This one is straightforward.
See attached.
Regression tested on x86_64.
OK for mainline? and then backport to 16.
Regards,
Jerry
---
fortran: ICE in DO CONCURRENT with DEFAULT(NONE) inside ASSOCIATE
Two bugs in check_default_none_expr caused a segfault when a DO CONCURRENT
with inline type-spec iterators (e.g. "integer :: i = 1:10") contained an
ASSOCIATE construct.
Bug 1: sym->ns->code was used to locate the ext.concur.forall_iterator
list. When a symbol's namespace is an ASSOCIATE body, sym->ns->code is an
EXEC_BLOCK node, not the DO CONCURRENT node; reading ext.concur from it
interprets the wrong union member and yields a garbage pointer. Fix: use
d->code instead, the DO CONCURRENT gfc_code node passed through the walker's
data parameter, which is always the correct node.
Bug 2: inline type-spec iterators are shadow iterators, stored internally
with a leading underscore prepended to the name. The comparison of the
iterator's symtree name against the user-visible sym->name must skip that
underscore by advancing iter_name one character when iter->shadow is set.
PR fortran/125529
Assisted by: Claude Sonnet 4.6
gcc/fortran/ChangeLog:
PR fortran/125529
* resolve.cc (check_default_none_expr): Use d->code instead of
sym->ns->code to locate the DO CONCURRENT forall_iterator list,
avoiding a wrong-union-member read when the symbol's namespace is
an ASSOCIATE body. Skip leading underscore when comparing iterator
names for shadow iterators.
gcc/testsuite/ChangeLog:
PR fortran/125529
* gfortran.dg/do_concurrent_assoc_default_none.f90: New test.From d184c3c93f3de6132b8ac466b3ecb0f0a33dda4c Mon Sep 17 00:00:00 2001
From: Jerry DeLisle <[email protected]>
Date: Wed, 27 May 2026 21:00:19 -0700
Subject: [PATCH] fortran: ICE in DO CONCURRENT with DEFAULT(NONE) inside
ASSOCIATE
Two bugs in check_default_none_expr caused a segfault when a DO CONCURRENT
with inline type-spec iterators (e.g. "integer :: i = 1:10") contained an
ASSOCIATE construct.
Bug 1: sym->ns->code was used to locate the ext.concur.forall_iterator
list. When a symbol's namespace is an ASSOCIATE body, sym->ns->code is an
EXEC_BLOCK node, not the DO CONCURRENT node; reading ext.concur from it
interprets the wrong union member and yields a garbage pointer. Fix: use
d->code instead, the DO CONCURRENT gfc_code node passed through the walker's
data parameter, which is always the correct node.
Bug 2: inline type-spec iterators are shadow iterators, stored internally
with a leading underscore prepended to the name. The comparison of the
iterator's symtree name against the user-visible sym->name must skip that
underscore by advancing iter_name one character when iter->shadow is set.
PR fortran/125529
Assisted by: Claude Sonnet 4.6
gcc/fortran/ChangeLog:
PR fortran/125529
* resolve.cc (check_default_none_expr): Use d->code instead of
sym->ns->code to locate the DO CONCURRENT forall_iterator list,
avoiding a wrong-union-member read when the symbol's namespace is
an ASSOCIATE body. Skip leading underscore when comparing iterator
names for shadow iterators.
gcc/testsuite/ChangeLog:
PR fortran/125529
* gfortran.dg/do_concurrent_assoc_default_none.f90: New test.
---
gcc/fortran/resolve.cc | 24 ++++++++-----
.../do_concurrent_assoc_default_none.f90 | 35 +++++++++++++++++++
2 files changed, 51 insertions(+), 8 deletions(-)
create mode 100644 gcc/testsuite/gfortran.dg/do_concurrent_assoc_default_none.f90
diff --git a/gcc/fortran/resolve.cc b/gcc/fortran/resolve.cc
index df91671e171..b9078fd8722 100644
--- a/gcc/fortran/resolve.cc
+++ b/gcc/fortran/resolve.cc
@@ -8597,15 +8597,23 @@ check_default_none_expr (gfc_expr **e, int *, void *data)
ns2 = ns2->parent;
}
- /* A DO CONCURRENT iterator cannot appear in a locality spec. */
- if (sym->ns->code->ext.concur.forall_iterator)
+ /* A DO CONCURRENT iterator cannot appear in a locality spec.
+ Use d->code (the DO CONCURRENT node) rather than sym->ns->code,
+ which may be a different code type (e.g. EXEC_ASSOCIATE) whose
+ ext union would be read incorrectly. */
+ for (gfc_forall_iterator *iter = d->code->ext.concur.forall_iterator;
+ iter; iter = iter->next)
{
- gfc_forall_iterator *iter
- = sym->ns->code->ext.concur.forall_iterator;
- for (; iter; iter = iter->next)
- if (iter->var->symtree
- && strcmp(sym->name, iter->var->symtree->name) == 0)
- return 0;
+ if (!iter->var || !iter->var->symtree)
+ continue;
+ const char *iter_name = iter->var->symtree->name;
+ /* Shadow iterators (from inline type-spec: integer :: i = ...)
+ store the iterator with a leading underscore internally; the
+ user-visible name does not have the underscore. */
+ if (iter->shadow)
+ iter_name++;
+ if (strcmp (sym->name, iter_name) == 0)
+ return 0;
}
/* A named constant is not a variable, so skip test. */
diff --git a/gcc/testsuite/gfortran.dg/do_concurrent_assoc_default_none.f90 b/gcc/testsuite/gfortran.dg/do_concurrent_assoc_default_none.f90
new file mode 100644
index 00000000000..f1cb2718907
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/do_concurrent_assoc_default_none.f90
@@ -0,0 +1,35 @@
+! { dg-do compile }
+! { dg-options "-std=f2018" }
+!
+! PR fortran/125515
+! DO CONCURRENT with inline type-spec iterators and DEFAULT(NONE) inside an
+! ASSOCIATE block used to ICE (segfault in check_default_none_expr). The
+! bug was that sym->ns->code was used instead of d->code to locate the
+! forall_iterator list; for symbols in an ASSOCIATE or enclosing procedure
+! namespace the wrong union member was read, giving a garbage pointer.
+! Additionally, shadow iterators store their name with a leading underscore
+! internally, so the comparison against user-visible names must strip it.
+
+subroutine test_associate_default_none (a, b)
+ implicit none
+ integer, intent(in) :: a(10,10)
+ integer, intent(out) :: b(10,10)
+ do concurrent (integer :: i = 1:10, j = 1:10) &
+ default(none) shared(a, b)
+ associate (tmp => a(i,j))
+ b(i,j) = tmp + 1
+ end associate
+ end do
+end subroutine
+
+! Verify that a genuinely missing variable still gets an error.
+subroutine test_missing_var (a, b, c)
+ implicit none
+ integer, intent(in) :: a(10,10), c
+ integer, intent(out) :: b(10,10)
+ do concurrent (integer :: i = 1:10, j = 1:10) &
+ default(none) shared(a, b)
+ ! { dg-error "not specified in a locality spec" "" { target *-*-* } .-1 }
+ b(i,j) = a(i,j) + c ! { dg-error "not specified in a locality spec" }
+ end do
+end subroutine
--
2.54.0
