https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78472
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |lto, wrong-code
Status|UNCONFIRMED |ASSIGNED
Last reconfirmed| |2016-11-22
CC| |hubicka at gcc dot gnu.org
Assignee|unassigned at gcc dot gnu.org |rguenth at gcc dot
gnu.org
Ever confirmed|0 |1
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed. The C variant has a FIELD_DECL for the :0 while the C++ variant
does not have that. This makes them not share a canonical type.
Thus this is a possible wrong-code C/C++ interoperability issue.
(gdb) p debug_tree ((tree)0x7ffff6a6c1c8)
<field_decl 0x7ffff6a6c1c8 D.3900
type <integer_type 0x7ffff6a6dc78 unsigned QI
size <integer_cst 0x7ffff6887ca8 constant 8>
unit size <integer_cst 0x7ffff6887cc0 constant 1>
align 8 symtab 0 alias set -1 canonical type 0x7ffff6a6dc78 precision
0>
unsigned external bit-field nonaddressable QI file x.h line 4 col 1
size <integer_cst 0x7ffff6887c30 type <integer_type 0x7ffff688b2a0
bitsizetype> constant 0>
unit size <integer_cst 0x7ffff6887be8 type <integer_type 0x7ffff688b1f8
sizetype> constant 0>
align 32 offset_align 128 offset <integer_cst 0x7ffff6887be8 0>
so the field has zero size. Ignoring zero-sized fields in canonical type
hashing/comparing works. Testing a patch.
