sl@0: /* crypto/bn/bn_asm.c */
sl@0: /* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
sl@0:  * All rights reserved.
sl@0:  *
sl@0:  * This package is an SSL implementation written
sl@0:  * by Eric Young (eay@cryptsoft.com).
sl@0:  * The implementation was written so as to conform with Netscapes SSL.
sl@0:  * 
sl@0:  * This library is free for commercial and non-commercial use as long as
sl@0:  * the following conditions are aheared to.  The following conditions
sl@0:  * apply to all code found in this distribution, be it the RC4, RSA,
sl@0:  * lhash, DES, etc., code; not just the SSL code.  The SSL documentation
sl@0:  * included with this distribution is covered by the same copyright terms
sl@0:  * except that the holder is Tim Hudson (tjh@cryptsoft.com).
sl@0:  * 
sl@0:  * Copyright remains Eric Young's, and as such any Copyright notices in
sl@0:  * the code are not to be removed.
sl@0:  * If this package is used in a product, Eric Young should be given attribution
sl@0:  * as the author of the parts of the library used.
sl@0:  * This can be in the form of a textual message at program startup or
sl@0:  * in documentation (online or textual) provided with the package.
sl@0:  * 
sl@0:  * Redistribution and use in source and binary forms, with or without
sl@0:  * modification, are permitted provided that the following conditions
sl@0:  * are met:
sl@0:  * 1. Redistributions of source code must retain the copyright
sl@0:  *    notice, this list of conditions and the following disclaimer.
sl@0:  * 2. Redistributions in binary form must reproduce the above copyright
sl@0:  *    notice, this list of conditions and the following disclaimer in the
sl@0:  *    documentation and/or other materials provided with the distribution.
sl@0:  * 3. All advertising materials mentioning features or use of this software
sl@0:  *    must display the following acknowledgement:
sl@0:  *    "This product includes cryptographic software written by
sl@0:  *     Eric Young (eay@cryptsoft.com)"
sl@0:  *    The word 'cryptographic' can be left out if the rouines from the library
sl@0:  *    being used are not cryptographic related :-).
sl@0:  * 4. If you include any Windows specific code (or a derivative thereof) from 
sl@0:  *    the apps directory (application code) you must include an acknowledgement:
sl@0:  *    "This product includes software written by Tim Hudson (tjh@cryptsoft.com)"
sl@0:  * 
sl@0:  * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
sl@0:  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
sl@0:  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
sl@0:  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
sl@0:  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
sl@0:  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
sl@0:  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
sl@0:  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
sl@0:  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
sl@0:  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
