[1] = pws[gid].i[13]; w3[2] = pws[gid].i[14]; w3[3] = pws[gid].i[15]; - - /** - * keyfile - */ + w4[0] = pws[gid].i[16]; + w4[1] = pws[gid].i[17]; + w4[2] = pws[gid].i[18]; + w4[3] = pws[gid].i[19]; + w5[0] = pws[gid].i[20]; + w5[1] = pws[gid].i[21]; + w5[2] = pws[gid].i[22]; + w5[3] = pws[gid].i[23]; + w6[0] = pws[gid].i[24]; + w6[1] = pws[gid].i[25]; + w6[2] = pws[gid].i[26]; + w6[3] = pws[gid].i[27]; + w7[0] = pws[gid].i[28]; + w7[1] = pws[gid].i[29]; + w7[2] = pws[gid].i[30]; + w7[3] = pws[gid].i[31]; w0[0] = u8add (w0[0], esalt_bufs[digests_offset].keyfile_buf[ 0]); w0[1] = u8add (w0[1], esalt_bufs[digests_offset].keyfile_buf[ 1]); @@ -364,98 +172,121 @@ __kernel void m06223_init (__global pw_t *pws, __global const kernel_rule_t *rul w3[2] = u8add (w3[2], esalt_bufs[digests_offset].keyfile_buf[14]); w3[3] = u8add (w3[3], esalt_bufs[digests_offset].keyfile_buf[15]); - /** - * salt - */ - - u64 salt_buf[16]; - - // swap fehlt - - salt_buf[ 0] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 0])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 1]); - salt_buf[ 1] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 2])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 3]); - salt_buf[ 2] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 4])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 5]); - salt_buf[ 3] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 6])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 7]); - salt_buf[ 4] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 8])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[ 9]); - salt_buf[ 5] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[10])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[11]); - salt_buf[ 6] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[12])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[13]); - salt_buf[ 7] = ((u64) swap32 (esalt_bufs[digests_offset].salt_buf[14])) << 32 | (u64) swap32 (esalt_bufs[digests_offset].salt_buf[15]); - salt_buf[ 8] = 0; - salt_buf[ 9] = 0; - salt_buf[10] = 0; - salt_buf[11] = 0; - salt_buf[12] = 0; - salt_buf[13] = 0; - salt_buf[14] = 0; - salt_buf[15] = (128 + 64 + 4) * 8; - - u64 w[16]; - - w[ 0] = ((u64) swap32 (w0[0])) << 32 | (u64) swap32 (w0[1]); - w[ 1] = ((u64) swap32 (w0[2])) << 32 | (u64) swap32 (w0[3]); - w[ 2] = ((u64) swap32 (w1[0])) << 32 | (u64) swap32 (w1[1]); - w[ 3] = ((u64) swap32 (w1[2])) << 32 | (u64) swap32 (w1[3]); - w[ 4] = ((u64) swap32 (w2[0])) << 32 | (u64) swap32 (w2[1]); - w[ 5] = ((u64) swap32 (w2[2])) << 