[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

--- Comment #4 from dodji at redhat dot com ---
"bochecha at daitauha dot fr" <sourceware-bugzilla@sourceware.org>
writes:

> Unfortunately, this doesn't seem to be entirely fixed. :-/

[...]

> But with abidiff from Git master, I get:
>
> $ ./tools/abidiff --no-added-syms --drop-private-types --headers-dir1
> $T1/usr/include --headers-dir2 $T2/usr/include --debug-info-dir1
> $T1/usr/lib/debug --debug-info-dir2 $T2/usr/lib/debug
> $T1/usr/lib/x86_64-linux-gnu/libgtk-3.so.0.2200.30
> $T2/usr/lib/x86_64-linux-gnu/libgtk-3.so.0.2400.0
> Functions changes summary: 0 Removed, 1 Changed (37 filtered out), 0 Added (31
> filtered out) functions
> Variables changes summary: 0 Removed, 0 Changed, 0 Added variable
>
> 1 function with some indirect sub-type change:

[...]

Hmmh, I am afraid I couldn't reproduce by just recompiling Gtk 3.22.30
and Gtk 3.24 and running "make install".  It looks like something else
is happening when you are editing the binaries to split debug info
(maybe at debug info compression time?).  I'll have to get my hands on
you files libgtk-3.so.0.2200.30 libgtk-3.so.0.2400.0 as well as their
.debug corresponding files, so that I can inspect them.

For the record, here is what I am getting here:

$ ./build/tools/abidiff --no-added-syms --drop-private-types \
                        --hd1 ~/.local/gtk-3.22.30/usr/local/include \
                        --hd2 ~/.local/gtk-3.24.0/usr/local/include \
                       
~/.local/gtk-3.22.30/usr/local/lib/libgtk-3.so.0.2200.30 \
                        ~/.local/gtk-3.24.0/usr/local/lib/libgtk-3.so.0.2400.0

Functions changes summary: 0 Removed, 0 Changed, 0 Added (31 filtered out)
functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

$

In general, I think it's always better to attach the exact binaries you
are seeing issues on because there can be subtle variations in our
environments when I we to build the binaries on our respective systems.

Thanks.

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

[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

dodji at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2018-09-21
            Summary|Libabigail seems to be      |confusion when a type
                   |confused by Gtk             |definition DIE is matched
                   |                            |by a supprspec and its decl
                   |                            |DIEs aren't
     Ever confirmed|0                           |1

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

[Bug default/23641] Libabigail seems to be confused by Gtk

[bochecha at daitauha dot fr]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

--- Comment #1 from Mathieu Bridon <bochecha at daitauha dot fr> ---
I'll add that I've seen similar abidiff reports when checking multiple commits
in the Gtk repository.

I originally encountered this when upgrading Gtk from 3.22.30 to 3.24.0 in the
Freedesktop Sdk:
https://gitlab.com/freedesktop-sdk/freedesktop-sdk/-/jobs/96780305

I thought maybe that was an actual ABI break, and spent the afternoon bisecting
the changes in Git between those 2 releases… only to find an ABI break in a
commit which changed an m4 file provided for Gtk consumers (i.e not used when
building libgtk-3.so.0).

I have since reproduced similar abidiff reports (always GtkFooPrivate at offset
192 of some GtkFoo class) for many commits in the Gtk repository, whether they
changed the code or not, leading me to think this might be a libabigail issue.

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

[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

--- Comment #7 from dodji at redhat dot com ---
"bochecha at daitauha dot fr" <sourceware-bugzilla@sourceware.org>
writes:

> https://sourceware.org/bugzilla/show_bug.cgi?id=23641
>
> --- Comment #6 from Mathieu Bridon <bochecha at daitauha dot fr> ---
> Here's a tarball containing the two stripped libraries, debuginfos and headers:
> https://www.daitauha.fr/static/abigail/gtk3-reproducer.tar.xz
>
> It's too big to be added to Bugzilla unfortunately.

