Return-path: Received: from mail.candelatech.com ([208.74.158.172]:33177 "EHLO ns3.lanforge.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754322Ab3BUXxV (ORCPT ); Thu, 21 Feb 2013 18:53:21 -0500 Message-ID: <5126B368.9010209@candelatech.com> (sfid-20130222_005325_365483_8B0BB9D3) Date: Thu, 21 Feb 2013 15:53:12 -0800 From: Ben Greear MIME-Version: 1.0 To: Adrian Chadd CC: Sujith Manoharan , "linux-wireless@vger.kernel.org" Subject: Re: 3.7.6+: ath9k: tx logic locks up after taking attenuation very high. References: <511935F2.8080103@candelatech.com> <20761.34987.125000.908722@gargle.gargle.HOWL> <511BC243.8010409@candelatech.com> <20770.57339.20017.225929@gargle.gargle.HOWL> <51255360.6060603@candelatech.com> <51257304.10109@candelatech.com> <20773.31380.740958.31338@gargle.gargle.HOWL> <5125B39F.3010200@candelatech.com> <51265AC3.5030304@candelatech.com> <5126A993.8060408@candelatech.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Sender: linux-wireless-owner@vger.kernel.org List-ID: On 02/21/2013 03:49 PM, Adrian Chadd wrote: > .. so the current code correctly soft fails all of those frames and > puts them back onto the relevant software queues, right? I might have been mistaken. Further testing makes me think the pending-frames are properly dealt with. I see them show back up in the tx-complete logic about 30 second after the first chip reset logic hits. That seems quite long to me, but then again, it appears that without a chip reset, they never show back up in the tx-complete logic at all (and thus the hang). I'm testing now with just the pending-frames lockup check with a plain chip reset as work-around (and no fiddling with pending-frames counter). Thanks, Ben > > > Adrian > -- Ben Greear Candela Technologies Inc http://www.candelatech.com