Re: priviledge escalation techniques

[miguel.dilaj () pharma novartis com]

Hi again jnf!

I'll provide you 2 answers, then you can investigate yourself.

a) It's perfectly possible for a process to run with high privileges, and 
drop the privileges when in the need to do something else. In fact is not 
only possible, it's common practice, both in the *nix and Windows world.
I blame the fact that WHEN NO ONE IS LOGGED IN, the system is still 
monitoring the 5 SHIFT sequence, and runs sethc.exe as SYSTEM in that case 
(you can even launch explorer.exe and have the whole enchilada as SYSTEM). 
I'm not saying that I'm surprised, considering that the guys at M$ have 
thrown everything but the kitchen sink into system space...

b) It's perfectly possible to monitor keystrokes even without 
administrative privileges, thanks to the way Windows is built. Feel free 
to try the keylogging functionality of the spanish tool VeoVeo 
(www.hackindex.org) as a normal user. If you don't understand spanish, 
don't panic, I made a translation to english, available at 
http://usuarios.lycos.es/n3kr0m4nc3r/tools/
I know VeoVeo it's not perfect, but it shows the idea, and the source is 
available if you are not happy with it.

I hope you don't think that the above are also silly statements...
Cheers,

Miguel Dilaj (Nekromancer, the humorous one)
Vice-President of IT Security Research, OISSG (the funny organization)
We need YOU at www.oissg.org!






jnf <lists () innocence-lost net>
17/01/2005 19:45

 
        To:     Miguel Dilaj/PH/Novartis@PH
        cc:     pen-test () securityfocus com
        Subject:        Re: priviledge escalation techniques


> and the guys at Micro$oft comit the cardinal mistake of not
> making IT check if SHIFT was pressed 5 times, but to include that in 
some
> other part of the OS (kernel? ;-)

And while I sit here eating lunch it occured to me how silly of a
statement that was- consider which is more of an acceptible risk-

scenario 1) sethc.exe is run as a normal user, or rather as the user
logged in- it does not run with any special capabilities, the keyboard
driver or whatever intercepts and detects shift pressed 5 times, or held
for X seconds- however
IF someone managed to override your DAC's/file permissions then they can
overwrite the program, however if this occurs- the game is already up
because you had a more critical flaw some place else, and that is really
the way that you lost control.

scenario 2) sethc.exe is always running and monitoring keystrokes looking
for any sequence of keystrokes that it recognizes, in order to do this
either any user has to be able to 'sniff keystrokes', OR, it has to run
with special access allowing the window for abuse to grow bigger- in
addition to this the kernel has to take extra steps in order to pass every
keystroke to userspace, which is going to degrade performance. So here,
the simple program is now running with elevated status and becomes a huge
potential for abuse.

> From a perspective of security, which is a better design? scenario 2 is
basically what you are suggesting. I love IT Security as well, but its not
nearly as humorous as It Security 'Professionals'


cheers,
jnf