bn_asm.c

RMI的处理器au1200系列所用的BOOTLOAD,包括SD卡启动USB启动硬盘启动网络启动,并初始化硬件的所有参数,支持内核调试.

	r[1] = (t1 - t2 - c) & BN_MASK2;
	if (t1 != t2)
	    c = (t1 < t2);
	if (--n <= 0)
	    break;

	t1 = a[2];
	t2 = b[2];
	r[2] = (t1 - t2 - c) & BN_MASK2;
	if (t1 != t2)
	    c = (t1 < t2);
	if (--n <= 0)
	    break;

	t1 = a[3];
	t2 = b[3];
	r[3] = (t1 - t2 - c) & BN_MASK2;
	if (t1 != t2)
	    c = (t1 < t2);
	if (--n <= 0)
	    break;

	a += 4;
	b += 4;
	r += 4;
    }
    return (c);
}

#ifdef BN_MUL_COMBA

#undef bn_mul_comba8
#undef bn_mul_comba4
#undef bn_sqr_comba8
#undef bn_sqr_comba4

/* 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) */
/* 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) */

#ifdef BN_LLONG
#define mul_add_c(a,b,c0,c1,c2) \
	t=(BN_ULLONG)a*b; \
	t1=(BN_ULONG)Lw(t); \
	t2=(BN_ULONG)Hw(t); \
	c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;

#define mul_add_c2(a,b,c0,c1,c2) \
	t=(BN_ULLONG)a*b; \
	tt=(t+t)&BN_MASK; \
	if (tt < t) c2++; \
	t1=(BN_ULONG)Lw(tt); \
	t2=(BN_ULONG)Hw(tt); \
	c0=(c0+t1)&BN_MASK2;  \
	if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;

#define sqr_add_c(a,i,c0,c1,c2) \
	t=(BN_ULLONG)a[i]*a[i]; \
	t1=(BN_ULONG)Lw(t); \
	t2=(BN_ULONG)Hw(t); \
	c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;

#define sqr_add_c2(a,i,j,c0,c1,c2) \
	mul_add_c2((a)[i],(a)[j],c0,c1,c2)

#elif defined(BN_UMULT_HIGH)

#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 sqr_add_c(a,i,c0,c1,c2)	{	\
	BN_ULONG ta=(a)[i];		\
	t1 = ta * ta;			\
	t2 = BN_UMULT_HIGH(ta,ta);	\
	c0 += t1; t2 += (c0<t1)?1:0;	\
	c1 += t2; c2 += (c1<t2)?1:0;	\
	}

#define sqr_add_c2(a,i,j,c0,c1,c2)	\
	mul_add_c2((a)[i],(a)[j],c0,c1,c2)

#else				/* !BN_LLONG */
#define mul_add_c(a,b,c0,c1,c2) \
	t1=LBITS(a); t2=HBITS(a); \
	bl=LBITS(b); bh=HBITS(b); \
	mul64(t1,t2,bl,bh); \
	c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;

#define mul_add_c2(a,b,c0,c1,c2) \
	t1=LBITS(a); t2=HBITS(a); \
	bl=LBITS(b); bh=HBITS(b); \
	mul64(t1,t2,bl,bh); \
	if (t2 & BN_TBIT) c2++; \
	t2=(t2+t2)&BN_MASK2; \
	if (t1 & BN_TBIT) t2++; \
	t1=(t1+t1)&BN_MASK2; \
	c0=(c0+t1)&BN_MASK2;  \
	if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \
	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;

#define sqr_add_c(a,i,c0,c1,c2) \
	sqr64(t1,t2,(a)[i]); \
	c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \
	c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++;

#define sqr_add_c2(a,i,j,c0,c1,c2) \
	mul_add_c2((a)[i],(a)[j],c0,c1,c2)
#endif				/* !BN_LLONG */

void bn_mul_comba8(BN_ULONG * r, BN_ULONG * a, BN_ULONG * b)
{
#ifdef BN_LLONG
    BN_ULLONG t;
#else
    BN_ULONG bl, bh;
#endif
    BN_ULONG t1, t2;
    BN_ULONG c1, c2, c3;

