Red Hat Bugzilla – Bug 17205
FP comparison optimization bug on i686.
Last modified: 2008-05-01 11:37:58 EDT
When you use -O -march=i686 on gcc.c-torture/execute/ieee/920810-1.c,
the testcase will fail. I believe it is a regression, at least for me.
It may have something to do FP comparison optimization on i686. I
will take a look when I find time.
Seems to happen both in our gcc and vanilla CVS head, looks like
combiner "optimizes" it out. Will try to debug it.
Should be fixed in gcc-2.96-60 in rawhide.