/** * Author......: See docs/credits.txt * License.....: MIT */ //#define NEW_SIMD_CODE #ifdef KERNEL_STATIC #include "inc_vendor.h" #include "inc_types.h" #include "inc_platform.cl" #include "inc_common.cl" #include "inc_hash_md5.cl" #endif #define COMPARE_S "inc_comp_single.cl" #define COMPARE_M "inc_comp_multi.cl" typedef struct md5crypt_tmp { u32 digest_buf[4]; } md5crypt_tmp_t; #define md5crypt_magic 0x00243124u DECLSPEC void memcat16 (u32 *block0, u32 *block1, u32 *block2, u32 *block3, const u32 offset, const u32 *append) { u32 tmp0; u32 tmp1; u32 tmp2; u32 tmp3; u32 tmp4; #if (defined IS_AMD || defined IS_HIP) || defined IS_GENERIC u32 in0 = append[0]; u32 in1 = append[1]; u32 in2 = append[2]; u32 in3 = append[3]; tmp0 = hc_bytealign ( 0, in0, offset); tmp1 = hc_bytealign (in0, in1, offset); tmp2 = hc_bytealign (in1, in2, offset); tmp3 = hc_bytealign (in2, in3, offset); tmp4 = hc_bytealign (in3, 0, offset); #endif #ifdef IS_NV const int offset_mod_4 = offset & 3; const int offset_minus_4 = 4 - offset_mod_4; const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff; u32 in0 = append[0]; u32 in1 = append[1]; u32 in2 = append[2]; u32 in3 = append[3]; tmp0 = hc_byte_perm ( 0, in0, selector); tmp1 = hc_byte_perm (in0, in1, selector); tmp2 = hc_byte_perm (in1, in2, selector); tmp3 = hc_byte_perm (in2, in3, selector); tmp4 = hc_byte_perm (in3, 0, selector); #endif const u32 div = offset / 4; switch (div) { case 0: block0[0] |= tmp0; block0[1] = tmp1; block0[2] = tmp2; block0[3] = tmp3; block1[0] = tmp4; break; case 1: block0[1] |= tmp0; block0[2] = tmp1; block0[3] = tmp2; block1[0] = tmp3; block1[1] = tmp4; break; case 2: block0[2] |= tmp0; block0[3] = tmp1; block1[0] = tmp2; block1[1] = tmp3; block1[2] = tmp4; break; case 3: block0[3] |= tmp0; block1[0] = tmp1; block1[1] = tmp2; block1[2] = tmp3; block1[3] = tmp4; break; case 4: block1[0] |= tmp0; block1[1] = tmp1; block1[2] = tmp2; block1[3] = tmp3; block2[0] = tmp4; break; case 5: block1[1] |= tmp0; block1[2] = tmp1; block1[3] = tmp2; block2[0] = tmp3; block2[1] = tmp4; break; case 6: block1[2] |= tmp0; block1[3] = tmp1; block2[0] = tmp2; block2[1] = tmp3; block2[2] = tmp4; break; case 7: block1[3] |= tmp0; block2[0] = tmp1; block2[1] = tmp2; block2[2] = tmp3; block2[3] = tmp4; break; case 8: block2[0] |= tmp0; block2[1] = tmp1; block2[2] = tmp2; block2[3] = tmp3; block3[0] = tmp4; break; case 9: block2[1] |= tmp0; block2[2] = tmp1; block2[3] = tmp2; block3[0] = tmp3; block3[1] = tmp4; break; } } DECLSPEC void memcat16_x80 (u32 *block0, u32 *block1, u32 *block2, u32 *block3, const u32 offset, const u32 *append) { u32 tmp0; u32 tmp1; u32 tmp2; u32 tmp3; u32 tmp4; #if (defined IS_AMD || defined IS_HIP) || defined IS_GENERIC u32 in0 = append[0]; u32 in1 = append[1]; u32 in2 = append[2]; u32 in3 = append[3]; u32 in4 = 0x80; tmp0 = hc_bytealign ( 0, in0, offset); tmp1 = hc_bytealign (in0, in1, offset); tmp2 = hc_bytealign (in1, in2, offset); tmp3 = hc_bytealign (in2, in3, offset); tmp4 = hc_bytealign (in3, in4, offset); #endif #ifdef IS_NV const int offset_mod_4 = offset & 3; const int offset_minus_4 = 4 - offset_mod_4; const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff; u32 in0 = append[0]; u32 in1 = append[1]; u32 in2 = append[2]; u32 in3 = append[3]; u32 in4 = 0x80; tmp0 = hc_byte_perm ( 0, in0, selector); tmp1 = hc_byte_perm (in0, in1, selector); tmp2 = hc_byte_perm (in1, in2, selector); tmp3 = hc_byte_perm (in2, in3, selector); tmp4 = hc_byte_perm (in3, in4, selector); #endif const u32 div = offset / 4; switch (div) { case 0: block0[0] |= tmp0; block0[1] = tmp1; block0[2] = tmp2; block0[3] = tmp3; block1[0] = tmp4; break; case 1: block0[1] |= tmp0; block0[2] = tmp1; block0[3] = tmp2; block1[0] = tmp3; block1[1] = tmp4; break; case 2: block0[2] |= tmp0; block0[3] = tmp1; block1[0] = tmp2; block1[1] = tmp3; block1[2] = tmp4; break; case 3: block0[3] |= tmp0; block1[0] = tmp1; block1[1] = tmp2; block1[2] = tmp3; block1[3] = tmp4; break; case 4: block1[0] |= tmp0; block1[1] = tmp1; block1[2] = tmp2; block1[3] = tmp3; block2[0] = tmp4; break; case 5: block1[1] |= tmp0; block1[2] = tmp1; block1[3] = tmp2; block2[0] = tmp3; block2[1] = tmp4; break; case 6: block1[2] |= tmp0; block1[3] = tmp1; block2[0] = tmp2; block2[1] = tmp3; block2[2] = tmp4; break; case 7: block1[3] |= tmp0; block2[0] = tmp1; block2[1] = tmp2; block2[2] = tmp3; block2[3] = tmp4; break; case 8: block2[0] |= tmp0; block2[1] = tmp1; block2[2] = tmp2; block2[3] = tmp3; block3[0] = tmp4; break; case 9: block2[1] |= tmp0; block2[2] = tmp1; block2[3] = tmp2; block3[0] = tmp3; block3[1] = tmp4; break; } } DECLSPEC void memcat8 (u32 *block0, u32 *block1, u32 *block2, u32 *block3, const u32 offset, const u32 *append) { u32 tmp0; u32 tmp1; u32 tmp2; #if (defined IS_AMD || defined IS_HIP) || defined IS_GENERIC u32 in0 = append[0]; u32 in1 = append[1]; tmp0 = hc_bytealign ( 0, in0, offset); tmp1 = hc_bytealign (in0, in1, offset); tmp2 = hc_bytealign (in1, 0, offset); #endif #ifdef IS_NV const int offset_mod_4 = offset & 3; const int offset_minus_4 = 4 - offset_mod_4; const int selector = (0x76543210 >> (offset_minus_4 * 4)) & 0xffff; u32 in0 = append[0]; u32 in1 = append[1]; tmp0 = hc_byte_perm ( 0, in0, selector); tmp1 = hc_byte_perm (in0, in1, selector); tmp2 = hc_byte_perm (in1, 0, selector); #endif const u32 div = offset / 4; switch (div) { case 0: block0[0] |= tmp0; block0[1] = tmp1; block0[2] = tmp2; break; case 1: block0[1] |= tmp0; block0[2] = tmp1; block0[3] = tmp2; break; case 2: block0[2] |= tmp0; block0[3] = tmp1; block1[0] = tmp2; break; case 3: block0[3] |= tmp0; block1[0] = tmp1; block1[1] = tmp2; break; case 4: block1[0] |= tmp0; block1[1] = tmp1; block1[2] = tmp2; break; case 5: block1[1] |= tmp0; block1[2] = tmp1; block1[3] = tmp2; break; case 6: block1[2] |= tmp0; block1[3] = tmp1; block2[0] = tmp2; break; case 7: block1[3] |= tmp0; block2[0] = tmp1; block2[1] = tmp2; break; case 8: block2[0] |= tmp0; block2[1] = tmp1; block2[2] = tmp2; break; case 9: block2[1] |= tmp0; block2[2] = tmp1; block2[3] = tmp2; break; case 10: block2[2] |= tmp0; block2[3] = tmp1; block3[0] = tmp2; break; case 11: block2[3] |= tmp0; block3[0] = tmp1; block3[1] = tmp2; break; } } DECLSPEC void append_sign (u32 *block0, u32 *block1, const u32 block_len) { switch (block_len) { case 0: block0[0] = md5crypt_magic; break; case 1: block0[0] = block0[0] | md5crypt_magic << 8u; block0[1] = md5crypt_magic >> 24u; break; case 2: block0[0] = block0[0] | md5crypt_magic << 16u; block0[1] = md5crypt_magic >> 