1/*
2 * Cast5 Cipher 16-way parallel algorithm (AVX/x86_64)
3 *
4 * Copyright (C) 2012 Johannes Goetzfried
5 *     <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
6 *
7 * Copyright © 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
8 *
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.
13 *
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.
18 *
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
22 * USA
23 *
24 */
25
26#include <linux/linkage.h>
27#include <asm/frame.h>
28
29.file "cast5-avx-x86_64-asm_64.S"
30
31.extern cast_s1
32.extern cast_s2
33.extern cast_s3
34.extern cast_s4
35
36/* structure of crypto context */
37#define km	0
38#define kr	(16*4)
39#define rr	((16*4)+16)
40
41/* s-boxes */
42#define s1	cast_s1
43#define s2	cast_s2
44#define s3	cast_s3
45#define s4	cast_s4
46
47/**********************************************************************
48  16-way AVX cast5
49 **********************************************************************/
50#define CTX %rdi
51
52#define RL1 %xmm0
53#define RR1 %xmm1
54#define RL2 %xmm2
55#define RR2 %xmm3
56#define RL3 %xmm4
57#define RR3 %xmm5
58#define RL4 %xmm6
59#define RR4 %xmm7
60
61#define RX %xmm8
62
63#define RKM  %xmm9
64#define RKR  %xmm10
65#define RKRF %xmm11
66#define RKRR %xmm12
67
68#define R32  %xmm13
69#define R1ST %xmm14
70
71#define RTMP %xmm15
72
73#define RID1  %rbp
74#define RID1d %ebp
75#define RID2  %rsi
76#define RID2d %esi
77
78#define RGI1   %rdx
79#define RGI1bl %dl
80#define RGI1bh %dh
81#define RGI2   %rcx
82#define RGI2bl %cl
83#define RGI2bh %ch
84
85#define RGI3   %rax
86#define RGI3bl %al
87#define RGI3bh %ah
88#define RGI4   %rbx
89#define RGI4bl %bl
90#define RGI4bh %bh
91
92#define RFS1  %r8
93#define RFS1d %r8d
94#define RFS2  %r9
95#define RFS2d %r9d
96#define RFS3  %r10
97#define RFS3d %r10d
98
99
100#define lookup_32bit(src, dst, op1, op2, op3, interleave_op, il_reg) \
101	movzbl		src ## bh,     RID1d;    \
102	movzbl		src ## bl,     RID2d;    \
103	shrq $16,	src;                     \
104	movl		s1(, RID1, 4), dst ## d; \
105	op1		s2(, RID2, 4), dst ## d; \
106	movzbl		src ## bh,     RID1d;    \
107	movzbl		src ## bl,     RID2d;    \
108	interleave_op(il_reg);			 \
109	op2		s3(, RID1, 4), dst ## d; \
110	op3		s4(, RID2, 4), dst ## d;
111
112#define dummy(d) /* do nothing */
113
114#define shr_next(reg) \
115	shrq $16,	reg;
116
117#define F_head(a, x, gi1, gi2, op0) \
118	op0	a,	RKM,  x;                 \
119	vpslld	RKRF,	x,    RTMP;              \
120	vpsrld	RKRR,	x,    x;                 \
121	vpor	RTMP,	x,    x;                 \
122	\
123	vmovq		x,    gi1;               \
124	vpextrq $1,	x,    gi2;
125
126#define F_tail(a, x, gi1, gi2, op1, op2, op3) \
127	lookup_32bit(##gi1, RFS1, op1, op2, op3, shr_next, ##gi1); \
