xref: /linux/arch/x86/crypto/serpent-sse2-i586-asm_32.S (revision ca55b2fef3a9373fcfc30f82fd26bc7fccbda732)
1/*
2 * Serpent Cipher 4-way parallel algorithm (i586/SSE2)
3 *
4 * Copyright (C) 2011 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
5 *
6 * Based on crypto/serpent.c by
7 *  Copyright (C) 2002 Dag Arne Osvik <osvik@ii.uib.no>
8 *                2003 Herbert Valerio Riedel <hvr@gnu.org>
9 *
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License as published by
12 * the Free Software Foundation; either version 2 of the License, or
13 * (at your option) any later version.
14 *
15 * This program is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18 * GNU General Public License for more details.
19 *
20 * You should have received a copy of the GNU General Public License
21 * along with this program; if not, write to the Free Software
22 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307
23 * USA
24 *
25 */
26
27#include <linux/linkage.h>
28
29.file "serpent-sse2-i586-asm_32.S"
30.text
31
32#define arg_ctx 4
33#define arg_dst 8
34#define arg_src 12
35#define arg_xor 16
36
37/**********************************************************************
38  4-way SSE2 serpent
39 **********************************************************************/
40#define CTX %edx
41
42#define RA %xmm0
43#define RB %xmm1
44#define RC %xmm2
45#define RD %xmm3
46#define RE %xmm4
47
48#define RT0 %xmm5
49#define RT1 %xmm6
50
51#define RNOT %xmm7
52
53#define get_key(i, j, t) \
54	movd (4*(i)+(j))*4(CTX), t; \
55	pshufd $0, t, t;
56
57#define K(x0, x1, x2, x3, x4, i) \
58	get_key(i, 0, x4); \
59	get_key(i, 1, RT0); \
60	get_key(i, 2, RT1); \
61	pxor x4,		x0; \
62	pxor RT0,		x1; \
63	pxor RT1,		x2; \
64	get_key(i, 3, x4); \
65	pxor x4,		x3;
66
67#define LK(x0, x1, x2, x3, x4, i) \
68	movdqa x0,		x4; \
69	pslld $13,		x0; \
70	psrld $(32 - 13),	x4; \
71	por x4,			x0; \
72	pxor x0,		x1; \
73	movdqa x2,		x4; \
74	pslld $3,		x2; \
75	psrld $(32 - 3),	x4; \
76	por x4,			x2; \
77	pxor x2,		x1; \
78	movdqa x1,		x4; \
79	pslld $1,		x1; \
80	psrld $(32 - 1),	x4; \
81	por x4,			x1; \
82	movdqa x0,		x4; \
83	pslld $3,		x4; \
84	pxor x2,		x3; \
85	pxor x4,		x3; \
86	movdqa x3,		x4; \
87	pslld $7,		x3; \
88	psrld $(32 - 7),	x4; \
89	por x4,			x3; \
90	movdqa x1,		x4; \
91	pslld $7,		x4; \
92	pxor x1,		x0; \
93	pxor x3,		x0; \
94	pxor x3,		x2; \
95	pxor x4,		x2; \
96	movdqa x0,		x4; \
97	get_key(i, 1, RT0); \
98	pxor RT0,		x1; \
99	get_key(i, 3, RT0); \
100	pxor RT0,		x3; \
101	pslld $5,		x0; \
102	psrld $(32 - 5),	x4; \
103	por x4,			x0; \
104	movdqa x2,		x4; \
105	pslld $22,		x2; \
106	psrld $(32 - 22),	x4; \
107	por x4,			x2; \
108	get_key(i, 0, RT0); \