    c1 = 0;
    c2 = 0;
    c3 = 0;
    mul_add_c(a[0], b[0], c1, c2, c3);
    r[0] = c1;
    c1 = 0;
    mul_add_c(a[0], b[1], c2, c3, c1);
    mul_add_c(a[1], b[0], c2, c3, c1);
    r[1] = c2;
    c2 = 0;
    mul_add_c(a[2], b[0], c3, c1, c2);
    mul_add_c(a[1], b[1], c3, c1, c2);
    mul_add_c(a[0], b[2], c3, c1, c2);
    r[2] = c3;
    c3 = 0;
    mul_add_c(a[0], b[3], c1, c2, c3);
    mul_add_c(a[1], b[2], c1, c2, c3);
    mul_add_c(a[2], b[1], c1, c2, c3);
    mul_add_c(a[3], b[0], c1, c2, c3);
    r[3] = c1;
    c1 = 0;
    mul_add_c(a[4], b[0], c2, c3, c1);
    mul_add_c(a[3], b[1], c2, c3, c1);
    mul_add_c(a[2], b[2], c2, c3, c1);
    mul_add_c(a[1], b[3], c2, c3, c1);
    mul_add_c(a[0], b[4], c2, c3, c1);
    r[4] = c2;
    c2 = 0;
    mul_add_c(a[0], b[5], c3, c1, c2);
    mul_add_c(a[1], b[4], c3, c1, c2);
    mul_add_c(a[2], b[3], c3, c1, c2);
    mul_add_c(a[3], b[2], c3, c1, c2);
    mul_add_c(a[4], b[1], c3, c1, c2);
    mul_add_c(a[5], b[0], c3, c1, c2);
    r[5] = c3;
    c3 = 0;
    mul_add_c(a[6], b[0], c1, c2, c3);
    mul_add_c(a[5], b[1], c1, c2, c3);
    mul_add_c(a[4], b[2], c1, c2, c3);
    mul_add_c(a[3], b[3], c1, c2, c3);
    mul_add_c(a[2], b[4], c1, c2, c3);
    mul_add_c(a[1], b[5], c1, c2, c3);
    mul_add_c(a[0], b[6], c1, c2, c3);
    r[6] = c1;
    c1 = 0;
    mul_add_c(a[0], b[7], c2, c3, c1);
    mul_add_c(a[1], b[6], c2, c3, c1);
    mul_add_c(a[2], b[5], c2, c3, c1);
    mul_add_c(a[3], b[4], c2, c3, c1);
    mul_add_c(a[4], b[3], c2, c3, c1);
    mul_add_c(a[5], b[2], c2, c3, c1);
    mul_add_c(a[6], b[1], c2, c3, c1);
    mul_add_c(a[7], b[0], c2, c3, c1);
    r[7] = c2;
    c2 = 0;
    mul_add_c(a[7], b[1], c3, c1, c2);
    mul_add_c(a[6], b[2], c3, c1, c2);
    mul_add_c(a[5], b[3], c3, c1, c2);
    mul_add_c(a[4], b[4], c3, c1, c2);
    mul_add_c(a[3], b[5], c3, c1, c2);
    mul_add_c(a[2], b[6], c3, c1, c2);
    mul_add_c(a[1], b[7], c3, c1, c2);
    r[8] = c3;
    c3 = 0;
    mul_add_c(a[2], b[7], c1, c2, c3);
    mul_add_c(a[3], b[6], c1, c2, c3);
    mul_add_c(a[4], b[5], c1, c2, c3);
    mul_add_c(a[5], b[4], c1, c2, c3);
    mul_add_c(a[6], b[3], c1, c2, c3);
    mul_add_c(a[7], b[2], c1, c2, c3);
    r[9] = c1;
    c1 = 0;
    mul_add_c(a[7], b[3], c2, c3, c1);
    mul_add_c(a[6], b[4], c2, c3, c1);
    mul_add_c(a[5], b[5], c2, c3, c1);
    mul_add_c(a[4], b[6], c2, c3, c1);
    mul_add_c(a[3], b[7], c2, c3, c1);
    r[10] = c2;
    c2 = 0;
    mul_add_c(a[4], b[7], c3, c1, c2);
    mul_add_c(a[5], b[6], c3, c1, c2);
    mul_add_c(a[6], b[5], c3, c1, c2);
    mul_add_c(a[7], b[4], c3, c1, c2);
    r[11] = c3;
    c3 = 0;
    mul_add_c(a[7], b[5], c1, c2, c3);
    mul_add_c(a[6], b[6], c1, c2, c3);
    mul_add_c(a[5], b[7], c1, c2, c3);
    r[12] = c1;
    c1 = 0;
    mul_add_c(a[6], b[7], c2, c3, c1);
    mul_add_c(a[7], b[6], c2, c3, c1);
    r[13] = c2;
    c2 = 0;
    mul_add_c(a[7], b[7], c3, c1, c2);
    r[14] = c3;
    r[15] = c1;
}

