From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mtiwmhc13.worldnet.att.net ([204.127.131.117]:35727 "EHLO mtiwmhc13.worldnet.att.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754025AbZAMT7y (ORCPT ); Tue, 13 Jan 2009 14:59:54 -0500 Message-ID: <496CF2B3.1000402@lwfinger.net> (sfid-20090113_210902_480524_DD2F5520) Date: Tue, 13 Jan 2009 13:59:47 -0600 From: Larry Finger MIME-Version: 1.0 To: Christian Lamparter CC: Artur Skawina , linux-wireless@vger.kernel.org Subject: Re: wireless-testing, p54 and sinus 154 data no longer works References: <494698AF.4020204@gmail.com> <200901131449.41189.chunkeey@web.de> <496CC514.3030307@gmail.com> <200901131906.13096.chunkeey@web.de> In-Reply-To: <200901131906.13096.chunkeey@web.de> Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: Christian Lamparter wrote: > well, you have a 32-bit architecture, unlike me or larry. > So, with a bit of luck the skbs always fitted into a single page and you never > had problems with memory fragmentation. Whether it is a 32- or 64-bit kernel, it has a 4K page size. If that were not true, the rx_mtu reduction would not have triggered. I have additional information obtained from a kmemleak kernel. Driver p54usb and mac80211 do not show any memory leaks, at least in STA mode. I tested this by running normal traffic for about 15 minutes, and then rmmod'ed p54usb. If kernel leaks are the problem, then they are arising from hostapd. Userland memory leaks can be checked with valgrind. Larry