From dbbdb7e5ac4f7db2dd503c1820b91fbbcfc37c92 Mon Sep 17 00:00:00 2001 From: Jens Steube Date: Sun, 16 Jun 2019 11:41:42 +0200 Subject: [PATCH] WipZip cracking: Added two byte early reject, resulting in higher cracking speed --- OpenCL/m13600-pure.cl | 117 +++++++++++++++++++------------------ docs/changes.txt | 1 + src/modules/module_13600.c | 21 +++++-- 3 files changed, 77 insertions(+), 62 deletions(-) diff --git a/OpenCL/m13600-pure.cl b/OpenCL/m13600-pure.cl index 944215e82..915b7a5f6 100644 --- a/OpenCL/m13600-pure.cl +++ b/OpenCL/m13600-pure.cl @@ -3,7 +3,7 @@ * License.....: MIT */ -#define NEW_SIMD_CODE +//#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" @@ -43,7 +43,7 @@ typedef struct zip2 } zip2_t; -DECLSPEC void hmac_sha1_run_V (u32x *w0, u32x *w1, u32x *w2, u32x *w3, u32x *ipad, u32x *opad, u32x *digest) +DECLSPEC void hmac_sha1_run (u32 *w0, u32 *w1, u32 *w2, u32 *w3, u32 *ipad, u32 *opad, u32 *digest) { digest[0] = ipad[0]; digest[1] = ipad[1]; @@ -51,7 +51,7 @@ DECLSPEC void hmac_sha1_run_V (u32x *w0, u32x *w1, u32x *w2, u32x *w3, u32x *ipa digest[3] = ipad[3]; digest[4] = ipad[4]; - sha1_transform_vector (w0, w1, w2, w3, digest); + sha1_transform (w0, w1, w2, w3, digest); w0[0] = digest[0]; w0[1] = digest[1]; @@ -76,7 +76,7 @@ DECLSPEC void hmac_sha1_run_V (u32x *w0, u32x *w1, u32x *w2, u32x *w3, u32x *ipa digest[3] = opad[3]; digest[4] = opad[4]; - sha1_transform_vector (w0, w1, w2, w3, digest); + sha1_transform (w0, w1, w2, w3, digest); } KERNEL_FQ void m13600_init (KERN_ATTR_TMPS_ESALT (pbkdf2_sha1_tmp_t, zip2_t)) @@ -131,27 +131,23 @@ KERNEL_FQ void m13600_init (KERN_ATTR_TMPS_ESALT (pbkdf2_sha1_tmp_t, zip2_t)) const u32 mode = esalt_bufs[digests_offset].mode; - u32 iter_start; - u32 iter_stop; - u32 count_start; + int iter_start; + int iter_stop; switch (mode) { case 1: iter_start = 0; iter_stop = 2; - count_start = 1; break; case 2: iter_start = 1; iter_stop = 3; - count_start = 2; break; case 3: iter_start = 1; iter_stop = 4; - count_start = 2; break; } - for (u32 i = iter_start, j = count_start; i < iter_stop; i++, j++) + for (int i = iter_stop - 1, j = iter_stop; i >= iter_start; i--, j--) { sha1_hmac_ctx_t sha1_hmac_ctx2 = sha1_hmac_ctx; @@ -196,70 +192,68 @@ KERNEL_FQ void m13600_loop (KERN_ATTR_TMPS_ESALT (pbkdf2_sha1_tmp_t, zip2_t)) { const u64 gid = get_global_id (0); - if ((gid * VECT_SIZE) >= gid_max) return; + if (gid >= gid_max) return; - u32x ipad[5]; - u32x opad[5]; + u32 ipad[5]; + u32 opad[5]; - ipad[0] = packv (tmps, ipad, gid, 0); - ipad[1] = packv (tmps, ipad, gid, 1); - ipad[2] = packv (tmps, ipad, gid, 2); - ipad[3] = packv (tmps, ipad, gid, 3); - ipad[4] = packv (tmps, ipad, gid, 4); + 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]; - 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[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]; + + const