Return-path: Received: from mail-yw0-f176.google.com ([209.85.211.176]:48290 "EHLO mail-yw0-f176.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752666Ab0AYRjl convert rfc822-to-8bit (ORCPT ); Mon, 25 Jan 2010 12:39:41 -0500 Received: by ywh6 with SMTP id 6so3299067ywh.4 for ; Mon, 25 Jan 2010 09:39:41 -0800 (PST) MIME-Version: 1.0 In-Reply-To: <4B5CD1FF.4090209@twilightblue.net> References: <4B5CD1FF.4090209@twilightblue.net> Date: Mon, 25 Jan 2010 12:39:40 -0500 Message-ID: Subject: Re: Bug: 2.6.32 ath5k, associating with WEP128 AP is unreliable From: Bob Copeland To: Don Darling Cc: linux-wireless@vger.kernel.org Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-wireless-owner@vger.kernel.org List-ID: On Sun, Jan 24, 2010 at 6:04 PM, Don Darling wrote: > Hello folks, > > I recently tried upgrading to kernel version 2.6.32, but after this upgrade > my wireless card became flaky when making connections. In cases where a > connection can be established it is ?stable, but there are problems when > making wireless associations. Downgrading to kernel version 2.6.31 resolves > this issue. ?I'm not sure at the moment which versions of the ath5k driver > maps to these kernel versions. The (vanilla) kernel version is sufficient to know which driver it is. > I'm assuming this issue is related to the ath5k driver itself -- the network > I'm using is only WEP128, and doesn't need wpa_supplicant. Huh, I didn't think any mac80211 driver supported WEP 128. Are you sure you weren't using madwifi before? Or maybe WEP 104? Are you using any modparams with ath5k? > In fact, I have > fewer problems when I connect to WPA+TKIP networks with wpa_supplicant. This is somewhat more secure, so I'd recommend it, or better CCMP. > Reverting to 2.6.31 resolves this issue. It will connect the first time, and > I can disconnect/re-connect several times without issue. I can't think of anything that changed in this time frame with ath5k specifically having to do with crypto. If you were using nohwcrypt=1, it could be a problem with mac80211. Otherwise, bisecting might be the best route to pin it down. Also on the off-chance that it's not crypto related, you may try the suggestions in the following to get a better picture of whwat is going on: http://wireless.kernel.org/en/users/Documentation/Reporting_bugs -- Bob Copeland %% www.bobcopeland.com