109	pxor RT0,		x0; \
110	get_key(i, 2, RT0); \
111	pxor RT0,		x2;
112
113#define KL(x0, x1, x2, x3, x4, i) \
114	K(x0, x1, x2, x3, x4, i); \
115	movdqa x0,		x4; \
116	psrld $5,		x0; \
117	pslld $(32 - 5),	x4; \
118	por x4,			x0; \
119	movdqa x2,		x4; \
120	psrld $22,		x2; \
121	pslld $(32 - 22),	x4; \
122	por x4,			x2; \
123	pxor x3,		x2; \
124	pxor x3,		x0; \
125	movdqa x1,		x4; \
126	pslld $7,		x4; \
127	pxor x1,		x0; \
128	pxor x4,		x2; \
129	movdqa x1,		x4; \
130	psrld $1,		x1; \
131	pslld $(32 - 1),	x4; \
132	por x4,			x1; \
133	movdqa x3,		x4; \
134	psrld $7,		x3; \
135	pslld $(32 - 7),	x4; \
136	por x4,			x3; \
137	pxor x0,		x1; \
138	movdqa x0,		x4; \
139	pslld $3,		x4; \
140	pxor x4,		x3; \
141	movdqa x0,		x4; \
142	psrld $13,		x0; \
143	pslld $(32 - 13),	x4; \
144	por x4,			x0; \
145	pxor x2,		x1; \
146	pxor x2,		x3; \
147	movdqa x2,		x4; \
148	psrld $3,		x2; \
149	pslld $(32 - 3),	x4; \
150	por x4,			x2;
151
152#define S0(x0, x1, x2, x3, x4) \
153	movdqa x3,		x4; \
154	por x0,			x3; \
155	pxor x4,		x0; \
156	pxor x2,		x4; \
157	pxor RNOT,		x4; \
158	pxor x1,		x3; \
159	pand x0,		x1; \
160	pxor x4,		x1; \
161	pxor x0,		x2; \
162	pxor x3,		x0; \
163	por x0,			x4; \
164	pxor x2,		x0; \
165	pand x1,		x2; \
166	pxor x2,		x3; \
167	pxor RNOT,		x1; \
168	pxor x4,		x2; \
169	pxor x2,		x1;
170
171#define S1(x0, x1, x2, x3, x4) \
172	movdqa x1,		x4; \
173	pxor x0,		x1; \
174	pxor x3,		x0; \
175	pxor RNOT,		x3; \
176	pand x1,		x4; \
177	por x1,			x0; \
178	pxor x2,		x3; \
179	pxor x3,		x0; \
180	pxor x3,		x1; \
181	pxor x4,		x3; \
182	por x4,			x1; \
183	pxor x2,		x4; \
184	pand x0,		x2; \
185	pxor x1,		x2; \
186	por x0,			x1; \
187	pxor RNOT,		x0; \
188	pxor x2,		x0; \
189	pxor x1,		x4;
190
191#define S2(x0, x1, x2, x3, x4) \
192	pxor RNOT,		x3; \
193	pxor x0,		x1; \
194	movdqa x0,		x4; \
195	pand x2,		x0; \
196	pxor x3,		x0; \
197	por x4,			x3; \
198	pxor x1,		x2; \
199	pxor x1,		x3; \
200	pand x0,		x1; \
201	pxor x2,		x0; \
202	pand x3,		x2; \
203	por x1,			x3; \
204	pxor RNOT,		x0; \
205	pxor x0,		x3; \
206	pxor x0,		x4; \
207	pxor x2,		x0; \
208	por x2,			x1;
209
210#define S3(x0, x1, x2, x3, x4) \
211	movdqa x1,		x4; \
212	pxor x3,		x1; \
213	por x0,			x3; \
214	pand x0,		x4; \
215	pxor x2,		x0; \
216	pxor x1,		x2; \
217	pand x3,		x1; \
218	pxor x3,		x2; \
219	por x4,			x0; \
220	pxor x3,		x4; \
221	pxor x0,		x1; \
222	pand x3,		x0; \
223	pand x4,		x3; \
224	pxor x2,		x3; \
225	por x1,			x4; \
226	pand x1,		x2; \
227	pxor x3,		x4; \
228	pxor x3,		x0; \
229	pxor x2,		x3;
230
231#define S4(x0, x1, x2, x3, x4) \
232	movdqa x3,		x4; \
233	pand x0,		x3; \
234	pxor x4,		x0; \
235	pxor x2,		x3; \
236	por x4,			x2; \