16u; break; case 3: block0[0] = block0[0] | md5crypt_magic << 24u; block0[1] = md5crypt_magic >> 8u; break; case 4: block0[1] = md5crypt_magic; break; case 5: block0[1] = block0[1] | md5crypt_magic << 8u; block0[2] = md5crypt_magic >> 24u; break; case 6: block0[1] = block0[1] | md5crypt_magic << 16u; block0[2] = md5crypt_magic >> 16u; break; case 7: block0[1] = block0[1] | md5crypt_magic << 24u; block0[2] = md5crypt_magic >> 8u; break; case 8: block0[2] = md5crypt_magic; break; case 9: block0[2] = block0[2] | md5crypt_magic << 8u; block0[3] = md5crypt_magic >> 24u; break; case 10: block0[2] = block0[2] | md5crypt_magic << 16u; block0[3] = md5crypt_magic >> 16u; break; case 11: block0[2] = block0[2] | md5crypt_magic << 24u; block0[3] = md5crypt_magic >> 8u; break; case 12: block0[3] = md5crypt_magic; break; case 13: block0[3] = block0[3] | md5crypt_magic << 8u; block1[0] = md5crypt_magic >> 24u; break; case 14: block0[3] = block0[3] | md5crypt_magic << 16u; block1[0] = md5crypt_magic >> 16u; break; case 15: block0[3] = block0[3] | md5crypt_magic << 24u; block1[0] = md5crypt_magic >> 8u; break; } } DECLSPEC void append_1st (u32 *block0, u32 *block1, u32 *block2, u32 *block3, const u32 block_len, const u32 append) { switch (block_len) { case 0: block0[0] = append; break; case 1: block0[0] = block0[0] | append << 8; break; case 2: block0[0] = block0[0] | append << 16; break; case 3: block0[0] = block0[0] | append << 24; break; case 4: block0[1] = append; break; case 5: block0[1] = block0[1] | append << 8; break; case 6: block0[1] = block0[1] | append << 16; break; case 7: block0[1] = block0[1] | append << 24; break; case 8: block0[2] = append; break; case 9: block0[2] = block0[2] | append << 8; break; case 10: block0[2] = block0[2] | append << 16; break; case 11: block0[2] = block0[2] | append << 24; break; case 12: block0[3] = append; break; case 13: block0[3] = block0[3] | append << 8; break; case 14: block0[3] = block0[3] | append << 16; break; case 15: block0[3] = block0[3] | append << 24; break; case 16: block1[0] = append; break; case 17: block1[0] = block1[0] | append << 8; break; case 18: block1[0] = block1[0] | append << 16; break; case 19: block1[0] = block1[0] | append << 24; break; case 20: block1[1] = append; break; case 21: block1[1] = block1[1] | append << 8; break; case 22: block1[1] = block1[1] | append << 16; break; case 23: block1[1] = block1[1] | append << 24; break; case 24: block1[2] = append; break; case 25: block1[2] = block1[2] | append << 8; break; case 26: block1[2] = block1[2] | append << 16; break; case 27: block1[2] = block1[2] | append << 24; break; case 28: block1[3] = append; break; case 29: block1[3] = block1[3] | append << 8; break; case 30: block1[3] = block1[3] | append << 16; break; case 31: block1[3] = block1[3] | append << 24; break; case 32: block2[0] = append; break; case 33: block2[0] = block2[0] | append << 8; break; case 34: block2[0] = block2[0] | append << 16; break; case 35: block2[0] = block2[0] | append << 24; break; case 36: block2[1] = append; break; case 37: block2[1] = block2[1] | append << 8; break; case 38: block2[1] = block2[1] | append << 16; break; case 39: block2[1] = block2[1] | append << 24; break; case 40: block2[2] = append; break; case 41: