RE: Future Linux devel. Kernels

From: Khimenko Victor (khim@sch57.msk.ru)
Date: Mon May 08 2000 - 09:09:33 EST


In <Pine.LNX.4.21.0005081431370.13513-100000@base.jdimedia.nl> Igmar Palsenberg (maillist@chello.nl) wrote:

>> > It doesn't work.
>>
>> It works beautifully. As long as intruder does not know where exactly
>> traps are placed he can not avoid traps. Will it work as long time defence
>> against scilled cracker SPECIALLY directed against you ? Probably not.
>> Will it stop most crackers ? For sure. As long as traps are NOT common and
>> thus not known to majority of crackers!

IP> Traps are common, as long as the intruder knows what he is looking for.

Exactly. And if he is not ? Even VERY simple local traps will work against
highly skilled intruder as long as he (she) not discover them (by reading
code and just scanning system). And it'll requite LOTS of time.

IP> A very better idea is to secure programs, and avoid programs running as
IP> root.

In ideal world - yes. In real world it does not work: programs are created
by humans and thus holes are inavoidable in programs of decent size.

IP> BSDI also has a mode like this, the kernel secure levels. Basically means
IP> that some operations are disabled, and the only was to switch the level is
IP> from init 1 :-))

So you just need to change code of init in memory via ptrace ? Not such a big
problem...

IP> The 'main' risk if someone gets in that he replaces system bins.. So the
IP> only way to detect this is a proper logging system, that cannot be
IP> modified without someone noticing.

It depends from system heavily. What about router where /sbin/init just set
routing table and then just exit() ? Not very "normal" system but enough to
get point...

>> > The main 'problem' is that someone that has root is god.
>>
>> For now (till "Trusted Linux" not invented).
>>
>> > The only was to make sure we nail the guy is to make sure we can
>> > trace what he did.
>> >
>> Exactly.
>>
>> > If the guy (girl) really know what he is doing he is able to wipe his
>> > traces..
>> >
>> Hah. ONLY if he (she) will feel NEED to track "that style traces as well".
>> See above.

IP> Not wiping your traces degrades someone to script kiddie.

Once. More. YOU. CAN. NOT. WIPE. TRACES. IF. YOU. DO. NOT. KNOW. WHERE. THEY.
ARE. LOGGED. Yes, it does not work as long term defence. But it works and
works GREAT as short term defence even against VERY skillfull cracker (the
more skilled cracker is the less time he (she) will need to detect unusual
tracking system but in many magabytes of code active in current systems even
skilled cracker will need days to detect all places). Of course if code is
in standard kernel with open sources he can do this all at home before
starting attack so it's useless there.

IP> And I hate those..

Becouse most damage was done by script kiddies ? Yes, they are not very skilled
but there are A LOT OF them.

>> > Nothing is safe agains an editor and someone who has root and know about
>> > /dev/kmem
>> >
>> Again: if /dev/kmem is readable on system :-)

IP> It is as root..

In "Trusted XXX" (sometime in future "Trusted Linux" will be created) there
are NO omnipotent root. Even in CURRENT linux you can disable ability to
access hardware. Then EVERYONE (even root) will be unable to read /dev/kmem ...

IP> And it it is not as root, what's the use of /dev/kmem ?

Even if it's as root /dev/kmem can be not usable. Unfortunatelly currently
X server will not work in such mode as well but not all systems out there
need X server in first place...

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Mon May 15 2000 - 21:00:11 EST