sl@0:  * SUCH DAMAGE.
sl@0:  * 
sl@0:  * The licence and distribution terms for any publically available version or
sl@0:  * derivative of this code cannot be changed.  i.e. this code cannot simply be
sl@0:  * copied and put under another distribution licence
sl@0:  * [including the GNU Public Licence.]
sl@0:  */
sl@0: 
sl@0: #ifndef BN_DEBUG
sl@0: # undef NDEBUG /* avoid conflicting definitions */
sl@0: # define NDEBUG
sl@0: #endif
sl@0: 
sl@0: #include <stdio.h>
sl@0: #include <assert.h>
sl@0: #include "cryptlib.h"
sl@0: #include "bn_lcl.h"
sl@0: 
sl@0: #if defined(BN_LLONG) || defined(BN_UMULT_HIGH)
sl@0: 
sl@0: EXPORT_C BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
sl@0: 	{
sl@0: 	BN_ULONG c1=0;
sl@0: 
sl@0: 	assert(num >= 0);
sl@0: 	if (num <= 0) return(c1);
sl@0: 
sl@0: 	while (num&~3)
sl@0: 		{
sl@0: 		mul_add(rp[0],ap[0],w,c1);
sl@0: 		mul_add(rp[1],ap[1],w,c1);
sl@0: 		mul_add(rp[2],ap[2],w,c1);
sl@0: 		mul_add(rp[3],ap[3],w,c1);
sl@0: 		ap+=4; rp+=4; num-=4;
sl@0: 		}
sl@0: 	if (num)
sl@0: 		{
sl@0: 		mul_add(rp[0],ap[0],w,c1); if (--num==0) return c1;
sl@0: 		mul_add(rp[1],ap[1],w,c1); if (--num==0) return c1;
sl@0: 		mul_add(rp[2],ap[2],w,c1); return c1;
sl@0: 		}
sl@0: 	
sl@0: 	return(c1);
sl@0: 	} 
sl@0: 
sl@0: EXPORT_C BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
sl@0: 	{
sl@0: 	BN_ULONG c1=0;
sl@0: 
sl@0: 	assert(num >= 0);
sl@0: 	if (num <= 0) return(c1);
sl@0: 
sl@0: 	while (num&~3)
sl@0: 		{
sl@0: 		mul(rp[0],ap[0],w,c1);
sl@0: 		mul(rp[1],ap[1],w,c1);
sl@0: 		mul(rp[2],ap[2],w,c1);
sl@0: 		mul(rp[3],ap[3],w,c1);
sl@0: 		ap+=4; rp+=4; num-=4;
sl@0: 		}
sl@0: 	if (num)
sl@0: 		{
sl@0: 		mul(rp[0],ap[0],w,c1); if (--num == 0) return c1;
sl@0: 		mul(rp[1],ap[1],w,c1); if (--num == 0) return c1;
sl@0: 		mul(rp[2],ap[2],w,c1);
sl@0: 		}
sl@0: 	return(c1);
sl@0: 	} 
sl@0: 
sl@0: EXPORT_C void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n)
sl@0:         {
sl@0: 	assert(n >= 0);
sl@0: 	if (n <= 0) return;
sl@0: 	while (n&~3)
sl@0: 		{
sl@0: 		sqr(r[0],r[1],a[0]);
sl@0: 		sqr(r[2],r[3],a[1]);
sl@0: 		sqr(r[4],r[5],a[2]);
sl@0: 		sqr(r[6],r[7],a[3]);
sl@0: 		a+=4; r+=8; n-=4;
sl@0: 		}
sl@0: 	if (n)
sl@0: 		{
sl@0: 		sqr(r[0],r[1],a[0]); if (--n == 0) return;
sl@0: 		sqr(r[2],r[3],a[1]); if (--n == 0) return;
sl@0: 		sqr(r[4],r[5],a[2]);
sl@0: 		}
sl@0: 	}
sl@0: 
sl@0: #else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
sl@0: 
sl@0: EXPORT_C BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
sl@0: 	{
sl@0: 	BN_ULONG c=0;
sl@0: 	BN_ULONG bl,bh;
sl@0: 
sl@0: 	assert(num >= 0);
sl@0: 	if (num <= 0) return((BN_ULONG)0);
sl@0: 
sl@0: 	bl=LBITS(w);
sl@0: 	bh=HBITS(w);
sl@0: 
sl@0: 	for (;;)
sl@0: 		{
sl@0: 		mul_add(rp[0],ap[0],bl,bh,c);
sl@0: 		if (--num == 0) break;
sl@0: 		mul_add(rp[1],ap[1],bl,bh,c);
sl@0: 		if (--num == 0) break;
sl@0: 		mul_add(rp[2],ap[2],bl,bh,c);
sl@0: 		if (--num == 0) break;
sl@0: 		mul_add(rp[3],ap[3],bl,bh,c);
sl@0: 		if (--num == 0) break;
sl@0: 		ap+=4;
sl@0: 		rp+=4;
sl@0: 		}
sl@0: 	return(c);
sl@0: 	} 
sl@0: 