237	pxor x1,		x0; \
238	pxor x3,		x4; \
239	por x0,			x2; \
240	pxor x1,		x2; \
241	pand x0,		x1; \
242	pxor x4,		x1; \
243	pand x2,		x4; \
244	pxor x3,		x2; \
245	pxor x0,		x4; \
246	por x1,			x3; \
247	pxor RNOT,		x1; \
248	pxor x0,		x3;
249
250#define S5(x0, x1, x2, x3, x4) \
251	movdqa x1,		x4; \
252	por x0,			x1; \
253	pxor x1,		x2; \
254	pxor RNOT,		x3; \
255	pxor x0,		x4; \
256	pxor x2,		x0; \
257	pand x4,		x1; \
258	por x3,			x4; \
259	pxor x0,		x4; \
260	pand x3,		x0; \
261	pxor x3,		x1; \
262	pxor x2,		x3; \
263	pxor x1,		x0; \
264	pand x4,		x2; \
265	pxor x2,		x1; \
266	pand x0,		x2; \
267	pxor x2,		x3;
268
269#define S6(x0, x1, x2, x3, x4) \
270	movdqa x1,		x4; \
271	pxor x0,		x3; \
272	pxor x2,		x1; \
273	pxor x0,		x2; \
274	pand x3,		x0; \
275	por x3,			x1; \
276	pxor RNOT,		x4; \
277	pxor x1,		x0; \
278	pxor x2,		x1; \
279	pxor x4,		x3; \
280	pxor x0,		x4; \
281	pand x0,		x2; \
282	pxor x1,		x4; \
283	pxor x3,		x2; \
284	pand x1,		x3; \
285	pxor x0,		x3; \
286	pxor x2,		x1;
287
288#define S7(x0, x1, x2, x3, x4) \
289	pxor RNOT,		x1; \
290	movdqa x1,		x4; \
291	pxor RNOT,		x0; \
292	pand x2,		x1; \
293	pxor x3,		x1; \
294	por x4,			x3; \
295	pxor x2,		x4; \
296	pxor x3,		x2; \
297	pxor x0,		x3; \
298	por x1,			x0; \
299	pand x0,		x2; \
300	pxor x4,		x0; \
301	pxor x3,		x4; \
302	pand x0,		x3; \
303	pxor x1,		x4; \
304	pxor x4,		x2; \
305	pxor x1,		x3; \
306	por x0,			x4; \
307	pxor x1,		x4;
308
309#define SI0(x0, x1, x2, x3, x4) \
310	movdqa x3,		x4; \
311	pxor x0,		x1; \
312	por x1,			x3; \
313	pxor x1,		x4; \
314	pxor RNOT,		x0; \
315	pxor x3,		x2; \
316	pxor x0,		x3; \
317	pand x1,		x0; \
318	pxor x2,		x0; \
319	pand x3,		x2; \
320	pxor x4,		x3; \
321	pxor x3,		x2; \
322	pxor x3,		x1; \
323	pand x0,		x3; \
324	pxor x0,		x1; \
325	pxor x2,		x0; \
326	pxor x3,		x4;
327
328#define SI1(x0, x1, x2, x3, x4) \
329	pxor x3,		x1; \
330	movdqa x0,		x4; \
331	pxor x2,		x0; \
332	pxor RNOT,		x2; \
333	por x1,			x4; \
334	pxor x3,		x4; \
335	pand x1,		x3; \
336	pxor x2,		x1; \
337	pand x4,		x2; \
338	pxor x1,		x4; \
339	por x3,			x1; \
340	pxor x0,		x3; \
341	pxor x0,		x2; \
342	por x4,			x0; \
343	pxor x4,		x2; \
344	pxor x0,		x1; \
345	pxor x1,		x4;
346
347#define SI2(x0, x1, x2, x3, x4) \
348	pxor x1,		x2; \
349	movdqa x3,		x4; \
350	pxor RNOT,		x3; \
351	por x2,			x3; \
352	pxor x4,		x2; \
353	pxor x0,		x4; \
354	pxor x1,		x3; \
355	por x2,			x1; \
356	pxor x0,		x2; \
357	pxor x4,		x1; \
358	por x3,			x4; \
359	pxor x3,		x2; \
360	pxor x2,		x4; \
361	pand x1,		x2; \
362	pxor x3,		x2; \
363	pxor x4,		x3; \
364	pxor x0,		x4;
365
366#define SI3(x0, x1, x2, x3, x4) \
367	pxor x1,		x2; \
368	movdqa x1,		x4; \
369	pand x2,		x1; \