Thanks!!

I am reproducing the issue on my end now.

/me is back to the blackboard.

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

[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

dodji at redhat dot com changed:

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

--- Comment #2 from dodji at redhat dot com ---
This issue should be fixed in the master branch by commit
https://sourceware.org/git/gitweb.cgi?p=libabigail.git;a=commit;h=90b4e7676c20c78ebebff14f8d4d58b8509f1012
and the fix should be available in the coming 1.5 version of libabigail.

Sorry for the inconvienience.

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

[Bug default/23641] New: Libabigail seems to be confused by Gtk

[bochecha at daitauha dot fr]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

            Bug ID: 23641
           Summary: Libabigail seems to be confused by Gtk
           Product: libabigail
           Version: unspecified
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: default
          Assignee: dodji at redhat dot com
          Reporter: bochecha at daitauha dot fr
                CC: libabigail at sourceware dot org
  Target Milestone: ---

Here is a fun reproducer.

Clone the Gtk repository:

  $ git clone https://gitlab.gnome.org/GNOME/gtk.git
  $ cd gtk

First, build and install 3.24.0:

  $ git checkout 3.24.0
  $ git clean -dfx
  $ ./autogen.sh
  $ make
  $ make install DESTDIR=$(pwd)/../gtk-3.24.0

Now, do the exact same thing with the commit just before:

  $ git checkout HEAD
  $ git clean -dfx
  $ ./autogen.sh
  $ make
  $ make install DESTDIR=$(pwd)/../gtk-3.24.0-previous

Finally, compare their ABI:

  $ abidiff --no-added-syms --drop-private-types \
      --headers-dir1 ../gtk-3.24.0/usr/local/include/ \
      --headers-dir2 ../gtk-3.24.0-previous/usr/local/include/ \
      ../gtk-3.24.0/usr/local/lib/libgtk-3.so.0 \
      ../gtk-3.24.0-previous/usr/local/lib/libgtk-3.so.0
  Functions changes summary: 0 Removed, 3 Changed (91 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

  3 functions with some indirect sub-type change:

    [C]'function GtkWidget* gtk_action_bar_get_center_widget(GtkActionBar*)' at
gtkactionbar.c:485:1 has some indirect sub-type changes:
      return type changed:
        in pointed to type 'typedef GtkWidget' at gtktypes.h:46:1:
          underlying type 'struct _GtkWidget' at gtkwidget.h:133:1 changed:
            type size hasn't changed
            1 data member deletion:
              'GtkWidgetPrivate* _GtkWidget::priv', at offset 192 (in bits) at
gtkwidget.h:139:1


    [C]'function void
gtk_icon_view_convert_widget_to_bin_window_coords(GtkIconView*, gint, gint,
gint*, gint*)' at gtkiconview.c:4417:1 has some indirect sub-type changes:
      parameter 1 of type 'GtkIconView*' has sub-type changes:
        in pointed to type 'typedef GtkIconView' at gtkiconview.h:41:1:
          underlying type 'struct _GtkIconView' at gtkiconview.h:79:1 changed:
            type size hasn't changed
            1 data member change:
             type of 'GtkContainer _GtkIconView::parent' changed:
               underlying type 'struct _GtkContainer' at gtkcontainer.h:49:1
changed:
                 type size hasn't changed
                 1 data member change:
                  type of 'GtkWidget _GtkContainer::widget' changed:
                    underlying type 'struct _GtkWidget' at gtkwidget.h:133:1
changed:
                      type size hasn't changed
                      1 data member insertion:
                        'GtkWidgetPrivate* _GtkWidget::priv', at offset 192 (in
bits) at gtkwidget.h:139:1



    [C]'function void gtk_size_group_add_widget(GtkSizeGroup*, GtkWidget*)' at
gtksizegroup.c:474:1 has some indirect sub-type changes:
      parameter 1 of type 'GtkSizeGroup*' has sub-type changes:
        in pointed to type 'typedef GtkSizeGroup' at gtksizegroup.h:38:1:
          underlying type 'struct _GtkSizeGroup' at gtksizegroup.h:42:1
changed:
            type size hasn't changed
            1 data member insertion:
              'GtkSizeGroupPrivate* _GtkSizeGroup::priv', at offset 192 (in
bits) at gtksizegroup.h:47:1

So libabigail thinks the ABI changed between those 2 versions.