sl@0: EXPORT_C BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
sl@0: 	{
sl@0: 	BN_ULONG carry=0;
sl@0: 	BN_ULONG bl,bh;
sl@0: 
sl@0: 	assert(num >= 0);
sl@0: 	if (num <= 0) return((BN_ULONG)0);
sl@0: 
sl@0: 	bl=LBITS(w);
sl@0: 	bh=HBITS(w);
sl@0: 
sl@0: 	for (;;)
sl@0: 		{
sl@0: 		mul(rp[0],ap[0],bl,bh,carry);
sl@0: 		if (--num == 0) break;
sl@0: 		mul(rp[1],ap[1],bl,bh,carry);
sl@0: 		if (--num == 0) break;
sl@0: 		mul(rp[2],ap[2],bl,bh,carry);
sl@0: 		if (--num == 0) break;
sl@0: 		mul(rp[3],ap[3],bl,bh,carry);
sl@0: 		if (--num == 0) break;
sl@0: 		ap+=4;
sl@0: 		rp+=4;
sl@0: 		}
sl@0: 	return(carry);
sl@0: 	} 
sl@0: 
sl@0: EXPORT_C void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n)
sl@0:         {
sl@0: 	assert(n >= 0);
sl@0: 	if (n <= 0) return;
sl@0: 	for (;;)
sl@0: 		{
sl@0: 		sqr64(r[0],r[1],a[0]);
sl@0: 		if (--n == 0) break;
sl@0: 
sl@0: 		sqr64(r[2],r[3],a[1]);
sl@0: 		if (--n == 0) break;
sl@0: 
sl@0: 		sqr64(r[4],r[5],a[2]);
sl@0: 		if (--n == 0) break;
sl@0: 
sl@0: 		sqr64(r[6],r[7],a[3]);
sl@0: 		if (--n == 0) break;
sl@0: 
sl@0: 		a+=4;
sl@0: 		r+=8;
sl@0: 		}
sl@0: 	}
sl@0: 
sl@0: #endif /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
sl@0: 
sl@0: #if defined(BN_LLONG) && defined(BN_DIV2W)
sl@0: 
sl@0: EXPORT_C BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
sl@0: 	{
sl@0: 	return((BN_ULONG)(((((BN_ULLONG)h)<<BN_BITS2)|l)/(BN_ULLONG)d));
sl@0: 	}
sl@0: 
sl@0: #else
sl@0: 
sl@0: /* Divide h,l by d and return the result. */
sl@0: /* I need to test this some more :-( */
sl@0: EXPORT_C BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
sl@0: 	{
sl@0: 	BN_ULONG dh,dl,q,ret=0,th,tl,t;
sl@0: 	int i,count=2;
sl@0: 
sl@0: 	if (d == 0) return(BN_MASK2);
sl@0: 
sl@0: 	i=BN_num_bits_word(d);
sl@0: 	assert((i == BN_BITS2) || (h <= (BN_ULONG)1<<i));
sl@0: 
sl@0: 	i=BN_BITS2-i;
sl@0: 	if (h >= d) h-=d;
sl@0: 
sl@0: 	if (i)
sl@0: 		{
sl@0: 		d<<=i;
sl@0: 		h=(h<<i)|(l>>(BN_BITS2-i));
sl@0: 		l<<=i;
sl@0: 		}
sl@0: 	dh=(d&BN_MASK2h)>>BN_BITS4;
sl@0: 	dl=(d&BN_MASK2l);
sl@0: 	for (;;)
sl@0: 		{
sl@0: 		if ((h>>BN_BITS4) == dh)
sl@0: 			q=BN_MASK2l;
sl@0: 		else
sl@0: 			q=h/dh;
sl@0: 
sl@0: 		th=q*dh;
sl@0: 		tl=dl*q;
sl@0: 		for (;;)
sl@0: 			{
sl@0: 			t=h-th;
sl@0: 			if ((t&BN_MASK2h) ||
sl@0: 				((tl) <= (
sl@0: 					(t<<BN_BITS4)|
sl@0: 					((l&BN_MASK2h)>>BN_BITS4))))
sl@0: 				break;
sl@0: 			q--;
sl@0: 			th-=dh;
sl@0: 			tl-=dl;
sl@0: 			}
sl@0: 		t=(tl>>BN_BITS4);
sl@0: 		tl=(tl<<BN_BITS4)&BN_MASK2h;
sl@0: 		th+=t;
sl@0: 
sl@0: 		if (l < tl) th++;
sl@0: 		l-=tl;
sl@0: 		if (h < th)
sl@0: 			{
sl@0: 			h+=d;
sl@0: 			q--;
sl@0: 			}
sl@0: 		h-=th;
sl@0: 
sl@0: 		if (--count == 0) break;
sl@0: 
sl@0: 		ret=q<<BN_BITS4;
sl@0: 		h=((h<<BN_BITS4)|(l>>BN_BITS4))&BN_MASK2;
sl@0: 		l=(l&BN_MASK2l)<<BN_BITS4;
sl@0: 		}
sl@0: 	ret|=q;
sl@0: 	return(ret);
sl@0: 	}
sl@0: #endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */
sl@0: 
sl@0: #ifdef BN_LLONG
sl@0: EXPORT_C BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
sl@0:         {
sl@0: 	BN_ULLONG ll=0;
