[PATCH v2] [BZ 21130] ldbl-128: Fix y0 and y1 for -Inf input

[PATCH v2] [BZ 21130] ldbl-128: Fix y0 and y1 for -Inf input

[Gabriel F. T. Gomes]

The Bessel functions of the second type (Yn) are not defined for
negative input and should return NAN with the "invalid" exception
raised, in these cases.  However, current code checks for infinity and
return zero, regardless of the sign.  This error is exposed for long
double when linking with -lieee.  Without this flag, the error is not
exposed, because the wrappers for these functions, which use
__kernel_standard functionality, return the correct value.

Tested for powerpc64le.

2017-01-10  Gabriel F. T. Gomes  <gftg@linux.vnet.ibm.com>

	[BZ 21130]
	* sysdeps/ieee754/ldbl-128/e_j0l.c (__ieee754_y0l): Return NAN
	with the "invalid" exception raised when x is -Inf.
	* sysdeps/ieee754/ldbl-128/e_j1l.c (__ieee754_y1l): Likewise.
---
 sysdeps/ieee754/ldbl-128/e_j0l.c | 7 +------
 sysdeps/ieee754/ldbl-128/e_j1l.c | 7 +------
 2 files changed, 2 insertions(+), 12 deletions(-)

diff --git a/sysdeps/ieee754/ldbl-128/e_j0l.c b/sysdeps/ieee754/ldbl-128/e_j0l.c
index d711007..855b5a5 100644
--- a/sysdeps/ieee754/ldbl-128/e_j0l.c
+++ b/sysdeps/ieee754/ldbl-128/e_j0l.c
@@ -829,12 +829,7 @@ _Float128
   _Float128 xx, xinv, z, p, q, c, s, cc, ss;
 
   if (! isfinite (x))
-    {
-      if (x != x)
-	return x + x;
-      else
-	return 0;
-    }
+    return 1 / (x + x * x);
   if (x <= 0)
     {
       if (x < 0)
diff --git a/sysdeps/ieee754/ldbl-128/e_j1l.c b/sysdeps/ieee754/ldbl-128/e_j1l.c
index 9e78230..db8dca0 100644
--- a/sysdeps/ieee754/ldbl-128/e_j1l.c
+++ b/sysdeps/ieee754/ldbl-128/e_j1l.c
@@ -847,12 +847,7 @@ __ieee754_y1l (_Float128 x)
   _Float128 xx, xinv, z, p, q, c, s, cc, ss;
 
   if (! isfinite (x))
-    {
-      if (x != x)
-	return x + x;
-      else
-	return 0;
-    }
+    return 1 / (x + x * x);
   if (x <= 0)
     {
       if (x < 0)
-- 
2.4.11

Re: [PATCH v2] [BZ 21130] ldbl-128: Fix y0 and y1 for -Inf input

[Joseph Myers]

On Fri, 10 Feb 2017, Gabriel F. T. Gomes wrote:

> The Bessel functions of the second type (Yn) are not defined for
> negative input and should return NAN with the "invalid" exception
> raised, in these cases.  However, current code checks for infinity and
> return zero, regardless of the sign.  This error is exposed for long
> double when linking with -lieee.  Without this flag, the error is not
> exposed, because the wrappers for these functions, which use
> __kernel_standard functionality, return the correct value.
> 
> Tested for powerpc64le.

OK.

> 	[BZ 21130]

Note that there is a # before the bug number in the ChangeLog syntax, 
"[BZ #21130]".

-- 
Joseph S. Myers
joseph@codesourcery.com

Re: [PATCH v2] [BZ 21130] ldbl-128: Fix y0 and y1 for -Inf input

[Gabriel F. T. Gomes]

On Fri, 10 Feb 2017 20:55:13 +0000
Joseph Myers <joseph@codesourcery.com> wrote:

> On Fri, 10 Feb 2017, Gabriel F. T. Gomes wrote:
> 
> > The Bessel functions of the second type (Yn) are not defined for
> > negative input and should return NAN with the "invalid" exception
> > raised, in these cases.  However, current code checks for infinity and
> > return zero, regardless of the sign.  This error is exposed for long
> > double when linking with -lieee.  Without this flag, the error is not
> > exposed, because the wrappers for these functions, which use
> > __kernel_standard functionality, return the correct value.
> > 
> > Tested for powerpc64le.  
> 
> OK.
> 
> > 	[BZ 21130]  
> 
> Note that there is a # before the bug number in the ChangeLog syntax, 
> "[BZ #21130]".
> 

Thanks.  Pushed with this change as b987917e6aa7.