1 /* crypto/bn/bn_asm.c */
2 /* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
5 * This package is an SSL implementation written
6 * by Eric Young (eay@cryptsoft.com).
7 * The implementation was written so as to conform with Netscapes SSL.
9 * This library is free for commercial and non-commercial use as long as
10 * the following conditions are aheared to. The following conditions
11 * apply to all code found in this distribution, be it the RC4, RSA,
12 * lhash, DES, etc., code; not just the SSL code. The SSL documentation
13 * included with this distribution is covered by the same copyright terms
14 * except that the holder is Tim Hudson (tjh@cryptsoft.com).
16 * Copyright remains Eric Young's, and as such any Copyright notices in
17 * the code are not to be removed.
18 * If this package is used in a product, Eric Young should be given attribution
19 * as the author of the parts of the library used.
20 * This can be in the form of a textual message at program startup or
21 * in documentation (online or textual) provided with the package.
23 * Redistribution and use in source and binary forms, with or without
24 * modification, are permitted provided that the following conditions
26 * 1. Redistributions of source code must retain the copyright
27 * notice, this list of conditions and the following disclaimer.
28 * 2. Redistributions in binary form must reproduce the above copyright
29 * notice, this list of conditions and the following disclaimer in the
30 * documentation and/or other materials provided with the distribution.
31 * 3. All advertising materials mentioning features or use of this software
32 * must display the following acknowledgement:
33 * "This product includes cryptographic software written by
34 * Eric Young (eay@cryptsoft.com)"
35 * The word 'cryptographic' can be left out if the rouines from the library
36 * being used are not cryptographic related :-).
37 * 4. If you include any Windows specific code (or a derivative thereof) from
38 * the apps directory (application code) you must include an acknowledgement:
39 * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
41 * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
42 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
43 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
44 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
45 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
46 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
47 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
48 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
