From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from aserp2120.oracle.com ([141.146.126.78]:45156 "EHLO aserp2120.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933925AbeF1R3v (ORCPT ); Thu, 28 Jun 2018 13:29:51 -0400 Date: Thu, 28 Jun 2018 10:29:37 -0700 From: "Darrick J. Wong" Subject: [PATCH 11/10] xfs_repair: clear extent size hints when clearing inode core Message-ID: <20180628172937.GI5711@magnolia> References: <153006766483.20121.9285982017465570544.stgit@magnolia> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <153006766483.20121.9285982017465570544.stgit@magnolia> Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: sandeen@redhat.com Cc: linux-xfs@vger.kernel.org From: Darrick J. Wong In kernel 4.18 we become more strict about what can be in the extent size hint fields, even for freed inodes. Therefore, if repair decides to clear out an inode core, zero the hint fields and clear the flags so that the kernel won't trip over the cleared inode if and when it tries to read the chunk. Signed-off-by: Darrick J. Wong --- repair/dinode.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/repair/dinode.c b/repair/dinode.c index f9b2bac0..4118db7c 100644 --- a/repair/dinode.c +++ b/repair/dinode.c @@ -194,6 +194,11 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) dinoc->di_anextents = 0; } + if (be32_to_cpu(dinoc->di_extsize) != 0) { + __dirty_no_modify_ret(dirty); + dinoc->di_extsize = 0; + } + if (dinoc->di_version > 1 && be32_to_cpu(dinoc->di_nlink) != 0) { __dirty_no_modify_ret(dirty); @@ -237,6 +242,11 @@ clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num) dinoc->di_changecount = 0; } + if (be32_to_cpu(dinoc->di_cowextsize) != 0) { + __dirty_no_modify_ret(dirty); + dinoc->di_cowextsize = 0; + } + return dirty; }