[PATCH] Fixed inaccuracy of j0f (BZ #28185)

[Joseph Myers]

On Wed, 4 Aug 2021, Paul Zimmermann wrote:

> Would it make sense to add just one entry (the one with the largest error)
> with xfail-rounding:double,float128,float32x,float64,float64x,ldouble?

xfail-rounding is only if the problems are only in non-default rounding 
modes, if there are large errors in round-to-nearest you should use xfail 
(and note that the arguments are the format names as used by 
gen-auto-libm-tests, e.g. binary64, not the names used in the testcase 
names).

-- 
Joseph S. Myers
joseph@codesourcery.com