2 * Twofish Cipher 8-way parallel algorithm (AVX/x86_64)
4 * Copyright (C) 2012 Johannes Goetzfried
5 * <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
7 * Copyright © 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2 of the License, or
12 * (at your option) any later version.
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
19 * You should have received a copy of the GNU General Public License
20 * along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
26 #include <linux/linkage.h>
27 #include "glue_helper-asm-avx.S"
29 .file "twofish-avx-x86_64-asm_64.S"
35 .byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
39 /* structure of crypto context */
47 /**********************************************************************
49 **********************************************************************/
101 #define lookup_32bit(t0, t1, t2, t3, src, dst, interleave_op, il_reg) \
102 movzbl src ## bl, RID1d; \
103 movzbl src ## bh, RID2d; \
105 movl t0(CTX, RID1, 4), dst ## d; \
106 movl t1(CTX, RID2, 4), RID2d; \
107 movzbl src ## bl, RID1d; \
108 xorl RID2d, dst ## d; \
109 movzbl src ## bh, RID2d; \
110 interleave_op(il_reg); \
111 xorl t2(CTX, RID1, 4), dst ## d; \
112 xorl t3(CTX, RID2, 4), dst ## d;
114 #define dummy(d) /* do nothing */
116 #define shr_next(reg) \
119 #define G(gi1, gi2, x, t0, t1, t2, t3) \
120 lookup_32bit(t0, t1, t2, t3, ##gi1, RGS1, shr_next, ##gi1); \
121 lookup_32bit(t0, t1, t2, t3, ##gi2, RGS3, shr_next, ##gi2); \
123 lookup_32bit(t0, t1, t2, t3, ##gi1, RGS2, dummy, none); \
126 lookup_32bit(t0, t1, t2, t3, ##gi2, RGS1, dummy, none); \
130 #define round_head_2(a, b, x1, y1, x2, y2) \
131 vmovq b ## 1, RGI3; \
132 vpextrq $1, b ## 1, RGI4; \
134 G(RGI1, RGI2, x1, s0, s1, s2, s3); \
135 vmovq a ## 2, RGI1; \
136 vpextrq $1, a ## 2, RGI2; \
138 vpinsrq $1, RGS3, x1, x1; \
140 G(RGI3, RGI4, y1, s1, s2, s3, s0); \
141 vmovq b ## 2, RGI3; \
142 vpextrq $1, b ## 2, RGI4; \
144 vpinsrq $1, RGS3, y1, y1; \
146 G(RGI1, RGI2, x2, s0, s1, s2, s3); \
148 vpinsrq $1, RGS3, x2, x2; \
150 G(RGI3, RGI4, y2, s1, s2, s3, s0); \
152 vpinsrq $1, RGS3, y2, y2;
154 #define encround_tail(a, b, c, d, x, y, prerotate) \
162 vpslld $(32 - 1), c, c; \
166 #define decround_tail(a, b, c, d, x, y, prerotate) \
175 vpslld $(32 - 1), d, d; \
178 #define rotate_1l(x) \
180 vpsrld $(32 - 1), x, x; \
183 #define preload_rgi(c) \
187 #define encrypt_round(n, a, b, c, d, preload, prerotate) \
188 vbroadcastss (k+4*(2*(n)))(CTX), RK1; \
189 vbroadcastss (k+4*(2*(n)+1))(CTX), RK2; \
190 round_head_2(a, b, RX0, RY0, RX1, RY1); \
191 encround_tail(a ## 1, b ## 1, c ## 1, d ## 1, RX0, RY0, prerotate); \
193 encround_tail(a ## 2, b ## 2, c ## 2, d ## 2, RX1, RY1, prerotate);
195 #define decrypt_round(n, a, b, c, d, preload, prerotate) \
196 vbroadcastss (k+4*(2*(n)))(CTX), RK1; \
197 vbroadcastss (k+4*(2*(n)+1))(CTX), RK2; \
198 round_head_2(a, b, RX0, RY0, RX1, RY1); \
199 decround_tail(a ## 1, b ## 1, c ## 1, d ## 1, RX0, RY0, prerotate); \
201 decround_tail(a ## 2, b ## 2, c ## 2, d ## 2, RX1, RY1, prerotate);
203 #define encrypt_cycle(n) \
204 encrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l); \
205 encrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l);