128	lookup_32bit(##gi2, RFS3, op1, op2, op3, shr_next, ##gi2); \
129	\
130	lookup_32bit(##gi1, RFS2, op1, op2, op3, dummy, none);     \
131	shlq $32,	RFS2;                                      \
132	orq		RFS1, RFS2;                                \
133	lookup_32bit(##gi2, RFS1, op1, op2, op3, dummy, none);     \
134	shlq $32,	RFS1;                                      \
135	orq		RFS1, RFS3;                                \
136	\
137	vmovq		RFS2, x;                                   \
138	vpinsrq $1,	RFS3, x, x;
139
140#define F_2(a1, b1, a2, b2, op0, op1, op2, op3) \
141	F_head(b1, RX, RGI1, RGI2, op0);              \
142	F_head(b2, RX, RGI3, RGI4, op0);              \
143	\
144	F_tail(b1, RX, RGI1, RGI2, op1, op2, op3);    \
145	F_tail(b2, RTMP, RGI3, RGI4, op1, op2, op3);  \
146	\
147	vpxor		a1, RX,   a1;                 \
148	vpxor		a2, RTMP, a2;
149
150#define F1_2(a1, b1, a2, b2) \
151	F_2(a1, b1, a2, b2, vpaddd, xorl, subl, addl)
152#define F2_2(a1, b1, a2, b2) \
153	F_2(a1, b1, a2, b2, vpxor, subl, addl, xorl)
154#define F3_2(a1, b1, a2, b2) \
155	F_2(a1, b1, a2, b2, vpsubd, addl, xorl, subl)
156
157#define subround(a1, b1, a2, b2, f) \
158	F ## f ## _2(a1, b1, a2, b2);
159
160#define round(l, r, n, f) \
161	vbroadcastss 	(km+(4*n))(CTX), RKM;        \
162	vpand		R1ST,            RKR,  RKRF; \
163	vpsubq		RKRF,            R32,  RKRR; \
164	vpsrldq $1,	RKR,             RKR;        \
165	subround(l ## 1, r ## 1, l ## 2, r ## 2, f); \
166	subround(l ## 3, r ## 3, l ## 4, r ## 4, f);
167
168#define enc_preload_rkr() \
169	vbroadcastss	.L16_mask,                RKR;      \
170	/* add 16-bit rotation to key rotations (mod 32) */ \
171	vpxor		kr(CTX),                  RKR, RKR;
172
173#define dec_preload_rkr() \
174	vbroadcastss	.L16_mask,                RKR;      \
175	/* add 16-bit rotation to key rotations (mod 32) */ \
176	vpxor		kr(CTX),                  RKR, RKR; \
177	vpshufb		.Lbswap128_mask,          RKR, RKR;
178
179#define transpose_2x4(x0, x1, t0, t1) \
180	vpunpckldq		x1, x0, t0; \
181	vpunpckhdq		x1, x0, t1; \
182	\
183	vpunpcklqdq		t1, t0, x0; \
184	vpunpckhqdq		t1, t0, x1;
185
186#define inpack_blocks(x0, x1, t0, t1, rmask) \
187	vpshufb rmask, 	x0,	x0; \
188	vpshufb rmask, 	x1,	x1; \
189	\
190	transpose_2x4(x0, x1, t0, t1)
191
192#define outunpack_blocks(x0, x1, t0, t1, rmask) \
193	transpose_2x4(x0, x1, t0, t1) \
194	\
195	vpshufb rmask,	x0, x0;           \
196	vpshufb rmask,	x1, x1;
197
198.data
199
200.align 16
201.Lbswap_mask:
202	.byte 3, 2, 1, 0, 7, 6, 5, 4, 11, 10, 9, 8, 15, 14, 13, 12
203.Lbswap128_mask:
204	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
205.Lbswap_iv_mask:
206	.byte 7, 6, 5, 4, 3, 2, 1, 0, 7, 6, 5, 4, 3, 2, 1, 0
207.L16_mask:
208	.byte 16, 16, 16, 16
209.L32_mask:
210	.byte 32, 0, 0, 0
