[Bug fortran/60088] New: Segfault when using quad precision and -march=native on gfortran

[thatcadguy at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60088

            Bug ID: 60088
           Summary: Segfault when using quad precision and -march=native
                    on  gfortran
           Product: gcc
           Version: 4.7.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: thatcadguy at gmail dot com

Created attachment 32061
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32061&action=edit
zip file of the code, makefile, and gcc output

I went to test the attached code with quad precision by changing
SELECTED_REAL_KIND to 33 digits in precision.f90; I got a segfault when trying
to run the compiled sip_test executable. Everything runs fine with double
precision (SELECTED_REAL_KIND @ 15 digits).

This same behavior also occurs with MinGW gfortran 4.8.1 on Windows 7 64-bit.

I noticed that if I take away '-march=native' that the segfault goes away. My
laptop at work with MinGW is a Core i7, this laptop with linux is a Core Duo.

I've attached the code and the gcc -v output as well as the -march=native
flags.