In taking a look at PR fortran/77632, I stumbled acrossed
the testcase in the attached patch cause an ICE. I originally
thought that it was realated to the topic of the PR, but is
in fact an unrelated bug.
If a variable is a target in a pointer initialization, then
it must have the SAVE attribute. A variable in PROGRAM,
MODULE, or SUBMODULE scope is implicitly SAVEd. So, the
patch explicitly sets the save attribute to SAVE_IMPLICIT.
2019-06-20 Steven G. Kargl <ka...@gcc.gnu.org>
PR fortran/77632
* /decl.c (variable_decl): Mark a variable that is a target in pointer
initialization when in PROGRAM, MODULE, or SUBMODULE scope with an
implicit save.
2019-06-20 Steven G. Kargl <ka...@gcc.gnu.org>
PR fortran/77632
* gfortran.dg/pr77632_1.f90: New test.
--
Steve
Index: gcc/fortran/decl.c
===================================================================
--- gcc/fortran/decl.c (revision 272523)
+++ gcc/fortran/decl.c (working copy)
@@ -2779,6 +2779,16 @@ variable_decl (int elem)
m = match_pointer_init (&initializer, 0);
if (m != MATCH_YES)
goto cleanup;
+
+ /* The target of a pointer initialization must have the SAVE
+ attribute. A variable in PROGRAM, MODULE, or SUBMODULE scope
+ is implicit SAVEd. Explicitly, set the SAVE_IMPLICIT value. */
+ if (initializer->expr_type == EXPR_VARIABLE
+ && initializer->symtree->n.sym->attr.save == SAVE_NONE
+ && (gfc_current_state () == COMP_PROGRAM
+ || gfc_current_state () == COMP_MODULE
+ || gfc_current_state () == COMP_SUBMODULE))
+ initializer->symtree->n.sym->attr.save = SAVE_IMPLICIT;
}
else if (gfc_match_char ('=') == MATCH_YES)
{
Index: gcc/testsuite/gfortran.dg/pr77632_1.f90
===================================================================
--- gcc/testsuite/gfortran.dg/pr77632_1.f90 (nonexistent)
+++ gcc/testsuite/gfortran.dg/pr77632_1.f90 (working copy)
@@ -0,0 +1,7 @@
+! { dg-do run }
+program foo
+ implicit none
+ real, target :: a
+ real, pointer :: b => a
+ if (associated(b, a) .eqv. .false.) stop 1
+end program foo