void bn_mul_comba4(BN_ULONG * r, BN_ULONG * a, BN_ULONG * b)
{
#ifdef BN_LLONG
    BN_ULLONG t;
#else
    BN_ULONG bl, bh;
#endif
    BN_ULONG t1, t2;
    BN_ULONG c1, c2, c3;

    c1 = 0;
    c2 = 0;
    c3 = 0;
    mul_add_c(a[0], b[0], c1, c2, c3);
    r[0] = c1;
    c1 = 0;
    mul_add_c(a[0], b[1], c2, c3, c1);
    mul_add_c(a[1], b[0], c2, c3, c1);
    r[1] = c2;
    c2 = 0;
    mul_add_c(a[2], b[0], c3, c1, c2);
    mul_add_c(a[1], b[1], c3, c1, c2);
    mul_add_c(a[0], b[2], c3, c1, c2);
    r[2] = c3;
    c3 = 0;
    mul_add_c(a[0], b[3], c1, c2, c3);
    mul_add_c(a[1], b[2], c1, c2, c3);
    mul_add_c(a[2], b[1], c1, c2, c3);
    mul_add_c(a[3], b[0], c1, c2, c3);
    r[3] = c1;
    c1 = 0;
    mul_add_c(a[3], b[1], c2, c3, c1);
    mul_add_c(a[2], b[2], c2, c3, c1);
    mul_add_c(a[1], b[3], c2, c3, c1);
    r[4] = c2;
    c2 = 0;
    mul_add_c(a[2], b[3], c3, c1, c2);
    mul_add_c(a[3], b[2], c3, c1, c2);
    r[5] = c3;
    c3 = 0;
    mul_add_c(a[3], b[3], c1, c2, c3);
    r[6] = c1;
    r[7] = c2;
}

void bn_sqr_comba8(BN_ULONG * r, BN_ULONG * a)
{
#ifdef BN_LLONG
    BN_ULLONG t, tt;
#else
    BN_ULONG bl, bh;
#endif
    BN_ULONG t1, t2;
    BN_ULONG c1, c2, c3;

    c1 = 0;
    c2 = 0;
    c3 = 0;
    sqr_add_c(a, 0, c1, c2, c3);
    r[0] = c1;
    c1 = 0;
    sqr_add_c2(a, 1, 0, c2, c3, c1);
    r[1] = c2;
    c2 = 0;
    sqr_add_c(a, 1, c3, c1, c2);
    sqr_add_c2(a, 2, 0, c3, c1, c2);
    r[2] = c3;
    c3 = 0;
    sqr_add_c2(a, 3, 0, c1, c2, c3);
    sqr_add_c2(a, 2, 1, c1, c2, c3);
    r[3] = c1;
    c1 = 0;
    sqr_add_c(a, 2, c2, c3, c1);
    sqr_add_c2(a, 3, 1, c2, c3, c1);
    sqr_add_c2(a, 4, 0, c2, c3, c1);
    r[4] = c2;
    c2 = 0;
    sqr_add_c2(a, 5, 0, c3, c1, c2);
    sqr_add_c2(a, 4, 1, c3, c1, c2);
    sqr_add_c2(a, 3, 2, c3, c1, c2);
    r[5] = c3;
    c3 = 0;
    sqr_add_c(a, 3, c1, c2, c3);
    sqr_add_c2(a, 4, 2, c1, c2, c3);
    sqr_add_c2(a, 5, 1, c1, c2, c3);
    sqr_add_c2(a, 6, 0, c1, c2, c3);
    r[6] = c1;
    c1 = 0;
    sqr_add_c2(a, 7, 0, c2, c3, c1);
    sqr_add_c2(a, 6, 1, c2, c3, c1);
    sqr_add_c2(a, 5, 2, c2, c3, c1);
    sqr_add_c2(a, 4, 3, c2, c3, c1);
    r[7] = c2;
    c2 = 0;
    sqr_add_c(a, 4, c3, c1, c2);
    sqr_add_c2(a, 5, 3, c3, c1, c2);
    sqr_add_c2(a, 6, 2, c3, c1, c2);
    sqr_add_c2(a, 7, 1, c3, c1, c2);
    r[8] = c3;
    c3 = 0;
    sqr_add_c2(a, 7, 2, c1, c2, c3);
    sqr_add_c2(a, 6, 3, c1, c2, c3);
    sqr_add_c2(a, 5, 4, c1, c2, c3);
    r[9] = c1;
    c1 = 0;
    sqr_add_c(a, 5, c2, c3, c1);
    sqr_add_c2(a, 6, 4, c2, c3, c1);
    sqr_add_c2(a, 7, 3, c2, c3, c1);
    r[10] = c2;
    c2 = 0;
    sqr_add_c2(a, 7, 4, c3, c1, c2);
    sqr_add_c2(a, 6, 5, c3, c1, c2);
    r[11] = c3;
    c3 = 0;
    sqr_add_c(a, 6, c1, c2, c3);
    sqr_add_c2(a, 7, 5, c1, c2, c3);
    r[12] = c1;
    c1 = 0;
    sqr_add_c2(a, 7, 6, c2, c3, c1);
    r[13] = c2;
    c2 = 0;
    sqr_add_c(a, 7, c3, c1, c2);
    r[14] = c3;
    r[15] = c1;
}