sl@0: 
sl@0: 	assert(n >= 0);
sl@0: 	if (n <= 0) return((BN_ULONG)0);
sl@0: 
sl@0: 	for (;;)
sl@0: 		{
sl@0: 		ll+=(BN_ULLONG)a[0]+b[0];
sl@0: 		r[0]=(BN_ULONG)ll&BN_MASK2;
sl@0: 		ll>>=BN_BITS2;
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		ll+=(BN_ULLONG)a[1]+b[1];
sl@0: 		r[1]=(BN_ULONG)ll&BN_MASK2;
sl@0: 		ll>>=BN_BITS2;
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		ll+=(BN_ULLONG)a[2]+b[2];
sl@0: 		r[2]=(BN_ULONG)ll&BN_MASK2;
sl@0: 		ll>>=BN_BITS2;
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		ll+=(BN_ULLONG)a[3]+b[3];
sl@0: 		r[3]=(BN_ULONG)ll&BN_MASK2;
sl@0: 		ll>>=BN_BITS2;
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		a+=4;
sl@0: 		b+=4;
sl@0: 		r+=4;
sl@0: 		}
sl@0: 	return((BN_ULONG)ll);
sl@0: 	}
sl@0: #else /* !BN_LLONG */
sl@0: EXPORT_C BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
sl@0:         {
sl@0: 	BN_ULONG c,l,t;
sl@0: 
sl@0: 	assert(n >= 0);
sl@0: 	if (n <= 0) return((BN_ULONG)0);
sl@0: 
sl@0: 	c=0;
sl@0: 	for (;;)
sl@0: 		{
sl@0: 		t=a[0];
sl@0: 		t=(t+c)&BN_MASK2;
sl@0: 		c=(t < c);
sl@0: 		l=(t+b[0])&BN_MASK2;
sl@0: 		c+=(l < t);
sl@0: 		r[0]=l;
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		t=a[1];
sl@0: 		t=(t+c)&BN_MASK2;
sl@0: 		c=(t < c);
sl@0: 		l=(t+b[1])&BN_MASK2;
sl@0: 		c+=(l < t);
sl@0: 		r[1]=l;
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		t=a[2];
sl@0: 		t=(t+c)&BN_MASK2;
sl@0: 		c=(t < c);
sl@0: 		l=(t+b[2])&BN_MASK2;
sl@0: 		c+=(l < t);
sl@0: 		r[2]=l;
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		t=a[3];
sl@0: 		t=(t+c)&BN_MASK2;
sl@0: 		c=(t < c);
sl@0: 		l=(t+b[3])&BN_MASK2;
sl@0: 		c+=(l < t);
sl@0: 		r[3]=l;
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		a+=4;
sl@0: 		b+=4;
sl@0: 		r+=4;
sl@0: 		}
sl@0: 	return((BN_ULONG)c);
sl@0: 	}
sl@0: #endif /* !BN_LLONG */
sl@0: 
sl@0: EXPORT_C BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
sl@0:         {
sl@0: 	BN_ULONG t1,t2;
sl@0: 	int c=0;
sl@0: 
sl@0: 	assert(n >= 0);
sl@0: 	if (n <= 0) return((BN_ULONG)0);
sl@0: 
sl@0: 	for (;;)
sl@0: 		{
sl@0: 		t1=a[0]; t2=b[0];
sl@0: 		r[0]=(t1-t2-c)&BN_MASK2;
sl@0: 		if (t1 != t2) c=(t1 < t2);
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		t1=a[1]; t2=b[1];
sl@0: 		r[1]=(t1-t2-c)&BN_MASK2;
sl@0: 		if (t1 != t2) c=(t1 < t2);
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		t1=a[2]; t2=b[2];
sl@0: 		r[2]=(t1-t2-c)&BN_MASK2;
sl@0: 		if (t1 != t2) c=(t1 < t2);
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		t1=a[3]; t2=b[3];
sl@0: 		r[3]=(t1-t2-c)&BN_MASK2;
sl@0: 		if (t1 != t2) c=(t1 < t2);
sl@0: 		if (--n <= 0) break;
sl@0: 
sl@0: 		a+=4;
sl@0: 		b+=4;
sl@0: 		r+=4;
sl@0: 		}
sl@0: 	return(c);
sl@0: 	}
sl@0: 
sl@0: #ifdef BN_MUL_COMBA
sl@0: 
sl@0: #undef bn_mul_comba8
sl@0: #undef bn_mul_comba4
sl@0: #undef bn_sqr_comba8
sl@0: #undef bn_sqr_comba4
sl@0: 
sl@0: /* mul_add_c(a,b,c0,c1,c2)  -- c+=a*b for three word number c=(c2,c1,c0) */
sl@0: /* mul_add_c2(a,b,c0,c1,c2) -- c+=2*a*b for three word number c=(c2,c1,c0) */
