From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Hansen Subject: Re: [PATCH 1/1] net/unix: Allow Unix sockets to be treated like normal files Date: Wed, 27 Oct 2010 20:50:17 -0600 Message-ID: <4CC8E4E9.2060209@jeffhansen.com> References: <1288232669-8927-1-git-send-email-x@jeffhansen.com> <20101027.193212.193709254.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org To: David Miller Return-path: Received: from out01.mta.xmission.com ([166.70.13.231]:49717 "EHLO out01.mta.xmission.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932744Ab0J1CuX (ORCPT ); Wed, 27 Oct 2010 22:50:23 -0400 In-Reply-To: <20101027.193212.193709254.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: Dave, I agree that on larger systems this doesn't really make sense, but on embedded platforms this can save some code space since applications can get rid of their FIFO listeners and have strictly socket listeners. That's why I made it an option that could be disabled by default. Do you know who originally suggested that "creepy crawlies" are introduced by allowing sockets to be opened? I'd be interested to know how this could affect security, if at all. -Jeff On 10/27/2010 08:32 PM, David Miller wrote: > From: x@jeffhansen.com > Date: Thu, 28 Oct 2010 02:24:29 +0000 > >> Resent. Is there anything else I need to do for this patch to get reviewed >> and/or merged? Any comments by anyone? > I don't like this idea at all. > > I remember there is a reason why similar things are not allowed for > sockets, it causes all sorts of problems although I forget the exact > details. > > Take a look at net/socket.c:sock_no_open(), for example. > -- --------------------------------------------------- "If someone's gotta do it, it might as well be me." x@jeffhansen.com