#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
* machine.
*/
-#ifdef _WIN64
+#if defined(_WIN64) || !defined(__LP64__)
#define BN_ULONG unsigned long long
#else
#define BN_ULONG unsigned long
#undef mul
#undef mul_add
-/*
+/*-
* "m"(a), "+m"(r) is the way to favor DirectPath ยต-code;
* "g"(0) let the compiler to decide where does it
* want to keep the value of zero;
: "cc"); \
(r)=carry, carry=high; \
} while (0)
-
+#undef sqr
#define sqr(r0,r1,a) \
asm ("mulq %2" \
: "=a"(r0),"=d"(r1) \
}
BN_ULONG bn_add_words (BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int n)
-{ BN_ULONG ret=0,i=0;
+{ BN_ULONG ret;
+ size_t i=0;
if (n <= 0) return 0;
- asm (
- " subq %2,%2 \n"
+ asm volatile (
+ " subq %0,%0 \n" /* clear carry */
+ " jmp 1f \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"
+ " lea 1(%2),%2 \n"
" loop 1b \n"
" sbbq %0,%0 \n"
- : "=&a"(ret),"+c"(n),"=&r"(i)
+ : "=&r"(ret),"+c"(n),"+r"(i)
: "r"(rp),"r"(ap),"r"(bp)
- : "cc"
+ : "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;
+{ BN_ULONG ret;
+ size_t i=0;
if (n <= 0) return 0;
- asm (
- " subq %2,%2 \n"
+ asm volatile (
+ " subq %0,%0 \n" /* clear borrow */
+ " jmp 1f \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"
+ " lea 1(%2),%2 \n"
" loop 1b \n"
" sbbq %0,%0 \n"
- : "=&a"(ret),"+c"(n),"=&r"(i)
+ : "=&r"(ret),"+c"(n),"+r"(i)
: "r"(rp),"r"(ap),"r"(bp)
- : "cc"
+ : "cc", "memory"
);
return ret&1;
/* 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) */
+/*
+ * Keep in mind that carrying into high part of multiplication result
+ * can not overflow, because it cannot be all-ones.
+ */
#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; \
- }
+#define mul_add_c(a,b,c0,c1,c2) do { \
+ BN_ULONG ta = (a), tb = (b); \
+ BN_ULONG lo, hi; \
+ BN_UMULT_LOHI(lo,hi,ta,tb); \
+ c0 += lo; hi += (c0<lo)?1:0; \
+ c1 += hi; c2 += (c1<hi)?1:0; \
+ } while(0)
+
+#define mul_add_c2(a,b,c0,c1,c2) do { \
+ BN_ULONG ta = (a), tb = (b); \
+ BN_ULONG lo, hi, tt; \
+ BN_UMULT_LOHI(lo,hi,ta,tb); \
+ c0 += lo; tt = hi+((c0<lo)?1:0); \
+ c1 += tt; c2 += (c1<tt)?1:0; \
+ c0 += lo; hi += (c0<lo)?1:0; \
+ c1 += hi; c2 += (c1<hi)?1:0; \
+ } while(0)
+
+#define sqr_add_c(a,i,c0,c1,c2) do { \
+ BN_ULONG ta = (a)[i]; \
+ BN_ULONG lo, hi; \
+ BN_UMULT_LOHI(lo,hi,ta,ta); \
+ c0 += lo; hi += (c0<lo)?1:0; \
+ c1 += hi; c2 += (c1<hi)?1:0; \
+ } while(0)
#else
#define mul_add_c(a,b,c0,c1,c2) do { \
+ BN_ULONG t1,t2; \
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"); \
+ asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \
+ : "+r"(c0),"+r"(c1),"+r"(c2) \
+ : "r"(t1),"r"(t2),"g"(0) \
+ : "cc"); \
} while (0)
#define sqr_add_c(a,i,c0,c1,c2) do { \
+ BN_ULONG t1,t2; \
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"); \
+ asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \
+ : "+r"(c0),"+r"(c1),"+r"(c2) \
+ : "r"(t1),"r"(t2),"g"(0) \
+ : "cc"); \
} while (0)
#define mul_add_c2(a,b,c0,c1,c2) do { \
+ BN_ULONG t1,t2; \
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"); \
+ asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \
+ : "+r"(c0),"+r"(c1),"+r"(c2) \
+ : "r"(t1),"r"(t2),"g"(0) \
+ : "cc"); \
+ asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \
+ : "+r"(c0),"+r"(c1),"+r"(c2) \
+ : "r"(t1),"r"(t2),"g"(0) \
+ : "cc"); \
} while (0)
#endif
void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
{
- BN_ULONG t1,t2;
BN_ULONG c1,c2,c3;
c1=0;
void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b)
{
- BN_ULONG t1,t2;
BN_ULONG c1,c2,c3;
c1=0;
void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a)
{
- BN_ULONG t1,t2;
BN_ULONG c1,c2,c3;
c1=0;
void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a)
{
- BN_ULONG t1,t2;
BN_ULONG c1,c2,c3;
c1=0;