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 © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.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 <asm/frame.h>
28 #include "glue_helper-asm-avx.S"
30 .file "twofish-avx-x86_64-asm_64.S"
36 .byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
37 .Lxts_gf128mul_and_shl1_mask:
38 .byte 0x87, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
42 /* structure of crypto context */
50 /**********************************************************************
52 **********************************************************************/
104 #define lookup_32bit(t0, t1, t2, t3, src, dst, interleave_op, il_reg) \
105 movzbl src ## bl, RID1d; \
106 movzbl src ## bh, RID2d; \
108 movl t0(CTX, RID1, 4), dst ## d; \
109 movl t1(CTX, RID2, 4), RID2d; \
110 movzbl src ## bl, RID1d; \
111 xorl RID2d, dst ## d; \
112 movzbl src ## bh, RID2d; \
113 interleave_op(il_reg); \
114 xorl t2(CTX, RID1, 4), dst ## d; \
115 xorl t3(CTX, RID2, 4), dst ## d;
117 #define dummy(d) /* do nothing */
119 #define shr_next(reg) \
122 #define G(gi1, gi2, x, t0, t1, t2, t3) \
123 lookup_32bit(t0, t1, t2, t3, ##gi1, RGS1, shr_next, ##gi1); \
124 lookup_32bit(t0, t1, t2, t3, ##gi2, RGS3, shr_next, ##gi2); \
126 lookup_32bit(t0, t1, t2, t3, ##gi1, RGS2, dummy, none); \
129 lookup_32bit(t0, t1, t2, t3, ##gi2, RGS1, dummy, none); \
133 #define round_head_2(a, b, x1, y1, x2, y2) \
134 vmovq b ## 1, RGI3; \
135 vpextrq $1, b ## 1, RGI4; \
137 G(RGI1, RGI2, x1, s0, s1, s2, s3); \
138 vmovq a ## 2, RGI1; \
139 vpextrq $1, a ## 2, RGI2; \
141 vpinsrq $1, RGS3, x1, x1; \
143 G(RGI3, RGI4, y1, s1, s2, s3, s0); \
144 vmovq b ## 2, RGI3; \
145 vpextrq $1, b ## 2, RGI4; \
147 vpinsrq $1, RGS3, y1, y1; \
149 G(RGI1, RGI2, x2, s0, s1, s2, s3); \
151 vpinsrq $1, RGS3, x2, x2; \
153 G(RGI3, RGI4, y2, s1, s2, s3, s0); \
155 vpinsrq $1, RGS3, y2, y2;
157 #define encround_tail(a, b, c, d, x, y, prerotate) \
165 vpslld $(32 - 1), c, c; \
169 #define decround_tail(a, b, c, d, x, y, prerotate) \
178 vpslld $(32 - 1), d, d; \
181 #define rotate_1l(x) \
183 vpsrld $(32 - 1), x, x; \
186 #define preload_rgi(c) \
190 #define encrypt_round(n, a, b, c, d, preload, prerotate) \
191 vbroadcastss (k+4*(2*(n)))(CTX), RK1; \
192 vbroadcastss (k+4*(2*(n)+1))(CTX), RK2; \
193 round_head_2(a, b, RX0, RY0, RX1, RY1); \
194 encround_tail(a ## 1, b ## 1, c ## 1, d ## 1, RX0, RY0, prerotate); \
196 encround_tail(a ## 2, b ## 2, c ## 2, d ## 2, RX1, RY1, prerotate);
198 #define decrypt_round(n, a, b, c, d, preload, prerotate) \
199 vbroadcastss (k+4*(2*(n)))(CTX), RK1; \
200 vbroadcastss (k+4*(2*(n)+1))(CTX), RK2; \
201 round_head_2(a, b, RX0, RY0, RX1, RY1); \
202 decround_tail(a ## 1, b ## 1, c ## 1, d ## 1, RX0, RY0, prerotate); \
204 decround_tail(a ## 2, b ## 2, c ## 2, d ## 2, RX1, RY1, prerotate);
206 #define encrypt_cycle(n) \
207 encrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l); \
208 encrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l);
210 #define encrypt_cycle_last(n) \
211 encrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l); \
212 encrypt_round(((2*n) + 1), RC, RD, RA, RB, dummy, dummy);
214 #define decrypt_cycle(n) \
215 decrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l); \
216 decrypt_round((2*n), RA, RB, RC, RD, preload_rgi, rotate_1l);
218 #define decrypt_cycle_last(n) \
219 decrypt_round(((2*n) + 1), RC, RD, RA, RB, preload_rgi, rotate_1l); \
220 decrypt_round((2*n), RA, RB, RC, RD, dummy, dummy);
222 #define transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
