Return-path: Received: from Viola.Opus1.COM ([192.245.12.8]:1535 "EHLO Viola.Opus1.COM" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751669AbYI3NqI (ORCPT ); Tue, 30 Sep 2008 09:46:08 -0400 Received: from [10.1.1.4] ([204.17.34.178]) by Opus1.COM (PMDF V6.2-X27 #9830) with ESMTPSA id <01N05W5YYT808YT9SM@Opus1.COM> for linux-wireless@vger.kernel.org; Tue, 30 Sep 2008 06:46:04 -0700 (MST) Date: Tue, 30 Sep 2008 06:46:04 -0700 From: gavron@wetwork.net Subject: Re: [RFC/T] b43: to few loop tries in do_dummy_tx In-reply-to: <200809301528.26304.mb@bu3sch.de> To: Michael Buesch Cc: bcm43xx-dev@lists.berlios.de, wireless , Peter Stuge Message-id: <48E22D9C.9090404@wetwork.net> (sfid-20080930_154614_511803_31971DC3) MIME-version: 1.0 Content-type: text/plain; charset=ISO-8859-1; format=flowed References: <48E11F1E.50705@lwfinger.net> <48E14879.4050300@lwfinger.net> <20080930055034.16412.qmail@stuge.se> <200809301528.26304.mb@bu3sch.de> Sender: linux-wireless-owner@vger.kernel.org List-ID: Michael Buesch wrote: > On Tuesday 30 September 2008 07:50:34 Peter Stuge wrote: > >> Larry Finger wrote: >> >>>> Which specs? >>>> >>> The ones generated by the reverse engineers. See >>> http://bcm-v4.sipsolutions.net/. >>> >> Nice work, but as it's a spec of another driver implementation rather >> than hardware (or even the firmware API) I don't think it should be >> so authoritative. If other values are clearly better why not use >> them? >> > > What crap are you smoking? > The b43 and b43-legacy driver are _based_ on these specifications. > There are no other specs available. > > If I understand him correctly he's suggesting that there could be BETTER values than those used by the reference driver. In other words, yes, B43/B43-Legacy are based on the RE of the Windows driver but perhaps there are better values that improve behavior beyond that of the original driver. He didn't say the following but I will: It's also true that there are edge cases that RE won't catch without repeated arduous testing in adverse conditions, and there may be code in the reference driver that will therefore won't end up in the specs. This means that behavioral improvements and/or performance gains in B43/B43-Legacy that can be gained without getting into those edge cases are worthy of consideration (or maybe specially labeled code). Just my two farthings worth. E -- Legal Disclaimer that you are now contractually bound to under all laws with no recourse: http://attrition.org/security/rants/z/disclaimers.html