/** * Author......: Netherlands Forensic Institute * License.....: MIT */ /* Pseudocode: 1. sha256(sha256(password=masterkey)||keyfile) = argon.in 2. argon2(salt=transformseed, password=argon2.in) = argon2.out 2. sha512(masterseed||argon2.out||0x01) = final 3. sha512(0xFFFFFFFFFFFFFFFF||final) = out 4. hmac_sha256(init=out, data=header) = header_hmac 5. compare header_hmac to hash */ #ifdef KERNEL_STATIC #include M2S(INCLUDE_PATH/inc_platform.cl) #include M2S(INCLUDE_PATH/inc_common.cl) #include M2S(INCLUDE_PATH/inc_hash_blake2b.cl) #include M2S(INCLUDE_PATH/inc_hash_argon2.cl) #include M2S(INCLUDE_PATH/inc_hash_sha512.cl) #include M2S(INCLUDE_PATH/inc_hash_sha256.cl) #endif #define COMPARE_S M2S(INCLUDE_PATH/inc_comp_single.cl) #define COMPARE_M M2S(INCLUDE_PATH/inc_comp_multi.cl) typedef struct argon2_tmp { u32 state[4]; } argon2_tmp_t; typedef struct keepass4 { u32 masterseed[8]; u32 header[64]; /* key-file handling */ u32 keyfile_len; u32 keyfile[8]; } keepass4_t; typedef struct merged_options { argon2_options_t argon2_options; keepass4_t keepass4; } merged_options_t; KERNEL_FQ KERNEL_FA void m34300_init (KERN_ATTR_TMPS_ESALT (argon2_tmp_t, merged_options_t)) { const u64 gid = get_global_id (0); if (gid >= GID_CNT) return; const u32 gd4 = gid / 4; const u32 gm4 = gid % 4; GLOBAL_AS void *V; switch (gm4) { case 0: V = d_extra0_buf; break; case 1: V = d_extra1_buf; break; case 2: V = d_extra2_buf; break; case 3: V = d_extra3_buf; break; } const argon2_options_t argon2_options = esalt_bufs[DIGESTS_OFFSET_HOST].argon2_options; GLOBAL_AS argon2_block_t *argon2_block = get_argon2_block (&argon2_options, V, gd4); GLOBAL_AS const keepass4_t *keepass4 = &esalt_bufs[DIGESTS_OFFSET_HOST].keepass4; sha256_ctx_t ctx0; sha256_init (&ctx0); sha256_update_global_swap (&ctx0, pws[gid].i, pws[gid].pw_len); sha256_final (&ctx0); sha256_ctx_t ctx; sha256_init (&ctx); ctx.w0[0] = ctx0.h[0]; ctx.w0[1] = ctx0.h[1]; ctx.w0[2] = ctx0.h[2]; ctx.w0[3] = ctx0.h[3]; ctx.w1[0] = ctx0.h[4]; ctx.w1[1] = ctx0.h[5]; ctx.w1[2] = ctx0.h[6]; ctx.w1[3] = ctx0.h[7]; ctx.len = 32; if (keepass4->keyfile_len) { ctx.w2[0] = keepass4->keyfile[0]; ctx.w2[1] = keepass4->keyfile[1]; ctx.w2[2] = keepass4->keyfile[2]; ctx.w2[3] = keepass4->keyfile[3]; ctx.w3[0] = keepass4->keyfile[4]; ctx.w3[1] = keepass4->keyfile[5]; ctx.w3[2] = keepass4->keyfile[6]; ctx.w3[3] = keepass4->keyfile[7]; ctx.len += 32; sha256_transform (ctx.w0, ctx.w1, ctx.w2, ctx.w3, ctx.h); ctx.w0[0] = 0; ctx.w0[1] = 0; ctx.w0[2] = 0; ctx.w0[3] = 0; ctx.w1[0] = 0; ctx.w1[1] = 0; ctx.w1[2] = 0; ctx.w1[3] = 0; ctx.w2[0] = 0; ctx.w2[1] = 0; ctx.w2[2] = 0; ctx.w2[3] = 0; ctx.w3[0] = 0; ctx.w3[1] = 0; ctx.w3[2] = 