155714Skris/* crypto/bn/bn_sqr.c */ 255714Skris/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com) 355714Skris * All rights reserved. 455714Skris * 555714Skris * This package is an SSL implementation written 655714Skris * by Eric Young (eay@cryptsoft.com). 755714Skris * The implementation was written so as to conform with Netscapes SSL. 855714Skris * 955714Skris * This library is free for commercial and non-commercial use as long as 1055714Skris * the following conditions are aheared to. The following conditions 1155714Skris * apply to all code found in this distribution, be it the RC4, RSA, 1255714Skris * lhash, DES, etc., code; not just the SSL code. The SSL documentation 1355714Skris * included with this distribution is covered by the same copyright terms 1455714Skris * except that the holder is Tim Hudson (tjh@cryptsoft.com). 1555714Skris * 1655714Skris * Copyright remains Eric Young's, and as such any Copyright notices in 1755714Skris * the code are not to be removed. 1855714Skris * If this package is used in a product, Eric Young should be given attribution 1955714Skris * as the author of the parts of the library used. 2055714Skris * This can be in the form of a textual message at program startup or 2155714Skris * in documentation (online or textual) provided with the package. 2255714Skris * 2355714Skris * Redistribution and use in source and binary forms, with or without 2455714Skris * modification, are permitted provided that the following conditions 2555714Skris * are met: 2655714Skris * 1. Redistributions of source code must retain the copyright 2755714Skris * notice, this list of conditions and the following disclaimer. 2855714Skris * 2. Redistributions in binary form must reproduce the above copyright 2955714Skris * notice, this list of conditions and the following disclaimer in the 3055714Skris * documentation and/or other materials provided with the distribution. 3155714Skris * 3. All advertising materials mentioning features or use of this software 3255714Skris * must display the following acknowledgement: 3355714Skris * "This product includes cryptographic software written by 3455714Skris * Eric Young (eay@cryptsoft.com)" 3555714Skris * The word 'cryptographic' can be left out if the rouines from the library 3655714Skris * being used are not cryptographic related :-). 3755714Skris * 4. If you include any Windows specific code (or a derivative thereof) from 3855714Skris * the apps directory (application code) you must include an acknowledgement: 3955714Skris * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" 4055714Skris * 4155714Skris * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND 4255714Skris * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 4355714Skris * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 4455714Skris * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 4555714Skris * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 4655714Skris * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 4755714Skris * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 4855714Skris * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 4955714Skris * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 5055714Skris * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 5155714Skris * SUCH DAMAGE. 5255714Skris * 5355714Skris * The licence and distribution terms for any publically available version or 5455714Skris * derivative of this code cannot be changed. i.e. this code cannot simply be 5555714Skris * copied and put under another distribution licence 5655714Skris * [including the GNU Public Licence.] 