x86_64-mont5.pl: add missing Win64 support.

[openssl.git] / crypto / bn / bn_asm.c

diff --git a/crypto/bn/bn_asm.c b/crypto/bn/bn_asm.c
index 85b203d25dc65b56d9e55c0133fc743ecaaa3326..7b1280ae61456f7ccd1bdce22d5431a4e24bd840 100644 (file)
--- a/crypto/bn/bn_asm.c
+++ b/crypto/bn/bn_asm.c
@@ -63,12 +63,13 @@
 
 #include <stdio.h>
 #include <assert.h>
+#include <openssl/crypto.h>
 #include "cryptlib.h"
 #include "bn_lcl.h"
 
 #if defined(BN_LLONG) || defined(BN_UMULT_HIGH)
 
-BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, size_t num, BN_ULONG w)
+BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
        {
        BN_ULONG c1=0;
 
@@ -94,7 +95,7 @@ BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, size_t num, BN_ULONG
        return(c1);
        } 
 
-BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, size_t num, BN_ULONG w)
+BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
        {
        BN_ULONG c1=0;
 
@@ -119,7 +120,7 @@ BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, size_t num, BN_ULONG w)
        return(c1);
        } 
 
-void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, size_t n)
+void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n)
         {
        assert(n >= 0);
        if (n <= 0) return;
@@ -143,7 +144,7 @@ void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, size_t n)
 
 #else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */
 
-BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, size_t num, BN_ULONG w)
+BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
        {
        BN_ULONG c=0;
        BN_ULONG bl,bh;
@@ -172,7 +173,7 @@ BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, size_t num, BN_ULONG
        return(c);
        } 
 
-BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, size_t num, BN_ULONG w)
+BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w)
        {
        BN_ULONG carry=0;
        BN_ULONG bl,bh;
@@ -201,7 +202,7 @@ BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, size_t num, BN_ULONG w)
        return(carry);
        } 
 
-void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, size_t n)
+void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n)
         {
        assert(n >= 0);
        if (n <= 0) return;
@@ -303,7 +304,7 @@ BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d)
 #endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */
 
 #ifdef BN_LLONG
-BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, size_t n)
+BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
         {
        BN_ULLONG ll=0;
 
@@ -338,7 +339,7 @@ BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, size_t
        return((BN_ULONG)ll);
        }
 #else /* !BN_LLONG */
-BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, size_t n)
+BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
         {
        BN_ULONG c,l,t;
 
@@ -390,7 +391,7 @@ BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, size_t
        }
 #endif /* !BN_LLONG */
 
-BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, size_t n)
+BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n)
         {
        BN_ULONG t1,t2;
        int c=0;
@@ -428,10 +429,12 @@ BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, size_t
 
 #if defined(BN_MUL_COMBA) && !defined(OPENSSL_SMALL_FOOTPRINT)
 
+#ifndef OPENSSL_FIPSCANISTER
 #undef bn_mul_comba8
 #undef bn_mul_comba4
 #undef bn_sqr_comba8
 #undef bn_sqr_comba4
+#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) */
@@ -860,13 +863,35 @@ int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_U
 
        n0 = *n0p;
 
-       tp[num]   = bn_mul_words(tp,ap,num,bp[0]);
+       c0 = 0;
+       ml = bp[0];
+#ifdef mul64
+       mh = HBITS(ml);
+       ml = LBITS(ml);
+       for (j=0;j<num;++j)
+               mul(tp[j],ap[j],ml,mh,c0);
+#else
+       for (j=0;j<num;++j)
+               mul(tp[j],ap[j],ml,c0);
+#endif
+
+       tp[num]   = c0;
        tp[num+1] = 0;
        goto enter;
 
        for(i=0;i<num;i++)
                {
-               c0 = bn_mul_add_words(tp,ap,num,bp[i]);
+               c0 = 0;
+               ml = bp[i];
+#ifdef mul64
+               mh = HBITS(ml);
+               ml = LBITS(ml);
+               for (j=0;j<num;++j)
+                       mul_add(tp[j],ap[j],ml,mh,c0);
+#else
+               for (j=0;j<num;++j)
+                       mul_add(tp[j],ap[j],ml,c0);
+#endif
                c1 = (tp[num] + c0)&BN_MASK2;
                tp[num]   = c1;
                tp[num+1] = (c1<c0?1:0);
@@ -924,14 +949,16 @@ int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_U
 #else /* !BN_MUL_COMBA */
 
 /* hmm... is it faster just to do a multiply? */
+#ifndef OPENSSL_FIPSCANISTER
 #undef bn_sqr_comba4
+#undef bn_sqr_comba8
+#endif
 void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a)
        {
        BN_ULONG t[8];
        bn_sqr_normal(r,a,4,t);
        }
 
-#undef bn_sqr_comba8
 void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a)
        {
        BN_ULONG t[16];