I am the author of the message forwarded regarding the virus feedback type and 
the _report subdomain. I decided to now join the mailing list and comment on a 
couple of points raised. This mail may not get properly threaded though for the 
lack of in-reply-to and reference fields that I can't insert into the mailer.

ad Point evidence
-----------------

I see the name of the virus together with the name of the virus scanner 
producing the identification as equivalent to the actual byte stream, I 
actually compare it to a compression utility in that sense, just that there's a 
lot better compression factor.

If we start to discuss, whether that combination of virus name and virus 
scanner is really sufficient as evidence, then we will also need to raise the 
question, whether the full message is any evidence at all. The message could be 
tampered with (could be modified), the message could never ever have been sent, 
... The MARF could actually be sent by a bad guy trying to lure the ISP's abuse 
department into taking action against an innocent customer! With that in mind, 
it does make perfect sense to publish the IP address of permitted MARF senders 
for a particular domain, similiar to SPF.


ad point prevent sending malware in first place
-----------------------------------------------

Of course, that would be the ideal case. However, as always with regulations 
and law, there are always people not adhering and sending such stuff 
nonetheless.

Now, in order to enforce a regulation is it possible to violate the regulation? 
Certainly not. So, as we agree that malware should not be transmitted over the 
Internet (and Netiquette as well as RFCs actually require to not send such 
abuse), why would we then ENFORCE sending such malware in abuse reports? We 
agree, malware should not be sent, hence the abuse reporting mechanism should 
provide for that. That's the generic idea behind my proposal.

ad ARIN database:
-----------------

thanks, Alessandro, I am delighted about that announcement.

ad input from malware folks, drop malware reporting altogether:
---------------------------------------------------------------

I suspect that just like me those good folks didn't see a lot of need so far, 
more or less free format mails were generated so far. However, a number of ISPs 
have started to require ARF format for all mail arriving at their abuse 
department, so malware reporting has started to become an issue and feedback 
will probably start to flow in.

MARF represents mail abuse reporting format, not just spam reporting format, 
hence it should be able to cater for proper malware reporting, too. By 
dropping, or not supporting a proper malware reporting, MARF could actually 
contribute to malware authors escaping any form of control/abuse handling.

ad "Reported-CVE-Id":
---------------------

Full support on my side, I actually forgot about that. I'd treat this as an 
optional field for those virus scanners, the database of which permit to 
identify attached CVE IDs. The scanners I am aware of currently do not cater 
for that facility.

ad encryption of malware:
-------------------------

Bandwidth is an issue despite encryption, as I said above, identifying the 
virus can be seen as a very effective compression, and at the same time removes 
the requirement of sending malware across the internet (to a possibly 
unexpecting receiver).

The malware may well be in the order of several 100K, maybe even MBs of 
attachment, while the virusname is a matter of possibly 128, maybe 512 bytes. 
Is that line as well as mailbox congestion really worth sending an attachment, 
that is fully identified by the virusname anyway so that the recipient could go 
in his database, fetch a sample of that thing and knows the identified malware 
exactly byte by byte?

Who does need and can take use of the actual malware body? Those who write 
virus scanners, obviously, they however had according (expecting) reporting 
channels already via their websites.

What would the ISP do with the binary file he gets sent? The human eye wouldn't 
see anything, he might run his own virus scanner over it to get the same result 
the sender already had, and all we achieve is using up bandwidth on Internet 
and CPU power. And I'd need to again raise the question whether that attachment 
has any value as evidence whatsoever.

If we don't trust that the MARF report contains accurate data, why would we 
assume the contained message is accurate? It could as well be just produced 
from any place without any actual mail having been received.

If the receiver of the MARF report doubts the identification, he'd certainly 
inquire with the sender and they both will agree how to produce something that 
both accept as reliable evidence.

ad FBL:
-------

The MARF contains a clear free text section, feedback section and evidence 
section for good reason. The MARF can that way be processed by either any human 
monitoring an abuse mailbox, an automatic handling tool and/or both. So, why 
then restrict the use of MARF to only those who actually expect a MARF? Would 
the abuse department not requiring nor being prepared for MARF not be 
interested in seeing the spam mail, or know about what malware was sent from 
where? So, whether or not it is MARF, the abuse report does need to contain 
such stuff. So, why not send MARF (and "promote" it that way) on ANY abuse 
report if the complainer's reporting utility is capable of MARF?

However, then MARF must not assume that the recipient is expecting and capable 
of handling such unsafe content, which again is another argument of shaping the 
evidence section according to need.


Sorry again for the long sermon, and sorry for most likely breaking the 
threading of this discussion.
_______________________________________________
marf mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/marf

Reply via email to