5755714Skris */ 5855714Skris 5955714Skris#include <stdio.h> 6055714Skris#include "cryptlib.h" 6155714Skris#include "bn_lcl.h" 6255714Skris 6355714Skris/* r must not be a */ 6455714Skris/* I've just gone over this and it is now %20 faster on x86 - eay - 27 Jun 96 */ 65109998Smarkmint BN_sqr(BIGNUM *r, const BIGNUM *a, BN_CTX *ctx) 6655714Skris { 6755714Skris int max,al; 6859191Skris int ret = 0; 6955714Skris BIGNUM *tmp,*rr; 7055714Skris 7155714Skris#ifdef BN_COUNT 72109998Smarkm fprintf(stderr,"BN_sqr %d * %d\n",a->top,a->top); 7355714Skris#endif 7455714Skris bn_check_top(a); 7555714Skris 7655714Skris al=a->top; 7755714Skris if (al <= 0) 7855714Skris { 7955714Skris r->top=0; 80279264Sdelphij r->neg = 0; 81160814Ssimon return 1; 8255714Skris } 8355714Skris 8459191Skris BN_CTX_start(ctx); 8559191Skris rr=(a != r) ? r : BN_CTX_get(ctx); 8659191Skris tmp=BN_CTX_get(ctx); 87160814Ssimon if (!rr || !tmp) goto err; 8859191Skris 89160814Ssimon max = 2 * al; /* Non-zero (from above) */ 90160814Ssimon if (bn_wexpand(rr,max) == NULL) goto err; 9155714Skris 9255714Skris if (al == 4) 9355714Skris { 9455714Skris#ifndef BN_SQR_COMBA 9555714Skris BN_ULONG t[8]; 9655714Skris bn_sqr_normal(rr->d,a->d,4,t); 9755714Skris#else 9855714Skris bn_sqr_comba4(rr->d,a->d); 9955714Skris#endif 10055714Skris } 10155714Skris else if (al == 8) 10255714Skris { 10355714Skris#ifndef BN_SQR_COMBA 10455714Skris BN_ULONG t[16]; 10555714Skris bn_sqr_normal(rr->d,a->d,8,t); 10655714Skris#else 10755714Skris bn_sqr_comba8(rr->d,a->d); 10855714Skris#endif 10955714Skris } 11055714Skris else 11155714Skris { 11255714Skris#if defined(BN_RECURSION) 11355714Skris if (al < BN_SQR_RECURSIVE_SIZE_NORMAL) 11455714Skris { 11555714Skris BN_ULONG t[BN_SQR_RECURSIVE_SIZE_NORMAL*2]; 11655714Skris bn_sqr_normal(rr->d,a->d,al,t); 11755714Skris } 11855714Skris else 11955714Skris { 12055714Skris int j,k; 12155714Skris 12255714Skris j=BN_num_bits_word((BN_ULONG)al); 12355714Skris j=1<<(j-1); 12455714Skris k=j+j; 12555714Skris if (al == j) 12655714Skris { 12759191Skris if (bn_wexpand(tmp,k*2) == NULL) goto err; 12855714Skris bn_sqr_recursive(rr->d,a->d,al,tmp->d); 12955714Skris } 13055714Skris else 13155714Skris { 13259191Skris if (bn_wexpand(tmp,max) == NULL) goto err; 13355714Skris bn_sqr_normal(rr->d,a->d,al,tmp->d); 13455714Skris } 13555714Skris } 13655714Skris#else 13759191Skris if (bn_wexpand(tmp,max) == NULL) goto err; 13855714Skris bn_sqr_normal(rr->d,a->d,al,tmp->d); 13955714Skris#endif 14055714Skris } 14155714Skris 142109998Smarkm rr->neg=0; 143160814Ssimon /* If the most-significant half of the top word of 'a' is zero, then 144160814Ssimon * the square of 'a' will max-1 words. */ 145160814Ssimon if(a->d[al - 1] == (a->d[al - 1] & BN_MASK2l)) 146160814Ssimon rr->top = max - 1; 147160814Ssimon else 148160814Ssimon rr->top = max; 14955714Skris if (rr != r) BN_copy(r,rr); 15059191Skris ret = 1; 15159191Skris err: 152160814Ssimon bn_check_top(rr); 153160814Ssimon bn_check_top(tmp); 15459191Skris BN_CTX_end(ctx); 15559191Skris return(ret); 15655714Skris } 15755714Skris 15855714Skris/* tmp must have 2*n words */ 159109998Smarkmvoid bn_sqr_normal(BN_ULONG *r, const BN_ULONG *a, int n, BN_ULONG *tmp) 16055714Skris { 16155714Skris int i,j,max; 162109998Smarkm const BN_ULONG *ap; 163109998Smarkm BN_ULONG *rp; 16455714Skris 16555714Skris max=n*2; 16655714Skris ap=a; 16755714Skris rp=r; 16855714Skris rp[0]=rp[max-1]=0; 16955714Skris rp++; 17055714Skris j=n; 17155714Skris 17255714Skris if (--j > 0) 17355714Skris { 17455714Skris ap++; 17555714Skris rp[j]=bn_mul_words(rp,ap,j,ap[-1]); 17655714Skris rp+=2; 17755714Skris } 17855714Skris 17955714Skris for (i=n-2; i>0; i--) 