From mboxrd@z Thu Jan 1 00:00:00 1970 From: Theodore Ts'o Subject: Re: metadata_csum set but no space in dir leaf for checksum Date: Sun, 14 Oct 2012 10:55:08 -0400 Message-ID: <20121014145508.GC6207@thunk.org> References: <20121013050554.9442.qmail@science.horizon.com> <65D4536C-0F58-4A26-A562-B665EC6CBBB2@dilger.ca> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: George Spelvin , linux-ext4@vger.kernel.org To: Andreas Dilger Return-path: Received: from li9-11.members.linode.com ([67.18.176.11]:55443 "EHLO imap.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754760Ab2JNOzN (ORCPT ); Sun, 14 Oct 2012 10:55:13 -0400 Content-Disposition: inline In-Reply-To: <65D4536C-0F58-4A26-A562-B665EC6CBBB2@dilger.ca> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Sun, Oct 14, 2012 at 12:43:47AM -0600, Andreas Dilger wrote: > > This means there was an existing directory block that was completely full and > could not have the checksum added. It isn't harmful, the chance of having > silent corruption in this specific block is very small. Was this a freshly created ext4 file system with the metadata_csum checksum, or was this a previously existing ext4 file system where the metadata_csum feature was added later? I've pushed an update to the e2fsprogs repository which allows htree and "ls -c" to actually show us the directory leaf block checksums. Previously, they were hidden, which means that it's hard to tell whether a directory had all of its directory blocks properly checksummed or not. > The message itself should be quieted though, since there isn't anything to be > helped by printing it continuously. I guess there is something missing in > e2fsck that it doesn't add a checksum to this block, however. Either that, or somehow there is a case where the directory tail containing the checksum is getting overwritten, or a directory block isn't getting a checksum after a htree node split. The e2fsprogs changes will hopefully make it easier for us to figure out what is going on. - Ted