Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <6cde43c8-3300-9269-7a8a-8ff6e8b1e287@gmail.com>
Date:   Wed, 22 Mar 2023 20:27:15 +0300
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.8.0
Subject: Re: A new special orphan inode 12 in ext4 only?
Content-Language: en-US
To:     Qu Wenruo <quwenruo.btrfs@gmx.com>,
        Ext4 <linux-ext4@vger.kernel.org>,
        "linux-btrfs@vger.kernel.org" <linux-btrfs@vger.kernel.org>
References: <4034e634-59d3-e9a5-a1c5-1f275d8e2832@gmx.com>
From:   Andrei Borzenkov <arvidjaar@gmail.com>
In-Reply-To: <4034e634-59d3-e9a5-a1c5-1f275d8e2832@gmx.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 22.03.2023 12:09, Qu Wenruo wrote:
> Hi,
> 
> Recently I observed newer mkfs.ext4 seems to create a new orphan inode
> 12, with some file extents.
> 
> Which seems to have no direct parent directory, thus tools like
> btrfs-convert would also follow the ext4 inodes by creating an orphan
> inode too.
> 
> On the other hand, if I go mkfs.ext3, the mysterious inode seems to be gone.
> 
> Is this inode 12 a known special inode?

This is orphan file. It is normal file; mke2fs creates first normal 
inode for lost+found (11) and if enabled creates orphan file next which 
gets next inode number (12). Inode number is recorded in superblock as 
s_orphan_file_num.

/*27c*/ __le16  s_encoding;             /* Filename charset encoding */
         __le16  s_encoding_flags;       /* Filename charset encoding 
flags */
         __le32  s_orphan_file_inum;     /* Inode for tracking orphan 
inodes */


> If so, how can we avoid such special inode?
> (s_special_ino is still 11, thus checking against that value doesn't
> seem to help).
> 
> 
> Some details of btrfs-convert:
> 
> It goes with ext2fs_open_inode_scan() to iterate all inodes of an ext4.
> 
> And if we hit an directory inode, we iterate the directory by using
> ext2fs_dir_iterate2() to insert the dir entries between parent and child
> inodes.
> 
> So if we hit an inode without any parent dir, an equivalent btrfs inode
> would still be created, but btrfs-check would complain about such orphan
> inode.
> 
> Thanks,
> Qu