[Bug middle-end/84309] [8 Regression] Wrong-code with -ffast-math

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84309

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Eric Gallager from comment #4)
> (In reply to Jakub Jelinek from comment #2)
> > Isn't this just an invalid test?  I mean, when working with floating point,
> > especially when using -ffast-math, one needs to accept some inaccuracy.
> > Here we compute
> > 0x1.ffffffffffff9p+11
> > which is 7ulps from the correct result, I'd say for -ffast-math that is
> > within the acceptability limit.
> 
> There's some warnings for it:

The warnings are completely useless.