glibc/sysdeps/ieee754/ldbl-128ibm/e_hypotl.c
Joseph Myers f6987f5aa4 Fix hypot missing underflows (bug 18803).
Similar to various other bugs in this area, hypot functions can fail
to raise the underflow exception when the result is tiny and inexact
but one or more low bits of the intermediate result that is scaled
down (or, in the i386 case, converted from a wider evaluation format)
are zero.  This patch forces the exception in a similar way to
previous fixes.

Note that this issue cannot arise for implementations of hypotf using
double (or wider) for intermediate evaluation (if hypotf should
underflow, that means the double square root is being computed of some
number of the form N*2^-298, for 0 < N < 2^46, which is exactly
represented as a double, and whatever the rounding mode such a square
root cannot have a mantissa with all zeroes after the initial 23
bits).  Thus no changes are made to hypotf implementations in this
patch, only to hypot and hypotl.

Tested for x86_64, x86, mips64 and powerpc.

	[BZ #18803]
	* sysdeps/i386/fpu/e_hypot.S: Use DEFINE_DBL_MIN.
	(MO): New macro.
	(__ieee754_hypot) [PIC]: Load PIC register.
	(__ieee754_hypot): Use DBL_NARROW_EVAL_UFLOW_NONNEG instead of
	DBL_NARROW_EVAL.
	* sysdeps/ieee754/dbl-64/e_hypot.c (__ieee754_hypot): Use
	math_check_force_underflow_nonneg in case where result might be
	tiny.
	* sysdeps/ieee754/ldbl-128/e_hypotl.c (__ieee754_hypotl):
	Likewise.
	* sysdeps/ieee754/ldbl-128ibm/e_hypotl.c (__ieee754_hypotl):
	Likewise.
	* sysdeps/ieee754/ldbl-96/e_hypotl.c (__ieee754_hypotl): Likewise.
	* sysdeps/powerpc/fpu/e_hypot.c (__ieee754_hypot): Likewise.
	* math/auto-libm-test-in: Add more tests of hypot.
	* math/auto-libm-test-out: Regenerated.
2015-09-24 23:43:57 +00:00

137 lines
3.5 KiB
C

/* @(#)e_hypotl.c 5.1 93/09/24 */
/*
* ====================================================
* Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
*
* Developed at SunPro, a Sun Microsystems, Inc. business.
* Permission to use, copy, modify, and distribute this
* software is freely granted, provided that this notice
* is preserved.
* ====================================================
*/
/* __ieee754_hypotl(x,y)
*
* Method :
* If (assume round-to-nearest) z=x*x+y*y
* has error less than sqrtl(2)/2 ulp, than
* sqrtl(z) has error less than 1 ulp (exercise).
*
* So, compute sqrtl(x*x+y*y) with some care as
* follows to get the error below 1 ulp:
*
* Assume x>y>0;
* (if possible, set rounding to round-to-nearest)
* 1. if x > 2y use
* x1*x1+(y*y+(x2*(x+x1))) for x*x+y*y
* where x1 = x with lower 53 bits cleared, x2 = x-x1; else
* 2. if x <= 2y use
* t1*y1+((x-y)*(x-y)+(t1*y2+t2*y))
* where t1 = 2x with lower 53 bits cleared, t2 = 2x-t1,
* y1= y with lower 53 bits chopped, y2 = y-y1.
*
* NOTE: scaling may be necessary if some argument is too
* large or too tiny
*
* Special cases:
* hypotl(x,y) is INF if x or y is +INF or -INF; else
* hypotl(x,y) is NAN if x or y is NAN.
*
* Accuracy:
* hypotl(x,y) returns sqrtl(x^2+y^2) with error less
* than 1 ulps (units in the last place)
*/
#include <math.h>
#include <math_private.h>
long double
__ieee754_hypotl(long double x, long double y)
{
long double a,b,a1,a2,b1,b2,w,kld;
int64_t j,k,ha,hb;
double xhi, yhi, hi, lo;
xhi = ldbl_high (x);
EXTRACT_WORDS64 (ha, xhi);
yhi = ldbl_high (y);
EXTRACT_WORDS64 (hb, yhi);
ha &= 0x7fffffffffffffffLL;
hb &= 0x7fffffffffffffffLL;
if(hb > ha) {a=y;b=x;j=ha; ha=hb;hb=j;} else {a=x;b=y;}
a = fabsl(a); /* a <- |a| */
b = fabsl(b); /* b <- |b| */
if((ha-hb)>0x0780000000000000LL) {return a+b;} /* x/y > 2**120 */
k=0;
kld = 1.0L;
if(ha > 0x5f30000000000000LL) { /* a>2**500 */
if(ha >= 0x7ff0000000000000LL) { /* Inf or NaN */
w = a+b; /* for sNaN */
if(ha == 0x7ff0000000000000LL)
w = a;
if(hb == 0x7ff0000000000000LL)
w = b;
return w;
}
/* scale a and b by 2**-600 */
a *= 0x1p-600L;
b *= 0x1p-600L;
k = 600;
kld = 0x1p+600L;
}
else if(hb < 0x23d0000000000000LL) { /* b < 2**-450 */
if(hb <= 0x000fffffffffffffLL) { /* subnormal b or 0 */
if(hb==0) return a;
a *= 0x1p+1022L;
b *= 0x1p+1022L;
k = -1022;
kld = 0x1p-1022L;
} else { /* scale a and b by 2^600 */
a *= 0x1p+600L;
b *= 0x1p+600L;
k = -600;
kld = 0x1p-600L;
}
}
/* medium size a and b */
w = a-b;
if (w>b) {
ldbl_unpack (a, &hi, &lo);
a1 = hi;
a2 = lo;
/* a*a + b*b
= (a1+a2)*a + b*b
= a1*a + a2*a + b*b
= a1*(a1+a2) + a2*a + b*b
= a1*a1 + a1*a2 + a2*a + b*b
= a1*a1 + a2*(a+a1) + b*b */
w = __ieee754_sqrtl(a1*a1-(b*(-b)-a2*(a+a1)));
} else {
a = a+a;
ldbl_unpack (b, &hi, &lo);
b1 = hi;
b2 = lo;
ldbl_unpack (a, &hi, &lo);
a1 = hi;
a2 = lo;
/* a*a + b*b
= a*a + (a-b)*(a-b) - (a-b)*(a-b) + b*b
= a*a + w*w - (a*a - 2*a*b + b*b) + b*b
= w*w + 2*a*b
= w*w + (a1+a2)*b
= w*w + a1*b + a2*b
= w*w + a1*(b1+b2) + a2*b
= w*w + a1*b1 + a1*b2 + a2*b */
w = __ieee754_sqrtl(a1*b1-(w*(-w)-(a1*b2+a2*b)));
}
if(k!=0)
{
w *= kld;
math_check_force_underflow_nonneg (w);
return w;
}
else
return w;
}
strong_alias (__ieee754_hypotl, __hypotl_finite)