207 #define encrypt_cycle_last(n) \
208 encrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l); \
209 encrypt_round(((2*n) + 1), RC, RD, RA, RB, dummy, dummy);
211 #define decrypt_cycle(n) \
212 decrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l); \
213 decrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l);
215 #define decrypt_cycle_last(n) \
216 decrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l); \
217 decrypt_round((2*n), RA, RB, RC, RD, dummy, dummy);
219 #define transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
220 vpunpckldq x1, x0, t0; \
221 vpunpckhdq x1, x0, t2; \
222 vpunpckldq x3, x2, t1; \
223 vpunpckhdq x3, x2, x3; \
225 vpunpcklqdq t1, t0, x0; \
226 vpunpckhqdq t1, t0, x1; \
227 vpunpcklqdq x3, t2, x2; \
228 vpunpckhqdq x3, t2, x3;
230 #define inpack_blocks(x0, x1, x2, x3, wkey, t0, t1, t2) \
231 vpxor x0, wkey, x0; \
232 vpxor x1, wkey, x1; \
233 vpxor x2, wkey, x2; \
234 vpxor x3, wkey, x3; \
236 transpose_4x4(x0, x1, x2, x3, t0, t1, t2)
238 #define outunpack_blocks(x0, x1, x2, x3, wkey, t0, t1, t2) \
239 transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
241 vpxor x0, wkey, x0; \
242 vpxor x1, wkey, x1; \
243 vpxor x2, wkey, x2; \
250 * RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: blocks
252 * RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2: encrypted blocks
261 inpack_blocks(RA1, RB1, RC1, RD1, RK1, RX0, RY0, RK2);
264 inpack_blocks(RA2, RB2, RC2, RD2, RK1, RX0, RY0, RK2);
274 encrypt_cycle_last(7);
276 vmovdqu (w+4*4)(CTX), RK1;
282 outunpack_blocks(RC1, RD1, RA1, RB1, RK1, RX0, RY0, RK2);
283 outunpack_blocks(RC2, RD2, RA2, RB2, RK1, RX0, RY0, RK2);
286 ENDPROC(__twofish_enc_blk8)
292 * RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2: encrypted blocks
294 * RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: decrypted blocks
297 vmovdqu (w+4*4)(CTX), RK1;
302 inpack_blocks(RC1, RD1, RA1, RB1, RK1, RX0, RY0, RK2);
305 inpack_blocks(RC2, RD2, RA2, RB2, RK1, RX0, RY0, RK2);
315 decrypt_cycle_last(0);
317 vmovdqu (w)(CTX), RK1;
322 outunpack_blocks(RA1, RB1, RC1, RD1, RK1, RX0, RY0, RK2);
323 outunpack_blocks(RA2, RB2, RC2, RD2, RK1, RX0, RY0, RK2);
326 ENDPROC(__twofish_dec_blk8)
328 ENTRY(twofish_ecb_enc_8way)
337 load_8way(%rdx, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
339 call __twofish_enc_blk8;
341 store_8way(%r11, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
344 ENDPROC(twofish_ecb_enc_8way)
346 ENTRY(twofish_ecb_dec_8way)
355 load_8way(%rdx, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
357 call __twofish_dec_blk8;
359 store_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
362 ENDPROC(twofish_ecb_dec_8way)
364 ENTRY(twofish_cbc_dec_8way)
376 load_8way(%rdx, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
378 call __twofish_dec_blk8;
380 store_cbc_8way(%r12, %r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
385 ENDPROC(twofish_cbc_dec_8way)
387 ENTRY(twofish_ctr_8way)
392 * %rcx: iv (little endian, 128bit)
400 load_ctr_8way(%rcx, .Lbswap128_mask, RA1, RB1, RC1, RD1, RA2, RB2, RC2,
403 call __twofish_enc_blk8;
405 store_ctr_8way(%r12, %r11, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
410 ENDPROC(twofish_ctr_8way)