Re: MPS symbols

[Gerd Möllmann]

Helmut Eller <eller.helmut@gmail.com> writes:

> On Fri, Jun 28 2024, Gerd Möllmann wrote:
>
>> Gerd Möllmann <gerd.moellmann@gmail.com> writes:
>>
>>> Forgot something. While looking at the scanner, I wondered if we could
>>> remove the is_mps check from fix_lisp_obj now that we have the dump
>>> in MPS...
>>
>> Let's find out, pushed. Works for me :-)
>
> I think this uncovered a bug.  Now this
>
>   emacs -Q -eval '(progn (view-hello-file) (forward-line 7) (dotimes (_ 5)
>   (redisplay) (igc--collect) (forward-line)))' -f kill-emacs
>
> causes a SIGSEGV in fix_font,
>
>       case FONT_OBJECT_MAX:
>       {
>         struct font *f = (struct font *)v;
>         const Lisp_Object *type = &f->driver->type;
>         IGC_FIX12_OBJ (ss, igc_const_cast (Lisp_Object *, type));
>       }
>
> It turns out that f->driver is the variable xfont_driver and it is
> actually const and the type field is a builtin symbol.  So it looks like
> we can remove fix_font completely.

Hm, okay. Can you see why is segfaults? I'm wondering because I'd have
thought that we'd crash all over the place if we couldn't handle
built-in symbols...