https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71703
Pierre de Buyl <bugzilla at pdebuyl dot be> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |bugzilla at pdebuyl dot be
--- Comment #8 from Pierre de Buyl <bugzilla at pdebuyl dot be> ---
Hello,
I don't know if this is related (I am new to bug submission here).
Test case:
program test_ice
implicit none
type test_t
integer, allocatable :: a(:)
end type test_t
type(test_t) :: v
integer, parameter :: n = 100
allocate(v%a(n))
v%a = 0
end program test_ice
compiled with gfortran 5.3.1 and the "-fcheck=mem" flag generates this output
test_ice.f90:13:0:
v%a = 0
1
internal compiler error: in wide_int_to_tree, at tree.c:1464
0x968950 wide_int_to_tree(tree_node*,
generic_wide_int<wide_int_ref_storage<false> > const&)
../../src/gcc/tree.c:1464
0x10a6543 build_int_cst(tree_node*, long)
../../src/gcc/tree.c:1272
0x64e9d3 gfc_trans_assignment_1
../../src/gcc/fortran/trans-expr.c:9146
0xd48b00 trans_code
../../src/gcc/fortran/trans.c:1711
0xd48b00 gfc_trans_code(gfc_code*)
../../src/gcc/fortran/trans.c:2020
0xd4d522 gfc_generate_function_code(gfc_namespace*)
../../src/gcc/fortran/trans-decl.c:5941
0xd2b045 translate_all_program_units
../../src/gcc/fortran/parse.c:5406
0xd2b045 gfc_parse_file()
../../src/gcc/fortran/parse.c:5603
0x11f8615 gfc_be_parse_file
../../src/gcc/fortran/f95-lang.c:229
Switching to gfortran 4.8.4 or 4.9.2, or removing "-fcheck=all" allows for
compilation and execution.
Also, adding "(:)" to the member "a" as
v%a(:) = 0
also compiles with "-fcheck=all" on gfortran 5.3.1
I found this issue while trying a "-fcheck=all" compilation on a larger
codebase and reduced the bug to the example above.
Is this of interest here?
Anyway, thanks to everyone working on gcc and gfortran!
Regards,
Pierre
