As it's security fix, will be the patch provided via
http://www.net-snmp.org/official_patches/ ?

Radek

On Thu, 2005-06-30 at 12:14 -0700, Wes Hardaker wrote:
> >>>>> On Thu, 30 Jun 2005 11:45:15 +0400, Roman Tsiroulnikov <[EMAIL 
> >>>>> PROTECTED]> said:
> 
> Roman> Thank you for fast reply!
> 
> No problem.  Thanks for explaining the problem so well!  Anytime
> people give us a well documented reproducible bug it makes it much
> easier to track down the problem.  And then you supplied a patch as
> well!
> 
> However, the patch you actually supplied broke things in other ways.
> In particular, you check for the expected pdu length being <= 0.  The
> check_packet routine actually returns a -1 for serious error, where
> the connection should be dropped, and a 0 for "I don't even have
> enough data to determine how much data I need to expect".  Thus,
> really slow trickles over slow/broken networks will cause problems.
> So the patch wasn't applied directly as is, but modified so that it
> only closes the stream iff a -1 was returned.
> 
> Roman> Yes, at this time, it's reproducible only via TCP, but, in principle, 
> Roman> there was no matter, what transport we are using.
> 
> It would break for all stream-based protocols.  However, you original
> message said that UDP was an issue to, but it was not as it's not a
> stream based protocol...
> 
-- 
Radek Vokál <[EMAIL PROTECTED]>

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to