sl@0: /* sqr_add_c(a,i,c0,c1,c2)  -- c+=a[i]^2 for three word number c=(c2,c1,c0) */
sl@0: /* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */
sl@0: 
sl@0: #ifdef BN_LLONG
sl@0: #define mul_add_c(a,b,c0,c1,c2) \
sl@0: 	t=(BN_ULLONG)a*b; \
sl@0: 	t1=(BN_ULONG)Lw(t); \
sl@0: 	t2=(BN_ULONG)Hw(t); \
sl@0: 	c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
sl@0: 	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
sl@0: 
sl@0: #define mul_add_c2(a,b,c0,c1,c2) \
sl@0: 	t=(BN_ULLONG)a*b; \
sl@0: 	tt=(t+t)&BN_MASK; \
sl@0: 	if (tt < t) c2++; \
sl@0: 	t1=(BN_ULONG)Lw(tt); \
sl@0: 	t2=(BN_ULONG)Hw(tt); \
sl@0: 	c0=(c0+t1)&BN_MASK2;  \
sl@0: 	if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
sl@0: 	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
sl@0: 
sl@0: #define sqr_add_c(a,i,c0,c1,c2) \
sl@0: 	t=(BN_ULLONG)a[i]*a[i]; \
sl@0: 	t1=(BN_ULONG)Lw(t); \
sl@0: 	t2=(BN_ULONG)Hw(t); \
sl@0: 	c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
sl@0: 	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
sl@0: 
sl@0: #define sqr_add_c2(a,i,j,c0,c1,c2) \
sl@0: 	mul_add_c2((a)[i],(a)[j],c0,c1,c2)
sl@0: 
sl@0: #elif defined(BN_UMULT_LOHI)
sl@0: 
sl@0: #define mul_add_c(a,b,c0,c1,c2)	{	\
sl@0: 	BN_ULONG ta=(a),tb=(b);		\
sl@0: 	BN_UMULT_LOHI(t1,t2,ta,tb);	\
sl@0: 	c0 += t1; t2 += (c0<t1)?1:0;	\
sl@0: 	c1 += t2; c2 += (c1<t2)?1:0;	\
sl@0: 	}
sl@0: 
sl@0: #define mul_add_c2(a,b,c0,c1,c2) {	\
sl@0: 	BN_ULONG ta=(a),tb=(b),t0;	\
sl@0: 	BN_UMULT_LOHI(t0,t1,ta,tb);	\
sl@0: 	t2 = t1+t1; c2 += (t2<t1)?1:0;	\
sl@0: 	t1 = t0+t0; t2 += (t1<t0)?1:0;	\
sl@0: 	c0 += t1; t2 += (c0<t1)?1:0;	\
sl@0: 	c1 += t2; c2 += (c1<t2)?1:0;	\
sl@0: 	}
sl@0: 
sl@0: #define sqr_add_c(a,i,c0,c1,c2)	{	\
sl@0: 	BN_ULONG ta=(a)[i];		\
sl@0: 	BN_UMULT_LOHI(t1,t2,ta,ta);	\
sl@0: 	c0 += t1; t2 += (c0<t1)?1:0;	\
sl@0: 	c1 += t2; c2 += (c1<t2)?1:0;	\
sl@0: 	}
sl@0: 
sl@0: #define sqr_add_c2(a,i,j,c0,c1,c2)	\
sl@0: 	mul_add_c2((a)[i],(a)[j],c0,c1,c2)
sl@0: 
sl@0: #elif defined(BN_UMULT_HIGH)
sl@0: 
sl@0: #define mul_add_c(a,b,c0,c1,c2)	{	\
sl@0: 	BN_ULONG ta=(a),tb=(b);		\
sl@0: 	t1 = ta * tb;			\
sl@0: 	t2 = BN_UMULT_HIGH(ta,tb);	\
sl@0: 	c0 += t1; t2 += (c0<t1)?1:0;	\
sl@0: 	c1 += t2; c2 += (c1<t2)?1:0;	\
sl@0: 	}
sl@0: 
sl@0: #define mul_add_c2(a,b,c0,c1,c2) {	\
sl@0: 	BN_ULONG ta=(a),tb=(b),t0;	\
sl@0: 	t1 = BN_UMULT_HIGH(ta,tb);	\
sl@0: 	t0 = ta * tb;			\
sl@0: 	t2 = t1+t1; c2 += (t2<t1)?1:0;	\
sl@0: 	t1 = t0+t0; t2 += (t1<t0)?1:0;	\
sl@0: 	c0 += t1; t2 += (c0<t1)?1:0;	\
sl@0: 	c1 += t2; c2 += (c1<t2)?1:0;	\
sl@0: 	}
sl@0: 
sl@0: #define sqr_add_c(a,i,c0,c1,c2)	{	\
sl@0: 	BN_ULONG ta=(a)[i];		\
sl@0: 	t1 = ta * ta;			\
sl@0: 	t2 = BN_UMULT_HIGH(ta,ta);	\
sl@0: 	c0 += t1; t2 += (c0<t1)?1:0;	\
sl@0: 	c1 += t2; c2 += (c1<t2)?1:0;	\
sl@0: 	}
sl@0: 
sl@0: #define sqr_add_c2(a,i,j,c0,c1,c2)	\
sl@0: 	mul_add_c2((a)[i],(a)[j],c0,c1,c2)
sl@0: 
sl@0: #else /* !BN_LLONG */
sl@0: #define mul_add_c(a,b,c0,c1,c2) \
sl@0: 	t1=LBITS(a); t2=HBITS(a); \
sl@0: 	bl=LBITS(b); bh=HBITS(b); \
sl@0: 	mul64(t1,t2,bl,bh); \
sl@0: 	c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
