From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from li9-11.members.linode.com ([67.18.176.11]:34703 "EHLO test.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753553Ab1LMR0p (ORCPT ); Tue, 13 Dec 2011 12:26:45 -0500 Date: Tue, 13 Dec 2011 12:26:42 -0500 From: Ted Ts'o To: linux-kernel@vger.kernel.org Cc: ilw@linux.intel.com, linux-wireless@vger.kernel.org, Johannes Berg Subject: Re: REGRESSION: v3.2-rcX: iwlagn refuses to associate with my AP's Message-ID: <20111213172642.GA4962@thunk.org> (sfid-20111213_182706_919934_1757839A) 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 Mon, Dec 12, 2011 at 02:42:53PM -0500, Theodore Ts'o wrote: > Hi there, > > I recently discovered a regression in the iwlagn driver (works in v3.1 > but not in v3.2-rc2... I'm going to continue to bisect, but I thought > I'd give a heads up now). I'm using a Lenovo T410, with the following > wireless card: > > [ 11.197410] iwlwifi 0000:03:00.0: HW Revision ID = 0x35 > [ 11.197637] iwlwifi 0000:03:00.0: Detected Intel(R) Centrino(R) Advanced-N + WiMAX 6250 AGN, REV=0x84 > > The failure to associate with access points happens at work, with > whatever AP's are in use at the Cambridge Google office. When I tried > v3.2-rc5 at home, I was able to associate with a consumer-grade NetGear > AP, although it was flaky --- that is, it completely failed to associate > initially, but then I tried rebooting and I was eventually able to get > it to work. At work, I was able to reproduce the problem with a > v3.2-rc5 and v3.2-rc2 kernel, but the problem did not manifest itself > with a v3.1 kernel. OK, a follow up. A bisection puts the finger of blame very firmly at: debcf73 iwlagn: handle GO powersave Specifically, I can't associate with *any* access points (encrypted or not) if I compile a kernel at commit debcf73, but if I compile a kernel with its parent (commit 8ad71be), it works fine. However, as a puzzler, when I tried reverted compiling 3.2-rc5 with debcf73 reverted, it did not make the problem go away. So there's something more going on here. I could try doing a bisect with reverts of debcf73 at each step, trying to find the problem, but I'm hoping this is enough of a hint that someone can tell me either that this is fixed already in the linux-wireless tree, or that they know what's wrong and can provide a test patch... Thanks, - Ted