Return-path: Received: from smtp01.msg.oleane.net ([62.161.4.1]:34556 "EHLO smtp01.msg.oleane.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751665AbbCXImE convert rfc822-to-8bit (ORCPT ); Tue, 24 Mar 2015 04:42:04 -0400 From: "voncken" To: "'Michal Kazior'" Cc: "'linux-wireless'" References: <773DB8A82AB6A046AE0195C68612A319019FEF93@sbs2003.acksys.local> In-Reply-To: Subject: RE: ath10k: What will happens when radar is detected ? Date: Tue, 24 Mar 2015 09:41:59 +0100 Message-ID: <01eb01d0660e$650becc0$2f23c640$@acksys.fr> (sfid-20150324_094210_304494_B88695E4) MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Sender: linux-wireless-owner@vger.kernel.org List-ID: > -----Message d'origine----- > De : Michal Kazior [mailto:michal.kazior@tieto.com] > Envoyé : mardi 24 mars 2015 07:56 > À : Cedric VONCKEN > Cc : linux-wireless > Objet : Re: ath10k: What will happens when radar is detected ? > > On 23 March 2015 at 16:16, Cedric VONCKEN > wrote: > > In 802.11ac standard, it is possible to dynamically reduce the channel > > width used. > > My question is: > > If I use a channel with 80 or 40 MHz width, what will happen > > if I detect radar? > > The ath10k card/driver reduces automatically the > > channel width. > > Currently entire chandef occupied will be marked as unavailable and AP > will stop/switch to a different non-overlapping chandef via CSA. Thanks for your answer. > > I guess it should be possible to implement what you suggest, i.e. > change only chandef width when radar is narrow and located at a suitable > part of the chandef. This would require hw to be capable of detecting > radar center freq and width accurately. I'm not sure if QCA988X is > capable of that although firmware interface seems to be able to carry > this kind of information already. > > I wonder why would you want this behaviour in the first place? > Wouldn't this actually end up with having less bandwidth and lower > throughput (which is already penalized when radar detection is active)? In an industrial environment we prefer to reduce the throughput but keep the link without gap. I think it can be a parameter in AP software for example. Cédric