From:   zhanchengbin <zhanchengbin1@huawei.com>
To:     <tytso@mit.edu>, <jack@suse.com>
CC:     <linux-ext4@vger.kernel.org>, <yi.zhang@huawei.com>,
        <linfeilong@huawei.com>, <liuzhiqiang26@huawei.com>,
        zhanchengbin <zhanchengbin1@huawei.com>
Subject: [PATCH v4 0/2] fix extents need to be restored when ext4_ext_insert_extent failed
Date:   Mon, 13 Feb 2023 12:05:20 +0800
Message-ID: <20230213040522.3339406-1-zhanchengbin1@huawei.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 7BIT
Content-Type:   text/plain; charset=US-ASCII
Precedence: bulk

Inside the ext4_ext_insert_extent function, every error returned will
not destroy the consistency of the tree. Even if it fails after changing
half of the tree, can also ensure that the tree is self-consistent, like
function ext4_ext_create_new_leaf.
After ext4_ext_insert_extent fails, update extent status tree depends on
the incoming split_flag. So restore the len of extent to be split when
ext4_ext_insert_extent return failed in ext4_split_extent_at.

Diff v2 Vs v1:
1) return directly after inserting successfully
2) restore the length of extent in memory after inserting unsuccessfully

Diff v3 Vs v2:
Sorry for not taking into account the successful extent insertion. and I
reanalyzed the ext4_ext_insert_extent function and modified the conditions
for restoring the length.

Diff v4 Vs v3:
Clear the verified flag from the modified bh when failed inext4_ext_rm_idx
or ext4_ext_correct_indexes.

zhanchengbin (2):
  ext4: fix inode tree inconsistency caused by ENOMEM in
    ext4_split_extent_at
  ext4: clear the verified flag of the modified leaf or idx if error

 fs/ext4/extents.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

-- 
2.31.1