sl@0: 	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
sl@0: 
sl@0: #define mul_add_c2(a,b,c0,c1,c2) \
sl@0: 	t1=LBITS(a); t2=HBITS(a); \
sl@0: 	bl=LBITS(b); bh=HBITS(b); \
sl@0: 	mul64(t1,t2,bl,bh); \
sl@0: 	if (t2 & BN_TBIT) c2++; \
sl@0: 	t2=(t2+t2)&BN_MASK2; \
sl@0: 	if (t1 & BN_TBIT) t2++; \
sl@0: 	t1=(t1+t1)&BN_MASK2; \
sl@0: 	c0=(c0+t1)&BN_MASK2;  \
sl@0: 	if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
sl@0: 	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
sl@0: 
sl@0: #define sqr_add_c(a,i,c0,c1,c2) \
sl@0: 	sqr64(t1,t2,(a)[i]); \
sl@0: 	c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
sl@0: 	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;
sl@0: 
sl@0: #define sqr_add_c2(a,i,j,c0,c1,c2) \
sl@0: 	mul_add_c2((a)[i],(a)[j],c0,c1,c2)
sl@0: #endif /* !BN_LLONG */
sl@0: 
sl@0: EXPORT_C void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
sl@0: 	{
sl@0: #ifdef BN_LLONG
sl@0: 	BN_ULLONG t;
sl@0: #else
sl@0: 	BN_ULONG bl,bh;
sl@0: #endif
sl@0: 	BN_ULONG t1,t2;
sl@0: 	BN_ULONG c1,c2,c3;
sl@0: 
sl@0: 	c1=0;
sl@0: 	c2=0;
sl@0: 	c3=0;
sl@0: 	mul_add_c(a[0],b[0],c1,c2,c3);
sl@0: 	r[0]=c1;
sl@0: 	c1=0;
sl@0: 	mul_add_c(a[0],b[1],c2,c3,c1);
sl@0: 	mul_add_c(a[1],b[0],c2,c3,c1);
sl@0: 	r[1]=c2;
sl@0: 	c2=0;
sl@0: 	mul_add_c(a[2],b[0],c3,c1,c2);
sl@0: 	mul_add_c(a[1],b[1],c3,c1,c2);
sl@0: 	mul_add_c(a[0],b[2],c3,c1,c2);
sl@0: 	r[2]=c3;
sl@0: 	c3=0;
sl@0: 	mul_add_c(a[0],b[3],c1,c2,c3);
sl@0: 	mul_add_c(a[1],b[2],c1,c2,c3);
sl@0: 	mul_add_c(a[2],b[1],c1,c2,c3);
sl@0: 	mul_add_c(a[3],b[0],c1,c2,c3);
sl@0: 	r[3]=c1;
sl@0: 	c1=0;
sl@0: 	mul_add_c(a[4],b[0],c2,c3,c1);
sl@0: 	mul_add_c(a[3],b[1],c2,c3,c1);
sl@0: 	mul_add_c(a[2],b[2],c2,c3,c1);
sl@0: 	mul_add_c(a[1],b[3],c2,c3,c1);
sl@0: 	mul_add_c(a[0],b[4],c2,c3,c1);
sl@0: 	r[4]=c2;
sl@0: 	c2=0;
sl@0: 	mul_add_c(a[0],b[5],c3,c1,c2);
sl@0: 	mul_add_c(a[1],b[4],c3,c1,c2);
sl@0: 	mul_add_c(a[2],b[3],c3,c1,c2);
sl@0: 	mul_add_c(a[3],b[2],c3,c1,c2);
sl@0: 	mul_add_c(a[4],b[1],c3,c1,c2);
sl@0: 	mul_add_c(a[5],b[0],c3,c1,c2);
sl@0: 	r[5]=c3;
sl@0: 	c3=0;
sl@0: 	mul_add_c(a[6],b[0],c1,c2,c3);
sl@0: 	mul_add_c(a[5],b[1],c1,c2,c3);
sl@0: 	mul_add_c(a[4],b[2],c1,c2,c3);
sl@0: 	mul_add_c(a[3],b[3],c1,c2,c3);
sl@0: 	mul_add_c(a[2],b[4],c1,c2,c3);
sl@0: 	mul_add_c(a[1],b[5],c1,c2,c3);
sl@0: 	mul_add_c(a[0],b[6],c1,c2,c3);
sl@0: 	r[6]=c1;
sl@0: 	c1=0;
sl@0: 	mul_add_c(a[0],b[7],c2,c3,c1);
sl@0: 	mul_add_c(a[1],b[6],c2,c3,c1);
sl@0: 	mul_add_c(a[2],b[5],c2,c3,c1);
sl@0: 	mul_add_c(a[3],b[4],c2,c3,c1);
sl@0: 	mul_add_c(a[4],b[3],c2,c3,c1);
sl@0: 	mul_add_c(a[5],b[2],c2,c3,c1);
sl@0: 	mul_add_c(a[6],b[1],c2,c3,c1);
sl@0: 	mul_add_c(a[7],b[0],c2,c3,c1);
sl@0: 	r[7]=c2;
sl@0: 	c2=0;
sl@0: 	mul_add_c(a[7],b[1],c3,c1,c2);
sl@0: 	mul_add_c(a[6],b[2],c3,c1,c2);
sl@0: 	mul_add_c(a[5],b[3],c3,c1,c2);
sl@0: 	mul_add_c(a[4],b[4],c3,c1,c2);
sl@0: 	mul_add_c(a[3],b[5],c3,c1,c2);
sl@0: 	mul_add_c(a[2],b[6],c3,c1,c2);
sl@0: 	mul_add_c(a[1],b[7],c3,c1,c2);
sl@0: 	r[8]=c3;
sl@0: 	c3=0;
sl@0: 	mul_add_c(a[2],b[7],c1,c2,c3);