32 | (u64) swap32 (w2[3]); - w[ 6] = ((u64) swap32 (w3[0])) << 32 | (u64) swap32 (w3[1]); - w[ 7] = ((u64) swap32 (w3[2])) << 32 | (u64) swap32 (w3[3]); - w[ 8] = 0; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = 0; - - u64 ipad[8]; - u64 opad[8]; - - hmac_init (w, ipad, opad); - - tmps[gid].ipad[0] = ipad[0]; - tmps[gid].ipad[1] = ipad[1]; - tmps[gid].ipad[2] = ipad[2]; - tmps[gid].ipad[3] = ipad[3]; - tmps[gid].ipad[4] = ipad[4]; - tmps[gid].ipad[5] = ipad[5]; - tmps[gid].ipad[6] = ipad[6]; - tmps[gid].ipad[7] = ipad[7]; - - tmps[gid].opad[0] = opad[0]; - tmps[gid].opad[1] = opad[1]; - tmps[gid].opad[2] = opad[2]; - tmps[gid].opad[3] = opad[3]; - tmps[gid].opad[4] = opad[4]; - tmps[gid].opad[5] = opad[5]; - tmps[gid].opad[6] = opad[6]; - tmps[gid].opad[7] = opad[7]; + w0[0] = swap32_S (w0[0]); + w0[1] = swap32_S (w0[1]); + w0[2] = swap32_S (w0[2]); + w0[3] = swap32_S (w0[3]); + w1[0] = swap32_S (w1[0]); + w1[1] = swap32_S (w1[1]); + w1[2] = swap32_S (w1[2]); + w1[3] = swap32_S (w1[3]); + w2[0] = swap32_S (w2[0]); + w2[1] = swap32_S (w2[1]); + w2[2] = swap32_S (w2[2]); + w2[3] = swap32_S (w2[3]); + w3[0] = swap32_S (w3[0]); + w3[1] = swap32_S (w3[1]); + w3[2] = swap32_S (w3[2]); + w3[3] = swap32_S (w3[3]); + w4[0] = swap32_S (w4[0]); + w4[1] = swap32_S (w4[1]); + w4[2] = swap32_S (w4[2]); + w4[3] = swap32_S (w4[3]); + w5[0] = swap32_S (w5[0]); + w5[1] = swap32_S (w5[1]); + w5[2] = swap32_S (w5[2]); + w5[3] = swap32_S (w5[3]); + w6[0] = swap32_S (w6[0]); + w6[1] = swap32_S (w6[1]); + w6[2] = swap32_S (w6[2]); + w6[3] = swap32_S (w6[3]); + w7[0] = swap32_S (w7[0]); + w7[1] = swap32_S (w7[1]); + w7[2] = swap32_S (w7[2]); + w7[3] = swap32_S (w7[3]); + + sha512_hmac_ctx_t sha512_hmac_ctx; + + sha512_hmac_init (&sha512_hmac_ctx, w0, w1, w2, w3, w5, w5, w6, w7); + + tmps[gid].ipad[0] = sha512_hmac_ctx.ipad.h[0]; + tmps[gid].ipad[1] = sha512_hmac_ctx.ipad.h[1]; + tmps[gid].ipad[2] = sha512_hmac_ctx.ipad.h[2]; + tmps[gid].ipad[3] = sha512_hmac_ctx.ipad.h[3]; + tmps[gid].ipad[4] = sha512_hmac_ctx.ipad.h[4]; + tmps[gid].ipad[5] = sha512_hmac_ctx.ipad.h[5]; + tmps[gid].ipad[6] = sha512_hmac_ctx.ipad.h[6]; + tmps[gid].ipad[7] = sha512_hmac_ctx.ipad.h[7]; + + tmps[gid].opad[0] = sha512_hmac_ctx.opad.h[0]; + tmps[gid].opad[1] = sha512_hmac_ctx.opad.h[1]; + tmps[gid].opad[2] = sha512_hmac_ctx.opad.h[2]; + tmps[gid].opad[3] = sha512_hmac_ctx.opad.h[3]; + tmps[gid].opad[4] = sha512_hmac_ctx.opad.h[4]; + tmps[gid].opad[5] = sha512_hmac_ctx.opad.h[5]; + tmps[gid].opad[6] = sha512_hmac_ctx.opad.h[6]; + tmps[gid].opad[7] = sha512_hmac_ctx.opad.h[7]; + + sha512_hmac_update_global_swap (&sha512_hmac_ctx, esalt_bufs[digests_offset].salt_buf, 64); for (u32 i = 0, j = 1; i < 24; i += 8, j += 1) { - salt_buf[8] = (u64) j << 32 | (u64) 0x80000000; - - u64 dgst[8]; - - hmac_run (salt_buf, ipad, opad, dgst); - - tmps[gid].dgst[i + 0] = dgst[0]; - tmps[gid].dgst[i + 1] = dgst[1]; - tmps[gid].dgst[i + 2] = dgst[2]; - tmps[gid].dgst[i + 3] = dgst[3]; - tmps[gid].dgst[i + 4] = dgst[4]; - tmps[gid].dgst[i + 5] = dgst[5]; - tmps[gid].dgst[i + 6] = dgst[6]; - tmps[gid].dgst[i + 7] = dgst[7]; - - tmps[gid].out[i + 0] = dgst[0]; - tmps[gid].out[i + 1] = dgst[1]; - tmps[gid].out[i + 2] = dgst[2]; - tmps[gid].out[i + 3] = dgst[3]; - tmps[gid].out[i + 4] = dgst[4]; - tmps[gid].out[i + 5] = dgst[5]; - tmps[gid].out[i + 6] = dgst[6]; - tmps[gid].out[i + 7] = dgst[7]; + sha512_hmac_ctx_t sha512_hmac_ctx2 = sha512_hmac_ctx; + + w0[0] = j; + w0[1] = 0; + w0[2] = 0; + w0[3] = 0; + w1[0] = 0; + w1[1] = 0; + w1[2] = 0; + w1[3] = 0; + w2[0] = 0; + w2[1] = 0; + w2[2] = 0; + w2[3] = 0; + w3[0] = 0; + w3[1] = 0; + w3[2] = 0; + w3[3] = 0; + w4[0] = 0; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = 0; + + sha512_hmac_update_128 (&sha512_hmac_ctx2, w0, w1, w2, w3, w4, w5, w6, w7, 4); + + sha512_hmac_final (&sha512_hmac_ctx2); + + tmps[gid].dgst[i + 0] = sha512_hmac_ctx2.opad.h[0]; + tmps[gid].dgst[i + 1] = sha512_hmac_ctx2.opad.h[1]; + tmps[gid].dgst[i + 2] = sha512_hmac_ctx2.opad.h[2]; + tmps[gid].dgst[i + 3] = sha512_hmac_ctx2.opad.h[3]; + tmps[gid].dgst[i + 4] = sha512_hmac_ctx2.opad.h[4]; + tmps[gid].dgst[i + 5] = sha512_hmac_ctx2.opad.h[5]; + tmps[gid].dgst[i + 6] = sha512_hmac_ctx2.opad.h[6]; + tmps[gid].dgst[i + 7] = sha512_hmac_ctx2.opad.h[7]; + + tmps[gid].out[i + 0] = tmps[gid].dgst[i + 0]; + tmps[gid].out[i + 1] = tmps[gid].dgst[i + 1]; + tmps[gid].out[i + 2] = tmps[gid].dgst[i + 2]; + tmps[gid].out[i + 3] = tmps[gid].dgst[i + 3]; + tmps[gid].out[i + 4] = tmps[gid].dgst[i + 4]; + tmps[gid].out[i + 5] = tmps[gid].dgst[i + 5]; + tmps[gid].out[i + 6] = tmps[gid].dgst[i + 6]; + tmps[gid].out[i + 7] = tmps[gid].dgst[i + 7]; } } @@ -463,76 +294,97 @@ __kernel void m06223_loop (__global pw_t *pws, __global const kernel_rule_t *rul { const u32 gid = get_global_id (0); - if (gid >= gid_max) return; - - u64 ipad[8]; - - ipad[0] = tmps[gid].ipad[0]; - ipad[1] = tmps[gid].ipad[1]; - ipad[2] = tmps[gid].ipad[2]; - ipad[3] = tmps[gid].ipad[3]; - ipad[4] = tmps[gid].ipad[4]; - ipad[5] = tmps[gid].ipad[5]; - ipad[6] = tmps[gid].ipad[6]; - ipad[7] = tmps[gid].ipad[7]; - - u64 