From 8ef8dfdc7299ac1c30869ec6ad4d11276f0cf5db Mon Sep 17 00:00:00 2001 From: bg Date: Thu, 22 Jan 2009 11:59:53 +0000 Subject: [PATCH] faster aes decryption --- aes_dec-asm_faster.S | 459 ++++++++++++++++++++++++++++++++++++++ aes_dec.c | 2 +- avr-makefile.inc | 3 +- host/gf256_table_gen.c | 8 +- host/gf256mul.c | 17 +- mkfiles/aes_asm_faster.mk | 14 ++ 6 files changed, 491 insertions(+), 12 deletions(-) create mode 100644 aes_dec-asm_faster.S create mode 100644 mkfiles/aes_asm_faster.mk diff --git a/aes_dec-asm_faster.S b/aes_dec-asm_faster.S new file mode 100644 index 0000000..08d3b97 --- /dev/null +++ b/aes_dec-asm_faster.S @@ -0,0 +1,459 @@ +/* aes_dec-asm.S */ +/* + This file is part of the Crypto-avr-lib/microcrypt-lib. + Copyright (C) 2008, 2009 Daniel Otte (daniel.otte@rub.de) + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . +*/ +/** + * \file aes_dec-asm.S + * \email daniel.otte@rub.de + * \author Daniel Otte + * \date 2009-01-10 + * \license GPLv3 or later + * + */ + +#include "avr-asm-macros.S" +A = 28 +B = 29 +P = 0 +xREDUCER = 25 + +.global aes256_dec +aes256_dec: + ldi r20, 14 + rjmp aes_decrypt_core + +.global aes192_dec +aes192_dec: + ldi r20, 12 + rjmp aes_decrypt_core + +.global aes128_dec +aes128_dec: + ldi r20, 10 + + +/* + void aes_decrypt_core(aes_cipher_state_t* state, const aes_genctx_t* ks, uint8_t rounds) +*/ +T0= 2 +T1= 3 +T2= 4 +T3= 5 +T4 = 6 +T5 = 7 +ST00 = 8 +ST01 = 9 +ST02 = 10 +ST03 = 11 +ST10 = 12 +ST11 = 13 +ST12 = 14 +ST13 = 15 +ST20 = 16 +ST21 = 17 +ST22 = 18 +ST23 = 19 +ST30 = 20 +ST31 = 21 +ST32 = 22 +ST33 = 23 +CTR = 24 +/* + * param state: r24:r25 + * param ks: r22:r23 + * param rounds: r20 + */ +.global aes_decrypt_core +aes_decrypt_core: + push_range 2, 17 + push r28 + push r29 + push r24 + push r25 + movw r26, r22 + movw r30, r24 + mov CTR, r20 + inc r20 + swap r20 /* r20*16 */ + add r26, r20 + adc r27, r1 + clt +; ldi CTR, 2 + .irp param, ST00, ST01, ST02, ST03, ST10, ST11, ST12, ST13, ST20, ST21, ST22, ST23, ST30, ST31, ST32, ST33 + ld \param, Z+ + .endr + + ldi xREDUCER, 0x1b /* load reducer */ + ldi r31, hi8(aes_invsbox) + + + .irp param, ST33, ST32, ST31, ST30, ST23, ST22, ST21, ST20, ST13, ST12, ST11, ST10, ST03, ST02, ST01, ST00 + ld r0, -X + eor \param, r0 + .endr +1: + dec CTR + brne 2f + set +2: + /* substitute and invShift */ + .irp param, ST00, ST10, ST20, ST30 + mov r30, \param + lpm \param, Z + .endr + mov r30, ST31 + lpm T0, Z + mov r30, ST21 + lpm ST31, Z + mov r30, ST11 + lpm ST21, Z + mov r30, ST01 + lpm ST11, Z + mov ST01, T0 + + mov r30, ST32 + lpm T0, Z + mov r30, ST22 + lpm T1,Z + mov r30, ST12 + lpm