From: Eric Sandeen Subject: Re: [PATCH] ext3: truncate block allocated on a failed ext3_write_begin Date: Thu, 28 Aug 2008 12:56:28 -0500 Message-ID: <48B6E6CC.7030704@redhat.com> References: <1219944473-8770-1-git-send-email-aneesh.kumar@linux.vnet.ibm.com> <1219944473-8770-2-git-send-email-aneesh.kumar@linux.vnet.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: cmm@us.ibm.com, tytso@mit.edu, linux-ext4@vger.kernel.org To: "Aneesh Kumar K.V" Return-path: Received: from mx2.redhat.com ([66.187.237.31]:46207 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753222AbYH1SOA (ORCPT ); Thu, 28 Aug 2008 14:14:00 -0400 In-Reply-To: <1219944473-8770-2-git-send-email-aneesh.kumar@linux.vnet.ibm.com> Sender: linux-ext4-owner@vger.kernel.org List-ID: Aneesh Kumar K.V wrote: > For blocksize < pagesize we need to remove blocks that > got allocte in block_wirte_begin if we fail with ENOSPC > for later blocks. block_write_begin internally does > this if it allocated page locally. This make sure > we don't have blocks outisde inode.i_size during > ENOSPC I think this is good; here's an easy testcase: # dd if=/dev/zero of=fsfile bs=1k count=2048 # mkfs.ext3 -F fsfile # mkdir mnt # mount -o loop fsfile mnt/ # dd if=/dev/zero of=mnt/1kfile bs=1k count=1 # cd mnt/ # dd if=/dev/zero of=bigfile bs=4k # cd .. # umount mnt # e2fsck -f fsfile e2fsck 1.39 (29-May-2006) Pass 1: Checking inodes, blocks, and sizes Inode 13, i_size is 974848, should be 976896. Fix? ... can you test with that, unless you already have a testcase you've used? Assuming it fixes it (it should) you can add: Acked-by: Eric Sandeen -Eric > Signed-off-by: Aneesh Kumar K.V > --- > fs/ext3/inode.c | 7 +++++++ > 1 files changed, 7 insertions(+), 0 deletions(-) > > diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c > index ebfec4d..f8424ad 100644 > --- a/fs/ext3/inode.c > +++ b/fs/ext3/inode.c > @@ -1186,6 +1186,13 @@ static int ext3_write_begin(struct file *file, struct address_space *mapping, > ext3_journal_stop(handle); > unlock_page(page); > page_cache_release(page); > + /* > + * block_write_begin may have instantiated a few blocks > + * outside i_size. Trim these off again. Don't need > + * i_size_read because we hold i_mutex. > + */ > + if (pos + len > inode->i_size) > + vmtruncate(inode, inode->i_size); > } > if (ret == -ENOSPC && ext3_should_retry_alloc(inode->i_sb, &retries)) > goto retry;