49 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
50 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
53 * The licence and distribution terms for any publically available version or
54 * derivative of this code cannot be changed. i.e. this code cannot simply be
55 * copied and put under another distribution licence
56 * [including the GNU Public Licence.]
65 BN_ULONG bn_mul_add_words(rp,ap,num,w)
73 if (num <= 0) return(c1);
77 mul_add(rp[0],ap[0],w,c1);
78 if (--num == 0) break;
79 mul_add(rp[1],ap[1],w,c1);
80 if (--num == 0) break;
81 mul_add(rp[2],ap[2],w,c1);
82 if (--num == 0) break;
83 mul_add(rp[3],ap[3],w,c1);
84 if (--num == 0) break;
92 BN_ULONG bn_mul_words(rp,ap,num,w)
100 if (num <= 0) return(c1);
104 mul(rp[0],ap[0],w,c1);
105 if (--num == 0) break;
106 mul(rp[1],ap[1],w,c1);
107 if (--num == 0) break;
108 mul(rp[2],ap[2],w,c1);
109 if (--num == 0) break;
110 mul(rp[3],ap[3],w,c1);
111 if (--num == 0) break;
118 void bn_sqr_words(r,a,n)
128 t=(BN_ULLONG)(a[0])*(a[0]);
129 r[0]=Lw(t); r[1]=Hw(t);
132 t=(BN_ULLONG)(a[1])*(a[1]);
133 r[2]=Lw(t); r[3]=Hw(t);
136 t=(BN_ULLONG)(a[2])*(a[2]);
137 r[4]=Lw(t); r[5]=Hw(t);
140 t=(BN_ULLONG)(a[3])*(a[3]);
141 r[6]=Lw(t); r[7]=Hw(t);
151 BN_ULONG bn_mul_add_words(rp,ap,num,w)
160 if (num <= 0) return((BN_ULONG)0);
167 mul_add(rp[0],ap[0],bl,bh,c);
168 if (--num == 0) break;
169 mul_add(rp[1],ap[1],bl,bh,c);
170 if (--num == 0) break;
171 mul_add(rp[2],ap[2],bl,bh,c);
172 if (--num == 0) break;
173 mul_add(rp[3],ap[3],bl,bh,c);
174 if (--num == 0) break;
181 BN_ULONG bn_mul_words(rp,ap,num,w)
190 if (num <= 0) return((BN_ULONG)0);
197 mul(rp[0],ap[0],bl,bh,carry);
198 if (--num == 0) break;
199 mul(rp[1],ap[1],bl,bh,carry);
200 if (--num == 0) break;
201 mul(rp[2],ap[2],bl,bh,carry);
202 if (--num == 0) break;
203 mul(rp[3],ap[3],bl,bh,carry);
204 if (--num == 0) break;
211 void bn_sqr_words(r,a,n)
219 sqr64(r[0],r[1],a[0]);
222 sqr64(r[2],r[3],a[1]);
225 sqr64(r[4],r[5],a[2]);
228 sqr64(r[6],r[7],a[3]);
238 #if defined(BN_LLONG) && defined(BN_DIV2W)
240 BN_ULONG bn_div_words(h,l,d)
243 return((BN_ULONG)(((((BN_ULLONG)h)<<BN_BITS2)|l)/(BN_ULLONG)d));
248 /* Divide h-l by d and return the result. */
249 /* I need to test this some more :-( */
250 BN_ULONG bn_div_words(h,l,d)
253 BN_ULONG dh,dl,q,ret=0,th,tl,t;
256 if (d == 0) return(BN_MASK2);
258 i=BN_num_bits_word(d);
259 if ((i != BN_BITS2) && (h > (BN_ULONG)1<<i))
261 #if !defined(NO_STDIO) && !defined(WIN16)
262 fprintf(stderr,"Division would overflow (%d)\n",i);
272 h=(h<<i)|(l>>(BN_BITS2-i));
275 dh=(d&BN_MASK2h)>>BN_BITS4;