ST32, Z + mov r30, ST02 + lpm ST22, Z + mov ST12, T0 + mov ST02, T1 + + mov r30, ST03 + lpm T0, Z + mov r30, ST13 + lpm ST03, Z + mov r30, ST23 + lpm ST13, Z + mov r30, ST33 + lpm ST23, Z + mov ST33, T0 + + /* key addition */ + .irp param, ST33, ST32, ST31, ST30, ST23, ST22, ST21, ST20, ST13, ST12, ST11, ST10, ST03, ST02, ST01, ST00 + ld r0, -X + eor \param, r0 + .endr + brtc 2f +exit: + pop r31 + pop r30 + st Z+, ST00 + st Z+, ST01 + st Z+, ST02 + st Z+, ST03 + st Z+, ST10 + st Z+, ST11 + st Z+, ST12 + st Z+, ST13 + st Z+, ST20 + st Z+, ST21 + st Z+, ST22 + st Z+, ST23 + st Z+, ST30 + st Z+, ST31 + st Z+, ST32 + st Z+, ST33 + pop r29 + pop r28 + pop_range 2, 17 + ret +2: + /* inv column (row) mixing*/ + /* invMixCol (Row) 1 */ + /* preparing */ + ldi r31, hi8(lut_gf256mul_0x09) + mov T0, ST03 + eor T0, ST02 ; T0 = t + mov T1, ST00 + eor T1, ST01 ; T1 = u + mov r30, T0 + eor r30, T1 + lpm T2, Z ; T2 = v' + + ldi r31, hi8(lut_gf256mul_0x04) + mov r30, ST02 + eor r30, ST00 + lpm T3, Z + eor T3, T2; T3 = w + + mov r30, ST03 + eor r30, ST01 + lpm P, Z ; T2 = v + eor T2, P + + /* now the big move */ + mov T4, ST00 + eor T4, ST03 + lsl T4 + brcc 3f + eor T4, xREDUCER +3: eor T4, T2 + eor ST03, T4 + + mov T4, ST02 + eor T4, ST01 + lsl T4 + brcc 3f + eor T4, xREDUCER +3: eor T4, T2 + eor ST01, T4 + + lsl T0 + brcc 3f + eor T0, xREDUCER +3: eor T0, T3 + eor ST02, T0 + + lsl T1 + brcc 3f + eor T1, xREDUCER +3: eor T1, T3 + eor ST00, T1 + + /* invMixCol (Row) 2 */ + /* preparing */ + ldi r31, hi8(lut_gf256mul_0x09) + mov T0, ST13 + eor T0, ST12 ; T0 = t + mov T1, ST10 + eor T1, ST11 ; T1 = u + mov r30, T0 + eor r30, T1 + + lpm T2, Z ; T2 = v' + + ldi r31, hi8(lut_gf256mul_0x04) + mov r30, ST12 + eor r30, ST10 + lpm T3, Z + eor T3, T2; T3 = w + + mov r30, ST13 + eor r30, ST11 + lpm P, Z + eor T2, P ; T2 = v + + /* now the big move */ + mov T4, ST10 + eor T4, ST13 + lsl T4 + brcc 3f + eor T4, xREDUCER +3: eor T4, T2 + eor ST13, T4 + + mov T4, ST12 + eor T4, ST11 + lsl T4 + brcc 3f + eor T4, xREDUCER +3: eor T4, T2 + eor ST11, T4 + + lsl T0 + brcc 3f + eor T0, xREDUCER +3: eor T0, T3 + eor ST12, T0 + + lsl T1 + brcc 3f + eor T1, xREDUCER +3: eor T1, T3 + eor ST10, T1 + + /* invMixCol (Row) 2 */ + /* preparing */ + ldi r31, hi8(lut_gf256mul_0x09) + mov T0, ST23 + eor T0, ST22 ; T0 = t + mov T1, ST20 + eor T1, ST21 ; T1 = u + mov r30, T0 + eor r30, T1 + + lpm T2, Z ; T2 = v' + + ldi r31, hi8(lut_gf256mul_0x04) + mov r30, ST22 + eor r30, ST20 + lpm T3, Z + eor T3, T2; T3 = w + + mov r30, ST23 + eor r30, ST21 + lpm P, Z + eor T2, P ; T2 = v + + /* now the big move */ + mov T4, ST20 + eor