https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118705
Paul Thomas <pault at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Ever confirmed|0 |1
Summary|Issue compiling derived |[17 regression] Issue
|types when split into |compiling derived types
|submodules |when split into submodules
CC| |pault at gcc dot gnu.org
Status|UNCONFIRMED |NEW
Last reconfirmed| |2026-06-07
--- Comment #4 from Paul Thomas <pault at gcc dot gnu.org> ---
This is fixed in gcc-16. Regression caused by patch for pr125527.
This PR is cured by:
index a4c3c37104a..3f10ba26c3c 100644
--- a/gcc/fortran/class.cc
+++ b/gcc/fortran/class.cc
@@ -2491,6 +2491,7 @@ gfc_find_derived_vtab (gfc_symbol *derived)
/* Work in the gsymbol namespace if the top-level namespace is a module.
This ensures that the vtable is unique, which is required since we use
its address in SELECT TYPE. */
+ gfc_namespace *module_ns = ns;
if (gsym && gsym->ns && ns && ns->proc_name
&& ns->proc_name->attr.flavor == FL_MODULE)
ns = gsym->ns;
@@ -2516,6 +2517,11 @@ gfc_find_derived_vtab (gfc_symbol *derived)
gfc_find_symbol (name, ns, 0, &vtab);
if (vtab == NULL)
gfc_find_symbol (name, derived->ns, 0, &vtab);
+ /* Also look in the original top-level module/submodule namespace so
that
+ the module variable translation and the procedure body translation
+ find the same frontend symbol and thus the same backend decl. */
+ if (vtab == NULL && module_ns != ns)
+ gfc_find_symbol (name, module_ns, 0, &vtab);
if (vtab == NULL)
{
