>>Could somebody with better knowledge of Linux kernel enlighten me what
>>will happen if attacker tries to install an adore-based kit (or other LKM
>>kit) on a box already trojaned with LKM? I suppose new adore will take
>>control from previous adore since it will remap kernel calls elsewhere,
>>right? Or am I gravely confused here? ;-)

Adore acts in a non-destructive manner by wrapping 
the functions that it replaced. It is possible to 
layer adore modules -- however, depending on the 
configuration of the 2nd adore module, the functions 
of the first adore module may not be reachable (they 
would be intercepted by the outermost (or last loaded)) 
module.


>> Any way to make sure my adore stays put? I looked at what StJude module
>> is doing and it looks promising, but maybe something else can help?

StJude, alone, does not protect the kernel from LKM 
rootkits. StMichael[1], which was integrated into StJude 
at version 0.20, does do checks for attacks/changes on/of 
the kernel's integrity after its load.

If you are simply trying to keep adore out of the kernel, 
use StMichael with cloaking, and checksumming support. If 
you are using adore as a 'backdoor' into a honeypot, you 
can use StMichael too -- first load your adore, then 
StMichael[1]. Any subsequent kernel module rootkit loads 
will be detected and, in many circumstances, disabled. 
Silent mode would be appropriate here.

[1]
http://sourceforge.net/project/showfiles.php?group_id=13777&release_id=82067
[2] If you use StMichael 0.10, one modification is necessary -- you
    have to comment out the Bounds checking in StMichael_lkm.c, in
    order to bypass the on-load bounds checking of entries within
    the system call table. 

    The bounds checking will check to ensure that the code for the
    system calls referenced in sys_call_table[] is within the text
    portion of the kernel. Adore's wrapped system calls would show
    up as being outside of the kernel text, since its a module, and
    cause StMichael to generate an error message and abort its load.


Thanks,

--Tim


Tim Lawless, CISSP
EDS Global Information Assurance Services
* [EMAIL PROTECTED]

The Views and Opinions Expressed within this 
document do not necessarily represent the 
views and opinions held by EDS GIAS, or EDS. 


Reply via email to