[Bug default/20420] New: abipkgdiff aborted((core dumped)) while running on glibc package

[Bug default/20420] New: abipkgdiff aborted((core dumped)) while running on glibc package

[sinny at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=20420

            Bug ID: 20420
           Summary: abipkgdiff aborted((core dumped)) while running on
                    glibc package
           Product: libabigail
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: default
          Assignee: dodji at redhat dot com
          Reporter: sinny at redhat dot com
                CC: libabigail at sourceware dot org
  Target Milestone: ---

Created attachment 9407
  --> https://sourceware.org/bugzilla/attachment.cgi?id=9407&action=edit
abipkgdiff output for glibc package

abipkgdiff aborted while running between same version of glibc package.
Following package were used:

glibc pkackage -
https://kojipkgs.fedoraproject.org//packages/glibc/2.23.90/30.fc25/armv7hl/glibc-2.23.90-30.fc25.armv7hl.rpm
glibc debuginfo package -
https://kojipkgs.fedoraproject.org//packages/glibc/2.23.90/30.fc25/armv7hl/glibc-debuginfo-2.23.90-30.fc25.armv7hl.rpm

Command used:
-------------
$ abipkgdiff --dso-only --d1
/var/tmp/taskotron/skumari/task-_f0DGL/stable/glibc-debuginfo-2.23.90-30.fc25.armv7hl.rpm
--d2
/var/tmp/taskotron/skumari/task-_f0DGL/update/glibc-debuginfo-2.23.90-30.fc25.armv7hl.rpm
/var/tmp/taskotron/skumari/task-_f0DGL/stable/glibc-2.23.90-30.fc25.armv7hl.rpm
/var/tmp/taskotron/skumari/task-_f0DGL/update/glibc-2.23.90-30.fc25.armv7hl.rpm

Output:
-------
abipkgdiff: ../../src/abg-comparison.cc:4280: virtual void
abigail::comparison::qualified_type_diff::report(std::ostream&, const string&)
const: Assertion `d->to_be_reported()' failed.
Aborted (core dumped)

Expected Output
---------------
abipkgdiff should complete without abort message

Additional information:
----------------------
I have attached in bugzilla detailed output obtained while running with gdb and
--verbose option

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

[Bug default/20420] Wrong ODR-based type comparison optimization on qualified type

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=20420

dodji at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
            Summary|abipkgdiff aborted((core    |Wrong ODR-based type
                   |dumped)) while running on   |comparison optimization on
                   |glibc package               |qualified type

--- Comment #1 from dodji at redhat dot com ---
It looks like the problem happens during type canonicalizing.  Basically, a
const anonymous struct (a qualified type) from the second ABI corpus is wrongly
being considered equivalent to another (different) const anonymous struct from
the first ABI corpus.

The comparison is made on the name of these types because the types have
wrongly been elected to the ODR (One Definition Rule) based optimization.

I guess the fix is that the ODR-based optimization (for comparing types during
type canonicalizing) should only happen on built-in types as well non-anonymous
classes and enums.  Not qualified types, for instance.

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

[Bug default/20420] Wrong ODR-based type comparison optimization on qualified type

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=20420

--- Comment #2 from dodji at redhat dot com ---
(In reply to dodji from comment #1)

> Basically, a const anonymous struct (a qualified type)
> from the second ABI corpus is wrongly being considered equivalent
> to another (different) const anonymous struct from the first ABI corpus.

Sorry, this should rather be:

Basically, a const anonymous struct (a qualified type)
from the second ABI corpus is wrongly being considered equivalent
to another (different) const anonymous struct from that same corpus.

Of course the first corpus and the second corpus happen to be identical in the
particular test case of this problem report, but I think the underlying problem
exists independently from that fact.

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

[Bug default/20420] Wrong ODR-based type comparison optimization on qualified type

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=20420

dodji at redhat dot com changed:

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

--- Comment #3 from dodji at redhat dot com ---
This issue should be fixed by commit 237734d in the master branch of the git
repository.

Thank you for taking time to report the problem.

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