https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83184
--- Comment #2 from Fritz Reese <foreese at gcc dot gnu.org> --- This is really two separate bugs; the first being that the locus of the variable expression is unset from match_old_style_init(), which can be seen in the simpler test case: $ cat z1.f90 integer n1(..) /1/ end $ gfortran z1.f90 z1.f90:1:13: integer n(..) /1/ 1 Error: Assumed-rank array at (1) must be a dummy argument f951: internal compiler error: Segmentation fault 0xd9b2af crash_signal /data/midas/foreese/src/gcc-trunk/gcc/toplev.c:325 0x8c868c traverse_data_list /data/midas/foreese/src/gcc-trunk/gcc/fortran/resolve.c:15450 0x8c868c traverse_data_var /data/midas/foreese/src/gcc-trunk/gcc/fortran/resolve.c:15530 0x8ba7c3 resolve_data /data/midas/foreese/src/gcc-trunk/gcc/fortran/resolve.c:15587 0x8ba7c3 resolve_types /data/midas/foreese/src/gcc-trunk/gcc/fortran/resolve.c:16405 0x8bf4ac gfc_resolve(gfc_namespace*) /data/midas/foreese/src/gcc-trunk/gcc/fortran/resolve.c:16494 0x8ad424 resolve_all_program_units /data/midas/foreese/src/gcc-trunk/gcc/fortran/parse.c:6042 0x8ad424 gfc_parse_file() /data/midas/foreese/src/gcc-trunk/gcc/fortran/parse.c:6292 0x8f326f gfc_be_parse_file /data/midas/foreese/src/gcc-trunk/gcc/fortran/f95-lang.c:204 The second bug is the out-of-memory issue from within match_clist_expr().