Re: bin/49868: tftpd(8) doesn't play well with clients that return acknowledgements to the broadcast address

[David Holland <dholland-bugs%netbsd.org@localhost>]

The following reply was made to PR bin/49868; it has been noted by GNATS.

From: David Holland <dholland-bugs%netbsd.org@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: bin/49868: tftpd(8) doesn't play well with clients that return
 acknowledgements to the broadcast address
Date: Sun, 11 Oct 2015 03:22:29 +0000

 On Fri, May 01, 2015 at 12:45:01AM +0000, Robert Elz wrote:
  > The one issue I see with this change is that switching from a
  > connect/send implementation to sendto (which happens in the patch
  > independently of the new switch) means that netstat won't identify
  > the clients any more.  It would be kind of nice if that didn't
  > happen unconditionally -
 
 There's another reason: as I recall if you have a multihomed machine
 serving tftp, bad things happen if you don't use UDP connect to
 deliver packets. That is: if you don't bind the source address of the
 send socket, the packets might come from the wrong address; and if you
 try to, you end up having to scan routing tables and other horrors to
 figure out which source address to use. Or something like that.
 
 Anyway I would suggest leaving the connect/send alone unless the
 broadcast stuff is specifically needd.
 
 -- 
 David A. Holland
 dholland%netbsd.org@localhost