Re: firewall securing outgoing traffic?
From: Dimitri Maziuk (dima@127.0.0.1)Date: 01/29/02
- Next message: Juha Laiho: "Re: Tripwire problem (st_ctime)"
- Previous message: Enrique: "Re: how to use the custom picture in graphical lilo in 7.x redhat series"
- In reply to: Yan Seiner: "Re: firewall securing outgoing traffic?"
- Next in thread: Bruno Wolff III: "Re: firewall securing outgoing traffic?"
- Reply: Bruno Wolff III: "Re: firewall securing outgoing traffic?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
From: Dimitri Maziuk <dima@127.0.0.1> Date: 29 Jan 2002 19:54:40 GMT
On Tue, 29 Jan 2002 05:06:11 -0500, Yan Seiner wrote:
> In article <3c55ca7a.21999844@news.online.de>, "Joe"
><NO!SPAM!hte123@gmx.net> wrote:
>
><snip>
>> Linux has IMHO no implicit protection from that behaviour? As far as i
>> know by now, a program for example data can read out data from the tmp
>> dir of the current user.
>>
>
> Yes and no. While an app may read from /tmp, very little actual data is
> in temp. Most of the information that spyware would want is in log files
> in /var/log; these are (or should be) only readable by root. So the job
> is much harder for spyware. In fact, most of the time /tmp is empty.
>
> As for software that "calls home" - that's much harder to block. Typical
> firewalls only block ports; the assumption is that you want to protect
> services.
This doesn't really work: presumably the code that calls home is
hidden inside some application that _you've_installed_because_you
_wanted_to_use_it. If it needs network access for legitimate
reasons, you can't block it (or you can't use it).
Similarly, it can use a "legitimate" port when it calls home (e.g.
it calls a CGI script), so trying to block on per-port basis won't
work either.
The solution is not technical: if an open-source app calls home
behind the scenes, I'd expect to see a lot of stink in usual places
(plus a patch to disable that behaviour) pretty soon after the app
is released. If it's a closed-source app, well, you get what you
paid for.
> I like the SGID idea - set up a special group that is the only group
> allowed access out. Then run the few programs that need to communicate
> with the outside SGID that group. Of course, this offers NO protection
> if you have win boxes NATed behind your linux firewall.
Debian does this with access to devices, like floppy, audio, cdrom.
Unfortunately, unix u/g/o model doesn't work too well here:
1. you can easily get to the point where an app must have >1 different
[E|S]GIDs at the same time,
2. administration nightmare -- too many groups, too many SGID binaries.
Dima
- Next message: Juha Laiho: "Re: Tripwire problem (st_ctime)"
- Previous message: Enrique: "Re: how to use the custom picture in graphical lilo in 7.x redhat series"
- In reply to: Yan Seiner: "Re: firewall securing outgoing traffic?"
- Next in thread: Bruno Wolff III: "Re: firewall securing outgoing traffic?"
- Reply: Bruno Wolff III: "Re: firewall securing outgoing traffic?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|