u32 verify_bytes = esalt_bufs[digests_offset].verify_bytes; const u32 mode = esalt_bufs[digests_offset].mode; - u32 iter_start; - u32 iter_stop; - u32 count_start; + int iter_start; + int iter_stop; switch (mode) { case 1: iter_start = 0; iter_stop = 2; - count_start = 1; break; case 2: iter_start = 1; iter_stop = 3; - count_start = 2; break; case 3: iter_start = 1; iter_stop = 4; - count_start = 2; break; } - for (u32 i = iter_start, j = count_start; i < iter_stop; i++, j++) + for (int i = iter_stop - 1; i >= iter_start; i--) { const u32 i5 = i * 5; - u32x dgst[5]; - u32x out[5]; + u32 dgst[5]; + u32 out[5]; - dgst[0] = packv (tmps, dgst, gid, i5 + 0); - dgst[1] = packv (tmps, dgst, gid, i5 + 1); - dgst[2] = packv (tmps, dgst, gid, i5 + 2); - dgst[3] = packv (tmps, dgst, gid, i5 + 3); - dgst[4] = packv (tmps, dgst, gid, i5 + 4); + dgst[0] = tmps[gid].dgst[i5 + 0]; + dgst[1] = tmps[gid].dgst[i5 + 1]; + dgst[2] = tmps[gid].dgst[i5 + 2]; + dgst[3] = tmps[gid].dgst[i5 + 3]; + dgst[4] = tmps[gid].dgst[i5 + 4]; - out[0] = packv (tmps, out, gid, i5 + 0); - out[1] = packv (tmps, out, gid, i5 + 1); - out[2] = packv (tmps, out, gid, i5 + 2); - out[3] = packv (tmps, out, gid, i5 + 3); - out[4] = packv (tmps, out, gid, i5 + 4); + out[0] = tmps[gid].out[i5 + 0]; + out[1] = tmps[gid].out[i5 + 1]; + out[2] = tmps[gid].out[i5 + 2]; + out[3] = tmps[gid].out[i5 + 3]; + out[4] = tmps[gid].out[i5 + 4]; for (u32 j = 0; j < loop_cnt; j++) { - u32x w0[4]; - u32x w1[4]; - u32x w2[4]; - u32x w3[4]; + u32 w0[4]; + u32 w1[4]; + u32 w2[4]; + u32 w3[4]; w0[0] = dgst[0]; w0[1] = dgst[1]; @@ -278,7 +272,7 @@ KERNEL_FQ void m13600_loop (KERN_ATTR_TMPS_ESALT (pbkdf2_sha1_tmp_t, zip2_t)) w3[2] = 0; w3[3] = (64 + 20) * 8; - hmac_sha1_run_V (w0, w1, w2, w3, ipad, opad, dgst); + hmac_sha1_run (w0, w1, w2, w3, ipad, opad, dgst); out[0] ^= dgst[0]; out[1] ^= dgst[1]; @@ -287,17 +281,24 @@ KERNEL_FQ void m13600_loop (KERN_ATTR_TMPS_ESALT (pbkdf2_sha1_tmp_t, zip2_t)) out[4] ^= dgst[4]; } - unpackv (tmps, dgst, gid, i5 + 0, dgst[0]); - unpackv (tmps, dgst, gid, i5 + 1, dgst[1]); - unpackv (tmps, dgst, gid, i5 + 2, dgst[2]); - unpackv (tmps, dgst, gid, i5 + 3, dgst[3]); - unpackv (tmps, dgst, gid, i5 + 4, dgst[4]); + if (i == iter_stop - 1) // 2 byte optimization check + { + if (mode == 1) if ((out[3] >> 16) != verify_bytes) break; + if (mode == 2) if ((out[2] >> 16) != verify_bytes) break; + if (mode == 3) if ((out[1] >> 16) != verify_bytes) break; + } - unpackv (tmps, out, gid, i5 + 0, out[0]); - unpackv (tmps, out, gid, i5 + 1, out[1]); - unpackv (tmps, out, gid, i5 + 2, out[2]); - unpackv (tmps, out, gid, i5 + 3, out[3]); - unpackv (tmps, out, gid, i5 + 4, out[4]); + tmps[gid].dgst[i5 + 0] = dgst[0]; + tmps[gid].dgst[i5 + 1] = dgst[1]; + tmps[gid].dgst[i5 + 2] = dgst[2]; + tmps[gid].dgst[i5 + 3] = dgst[3]; + tmps[gid].dgst[i5 + 4] = dgst[4]; + + tmps[gid].out[i5 + 0] = out[0]; + tmps[gid].out[i5 + 1] = out[1]; + tmps[gid].out[i5 + 2] = out[2]; + tmps[gid].out[i5 + 3] = out[3]; + tmps[gid].out[i5 + 4] = out[4]; } } diff --git a/docs/changes.txt b/docs/changes.txt index d11b9d156..d28b45430 100644 --- a/docs/changes.txt +++ b/docs/changes.txt @@ -72,6 +72,7 @@ - Startup Screen: Provide an estimate of host memory requirements for the requested attack - Tuning Database: Updated hashcat.hctune with new models and refreshed vector width values - WPA/WPA2 cracking: In the potfile, replace password with PMK in order to detect already cracked networks across all WPA modes +- WipZip cracking: Added two byte early reject, resulting in higher cracking speed ## ## Technical diff --git a/src/modules/module_13600.c b/src/modules/module_13600.c index 1d048060e..562841524 100644 --- a/src/modules/module_13600.c +++ b/src/modules/module_13600.c @@ -19,8 +19,7 @@ static const u32 DGST_SIZE = DGST_SIZE_4_4; static const u32 HASH_CATEGORY = HASH_CATEGORY_ARCHIVE; static const char *HASH_NAME = "WinZip"; static const u64 KERN_TYPE = 13600; -static const u32 OPTI_TYPE = OPTI_TYPE_ZERO_BYTE - | OPTI_TYPE_SLOW_HASH_SIMD_LOOP; +static const u32 OPTI_TYPE = OPTI_TYPE_ZERO_BYTE; static const u64 OPTS_TYPE = OPTS_TYPE_PT_GENERATE_LE; static const u32 SALT_TYPE = SALT_TYPE_EMBEDDED; static const char *ST_PASS = "hashcat"; @@ -85,6 +84,20 @@ u64 module_tmp_size (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED c return tmp_size; } +u32 module_kernel_loops_min (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) +{ + const u32 kernel_loops_min = ROUNDS_ZIP2 - 1; + + return kernel_loops_min; +} + +u32 module_kernel_loops_max (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) +{ + const u32 kernel_loops_max = ROUNDS_ZIP2 - 1; + + return kernel_loops_max; +} + u32 module_pw_max (MAYBE_UNUSED const hashconfig_t *hashconfig, MAYBE_UNUSED const user_options_t *user_options, MAYBE_UNUSED const user_options_extra_t *user_options_extra) { // this overrides the reductions of PW_MAX in case optimized kernel is selected @@ -432,8 +445,8 @@ void module_init (module_ctx_t *module_ctx) module_ctx->module_jit_cache_disable = MODULE_DEFAULT; module_ctx->module_kernel_accel_max = MODULE_DEFAULT; module_ctx->module_kernel_accel_min = MODULE_DEFAULT; - module_ctx->module_kernel_loops_max = MODULE_DEFAULT; - module_ctx->module_kernel_loops_min = MODULE_DEFAULT; + module_ctx->module_kernel_loops_max = module_kernel_loops_max; + module_ctx->module_kernel_loops_min = module_kernel_loops_min; module_ctx->module_kernel_threads_max = MODULE_DEFAULT; module_ctx->module_kernel_threads_min = MODULE_DEFAULT; module_ctx->module_kern_type = module_kern_type;