Dear All,
The attached patch allows the original testcase to compile but
attempting to use it runs into segfaults at runtime. The initializer
for 'decay_t' has a bad FIELD_DECL for 'decay_gen_t' since it lacks
the DECL_SIZE(_UNIT) fields but the TYPE_SIZE(_UNIT) fields are OK.
The patch runs through the constructor and restores the missing
fields, thus preventing the ICE in varasm.c.
Taking the reduced testcase of comment #4 and adding a main program:
module decays
implicit none
interface
real elemental function iface (arg)
real, intent(in) :: arg
end function
end interface
type :: decay_term_t
type(decay_t), pointer :: unstable_product
end type
type :: decay_gen_t
procedure(iface), nopass, pointer :: obs1_int
type(decay_term_t), allocatable :: term
end type
type :: rng_t
integer :: i
end type
type, extends (decay_gen_t) :: decay_t
class(rng_t), allocatable :: rng
end type
class(decay_t), allocatable :: object
end
use decays
type(decay_t), pointer :: template
allocate (template)
allocate (template%rng)
template%obs1_int => cos
print *, template%obs1_int (3.14159/2.)
allocate (object, source = template)
! print *, object%obs1_int (3.14159/2.) ! This should be OK but it
segfaults at runtime.
object%obs1_int => sin
print *, object%obs1_int (3.14159/2.)
end
This runs correctly. However, the commented out line causes a
segfault in runtime. I haven't attempted yet to understand why this
should be the case but it is obvious that the _copy procedure for
decay_t has something wrong with it.
The patch, therefore is neither ready in principle to be committed and
uncouvers other nasties. I have two questions that I hope that
somebody can answer:
(i) How or why is it possible for build_constructor to produce
incomplete FIELD_DECLS?; and
(ii) What is wrong with the _copy procedure?
I cannot return to the fray for slightly more than a week.
Best regards
Paul
Index: gcc/fortran/trans-expr.c
===================================================================
*** gcc/fortran/trans-expr.c (revision 208048)
--- gcc/fortran/trans-expr.c (working copy)
*************** gfc_conv_structure (gfc_se * se, gfc_exp
*** 6175,6180 ****
--- 6175,6201 ----
se->expr = build_constructor (type, v);
if (init)
TREE_CONSTANT (se->expr) = 1;
+
+ /* Verify that the DECL_SIZE fields have been set so that the
+ constructor is ready for varasm.c(output_constructor_regular_field).
+ TODO: Find out why this is very occasionally necessary.
+ See PR59198. */
+ type = TREE_TYPE (se->expr);
+ if (TREE_CODE (type) == RECORD_TYPE)
+ {
+ tmp = TYPE_FIELDS (type);
+ while (tmp)
+ {
+ if (TREE_CODE (tmp) == FIELD_DECL
+ && DECL_SIZE_UNIT (tmp) == NULL_TREE)
+ {
+ DECL_SIZE (tmp) = TYPE_SIZE (TREE_TYPE (tmp));
+ DECL_SIZE_UNIT (tmp) = TYPE_SIZE_UNIT (TREE_TYPE (tmp));
+ }
+ tmp = TREE_CHAIN (tmp);
+ }
+ }
+
}
