From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mail-ew0-f219.google.com ([209.85.219.219]:59022 "EHLO mail-ew0-f219.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751172AbZLWM2L (ORCPT ); Wed, 23 Dec 2009 07:28:11 -0500 Received: by ewy19 with SMTP id 19so6258401ewy.21 for ; Wed, 23 Dec 2009 04:28:10 -0800 (PST) MIME-Version: 1.0 In-Reply-To: <1261061645.10356.114.camel@johannes.local> References: <1261061645.10356.114.camel@johannes.local> Date: Wed, 23 Dec 2009 14:28:10 +0200 Message-ID: Subject: Re: Backoff algorithm From: Andrey Lukyanenko To: Johannes Berg Cc: linux-wireless@vger.kernel.org Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: Thanks a lot for your help. I have found OpenFWWF project, which has open source firmware for Broadcom chipsets. Unfortunately it does not support CTS/RTS, AP. What do you mean by " influencing the value you give the hw"? Are you talking about direct access to registers? How can I influence. I have heard about tos-mac (TinyOS) which uses directly registers but it is for sensors. Do you mean something similar? In our group we have five ar5213 chips, I hope I can try to use it. On Thu, Dec 17, 2009 at 4:54 PM, Johannes Berg wrote: > On Thu, 2009-12-17 at 16:10 +0200, Andrey Lukyanenko wrote: >> Hello Everyone, >> >> I wanted to find implementation of backoff protocols for different platforms. >> I tried to find it in various open-source drivers in linux-sources, madwifi and >> stlc45xx. Everywhere I could only find some initialization of CWmax, CWmin, ... >> but not the actual algorithm (something like, cw = min(cw<<1 + 1, cwmax) ), >> except ath_rate sample/mistrel where the operation is used for "best >> rate measurement" >> as I understood. >> Am I correct to assume that the protocol is implemented in firmwhere and there >> is no open source driver which I could use to "play" directly with it? > > Yes, basically. It's not something you can really implement in any > driver anyway as it's very timing sensitive. > > ar9170 even has this in hardware, afaik, although you can of course play > with it in b43 firmware and/or ar9170 by influencing the value you give > the hw. > > johannes > -- Best regards, Andrey Lukyanenko.