linux-ext4 - Re: ext4 crypto: Do not select from EXT4_FS

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]

Message-ID: <20150503211118.GK10014@thunk.org>
Date:	Sun, 3 May 2015 17:11:18 -0400
From:	Theodore Ts'o <tytso@....edu>
To:	Anssi Hannula <anssi.hannula@....fi>
Cc:	Herbert Xu <herbert@...dor.apana.org.au>,
	Andreas Dilger <adilger.kernel@...ger.ca>,
	linux-ext4@...r.kernel.org, Michael Halcrow <mhalcrow@...gle.com>
Subject: Re: ext4 crypto: Do not select from EXT4_FS_ENCRYPTION

On Sun, May 03, 2015 at 09:29:02PM +0300, Anssi Hannula wrote:
> > I believe the situation which is causing concern is when someone wants
> > to build a kernel where EXT4_FS=y, but they want the cryptographic
> > algorithms to be modules.  In that case, since EXT4_FS_ENCRYPTION is
> > 'y', it forces the all of the crypto modules to be built into the
> > kernel, and so it forecloses that option from someone who is building
> > or packaging a kernel.
> 
> Ah, OK, so not "EXT4 itself as a module" like the commit message said :)
> 
> For the situation you described I don't see a better solution either.

Thanks for pointing out problem in the commit message.  I guess I
wasn't reading all that carefully, but started experimenting, and came
up with some case where, if they aren't lack _bugs_, do constrain
flexibility a little.  You are correct that various crypto modules can
still be built as modules even if ext4 is a module and
EXT4_FS_ENCRYPTION is 'y'.  The main issue that I was able to find is
that if ext4 is _not_ a module, then it also forces the crypto modules
to also be built in.

(Personally from a performance perspective, I'd always want to make
the common crypto modules always built in to avoid pressure on the TLB
cache, but I understand that distributions seem to like to build
_everything_ as modules (which is one of the reasons why I generally
don't use distro kernels myself.  :-)

In any case, I'll correct the commit message so that it describes the
problem which it addresses more clearly.

							- Ted
--
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