223 vpunpckldq x1, x0, t0; \
224 vpunpckhdq x1, x0, t2; \
225 vpunpckldq x3, x2, t1; \
226 vpunpckhdq x3, x2, x3; \
228 vpunpcklqdq t1, t0, x0; \
229 vpunpckhqdq t1, t0, x1; \
230 vpunpcklqdq x3, t2, x2; \
231 vpunpckhqdq x3, t2, x3;
233 #define inpack_blocks(x0, x1, x2, x3, wkey, t0, t1, t2) \
234 vpxor x0, wkey, x0; \
235 vpxor x1, wkey, x1; \
236 vpxor x2, wkey, x2; \
237 vpxor x3, wkey, x3; \
239 transpose_4x4(x0, x1, x2, x3, t0, t1, t2)
241 #define outunpack_blocks(x0, x1, x2, x3, wkey, t0, t1, t2) \
242 transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
244 vpxor x0, wkey, x0; \
245 vpxor x1, wkey, x1; \
246 vpxor x2, wkey, x2; \
253 * RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: blocks
255 * RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2: encrypted blocks
264 inpack_blocks(RA1, RB1, RC1, RD1, RK1, RX0, RY0, RK2);
267 inpack_blocks(RA2, RB2, RC2, RD2, RK1, RX0, RY0, RK2);
277 encrypt_cycle_last(7);
279 vmovdqu (w+4*4)(CTX), RK1;
285 outunpack_blocks(RC1, RD1, RA1, RB1, RK1, RX0, RY0, RK2);
286 outunpack_blocks(RC2, RD2, RA2, RB2, RK1, RX0, RY0, RK2);
289 ENDPROC(__twofish_enc_blk8)
295 * RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2: encrypted blocks
297 * RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: decrypted blocks
300 vmovdqu (w+4*4)(CTX), RK1;
305 inpack_blocks(RC1, RD1, RA1, RB1, RK1, RX0, RY0, RK2);
308 inpack_blocks(RC2, RD2, RA2, RB2, RK1, RX0, RY0, RK2);
318 decrypt_cycle_last(0);
320 vmovdqu (w)(CTX), RK1;
325 outunpack_blocks(RA1, RB1, RC1, RD1, RK1, RX0, RY0, RK2);
326 outunpack_blocks(RA2, RB2, RC2, RD2, RK1, RX0, RY0, RK2);
329 ENDPROC(__twofish_dec_blk8)
331 ENTRY(twofish_ecb_enc_8way)
341 load_8way(%rdx, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
343 call __twofish_enc_blk8;
345 store_8way(%r11, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
349 ENDPROC(twofish_ecb_enc_8way)
351 ENTRY(twofish_ecb_dec_8way)
361 load_8way(%rdx, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
363 call __twofish_dec_blk8;
365 store_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
369 ENDPROC(twofish_ecb_dec_8way)
371 ENTRY(twofish_cbc_dec_8way)
384 load_8way(%rdx, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
386 call __twofish_dec_blk8;
388 store_cbc_8way(%r12, %r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
394 ENDPROC(twofish_cbc_dec_8way)
396 ENTRY(twofish_ctr_8way)
401 * %rcx: iv (little endian, 128bit)
410 load_ctr_8way(%rcx, .Lbswap128_mask, RA1, RB1, RC1, RD1, RA2, RB2, RC2,
413 call __twofish_enc_blk8;
415 store_ctr_8way(%r12, %r11, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
421 ENDPROC(twofish_ctr_8way)
423 ENTRY(twofish_xts_enc_8way)
428 * %rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
434 /* regs <= src, dst <= IVs, regs <= regs xor IVs */
435 load_xts_8way(%rcx, %rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2,
436 RX0, RX1, RY0, .Lxts_gf128mul_and_shl1_mask);
438 call __twofish_enc_blk8;
440 /* dst <= regs xor IVs(in dst) */
441 store_xts_8way(%r11, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
445 ENDPROC(twofish_xts_enc_8way)
447 ENTRY(twofish_xts_dec_8way)
452 * %rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
458 /* regs <= src, dst <= IVs, regs <= regs xor IVs */
459 load_xts_8way(%rcx, %rdx, %rsi, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2,
460 RX0, RX1, RY0, .Lxts_gf128mul_and_shl1_mask);
462 call __twofish_dec_blk8;
464 /* dst <= regs xor IVs(in dst) */
465 store_xts_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
469 ENDPROC(twofish_xts_dec_8way)