opad[8]; - - opad[0] = tmps[gid].opad[0]; - opad[1] = tmps[gid].opad[1]; - opad[2] = tmps[gid].opad[2]; - opad[3] = tmps[gid].opad[3]; - opad[4] = tmps[gid].opad[4]; - opad[5] = tmps[gid].opad[5]; - opad[6] = tmps[gid].opad[6]; - opad[7] = tmps[gid].opad[7]; + if ((gid * VECT_SIZE) >= gid_max) return; + + u64x ipad[8]; + u64x opad[8]; + + ipad[0] = pack64v (tmps, ipad, gid, 0); + ipad[1] = pack64v (tmps, ipad, gid, 1); + ipad[2] = pack64v (tmps, ipad, gid, 2); + ipad[3] = pack64v (tmps, ipad, gid, 3); + ipad[4] = pack64v (tmps, ipad, gid, 4); + ipad[5] = pack64v (tmps, ipad, gid, 5); + ipad[6] = pack64v (tmps, ipad, gid, 6); + ipad[7] = pack64v (tmps, ipad, gid, 7); + + opad[0] = pack64v (tmps, opad, gid, 0); + opad[1] = pack64v (tmps, opad, gid, 1); + opad[2] = pack64v (tmps, opad, gid, 2); + opad[3] = pack64v (tmps, opad, gid, 3); + opad[4] = pack64v (tmps, opad, gid, 4); + opad[5] = pack64v (tmps, opad, gid, 5); + opad[6] = pack64v (tmps, opad, gid, 6); + opad[7] = pack64v (tmps, opad, gid, 7); for (u32 i = 0; i < 24; i += 8) { - u64 dgst[8]; - - dgst[0] = tmps[gid].dgst[i + 0]; - dgst[1] = tmps[gid].dgst[i + 1]; - dgst[2] = tmps[gid].dgst[i + 2]; - dgst[3] = tmps[gid].dgst[i + 3]; - dgst[4] = tmps[gid].dgst[i + 4]; - dgst[5] = tmps[gid].dgst[i + 5]; - dgst[6] = tmps[gid].dgst[i + 6]; - dgst[7] = tmps[gid].dgst[i + 7]; - - u64 out[8]; - - out[0] = tmps[gid].out[i + 0]; - out[1] = tmps[gid].out[i + 1]; - out[2] = tmps[gid].out[i + 2]; - out[3] = tmps[gid].out[i + 3]; - out[4] = tmps[gid].out[i + 4]; - out[5] = tmps[gid].out[i + 5]; - out[6] = tmps[gid].out[i + 6]; - out[7] = tmps[gid].out[i + 7]; + u64x dgst[8]; + u64x out[8]; + + dgst[0] = pack64v (tmps, dgst, gid, i + 0); + dgst[1] = pack64v (tmps, dgst, gid, i + 1); + dgst[2] = pack64v (tmps, dgst, gid, i + 2); + dgst[3] = pack64v (tmps, dgst, gid, i + 3); + dgst[4] = pack64v (tmps, dgst, gid, i + 4); + dgst[5] = pack64v (tmps, dgst, gid, i + 5); + dgst[6] = pack64v (tmps, dgst, gid, i + 6); + dgst[7] = pack64v (tmps, dgst, gid, i + 7); + + out[0] = pack64v (tmps, out, gid, i + 0); + out[1] = pack64v (tmps, out, gid, i + 1); + out[2] = pack64v (tmps, out, gid, i + 2); + out[3] = pack64v (tmps, out, gid, i + 3); + out[4] = pack64v (tmps, out, gid, i + 4); + out[5] = pack64v (tmps, out, gid, i + 5); + out[6] = pack64v (tmps, out, gid, i + 6); + out[7] = pack64v (tmps, out, gid, i + 7); for (u32 j = 0; j < loop_cnt; j++) { - u64 w[16]; - - w[ 0] = dgst[0]; - w[ 1] = dgst[1]; - w[ 2] = dgst[2]; - w[ 3] = dgst[3]; - w[ 