Return-path: Received: from charlotte.tuxdriver.com ([70.61.120.58]:37118 "EHLO smtp.tuxdriver.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751261AbYKLSqJ (ORCPT ); Wed, 12 Nov 2008 13:46:09 -0500 Date: Wed, 12 Nov 2008 13:35:10 -0500 From: "John W. Linville" To: Harvey Harrison Cc: linux-wireless@vger.kernel.org Subject: Re: [RFC PATCH 1/4] lib80211: absorb crypto bits from net/ieee80211 Message-ID: <20081112183509.GE2411@tuxdriver.com> (sfid-20081112_194614_425067_4DB5A030) References: <1226440265-21240-1-git-send-email-linville@tuxdriver.com> <1226508514.5368.21.camel@brick> <20081112180228.GC2411@tuxdriver.com> <1226513943.16649.0.camel@brick> <20081112182814.GD2411@tuxdriver.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <20081112182814.GD2411@tuxdriver.com> Sender: linux-wireless-owner@vger.kernel.org List-ID: On Wed, Nov 12, 2008 at 01:28:14PM -0500, John W. Linville wrote: > On Wed, Nov 12, 2008 at 10:19:03AM -0800, Harvey Harrison wrote: > > On Wed, 2008-11-12 at 13:02 -0500, John W. Linville wrote: > > > On Wed, Nov 12, 2008 at 08:48:34AM -0800, Harvey Harrison wrote: > > > > On Tue, 2008-11-11 at 16:51 -0500, John W. Linville wrote: > > > > > These bits are shared already between ipw2x00 and hostap, and could > > > > > probably be shared both more cleanly and with other drivers. This > > > > > commit simply relocates the code to lib80211 and adjusts the drivers > > > > > appropriately. > > > > > > > > > > Signed-off-by: John W. Linville > > > > > --- > > > > > > > > Wouldn't it have been better to start from the mac80211 crypto bits > > > > as lib80211 and modify hostap/ipw2x00 to use the 'new' lib80211 > > > > crypto bits? > > > > > > I'm sorry, but you forgot to attach your patches... > > > > > > > I will if it's desired, I just wanted to know if there was a particular > > reason for going this way. > > Those bits are already shared between ieee80211 and the hostap driver. > Unfortunately, they are not shared with a clean API. I don't think > hacking on mac80211's crypto code is worthwhile until these drivers > are given a bit more discipline in this area. IMHO, this is the > better way to start. Oh, and removing hostap's dependency on ieee80211 enables us to turn ieee80211 into libipw now instead of waiting even longer while we refactor all the crypto code, potentially destabilizing ipw2100, ipw2200, hostap and (potentially) mac80211 along the way. John -- John W. Linville Linux should be at the core linville@tuxdriver.com of your literate lifestyle.