X-Git-Url: https://www.flypig.org.uk/git/?a=blobdiff_plain;f=OpenCL%2Fm10900.cl;h=c1ec4c4155602667230bffbc109514841dde8ebc;hb=161a6eb4bc643d8e636e96eda613f5137d30da59;hp=d30290a0f3daae4723389eb8b96f15198f702e7b;hpb=e4465d7dcb88426532f6b0488cf9bc1cd3b06cc6;p=hashcat.git diff --git a/OpenCL/m10900.cl b/OpenCL/m10900.cl index d30290a..c1ec4c4 100644 --- a/OpenCL/m10900.cl +++ b/OpenCL/m10900.cl @@ -5,21 +5,17 @@ #define _PBKDF2_SHA256_ -#include "include/constants.h" -#include "include/kernel_vendor.h" +#define NEW_SIMD_CODE -#define DGST_R0 0 -#define DGST_R1 1 -#define DGST_R2 2 -#define DGST_R3 3 +#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 "include/kernel_functions.c" - -#include "types_ocl.c" -#include "common.c" - -#define COMPARE_S "check_single_comp4.c" -#define COMPARE_M "check_multi_comp4.c" +#define COMPARE_S "inc_comp_single.cl" +#define COMPARE_M "inc_comp_multi.cl" __constant u32 k_sha256[64] = { @@ -41,7 +37,7 @@ __constant u32 k_sha256[64] = SHA256C3c, SHA256C3d, SHA256C3e, SHA256C3f, }; -static void sha256_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[8]) +void sha256_transform_S (const u32 w0[4], const u32 w1[4], const u32 w2[4], const u32 w3[4], u32 digest[8]) { u32 a = digest[0]; u32 b = digest[1]; @@ -69,6 +65,195 @@ static void sha256_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], u32 we_t = w3[2]; u32 wf_t = w3[3]; + #define ROUND_EXPAND_S() \ + { \ + w0_t = SHA256_EXPAND_S (we_t, w9_t, w1_t, w0_t); \ + w1_t = SHA256_EXPAND_S (wf_t, wa_t, w2_t, w1_t); \ + w2_t = SHA256_EXPAND_S (w0_t, wb_t, w3_t, w2_t); \ + w3_t = SHA256_EXPAND_S (w1_t, wc_t, w4_t, w3_t); \ + w4_t = SHA256_EXPAND_S (w2_t, wd_t, w5_t, w4_t); \ + w5_t = SHA256_EXPAND_S (w3_t, we_t, w6_t, w5_t); \ + w6_t = SHA256_EXPAND_S (w4_t, wf_t, w7_t, w6_t); \ + w7_t = SHA256_EXPAND_S (w5_t, w0_t, w8_t, w7_t); \ + w8_t = SHA256_EXPAND_S (w6_t, w1_t, w9_t, w8_t); \ + w9_t = SHA256_EXPAND_S (w7_t, w2_t, wa_t, w9_t); \ + wa_t = SHA256_EXPAND_S (w8_t, w3_t, wb_t, wa_t); \ + wb_t = SHA256_EXPAND_S (w9_t, w4_t, wc_t, wb_t); \ + wc_t = SHA256_EXPAND_S (wa_t, w5_t, wd_t, wc_t); \ + wd_t = SHA256_EXPAND_S (wb_t, w6_t, we_t, wd_t); \ + we_t = SHA256_EXPAND_S (wc_t, w7_t, wf_t, we_t); \ + wf_t = SHA256_EXPAND_S (wd_t, w8_t, w0_t, wf_t); \ + } + + #define ROUND_STEP_S(i) \ + { \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w0_t, k_sha256[i + 0]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w1_t, k_sha256[i + 1]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, w2_t, k_sha256[i + 2]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, w3_t, k_sha256[i + 3]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, w4_t, k_sha256[i + 4]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, w5_t, k_sha256[i + 5]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, w6_t, k_sha256[i + 6]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, w7_t, k_sha256[i + 