[Bug default/28641] unspecified indirect subtype changes.

["woodard at redhat dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=28641

--- Comment #8 from Ben Woodard <woodard at redhat dot com> ---
Attached a very simple reproducer:

[ben@alien libabigail-test]$ cat llvm-constness.C
int a=0;

struct X {
  void foo();
  void foo() const;
};

void X::foo(){
  a++;
}

void X::foo() const {
  a--;
}
[ben@alien libabigail-test]$ glto/bin/abidiff llvm-constness.o.gcc
llvm-constness.o.llvm 
Functions changes summary: 0 Removed, 1 Changed, 0 Added function
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

1 function with some indirect sub-type change:

  [C] 'method void X::foo() const' at llvm-constness.C:5:1 has some indirect
sub-type changes:

Here is the DWARF breakdown:

gcc:
 [    72]      subprogram           abbrev: 10
               external             (flag_present) yes
               name                 (string) "foo"
               decl_file            (data1) llvm-constness.C (1)
               decl_line            (data1) 5
               decl_column          (data1) 8
               linkage_name         (strp) "_ZNK1X3fooEv"
               declaration          (flag_present) yes
               object_pointer       (ref4) [    82]
 [    82]        formal_parameter     abbrev: 2
                 type                 (ref4) [    98]
                 artificial           (flag_present) yes
 [    98]    pointer_type         abbrev: 3
             byte_size            (implicit_const) 8
             type                 (ref4) [    89]
 [    89]    const_type           abbrev: 1
             type                 (ref4) [    4d]
 [    4d]    structure_type       abbrev: 8
             name                 (string) "X"
             byte_size            (data1) 1
             decl_file            (data1) llvm-constness.C (1)
             decl_line            (data1) 3
             decl_column          (data1) 8
             sibling              (ref4) [    89]

clang++:
 [    60]      subprogram           abbrev: 5
               linkage_name         (strp) "_ZNK1X3fooEv"
               name                 (strp) "foo"
               decl_file            (data1) llvm-constness.C (1)
               decl_line            (data1) 5
               declaration          (flag_present) yes
               external             (flag_present) yes
 [    6b]        formal_parameter     abbrev: 6
                 type                 (ref4) [    77]
                 artificial           (flag_present) yes
 [    77]    pointer_type         abbrev: 7
             type                 (ref4) [    7c]
 [    7c]    const_type           abbrev: 8
             type                 (ref4) [    46]
 [    46]    structure_type       abbrev: 4
             calling_convention   (data1) pass_by_value (5)
             name                 (strp) "X"
             byte_size            (data1) 1
             decl_file            (data1) llvm-constness.C (1)
             decl_line            (data1) 3

Which looks pretty much the same and basically correct. I think that the first
time I looked at it I mistook the implicit_const on GCC's byte_size as the
thing that was different so there may not be a clang bug just a libabigail bug
where for some reason it doesn't pick up the fact that the this pointer is
const which seems to be the thing that indicates that the member function is
const.

-- 
You are receiving this mail because:
You are on the CC list for the bug.