lkml.org 
[lkml]   [2003]   [Aug]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH] loop: fixing cryptoloop troubles.
From
Date
Am Mo, 2003-08-11 um 15.11 schrieb Fruhwirth Clemens:

> > The main problem with CBC is that you can't really do it. It only works
> > when you have a constant stream of data because you always need the
> > result from the previous encryption which you don't have when doing
> > something in the middle of the block device.
>
> That's partially correct. As most block cipher operate on blocks of 16 bytes
> size it perfectly makes sence to use CBC on a 512 byte block.

Ah, you're right, I didn't look at it enough to see that the loop in
cipher.c only handles bsize bytes at once.

> > The cryptoloop code is doing things correctly. In ecb mode, every bio
> > could get converted at once, or every bvec.
>
> ECB mode is broken in 2.6.0-test[12].

I was speaking theoretically ;)

> It's a quite conservative patch. ECB processing can be optimized.

Yes, you're not restricted to split the request into 512 byte chunks.
But I don't think it's too much of a performance loss. Or should it
better be handled differently? Because I'm not doing it either in my
code.

> Definitly. loop.c is anyway ugly :). It would be nice to rip out the
> block-backend stuff of loop.c and recommend to use device mapper instead.
> loop.c will benefit from that for sure since it doesn't have to handle two
> different case in such a schizophrenic manner.

That's right. I could also write a losetup-like user space utility that
sets up a linear mapped device with the full size of the block device
and optionally uses encryption.

What do you think of this passphrase thing? I could optionally link a
against openssl or such a library to offer password hashing.

> I'll give it a try, promised :)

That'd be nice. :)

--
Christophe Saout <christophe@saout.de>
Please avoid sending me Word or PowerPoint attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:47    [W:0.058 / U:0.524 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site