But… the ABI check was done over this commit:
https://gitlab.gnome.org/GNOME/gtk/commit/ef1a1b5cb1010977d631dc1131b0db45a1e1e15f

How could that possibly break ABI, when it didn't change the code? (and of
course, you didn't change anything on your host between the 2 builds)

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

[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

dodji at redhat dot com changed:

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

--- Comment #8 from dodji at redhat dot com ---
This should hopefully be fixed now by commit
https://sourceware.org/git/gitweb.cgi?p=libabigail.git;a=commit;h=342ce8d25cf1bdb8618474f95d4b85cb1d94425c.

Thank you for your dedication.

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

[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[bochecha at daitauha dot fr]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

--- Comment #3 from Mathieu Bridon <bochecha at daitauha dot fr> ---
Unfortunately, this doesn't seem to be entirely fixed. :-/

When comparing Gtk 3.22.30 and Gtk 3.24 with abidiff 1.4, I get:

$ abidiff --no-added-syms --drop-private-types --headers-dir1 $T1/usr/include
--headers-dir2 $T2/usr/include --debug-info-dir1 $T1/usr/lib/debug
--debug-info-dir2 $T2/usr/lib/debug
$T1/usr/lib/x86_64-linux-gnu/libgtk-3.so.0.2200.30
$T2/usr/lib/x86_64-linux-gnu/libgtk-3.so.0.2400.0
Functions changes summary: 0 Removed, 2 Changed (58 filtered out), 0 Added (31
filtered out) functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

2 functions with some indirect sub-type change:

  [C]'function void gtk_enumerate_printers(GtkPrinterFunc, gpointer,
GDestroyNotify, gboolean)' at gtkprinter.c:1248:1 has some indirect sub-type
changes:
    parameter 1 of type 'typedef GtkPrinterFunc' has sub-type changes:
      underlying type 'typedef gboolean (GtkPrinter*, typedef gpointer)*'
changed:
        in pointed to type 'function type typedef gboolean (GtkPrinter*,
typedef gpointer)':
          parameter 1 of type 'GtkPrinter*' has sub-type changes:
            in pointed to type 'typedef GtkPrinter' at gtkprinter.h:81:1:
              underlying type 'struct _GtkPrinter' at gtkprinter.h:88:1
changed:
                type size hasn't changed
                1 data member insertion:
                  'GtkPrinterPrivate* _GtkPrinter::priv', at offset 192 (in
bits) at gtkprinter.h:93:1

  [C]'function void gtk_stack_add_named(GtkStack*, GtkWidget*, const gchar*)'
at gtkstack.c:1243:1 has some indirect sub-type changes:
    parameter 1 of type 'GtkStack*' has sub-type changes:
      in pointed to type 'typedef GtkStack' at gtkstack.h:41:1:
        underlying type 'struct _GtkStack' at gtkstack.h:67:1 changed:
          type size hasn't changed
          1 data member change:
           type of 'GtkContainer _GtkStack::parent_instance' changed:
             underlying type 'struct _GtkContainer' at gtkcontainer.h:49:1
changed:
               type size hasn't changed
               1 data member change:


    parameter 2 of type 'GtkWidget*' has sub-type changes:
      in pointed to type 'typedef GtkWidget' at gtktypes.h:46:1:
        underlying type 'struct _GtkWidget' at gtkwidget.h:133:1 changed:
          type size hasn't changed
          1 data member deletion:
            'GtkWidgetPrivate* _GtkWidget::priv', at offset 192 (in bits) at
gtkwidget.h:139:1

