about summary refs log tree commit diff
path: root/sysdeps/ieee754
diff options
context:
space:
mode:
authorAdhemerval Zanella <azanella@linux.vnet.ibm.com>2012-06-05 10:13:41 -0300
committerAdhemerval Zanella <azanella@linux.vnet.ibm.com>2012-06-05 10:16:49 -0300
commit34ae0b3270c67cae0c54ac98b693fdf7d010a206 (patch)
tree6ddad70ab157b943232a254477fe07ec96445331 /sysdeps/ieee754
parent1214ec8f4c2e8fd105a6d9159b85cf31c0c6c386 (diff)
downloadglibc-34ae0b3270c67cae0c54ac98b693fdf7d010a206.tar.gz
glibc-34ae0b3270c67cae0c54ac98b693fdf7d010a206.tar.xz
glibc-34ae0b3270c67cae0c54ac98b693fdf7d010a206.zip
Fix ldbl128ibm fmodl for subnormals.
Diffstat (limited to 'sysdeps/ieee754')
-rw-r--r--sysdeps/ieee754/ldbl-128ibm/e_fmodl.c27
-rw-r--r--sysdeps/ieee754/ldbl-128ibm/math_ldbl.h10
2 files changed, 19 insertions, 18 deletions
diff --git a/sysdeps/ieee754/ldbl-128ibm/e_fmodl.c b/sysdeps/ieee754/ldbl-128ibm/e_fmodl.c
index 10cda317a9..31e709b5d5 100644
--- a/sysdeps/ieee754/ldbl-128ibm/e_fmodl.c
+++ b/sysdeps/ieee754/ldbl-128ibm/e_fmodl.c
@@ -27,8 +27,8 @@ static const long double one = 1.0, Zero[] = {0.0, -0.0,};
 long double
 __ieee754_fmodl (long double x, long double y)
 {
-	int64_t n,hx,hy,hz,ix,iy,sx,i;
-	u_int64_t lx,ly,lz;
+	int64_t n,hx,hy,hz,ix,iy,sx;
+	u_int64_t lx,ly,lz, i;
 	int temp;
 
 	GET_LDOUBLE_WORDS64(hx,lx,x);
@@ -38,41 +38,42 @@ __ieee754_fmodl (long double x, long double y)
 	hy &= 0x7fffffffffffffffLL;		/* |y| */
 
     /* purge off exception values */
-	if((hy|(ly&0x7fffffffffffffff))==0||(hx>=0x7ff0000000000000LL)|| /* y=0,or x not finite */
-	  (hy>0x7ff0000000000000LL))	/* or y is NaN */
+	if(__builtin_expect((hy|(ly&0x7fffffffffffffff))==0 ||
+			    (hx>=0x7ff0000000000000LL)|| /* y=0,or x not finite */
+			    (hy>0x7ff0000000000000LL),0))	/* or y is NaN */
 	    return (x*y)/(x*y);
-	if(hx<=hy) {
+	if(__builtin_expect(hx<=hy,0)) {
 	    if((hx<hy)||(lx<ly)) return x;	/* |x|<|y| return x */
 	    if(lx==ly)
 		return Zero[(u_int64_t)sx>>63];	/* |x|=|y| return x*0*/
 	}
 
     /* determine ix = ilogb(x) */
-	if(hx<0x0010000000000000LL) {	/* subnormal x */
+	if(__builtin_expect(hx<0x0010000000000000LL,0)) {	/* subnormal x */
 	    if(hx==0) {
 		for (ix = -1043, i=lx; i>0; i<<=1) ix -=1;
 	    } else {
-		for (ix = -1022, i=hx<<19; i>0; i<<=1) ix -=1;
+		for (ix = -1022, i=(hx<<11); i>0; i<<=1) ix -=1;
 	    }
 	} else ix = (hx>>52)-0x3ff;
 
     /* determine iy = ilogb(y) */
-	if(hy<0x0010000000000000LL) {	/* subnormal y */
+	if(__builtin_expect(hy<0x0010000000000000LL,0)) {	/* subnormal y */
 	    if(hy==0) {
 		for (iy = -1043, i=ly; i>0; i<<=1) iy -=1;
 	    } else {
-		for (iy = -1022, i=hy<<19; i>0; i<<=1) iy -=1;
+		for (iy = -1022, i=(hy<<11); i>0; i<<=1) iy -=1;
 	    }
 	} else iy = (hy>>52)-0x3ff;
 
