bug#42113: 28.0.50; Segmentation fault in json-parse-file

[Eli Zaretskii]

> From: Philipp Stephani <p.stephani2@gmail.com>
> Date: Mon, 29 Jun 2020 12:07:33 +0200
> Cc: bug-gnu-emacs@gnu.org, Erik Hetzner <egh@e6h.org>, 42113@debbugs.gnu.org
> 
> > > The JSON functions are missing CHECK_STRING in several places where
> > > json_encode is called.
> >
> > In this specific case I think the test is there, but it is done a bit too 
> > late.  The call to check_string_without_embedded_nuls should be moved 
> > before json_encode.
> 
> Is that always guaranteed to be correct though? Is there a guarantee
> that encoding can never produce null bytes that weren't already
> present in the input?

It is true that you could in principle write a coding-system that
would produce null bytes (although it wouldn't be very useful).  But
in this case we are not talking about any  arbitrary coding-system, we
are talking specifically about UTF-8.