279 if ((h>>BN_BITS4) == dh)
290 ((l&BN_MASK2h)>>BN_BITS4))))
297 tl=(tl<<BN_BITS4)&BN_MASK2h;
309 if (--count == 0) break;
312 h=((h<<BN_BITS4)|(l>>BN_BITS4))&BN_MASK2;
313 l=(l&BN_MASK2l)<<BN_BITS4;
321 BN_ULONG bn_add_words(r,a,b,n)
328 if (n <= 0) return((BN_ULONG)0);
332 ll+=(BN_ULLONG)a[0]+b[0];
333 r[0]=(BN_ULONG)ll&BN_MASK2;
337 ll+=(BN_ULLONG)a[1]+b[1];
338 r[1]=(BN_ULONG)ll&BN_MASK2;
342 ll+=(BN_ULLONG)a[2]+b[2];
343 r[2]=(BN_ULONG)ll&BN_MASK2;
347 ll+=(BN_ULLONG)a[3]+b[3];
348 r[3]=(BN_ULONG)ll&BN_MASK2;
356 return((BN_ULONG)ll);
359 BN_ULONG bn_add_words(r,a,b,n)
366 if (n <= 0) return((BN_ULONG)0);
411 BN_ULONG bn_sub_words(r,a,b,n)
419 if (n <= 0) return((BN_ULONG)0);
424 r[0]=(t1-t2-c)&BN_MASK2;
425 if (t1 != t2) c=(t1 < t2);
429 r[1]=(t1-t2-c)&BN_MASK2;
430 if (t1 != t2) c=(t1 < t2);
434 r[2]=(t1-t2-c)&BN_MASK2;
435 if (t1 != t2) c=(t1 < t2);
439 r[3]=(t1-t2-c)&BN_MASK2;
440 if (t1 != t2) c=(t1 < t2);
458 #define mul_add_c(a,b,c0,c1,c2) \
460 t1=(BN_ULONG)Lw(t); \
461 t2=(BN_ULONG)Hw(t); \
462 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
463 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
465 #define mul_add_c2(a,b,c0,c1,c2) \
469 t1=(BN_ULONG)Lw(tt); \
470 t2=(BN_ULONG)Hw(tt); \
471 c0=(c0+t1)&BN_MASK2; \
472 if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
473 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
475 #define sqr_add_c(a,i,c0,c1,c2) \
476 t=(BN_ULLONG)a[i]*a[i]; \
477 t1=(BN_ULONG)Lw(t); \
478 t2=(BN_ULONG)Hw(t); \
479 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
480 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
482 #define sqr_add_c2(a,i,j,c0,c1,c2) \
483 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
485 #define mul_add_c(a,b,c0,c1,c2) \
486 t1=LBITS(a); t2=HBITS(a); \
487 bl=LBITS(b); bh=HBITS(b); \
488 mul64(t1,t2,bl,bh); \
489 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
490 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
492 #define mul_add_c2(a,b,c0,c1,c2) \
493 t1=LBITS(a); t2=HBITS(a); \
494 bl=LBITS(b); bh=HBITS(b); \
495 mul64(t1,t2,bl,bh); \
496 if (t2 & BN_TBIT) c2++; \
497 t2=(t2+t2)&BN_MASK2; \
498 if (t1 & BN_TBIT) t2++; \
499 t1=(t1+t1)&BN_MASK2; \
500 c0=(c0+t1)&BN_MASK2; \
501 if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
502 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
504 #define sqr_add_c(a,i,c0,c1,c2) \
505 sqr64(t1,t2,(a)[i]); \
506 c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
507 c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
509 #define sqr_add_c2(a,i,j,c0,c1,c2) \
