[Bug fortran/51535] Optimization problem with -finit-real=snan at -O1

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51535

--- Comment #4 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> As the status "accidentally fixed in 4.8 but do not know when and why"
> does not exist, you may well do that.

I get the SIGFPE with revision 192891 (2012-10-28) and 'In p_h_usstd!' with
revision 193261 (2012-11-06) and newer. Note that since the SIGFPE needs -O1,
it may have been fixed in the middle-end. But if you really want to find out,
you'll have to do the remaining bisection yourself.

BTW in comment #2 I ment '-O1 -ffpe-trap=zero,overflow,invalid -finit-real=snan
-mfpmath=387'.