7]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, a, b, c, d, e, f, g, h, w8_t, k_sha256[i + 8]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, h, a, b, c, d, e, f, g, w9_t, k_sha256[i + 9]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, g, h, a, b, c, d, e, f, wa_t, k_sha256[i + 10]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, f, g, h, a, b, c, d, e, wb_t, k_sha256[i + 11]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, e, f, g, h, a, b, c, d, wc_t, k_sha256[i + 12]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, d, e, f, g, h, a, b, c, wd_t, k_sha256[i + 13]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, c, d, e, f, g, h, a, b, we_t, k_sha256[i + 14]); \ + SHA256_STEP_S (SHA256_F0o, SHA256_F1o, b, c, d, e, f, g, h, a, wf_t, k_sha256[i + 15]); \ + } + + ROUND_STEP_S (0); + + #ifdef _unroll + #pragma unroll + #endif + for (int i = 16; i < 64; i += 16) + { + ROUND_EXPAND_S (); ROUND_STEP_S (i); + } + + digest[0] += a; + digest[1] += b; + digest[2] += c; + digest[3] += d; + digest[4] += e; + digest[5] += f; + digest[6] += g; + digest[7] += h; +} + +void hmac_sha256_pad_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[8], u32 opad[8]) +{ + w0[0] = w0[0] ^ 0x36363636; + w0[1] = w0[1] ^ 0x36363636; + w0[2] = w0[2] ^ 0x36363636; + w0[3] = w0[3] ^ 0x36363636; + w1[0] = w1[0] ^ 0x36363636; + w1[1] = w1[1] ^ 0x36363636; + w1[2] = w1[2] ^ 0x36363636; + w1[3] = w1[3] ^ 0x36363636; + w2[0] = w2[0] ^ 0x36363636; + w2[1] = w2[1] ^ 0x36363636; + w2[2] = w2[2] ^ 0x36363636; + w2[3] = w2[3] ^ 0x36363636; + w3[0] = w3[0] ^ 0x36363636; + w3[1] = w3[1] ^ 0x36363636; + w3[2] = w3[2] ^ 0x36363636; + w3[3] = w3[3] ^ 0x36363636; + + ipad[0] = SHA256M_A; + ipad[1] = SHA256M_B; + ipad[2] = SHA256M_C; + ipad[3] = SHA256M_D; + ipad[4] = SHA256M_E; + ipad[5] = SHA256M_F; + ipad[6] = SHA256M_G; + ipad[7] = SHA256M_H; + + sha256_transform_S (w0, w1, w2, w3, ipad); + + w0[0] = w0[0] ^ 0x6a6a6a6a; + w0[1] = w0[1] ^ 0x6a6a6a6a; + w0[2] = w0[2] ^ 0x6a6a6a6a; + w0[3] = w0[3] ^ 0x6a6a6a6a; + w1[0] = w1[0] ^ 0x6a6a6a6a; + w1[1] = w1[1] ^ 0x6a6a6a6a; + w1[2] = w1[2] ^ 0x6a6a6a6a; + w1[3] = w1[3] ^ 0x6a6a6a6a; + w2[0] = w2[0] ^ 0x6a6a6a6a; + w2[1] = w2[1] ^ 0x6a6a6a6a; + w2[2] = w2[2] ^ 0x6a6a6a6a; + w2[3] = w2[3] ^ 0x6a6a6a6a; + w3[0] = w3[0] ^ 0x6a6a6a6a; + w3[1] = w3[1] ^ 0x6a6a6a6a; + w3[2] = w3[2] ^ 0x6a6a6a6a; + w3[3] = w3[3] ^ 0x6a6a6a6a; + + opad[0] = SHA256M_A; + opad[1] = SHA256M_B; + opad[2] = SHA256M_C; + opad[3] = SHA256M_D; + opad[4] = SHA256M_E; + opad[5] = SHA256M_F; + opad[6] = SHA256M_G; + opad[7] = SHA256M_H; + + sha256_transform_S (w0, w1, w2, w3, opad); +} + +void hmac_sha256_run_S (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[8], u32 opad[8], u32 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]; + + sha256_transform_S (w0, w1, w2, w3, digest); + + w0[0] = digest[0]; + w0[1] = digest[1]; + w0[2] = digest[2]; + w0[3] = digest[3]; + w1[0] = digest[4]; + w1[1] = digest[5]; + w1[2] = digest[6]; + w1[3] = digest[7]; + w2[0] = 