From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hong Liu Subject: Re: [patch 1/2]d80211: hardware TKIP support for ipw3945 Date: Tue, 24 Oct 2006 16:38:54 +0800 Message-ID: <1161679134.7258.6.camel@devlinux-hong> References: <1161335976.21402.11.camel@devlinux-hong> <20061023144028.629dc5d8@griffin.suse.cz> <1161607680.6617.36.camel@ux156> <20061023145646.32bbd59f@griffin.suse.cz> <1161678053.7083.7.camel@devlinux-hong> <1161678908.2840.12.camel@ux156> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: Jiri Benc , "John W. Linville" , netdev Return-path: Received: from mga02.intel.com ([134.134.136.20]:42091 "EHLO mga02.intel.com") by vger.kernel.org with ESMTP id S1030223AbWJXIpo (ORCPT ); Tue, 24 Oct 2006 04:45:44 -0400 To: Johannes Berg In-Reply-To: <1161678908.2840.12.camel@ux156> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Tue, 2006-10-24 at 16:35, Johannes Berg wrote: > On Tue, 2006-10-24 at 16:20 +0800, Hong Liu wrote: > > > It's really strange bcm43xx only does phase2 mixing in hw. With all the > > code that does phase2 mixing, doing phase1 mixing is very cheap. > > Yeah, well, they have the tkip sbox in hw. > > > With Jiri's ieee80211_tx_ctrl_common patch, I think we can put the > > rc4key or phase1 key in ieee80211_tx_control with each TX packet. > > No, the phase1 key is not dependent on the packet so it should not be > there. It should be set via the regular key setting mechanism, instead > of the TKIP key the driver would be given the phase1 key if required. > > johannes The first time when we set the TKIP key, we can set the phase1 key if the driver requires. The problem is when IV16 wraps, who will generate the new phase1 key? If the stack need to do this, then we will need to call set_key in the packet TX path which I think may not be appropriate. Thanks, Hong