bug#34206: Pdumper giving a segfault

[Compro Prasad]

On Sat, Jan 26, 2019 at 1:36 PM Eli Zaretskii <eliz@gnu.org> wrote:
>
> > From: Compro Prasad <comproprasad@gmail.com>
> > Date: Sat, 26 Jan 2019 12:03:59 +0530
> >
> > Emacs source: official master branch (commit: 7960951d5d)
> > Emacs config: 
> > https://github.com/Compro-Prasad/spacemacs/tree/pdumper-is-in-master
> > (commit: f5eef2127ad4d)
> > Spacemacs config: https://github.com/Compro-Prasad/.spacemacs.d
> > (commit: 8a1eebc65f9a)
> >
> > Build command: ./configure --enable-checking='yes,glyphs'
> > --enable-check-lisp-object-type CFLAGS='-O0 -g3' --with-xwidgets &&
> > make -j2
> >
> > GDB log: https://pastebin.com/2uMPHCRV
>
> This seems to indicate your configuration is quite complex, and I
> cannot easily understand what are you doing that is different from
> the normal build and "emacs -Q" to start Emacs.  In particular, this:
>
>   $ gdb ./src/emacs-27.0.50.11
>   Reading symbols from ./src/emacs-27.0.50.11...done.
>   (gdb) r --dump-file="/home/compro/.emacs.d/.cache/dumps/spacemacs.pdmp"
>   Starting program: 
> /home/compro/Downloads/git.savannah.gnu.org/git/emacs/src/emacs-27.0.50.11 
> --dump-file="/home/compro/.emacs.d/.cache/dumps/spacemacs.pdmp"
>
> indicates that your pdump file is spacemacs.pdmp, but I don't know how
> that file was produced, and I cannot easily understand what is
> different in your build procedure from the normal upstream build of
> Emacs.
I don't know if you can help me but I have some more information on
how the dump might be produced.

https://github.com/Compro-Prasad/spacemacs/blob/pdumper-is-in-master/core/core-dumper.el#L68

I went a bit further and tried executing the following using `C-x C-e`
while running spacemacs:

   (dump-emacs-portable "~/test.pdmp")

It gave me: eval: unsupported object type in dump: window configuration

Is this a bug or I am supposed to run Emacs in batch mode while dumping?