sl@0: 	mul_add_c(a[3],b[6],c1,c2,c3);
sl@0: 	mul_add_c(a[4],b[5],c1,c2,c3);
sl@0: 	mul_add_c(a[5],b[4],c1,c2,c3);
sl@0: 	mul_add_c(a[6],b[3],c1,c2,c3);
sl@0: 	mul_add_c(a[7],b[2],c1,c2,c3);
sl@0: 	r[9]=c1;
sl@0: 	c1=0;
sl@0: 	mul_add_c(a[7],b[3],c2,c3,c1);
sl@0: 	mul_add_c(a[6],b[4],c2,c3,c1);
sl@0: 	mul_add_c(a[5],b[5],c2,c3,c1);
sl@0: 	mul_add_c(a[4],b[6],c2,c3,c1);
sl@0: 	mul_add_c(a[3],b[7],c2,c3,c1);
sl@0: 	r[10]=c2;
sl@0: 	c2=0;
sl@0: 	mul_add_c(a[4],b[7],c3,c1,c2);
sl@0: 	mul_add_c(a[5],b[6],c3,c1,c2);
sl@0: 	mul_add_c(a[6],b[5],c3,c1,c2);
sl@0: 	mul_add_c(a[7],b[4],c3,c1,c2);
sl@0: 	r[11]=c3;
sl@0: 	c3=0;
sl@0: 	mul_add_c(a[7],b[5],c1,c2,c3);
sl@0: 	mul_add_c(a[6],b[6],c1,c2,c3);
sl@0: 	mul_add_c(a[5],b[7],c1,c2,c3);
sl@0: 	r[12]=c1;
sl@0: 	c1=0;
sl@0: 	mul_add_c(a[6],b[7],c2,c3,c1);
sl@0: 	mul_add_c(a[7],b[6],c2,c3,c1);
sl@0: 	r[13]=c2;
sl@0: 	c2=0;
sl@0: 	mul_add_c(a[7],b[7],c3,c1,c2);
sl@0: 	r[14]=c3;
sl@0: 	r[15]=c1;
sl@0: 	}
sl@0: 
sl@0: EXPORT_C void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
sl@0: 	{
sl@0: #ifdef BN_LLONG
sl@0: 	BN_ULLONG t;
sl@0: #else
sl@0: 	BN_ULONG bl,bh;
sl@0: #endif
sl@0: 	BN_ULONG t1,t2;
sl@0: 	BN_ULONG c1,c2,c3;
sl@0: 
sl@0: 	c1=0;
sl@0: 	c2=0;
sl@0: 	c3=0;
sl@0: 	mul_add_c(a[0],b[0],c1,c2,c3);
sl@0: 	r[0]=c1;
sl@0: 	c1=0;
sl@0: 	mul_add_c(a[0],b[1],c2,c3,c1);
sl@0: 	mul_add_c(a[1],b[0],c2,c3,c1);
sl@0: 	r[1]=c2;
sl@0: 	c2=0;
sl@0: 	mul_add_c(a[2],b[0],c3,c1,c2);
sl@0: 	mul_add_c(a[1],b[1],c3,c1,c2);
sl@0: 	mul_add_c(a[0],b[2],c3,c1,c2);
sl@0: 	r[2]=c3;
sl@0: 	c3=0;
sl@0: 	mul_add_c(a[0],b[3],c1,c2,c3);
sl@0: 	mul_add_c(a[1],b[2],c1,c2,c3);
sl@0: 	mul_add_c(a[2],b[1],c1,c2,c3);
sl@0: 	mul_add_c(a[3],b[0],c1,c2,c3);
sl@0: 	r[3]=c1;
sl@0: 	c1=0;
sl@0: 	mul_add_c(a[3],b[1],c2,c3,c1);
sl@0: 	mul_add_c(a[2],b[2],c2,c3,c1);
sl@0: 	mul_add_c(a[1],b[3],c2,c3,c1);
sl@0: 	r[4]=c2;
sl@0: 	c2=0;
sl@0: 	mul_add_c(a[2],b[3],c3,c1,c2);
sl@0: 	mul_add_c(a[3],b[2],c3,c1,c2);
sl@0: 	r[5]=c3;
sl@0: 	c3=0;
sl@0: 	mul_add_c(a[3],b[3],c1,c2,c3);
sl@0: 	r[6]=c1;
sl@0: 	r[7]=c2;
sl@0: 	}
sl@0: 
sl@0: EXPORT_C void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a)
sl@0: 	{
sl@0: #ifdef BN_LLONG
sl@0: 	BN_ULLONG t,tt;
sl@0: #else
sl@0: 	BN_ULONG bl,bh;
sl@0: #endif
sl@0: 	BN_ULONG t1,t2;
sl@0: 	BN_ULONG c1,c2,c3;
sl@0: 
sl@0: 	c1=0;
sl@0: 	c2=0;
sl@0: 	c3=0;
sl@0: 	sqr_add_c(a,0,c1,c2,c3);
sl@0: 	r[0]=c1;
sl@0: 	c1=0;
sl@0: 	sqr_add_c2(a,1,0,c2,c3,c1);
sl@0: 	r[1]=c2;
sl@0: 	c2=0;
sl@0: 	sqr_add_c(a,1,c3,c1,c2);
sl@0: 	sqr_add_c2(a,2,0,c3,c1,c2);
sl@0: 	r[2]=c3;
sl@0: 	c3=0;
sl@0: 	sqr_add_c2(a,3,0,c1,c2,c3);
sl@0: 	sqr_add_c2(a,2,1,c1,c2,c3);
sl@0: 	r[3]=c1;
sl@0: 	c1=0;
sl@0: 	sqr_add_c(a,2,c2,c3,c1);
sl@0: 	sqr_add_c2(a,3,1,c2,c3,c1);
sl@0: 	sqr_add_c2(a,4,0,c2,c3,c1);
sl@0: 	r[4]=c2;
sl@0: 	c2=0;
sl@0: 	sqr_add_c2(a,5,0,c3,c1,c2);
sl@0: 	sqr_add_c2(a,4,1,c3,c1,c2);
sl@0: 	sqr_add_c2(a,3,2,c3,c1,c2);
sl@0: 	r[5]=c3;
sl@0: 	c3=0;
sl@0: 	sqr_add_c(a,3,c1,c2,c3);
sl@0: 	sqr_add_c2(a,4,2,c1,c2,c3);
sl@0: 	sqr_add_c2(a,5,1,c1,c2,c3);
sl@0: 	sqr_add_c2(a,6,0,c1,c2,c3);