block2[2] = block2[2] | append << 8; break; case 42: block2[2] = block2[2] | append << 16; break; case 43: block2[2] = block2[2] | append << 24; break; case 44: block2[3] = append; break; case 45: block2[3] = block2[3] | append << 8; break; case 46: block2[3] = block2[3] | append << 16; break; case 47: block2[3] = block2[3] | append << 24; break; case 48: block3[0] = append; break; case 49: block3[0] = block3[0] | append << 8; break; case 50: block3[0] = block3[0] | append << 16; break; case 51: block3[0] = block3[0] | append << 24; break; case 52: block3[1] = append; break; case 53: block3[1] = block3[1] | append << 8; break; case 54: block3[1] = block3[1] | append << 16; break; case 55: block3[1] = block3[1] | append << 24; break; case 56: block3[2] = append; break; } } KERNEL_FQ void m00500_init (KERN_ATTR_TMPS (md5crypt_tmp_t)) { /** * base */ const u64 gid = get_global_id (0); if (gid >= gid_max) return; u32 w0[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]; const u32 pw_len = pws[gid].pw_len & 63; /** * salt */ u32 salt_buf[2]; salt_buf[0] = salt_bufs[SALT_POS].salt_buf[0]; salt_buf[1] = salt_bufs[SALT_POS].salt_buf[1]; const u32 salt_len = salt_bufs[SALT_POS].salt_len; /** * init */ //memcat16 (block0, block1, block2, block3, block_len, w0); //block_len += pw_len; u32 block_len = pw_len; u32 block0[4]; block0[0] = w0[0]; block0[1] = w0[1]; block0[2] = w0[2]; block0[3] = w0[3]; u32 block1[4]; block1[0] = 0; block1[1] = 0; block1[2] = 0; block1[3] = 0; u32 block2[4]; block2[0] = 0; block2[1] = 0; block2[2] = 0; block2[3] = 0; u32 block3[4]; block3[0] = 0; block3[1] = 0; block3[2] = 0; block3[3] = 0; memcat8 (block0, block1, block2, block3, block_len, salt_buf); block_len += salt_len; memcat16 (block0, block1, block2, block3, block_len, w0); block_len += pw_len; append_0x80_4x4 (block0, block1, block2, block3, block_len); block3[2] = block_len * 8; u32 digest[4]; digest[0] = MD5M_A; digest[1] = MD5M_B; digest[2] = MD5M_C; digest[3] = MD5M_D; md5_transform (block0, block1, block2, block3, digest); /* The password first, since that is what is most unknown */ /* Then our magic string */ /* Then the raw salt */ /* Then just as many characters of the MD5(pw,salt,pw) */ //memcat16 (block0, block1, block2, block3, block_len, w); //block_len += pw_len; block_len = pw_len; block0[0] = w0[0]; block0[1] = w0[1]; block0[2] = w0[2]; block0[3] = w0[3]; block1[0] = 0; block1[1] = 0; block1[2] = 0; block1[3] = 0; block2[0] = 0; block2[1] = 0; block2[2] = 0; block2[3] = 0; block3[0] = 0; block3[1] = 0; block3[2] = 0; block3[3] = 0; append_sign (block0, block1, block_len); block_len += 3; memcat8 (block0, block1, block2, block3, block_len, salt_buf); block_len += salt_len; truncate_block_4x4_le_S (digest, pw_len); memcat16 (block0, block1, block2, block3, block_len, digest); block_len += pw_len; /* Then something really weird... */ u32 append = block0[0] & 0xFF; for (u32 j = pw_len; j; j >>= 1) { if ((j & 1) == 0) { append_1st (block0, block1, block2, block3, block_len, append); } block_len++; } append_0x80_4x4 (block0, block1, block2, block3, block_len); block3[2] = block_len * 8; digest[0] = MD5M_A; digest[1] = MD5M_B; digest[2] = MD5M_C; digest[3] = MD5M_D; md5_transform (block0, block1, block2, block3, digest); tmps[gid].digest_buf[0] = digest[0]; tmps[gid].digest_buf[1] = digest[1]; tmps[gid].digest_buf[2] = digest[2]; tmps[gid].digest_buf[3] = digest[3]; } KERNEL_FQ void m00500_loop (KERN_ATTR_TMPS (md5crypt_tmp_t)) { /** * base */ const u64 gid = get_global_id (0); if (gid >= gid_max) return; u32 w0[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]; const u32 pw_len = pws[gid].pw_len & 63; u32 w0_x80[4]; w0_x80[0] = w0[0]; w0_x80[1] = w0[1]; w0_x80[2] = w0[2]; w0_x80[3] = w0[3]; append_0x80_1x4 (w0_x80, pw_len); /** * salt */ u32 salt_buf[2]; salt_buf[0] = salt_bufs[SALT_POS].salt_buf[0]; salt_buf[1] = salt_bufs[SALT_POS].salt_buf[1]; const u32 salt_len = salt_bufs[SALT_POS].salt_len; /** * digest */ u32 digest[4]; digest[0] = tmps[gid].digest_buf[0]; digest[1] = tmps[gid].digest_buf[1]; digest[2] = tmps[gid].digest_buf[2]; digest[3] = tmps[gid].digest_buf[3]; /** * loop */ /* and now, just to make sure things don't run too fast */ u32 block_len; u32 block0[4]; block0[0] = 0; block0[1] = 0; block0[2] = 0; block0[3] = 0; u32 block1[4]; block1[0] = 0; block1[1] = 0; block1[2] = 0; block1[3] = 0; u32 block2[4]; block2[0] = 0; block2[1] = 0; block2[2] = 0; block2[3] = 0; u32 block3[4]; block3[0] = 0; block3[1] = 0; block3[2] = 0; block3[3] = 0; for (u32 i = 0, j = loop_pos; i < loop_cnt; i++, j++) { block1[0] = 0; block1[1] = 0; block1[2] = 0; block1[3] = 0; block2[0] = 0; block2[1] = 0; block2[2] = 0; block2[3] = 0; block3[0] = 0; block3[1] = 0; const u32 j1 = (j & 1) ? 1 : 0; const u32 j3 = (j % 3) ? 1 : 0; const u32 j7 = (j % 7) ? 1 : 0; if (j1) { block0[0] = w0[0]; block0[1] = w0[1]; block0[2] = w0[2]; block0[3] = w0[3]; block_len = pw_len; if (j3) { memcat8 (block0, block1, block2, block3, block_len, salt_buf); block_len += salt_len; } if (j7) { memcat16 (block0, block1, block2, block3, block_len, w0); block_len += pw_len; } memcat16_x80 (block0, block1, block2, block3, block_len, digest); block_len += 16; } else { block0[0] = digest[0]; block0[1] = digest[1]; block0[2] = digest[2]; block0[3] = digest[3]; block_len = 16; if (j3 && j7) { block1[0] = salt_buf[0]; block1[1] = salt_buf[1]; block_len += salt_len; memcat16 (block0, block1, block2, block3, block_len, w0); block_len += pw_len; } else if (j3) { block1[0] = salt_buf[0]; block1[1] = salt_buf[1]; block_len += salt_len; } else if (j7) { block1[0] = w0[0]; block1[1] = w0[1]; block1[2] = w0[2]; block1[3] = w0[3]; block_len += pw_len; } memcat16 (block0, block1, block2, block3, block_len, w0_x80); block_len += pw_len; } block3[2] = block_len * 8; digest[0] = MD5M_A; digest[1] = MD5M_B; digest[2] = MD5M_C; digest[3] = MD5M_D; md5_transform (block0, block1, block2, block3, digest); } tmps[gid].digest_buf[0] = digest[0]; tmps[gid].digest_buf[1] = digest[1]; tmps[gid].digest_buf[2] = digest[2]; tmps[gid].digest_buf[3] = digest[3]; } KERNEL_FQ void m00500_comp (KERN_ATTR_TMPS (md5crypt_tmp_t)) { /** * modifier */ const u64 gid = get_global_id (0); if (gid >= gid_max) return; const u64 lid = get_local_id (0); /** * digest */ const u32 r0 = tmps[gid].digest_buf[DGST_R0]; const u32 r1 = tmps[gid].digest_buf[DGST_R1]; const u32 r2 = tmps[gid].digest_buf[DGST_R2]; const u32 r3 = tmps[gid].digest_buf[DGST_R3]; #define il_pos 0 #ifdef KERNEL_STATIC #include COMPARE_M #endif }