https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124450
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
2224 /* Now se.expr contains an element of the array. Take the address
and pass
2225 it to the IO routines. */
2226 tmp = gfc_build_addr_expr (NULL_TREE, se.expr);
2227 transfer_expr (&se, &cm->ts, tmp, NULL, NULL_TREE);
so code is NULL in some cases.
--- gcc/fortran/trans-io.cc.jj 2026-01-08 09:44:44.078248379 +0100
+++ gcc/fortran/trans-io.cc 2026-03-11 15:33:59.334057907 +0100
@@ -2513,7 +2513,9 @@ transfer_expr (gfc_se * se, gfc_typespec
if (c->attr.dimension)
{
- tmp = transfer_array_component (tmp, c, & code->loc);
+ tmp = transfer_array_component (tmp, c,
+ code ? &code->loc
+ : NULL);
gfc_add_expr_to_block (&se->pre, tmp);
}
else
would fix this (but I'm really surprised the tons of other code dereferences
don't trigger,
if (derived->attr.has_dtio_procs)
arg2 = get_dtio_proc (ts, code, &dtio_sub);
which dereferences it at the start of get_dtio_proc,
gfc_conv_derived_to_class (se, code->expr1,
dtio_sub->formal->sym, vptr, false,
false, "transfer");
or the earlier
if (ts->type == BT_VOID
&& code->expr1 && code->expr1->ts.type == BT_VOID
