Hi All,
This is a regression caused by my patch for PR109345 - r15-5083.
I overlooked the possibility that the unlimited polymorphic container might
be the component of a dummy derived type. The fix is simple and relatively
obvious.
Regards
Paul
Fortran: Fix regression caused by r15-5083 [PR117797]
2024-12-01 Paul Thomas <pa...@gcc.gnu.org>
gcc/fortran
PR fortran/117797
* trans-array.cc (gfc_get_array_span): Prevent unlimited poly-
morphic dummy variable expressions, if the array is a component
of a derived type, from using GFC_DECL_SAVED_DESCRIPTOR for the
class container. Use a gcc_assert to verify the presence of the
GFC_DECL_SAVED_DESCRIPTOR.
gcc/testsuite/
PR fortran/117797
* gfortran.dg/pr117797.f90 : New test.
diff --git a/gcc/fortran/trans-array.cc b/gcc/fortran/trans-array.cc
index a458af322ce..b874d5bbfa9 100644
--- a/gcc/fortran/trans-array.cc
+++ b/gcc/fortran/trans-array.cc
@@ -991,9 +991,11 @@ gfc_get_array_span (tree desc, gfc_expr *expr)
the element size need not be the same as the span. Obtain
the class container, which is simplified here by there being
no component references. */
- if (sym && sym->attr.dummy)
+ if (sym && sym->attr.dummy && TREE_CODE (desc) != COMPONENT_REF)
{
tmp = gfc_get_symbol_decl (sym);
+ gcc_assert (DECL_LANG_SPECIFIC (tmp)
+ && GFC_DECL_SAVED_DESCRIPTOR (tmp));
tmp = GFC_DECL_SAVED_DESCRIPTOR (tmp);
if (INDIRECT_REF_P (tmp))
tmp = TREE_OPERAND (tmp, 0);
