From owner-freebsd-hackers  Sat Aug 23 12:08:52 1997
Return-Path: <owner-freebsd-hackers>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id MAA09689
          for hackers-outgoing; Sat, 23 Aug 1997 12:08:52 -0700 (PDT)
Received: from gvr.gvr.org (root@gvr.gvr.org [194.151.74.97])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id MAA09681
          for <FreeBSD-hackers@FreeBSD.ORG>; Sat, 23 Aug 1997 12:08:47 -0700 (PDT)
Received: (from guido@localhost)
	by gvr.gvr.org (8.8.6/8.8.5) id VAA06497;
	Sat, 23 Aug 1997 21:08:41 +0200 (MET DST)
From: Guido van Rooij <guido@gvr.org>
Message-Id: <199708231908.VAA06497@gvr.gvr.org>
Subject: Re: broadcast question
In-Reply-To: <19970823143813.EY59755@uriah.heep.sax.de> from J Wunsch at "Aug 23, 97 02:38:13 pm"
To: joerg_wunsch@uriah.heep.sax.de
Date: Sat, 23 Aug 1997 21:08:41 +0200 (MET DST)
Cc: FreeBSD-hackers@FreeBSD.ORG
X-Mailer: ELM [version 2.4ME+ PL28 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

J Wunsch wrote:
> As Guido van Rooij wrote:
> 
> > If I have a host in a subnetted C-net, say 192.1.1.0/28
> > and I send an icmp echo request to the broadcast address of the C-net
> > (so NOT to 192.1.1.15) the 2.1.7 stack does send an echo reply
> > but the 2.2 stack does not.
> > So an all-subnets directed broadcast seems to be ignored on 2.2
> > stacks...I think this broken, yet fail to see where it was broken.
> 
> I think 2.1.x was broken.  What is an ``all-subnets'' broadcast in a
> CIDR world?  Forget about class A/B/C, forget the term `subnet', they
> don't exist anymore. (*) The address 192.1.1.255 is not in any way a
> valid broadcast address for net 192.1.1.0/28, it's a valid broadcast
> address for 192.1.1.128/25, or maybe for 192.1.1.240/28 (which are
> entirely different networks from 192.1.1.0/28).

I don't agree. See the first Stevens book. There actually are two kinds
of broadcasts. Anyway: even if it is a bug, I still can't find where
it was fixed. This makes me suspicious.

> 
> (*) The only meaning of the historic classes A/B/C that does still
> exist is that they determine the default netmask if nothing else has
> been specified in ifconfig(8), or route(8), etc.  This can sometimes
> be convenient to type.
> 
> 255.255.255.255 should work, i believe.
> 
> I'm surprised 2.1.x was still broken in this respect.

At least 2.1 and 2.2 and higher differ here ;-)

-Guido