From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from lager.deltatee.com ([207.54.116.66]:53614 "EHLO porter.deltatee.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1755452Ab0GZVkd (ORCPT ); Mon, 26 Jul 2010 17:40:33 -0400 Message-ID: <4C4E00CE.5050105@deltatee.com> Date: Mon, 26 Jul 2010 15:40:30 -0600 From: Logan Gunthorpe MIME-Version: 1.0 To: linux-wireless@vger.kernel.org CC: Luciano Coelho , "Levi, Shahar" Subject: Re: wl1271 with atmel-mci References: <4C488EEA.5020505@deltatee.com> In-Reply-To: <4C488EEA.5020505@deltatee.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Sender: linux-wireless-owner@vger.kernel.org List-ID: Hi Everyone, > I've had the wl1271 working for the most part on an Atmel > micro-controller, but I've been fighting with a bug for the past couple > of days. > > The problem happens whenever I try to transmit large packets of data > (using iperf, or nc, etc) the driver would hang and no longer be able to > send packets. tx_queue_len in the debugfs would then grow as I tried to > send packets but nothing would actually be sent to the device. At the > same time receiving packets would consistently work all of the time. > > When this occurred I usually got the following messages: > > atmel_mci atmel_mci.0: data CRC error > wl1271: ERROR sdio write failed (-84) - addr 0x14fd8, 1076 bytes, 1 Firstly, I'd like to let anyone who's interested know that the problem I had above turned out to be a hardware issue. I tried two newly populated boards which both turned out not to exhibit this problem. I now have the wl1271 working on two AT91SAM9G45-EKES development kits. Using iperf I have gathered some performance data. I am seeing between a wl1271 device and a host PC (connected via a dlink router): PC to Device: ~23.0Mbit/s Device to PC: ~21.7Mbit/s When running in Ad-Hoc mode between two devices I am getting between about 15Mbit/s and 17Mbit/s. Does anyone have any insight as to why the Ad-Hoc mode between two devices is noticeably slower? Can anything be done to improve this? Thanks, Logan