510 mul_add_c2((a)[i],(a)[j],c0,c1,c2)
513 void bn_mul_comba8(r,a,b)
527 mul_add_c(a[0],b[0],c1,c2,c3);
530 mul_add_c(a[0],b[1],c2,c3,c1);
531 mul_add_c(a[1],b[0],c2,c3,c1);
534 mul_add_c(a[2],b[0],c3,c1,c2);
535 mul_add_c(a[1],b[1],c3,c1,c2);
536 mul_add_c(a[0],b[2],c3,c1,c2);
539 mul_add_c(a[0],b[3],c1,c2,c3);
540 mul_add_c(a[1],b[2],c1,c2,c3);
541 mul_add_c(a[2],b[1],c1,c2,c3);
542 mul_add_c(a[3],b[0],c1,c2,c3);
545 mul_add_c(a[4],b[0],c2,c3,c1);
546 mul_add_c(a[3],b[1],c2,c3,c1);
547 mul_add_c(a[2],b[2],c2,c3,c1);
548 mul_add_c(a[1],b[3],c2,c3,c1);
549 mul_add_c(a[0],b[4],c2,c3,c1);
552 mul_add_c(a[0],b[5],c3,c1,c2);
553 mul_add_c(a[1],b[4],c3,c1,c2);
554 mul_add_c(a[2],b[3],c3,c1,c2);
555 mul_add_c(a[3],b[2],c3,c1,c2);
556 mul_add_c(a[4],b[1],c3,c1,c2);
557 mul_add_c(a[5],b[0],c3,c1,c2);
560 mul_add_c(a[6],b[0],c1,c2,c3);
561 mul_add_c(a[5],b[1],c1,c2,c3);
562 mul_add_c(a[4],b[2],c1,c2,c3);
563 mul_add_c(a[3],b[3],c1,c2,c3);
564 mul_add_c(a[2],b[4],c1,c2,c3);
565 mul_add_c(a[1],b[5],c1,c2,c3);
566 mul_add_c(a[0],b[6],c1,c2,c3);
569 mul_add_c(a[0],b[7],c2,c3,c1);
570 mul_add_c(a[1],b[6],c2,c3,c1);
571 mul_add_c(a[2],b[5],c2,c3,c1);
572 mul_add_c(a[3],b[4],c2,c3,c1);
573 mul_add_c(a[4],b[3],c2,c3,c1);
574 mul_add_c(a[5],b[2],c2,c3,c1);
575 mul_add_c(a[6],b[1],c2,c3,c1);
576 mul_add_c(a[7],b[0],c2,c3,c1);
579 mul_add_c(a[7],b[1],c3,c1,c2);
580 mul_add_c(a[6],b[2],c3,c1,c2);
581 mul_add_c(a[5],b[3],c3,c1,c2);
582 mul_add_c(a[4],b[4],c3,c1,c2);
583 mul_add_c(a[3],b[5],c3,c1,c2);
584 mul_add_c(a[2],b[6],c3,c1,c2);
585 mul_add_c(a[1],b[7],c3,c1,c2);
588 mul_add_c(a[2],b[7],c1,c2,c3);
589 mul_add_c(a[3],b[6],c1,c2,c3);
590 mul_add_c(a[4],b[5],c1,c2,c3);
591 mul_add_c(a[5],b[4],c1,c2,c3);
592 mul_add_c(a[6],b[3],c1,c2,c3);
593 mul_add_c(a[7],b[2],c1,c2,c3);
596 mul_add_c(a[7],b[3],c2,c3,c1);
597 mul_add_c(a[6],b[4],c2,c3,c1);
598 mul_add_c(a[5],b[5],c2,c3,c1);
599 mul_add_c(a[4],b[6],c2,c3,c1);
600 mul_add_c(a[3],b[7],c2,c3,c1);
603 mul_add_c(a[4],b[7],c3,c1,c2);
604 mul_add_c(a[5],b[6],c3,c1,c2);
605 mul_add_c(a[6],b[5],c3,c1,c2);
606 mul_add_c(a[7],b[4],c3,c1,c2);
609 mul_add_c(a[7],b[5],c1,c2,c3);
610 mul_add_c(a[6],b[6],c1,c2,c3);
611 mul_add_c(a[5],b[7],c1,c2,c3);
614 mul_add_c(a[6],b[7],c2,c3,c1);
615 mul_add_c(a[7],b[6],c2,c3,c1);
618 mul_add_c(a[7],b[7],c3,c1,c2);
623 void bn_mul_comba4(r,a,b)
637 mul_add_c(a[0],b[0],c1,c2,c3);
640 mul_add_c(a[0],b[1],c2,c3,c1);
641 mul_add_c(a[1],b[0],c2,c3,c1);
644 mul_add_c(a[2],b[0],c3,c1,c2);
645 mul_add_c(a[1],b[1],c3,c1,c2);
646 mul_add_c(a[0],b[2],c3,c1,c2);
649 mul_add_c(a[0],b[3],c1,c2,c3);