sl@0: 	r[6]=c1;
sl@0: 	c1=0;
sl@0: 	sqr_add_c2(a,7,0,c2,c3,c1);
sl@0: 	sqr_add_c2(a,6,1,c2,c3,c1);
sl@0: 	sqr_add_c2(a,5,2,c2,c3,c1);
sl@0: 	sqr_add_c2(a,4,3,c2,c3,c1);
sl@0: 	r[7]=c2;
sl@0: 	c2=0;
sl@0: 	sqr_add_c(a,4,c3,c1,c2);
sl@0: 	sqr_add_c2(a,5,3,c3,c1,c2);
sl@0: 	sqr_add_c2(a,6,2,c3,c1,c2);
sl@0: 	sqr_add_c2(a,7,1,c3,c1,c2);
sl@0: 	r[8]=c3;
sl@0: 	c3=0;
sl@0: 	sqr_add_c2(a,7,2,c1,c2,c3);
sl@0: 	sqr_add_c2(a,6,3,c1,c2,c3);
sl@0: 	sqr_add_c2(a,5,4,c1,c2,c3);
sl@0: 	r[9]=c1;
sl@0: 	c1=0;
sl@0: 	sqr_add_c(a,5,c2,c3,c1);
sl@0: 	sqr_add_c2(a,6,4,c2,c3,c1);
sl@0: 	sqr_add_c2(a,7,3,c2,c3,c1);
sl@0: 	r[10]=c2;
sl@0: 	c2=0;
sl@0: 	sqr_add_c2(a,7,4,c3,c1,c2);
sl@0: 	sqr_add_c2(a,6,5,c3,c1,c2);
sl@0: 	r[11]=c3;
sl@0: 	c3=0;
sl@0: 	sqr_add_c(a,6,c1,c2,c3);
sl@0: 	sqr_add_c2(a,7,5,c1,c2,c3);
sl@0: 	r[12]=c1;
sl@0: 	c1=0;
sl@0: 	sqr_add_c2(a,7,6,c2,c3,c1);
sl@0: 	r[13]=c2;
sl@0: 	c2=0;
sl@0: 	sqr_add_c(a,7,c3,c1,c2);
sl@0: 	r[14]=c3;
sl@0: 	r[15]=c1;
sl@0: 	}
sl@0: 
sl@0: EXPORT_C void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a)
sl@0: 	{
sl@0: #ifdef BN_LLONG
sl@0: 	BN_ULLONG t,tt;
sl@0: #else
sl@0: 	BN_ULONG bl,bh;
sl@0: #endif
sl@0: 	BN_ULONG t1,t2;
sl@0: 	BN_ULONG c1,c2,c3;
sl@0: 
sl@0: 	c1=0;
sl@0: 	c2=0;
sl@0: 	c3=0;
sl@0: 	sqr_add_c(a,0,c1,c2,c3);
sl@0: 	r[0]=c1;
sl@0: 	c1=0;
sl@0: 	sqr_add_c2(a,1,0,c2,c3,c1);
sl@0: 	r[1]=c2;
sl@0: 	c2=0;
sl@0: 	sqr_add_c(a,1,c3,c1,c2);
sl@0: 	sqr_add_c2(a,2,0,c3,c1,c2);
sl@0: 	r[2]=c3;
sl@0: 	c3=0;
sl@0: 	sqr_add_c2(a,3,0,c1,c2,c3);
sl@0: 	sqr_add_c2(a,2,1,c1,c2,c3);
sl@0: 	r[3]=c1;
sl@0: 	c1=0;
sl@0: 	sqr_add_c(a,2,c2,c3,c1);
sl@0: 	sqr_add_c2(a,3,1,c2,c3,c1);
sl@0: 	r[4]=c2;
sl@0: 	c2=0;
sl@0: 	sqr_add_c2(a,3,2,c3,c1,c2);
sl@0: 	r[5]=c3;
sl@0: 	c3=0;
sl@0: 	sqr_add_c(a,3,c1,c2,c3);
sl@0: 	r[6]=c1;
sl@0: 	r[7]=c2;
sl@0: 	}
sl@0: #else /* !BN_MUL_COMBA */
sl@0: 
sl@0: /* hmm... is it faster just to do a multiply? */
sl@0: #undef bn_sqr_comba4
sl@0: EXPORT_C void bn_sqr_comba4(BN_ULONG *r, BN_ULONG *a)
sl@0: 	{
sl@0: 	BN_ULONG t[8];
sl@0: 	bn_sqr_normal(r,a,4,t);
sl@0: 	}
sl@0: 
sl@0: #undef bn_sqr_comba8
sl@0: EXPORT_C void bn_sqr_comba8(BN_ULONG *r, BN_ULONG *a)
sl@0: 	{
sl@0: 	BN_ULONG t[16];
sl@0: 	bn_sqr_normal(r,a,8,t);
sl@0: 	}
sl@0: 
sl@0: EXPORT_C void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
sl@0: 	{
sl@0: 	r[4]=bn_mul_words(    &(r[0]),a,4,b[0]);
sl@0: 	r[5]=bn_mul_add_words(&(r[1]),a,4,b[1]);
sl@0: 	r[6]=bn_mul_add_words(&(r[2]),a,4,b[2]);
sl@0: 	r[7]=bn_mul_add_words(&(r[3]),a,4,b[3]);
sl@0: 	}
sl@0: 
sl@0: EXPORT_C void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
sl@0: 	{
sl@0: 	r[ 8]=bn_mul_words(    &(r[0]),a,8,b[0]);
sl@0: 	r[ 9]=bn_mul_add_words(&(r[1]),a,8,b[1]);
sl@0: 	r[10]=bn_mul_add_words(&(r[2]),a,8,b[2]);
sl@0: 	r[11]=bn_mul_add_words(&(r[3]),a,8,b[3]);
sl@0: 	r[12]=bn_mul_add_words(&(r[4]),a,8,b[4]);
sl@0: 	r[13]=bn_mul_add_words(&(r[5]),a,8,b[5]);
sl@0: 	r[14]=bn_mul_add_words(&(r[6]),a,8,b[6]);
sl@0: 	r[15]=bn_mul_add_words(&(r[7]),a,8,b[7]);
sl@0: 	}
sl@0: 
sl@0: #endif /* !BN_MUL_COMBA */