|Newbie questions about Pliant
how to install a firewall on FullPliant?
|Message posted by smoerk on 2001/08/07 18:17:55
|Is it possible to protect a FullPliant system with a firewall
on the same system? Or is it a firewall of no use for a
|Message posted by maybe Hubert Tonneau on 2001/08/07 18:47:52
|Yes it is:
When you provide some 'network/nat' or 'network/internet' or 'network/filter'
lines in the computer definition, they will be translated to Linux 'ipchains'
Basically, a device with 'networl/internet' or 'network/nat' will exclude
some ports (the one servicing NFS, Samba and Netatalk) because it's not safe
to export these on the internet,
and when you define a 'network/filter' line, you specify what the network
connected through this interface is allowed to reach on the Internet.
Anyway, if you feel more confortable with 'ipchains' direct access,
you still can create a /custom/your_domain/your_computer/boot.pli file that
will contain some Pliant 'execute "ipchains ..."' instructions.
See the definition of 'server1' in the provided computer.pdb file as an example.
|Message posted by maybe Hubert Tonneau on 2001/08/07 19:01:05
|Most firewall settings consist in restricting the access to some dangerous
or so considered ports. According to me, this is not really efficient.
Let's assume that in your network, one of the computers have been corrupted
as a result of running a nice looking small free application found on the
This computer can now be used by the hacker to go through your firewall and
access your entire network. He will just have to use an outside access to
port 80 on one of his servers on the Internet, and the firewall will let
informations go in and out. In facts this access can be a tunel that carries
packets in and out and is used to access all ports on all your network
So, a resticted ports only firewall is only really secured if you can trust
ALL the computers on your network.
With the 'network/filter' configuration in FullPliant, some parts of the
network (some physical subnetworks) will get restricted access to the Internet
(only a subset of trusted IPs addresses). So, a corrupted computer on such
a subnetwork cannot be used as a tunel through the firewall, to access and
attack other computers from the outside, and cannot tel the hacker server
that it is now under his control. So, the corrupted computer can automatically
attack other computers on the subnetwork, but cannot receive help from the
Finally, please keep in mind that security is a very complex issue ...
so don't even trust my design.
|Message posted by maybe smoerk on 2001/08/07 21:41:29
|thank you for the info. i'll go to hal2001 (www.hal2001.org) tomorrow and want to setup a FullPliant installation on my computer. i have no linux installed and also want to play with pliant, because i like the idea behind pliant. also i think it's fun to do it at HAL and maybe there are some pliant hackers.
but it's not unlikely that your system get compromised at HAL, so I asked before... :)
|Message posted by maybe Hubert Tonneau on 2001/08/07 22:51:13
|On each FullPliant system, there is a /pliant_security/site_secret.pdb
If you compromise the system security, then you can read it.
If you succeed to do it on http://pliant.cx/, please at least
email me it's content as a proof
... and if you are very kind, also tel me how you did it so that the
chalenge get's a bit harder next time.
On the other hand, deny of service is probably too easy to achieve at the
moment, so it would not be of any glory.
Now, there is a more serious chalenge, which is to compromise Pliant
secured channel, I mean not rely on stupid holes in the HTTP server
content, but succeed to break the secured channel directly (it's an
RSA + RC4 implementation, and as any implementation, it probably has
design flows and bugs)
If you succeed to do that, you should be abble to send me the content
of 'server2.heliogroup.fr' /pliant_security/site_secret.pdb
(at the same time, please send me it's public key because I lost it,
and it's 150 miles away from me)
I like chalenges, even if I know that I'm highly likely to loose this one.
I'd also like to improve FullPliant overall security, and since I'm not
a specialist of hacking (I spent all my time on the design), I would be
very happy to receive advises from advanced hackers.
One last word: please do attack http://pliant.cx/ which is server5.heliogroup.fr
but not server1.heliogroup.fr or server3.heliogroup.fr that are production
servers. As far as I'm spending most of my time developping free softwares,
I hope fair play from you.
|Message posted by maybe smoerk on 2001/08/07 23:42:04
|I reread my post and I didn't want to say that I or someone else want to attack your machine (pliant.cx), but that machines at HAL get attacked from outside.
I have no experience in hacking and I don't want to become a hacker, but I hope to learn more about security mechanism and how to prevent intrusion.
Hacking from HAL to the outside world and to computers which are not explicit announced for hacking is a bad idea (tm), because no one want to get in trouble with the police or bad media coverage.
If I'll get FullPliant installed and someone is interested in hacking my machine (looking for security holes), I'll share our experience.
|Message posted by maybe Hubert Tonneau on 2001/08/08 10:08:47
|As you pointed out, attacking a dedicated local machine is even better.
The only problem is that not all advanced hackers can attend a given
meeting, so having a computer somewhere on the Internet that is accepting
attacks is also valuable.
That's why I said that http://pliant.cx/ is a good candidate for that.
If hackers prefer to have a special chalenge computer, feel free to request:
I can set a dedicated one still connected to the Internet. Also one have to
specify if it should be something fairly easy to attack (just kike
http://pliant.cx/ which is exporting most of it's content), or faily hard
(only remote administration would be provided)
Of course, they are constrains:
- the hacker should say that he is going to try an attack (post a message on
the forum or on the mail) so that the chanlenge is real one (both sides can
- if succeeded, the attack should post a message saying that it succeeded,
and email me a message with the proof which is the content of
- the hacker should not modify files on the broken server other than what
he says to have modifyed
Provided these rules are respected, I find that an attack is a contribution,
and I like it. I should rather say I NEED it.
I also bet that advanced hackers probably aready have self established moral
rules that may be more mature that the few ones I've just written.
I can say it the other way round:
Pliant is very young code with advanced design in many areas,
so the problem is that I cannot seriously tel to the word that it is a
serious solution util it has been attacked.
The attacker is someone that has deaply studied the guts of the application,
and he is seeking for weak parts for the game, but as a result, he also knows
very well the strong parts compared to other similar products, so he has an
information that is vital for Pliant project success. Keep in mind that as
a purely free project, I have no marketing, so I need rock solid facts.
There are several areas that need to be checked:
- the secured channel
- the HTTP server secured policy (rights checking and ciphering of datas
sent to the browser so that it cannot provide malicious forms)
- potencial buffer overflows in the DNS server
- the 'ipchains' settings as a result of rules provided in the computer definition
What's interesting with Pliant is that there is no need to read millions of
lines of durty code written over 25 years, so it can be probably more fun to
play with. I hope so.