Return-path: Received: from hiems2.ing.unibs.it ([192.167.23.204]:44253 "EHLO hiems2.ing.unibs.it" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752666AbZBXQ3d (ORCPT ); Tue, 24 Feb 2009 11:29:33 -0500 Cc: Lorenzo Nava , wireless , Michael Buesch Message-Id: <03E07CC7-A971-4EDD-9363-628B61E94A92@ing.unibs.it> (sfid-20090224_172936_470162_3A928BAF) From: Francesco Gringoli To: Larry Finger In-Reply-To: <49A41F1D.9060300@lwfinger.net> Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Mime-Version: 1.0 (Apple Message framework v930.3) Subject: Re: More data on open-source firmware crash Date: Tue, 24 Feb 2009 17:29:28 +0100 References: <49A1A331.9080205@lwfinger.net> <80C13710-6BA9-4F14-95B5-82C3043CE93B@ing.unibs.it> <49A41F1D.9060300@lwfinger.net> Sender: linux-wireless-owner@vger.kernel.org List-ID: On Feb 24, 2009, at 5:23 PM, Larry Finger wrote: > > Francesco, > > I don't think the problem results from an overrun of the TX FIFO. In > my latest > test, I added code in routine handle_irq_transmit_status() that > would print a > message if the while loop count reached 16 or more. That print did > not occur. > The largest queue depth reached during the run was 14. When the first > out-of-order cookie was received, the largest depth in the last 64 > transactions > was only 5. > > I'm going to increase the number of transactions saved for my next > run. Any > other info that I should capture? What about drop info from dmesg? Another try could be halve the dma queue, I believe now it is 64 slots. Could you give a try setting it to 32? Cheers, -FG > > > Larry ------- Francesco Gringoli, PhD - Assistant Professor Dept. of Electrical Engineering for Automation University of Brescia via Branze, 38 25123 Brescia ITALY Ph: ++39.030.3715843 FAX: ++39.030.380014 WWW: http://www.ing.unibs.it/~gringoli