0x80000000; + w2[1] = 0; + w2[2] = 0; + w2[3] = 0; + w3[0] = 0; + w3[1] = 0; + w3[2] = 0; + w3[3] = (64 + 32) * 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]; + + sha256_transform_S (w0, w1, w2, w3, digest); +} + +void sha256_transform_V (const u32x w0[4], const u32x w1[4], const u32x w2[4], const u32x w3[4], u32x digest[8]) +{ + u32x a = digest[0]; + u32x b = digest[1]; + u32x c = digest[2]; + u32x d = digest[3]; + u32x e = digest[4]; + u32x f = digest[5]; + u32x g = digest[6]; + u32x h = digest[7]; + + u32x w0_t = w0[0]; + u32x w1_t = w0[1]; + u32x w2_t = w0[2]; + u32x w3_t = w0[3]; + u32x w4_t = w1[0]; + u32x w5_t = w1[1]; + u32x w6_t = w1[2]; + u32x w7_t = w1[3]; + u32x w8_t = w2[0]; + u32x w9_t = w2[1]; + u32x wa_t = w2[2]; + u32x wb_t = w2[3]; + u32x wc_t = w3[0]; + u32x wd_t = w3[1]; + u32x we_t = w3[2]; + u32x wf_t = w3[3]; + #define ROUND_EXPAND() \ { \ w0_t = SHA256_EXPAND (we_t, w9_t, w1_t, w0_t); \ @@ -111,7 +296,9 @@ static void sha256_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], ROUND_STEP (0); + #ifdef _unroll #pragma unroll + #endif for (int i = 16; i < 64; i += 16) { ROUND_EXPAND (); ROUND_STEP (i); @@ -127,7 +314,7 @@ static void sha256_transform (const u32 w0[4], const u32 w1[4], const u32 w2[4], digest[7] += h; } -static void hmac_sha256_pad (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[8], u32 opad[8]) +void hmac_sha256_pad_V (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x ipad[8], u32x opad[8]) { w0[0] = w0[0] ^ 0x36363636; w0[1] = w0[1] ^ 0x36363636; @@ -155,7 +342,7 @@ static void hmac_sha256_pad (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipa ipad[6] = SHA256M_G; ipad[7] = SHA256M_H; - sha256_transform (w0, w1, w2, w3, ipad); + sha256_transform_V (w0, w1, w2, w3, ipad); w0[0] = w0[0] ^ 0x6a6a6a6a; w0[1] = w0[1] ^ 0x6a6a6a6a; @@ -183,10 +370,10 @@ static void hmac_sha256_pad (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipa opad[6] = SHA256M_G; opad[7] = SHA256M_H; - sha256_transform (w0, w1, w2, w3, opad); + sha256_transform_V (w0, w1, w2, w3, opad); } -static void hmac_sha256_run (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipad[8], u32 opad[8], u32 digest[8]) +void hmac_sha256_run_V (u32x w0[4], u32x w1[4], u32x w2[4], u32x w3[4], u32x ipad[8], u32x opad[8], u32x digest[8]) { digest[0] = ipad[0]; digest[1] = ipad[1]; @@ -197,7 +384,7 @@ static void hmac_sha256_run (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipa digest[6] = ipad[6]; digest[7] = ipad[7]; - sha256_transform (w0, w1, w2, w3, digest); + sha256_transform_V (w0, w1, w2, w3, digest); w0[0] = digest[0]; w0[1] = digest[1]; @@ -225,10 +412,10 @@ static void hmac_sha256_run (u32 w0[4], u32 w1[4], u32 w2[4], u32 w3[4], u32 ipa digest[6] = opad[6]; digest[7] = opad[7]; - sha256_transform (w0, w1, w2, w3, digest); + sha256_transform_V (w0, w1, w2, w3, digest); } -__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_init (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha256_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_sha256_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max) +__kernel void m10900_init (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha256_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_sha256_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) { /** * base @@ -240,31 +427,31 @@ __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_init (__gl u32 w0[4]; - w0[0] = swap32 (pws[gid].i[ 0]); - w0[1] = swap32 (pws[gid].i[ 1]); - w0[2] = swap32 (pws[gid].i[ 2]); - w0[3] = swap32 (pws[gid].i[ 3]); + w0[0] = swap32_S (pws[gid].i[ 0]); + w0[1] = swap32_S (pws[gid].i[ 1]); + w0[2] = swap32_S (pws[gid].i[ 2]); + w0[3] = swap32_S (pws[gid].i[ 3]); u32 w1[4]; - w1[0] = swap32 (pws[gid].i[ 4]); - w1[1] = swap32 (pws[gid].i[ 5]); - w1[2] = swap32 (pws[gid].i[ 6]); - w1[3] = swap32 (pws[gid].i[ 7]); + w1[0] = swap32_S (pws[gid].i[ 4]); + w1[1] = swap32_S (pws[gid].i[ 5]); + w1[2] = swap32_S (pws[gid].i[ 6]); + w1[3] = swap32_S (pws[gid].i[ 7]); u32 w2[4]; - w2[0] = swap32 (pws[gid].i[ 8]); - w2[1] = swap32 (pws[gid].i[ 9]); - w2[2] = swap32 (pws[gid].i[10]); - w2[3] = swap32 (pws[gid].i[11]); + w2[0] = swap32_S (pws[gid].i[ 8]); + w2[1] = swap32_S (pws[gid].i[ 9]); + w2[2] = swap32_S (pws[gid].i[10]); + w2[3] = swap32_S (pws[gid].i[11]); u32 w3[4]; - w3[0] = swap32 (pws[gid].i[12]); - w3[1] = swap32 (pws[gid].i[13]); - w3[2] = swap32 (pws[gid].i[14]); - w3[3] = swap32 (pws[gid].i[15]); + w3[0] = swap32_S (pws[gid].i[12]); + w3[1] = swap32_S (pws[gid].i[13]); + w3[2] = swap32_S (pws[gid].i[14]); + w3[3] = swap32_S (pws[gid].i[15]); /** * salt @@ -277,27 +464,27 @@ __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_init (__gl u32 esalt_buf2[4]; u32 esalt_buf3[4]; - esalt_buf0[0] = swap32 (esalt_bufs[salt_pos].salt_buf[ 0]); - esalt_buf0[1] = swap32 (esalt_bufs[salt_pos].salt_buf[ 1]); - esalt_buf0[2] = swap32 (esalt_bufs[salt_pos].salt_buf[ 2]); - esalt_buf0[3] = swap32 (esalt_bufs[salt_pos].salt_buf[ 3]); - esalt_buf1[0] = swap32 (esalt_bufs[salt_pos].salt_buf[ 4]); - esalt_buf1[1] = swap32 (esalt_bufs[salt_pos].salt_buf[ 5]); - esalt_buf1[2] = swap32 (esalt_bufs[salt_pos].salt_buf[ 6]); - esalt_buf1[3] = swap32 (esalt_bufs[salt_pos].salt_buf[ 7]); - esalt_buf2[0] = swap32 (esalt_bufs[salt_pos].salt_buf[ 8]); - esalt_buf2[1] = swap32 (esalt_bufs[salt_pos].salt_buf[ 9]); - esalt_buf2[2] = swap32 (esalt_bufs[salt_pos].salt_buf[10]); - esalt_buf2[3] = swap32 (esalt_bufs[salt_pos].salt_buf[11]); - esalt_buf3[0] = swap32 (esalt_bufs[salt_pos].salt_buf[12]); - esalt_buf3[1] = swap32 (esalt_bufs[salt_pos].salt_buf[13]); + esalt_buf0[0] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 0]); + esalt_buf0[1] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 1]); + esalt_buf0[2] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 