aboutsummaryrefslogtreecommitdiffstats
path: root/crypto/bn/asm/x86_64-gcc.c
blob: 31476abebfb79b892cb5f5d86b0ffdf47811993a (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
#include "../bn_lcl.h"
#if !(defined(__GNUC__) && __GNUC__>=2)
# include "../bn_asm.c"	/* kind of dirty hack for Sun Studio */
#else
/*
 * x86_64 BIGNUM accelerator version 0.1, December 2002.
 *
 * Implemented by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL
 * project.
 *
 * Rights for redistribution and usage in source and binary forms are
 * granted according to the OpenSSL license. Warranty of any kind is
 * disclaimed.
 *
 * Q. Version 0.1? It doesn't sound like Andy, he used to assign real
 *    versions, like 1.0...
 * A. Well, that's because this code is basically a quick-n-dirty
 *    proof-of-concept hack. As you can see it's implemented with
 *    inline assembler, which means that you're bound to GCC and that
 *    there might be enough room for further improvement.
 *
 * Q. Why inline assembler?
 * A. x86_64 features own ABI which I'm not familiar with. This is
 *    why I decided to let the compiler take care of subroutine
 *    prologue/epilogue as well as register allocation. For reference.
 *    Win64 implements different ABI for AMD64, different from Linux.
 *
 * Q. How much faster does it get?
 * A. 'apps/openssl speed rsa dsa' output with no-asm:
 *
 *	                  sign    verify    sign/s verify/s
 *	rsa  512 bits   0.0006s   0.0001s   1683.8  18456.2
 *	rsa 1024 bits   0.0028s   0.0002s    356.0   6407.0
 *	rsa 2048 bits   0.0172s   0.0005s     58.0   1957.8
 *	rsa 4096 bits   0.1155s   0.0018s      8.7    555.6
 *	                  sign    verify    sign/s verify/s
 *	dsa  512 bits   0.0005s   0.0006s   2100.8   1768.3
 *	dsa 1024 bits   0.0014s   0.0018s    692.3    559.2
 *	dsa 2048 bits   0.0049s   0.0061s    204.7    165.0
 *
 *    'apps/openssl speed rsa dsa' output with this module:
 *
 *	                  sign    verify    sign/s verify/s
 *	rsa  512 bits   0.0004s   0.0000s   2767.1  33297.9
 *	rsa 1024 bits   0.0012s   0.0001s    867.4  14674.7
 *	rsa 2048 bits   0.0061s   0.0002s    164.0   5270.0
 *	rsa 4096 bits   0.0384s   0.0006s     26.1   1650.8
 *	                  sign    verify    sign/s verify/s
 *	dsa  512 bits   0.0002s   0.0003s   4442.2   3786.3
 *	dsa 1024 bits   0.0005s   0.0007s   1835.1   1497.4
 *	dsa 2048 bits   0.0016s   0.0020s    620.4    504.6
 *
 *    For the reference. IA-32 assembler implementation performs
 *    very much like 64-bit code compiled with no-asm on the same
 *    machine.
 */

#ifdef _WIN64
#define BN_ULONG unsigned long long
#else
#define BN_ULONG unsigned long
#endif

#undef mul
#undef mul_add
#undef sqr

/*
 * "m"(a), "+m"(r)	is the way to favor DirectPath -code;
 * "g"(0)		let the compiler to decide where does it
 *			want to keep the value of zero;
 */
#define mul_add(r,a,word,carry) do {	\
	register BN_ULONG high,low;	\
	asm ("mulq %3"			\
		: "=a"(low),"=d"(high)	\
		: "a"(word),"m"(a)	\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+r"(carry),"+d"(high)\
		: "a"(low),"g"(0)	\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+m"(r),"+d"(high)	\
		: "r"(carry),"g"(0)	\
		: "cc");		\
	carry=high;			\
	} while (0)

#define mul(r,a,word,carry) do {	\
	register BN_ULONG high,low;	\
	asm ("mulq %3"			\
		: "=a"(low),"=d"(high)	\
		: "a"(word),"g"(a)	\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+r"(carry),"+d"(high)\
		: "a"(low),"g"(0)	\
		: "cc");		\
	(r)=carry, carry=high;		\
	} while (0)

#define sqr(r0,r1,a)			\
	asm ("mulq %2"			\
		: "=a"(r0),"=d"(r1)	\
		: "a"(a)		\
		: "cc");

BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
	{
	BN_ULONG c1=0;

	if (num <= 0) return(c1);

	while (num&~3)
		{
		mul_add(rp[0],ap[0],w,c1);
		mul_add(rp[1],ap[1],w,c1);
		mul_add(rp[2],ap[2],w,c1);
		mul_add(rp[3],ap[3],w,c1);
		ap+=4; rp+=4; num-=4;
		}
	if (num)
		{
		mul_add(rp[0],ap[0],w,c1); if (--num==0) return c1;
		mul_add(rp[1],ap[1],w,c1); if (--num==0) return c1;
		mul_add(rp[2],ap[2],w,c1); return c1;
		}
	
	return(c1);
	} 

BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
	{
	BN_ULONG c1=0;

	if (num <= 0) return(c1);

	while (num&~3)
		{
		mul(rp[0],ap[0],w,c1);
		mul(rp[1],ap[1],w,c1);
		mul(rp[2],ap[2],w,c1);
		mul(rp[3],ap[3],w,c1);
		ap+=4; rp+=4; num-=4;
		}
	if (num)
		{
		mul(rp[0],ap[0],w,c1); if (--num == 0) return c1;
		mul(rp[1],ap[1],w,c1); if (--num == 0) return c1;
		mul(rp[2],ap[2],w,c1);
		}
	return(c1);
	} 

void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n)
        {
	if (n <= 0) return;

	while (n&~3)
		{
		sqr(r[0],r[1],a[0]);
		sqr(r[2],r[3],a[1]);
		sqr(r[4],r[5],a[2]);
		sqr(r[6],r[7],a[3]);
		a+=4; r+=8; n-=4;
		}
	if (n)
		{
		sqr(r[0],r[1],a[0]); if (--n == 0) return;
		sqr(r[2],r[3],a[1]); if (--n == 0) return;
		sqr(r[4],r[5],a[2]);
		}
	}

BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
{	BN_ULONG ret,waste;

	asm ("divq	%4"
		: "=a"(ret),"=d"(waste)
		: "a"(l),"d"(h),"g"(d)
		: "cc");

	return ret;
}

BN_ULONG bn_add_words (BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int n)
{ BN_ULONG ret=0,i=0;

	if (n <= 0) return 0;

	asm volatile (
	"	subq	%2,%2		\n"
	".p2align 4			\n"
	"1:	movq	(%4,%2,8),%0	\n"
	"	adcq	(%5,%2,8),%0	\n"
	"	movq	%0,(%3,%2,8)	\n"
	"	leaq	1(%2),%2	\n"
	"	loop	1b		\n"
	"	sbbq	%0,%0		\n"
		: "=&a"(ret),"+c"(n),"=&r"(i)
		: "r"(rp),"r"(ap),"r"(bp)
		: "cc", "memory"
	);

  return ret&1;
}

#ifndef SIMICS
BN_ULONG bn_sub_words (BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int n)
{ BN_ULONG ret=0,i=0;

	if (n <= 0) return 0;

	asm volatile (
	"	subq	%2,%2		\n"
	".p2align 4			\n"
	"1:	movq	(%4,%2,8),%0	\n"
	"	sbbq	(%5,%2,8),%0	\n"
	"	movq	%0,(%3,%2,8)	\n"
	"	leaq	1(%2),%2	\n"
	"	loop	1b		\n"
	"	sbbq	%0,%0		\n"
		: "=&a"(ret),"+c"(n),"=&r"(i)
		: "r"(rp),"r"(ap),"r"(bp)
		: "cc", "memory"
	);

  return ret&1;
}
#else
/* Simics 1.4<7 has buggy sbbq:-( */
#define BN_MASK2 0xffffffffffffffffL
BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n)
        {
	BN_ULONG t1,t2;
	int c=0;

	if (n <= 0) return((BN_ULONG)0);

	for (;;)
		{
		t1=a[0]; t2=b[0];
		r[0]=(t1-t2-c)&BN_MASK2;
		if (t1 != t2) c=(t1 < t2);
		if (--n <= 0) break;

		t1=a[1]; t2=b[1];
		r[1]=(t1-t2-c)&BN_MASK2;
		if (t1 != t2) c=(t1 < t2);
		if (--n <= 0) break;

		t1=a[2]; t2=b[2];
		r[2]=(t1-t2-c)&BN_MASK2;
		if (t1 != t2) c=(t1 < t2);
		if (--n <= 0) break;

		t1=a[3]; t2=b[3];
		r[3]=(t1-t2-c)&BN_MASK2;
		if (t1 != t2) c=(t1 < t2);
		if (--n <= 0) break;

		a+=4;
		b+=4;
		r+=4;
		}
	return(c);
	}
#endif

/* mul_add_c(a,b,c0,c1,c2)  -- c+=a*b for three word number c=(c2,c1,c0) */
/* mul_add_c2(a,b,c0,c1,c2) -- c+=2*a*b for three word number c=(c2,c1,c0) */
/* sqr_add_c(a,i,c0,c1,c2)  -- c+=a[i]^2 for three word number c=(c2,c1,c0) */
/* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */

