linux-ext4 - Re: [PATCH] ext4: avoid modifying checksum fields directly during checksum verification

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>] [day] [month] [year] [list]

Message-id: <1254354083.481464.1466035726062.JavaMail.weblogic@ep1ml102b>
Date:	Thu, 16 Jun 2016 00:08:46 +0000
From:	Daeho Jeong <daeho.jeong@...sung.com>
To:	Darrick J. Wong <darrick.wong@...cle.com>
Cc:	tytso@....edu <tytso@....edu>,
	linux-ext4@...r.kernel.org 
	<linux-ext4@...r.kernel.org>,
	yeongjin gil <youngjin.gil@...sung.com>
Subject: Re: [PATCH] ext4: avoid modifying checksum fields directly during checksum verification

Hi Darrick,

> Is something verifying the block at the same time the journal is
> writing the same block out via replay?  Or even just a regular commit?
> If that's the case, then yes, you're right, we can't touch a single
> bit on a metadata block without a transaction protecting it.  Oops.

As far as I know, if you want to modify a metadata block, you have to
get write access for a metadata block in advance in order to notify
your intention to jbd2 journaling module so that jbd2 can control all
the modifications to the metadata block and it can make a copied
version of the metadata block with copying out the original metadata
block if necessary.

However, if we touch a metadata block without jbd2's approval, jbd2 doesn't
know about what we are doing and it cannot do anything for us. So, we can
modify the content of the metadata block even being written back.