18055714Skris { 18155714Skris j--; 18255714Skris ap++; 18355714Skris rp[j]=bn_mul_add_words(rp,ap,j,ap[-1]); 18455714Skris rp+=2; 18555714Skris } 18655714Skris 18755714Skris bn_add_words(r,r,r,max); 18855714Skris 18955714Skris /* There will not be a carry */ 19055714Skris 19155714Skris bn_sqr_words(tmp,a,n); 19255714Skris 19355714Skris bn_add_words(r,r,tmp,max); 19455714Skris } 19555714Skris 19655714Skris#ifdef BN_RECURSION 19755714Skris/* r is 2*n words in size, 19868651Skris * a and b are both n words in size. (There's not actually a 'b' here ...) 19955714Skris * n must be a power of 2. 20055714Skris * We multiply and return the result. 20155714Skris * t must be 2*n words in size 20259191Skris * We calculate 20355714Skris * a[0]*b[0] 20455714Skris * a[0]*b[0]+a[1]*b[1]+(a[0]-a[1])*(b[1]-b[0]) 20555714Skris * a[1]*b[1] 20655714Skris */ 207109998Smarkmvoid bn_sqr_recursive(BN_ULONG *r, const BN_ULONG *a, int n2, BN_ULONG *t) 20855714Skris { 20955714Skris int n=n2/2; 21055714Skris int zero,c1; 21155714Skris BN_ULONG ln,lo,*p; 21255714Skris 21355714Skris#ifdef BN_COUNT 214109998Smarkm fprintf(stderr," bn_sqr_recursive %d * %d\n",n2,n2); 21555714Skris#endif 21655714Skris if (n2 == 4) 21755714Skris { 21855714Skris#ifndef BN_SQR_COMBA 21955714Skris bn_sqr_normal(r,a,4,t); 22055714Skris#else 22155714Skris bn_sqr_comba4(r,a); 22255714Skris#endif 22355714Skris return; 22455714Skris } 22555714Skris else if (n2 == 8) 22655714Skris { 22755714Skris#ifndef BN_SQR_COMBA 22855714Skris bn_sqr_normal(r,a,8,t); 22955714Skris#else 23055714Skris bn_sqr_comba8(r,a); 23155714Skris#endif 23255714Skris return; 23355714Skris } 23455714Skris if (n2 < BN_SQR_RECURSIVE_SIZE_NORMAL) 23555714Skris { 23655714Skris bn_sqr_normal(r,a,n2,t); 23755714Skris return; 23855714Skris } 23955714Skris /* r=(a[0]-a[1])*(a[1]-a[0]) */ 24055714Skris c1=bn_cmp_words(a,&(a[n]),n); 24155714Skris zero=0; 24255714Skris if (c1 > 0) 24355714Skris bn_sub_words(t,a,&(a[n]),n); 24455714Skris else if (c1 < 0) 24555714Skris bn_sub_words(t,&(a[n]),a,n); 24655714Skris else 24755714Skris zero=1; 24855714Skris 24955714Skris /* The result will always be negative unless it is zero */ 25055714Skris p= &(t[n2*2]); 25155714Skris 25255714Skris if (!zero) 25355714Skris bn_sqr_recursive(&(t[n2]),t,n,p); 25455714Skris else 25589837Skris memset(&(t[n2]),0,n2*sizeof(BN_ULONG)); 25655714Skris bn_sqr_recursive(r,a,n,p); 25755714Skris bn_sqr_recursive(&(r[n2]),&(a[n]),n,p); 25855714Skris 25955714Skris /* t[32] holds (a[0]-a[1])*(a[1]-a[0]), it is negative or zero 26055714Skris * r[10] holds (a[0]*b[0]) 26155714Skris * r[32] holds (b[1]*b[1]) 26255714Skris */ 26355714Skris 26455714Skris c1=(int)(bn_add_words(t,r,&(r[n2]),n2)); 26555714Skris 26655714Skris /* t[32] is negative */ 26755714Skris c1-=(int)(bn_sub_words(&(t[n2]),t,&(t[n2]),n2)); 26855714Skris 26955714Skris /* t[32] holds (a[0]-a[1])*(a[1]-a[0])+(a[0]*a[0])+(a[1]*a[1]) 27055714Skris * r[10] holds (a[0]*a[0]) 27155714Skris * r[32] holds (a[1]*a[1]) 27255714Skris * c1 holds the carry bits 27355714Skris */ 27455714Skris c1+=(int)(bn_add_words(&(r[n]),&(r[n]),&(t[n2]),n2)); 27555714Skris if (c1) 27655714Skris { 27755714Skris p= &(r[n+n2]); 27855714Skris lo= *p; 27955714Skris ln=(lo+c1)&BN_MASK2; 28055714Skris *p=ln; 28155714Skris 28255714Skris /* The overflow will stop before we over write 28355714Skris * words we should not overwrite */ 28455714Skris if (ln < (BN_ULONG)c1) 28555714Skris { 28655714Skris do { 28755714Skris p++; 28855714Skris lo= *p; 28955714Skris ln=(lo+1)&BN_MASK2; 29055714Skris *p=ln; 29155714Skris } while (ln == 0); 29255714Skris } 29355714Skris } 29455714Skris } 29555714Skris#endif 296