-ffast-math and mathinlines.h

[Ulrich Drepper]

I have generally no problem with this but the granularity of the patch
is not OK.  The -ffast-math flag should only enable additional
function which really behave differently then the libm functions.

Looking briefly through the mathinline.h file this at least excludes

	lrintf
	lrint
	lrintl
	llrintf
	llrint
	llrintl
	finite

Perhaps
	ldexpf
	ldexp
	ldexpl

-- 
---------------.      drepper at gnu.org  ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------