4] = dgst[4]; - w[ 5] = dgst[5]; - w[ 6] = dgst[6]; - w[ 7] = dgst[7]; - w[ 8] = 0x8000000000000000; - w[ 9] = 0; - w[10] = 0; - w[11] = 0; - w[12] = 0; - w[13] = 0; - w[14] = 0; - w[15] = (128 + 64) * 8; - - hmac_run (w, ipad, opad, dgst); + u32x w0[4]; + u32x w1[4]; + u32x w2[4]; + u32x w3[4]; + u32x w4[4]; + u32x w5[4]; + u32x w6[4]; + u32x w7[4]; + + w0[0] = h32_from_64 (dgst[0]); + w0[1] = l32_from_64 (dgst[0]); + w0[2] = h32_from_64 (dgst[1]); + w0[3] = l32_from_64 (dgst[1]); + w1[0] = h32_from_64 (dgst[2]); + w1[1] = l32_from_64 (dgst[2]); + w1[2] = h32_from_64 (dgst[3]); + w1[3] = l32_from_64 (dgst[3]); + w2[0] = h32_from_64 (dgst[4]); + w2[1] = l32_from_64 (dgst[4]); + w2[2] = h32_from_64 (dgst[5]); + w2[3] = l32_from_64 (dgst[5]); + w3[0] = h32_from_64 (dgst[6]); + w3[1] = l32_from_64 (dgst[6]); + w3[2] = h32_from_64 (dgst[7]); + w3[3] = l32_from_64 (dgst[7]); + w4[0] = 0x80000000; + w4[1] = 0; + w4[2] = 0; + w4[3] = 0; + w5[0] = 0; + w5[1] = 0; + w5[2] = 0; + w5[3] = 0; + w6[0] = 0; + w6[1] = 0; + w6[2] = 0; + w6[3] = 0; + w7[0] = 0; + w7[1] = 0; + w7[2] = 0; + w7[3] = (128 + 64) * 8; + + hmac_sha512_run_V (w0, w1, w2, w3, w4, w5, w6, w7, ipad, opad, dgst); out[0] ^= dgst[0]; out[1] ^= dgst[1]; @@ -544,23 +396,23 @@ __kernel void m06223_loop (__global pw_t *pws, __global const kernel_rule_t *rul out[7] ^= dgst[7]; } - tmps[gid].dgst[i + 0] = dgst[0]; - tmps[gid].dgst[i + 1] = dgst[1]; - tmps[gid].dgst[i + 2] = dgst[2]; - tmps[gid].dgst[i + 3] = dgst[3]; - tmps[gid].dgst[i + 4] = dgst[4]; - tmps[gid].dgst[i + 5] = dgst[5]; - tmps[gid].dgst[i + 6] = dgst[6]; - tmps[gid].dgst[i + 7] = dgst[7]; - - tmps[gid].out[i + 0] = out[0]; - tmps[gid].out[i + 1] = out[1]; - tmps[gid].out[i + 2] = out[2]; - tmps[gid].out[i + 3] = out[3]; - tmps[gid].out[i + 4] = out[4]; - tmps[gid].out[i + 5] = out[5]; - tmps[gid].out[i + 6] = out[6]; - tmps[gid].out[i + 7] = out[7]; + unpack64v (tmps, dgst, gid, i + 0, dgst[0]); + unpack64v (tmps, dgst, gid, i + 1, dgst[1]); + unpack64v (tmps, dgst, gid, i + 2, dgst[2]); + unpack64v (tmps, dgst, gid, i + 3, dgst[3]); + unpack64v (tmps, dgst, gid, i + 4, dgst[4]); + unpack64v (tmps, dgst, gid, i + 5, dgst[5]); + unpack64v (tmps, dgst, gid, i + 6, dgst[6]); + unpack64v (tmps, dgst, gid, i + 7, dgst[7]); + + unpack64v (tmps, out, gid, i + 0, out[0]); + unpack64v (tmps, out, gid, i + 1, out[1]); + unpack64v (tmps, out, gid, i + 2, out[2]); + unpack64v (tmps, out, gid, i + 3, out[3]); + unpack64v (tmps, out, gid, i + 4, out[4]); + unpack64v (tmps, out, gid, i + 