2]); + esalt_buf0[3] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 3]); + esalt_buf1[0] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 4]); + esalt_buf1[1] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 5]); + esalt_buf1[2] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 6]); + esalt_buf1[3] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 7]); + esalt_buf2[0] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 8]); + esalt_buf2[1] = swap32_S (esalt_bufs[salt_pos].salt_buf[ 9]); + esalt_buf2[2] = swap32_S (esalt_bufs[salt_pos].salt_buf[10]); + esalt_buf2[3] = swap32_S (esalt_bufs[salt_pos].salt_buf[11]); + esalt_buf3[0] = swap32_S (esalt_bufs[salt_pos].salt_buf[12]); + esalt_buf3[1] = swap32_S (esalt_bufs[salt_pos].salt_buf[13]); esalt_buf3[2] = 0; esalt_buf3[3] = (64 + salt_len + 4) * 8; u32 ipad[8]; u32 opad[8]; - hmac_sha256_pad (w0, w1, w2, w3, ipad, opad); + hmac_sha256_pad_S (w0, w1, w2, w3, ipad, opad); tmps[gid].ipad[0] = ipad[0]; tmps[gid].ipad[1] = ipad[1]; @@ -321,7 +508,7 @@ __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_init (__gl { u32 dgst[8]; - hmac_sha256_run (esalt_buf0, esalt_buf1, esalt_buf2, esalt_buf3, ipad, opad, dgst); + hmac_sha256_run_S (esalt_buf0, esalt_buf1, esalt_buf2, esalt_buf3, ipad, opad, dgst); tmps[gid].dgst[i + 0] = dgst[0]; tmps[gid].dgst[i + 1] = dgst[1]; @@ -343,64 +530,62 @@ __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_init (__gl } } -__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_loop (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha256_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_sha256_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max) +__kernel void m10900_loop (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha256_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_sha256_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) { const u32 gid = get_global_id (0); if (gid >= gid_max) return; - u32 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]; - - u32 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]; + u32x ipad[8]; + u32x opad[8]; + + ipad[0] = packv (tmps, ipad, gid, 0); + ipad[1] = packv (tmps, ipad, gid, 1); + ipad[2] = packv (tmps, ipad, gid, 2); + ipad[3] = packv (tmps, ipad, gid, 3); + ipad[4] = packv (tmps, ipad, gid, 4); + ipad[5] = packv (tmps, ipad, gid, 5); + ipad[6] = packv (tmps, ipad, gid, 6); + ipad[7] = packv (tmps, ipad, gid, 7); + + opad[0] = packv (tmps, opad, gid, 0); + opad[1] = packv (tmps, opad, gid, 1); + opad[2] = packv (tmps, opad, gid, 2); + opad[3] = packv (tmps, opad, gid, 3); + opad[4] = packv (tmps, opad, gid, 4); + opad[5] = packv (tmps, opad, gid, 5); + opad[6] = packv (tmps, opad, gid, 6); + opad[7] = packv (tmps, opad, gid, 7); for (u32 i = 0; i < 8; i += 8) { - u32 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]; - - u32 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]; + u32x dgst[8]; + u32x out[8]; + + dgst[0] = packv (tmps, dgst, gid, 