I think this is different than the existing Fortran failures. If '(ims:,jms:)'is changed to '(ims,jms)' or '(ims:ime,jms:jme)' things work.
Test case: SUBROUTINE foo ( dnew , ims, jms, ime, jme) IMPLICIT NONE INTEGER ims, jms, ime, jme REAL , DIMENSION(ims:,jms:) :: dnew INTEGER i i = dnew(10,10) END SUBROUTINE foo The failure: % gfortran -c x.f90 x.f90: In function 'foo': x.f90:1: internal compiler error: in gfc_conv_expr, at fortran/trans-expr.c:1807 Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://gcc.gnu.org/bugs.html> for instructions. -- Summary: ICE in gfc_conv_expr Product: gcc Version: 4.0.0 Status: UNCONFIRMED Severity: normal Priority: P2 Component: fortran AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: sje at cup dot hp dot com CC: gcc-bugs at gcc dot gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19045