home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Hacker 2
/
HACKER2.mdf
/
virus
/
virusl2
/
virusl2.26
< prev
next >
Wrap
Text File
|
1995-01-03
|
5KB
|
106 lines
VIRUS-L Digest Thursday, 26 Jan 1989 Volume 2 : Issue 26
Today's Topics:
nVir (init 29) (Mac)
Viral protection by checksum registration?
---------------------------------------------------------------------------
Date: Wed, 25 Jan 89 15:13:46 EST
From: SCOTT <LICHTBLS@DUVM.BITNET>
Subject: nVir (init 29) (Mac)
I have encountered this new strain of nVir on a bunch of Mac II's
with Interferon, but have not been able to successfully eradicate the
infection. Also Ferret, VirusRx, and virus detective are not able to
identify this virus. The virus also shows up as a code segment ID 255
or 256 which is 712 bytes long as previously noted. What is the best
way to eradicate this "thing"? Is this new strain of any potential
danger to documents saved on a different disk or will it just cause
memory problems when the infected machine is used?
Help!!
Scott.
------------------------------
Date: Wed, 25 Jan 89 14:44 MDT
From: Pete Klammer 303/556-3915 <PKLAMMER@CUDENVER.BITNET>
Subject: Viral protection by checksum registration?
We could have some protection against viruses if we could compare a
characteristic "signature" of a program file against a "register" of
known program signatures. The "signature" would have to be fairly
strong, and the problems of trusted registration and distribution of
copies of the register are non-trivial. Furthermore, a virus attack
can spread more quickly than registered-signature checking can be
done, but at least this method offers some assurance when we're
looking at a clean system. For that matter, it would let us know if
we're looking at identical copies of a known virus, vs. slightly
twiddled ones.
What I'm suggesting is that something like a checksum be defined, but
it must be long and complex enough, and include the file size, such
that a counterfeit file of the same size and signature which could
even execute at all, let alone do any useful viral-like damage, would
be too hard or too expensive to come up with. A checksum is too weak:
I can produce any checksum I want from any file if I have a few spare
bytes to "seed" with checksum-compensating values. Rather, the
algorithm for the "anti-viral-signature" of a file would have to be
more like a high-order cyclic redundancy check or one-way trap-door
encryption.
I'm also suggesting that a common, trustworthy repository for
registration of program files be set up. I could then know that
FORMAT.COM for PC-DOS 2.1 has signature "1140745HL2K6G76G724", and
FORMAT.COM for Zenith MS-DOS 3.1 has signature "1047HD2468K7G6762GR4",
and so forth. Over time, that could get to be a long list: how many
legitimate versions of C1.EXE (or whatever) for Microsoft C have
actually been distributed (3.0, 4.0, 4.01, 4.1, 5.0, 5.01?, 5.1...?).
And of course, versions of the "anti-viral-signature-checker" would
have to be registered, too.
With these tool, one could, on occasion or even constantly in "TSR
background spare time", scrutinize a file system for corruptions. The
"background" method is itself vulnerable to viral infiltration, but I
should still be able to boot up from a trusted write-protected* floppy
and scan my files whenever things get suspicious.
(*NOTE on that noisy subject: PC floppy drives implement write
protection in the hardware quite simply: the "write-enable" pin of the
floppy-disk-controller chip receives its signal from an AND gate --
i.e., whenever you ask to write, AND the write-enable notch is
detected, it writes. Commodore-64 drives (1541's) do not have such a
hardware AND gate, and in fact, their ROM firmware can be overridden
by executable code downloaded into into on-board RAM. [Speculation
now:] Since Apple ]['s do so much disk control, and so economically,
from inside the 6502 processor, I suspect Apple ][ write protection is
firmware-based, too. These kinds of implementations feed the
write-protect misunderstanding. REAL drives cannot write over a
write-protect tab.)
I recognize the anti-viral-signature method might be too cumbersome to
catch a virus in the act, but wouldn't it be worthwhile to have a way
to check if the ARC v5.13 or the MS-KERMIT v2.32A you just downloaded
from somewhere is clean or crooked?
* --poko Pete Klammer, Systems Programmer, (303)556-3915
* CU-Denver Computing Services / Campus Box 169
* 1200 Larimer St NC2506 / Denver CO 80204-5300
* BITNET: PKLAMMER@CUDENVER
* INTERNET: PKLAMMER@PIKES.COLORADO.EDU
* " I'm half Estonian, which makes up for the other half. "
[Ed. Ideas like this have been tossed around quite a bit, and they
certainly hold some promise (imho). They also have a lot of potential
logistics problems (e.g., who distributes the CRC program itself, and
how do we assure that *it* is not corrupt? a computing environment
in which everyone uses the same CRC (or checksum...) would be, as it
is now, relatively homogeneous - a virus could make use of this fact,
and propagate freely throughout the environment.). Comments?]
------------------------------
End of VIRUS-L Digest
*********************
Downloaded From P-80 International Information Systems 304-744-2253