0); + dgst[1] = packv (tmps, dgst, gid, 1); + dgst[2] = packv (tmps, dgst, gid, 2); + dgst[3] = packv (tmps, dgst, gid, 3); + dgst[4] = packv (tmps, dgst, gid, 4); + dgst[5] = packv (tmps, dgst, gid, 5); + dgst[6] = packv (tmps, dgst, gid, 6); + dgst[7] = packv (tmps, dgst, gid, 7); + + out[0] = packv (tmps, out, gid, 0); + out[1] = packv (tmps, out, gid, 1); + out[2] = packv (tmps, out, gid, 2); + out[3] = packv (tmps, out, gid, 3); + out[4] = packv (tmps, out, gid, 4); + out[5] = packv (tmps, out, gid, 5); + out[6] = packv (tmps, out, gid, 6); + out[7] = packv (tmps, out, gid, 7); for (u32 j = 0; j < loop_cnt; j++) { - u32 w0[4]; - u32 w1[4]; - u32 w2[4]; - u32 w3[4]; + u32x w0[4]; + u32x w1[4]; + u32x w2[4]; + u32x w3[4]; w0[0] = dgst[0]; w0[1] = dgst[1]; @@ -419,7 +604,7 @@ __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_loop (__gl w3[2] = 0; w3[3] = (64 + 32) * 8; - hmac_sha256_run (w0, w1, w2, w3, ipad, opad, dgst); + hmac_sha256_run_V (w0, w1, w2, w3, ipad, opad, dgst); out[0] ^= dgst[0]; out[1] ^= dgst[1]; @@ -431,27 +616,27 @@ __kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_loop (__gl 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]; + unpackv (tmps, dgst, gid, 0, dgst[0]); + unpackv (tmps, dgst, gid, 1, dgst[1]); + unpackv (tmps, dgst, gid, 2, dgst[2]); + unpackv (tmps, dgst, gid, 3, dgst[3]); + unpackv (tmps, dgst, gid, 4, dgst[4]); + unpackv (tmps, dgst, gid, 5, dgst[5]); + unpackv (tmps, dgst, gid, 6, dgst[6]); + unpackv (tmps, dgst, gid, 7, dgst[7]); + + unpackv (tmps, out, gid, 0, out[0]); + unpackv (tmps, out, gid, 1, out[1]); + unpackv (tmps, out, gid, 2, out[2]); + unpackv (tmps, out, gid, 3, out[3]); + unpackv (tmps, out, gid, 4, out[4]); + unpackv (tmps, out, gid, 5, out[5]); + unpackv (tmps, out, gid, 6, out[6]); + unpackv (tmps, out, gid, 7, out[7]); } } -__kernel void __attribute__((reqd_work_group_size (64, 1, 1))) m10900_comp (__global pw_t *pws, __global gpu_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha256_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_sha256_t *esalt_bufs, __global u32 *d_return_buf, __global u32 *d_scryptV_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 rules_cnt, const u32 digests_cnt, const u32 digests_offset, const u32 combs_mode, const u32 gid_max) +__kernel void m10900_comp (__global pw_t *pws, __global kernel_rule_t *rules_buf, __global comb_t *combs_buf, __global bf_t *bfs_buf, __global pbkdf2_sha256_tmp_t *tmps, __global void *hooks, __global u32 *bitmaps_buf_s1_a, __global u32 *bitmaps_buf_s1_b, __global u32 *bitmaps_buf_s1_c, __global u32 *bitmaps_buf_s1_d, __global u32 *bitmaps_buf_s2_a, __global u32 *bitmaps_buf_s2_b, __global u32 *bitmaps_buf_s2_c, __global u32 *bitmaps_buf_s2_d, __global plain_t *plains_buf, __global digest_t *digests_buf, __global u32 *hashes_shown, __global salt_t *salt_bufs, __global pbkdf2_sha256_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) { /** * base