Return-path: Received: from mail.deathmatch.net ([72.66.92.28]:4490 "EHLO mail.deathmatch.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932998Ab0GTXQI (ORCPT ); Tue, 20 Jul 2010 19:16:08 -0400 Date: Tue, 20 Jul 2010 19:16:29 -0400 From: Bob Copeland To: Derek Smithies Cc: Jonathan Guerin , "Ath5k Devel issues." , linux-wireless@vger.kernel.org Subject: Re: [ath5k-devel] ath5k Ad Hoc Association Message-ID: <20100720231629.GC7049@hash.localnet> References: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: Sender: linux-wireless-owner@vger.kernel.org List-ID: On Wed, Jul 21, 2010 at 10:11:14AM +1200, Derek Smithies wrote: >> Then on the other machine (Vista) I just located 'myibss' (so beacons >> are working) and selected it, then successfully pinged 192.168.10.1. > Wrong. > If the other machine located your node, then one of the four statements > is true: Ok, fine, I just set up a monitor interface and I see beacons (and probe responses) for my setup on that channel. > A scan should not be necessary - if both ends are put on channel 2462- a > scan should not be required - there is no channel changing. Yet - my > colleague noted a scan helped nodes to associate. There was a bug in older kernels that wouldn't fully configure the beacons until after a reset, which scanning happens to do. It's possible something similar is going on. Also, scanning might open a previously passive channel because we might get lucky and receive a beacon from another node on it. The 'iw phy phy0 info' can show this kind of thing. -- Bob Copeland %% www.bobcopeland.com