From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from brockman.in8.de ([85.214.220.56]:33806 "EHLO mail.in8.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753121Ab2GDNxg (ORCPT ); Wed, 4 Jul 2012 09:53:36 -0400 Message-ID: <4FF44ADE.8000207@jan-o-sch.net> Date: Wed, 04 Jul 2012 15:53:34 +0200 From: Jan Schmidt MIME-Version: 1.0 To: ziegler@uni-freiburg.de CC: Josef Bacik , "linux-btrfs@vger.kernel.org" Subject: Re: kmem_cache_destroy called for cache that still has objects References: <4FF1C140.3070705@jan-o-sch.net> In-Reply-To: <4FF1C140.3070705@jan-o-sch.net> Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-btrfs-owner@vger.kernel.org List-ID: Hi Martin, On 02.07.2012 17:41, Jan Schmidt wrote: > I'm recently seeing such buffer leaks as well. It's always about tree blocks, > and it's always just two or three refs left for a buffer. I can somehow > reproduce it and will be looking into it probably tomorrow. (May still be > unrelated, of course.) Okay, I've tracked my buffer leak down. It's most likely unrelated to yours, as you need Chris' current for-linus for that to occur. Second, you need to trigger a lot of tree mod log operations (i.e. by calling "btrfs inspect-internal" while modifying the trees). I'm sending a fix for the problem I found. However, there's no obvious reproducer for your problem yet. Can you reproduce it? -Jan