Re: CVE-2014-6271: remote code execution through bash

[Solar Designer <solar () openwall com>]

On Thu, Sep 25, 2014 at 07:40:50PM -0400, Chet Ramey wrote:
> On 9/25/14, 12:19 PM, Solar Designer wrote:
>
> > I think Florian's prefix-suffix patch is actually a better way to go
> > (right now, unless there's some drawback I am not yet aware of), and at
> > a later time function imports should require to be enabled with a
> > non-default option.
>
> Yes, some variant of that will be the next thing to work on.

I suggest that you use Florian's prefix-suffix convention as-is (not
necessarily the code, but same prefixes and suffixes, so that your
official change would be compatible with Florian's patch).  This is
because multiple distros have adopted Florian's patch already.  While
we'll be breaking compatibility for function exports/imports across
pre-patch and post-patch bash versions anyway, at least you can maintain
compatibility between distros' bash packages and your official releases.

Alexander