370	pxor x0,		x1; \
371	por x4,			x0; \
372	pxor x3,		x4; \
373	pxor x3,		x0; \
374	por x1,			x3; \
375	pxor x2,		x1; \
376	pxor x3,		x1; \
377	pxor x2,		x0; \
378	pxor x3,		x2; \
379	pand x1,		x3; \
380	pxor x0,		x1; \
381	pand x2,		x0; \
382	pxor x3,		x4; \
383	pxor x0,		x3; \
384	pxor x1,		x0;
385
386#define SI4(x0, x1, x2, x3, x4) \
387	pxor x3,		x2; \
388	movdqa x0,		x4; \
389	pand x1,		x0; \
390	pxor x2,		x0; \
391	por x3,			x2; \
392	pxor RNOT,		x4; \
393	pxor x0,		x1; \
394	pxor x2,		x0; \
395	pand x4,		x2; \
396	pxor x0,		x2; \
397	por x4,			x0; \
398	pxor x3,		x0; \
399	pand x2,		x3; \
400	pxor x3,		x4; \
401	pxor x1,		x3; \
402	pand x0,		x1; \
403	pxor x1,		x4; \
404	pxor x3,		x0;
405
406#define SI5(x0, x1, x2, x3, x4) \
407	movdqa x1,		x4; \
408	por x2,			x1; \
409	pxor x4,		x2; \
410	pxor x3,		x1; \
411	pand x4,		x3; \
412	pxor x3,		x2; \
413	por x0,			x3; \
414	pxor RNOT,		x0; \
415	pxor x2,		x3; \
416	por x0,			x2; \
417	pxor x1,		x4; \
418	pxor x4,		x2; \
419	pand x0,		x4; \
420	pxor x1,		x0; \
421	pxor x3,		x1; \
422	pand x2,		x0; \
423	pxor x3,		x2; \
424	pxor x2,		x0; \
425	pxor x4,		x2; \
426	pxor x3,		x4;
427
428#define SI6(x0, x1, x2, x3, x4) \
429	pxor x2,		x0; \
430	movdqa x0,		x4; \
431	pand x3,		x0; \
432	pxor x3,		x2; \
433	pxor x2,		x0; \
434	pxor x1,		x3; \
435	por x4,			x2; \
436	pxor x3,		x2; \
437	pand x0,		x3; \
438	pxor RNOT,		x0; \
439	pxor x1,		x3; \
440	pand x2,		x1; \
441	pxor x0,		x4; \
442	pxor x4,		x3; \
443	pxor x2,		x4; \
444	pxor x1,		x0; \
445	pxor x0,		x2;
446
447#define SI7(x0, x1, x2, x3, x4) \
448	movdqa x3,		x4; \
449	pand x0,		x3; \
450	pxor x2,		x0; \
451	por x4,			x2; \
452	pxor x1,		x4; \
453	pxor RNOT,		x0; \
454	por x3,			x1; \
455	pxor x0,		x4; \
456	pand x2,		x0; \
457	pxor x1,		x0; \
458	pand x2,		x1; \
459	pxor x2,		x3; \
460	pxor x3,		x4; \
461	pand x3,		x2; \
462	por x0,			x3; \
463	pxor x4,		x1; \
464	pxor x4,		x3; \
465	pand x0,		x4; \
466	pxor x2,		x4;
467
468#define transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
469	movdqa x0,		t2; \
470	punpckldq x1,		x0; \
471	punpckhdq x1,		t2; \
472	movdqa x2,		t1; \
473	punpckhdq x3,		x2; \
474	punpckldq x3,		t1; \
475	movdqa x0,		x1; \
476	punpcklqdq t1,		x0; \
477	punpckhqdq t1,		x1; \
478	movdqa t2,		x3; \
479	punpcklqdq x2,		t2; \
480	punpckhqdq x2,		x3; \
481	movdqa t2,		x2;
482
483#define read_blocks(in, x0, x1, x2, x3, t0, t1, t2) \
484	movdqu (0*4*4)(in),	x0; \
485	movdqu (1*4*4)(in),	x1; \
486	movdqu (2*4*4)(in),	x2; \
487	movdqu (3*4*4)(in),	x3; \
488	\
489	transpose_4x4(x0, x1, x2, x3, t0, t1, t2)
490
491#define write_blocks(out, x0, x1, x2, x3, t0, t1, t2) \
492	transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
