https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82721
Thomas Koenig <tkoenig at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |tkoenig at gcc dot gnu.org
--- Comment #7 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
Valgrind reports the problem as
asdf.f90:3:25:
3 | character(len(c)) :: b
| 1
Fehler: Symbol »b« bei (1) hat bereits grundlegenden Typen REAL
==28874== Invalid read of size 8
==28874== at 0x8CD014: check_host_association (resolve.c:5756)
==28874== by 0x8CD014: gfc_resolve_expr(gfc_expr*) (resolve.c:6879)
==28874== by 0x8D1A98: resolve_index_expr(gfc_expr*) (resolve.c:11927)
==28874== by 0x8D1BD4: resolve_charlen(gfc_charlen*) [clone .part.0]
(resolve.c:11979)
..
==28874== Address 0x5872a18 is 40 bytes inside a block of size 48 free'd
==28874== at 0x4C2B390: free (vg_replace_malloc.c:530)
==28874== by 0x8EF361: gfc_delete_symtree(gfc_symtree**, char const*)
(symbol.c:2962)
==28874== by 0x8F09F1: gfc_restore_last_undo_checkpoint() (symbol.c:3714)
==28874== by 0x8ABE87: reject_statement() (parse.c:2576)
==28874== by 0x8AC9C1: match_word (parse.c:70)
==28874== by 0x8AC9C1: decode_statement() (parse.c:376)
...
Looking at where the error is issued, gfc_add_type notices that
there already is a symbol, hands this up to build_sym, which returns
false to variable_decl, and when things finally go up all the way
to match_word, reject_statement() deletes the symtree.
