https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67982
--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
> Maybe r224201 ?
It is
Author: hubicka
Date: Sun Jun 7 21:30:58 2015 UTC (4 months, 1 week ago)
Changed paths: 13 (showing only 10; show all)
Log Message:
* alias.c (get_alias_set): Be ready for TYPE_CANONICAL
of ptr_type_node to not be ptr_to_node.
* tree.c (gimple_types_compatible_p): Do not match TREE_CODE of
TREE_TYPE of pointers.
* gimple-expr.c (useless_type_conversion): Reorder the check for
function pointers and TYPE_CANONICAL.
* lto.c (hash_canonical_type): Do not hash TREE_CODE of TREE_TYPE of
pointers.
* gfortran.dg/lto/bind_c-1_0.f90: New testcase.
* gfortran.dg/lto/bind_c-1_1.c: New testcase.
* gcc.dg/lto/c-compatible-types_0.c: Rename to ...
* gcc.dg/lto/c-compatible-types-1_0.c: this one; fix template
* gcc.dg/lto/c-compatible-types_1.c: Rename to ...
* gcc.dg/lto/c-compatible-types-1_1.c: this one; harden for
-fshort-enum.
so I doubt it.
> Otherwise this might need bisection.
Be my guest!