5, out[5]); + unpack64v (tmps, out, gid, i + 6, out[6]); + unpack64v (tmps, out, gid, i + 7, out[7]); } } diff --git a/include/interface.h b/include/interface.h index 3712a2b8e..94cce825d 100644 --- a/include/interface.h +++ b/include/interface.h @@ -291,7 +291,7 @@ typedef struct keepass typedef struct tc { - u32 salt_buf[16]; + u32 salt_buf[32]; u32 data_buf[112]; u32 keyfile_buf[16]; u32 signature; diff --git a/src/interface.c b/src/interface.c index 5dc53e12e..09fecd216 100644 --- a/src/interface.c +++ b/src/interface.c @@ -22034,6 +22034,7 @@ int hashconfig_init (hashcat_ctx_t *hashcat_ctx) hashconfig->dgst_size = DGST_SIZE_8_8; hashconfig->parse_func = truecrypt_parse_hash_1k; hashconfig->opti_type = OPTI_TYPE_ZERO_BYTE + | OPTI_TYPE_SLOW_HASH_SIMD_LOOP | OPTI_TYPE_USES_BITS_64; hashconfig->dgst_pos0 = 0; hashconfig->dgst_pos1 = 1; @@ -22052,6 +22053,7 @@ int hashconfig_init (hashcat_ctx_t *hashcat_ctx) hashconfig->dgst_size = DGST_SIZE_8_8; hashconfig->parse_func = truecrypt_parse_hash_1k; hashconfig->opti_type = OPTI_TYPE_ZERO_BYTE + | OPTI_TYPE_SLOW_HASH_SIMD_LOOP | OPTI_TYPE_USES_BITS_64; hashconfig->dgst_pos0 = 0; hashconfig->dgst_pos1 = 1; @@ -22070,6 +22072,7 @@ int hashconfig_init (hashcat_ctx_t *hashcat_ctx) hashconfig->dgst_size = DGST_SIZE_8_8; hashconfig->parse_func = truecrypt_parse_hash_1k; hashconfig->opti_type = OPTI_TYPE_ZERO_BYTE + | OPTI_TYPE_SLOW_HASH_SIMD_LOOP | OPTI_TYPE_USES_BITS_64; hashconfig->dgst_pos0 = 0; hashconfig->dgst_pos1 = 1; @@ -24629,14 +24632,14 @@ int hashconfig_init (hashcat_ctx_t *hashcat_ctx) case 15500: hashconfig->pw_max = 16; break; } + } - // some mode are converted to long passwords but without dropping performance, so they don't need the -L option set by user + // pw_max : some algo are converted to long password support but without dropping performance - switch (hashconfig->hash_mode) - { - case 5200: hashconfig->pw_max = PW_MAX; - break; - } + switch (hashconfig->hash_mode) + { + case 5200: hashconfig->pw_max = PW_MAX; + break; } // pw_max : algo specific hard max length @@ -24655,6 +24658,12 @@ int hashconfig_init (hashcat_ctx_t *hashcat_ctx) break; case 3200: hashconfig->pw_max = 72; // bcrypt max break; + case 6221: hashconfig->pw_max = 128; // PBKDF2-HMAC-SHA512 max + break; + case 6222: hashconfig->pw_max = 128; // PBKDF2-HMAC-SHA512 max + break; + case 6223: hashconfig->pw_max = 128; // PBKDF2-HMAC-SHA512 max + break; case 6400: hashconfig->pw_max = 64; // PBKDF2-HMAC-SHA256 max break; case 6500: hashconfig->pw_max = 128; // PBKDF2-HMAC-SHA512 max