On Mon, Apr 22, 2002 at 01:04:52AM -0700, Ask Bjoern Hansen wrote:
> The more correct variation would be to say that you could nuke
> PIPELINING from the EHLO and then do as I suggested.

Yeah, that'd work.  So far, the abuse I've seen has used HELO rather
than EHLO, not that it needs to.  In fact had they used EHLO I might
have interpreted it as a buggy sending MTA, except that the content is
always spam and the remote host is always in Asia.


> Actually, I just thought of a case where we might be broken with
> regards to PIPELINING.  Do we have to deal with a DATA segment even
> if we reject it?  (A client doing as you describe; if we give a
> 5xx code, is it our responsibility to not interpret their DATA part
>  as commands).  That should be checked with the RFC. :-/

The way I recall rfc1854, DATA can only appear at the end of a pipelined
command group -- after that the sender is required to wait for a 354
go-ahead or 5xx rejection before sending their data.  There's an
allowance for the receiving MTA issuing a go-ahead even if it's declined
every recipient, but the burden is on the sender.


-- 
Devin  \ aqua(at)devin.com, 1024D/E9ABFCD2;  http://www.devin.com
Carraway \ IRC: Requiem  GCS/CC/L s-:--- !a !tv C++++$ ULB+++$ O+@ P L+++

Reply via email to