#if 0
/* original macros are kept for reference purposes */
#define mul_add_c(a,b,c0,c1,c2) {	\
	BN_ULONG ta=(a),tb=(b);		\
	t1 = ta * tb;			\
	t2 = BN_UMULT_HIGH(ta,tb);	\
	c0 += t1; t2 += (c0<t1)?1:0;	\
	c1 += t2; c2 += (c1<t2)?1:0;	\
	}

#define mul_add_c2(a,b,c0,c1,c2) {	\
	BN_ULONG ta=(a),tb=(b),t0;	\
	t1 = BN_UMULT_HIGH(ta,tb);	\
	t0 = ta * tb;			\
	t2 = t1+t1; c2 += (t2<t1)?1:0;	\
	t1 = t0+t0; t2 += (t1<t0)?1:0;	\
	c0 += t1; t2 += (c0<t1)?1:0;	\
	c1 += t2; c2 += (c1<t2)?1:0;	\
	}
#else
#define mul_add_c(a,b,c0,c1,c2)	do {	\
	asm ("mulq %3"			\
		: "=a"(t1),"=d"(t2)	\
		: "a"(a),"m"(b)		\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+r"(c0),"+d"(t2)	\
		: "a"(t1),"g"(0)	\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+r"(c1),"+r"(c2)	\
		: "d"(t2),"g"(0)	\
		: "cc");		\
	} while (0)

#define sqr_add_c(a,i,c0,c1,c2)	do {	\
	asm ("mulq %2"			\
		: "=a"(t1),"=d"(t2)	\
		: "a"(a[i])		\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+r"(c0),"+d"(t2)	\
		: "a"(t1),"g"(0)	\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+r"(c1),"+r"(c2)	\
		: "d"(t2),"g"(0)	\
		: "cc");		\
	} while (0)

#define mul_add_c2(a,b,c0,c1,c2) do {	\
	asm ("mulq %3"			\
		: "=a"(t1),"=d"(t2)	\
		: "a"(a),"m"(b)		\
		: "cc");		\
	asm ("addq %0,%0; adcq %2,%1"	\
		: "+d"(t2),"+r"(c2)	\
		: "g"(0)		\
		: "cc");		\
	asm ("addq %0,%0; adcq %2,%1"	\
		: "+a"(t1),"+d"(t2)	\
		: "g"(0)		\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+r"(c0),"+d"(t2)	\
		: "a"(t1),"g"(0)	\
		: "cc");		\
	asm ("addq %2,%0; adcq %3,%1"	\
		: "+r"(c1),"+r"(c2)	\
		: "d"(t2),"g"(0)	\
		: "cc");		\
	} while (0)
#endif

#define sqr_add_c2(a,i,j,c0,c1,c2)	\
	mul_add_c2((a)[i],(a)[j],c0,c1,c2)

void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
	{
	BN_ULONG t1,t2;
	BN_ULONG c1,c2,c3;

