ISNAN bug in gfortran 4.3?

[Steve Kargl]

On Fri, Apr 24, 2009 at 02:25:40PM -0600, Matt Fago wrote:
> 
> ISNAN does seemingly work, I'm just puzzled by the memory corruption. I'm
> not using -ffast-math, only -fopenmp -O3.
> 
> I can look at -fdump-tree-original, but it seems there must be some hidden
> bug somewhere else. It's just odd that valgrind comes up clean without the
> ISNAN, and gfortran 4.4.0 and ifort 10.1 work as-is.
> 

ChangeLog shows all changes for isnan occurred during 2007.  Your
original email showed 

gcc version 4.3.0 20080428 (Red Hat 4.3.0-8) (GCC)

which is a RedHat-modified GCC.  Perhaps, you need to contact
RedHat.

-- 
Steve