493	\
494	movdqu x0, (0*4*4)(out); \
495	movdqu x1, (1*4*4)(out); \
496	movdqu x2, (2*4*4)(out); \
497	movdqu x3, (3*4*4)(out);
498
499#define xor_blocks(out, x0, x1, x2, x3, t0, t1, t2) \
500	transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
501	\
502	movdqu (0*4*4)(out),	t0; \
503	pxor t0,		x0; \
504	movdqu x0,		(0*4*4)(out); \
505	movdqu (1*4*4)(out),	t0; \
506	pxor t0,		x1; \
507	movdqu x1,		(1*4*4)(out); \
508	movdqu (2*4*4)(out),	t0; \
509	pxor t0,		x2; \
510	movdqu x2,		(2*4*4)(out); \
511	movdqu (3*4*4)(out),	t0; \
512	pxor t0,		x3; \
513	movdqu x3,		(3*4*4)(out);
514
515ENTRY(__serpent_enc_blk_4way)
516	/* input:
517	 *	arg_ctx(%esp): ctx, CTX
518	 *	arg_dst(%esp): dst
519	 *	arg_src(%esp): src
520	 *	arg_xor(%esp): bool, if true: xor output
521	 */
522
523	pcmpeqd RNOT, RNOT;
524
525	movl arg_ctx(%esp), CTX;
526
527	movl arg_src(%esp), %eax;
528	read_blocks(%eax, RA, RB, RC, RD, RT0, RT1, RE);
529
530					 K(RA, RB, RC, RD, RE, 0);
531	S0(RA, RB, RC, RD, RE);		LK(RC, RB, RD, RA, RE, 1);
532	S1(RC, RB, RD, RA, RE);		LK(RE, RD, RA, RC, RB, 2);
533	S2(RE, RD, RA, RC, RB);		LK(RB, RD, RE, RC, RA, 3);
534	S3(RB, RD, RE, RC, RA);		LK(RC, RA, RD, RB, RE, 4);
535	S4(RC, RA, RD, RB, RE);		LK(RA, RD, RB, RE, RC, 5);
536	S5(RA, RD, RB, RE, RC);		LK(RC, RA, RD, RE, RB, 6);
537	S6(RC, RA, RD, RE, RB);		LK(RD, RB, RA, RE, RC, 7);
538	S7(RD, RB, RA, RE, RC);		LK(RC, RA, RE, RD, RB, 8);
539	S0(RC, RA, RE, RD, RB);		LK(RE, RA, RD, RC, RB, 9);
540	S1(RE, RA, RD, RC, RB);		LK(RB, RD, RC, RE, RA, 10);
541	S2(RB, RD, RC, RE, RA);		LK(RA, RD, RB, RE, RC, 11);
542	S3(RA, RD, RB, RE, RC);		LK(RE, RC, RD, RA, RB, 12);
543	S4(RE, RC, RD, RA, RB);		LK(RC, RD, RA, RB, RE, 13);
544	S5(RC, RD, RA, RB, RE);		LK(RE, RC, RD, RB, RA, 14);
545	S6(RE, RC, RD, RB, RA);		LK(RD, RA, RC, RB, RE, 15);
546	S7(RD, RA, RC, RB, RE);		LK(RE, RC, RB, RD, RA, 16);
547	S0(RE, RC, RB, RD, RA);		LK(RB, RC, RD, RE, RA, 17);
548	S1(RB, RC, RD, RE, RA);		LK(RA, RD, RE, RB, RC, 18);
549	S2(RA, RD, RE, RB, RC);		LK(RC, RD, RA, RB, RE, 19);
550	S3(RC, RD, RA, RB, RE);		LK(RB, RE, RD, RC, RA, 20);
551	S4(RB, RE, RD, RC, RA);		LK(RE, RD, RC, RA, RB, 21);
552	S5(RE, RD, RC, RA, RB);		LK(RB, RE, RD, RA, RC, 22);
553	S6(RB, RE, RD, RA, RC);		LK(RD, RC, RE, RA, RB, 23);
554	S7(RD, RC, RE, RA, RB);		LK(RB, RE, RA, RD, RC, 24);
555	S0(RB, RE, RA, RD, RC);		LK(RA, RE, RD, RB, RC, 25);
556	S1(RA, RE, RD, RB, RC);		LK(RC, RD, RB, RA, RE, 26);
557	S2(RC, RD, RB, RA, RE);		LK(RE, RD, RC, RA, RB, 27);
558	S3(RE, RD, RC, RA, RB);		LK(RA, RB, RD, RE, RC, 28);
559	S4(RA, RB, RD, RE, RC);		LK(RB, RD, RE, RC, RA, 29);
560	S5(RB, RD, RE, RC, RA);		LK(RA, RB, RD, RC, RE, 30);
