[PATCH] Fix the inaccuracy of j0f (BZ 14469) and y0f (BZ 14471) (v3)

[Paul Zimmermann]

thank you Joseph for your review, I will submit a new version.

Paul

> Date: Thu, 28 Jan 2021 18:56:26 +0000
> From: Joseph Myers <joseph@codesourcery.com>
> 
> On Thu, 28 Jan 2021, Paul Zimmermann wrote:
> 
> > -# the next value generates larger error bounds on x86_64 (binary32)
> > -j0 0x2.602774p+0 xfail-rounding:ibm128-libgcc
> 
> I don't think libm test inputs should generally be removed, they're still 
> useful even if other inputs give larger errors.
> 
> > -# the next value generates larger error bounds on x86_64 (binary32)
> > -y0 0xd.3432bp-4
> 
> Likewise.
> 
> > +# the next two values yield the largest error (9 ulps) for binary32 (x86_64)
> > +y0 0x1.33eaacp+5 xfail:binary64 xfail:intel96
> 
> The XFAILs should have a comment referencing the corresponding open bug.
> 
> -- 
> Joseph S. Myers
> joseph@codesourcery.com
>