Re: [Emacs-diffs] trunk r117741: Handle C stack overflow caused by too nested Lisp evaluation.

[Dmitry Antipov]

On 08/26/2014 10:37 PM, Stefan Monnier wrote:

> > +  DEFVAR_LISP ("top-level-message", Vtop_level_message,
> > +              doc: /* Message displayed by `normal-top-level'.  */);
>
> Why does this need to be exported to Elisp?

This is just to allow 'normal-top-level' to display an appropriate message.

> > +  /* Hard GC error may lead to stack overflow caused by
> > +     too nested calls to mark_object.  No way to survive.  */
>
> I don't see why there's no way to survive.  Isn't it safe to interrupt
> GC within the mark phase and re-run it?

Maybe (at least, it requires some more work). But if we get SIGSEGV in GC,
most likely we have 1) too deep data structures or 2) fatal error,
probably invalid pointer. Since I don't see a way to recover in either
case, I don't see how restarting GC can help.

Dmitry