void bn_sqr_comba4(BN_ULONG * r, BN_ULONG * a)
{
#ifdef BN_LLONG
    BN_ULLONG t, tt;
#else
    BN_ULONG bl, bh;
#endif
    BN_ULONG t1, t2;
    BN_ULONG c1, c2, c3;

    c1 = 0;
    c2 = 0;
    c3 = 0;
    sqr_add_c(a, 0, c1, c2, c3);
    r[0] = c1;
    c1 = 0;
    sqr_add_c2(a, 1, 0, c2, c3, c1);
    r[1] = c2;
    c2 = 0;
    sqr_add_c(a, 1, c3, c1, c2);
    sqr_add_c2(a, 2, 0, c3, c1, c2);
    r[2] = c3;
    c3 = 0;
    sqr_add_c2(a, 3, 0, c1, c2, c3);
    sqr_add_c2(a, 2, 1, c1, c2, c3);
    r[3] = c1;
    c1 = 0;
    sqr_add_c(a, 2, c2, c3, c1);
    sqr_add_c2(a, 3, 1, c2, c3, c1);
    r[4] = c2;
    c2 = 0;
    sqr_add_c2(a, 3, 2, c3, c1, c2);
    r[5] = c3;
    c3 = 0;
    sqr_add_c(a, 3, c1, c2, c3);
    r[6] = c1;
    r[7] = c2;
}
#else				/* !BN_MUL_COMBA */

/* hmm... is it faster just to do a multiply? */
#undef bn_sqr_comba4
void bn_sqr_comba4(BN_ULONG * r, 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, BN_ULONG * a)
{
    BN_ULONG t[16];
    bn_sqr_normal(r, a, 8, t);
}

void bn_mul_comba4(BN_ULONG * r, BN_ULONG * a, BN_ULONG * b)
{
    r[4] = bn_mul_words(&(r[0]), a, 4, b[0]);
    r[5] = bn_mul_add_words(&(r[1]), a, 4, b[1]);
    r[6] = bn_mul_add_words(&(r[2]), a, 4, b[2]);
    r[7] = bn_mul_add_words(&(r[3]), a, 4, b[3]);
}

void bn_mul_comba8(BN_ULONG * r, BN_ULONG * a, BN_ULONG * b)
{
    r[8] = bn_mul_words(&(r[0]), a, 8, b[0]);
    r[9] = bn_mul_add_words(&(r[1]), a, 8, b[1]);
    r[10] = bn_mul_add_words(&(r[2]), a, 8, b[2]);
    r[11] = bn_mul_add_words(&(r[3]), a, 8, b[3]);
    r[12] = bn_mul_add_words(&(r[4]), a, 8, b[4]);
    r[13] = bn_mul_add_words(&(r[5]), a, 8, b[5]);
    r[14] = bn_mul_add_words(&(r[6]), a, 8, b[6]);
    r[15] = bn_mul_add_words(&(r[7]), a, 8, b[7]);
}

#endif				/* !BN_MUL_COMBA */