But with abidiff from Git master, I get:

$ ./tools/abidiff --no-added-syms --drop-private-types --headers-dir1
$T1/usr/include --headers-dir2 $T2/usr/include --debug-info-dir1
$T1/usr/lib/debug --debug-info-dir2 $T2/usr/lib/debug
$T1/usr/lib/x86_64-linux-gnu/libgtk-3.so.0.2200.30
$T2/usr/lib/x86_64-linux-gnu/libgtk-3.so.0.2400.0
Functions changes summary: 0 Removed, 1 Changed (37 filtered out), 0 Added (31
filtered out) functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

1 function with some indirect sub-type change:

  [C]'function void gtk_enumerate_printers(GtkPrinterFunc, gpointer,
GDestroyNotify, gboolean)' at gtkprinter.c:1248:1 has some indirect sub-type
changes:
    parameter 1 of type 'typedef GtkPrinterFunc' has sub-type changes:
      underlying type 'typedef gboolean (GtkPrinter*, typedef gpointer)*'
changed:
        in pointed to type 'function type typedef gboolean (GtkPrinter*,
typedef gpointer)':
          parameter 1 of type 'GtkPrinter*' has sub-type changes:
            in pointed to type 'typedef GtkPrinter' at gtkprinter.h:81:1:
              underlying type 'struct _GtkPrinter' at gtkprinter.h:88:1
changed:
                type size hasn't changed
                1 data member insertion:
                  'GtkPrinterPrivate* _GtkPrinter::priv', at offset 192 (in
bits) at gtkprinter.h:93:1

So there's definitely some progress. :)

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

[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

dodji at redhat dot com changed:

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

--- Comment #5 from dodji at redhat dot com ---
I am reopening this

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

[Bug default/23641] confusion when a type definition DIE is matched by a supprspec and its decl DIEs aren't

[bochecha at daitauha dot fr]

https://sourceware.org/bugzilla/show_bug.cgi?id=23641

--- Comment #6 from Mathieu Bridon <bochecha at daitauha dot fr> ---
Here's a tarball containing the two stripped libraries, debuginfos and headers:
https://www.daitauha.fr/static/abigail/gtk3-reproducer.tar.xz

It's too big to be added to Bugzilla unfortunately.

In a libabigail clone, where I just built:

$ tar xf gtk3-reproducer.tar.xz 
$ ./tools/abidiff --no-added-syms --drop-private-types --headers-dir1
1/usr/include --headers-dir2 2/usr/include --debug-info-dir1 1/usr/lib/debug
--debug-info-dir2 2/usr/lib/debug
1/usr/lib/x86_64-linux-gnu/libgtk-3.so.0.2200.30
2/usr/lib/x86_64-linux-gnu/libgtk-3.so.0.2400.0
Functions changes summary: 0 Removed, 1 Changed (37 filtered out), 0 Added (31
filtered out) functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

1 function with some indirect sub-type change:

  [C]'function void gtk_enumerate_printers(GtkPrinterFunc, gpointer,
GDestroyNotify, gboolean)' at gtkprinter.c:1248:1 has some indirect sub-type
changes:
    parameter 1 of type 'typedef GtkPrinterFunc' has sub-type changes:
      underlying type 'typedef gboolean (GtkPrinter*, typedef gpointer)*'
changed:
        in pointed to type 'function type typedef gboolean (GtkPrinter*,
typedef gpointer)':
          parameter 1 of type 'GtkPrinter*' has sub-type changes:
            in pointed to type 'typedef GtkPrinter' at gtkprinter.h:81:1:
              underlying type 'struct _GtkPrinter' at gtkprinter.h:88:1
changed:
                type size hasn't changed
                1 data member insertion:
                  'GtkPrinterPrivate* _GtkPrinter::priv', at offset 192 (in
bits) at gtkprinter.h:93:1

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