Re: CVE-2016-0634 code execution in Bash prompt when expanding hostname

[Leo Famulari <leo@famulari.name>]

On Thu, Sep 22, 2016 at 12:42:15AM +0900, Ludovic Courtès wrote:
> John Darrington <john@darrington.wattle.id.au> skribis:
> 
> > On Tue, Sep 20, 2016 at 04:55:30PM -0400, Leo Famulari wrote:
> >      Any advice on how we should handle CVE-2016-0634?
> >      
> >      http://seclists.org/oss-sec/2016/q3/534
> >
> > Like the comment there says, it is only a problem if the machine has
> > already been owned,
> 
> … or if a privilege application like a DHCP client can be made to set
> the host name to $(something bad), which was apparently possible at some
> point.
> 
> > so I don't see what the issue is.  If there is an issue it is for the
> > bash maintainers to patch.

Perhaps it's not the most critical bug, but I don't think we can
effectively anticipate the full impact of this (or any) bug. It's better
to just fix it now that we know about it.

> Chet proposed a patch:
> 
>   http://seclists.org/oss-sec/2016/q3/att-538/prompt-string-comsub.patch

I've asked Chet if he will add the patch to the bash-4.3-patches FTP
directory. If not, we can apply it the "normal" way.

> 
> IIUC, the just-released 4.4 isn’t affected, right?

Right.

> We should at least update it in core-updates, but core-updates won’t be
> merged until we have fixed that Binutils/MIPS issue (which shouldn’t be
> too hard, but we never know!).

I spent some time looking at the Bash package definition, but I'm stuck
on how to handle all the Bash %patch-series machinery. There are
currently no patches for Bash 4.4, nor is there an FTP directory
corresponding to a future patch series.

Does anyone have advice on how to proceed?