	c1=0;
	c2=0;
	c3=0;
	mul_add_c(a[0],b[0],c1,c2,c3);
	r[0]=c1;
	c1=0;
	mul_add_c(a[0],b[1],c2,c3,c1);
	mul_add_c(a[1],b[0],c2,c3,c1);
	r[1]=c2;
	c2=0;
	mul_add_c(a[2],b[0],c3,c1,c2);
	mul_add_c(a[1],b[1],c3,c1,c2);
	mul_add_c(a[0],b[2],c3,c1,c2);
	r[2]=c3;
	c3=0;
	mul_add_c(a[0],b[3],c1,c2,c3);
	mul_add_c(a[1],b[2],c1,c2,c3);
	mul_add_c(a[2],b[1],c1,c2,c3);
	mul_add_c(a[3],b[0],c1,c2,c3);
	r[3]=c1;
	c1=0;
	mul_add_c(a[4],b[0],c2,c3,c1);
	mul_add_c(a[3],b[1],c2,c3,c1);
	mul_add_c(a[2],b[2],c2,c3,c1);
	mul_add_c(a[1],b[3],c2,c3,c1);
	mul_add_c(a[0],b[4],c2,c3,c1);
	r[4]=c2;
	c2=0;
	mul_add_c(a[0],b[5],c3,c1,c2);
	mul_add_c(a[1],b[4],c3,c1,c2);
	mul_add_c(a[2],b[3],c3,c1,c2);
	mul_add_c(a[3],b[2],c3,c1,c2);
	mul_add_c(a[4],b[1],c3,c1,c2);
	mul_add_c(a[5],b[0],c3,c1,c2);
	r[5]=c3;
	c3=0;
	mul_add_c(a[6],b[0],c1,c2,c3);
	mul_add_c(a[5],b[1],c1,c2,c3);
	mul_add_c(a[4],b[2],c1,c2,c3);
	mul_add_c(a[3],b[3],c1,c2,c3);
	mul_add_c(a[2],b[4],c1,c2,c3);
	mul_add_c(a[1],b[5],c1,c2,c3);
	mul_add_c(a[0],b[6],c1,c2,c3);
	r[6]=c1;
	c1=0;
	mul_add_c(a[0],b[7],c2,c3,c1);
	mul_add_c(a[1],b[6],c2,c3,c1);
	mul_add_c(a[2],b[5],c2,c3,c1);
	mul_add_c(a[3],b[4],c2,c3,c1);
	mul_add_c(a[4],b[3],c2,c3,c1);
	mul_add_c(a[5],b[2],c2,c3,c1);
	mul_add_c(a[6],b[1],c2,c3,c1);
	mul_add_c(a[7],b[0],c2,c3,c1);
	r[7]=c2;
	c2=0;
	mul_add_c(a[7],b[1],c3,c1,c2);
	mul_add_c(a[6],b[2],c3,c1,c2);
	mul_add_c(a[5],b[3],c3,c1,c2);
	mul_add_c(a[4],b[4],c3,c1,c2);
	mul_add_c(a[3],b[5],c3,c1,c2);
	mul_add_c(a[2],b[6],c3,c1,c2);
	mul_add_c(a[1],b[7],c3,c1,c2);
	r[8]=c3;
	c3=0;
	mul_add_c(a[2],b[7],c1,c2,c3);
	mul_add_c(a[3],b[6],c1,c2,c3);
	mul_add_c(a[4],b[5],c1,c2,c3);
	mul_add_c(a[5],b[4],c1,c2,c3);
	mul_add_c(a[6],b[3],c1,c2,c3);
	mul_add_c(a[7],b[2],c1,c2,c3);
	r[9]=c1;
	c1=0;
	mul_add_c(a[7],b[3],c2,c3,c1);
	mul_add_c(a[6],b[4],c2,c3,c1);
	mul_add_c(a[5],b[5],c2,c3,c1);
	mul_add_c(a[4],b[6],c2,c3,c1);
	mul_add_c(a[3],b[7],c2,c3,c1);
	r[10]=c2;
	c2=0;
	mul_add_c(a[4],b[7],c3,c1,c2);
	mul_add_c(a[5],b[6],c3,c1,c2);
	mul_add_c(a[6],b[5],c3,c1,c2);
	mul_add_c(a[7],b[4],c3,c1,c2);
	r[11]=c3;
	c3=0;
	mul_add_c(a[7],b[5],c1,c2,c3);
	mul_add_c(a[6],b[6],c1,c2,c3);
	mul_add_c(a[5],b[7],c1,c2,c3);
	r[12]=c1;
	c1=0;
	mul_add_c(a[6],b[7],c2,c3,c1);
	mul_add_c(a[7],b[6],c2,c3,c1);
	r[13]=c2;
	c2=0;
	mul_add_c(a[7],b[7],c3,c1,c2);
	r[14]=c3;
	r[15]=c1;
	}

void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
	{
	BN_ULONG t1,t2;
	BN_ULONG c1,c2,c3;

	c1=0;
	c2=0;
	c3=0;
	mul_add_c(a[0],b[0],c1,c2,c3);
	r[0]=c1;
	c1=0;
	mul_add_c(a[0],b[1],c2,c3,c1);
	mul_add_c(a[1],b[0],c2,c3,c1);
	r[1]=c2;
	c2=0;
	mul_add_c(a[2],b[0],c3,c1,c2);
	mul_add_c(a[1],b[1],c3,c1,c2);
	mul_add_c(a[0],b[2],c3,c1,c2);
	r[2]=c3;
	c3=0;
	mul_add_c(a[0],b[3],c1,c2,c3);
	mul_add_c(a[1],b[2],c1,c2,c3);
	mul_add_c(a[2],b[1],c1,c2,c3);
	mul_add_c(a[3],b[0],c1,c2,c3);
	r[3]=c1;
	c1=0;
	mul_add_c(a[3],b[1],c2,c3,c1);
	mul_add_c(a[2],b[2],c2,c3,c1);
	mul_add_c(a[1],b[3],c2,c3,c1);
	r[4]=c2;
	c2=0;
	mul_add_c(a[2],b[3],c3,c1,c2);
	mul_add_c(a[3],b[2],c3,c1,c2);
	r[5]=c3;
	c3=0;
	mul_add_c(a[3],b[3],c1,c2,c3);
	r[6]=c1;
	r[7]=c2;
	}