211.Lfirst_mask:
212	.byte 0x1f, 0, 0, 0
213
214.text
215
216.align 16
217__cast5_enc_blk16:
218	/* input:
219	 *	%rdi: ctx, CTX
220	 *	RL1: blocks 1 and 2
221	 *	RR1: blocks 3 and 4
222	 *	RL2: blocks 5 and 6
223	 *	RR2: blocks 7 and 8
224	 *	RL3: blocks 9 and 10
225	 *	RR3: blocks 11 and 12
226	 *	RL4: blocks 13 and 14
227	 *	RR4: blocks 15 and 16
228	 * output:
229	 *	RL1: encrypted blocks 1 and 2
230	 *	RR1: encrypted blocks 3 and 4
231	 *	RL2: encrypted blocks 5 and 6
232	 *	RR2: encrypted blocks 7 and 8
233	 *	RL3: encrypted blocks 9 and 10
234	 *	RR3: encrypted blocks 11 and 12
235	 *	RL4: encrypted blocks 13 and 14
236	 *	RR4: encrypted blocks 15 and 16
237	 */
238
239	pushq %rbp;
240	pushq %rbx;
241
242	vmovdqa .Lbswap_mask, RKM;
243	vmovd .Lfirst_mask, R1ST;
244	vmovd .L32_mask, R32;
245	enc_preload_rkr();
246
247	inpack_blocks(RL1, RR1, RTMP, RX, RKM);
248	inpack_blocks(RL2, RR2, RTMP, RX, RKM);
249	inpack_blocks(RL3, RR3, RTMP, RX, RKM);
250	inpack_blocks(RL4, RR4, RTMP, RX, RKM);
251
252	round(RL, RR, 0, 1);
253	round(RR, RL, 1, 2);
254	round(RL, RR, 2, 3);
255	round(RR, RL, 3, 1);
256	round(RL, RR, 4, 2);
257	round(RR, RL, 5, 3);
258	round(RL, RR, 6, 1);
259	round(RR, RL, 7, 2);
260	round(RL, RR, 8, 3);
261	round(RR, RL, 9, 1);
262	round(RL, RR, 10, 2);
263	round(RR, RL, 11, 3);
264
265	movzbl rr(CTX), %eax;
266	testl %eax, %eax;
267	jnz .L__skip_enc;
268
269	round(RL, RR, 12, 1);
270	round(RR, RL, 13, 2);
271	round(RL, RR, 14, 3);
272	round(RR, RL, 15, 1);
273
274.L__skip_enc:
275	popq %rbx;
276	popq %rbp;
277
278	vmovdqa .Lbswap_mask, RKM;
279
280	outunpack_blocks(RR1, RL1, RTMP, RX, RKM);
281	outunpack_blocks(RR2, RL2, RTMP, RX, RKM);
282	outunpack_blocks(RR3, RL3, RTMP, RX, RKM);
283	outunpack_blocks(RR4, RL4, RTMP, RX, RKM);
284
285	ret;
286ENDPROC(__cast5_enc_blk16)
287
288.align 16
289__cast5_dec_blk16:
290	/* input:
291	 *	%rdi: ctx, CTX
292	 *	RL1: encrypted blocks 1 and 2
293	 *	RR1: encrypted blocks 3 and 4
294	 *	RL2: encrypted blocks 5 and 6
295	 *	RR2: encrypted blocks 7 and 8
296	 *	RL3: encrypted blocks 9 and 10
297	 *	RR3: encrypted blocks 11 and 12
298	 *	RL4: encrypted blocks 13 and 14
299	 *	RR4: encrypted blocks 15 and 16
300	 * output:
301	 *	RL1: decrypted blocks 1 and 2
302	 *	RR1: decrypted blocks 3 and 4
303	 *	RL2: decrypted blocks 5 and 6
304	 *	RR2: decrypted blocks 7 and 8
305	 *	RL3: decrypted blocks 9 and 10
306	 *	RR3: decrypted blocks 11 and 12
307	 *	RL4: decrypted blocks 13 and 14
308	 *	RR4: decrypted blocks 15 and 16
309	 */
310
311	pushq %rbp;
312	pushq %rbx;
313
314	vmovdqa .Lbswap_mask, RKM;
315	vmovd .Lfirst_mask, R1ST;
316	vmovd .L32_mask, R32;
317	dec_preload_rkr();
318
319	inpack_blocks(RL1, RR1, RTMP, RX, RKM);
320	inpack_blocks(RL2, RR2, RTMP, RX, RKM);
321	inpack_blocks(RL3, RR3, RTMP, RX, RKM);
