https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82993
kargl at gcc dot gnu.org changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |kargl at gcc dot gnu.org
--- Comment #3 from kargl at gcc dot gnu.org ---
(In reply to G. Steinmetz from comment #0)
> With a wrong parameter definition :
>
> $ cat z1.f90
> program p
> type t
> real :: a(3)
> end type
> type(t), parameter :: z = 1
> print *, z%a(1)
> print *, z%a
> end
>
>
> $ gfortran-8-20171112 -c z1.f90
> f951: internal compiler error: Segmentation fault
> 0xb61c7f crash_signal
> ../../gcc/toplev.c:325
> 0x69268b free_expr0
> ../../gcc/fortran/expr.c:445
> 0x6926ad gfc_free_expr(gfc_expr*)
> ../../gcc/fortran/expr.c:518
> 0x6efd59 gfc_match_rvalue(gfc_expr**)
> ../../gcc/fortran/primary.c:3250
> 0x6c655e match_primary
> ../../gcc/fortran/matchexp.c:157
> 0x6c655e match_level_1
> ../../gcc/fortran/matchexp.c:211
> 0x6c655e match_mult_operand
> ../../gcc/fortran/matchexp.c:267
> 0x6c67a8 match_add_operand
> ../../gcc/fortran/matchexp.c:356
> 0x6c6a3c match_level_2
> ../../gcc/fortran/matchexp.c:480
> 0x6c6b92 match_level_3
> ../../gcc/fortran/matchexp.c:551
> 0x6c6ca4 match_level_4
> ../../gcc/fortran/matchexp.c:599
> 0x6c6ca4 match_and_operand
> ../../gcc/fortran/matchexp.c:693
> 0x6c6e62 match_or_operand
> ../../gcc/fortran/matchexp.c:722
> 0x6c6f52 match_equiv_operand
> ../../gcc/fortran/matchexp.c:765
> 0x6c7042 match_level_5
> ../../gcc/fortran/matchexp.c:811
> 0x6c63b1 gfc_match_expr(gfc_expr**)
> ../../gcc/fortran/matchexp.c:870
> 0x6af7a9 match_io_element
> ../../gcc/fortran/io.c:3542
> 0x6af9f3 match_io_list
> ../../gcc/fortran/io.c:3581
> 0x6b4214 match_io
> ../../gcc/fortran/io.c:4242
> 0x6b592a gfc_match_print()
> ../../gcc/fortran/io.c:4298
Ugh. This one is real ugly. :(
If both print statement are commented out, one gets
% gfcx -c a.f90
a.f90:5:28:
type(t), parameter :: z = 1
1
Error: Incompatible derived type in PARAMETER at (1)
a.f90:5:28:
type(t), parameter :: z = 1
1
Error: Can't convert INTEGER(4) to TYPE(t) at (1)
This is the good news!
If the first print statement is uncomment, we get Gerhard's
trace above. If the first print is again commented out and
the second statement is uncommented, we get
% gfcx -c a.f90
a.f90:5:28:
type(t), parameter :: z = 1
1
Error: Incompatible derived type in PARAMETER at (1)
a.f90:5:28:
type(t), parameter :: z = 1
1
Error: Can't convert INTEGER(4) to TYPE(t) at (1)
(null):0: confused by earlier errors, bailing out
AFAICT, when the statement "type(t), parameter :: z = 1"
is rejected, the gfc_expr for z = 1 is not properly freed.
I see
(gdb) p *p
$8 = {op = EXEC_ASSIGN, block = 0x0, next = 0x0, loc = {nextc = 0x0,
lb = 0x0}, here = 0x0, label1 = 0x0, label2 = 0x0, label3 = 0x0,
symtree = 0x0, expr1 = 0x201d837e0, expr2 = 0x0, expr3 = 0x0, expr4 = 0x0,
resolved_sym = 0x0, resolved_isym = 0x0, ext = {actual = 0x0,
iterator = 0x0, alloc = {ts = {type = BT_UNKNOWN, kind = 0, u = {
derived = 0x0, cl = 0x0, pad = 0}, interface = 0x0,
is_c_interop = 0, is_iso_c = 0, f90_type = BT_UNKNOWN,
deferred = false, interop_kind = 0x0}, list = 0x0,
arr_spec_from_expr3 = 0}, block = {ns = 0x0, assoc = 0x0,
case_list = 0x0}, open = 0x0, close = 0x0, filepos = 0x0, inquire = 0x0,
wait = 0x0, dt = 0x0, forall_iterator = 0x0, which_construct = 0x0,
stop_code = 0, entry = 0x0, oacc_declare = 0x0, omp_clauses = 0x0,
omp_name = 0x0, omp_namelist = 0x0, omp_bool = false,
omp_atomic = GFC_OMP_ATOMIC_UPDATE}, cycle_label = 0x0, exit_label = 0x0}
(gdb) p *p->expr1
$3 = {expr_type = EXPR_FUNCTION, ts = {type = BT_UNKNOWN, kind = 0, u = {
derived = 0x0, cl = 0x0, pad = 0}, interface = 0x0, is_c_interop = 0,
is_iso_c = 0, f90_type = BT_UNKNOWN, deferred = false,
interop_kind = 0x0}, rank = 0, shape = 0x0, symtree = 0x201d805d0,
ref = 0x0, where = {nextc = 0x201dc50cc, lb = 0x201dc50a0}, base_expr = 0x0,
is_boz = 0, is_snan = 0, error = 0, user_operator = 0, mold = 0,
must_finalize = 0, representation = {length = 0, string = 0x0}, value = {
logical = 30934528, iokind = 30934528, integer = {{_mp_alloc = 30934528,
_mp_size = 2, _mp_d = 0x0}}, real = {{_mpfr_prec = 8620869120,
_mpfr_sign = 0, _mpfr_exp = 0, _mpfr_d = 0x0}}, complex = {{re = {{
_mpfr_prec = 8620869120, _mpfr_sign = 0, _mpfr_exp = 0,
_mpfr_d = 0x0}}, im = {{_mpfr_prec = 0, _mpfr_sign = 0,
_mpfr_exp = 0, _mpfr_d = 0x0}}}}, op = {op = 30934528, uop = 0x0,
op1 = 0x0, op2 = 0x0}, function = {actual = 0x201d80600, name = 0x0,
isym = 0x0, esym = 0x0}, compcall = {actual = 0x201d80600, name = 0x0,
base_object = 0x0, tbp = 0x0, ignore_pass = 0, assign = 0}, character = {
length = 30934528, string = 0x0}, constructor = 0x201d80600},
where p->expr1 loos like garbage on the stack.