650 mul_add_c(a[1],b[2],c1,c2,c3);
651 mul_add_c(a[2],b[1],c1,c2,c3);
652 mul_add_c(a[3],b[0],c1,c2,c3);
655 mul_add_c(a[3],b[1],c2,c3,c1);
656 mul_add_c(a[2],b[2],c2,c3,c1);
657 mul_add_c(a[1],b[3],c2,c3,c1);
660 mul_add_c(a[2],b[3],c3,c1,c2);
661 mul_add_c(a[3],b[2],c3,c1,c2);
664 mul_add_c(a[3],b[3],c1,c2,c3);
669 void bn_sqr_comba8(r,a)
683 sqr_add_c(a,0,c1,c2,c3);
686 sqr_add_c2(a,1,0,c2,c3,c1);
689 sqr_add_c(a,1,c3,c1,c2);
690 sqr_add_c2(a,2,0,c3,c1,c2);
693 sqr_add_c2(a,3,0,c1,c2,c3);
694 sqr_add_c2(a,2,1,c1,c2,c3);
697 sqr_add_c(a,2,c2,c3,c1);
698 sqr_add_c2(a,3,1,c2,c3,c1);
699 sqr_add_c2(a,4,0,c2,c3,c1);
702 sqr_add_c2(a,5,0,c3,c1,c2);
703 sqr_add_c2(a,4,1,c3,c1,c2);
704 sqr_add_c2(a,3,2,c3,c1,c2);
707 sqr_add_c(a,3,c1,c2,c3);
708 sqr_add_c2(a,4,2,c1,c2,c3);
709 sqr_add_c2(a,5,1,c1,c2,c3);
710 sqr_add_c2(a,6,0,c1,c2,c3);
713 sqr_add_c2(a,7,0,c2,c3,c1);
714 sqr_add_c2(a,6,1,c2,c3,c1);
715 sqr_add_c2(a,5,2,c2,c3,c1);
716 sqr_add_c2(a,4,3,c2,c3,c1);
719 sqr_add_c(a,4,c3,c1,c2);
720 sqr_add_c2(a,5,3,c3,c1,c2);
721 sqr_add_c2(a,6,2,c3,c1,c2);
722 sqr_add_c2(a,7,1,c3,c1,c2);
725 sqr_add_c2(a,7,2,c1,c2,c3);
726 sqr_add_c2(a,6,3,c1,c2,c3);
727 sqr_add_c2(a,5,4,c1,c2,c3);
730 sqr_add_c(a,5,c2,c3,c1);
731 sqr_add_c2(a,6,4,c2,c3,c1);
732 sqr_add_c2(a,7,3,c2,c3,c1);
735 sqr_add_c2(a,7,4,c3,c1,c2);
736 sqr_add_c2(a,6,5,c3,c1,c2);
739 sqr_add_c(a,6,c1,c2,c3);
740 sqr_add_c2(a,7,5,c1,c2,c3);
743 sqr_add_c2(a,7,6,c2,c3,c1);
746 sqr_add_c(a,7,c3,c1,c2);
751 void bn_sqr_comba4(r,a)
765 sqr_add_c(a,0,c1,c2,c3);
768 sqr_add_c2(a,1,0,c2,c3,c1);
771 sqr_add_c(a,1,c3,c1,c2);
772 sqr_add_c2(a,2,0,c3,c1,c2);
775 sqr_add_c2(a,3,0,c1,c2,c3);
776 sqr_add_c2(a,2,1,c1,c2,c3);
779 sqr_add_c(a,2,c2,c3,c1);
780 sqr_add_c2(a,3,1,c2,c3,c1);
783 sqr_add_c2(a,3,2,c3,c1,c2);
786 sqr_add_c(a,3,c1,c2,c3);
792 /* hmm... is it faster just to do a multiply? */
794 void bn_sqr_comba4(r,a)
798 bn_sqr_normal(r,a,4,t);
802 void bn_sqr_comba8(r,a)
806 bn_sqr_normal(r,a,8,t);
809 void bn_mul_comba4(r,a,b)
812 r[4]=bn_mul_words( &(r[0]),a,4,b[0]);
813 r[5]=bn_mul_add_words(&(r[1]),a,4,b[1]);
814 r[6]=bn_mul_add_words(&(r[2]),a,4,b[2]);
815 r[7]=bn_mul_add_words(&(r[3]),a,4,b[3]);
818 void bn_mul_comba8(r,a,b)
821 r[ 8]=bn_mul_words( &(r[0]),a,8,b[0]);
822 r[ 9]=bn_mul_add_words(&(r[1]),a,8,b[1]);
823 r[10]=bn_mul_add_words(&(r[2]),a,8,b[2]);
824 r[11]=bn_mul_add_words(&(r[3]),a,8,b[3]);
825 r[12]=bn_mul_add_words(&(r[4]),a,8,b[4]);
826 r[13]=bn_mul_add_words(&(r[5]),a,8,b[5]);
827 r[14]=bn_mul_add_words(&(r[6]),a,8,b[6]);
828 r[15]=bn_mul_add_words(&(r[7]),a,8,b[7]);
831 #endif /* BN_COMBA */