322	inpack_blocks(RL4, RR4, RTMP, RX, RKM);
323
324	movzbl rr(CTX), %eax;
325	testl %eax, %eax;
326	jnz .L__skip_dec;
327
328	round(RL, RR, 15, 1);
329	round(RR, RL, 14, 3);
330	round(RL, RR, 13, 2);
331	round(RR, RL, 12, 1);
332
333.L__dec_tail:
334	round(RL, RR, 11, 3);
335	round(RR, RL, 10, 2);
336	round(RL, RR, 9, 1);
337	round(RR, RL, 8, 3);
338	round(RL, RR, 7, 2);
339	round(RR, RL, 6, 1);
340	round(RL, RR, 5, 3);
341	round(RR, RL, 4, 2);
342	round(RL, RR, 3, 1);
343	round(RR, RL, 2, 3);
344	round(RL, RR, 1, 2);
345	round(RR, RL, 0, 1);
346
347	vmovdqa .Lbswap_mask, RKM;
348	popq %rbx;
349	popq %rbp;
350
351	outunpack_blocks(RR1, RL1, RTMP, RX, RKM);
352	outunpack_blocks(RR2, RL2, RTMP, RX, RKM);
353	outunpack_blocks(RR3, RL3, RTMP, RX, RKM);
354	outunpack_blocks(RR4, RL4, RTMP, RX, RKM);
355
356	ret;
357
358.L__skip_dec:
359	vpsrldq $4, RKR, RKR;
360	jmp .L__dec_tail;
361ENDPROC(__cast5_dec_blk16)
362
363ENTRY(cast5_ecb_enc_16way)
364	/* input:
365	 *	%rdi: ctx, CTX
366	 *	%rsi: dst
367	 *	%rdx: src
368	 */
369	FRAME_BEGIN
370
371	movq %rsi, %r11;
372
373	vmovdqu (0*4*4)(%rdx), RL1;
374	vmovdqu (1*4*4)(%rdx), RR1;
375	vmovdqu (2*4*4)(%rdx), RL2;
376	vmovdqu (3*4*4)(%rdx), RR2;
377	vmovdqu (4*4*4)(%rdx), RL3;
378	vmovdqu (5*4*4)(%rdx), RR3;
379	vmovdqu (6*4*4)(%rdx), RL4;
380	vmovdqu (7*4*4)(%rdx), RR4;
381
382	call __cast5_enc_blk16;
383
384	vmovdqu RR1, (0*4*4)(%r11);
385	vmovdqu RL1, (1*4*4)(%r11);
386	vmovdqu RR2, (2*4*4)(%r11);
387	vmovdqu RL2, (3*4*4)(%r11);
388	vmovdqu RR3, (4*4*4)(%r11);
389	vmovdqu RL3, (5*4*4)(%r11);
390	vmovdqu RR4, (6*4*4)(%r11);
391	vmovdqu RL4, (7*4*4)(%r11);
392
393	FRAME_END
394	ret;
395ENDPROC(cast5_ecb_enc_16way)
396
397ENTRY(cast5_ecb_dec_16way)
398	/* input:
399	 *	%rdi: ctx, CTX
400	 *	%rsi: dst
401	 *	%rdx: src
402	 */
403
404	FRAME_BEGIN
405	movq %rsi, %r11;
406
407	vmovdqu (0*4*4)(%rdx), RL1;
408	vmovdqu (1*4*4)(%rdx), RR1;
409	vmovdqu (2*4*4)(%rdx), RL2;
410	vmovdqu (3*4*4)(%rdx), RR2;
411	vmovdqu (4*4*4)(%rdx), RL3;
412	vmovdqu (5*4*4)(%rdx), RR3;
413	vmovdqu (6*4*4)(%rdx), RL4;
414	vmovdqu (7*4*4)(%rdx), RR4;
415
416	call __cast5_dec_blk16;
417
418	vmovdqu RR1, (0*4*4)(%r11);
419	vmovdqu RL1, (1*4*4)(%r11);
420	vmovdqu RR2, (2*4*4)(%r11);
421	vmovdqu RL2, (3*4*4)(%r11);
422	vmovdqu RR3, (4*4*4)(%r11);
423	vmovdqu RL3, (5*4*4)(%r11);
424	vmovdqu RR4, (6*4*4)(%r11);
425	vmovdqu RL4, (7*4*4)(%r11);
426
427	FRAME_END
428	ret;
429ENDPROC(cast5_ecb_dec_16way)
430
431ENTRY(cast5_cbc_dec_16way)
432	/* input:
433	 *	%rdi: ctx, CTX
434	 *	%rsi: dst
435	 *	%rdx: src
436	 */
437	FRAME_BEGIN
438
439	pushq %r12;
440
441	movq %rsi, %r11;
442	movq %rdx, %r12;
443
444	vmovdqu (0*16)(%rdx), RL1;
445	vmovdqu (1*16)(%rdx), RR1;
446	vmovdqu (2*16)(%rdx), RL2;
447	vmovdqu (3*16)(%rdx), RR2;
