[Bug default/19097] ABI identity problem insufficient data stored for _M_buf_locale type

[Bug default/19097] ABI identity problem insufficient data stored for _M_buf_locale type

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19097

dodji at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from dodji at redhat dot com ---
That assertion violation should now be fixed by a set of patches in the master
branch, the last one being
https://sourceware.org/git/gitweb.cgi?p=libabigail.git;a=commit;h=14ff32b322ebb127d7c281eaf703314964e32cec.

Thank you for reporting this problem.

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

[Bug default/19097] ABI identity problem insufficient data stored for _M_buf_locale type

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19097

--- Comment #2 from Ben Woodard <woodard at redhat dot com> ---
This one is currently hitting the n assertion.

hype356@ben:~/bin/abidw --abidiff
/collab/usr/global/tools/order/spack/opt/chaos_5_x86_64_ib/gcc@4.4.7/gcc@4.7.4-93774c82/lib64/libstdc++.so.6.0.17
abidw: ../../src/abg-reader.cc:1031: abigail::ir::type_base_sptr
abigail::xml_reader::read_context::build_or_get_type_decl(const std::string&,
bool): Assertion `n' failed.
Aborted (core dumped)

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

[Bug default/19097] ABI identity problem insufficient data stored for _M_buf_locale type

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19097

--- Comment #1 from Ben Woodard <woodard at redhat dot com> ---
I'm not 100% certain about this but I think that this could be the same
problem. Once again abidw doesn't print enough about what changed in the class
for me to tell what attribute about the class has changed. For this one, when
you fix the problem with _M_buf_locale then I'll also retest against libgo

hype260@ben:~/bin/abidw --abidiff
/collab/usr/global/tools/order/spack/opt/chaos_5_x86_64_ib/gcc@4.4.7/gcc@4.9.2-93774c82/lib64/libgo.so.5.0.0
2>&1 | head -30
Functions changes summary: 0 Removed, 201 Changed, 0 Added functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

201 functions with some indirect sub-type change:

  [C]'function error ast.Fprint(Writer, FileSet*, __anonymous_struct__,
FieldFilter)' has some indirect sub-type changes:
    parameter 2 of type 'FileSet*' has sub-type changes:
      in pointed to type 'typedef FileSet':
        underlying type 'struct __anonymous_struct__' changed:
          2 data member changes:
           type of '__anonymous_struct__ __anonymous_struct__::files' changed:
             1 data member change:
              type of 'File** __anonymous_struct__::__values' changed:
                in pointed to type 'File*':
                  in pointed to type 'typedef File':
                    underlying type 'struct __anonymous_struct__' changed:
                      1 data member change:
                       type of 'FileSet* __anonymous_struct__::set' changed:
                         in pointed to type 'typedef FileSet':
                           underlying type 'struct __anonymous_struct__'
changed; details are being reported

...

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

[Bug default/19097] New: ABI identity problem insufficient data stored for _M_buf_locale type

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=19097

            Bug ID: 19097
           Summary: ABI identity problem insufficient data stored for
                    _M_buf_locale type
           Product: libabigail
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: default
          Assignee: dodji at redhat dot com
          Reporter: woodard at redhat dot com
                CC: libabigail at sourceware dot org
  Target Milestone: ---

Created attachment 8692
  --> https://sourceware.org/bugzilla/attachment.cgi?id=8692&action=edit
File that reproduces the problem

It looks like some important detail about _M_buf_locale is being lost and that
creates a huge number of false positives regarding changes to the ABI in
libstdc++

ype260@ben:~/bin/abidw --abidiff
/collab/usr/global/tools/order/spack/opt/chaos_5_x86_64_ib/gcc@4.4.7/gcc@4.7.4-93774c82/lib64/libstdc++.so.6.0.17
2>&1 | head -40
Functions changes summary: 0 Removed, 119 Changed, 0 Added functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

119 functions with some indirect sub-type change:

  [C]'method virtual std::basic_filebuf<char, std::char_traits<char>
>::__streambuf_type* std::basic_filebuf<char, std::char_traits<char>
>::setbuf(std::basic_filebuf<char, std::char_traits<char> >::char_type*,
std::streamsize)' has some indirect sub-type changes:
    return type changed:
      in pointed to type 'typedef std::basic_filebuf<char,
std::char_traits<char> >::__streambuf_type':
        underlying type 'class std::basic_streambuf<char,
std::char_traits<char> >' changed:
          1 data member change:
           type of 'std::locale std::basic_streambuf<char,
std::char_traits<char> >::_M_buf_locale' changed:

  [C]'method std::basic_filebuf<wchar_t, std::char_traits<wchar_t>
>::__filebuf_type* std::basic_filebuf<wchar_t, std::char_traits<wchar_t>
>::close()' has some indirect sub-type changes:
    return type changed:
      in pointed to type 'typedef std::basic_filebuf<wchar_t,
std::char_traits<wchar_t> >::__filebuf_type':
        underlying type 'class std::basic_filebuf<wchar_t,
std::char_traits<wchar_t> >' changed:
          1 base class change:
            'class std::basic_streambuf<wchar_t, std::char_traits<wchar_t> >'
changed:
              1 data member change:
               type of 'std::locale std::basic_streambuf<wchar_t,
std::char_traits<wchar_t> >::_M_buf_locale' changed:
                 details were reported earlier


          1 member function change:
            'method virtual std::basic_filebuf<wchar_t,
std::char_traits<wchar_t> >::__streambuf_type* std::basic_filebuf<wchar_t,
std::char_traits<wchar_t> >::setbuf(std::basic_filebuf<wchar_t,
std::char_traits<wchar_t> >::char_type*, std::streamsize)' has some sub-type
changes:
              return type changed:
                in pointed to type 'typedef std::basic_filebuf<wchar_t,
std::char_traits<wchar_t> >::__streambuf_type':
                  underlying type 'class std::basic_streambuf<wchar_t,
std::char_traits<wchar_t> >' changed, as reported earlier


  [C]'method std::basic_filebuf<wchar_t, std::char_traits<wchar_t>
>::__filebuf_type* std::basic_filebuf<wchar_t, std::char_traits<wchar_t>
>::open(const char*, std::ios_base::openmode)' has some indirect sub-type
changes:
    return type changed:
      in pointed to type 'typedef std::basic_filebuf<wchar_t,
std::char_traits<wchar_t> >::__filebuf_type':
        underlying type 'class std::basic_filebuf<wchar_t,
std::char_traits<wchar_t> >' changed, as reported earlier

...

It is not clear from the message what abidw thinks changed in the
__M_buf_locale it just says that it changed. I also believe that in cases like
this it might be helpful to print what aspect about the change in type actually
changed.

this was done with the dodji/fixes branch

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