Amendments to pconsole fast-track [PSARC/2009/275 FastTrack timeout 05/08/2009]

[Nicolas Williams]

On Tue, May 05, 2009 at 11:07:46AM -0500, Norm Jacobs wrote:
> Nicolas Williams wrote:
> >On Mon, May 04, 2009 at 12:34:29PM -0500, Norm Jacobs wrote:
> >  
> >>James Carlson wrote:
> >>    
> >>>Norm Jacobs writes:
> >>> 
> >>>      
> >>>>TIOCSTI appears to require  elevated privilege.  streamio.c appears to 
> >>>>do the auth checking in the kernel using secpolicy_sti(), which equates 
> >>>>   
> >>>>        
> >>>If you issue it on the controlling tty for this process and you have
> >>>at least read access, then you won't need extra privileges.  But, yes,
> >>>if you want to do it on someone else's tty, then you'll need all
> >>>privileges, as it's an escalation threat.
> >>> 
> >>>      
> >>It appears to want to use other's ttys.
> >>    
> >
> >Er, why?
> >  
> Because that's what it does. It connects to a bunch of ttys, allows you 
> to type in it's "shell", and stuffs a copy of what you type out to all 
> of the ttys it's connected to. It monitors each "connection" through 
> individual xterms. Near as I can tell, it's for people that are too lazy 
> to cut/paste. :-)

Yes, but as has been pointed out, it could just master those ptys.
Sounds like pconsole's architecture is broken, not necessarily fatally
so since you can manage the privilege issue, but I now think of pconsole
as toxic.  Here's a question: can you cause pconsole to accidentally
clobber some other user's pty?

Nico
--