448	vmovdqu (4*16)(%rdx), RL3;
449	vmovdqu (5*16)(%rdx), RR3;
450	vmovdqu (6*16)(%rdx), RL4;
451	vmovdqu (7*16)(%rdx), RR4;
452
453	call __cast5_dec_blk16;
454
455	/* xor with src */
456	vmovq (%r12), RX;
457	vpshufd $0x4f, RX, RX;
458	vpxor RX, RR1, RR1;
459	vpxor 0*16+8(%r12), RL1, RL1;
460	vpxor 1*16+8(%r12), RR2, RR2;
461	vpxor 2*16+8(%r12), RL2, RL2;
462	vpxor 3*16+8(%r12), RR3, RR3;
463	vpxor 4*16+8(%r12), RL3, RL3;
464	vpxor 5*16+8(%r12), RR4, RR4;
465	vpxor 6*16+8(%r12), RL4, RL4;
466
467	vmovdqu RR1, (0*16)(%r11);
468	vmovdqu RL1, (1*16)(%r11);
469	vmovdqu RR2, (2*16)(%r11);
470	vmovdqu RL2, (3*16)(%r11);
471	vmovdqu RR3, (4*16)(%r11);
472	vmovdqu RL3, (5*16)(%r11);
473	vmovdqu RR4, (6*16)(%r11);
474	vmovdqu RL4, (7*16)(%r11);
475
476	popq %r12;
477
478	FRAME_END
479	ret;
480ENDPROC(cast5_cbc_dec_16way)
481
482ENTRY(cast5_ctr_16way)
483	/* input:
484	 *	%rdi: ctx, CTX
485	 *	%rsi: dst
486	 *	%rdx: src
487	 *	%rcx: iv (big endian, 64bit)
488	 */
489	FRAME_BEGIN
490
491	pushq %r12;
492
493	movq %rsi, %r11;
494	movq %rdx, %r12;
495
496	vpcmpeqd RTMP, RTMP, RTMP;
497	vpsrldq $8, RTMP, RTMP; /* low: -1, high: 0 */
498
499	vpcmpeqd RKR, RKR, RKR;
500	vpaddq RKR, RKR, RKR; /* low: -2, high: -2 */
501	vmovdqa .Lbswap_iv_mask, R1ST;
502	vmovdqa .Lbswap128_mask, RKM;
503
504	/* load IV and byteswap */
505	vmovq (%rcx), RX;
506	vpshufb R1ST, RX, RX;
507
508	/* construct IVs */
509	vpsubq RTMP, RX, RX;  /* le: IV1, IV0 */
510	vpshufb RKM, RX, RL1; /* be: IV0, IV1 */
511	vpsubq RKR, RX, RX;
512	vpshufb RKM, RX, RR1; /* be: IV2, IV3 */
513	vpsubq RKR, RX, RX;
514	vpshufb RKM, RX, RL2; /* be: IV4, IV5 */
515	vpsubq RKR, RX, RX;
516	vpshufb RKM, RX, RR2; /* be: IV6, IV7 */
517	vpsubq RKR, RX, RX;
518	vpshufb RKM, RX, RL3; /* be: IV8, IV9 */
519	vpsubq RKR, RX, RX;
520	vpshufb RKM, RX, RR3; /* be: IV10, IV11 */
521	vpsubq RKR, RX, RX;
522	vpshufb RKM, RX, RL4; /* be: IV12, IV13 */
523	vpsubq RKR, RX, RX;
524	vpshufb RKM, RX, RR4; /* be: IV14, IV15 */
525
526	/* store last IV */
527	vpsubq RTMP, RX, RX; /* le: IV16, IV14 */
528	vpshufb R1ST, RX, RX; /* be: IV16, IV16 */
529	vmovq RX, (%rcx);
530
531	call __cast5_enc_blk16;
532
533	/* dst = src ^ iv */
534	vpxor (0*16)(%r12), RR1, RR1;
535	vpxor (1*16)(%r12), RL1, RL1;
536	vpxor (2*16)(%r12), RR2, RR2;
537	vpxor (3*16)(%r12), RL2, RL2;
538	vpxor (4*16)(%r12), RR3, RR3;
539	vpxor (5*16)(%r12), RL3, RL3;
540	vpxor (6*16)(%r12), RR4, RR4;
541	vpxor (7*16)(%r12), RL4, RL4;
542	vmovdqu RR1, (0*16)(%r11);
543	vmovdqu RL1, (1*16)(%r11);
544	vmovdqu RR2, (2*16)(%r11);
545	vmovdqu RL2, (3*16)(%r11);
546	vmovdqu RR3, (4*16)(%r11);
547	vmovdqu RL3, (5*16)(%r11);
548	vmovdqu RR4, (6*16)(%r11);
549	vmovdqu RL4, (7*16)(%r11);
550
551	popq %r12;
552
553	FRAME_END
554	ret;
555ENDPROC(cast5_ctr_16way)
556