561	S6(RA, RB, RD, RC, RE);		LK(RD, RE, RB, RC, RA, 31);
562	S7(RD, RE, RB, RC, RA);		 K(RA, RB, RC, RD, RE, 32);
563
564	movl arg_dst(%esp), %eax;
565
566	cmpb $0, arg_xor(%esp);
567	jnz .L__enc_xor4;
568
569	write_blocks(%eax, RA, RB, RC, RD, RT0, RT1, RE);
570
571	ret;
572
573.L__enc_xor4:
574	xor_blocks(%eax, RA, RB, RC, RD, RT0, RT1, RE);
575
576	ret;
577ENDPROC(__serpent_enc_blk_4way)
578
579ENTRY(serpent_dec_blk_4way)
580	/* input:
581	 *	arg_ctx(%esp): ctx, CTX
582	 *	arg_dst(%esp): dst
583	 *	arg_src(%esp): src
584	 */
585
586	pcmpeqd RNOT, RNOT;
587
588	movl arg_ctx(%esp), CTX;
589
590	movl arg_src(%esp), %eax;
591	read_blocks(%eax, RA, RB, RC, RD, RT0, RT1, RE);
592
593					 K(RA, RB, RC, RD, RE, 32);
594	SI7(RA, RB, RC, RD, RE);	KL(RB, RD, RA, RE, RC, 31);
595	SI6(RB, RD, RA, RE, RC);	KL(RA, RC, RE, RB, RD, 30);
596	SI5(RA, RC, RE, RB, RD);	KL(RC, RD, RA, RE, RB, 29);
597	SI4(RC, RD, RA, RE, RB);	KL(RC, RA, RB, RE, RD, 28);
598	SI3(RC, RA, RB, RE, RD);	KL(RB, RC, RD, RE, RA, 27);
599	SI2(RB, RC, RD, RE, RA);	KL(RC, RA, RE, RD, RB, 26);
600	SI1(RC, RA, RE, RD, RB);	KL(RB, RA, RE, RD, RC, 25);
601	SI0(RB, RA, RE, RD, RC);	KL(RE, RC, RA, RB, RD, 24);
602	SI7(RE, RC, RA, RB, RD);	KL(RC, RB, RE, RD, RA, 23);
603	SI6(RC, RB, RE, RD, RA);	KL(RE, RA, RD, RC, RB, 22);
604	SI5(RE, RA, RD, RC, RB);	KL(RA, RB, RE, RD, RC, 21);
605	SI4(RA, RB, RE, RD, RC);	KL(RA, RE, RC, RD, RB, 20);
606	SI3(RA, RE, RC, RD, RB);	KL(RC, RA, RB, RD, RE, 19);
607	SI2(RC, RA, RB, RD, RE);	KL(RA, RE, RD, RB, RC, 18);
608	SI1(RA, RE, RD, RB, RC);	KL(RC, RE, RD, RB, RA, 17);
609	SI0(RC, RE, RD, RB, RA);	KL(RD, RA, RE, RC, RB, 16);
610	SI7(RD, RA, RE, RC, RB);	KL(RA, RC, RD, RB, RE, 15);
611	SI6(RA, RC, RD, RB, RE);	KL(RD, RE, RB, RA, RC, 14);
612	SI5(RD, RE, RB, RA, RC);	KL(RE, RC, RD, RB, RA, 13);
613	SI4(RE, RC, RD, RB, RA);	KL(RE, RD, RA, RB, RC, 12);
614	SI3(RE, RD, RA, RB, RC);	KL(RA, RE, RC, RB, RD, 11);
615	SI2(RA, RE, RC, RB, RD);	KL(RE, RD, RB, RC, RA, 10);
616	SI1(RE, RD, RB, RC, RA);	KL(RA, RD, RB, RC, RE, 9);
617	SI0(RA, RD, RB, RC, RE);	KL(RB, RE, RD, RA, RC, 8);
618	SI7(RB, RE, RD, RA, RC);	KL(RE, RA, RB, RC, RD, 7);
619	SI6(RE, RA, RB, RC, RD);	KL(RB, RD, RC, RE, RA, 6);
620	SI5(RB, RD, RC, RE, RA);	KL(RD, RA, RB, RC, RE, 5);
621	SI4(RD, RA, RB, RC, RE);	KL(RD, RB, RE, RC, RA, 4);
622	SI3(RD, RB, RE, RC, RA);	KL(RE, RD, RA, RC, RB, 3);
623	SI2(RE, RD, RA, RC, RB);	KL(RD, RB, RC, RA, RE, 2);
624	SI1(RD, RB, RC, RA, RE);	KL(RE, RB, RC, RA, RD, 1);
625	SI0(RE, RB, RC, RA, RD);	 K(RC, RD, RB, RE, RA, 0);
626
627	movl arg_dst(%esp), %eax;
628	write_blocks(%eax, RC, RD, RB, RE, RT0, RT1, RA);
629
630	ret;
631ENDPROC(serpent_dec_blk_4way)
632