     /* Make the IBM extended format 105 bit mantissa look like the ieee854 112
-       bit mantissa so the following operatations will give the correct
+       bit mantissa so the following operations will give the correct
        result.  */
 	ldbl_extract_mantissa(&hx, &lx, &temp, x);
 	ldbl_extract_mantissa(&hy, &ly, &temp, y);
 
     /* set up {hx,lx}, {hy,ly} and align y to x */
-	if(ix >= -1022)
+	if(__builtin_expect(ix >= -1022, 1))
 	    hx = 0x0001000000000000LL|(0x0000ffffffffffffLL&hx);
 	else {		/* subnormal x, shift x to normal */
 	    n = -1022-ix;
@@ -84,7 +85,7 @@ __ieee754_fmodl (long double x, long double y)
 		lx = 0;
 	    }
 	}
-	if(iy >= -1022)
+	if(__builtin_expect(iy >= -1022, 1))
 	    hy = 0x0001000000000000LL|(0x0000ffffffffffffLL&hy);
 	else {		/* subnormal y, shift y to normal */
 	    n = -1022-iy;
@@ -118,7 +119,7 @@ __ieee754_fmodl (long double x, long double y)
 	    hx = hx+hx+(lx>>63); lx = lx+lx;
 	    iy -= 1;
 	}
-	if(iy>= -1022) {	/* normalize output */
+	if(__builtin_expect(iy>= -1022,0)) {	/* normalize output */
 	    x = ldbl_insert_mantissa((sx>>63), iy, hx, lx);
 	} else {		/* subnormal output */
 	    n = -1022 - iy;
diff --git a/sysdeps/ieee754/ldbl-128ibm/math_ldbl.h b/sysdeps/ieee754/ldbl-128ibm/math_ldbl.h
index d055d6597e..be9ac71cb0 100644
--- a/sysdeps/ieee754/ldbl-128ibm/math_ldbl.h
+++ b/sysdeps/ieee754/ldbl-128ibm/math_ldbl.h
@@ -6,20 +6,20 @@
 #include <ieee754.h>
   
 static inline void
-ldbl_extract_mantissa (int64_t *hi64, u_int64_t *lo64, int *exp, long double x)
+ldbl_extract_mantissa (int64_t *hi64, uint64_t *lo64, int *exp, long double x)
 {
   /* We have 105 bits of mantissa plus one implicit digit.  Since
      106 bits are representable we use the first implicit digit for
      the number before the decimal point and the second implicit bit
      as bit 53 of the mantissa.  */
-  unsigned long long hi, lo;
+  uint64_t hi, lo;
   int ediff;
   union ibm_extended_long_double eldbl;
   eldbl.d = x;
   *exp = eldbl.ieee.exponent - IBM_EXTENDED_LONG_DOUBLE_BIAS;
 
-  lo = ((long long)eldbl.ieee.mantissa2 << 32) | eldbl.ieee.mantissa3;
-  hi = ((long long)eldbl.ieee.mantissa0 << 32) | eldbl.ieee.mantissa1;
+  lo = ((int64_t)eldbl.ieee.mantissa2 << 32) | eldbl.ieee.mantissa3;
+  hi = ((int64_t)eldbl.ieee.mantissa0 << 32) | eldbl.ieee.mantissa1;
   /* If the lower double is not a denomal or zero then set the hidden
      53rd bit.  */
   if (eldbl.ieee.exponent2 > 0x001)
@@ -31,8 +31,8 @@ ldbl_extract_mantissa (int64_t *hi64, u_int64_t *lo64, int *exp, long double x)
       ediff = eldbl.ieee.exponent - eldbl.ieee.exponent2;
       if (ediff > 53)
 	lo = lo >> (ediff-53);
+      hi |= (1ULL << 52);
     }
-  hi |= (1ULL << 52);
   
   if ((eldbl.ieee.negative != eldbl.ieee.negative2)
       && ((eldbl.ieee.exponent2 != 0) && (lo != 0LL)))