0; ctx.w3[3] = 0; } sha256_final (&ctx); pw_t pw; pw.i[ 0] = hc_swap32_S (ctx.h[0]); pw.i[ 1] = hc_swap32_S (ctx.h[1]); pw.i[ 2] = hc_swap32_S (ctx.h[2]); pw.i[ 3] = hc_swap32_S (ctx.h[3]); pw.i[ 4] = hc_swap32_S (ctx.h[4]); pw.i[ 5] = hc_swap32_S (ctx.h[5]); pw.i[ 6] = hc_swap32_S (ctx.h[6]); pw.i[ 7] = hc_swap32_S (ctx.h[7]); pw.i[ 8] = 0; pw.i[ 9] = 0; pw.i[10] = 0; pw.i[11] = 0; pw.i[12] = 0; pw.i[13] = 0; pw.i[14] = 0; pw.i[15] = 0; pw.pw_len = 32; // output of sha256 is always 32 bytes argon2_init_pg (&pw, &salt_bufs[SALT_POS_HOST], &argon2_options, argon2_block); } KERNEL_FQ KERNEL_FA void m34300_loop (KERN_ATTR_TMPS_ESALT (argon2_tmp_t, merged_options_t)) { const u64 gid = get_global_id (0); const u64 bid = get_group_id (0); const u64 lid = get_local_id (1); const u64 lsz = get_local_size (1); if (bid >= GID_CNT) return; const u32 argon2_thread = get_local_id (0); const u32 argon2_lsz = get_local_size (0); #ifdef ARGON2_PARALLELISM LOCAL_VK u64 shuffle_bufs[ARGON2_PARALLELISM][32]; #else LOCAL_VK u64 shuffle_bufs[32][32]; #endif LOCAL_AS u64 *shuffle_buf = shuffle_bufs[lid]; SYNC_THREADS(); const u32 bd4 = bid / 4; const u32 bm4 = bid % 4; GLOBAL_AS void *V; switch (bm4) { case 0: V = d_extra0_buf; break; case 1: V = d_extra1_buf; break; case 2: V = d_extra2_buf; break; case 3: V = d_extra3_buf; break; } argon2_options_t argon2_options = esalt_bufs[DIGESTS_OFFSET_HOST_BID].argon2_options; #ifdef IS_APPLE // it doesn't work on Apple, so we won't set it up #else #ifdef ARGON2_PARALLELISM argon2_options.parallelism = ARGON2_PARALLELISM; #endif #endif GLOBAL_AS argon2_block_t *argon2_block = get_argon2_block (&argon2_options, V, bd4); argon2_pos_t pos; pos.pass = (LOOP_POS / ARGON2_SYNC_POINTS); pos.slice = (LOOP_POS % ARGON2_SYNC_POINTS); for (u32 i = 0; i < LOOP_CNT; i++) { for (pos.lane = lid; pos.lane < argon2_options.parallelism; pos.lane += lsz) { argon2_fill_segment (argon2_block, &argon2_options, &pos, shuffle_buf, argon2_thread, argon2_lsz); } SYNC_THREADS (); pos.slice++; if (pos.slice == ARGON2_SYNC_POINTS) { pos.slice = 0; pos.pass++; } } } KERNEL_FQ KERNEL_FA void m34300_comp (KERN_ATTR_TMPS_ESALT (argon2_tmp_t, merged_options_t)) { const u64 gid = get_global_id (0); if (gid >= GID_CNT) return; const u32 gd4 = gid / 4; const u32 gm4 = gid % 4; GLOBAL_AS void *V; switch (gm4) { case 0: V = d_extra0_buf; break; case 1: V = d_extra1_buf; break; case 2: V = d_extra2_buf; break; case 3: V = d_extra3_buf; break; } const argon2_options_t argon2_options = esalt_bufs[DIGESTS_OFFSET_HOST].argon2_options; GLOBAL_AS argon2_block_t *argon2_block = get_argon2_block (&argon2_options, V, gd4); GLOBAL_AS const keepass4_t *keepass4 = &esalt_bufs[DIGESTS_OFFSET_HOST].keepass4; u32 out[8]; argon2_final (argon2_block, &argon2_options, out); sha512_ctx_t ctx; sha512_init (&ctx); ctx.w0[0] = hc_swap32_S (keepass4->masterseed[0]); ctx.w0[1] = hc_swap32_S (keepass4->masterseed[1]); ctx.w0[2] = hc_swap32_S (keepass4->masterseed[2]); ctx.w0[3] = hc_swap32_S (keepass4->masterseed[3]); ctx.w1[0] = hc_swap32_S (keepass4->masterseed[4]); ctx.w1[1] = hc_swap32_S (keepass4->masterseed[5]); ctx.w1[2] = hc_swap32_S (keepass4->masterseed[6]); ctx.w1[3] = hc_swap32_S (keepass4->masterseed[7]); ctx.w2[0] = hc_swap32_S (out[0]); ctx.w2[1] = hc_swap32_S (out[1]); ctx.w2[2] = hc_swap32_S (out[2]); ctx.w2[3] = hc_swap32_S (out[3]); ctx.w3[0] = hc_swap32_S (out[4]); ctx.w3[1] = hc_swap32_S (out[5]); ctx.w3[2] = hc_swap32_S (out[6]); ctx.w3[3] = hc_swap32_S (out[7]); ctx.w4[0] = 0x01000000; ctx.len = 32 + 32 + 1; sha512_final (&ctx); sha512_ctx_t ctx2; sha512_init (&ctx2); ctx2.w0[0] = 0xffffffff; ctx2.w0[1] = 0xffffffff; ctx2.w0[2] = h32_from_64_S (ctx.h[0]); ctx2.w0[3] = l32_from_64_S (ctx.h[0]); ctx2.w1[0] = h32_from_64_S (ctx.h[1]); ctx2.w1[1] = l32_from_64_S (ctx.h[1]); ctx2.w1[2] = h32_from_64_S (ctx.h[2]); ctx2.w1[3] = l32_from_64_S (ctx.h[2]); ctx2.w2[0] = h32_from_64_S (ctx.h[3]); ctx2.w2[1] = l32_from_64_S (ctx.h[3]); ctx2.w2[2] = h32_from_64_S (ctx.h[4]); ctx2.w2[3] = l32_from_64_S (ctx.h[4]); ctx2.w3[0] = h32_from_64_S (ctx.h[5]); ctx2.w3[1] = l32_from_64_S (ctx.h[5]); ctx2.w3[2] = h32_from_64_S (ctx.h[6]); ctx2.w3[3] = l32_from_64_S (ctx.h[6]); ctx2.w4[0] = h32_from_64_S (ctx.h[7]); ctx2.w4[1] = l32_from_64_S (ctx.h[7]); ctx2.len = 8 + 64; sha512_final (&ctx2); u32 outu32[16]; outu32[ 0] = h32_from_64_S (ctx2.h[0]); outu32[ 1] = l32_from_64_S (ctx2.h[0]); outu32[ 2] = h32_from_64_S (ctx2.h[1]); outu32[ 3] = l32_from_64_S (ctx2.h[1]); outu32[ 4] = h32_from_64_S (ctx2.h[2]); outu32[ 5] = l32_from_64_S (ctx2.h[2]); outu32[ 6] = h32_from_64_S (ctx2.h[3]); outu32[ 7] = l32_from_64_S (ctx2.h[3]); outu32[ 8] = h32_from_64_S (ctx2.h[4]); outu32[ 9] = l32_from_64_S (ctx2.h[4]); outu32[10] = h32_from_64_S (ctx2.h[5]); outu32[11] = l32_from_64_S (ctx2.h[5]); outu32[12] = h32_from_64_S (ctx2.h[6]); outu32[13] = l32_from_64_S (ctx2.h[6]); outu32[14] = h32_from_64_S (ctx2.h[7]); outu32[15] = l32_from_64_S (ctx2.h[7]); sha256_hmac_ctx_t ctx3; sha256_hmac_init (&ctx3, outu32, 64); sha256_hmac_update_global_swap (&ctx3, keepass4->header, 253); sha256_hmac_final (&ctx3); const u32 r0 = hc_swap32_S (ctx3.opad.h[0]); const u32 r1 = hc_swap32_S (ctx3.opad.h[1]); const u32 r2 = hc_swap32_S (ctx3.opad.h[2]); const u32 r3 = hc_swap32_S (ctx3.opad.h[3]); #define il_pos 0 #include COMPARE_M }