Return-path: Received: from mail-wm0-f46.google.com ([74.125.82.46]:35189 "EHLO mail-wm0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753759AbdERK3I (ORCPT ); Thu, 18 May 2017 06:29:08 -0400 Received: by mail-wm0-f46.google.com with SMTP id b84so196368615wmh.0 for ; Thu, 18 May 2017 03:29:08 -0700 (PDT) Subject: Re: [PATCH V2 0/9] nl80211: add support for PTK/GTK handshake offload To: Johannes Berg References: <1493808134-4074-1-git-send-email-arend.vanspriel@broadcom.com> <1495030794.2442.21.camel@sipsolutions.net> <1de42f39-1912-349b-e20d-4b5c3c44909f@broadcom.com> <1495099355.2553.1.camel@sipsolutions.net> Cc: linux-wireless , "hostap@lists.infradead.org" From: Arend Van Spriel Message-ID: <4e0672aa-51bc-115f-32b7-b1a8eb747e5b@broadcom.com> (sfid-20170518_122912_858528_27558AAB) Date: Thu, 18 May 2017 12:29:04 +0200 MIME-Version: 1.0 In-Reply-To: <1495099355.2553.1.camel@sipsolutions.net> Content-Type: text/plain; charset=utf-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: On 18-5-2017 11:22, Johannes Berg wrote: > On Thu, 2017-05-18 at 10:18 +0200, Arend Van Spriel wrote: >> >>> We should therefore probably set the expectation that wpa_s - if >>> it's new enough - always uses the offloaded functionality and >>> always sets the WANT_1X. Then this is even better with such >>> drivers, since they can immediately reject the connect() command if >>> want_1x isn't set. Getting back at this. With "always" you mean for every connect() regardless whether it is using 1X or PSK? >> Personally I am fine with this and it is how I tested it. So no >> network configuration or driver parameter in wpa_s.conf (mainly >> because I am lazy ;-) ). However, if the driver supports both offload >> and non-offload why not leave it up to user-space. Might be useful if >> people can try either way for example when debugging connection >> issues. > > If that should be possible though, then we'd probably want to have > feature bit both ways, otherwise we'll probably get hard-to-debug > failures at some point with this? You mean adding a nl80211 command in which user-space can indicate what features it supports? Do you want to use the same feature bits on both sides to easily determine the combined feature set? ext_feature does not really have much overlapping so not sure if it adds value. Regards, Arend