Return-path: Received: from mail-bw0-f227.google.com ([209.85.218.227]:41240 "EHLO mail-bw0-f227.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1759052AbZKFTpE convert rfc822-to-8bit (ORCPT ); Fri, 6 Nov 2009 14:45:04 -0500 Received: by bwz27 with SMTP id 27so1544917bwz.21 for ; Fri, 06 Nov 2009 11:45:08 -0800 (PST) MIME-Version: 1.0 In-Reply-To: <1257327017.28469.76.camel@johannes.local> References: <1257327017.28469.76.camel@johannes.local> Date: Fri, 6 Nov 2009 14:45:08 -0500 Message-ID: Subject: Re: iwlwifi connection troubles, maybe aggregation related From: Andrew Lutomirski To: Johannes Berg Cc: linux-wireless@vger.kernel.org, ilw@linux.intel.com Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-wireless-owner@vger.kernel.org List-ID: On Wed, Nov 4, 2009 at 4:30 AM, Johannes Berg wrote: > On Tue, 2009-11-03 at 18:38 -0500, Andrew Lutomirski wrote: > >> iwlagn 0000:03:00.0: iwl_tx_agg_start on ra = 00:21:d8:49:4a:52 tid = 0 > >> Turning on or off power management and fiddling with >> no_sleep_autoadjust makes no difference. ?Setting tx_agg_tid_enable to >> zero in debugfs while the connection was working seemed to make it a >> little more reliable (it lasted long enough to do "git pull" but not >> much longer). >> >> After running "iw dev wlan0 disconnect" a few times, I start to get >> errors like this: >> >> [18078.209635] iwlagn 0000:03:00.0: SENSITIVITY_CMD failed >> [18078.313461] iwlagn 0000:03:00.0: No space for Tx >> [18078.313467] iwlagn 0000:03:00.0: Error sending SENSITIVITY_CMD: >> enqueue_hcmd failed: -28 >> [18078.313470] iwlagn 0000:03:00.0: SENSITIVITY_CMD failed >> [18078.522409] iwlagn 0000:03:00.0: No space for Tx >> [18078.522414] iwlagn 0000:03:00.0: Error sending SENSITIVITY_CMD: >> enqueue_hcmd failed: -28 > > Sounds like the firmware messes up ... > > Maybe as a first workaround you could modprobe iwlagn with > 11n_disable=1. But I don't know at this point what the problem could be. 11n_disable50=1 seems to work. I'll reply again if it stops working. Do the Intel folks have any ideas? Thanks, Andy