linux-ext4 - Re: [PATCH, RFC] Ext3: Mount partition as read only if during orphan cleanup truncate fails to obtain journal handle.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Mon, 29 Oct 2012 11:49:39 +0100
From:	Jan Kara <jack@...e.cz>
To:	Ashish Sangwan <ashishsangwan2@...il.com>
Cc:	jack@...e.cz, linux-ext4@...r.kernel.org,
	Ashish Sangwan <ashish.sangwan2@...il.com>,
	Namjae Jeon <linkinjeon@...il.com>
Subject: Re: [PATCH, RFC] Ext3: Mount partition as read only if during
 orphan cleanup truncate fails to obtain journal handle.

On Wed 24-10-12 16:12:44, Ashish Sangwan wrote:
> During orphan cleanup while doing truncate, if we fail to obtain journal
> handle, the inode for which truncate was called would not be removed from
> both the on-disk and in-memory orphan lists as the call to ext3_orphan_del
> would not be executed.
> 
> This would have following consequences:
> a) As the inode is not removed from the on-disk list, truncate would be
> called again for the same inode. Each call would add the inode to the
> in-memory list. This operation would continue endlessly or until truncate
> is succeed.
> 
> b) If somehow, after some iterations, truncate is succeed, ext3_orphan_del
> will only remove the inode from in-memory list just 1 time. This will trigger
> j_assert during put super.
> 
> This patch handles both the problems. If truncate fails, first in-memory
> list is cleared and than the partition is mounted as read only.
> Failure to obtain journal handle during mount may suggest that journal
> device is corrupted.
  Could you really trigger any of the described bugs? Because
ext3_truncate() does take care to remove inode from orphan list even if
starting of a transaction fails (see out_notrans label in ext3_truncate()).
Also spitting additional messages should be unnecessary - the error which
caused truncate to fail should have logged the error message and also taken
appropriate action depending on filesystem configuration (i.e., remount fs
read-only, panic the system, or do nothing).

								Honza

> 
> Signed-off-by: Ashish Sangwan <ashish.sangwan2@...il.com>
> Signed-off-by: Namjae Jeon <linkinjeon@...il.com>
> ---
>  fs/ext3/super.c | 17 ++++++++++++++++-
>  1 file changed, 16 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/ext3/super.c b/fs/ext3/super.c
> index bd60e08..798cd6f 100644
> --- a/fs/ext3/super.c
> +++ b/fs/ext3/super.c
> @@ -1540,7 +1540,22 @@ static void ext3_orphan_cleanup (struct super_block * sb,
>  			jbd_debug(2, "truncating inode %lu to %Ld bytes\n",
>  				  inode->i_ino, inode->i_size);
>  			ext3_truncate(inode);
> -			nr_truncates++;
> +			if (list_empty(&EXT3_I(inode)->i_orphan)) {
> +				nr_truncates++;
> +			} else {
> +				/* Remove inode from in-memory orphan list */
> +				list_del_init(&EXT3_I(inode)->i_orphan);
> +				ext3_msg(sb, KERN_ERR, "Truncate failed for "
> +                                        "orphan  inode = %lu. Running e2fsck"
> +                                        " is recommended", inode->i_ino);
> +				if (!(s_flags & MS_RDONLY)) {
> +					ext3_msg(sb, KERN_INFO, "FS would be"
> +						" mounted as readonly");
> +					s_flags |= MS_RDONLY;
> +				}
> +				break;
> +			}
> +
>  		} else {
>  			printk(KERN_DEBUG
>  				"%s: deleting unreferenced inode %lu\n",
> -- 
> 1.7.11.4
> 
-- 
Jan Kara <jack@...e.cz>
SUSE Labs, CR
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html