http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59572
Bug ID: 59572
Summary: Not clear error message in smallest_mode_for_size in
handled case
Product: gcc
Version: 4.8.2
Status: UNCONFIRMED
Severity: enhancement
Priority: P3
Component: middle-end
Assignee: unassigned at gcc dot gnu.org
Reporter: christophe.curis at free dot fr
Hello,
I have received this Internal Compiler error:
----------------------------------------------------------------------------
<built-in>:0:0: internal compiler error: in smallest_mode_for_size, at
stor-layout.c:384
0x82ea77 smallest_mode_for_size
../../../src/gcc-4.8.2/gcc/stor-layout.c:384
0x831d94 smallest_mode_for_size
../../../src/gcc-4.8.2/gcc/stor-layout.c:495
0x831d94 layout_type(tree_node*)
../../../src/gcc-4.8.2/gcc/stor-layout.c:2081
0x8320b0 make_signed_type(int)
../../../src/gcc-4.8.2/gcc/stor-layout.c:2393
0x4da44d type_for_size
../../../src/gcc-4.8.2/gcc/vhdl/ortho-lang.c:649
0x4da4a6 type_for_mode
../../../src/gcc-4.8.2/gcc/vhdl/ortho-lang.c:669
0x9a5a14 build_common_builtin_nodes()
../../../src/gcc-4.8.2/gcc/tree.c:9795
0x4d9caa ortho_init
../../../src/gcc-4.8.2/gcc/vhdl/ortho-lang.c:328
0x4e67a6 lang_dependent_init
../../../src/gcc-4.8.2/gcc/toplev.c:1688
0x4e67a6 do_compile
../../../src/gcc-4.8.2/gcc/toplev.c:1850
----------------------------------------------------------------------------
I am trying to build gcc with VHDL support from GHDL:
http://sourceforge.net/projects/ghdl-updates/
This work quite well, except that in current case I am trying to build a
cross-compiler (--target=m68k-linux-gnu). All compilation flow works well until
this step:
----------------------------------------------------------------------------
make -C vhdl [...] ghdllib
../ghdl1 --std=87 -quiet -o std_standard.s --compile-standard
----------------------------------------------------------------------------
My current supposition is that while compiling the standards for VHDL
languages, it encouters a type that is not supported by the current arch I am
targetting (maybe a 64-bits word, or a floating point type), so if gcc's
smallest_mode_for_size function was able to provide a little bit of info on the
type it was being called for, it would help to investigate how to provide a
solution for it.
Regards,
Christophe.