T4, ST23 + lsl T4 + brcc 3f + eor T4, xREDUCER +3: eor T4, T2 + eor ST23, T4 + + mov T4, ST22 + eor T4, ST21 + lsl T4 + brcc 3f + eor T4, xREDUCER +3: eor T4, T2 + eor ST21, T4 + + lsl T0 + brcc 3f + eor T0, xREDUCER +3: eor T0, T3 + eor ST22, T0 + + lsl T1 + brcc 3f + eor T1, xREDUCER +3: eor T1, T3 + eor ST20, T1 + + /* invMixCol (Row) 3 */ + /* preparing */ + ldi r31, hi8(lut_gf256mul_0x09) + mov T0, ST33 + eor T0, ST32 ; T0 = t + mov T1, ST30 + eor T1, ST31 ; T1 = u + mov r30, T0 + eor r30, T1 + + lpm T2, Z ; T2 = v' + + ldi r31, hi8(lut_gf256mul_0x04) + mov r30, ST32 + eor r30, ST30 + lpm T3, Z + eor T3, T2; T3 = w + + mov r30, ST33 + eor r30, ST31 + lpm P, Z + eor T2, P ; T2 = v + + /* now the big move */ + mov T4, ST30 + eor T4, ST33 + lsl T4 + brcc 3f + eor T4, xREDUCER +3: eor T4, T2 + eor ST33, T4 + + mov T4, ST32 + eor T4, ST31 + lsl T4 + brcc 3f + eor T4, xREDUCER +3: eor T4, T2 + eor ST31, T4 + + lsl T0 + brcc 3f + eor T0, xREDUCER +3: eor T0, T3 + eor ST32, T0 + + lsl T1 + brcc 3f + eor T1, xREDUCER +3: eor T1, T3 + eor ST30, T1 + + ldi r31, hi8(aes_invsbox) + rjmp 1b + +.balign 256 + +lut_gf256mul_0x09: +.byte 0x00, 0x09, 0x12, 0x1B, 0x24, 0x2D, 0x36, 0x3F +.byte 0x48, 0x41, 0x5A, 0x53, 0x6C, 0x65, 0x7E, 0x77 +.byte 0x90, 0x99, 0x82, 0x8B, 0xB4, 0xBD, 0xA6, 0xAF +.byte 0xD8, 0xD1, 0xCA, 0xC3, 0xFC, 0xF5, 0xEE, 0xE7 +.byte 0x3B, 0x32, 0x29, 0x20, 0x1F, 0x16, 0x0D, 0x04 +.byte 0x73, 0x7A, 0x61, 0x68, 0x57, 0x5E, 0x45, 0x4C +.byte 0xAB, 0xA2, 0xB9, 0xB0, 0x8F, 0x86, 0x9D, 0x94 +.byte 0xE3, 0xEA, 0xF1, 0xF8, 0xC7, 0xCE, 0xD5, 0xDC +.byte 0x76, 0x7F, 0x64, 0x6D, 0x52, 0x5B, 0x40, 0x49 +.byte 0x3E, 0x37, 0x2C, 0x25, 0x1A, 0x13, 0x08, 0x01 +.byte 0xE6, 0xEF, 0xF4, 0xFD, 0xC2, 0xCB, 0xD0, 0xD9 +.byte 0xAE, 0xA7, 0xBC, 0xB5, 0x8A, 0x83, 0x98, 0x91 +.byte 0x4D, 0x44, 0x5F, 0x56, 0x69, 0x60, 0x7B, 0x72 +.byte 0x05, 0x0C, 0x17, 0x1E, 0x21, 0x28, 0x33, 0x3A +.byte 0xDD, 0xD4, 0xCF, 0xC6, 0xF9, 0xF0, 0xEB, 0xE2 +.byte 0x95, 0x9C, 0x87, 0x8E, 0xB1, 0xB8, 0xA3, 0xAA +.byte 0xEC, 0xE5, 0xFE, 0xF7, 0xC8, 0xC1, 0xDA, 0xD3 +.byte 0xA4, 0xAD, 0xB6, 0xBF, 0x80, 0x89, 0x92, 0x9B +.byte 0x7C, 0x75, 0x6E, 0x67, 0x58, 0x51, 0x4A, 0x43 +.byte 0x34, 0x3D, 0x26, 0x2F, 0x10, 0x19, 0x02, 0x0B +.byte 0xD7, 0xDE, 0xC5, 0xCC, 0xF3, 0xFA, 0xE1, 0xE8 +.byte 0x9F, 0x96, 0x8D, 0x84, 0xBB, 0xB2, 0xA9, 0xA0 +.byte 0x47, 0x4E, 0x55, 0x5C, 0x63, 0x6A, 0x71, 0x78 +.byte 0x0F, 0x06, 0x1D, 0x14, 0x2B, 0x22, 0x39, 0x30 +.byte 0x9A, 0x93, 0x88, 0x81, 0xBE, 0xB7, 0xAC, 0xA5 +.byte 0xD2, 0xDB, 0xC0, 0xC9, 0xF6, 0xFF, 0xE4, 0xED +.byte 0x0A, 0x03, 0x18, 0x11, 0x2E, 0x27, 0x3C, 0x35 +.byte 0x42, 0x4B, 0x50, 0x59, 