There exists what is known as "stack fingerprinting".  Although
TCP/IP is guided by a plethora of RFCs, however, implimentations
can have minor differences.  By sending carefully constructed
packets and looking at replies, it is sometimes possible to
identify the OS, and even kernel revision.

The most obvious methods are these:

#1 Take a challenge case where the RFCs leave a little grey aread
and an implementor must make some decisions on how the stack should
handle a rare or broken case.

#2 Various stacks have had known bugs in the past.  See if you can
cause the system to generate that.  An infamous example from years
past was that certain stacks left unneccessary garbage attached to
certain ICMP packets.  Thus, you send something that causes the system
to generate that type of traffic so you can look at it.

On Thu, Jan 27, 2000 at 11:32:57PM -0800, Perry Blalock wrote:
> Hello redhat-list,
> 
>   Anyone ever hear of Liunx, as an OS, being identified anywhere in a
>   TCP/IP packet header?  Reason I'm asking is that a certain software
>   claims to be able to glean that specific information from the
>   packet header, dunno which daemon, of a Linux machine.
> 
> Best regards,
>  Perry                          mailto:[EMAIL PROTECTED]
> 
> 
> 
> -- 
> To unsubscribe: mail [EMAIL PROTECTED] with "unsubscribe"
> as the Subject.
> 

-- 
J. Scott Kasten

jsk AT tetracon-eng DOT net

"That wasn't an attack.  It was preemptive retaliation!"


-- 
To unsubscribe: mail [EMAIL PROTECTED] with "unsubscribe"
as the Subject.

Reply via email to