From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.fusionio.com ([66.114.96.30]:41859 "EHLO mx1.fusionio.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760241Ab2FVNck (ORCPT ); Fri, 22 Jun 2012 09:32:40 -0400 Message-ID: <4FE473F6.5000109@fusionio.com> Date: Fri, 22 Jun 2012 09:32:38 -0400 From: Josef Bacik MIME-Version: 1.0 To: Dan Carpenter CC: "Chris L. Mason" , "linux-btrfs@vger.kernel.org" , "kernel-janitors@vger.kernel.org" Subject: Re: [patch] Btrfs: dereferencing free'd memory in panic message References: <20120622071433.GA27618@elgon.mountain> <4FE46E70.8070702@fusionio.com> <20120622133015.GF5390@mwanda> In-Reply-To: <20120622133015.GF5390@mwanda> Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Sender: linux-btrfs-owner@vger.kernel.org List-ID: On 06/22/2012 09:30 AM, Dan Carpenter wrote: > On Fri, Jun 22, 2012 at 09:09:04AM -0400, Josef Bacik wrote: >> On 06/22/2012 03:14 AM, Dan Carpenter wrote: >>> We free "node" and then dereference it in the panic message on the next >>> line. I considered moving the kfree() after the panic given that panic >>> can return under certain configurations, but in the end I decided it >>> doesn't matter if we leak a bit after a panic. >>> >>> Signed-off-by: Dan Carpenter >>> >>> diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c >>> index 790f492..c50d80a 100644 >>> --- a/fs/btrfs/relocation.c >>> +++ b/fs/btrfs/relocation.c >>> @@ -1239,7 +1239,6 @@ static int __must_check __add_reloc_root(struct btrfs_root *root) >>> node->bytenr, &node->rb_node); >>> spin_unlock(&rc->reloc_root_tree.lock); >>> if (rb_node) { >>> - kfree(node); >>> btrfs_panic(root->fs_info, -EEXIST, "Duplicate root found " >>> "for start=%llu while inserting into relocation " >>> "tree\n", node->bytenr); >> >> Except btrfs_panic can not panic the box if it's mounted to not >> panic on errors, so we still need to do the kfree afterwards. >> Thanks, > > Right. I mentioned that in my change log, but I figured a one time > memory leak was the least of our concerns in that case. I will > resend. This should probably return -EEXIST here as well yes? > Ah sorry I thought you were talking about BUG_ON() not actually stopping the box in some cases. But yes I'd like to not leak anyway and return -EEXIST. Thanks, Josef