0x66, 0x6F, 0x74, 0x7D +.byte 0xA1, 0xA8, 0xB3, 0xBA, 0x85, 0x8C, 0x97, 0x9E +.byte 0xE9, 0xE0, 0xFB, 0xF2, 0xCD, 0xC4, 0xDF, 0xD6 +.byte 0x31, 0x38, 0x23, 0x2A, 0x15, 0x1C, 0x07, 0x0E +.byte 0x79, 0x70, 0x6B, 0x62, 0x5D, 0x54, 0x4F, 0x46 + +lut_gf256mul_0x04: +.byte 0x00, 0x04, 0x08, 0x0C, 0x10, 0x14, 0x18, 0x1C +.byte 0x20, 0x24, 0x28, 0x2C, 0x30, 0x34, 0x38, 0x3C +.byte 0x40, 0x44, 0x48, 0x4C, 0x50, 0x54, 0x58, 0x5C +.byte 0x60, 0x64, 0x68, 0x6C, 0x70, 0x74, 0x78, 0x7C +.byte 0x80, 0x84, 0x88, 0x8C, 0x90, 0x94, 0x98, 0x9C +.byte 0xA0, 0xA4, 0xA8, 0xAC, 0xB0, 0xB4, 0xB8, 0xBC +.byte 0xC0, 0xC4, 0xC8, 0xCC, 0xD0, 0xD4, 0xD8, 0xDC +.byte 0xE0, 0xE4, 0xE8, 0xEC, 0xF0, 0xF4, 0xF8, 0xFC +.byte 0x1B, 0x1F, 0x13, 0x17, 0x0B, 0x0F, 0x03, 0x07 +.byte 0x3B, 0x3F, 0x33, 0x37, 0x2B, 0x2F, 0x23, 0x27 +.byte 0x5B, 0x5F, 0x53, 0x57, 0x4B, 0x4F, 0x43, 0x47 +.byte 0x7B, 0x7F, 0x73, 0x77, 0x6B, 0x6F, 0x63, 0x67 +.byte 0x9B, 0x9F, 0x93, 0x97, 0x8B, 0x8F, 0x83, 0x87 +.byte 0xBB, 0xBF, 0xB3, 0xB7, 0xAB, 0xAF, 0xA3, 0xA7 +.byte 0xDB, 0xDF, 0xD3, 0xD7, 0xCB, 0xCF, 0xC3, 0xC7 +.byte 0xFB, 0xFF, 0xF3, 0xF7, 0xEB, 0xEF, 0xE3, 0xE7 +.byte 0x36, 0x32, 0x3E, 0x3A, 0x26, 0x22, 0x2E, 0x2A +.byte 0x16, 0x12, 0x1E, 0x1A, 0x06, 0x02, 0x0E, 0x0A +.byte 0x76, 0x72, 0x7E, 0x7A, 0x66, 0x62, 0x6E, 0x6A +.byte 0x56, 0x52, 0x5E, 0x5A, 0x46, 0x42, 0x4E, 0x4A +.byte 0xB6, 0xB2, 0xBE, 0xBA, 0xA6, 0xA2, 0xAE, 0xAA +.byte 0x96, 0x92, 0x9E, 0x9A, 0x86, 0x82, 0x8E, 0x8A +.byte 0xF6, 0xF2, 0xFE, 0xFA, 0xE6, 0xE2, 0xEE, 0xEA +.byte 0xD6, 0xD2, 0xDE, 0xDA, 0xC6, 0xC2, 0xCE, 0xCA +.byte 0x2D, 0x29, 0x25, 0x21, 0x3D, 0x39, 0x35, 0x31 +.byte 0x0D, 0x09, 0x05, 0x01, 0x1D, 0x19, 0x15, 0x11 +.byte 0x6D, 0x69, 0x65, 0x61, 0x7D, 0x79, 0x75, 0x71 +.byte 0x4D, 0x49, 0x45, 0x41, 0x5D, 0x59, 0x55, 0x51 +.byte 0xAD, 0xA9, 0xA5, 0xA1, 0xBD, 0xB9, 0xB5, 0xB1 +.byte 0x8D, 0x89, 0x85, 0x81, 0x9D, 0x99, 0x95, 0x91 +.byte 0xED, 0xE9, 0xE5, 0xE1, 0xFD, 0xF9, 0xF5, 0xF1 +.byte 0xCD, 0xC9, 0xC5, 0xC1, 0xDD, 0xD9, 0xD5, 0xD1 + diff --git a/aes_dec.c b/aes_dec.c index c2ee5f2..48d6ab7 100644 --- a/aes_dec.c +++ b/aes_dec.c @@ -59,7 +59,7 @@ void aes_dec_round(aes_cipher_state_t* state, const aes_roundkey_t* k){ t = tmp[4*i+3] ^ tmp[4*i+2]; u = tmp[4*i+1] ^ tmp[4*i+0]; v = t ^ u; - v ^= gf256mul(0x08, v, 0x1b); + v = gf256mul(0x09, v, 0x1b); w = v ^ gf256mul(0x04, tmp[4*i+2] ^ tmp[4*i+0], 0x1b); v = v ^ gf256mul(0x04, tmp[4*i+3] ^ tmp[4*i+1], 0x1b); state->s[4*i+3] = tmp[4*i+3] ^ v ^ gf256mul(0x02, tmp[4*i+0] ^ tmp[4*i+3], 0x1b); diff --git a/avr-makefile.inc b/avr-makefile.inc index 741ede5..e65e585 100644 --- a/avr-makefile.inc +++ b/avr-makefile.inc @@ -1,8 +1,9 @@ MCU_TARGET = atmega644 OPTIMIZE = -Os +PROGRAMMER = stk500 # avr911 DEFS = -D$(call uc, $(MCU_TARGET)) -FLASHCMD = avrdude -p $(MCU_TARGET) -P /dev/ttyUSB0 -c avr911 -U flash:w:# no space at the end +FLASHCMD = avrdude -p $(MCU_TARGET) -P /dev/ttyUSB0 -c $(PROGRAMMER) -U flash:w:# no space at the end #FLASHCMD = avrdude -p $(MCU_TARGET) -c usbasp -U flash:w:# no space at the end DEP_DIR = deps/ BIN_DIR = bin/ diff --git a/host/gf256_table_gen.c b/host/gf256_table_gen.c index 0d55db4..0d002f9 100644 --- a/host/gf256_table_gen.c +++ b/host/gf256_table_gen.c @@ -58,6 +58,7 @@ void print_help(void){ " -u max upper limit value for multiplication table (default: 255)\n" " -f formatstring string for formating the output of a singel multiplicxation\n" " -o file outputfile\n" +); } int main(int argc, char** argv){ @@ -99,7 +100,7 @@ int main(int argc, char** argv){ char** eptr; FILE* of = stdout; while(c!=-1){ - c = getopt_long(argc, argv,"cCbBdhoaApPr:l:u:f:", + c = getopt_long(argc, argv,"cCbBdh8aApPr:l:u:f:", long_options, &option_index); switch(c){ case 'c': print_header=1; break; @@ -135,13 +136,16 @@ int main(int argc, char** argv){ } break; case 'f': fmt = optarg; break; + case -1: break; case '?': default: + printf ("unknown option '%c' (%2.2X) !\n\t", c, c); print_help(); + return 0; break; } } - if(!reducer = ul_reducer&0xff){ + if(!(reducer = ul_reducer&0xff)){ fprintf(stderr, "You must specify a reductionpolynomial!\n", argv[i]); return -1; } diff --git a/host/gf256mul.c b/host/gf256mul.c index 3da67cb..d10b69c 100644 --- a/host/gf256mul.c +++ b/host/gf256mul.c @@ -28,15 +28,16 @@ #include "gf256mul.h" uint8_t gf256mul(uint8_t a, uint8_t b, uint8_t reducer){ - uint8_t t,ret=0; - while(a){ - if(a&1) - ret ^= b; + uint8_t i; + uint8_t p=0,t; + for(i=0; i<8; ++i){ + if(b&1) + p ^= a; t=a&0x80; - b<<=1; + a<<=1; if(t) - b^=reducer; - a>>=1; + a ^= reducer; + b>>=1; } - return ret; + return p; } diff --git a/mkfiles/aes_asm_faster.mk b/mkfiles/aes_asm_faster.mk new file mode 100644 index 0000000..bd478f0 --- /dev/null +++ b/mkfiles/aes_asm_faster.mk @@ -0,0 +1,14 @@ +# Makefile for AES +ALGO_NAME := AES_FASTER + +# comment out the following line for removement of AES from the build process +BLOCK_CIPHERS += $(ALGO_NAME) + + +$(ALGO_NAME)_OBJ := aes_enc-asm.o aes_dec-asm_faster.o aes_sboxes-asm.o \ + aes_keyschedule-asm.o +$(ALGO_NAME)_TEST_BIN := main-aes-test.o debug.o uart.o serial-tools.o \ + nessie_bc_test.o nessie_common.o cli.o performance_test.o +$(ALGO_NAME)_NESSIE_TEST := test nessie +$(ALGO_NAME)_PERFORMANCE_TEST := performance + -- 2.39.5