void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a)
	{
	BN_ULONG t1,t2;
	BN_ULONG c1,c2,c3;

	c1=0;
	c2=0;
	c3=0;
	sqr_add_c(a,0,c1,c2,c3);
	r[0]=c1;
	c1=0;
	sqr_add_c2(a,1,0,c2,c3,c1);
	r[1]=c2;
	c2=0;
	sqr_add_c(a,1,c3,c1,c2);
	sqr_add_c2(a,2,0,c3,c1,c2);
	r[2]=c3;
	c3=0;
	sqr_add_c2(a,3,0,c1,c2,c3);
	sqr_add_c2(a,2,1,c1,c2,c3);
	r[3]=c1;
	c1=0;
	sqr_add_c(a,2,c2,c3,c1);
	sqr_add_c2(a,3,1,c2,c3,c1);
	sqr_add_c2(a,4,0,c2,c3,c1);
	r[4]=c2;
	c2=0;
	sqr_add_c2(a,5,0,c3,c1,c2);
	sqr_add_c2(a,4,1,c3,c1,c2);
	sqr_add_c2(a,3,2,c3,c1,c2);
	r[5]=c3;
	c3=0;
	sqr_add_c(a,3,c1,c2,c3);
	sqr_add_c2(a,4,2,c1,c2,c3);
	sqr_add_c2(a,5,1,c1,c2,c3);
	sqr_add_c2(a,6,0,c1,c2,c3);
	r[6]=c1;
	c1=0;
	sqr_add_c2(a,7,0,c2,c3,c1);
	sqr_add_c2(a,6,1,c2,c3,c1);
	sqr_add_c2(a,5,2,c2,c3,c1);
	sqr_add_c2(a,4,3,c2,c3,c1);
	r[7]=c2;
	c2=0;
	sqr_add_c(a,4,c3,c1,c2);
	sqr_add_c2(a,5,3,c3,c1,c2);
	sqr_add_c2(a,6,2,c3,c1,c2);
	sqr_add_c2(a,7,1,c3,c1,c2);
	r[8]=c3;
	c3=0;
	sqr_add_c2(a,7,2,c1,c2,c3);
	sqr_add_c2(a,6,3,c1,c2,c3);
	sqr_add_c2(a,5,4,c1,c2,c3);
	r[9]=c1;
	c1=0;
	sqr_add_c(a,5,c2,c3,c1);
	sqr_add_c2(a,6,4,c2,c3,c1);
	sqr_add_c2(a,7,3,c2,c3,c1);
	r[10]=c2;
	c2=0;
	sqr_add_c2(a,7,4,c3,c1,c2);
	sqr_add_c2(a,6,5,c3,c1,c2);
	r[11]=c3;
	c3=0;
	sqr_add_c(a,6,c1,c2,c3);
	sqr_add_c2(a,7,5,c1,c2,c3);
	r[12]=c1;
	c1=0;
	sqr_add_c2(a,7,6,c2,c3,c1);
	r[13]=c2;
	c2=0;
	sqr_add_c(a,7,c3,c1,c2);
	r[14]=c3;
	r[15]=c1;
	}

void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a)
	{
	BN_ULONG t1,t2;
	BN_ULONG c1,c2,c3;

	c1=0;
	c2=0;
	c3=0;
	sqr_add_c(a,0,c1,c2,c3);
	r[0]=c1;
	c1=0;
	sqr_add_c2(a,1,0,c2,c3,c1);
	r[1]=c2;
	c2=0;
	sqr_add_c(a,1,c3,c1,c2);
	sqr_add_c2(a,2,0,c3,c1,c2);
	r[2]=c3;
	c3=0;
	sqr_add_c2(a,3,0,c1,c2,c3);
	sqr_add_c2(a,2,1,c1,c2,c3);
	r[3]=c1;
	c1=0;
	sqr_add_c(a,2,c2,c3,c1);
	sqr_add_c2(a,3,1,c2,c3,c1);
	r[4]=c2;
	c2=0;
	sqr_add_c2(a,3,2,c3,c1,c2);
	r[5]=c3;
	c3=0;
	sqr_add_c(a,3,c1,c2,c3);
	r[6]=c1;
	r[7]=c2;
	}
#endif