Cryptographic fs support in the kernel with loadable modules ala loopback

Cryptographic fs support in the kernel with loadable modules ala loopback

Post by Klaus Fluori » Sun, 07 Feb 1999 04:00:00



For those of you who started your linux experience on a humble
slackware box, heres a poser.

        Several years ago, circa the release of the Linux 2.0.11
kernel, A set of patches for Crypto filesystem support _within_ the
kernel were uploaded to a BBS I was on at the time. As I recall (I'm
afraid I've lost the original message text), the patches allowed
IDEA/3DES encrypted ext2 filesystems to be mounted on a standalone
machine without the need to involve any external server daemons such
as NFS (ala CFS) or secondary authentication daemons. Basically, you
had a file system secured from outside server processes with one of
the strongest crytpto algorithms in the world.

        The message was posted (IIRC) to the cypherpunks mailing list,
and was discussed in earnest for some time - the initial
implementation was broken, and had to be further modified to allow
secure operation in IDEA mode.

        It operated like so - You made a raw file of roughly 50-100Mbs
or whatever size you wanted, formatted it, encrypted it with the
commands that came with the patches, and mounted it as a loopback file
system. In essence, it operated much like SecureDrive under MSDOS,
and was an excellent solution for single user workstation security.
The raw files were simply viewed as virtual devices and mounted as
such.

        Unfortunately, the patches proved to be less than portable
with subsequent versions of the kernel source, and so I had to leave
them by the wayside. To this day, I've yet to encounter another set of
tools that offered a similar solution. TCFS is unfortunately a little
above spec for what I need (It is after all, a standalone
workstation), and I wasn't happy with the way CFS could take a dive
very suddenly if the NFS daemon cacked it unexpectedly.

        Iain Goldbeg was involved with the development of the patches
As I recall, but I'm afraid I haven't been able to find any trace of
the original thread anywhere, or any hint of the patches having
existed. I'm going to search the BugTraq archive shortly after this
post, so that may turn up some gold, but in the meantime (and the
eventuality of an unsuccesful hunt) can anyone shed some light on this
or offer some pointers?

Any help, pointers, or info would be more than appreciated,

        Yours

Klaus Fluoride                       http://www.venona.freeserve.co.uk

 
 
 

Cryptographic fs support in the kernel with loadable modules ala loopback

Post by Andrew McDona » Mon, 08 Feb 1999 04:00:00



>It operated like so - You made a raw file of roughly 50-100Mbs
>or whatever size you wanted, formatted it, encrypted it with the
>commands that came with the patches, and mounted it as a loopback file
>system. In essence, it operated much like SecureDrive under MSDOS,
>and was an excellent solution for single user workstation security.
>The raw files were simply viewed as virtual devices and mounted as
>such.

Sounds like you are after the international kernel patch.
This adds strong encryption to the loopback block device.
You can find the patches at http://www.kerneli.org/

Andrew

--
Andrew McDonald

http://ban.joh.cam.ac.uk/~adm36/

 
 
 

Cryptographic fs support in the kernel with loadable modules ala loopback

Post by Karsten Patzwal » Mon, 08 Feb 1999 04:00:00



>    Iain Goldbeg was involved with the development of the patches
>As I recall, but I'm afraid I haven't been able to find any trace of
>the original thread anywhere, or any hint of the patches having
>existed. I'm going to search the BugTraq archive shortly after this
>post, so that may turn up some gold, but in the meantime (and the
>eventuality of an unsuccesful hunt) can anyone shed some light on this
>or offer some pointers?

IIRC there was a patch like this on http://www.counterpane.com. I'm not
quite sure, please correct me if I'm wrong... This one should support XOR,
IDEA, DES and Blowfish-support (I prefer the last). I wasn't able to install
the patch on my system anyway, but this was maybe my fault.

--

 
 
 

Cryptographic fs support in the kernel with loadable modules ala loopback

Post by Konrad Ant » Tue, 09 Feb 1999 04:00:00


Karsten Patzwaldt schrieb:


>>        Iain Goldbeg was involved with the development of the patches
>>As I recall, but I'm afraid I haven't been able to find any trace of
>>the original thread anywhere, or any hint of the patches having
>>existed. I'm going to search the BugTraq archive shortly after this
>>post, so that may turn up some gold, but in the meantime (and the
>>eventuality of an unsuccesful hunt) can anyone shed some light on this
>>or offer some pointers?

>IIRC there was a patch like this on http://www.counterpane.com. I'm not
>quite sure, please correct me if I'm wrong... This one should support XOR,
>IDEA, DES and Blowfish-support (I prefer the last). I wasn't able to install
>the patch on my system anyway, but this was maybe my fault.

I've found *some* patches for IDEA/DES-encrypted loopback devices on

  ftp://ftp.is.co.za/pub/linux/kernel/crypto/
(directory might be wrong, I'm typing out of my memory)

and they apply neatly to kernel 2.0.34, which can be patched to 2.0.35 using
an ordinary linux kernel patch afterwards.

The patches seem to work here.

HTH
Konrad.

PS:
I don't know if you are legally allowed to download any 'dangerous'
software from South Africa. Please care about that yourself if you want to;)

--
 n_n_n_n_n     LOOK, DOGBERT, I'VE INVENTED A MACHINE THAT POSTS USELESS
 |       |   / MESSAGES INTO NEWSGROUPS AND STARTS FLAMEWARS.
 |---()()|  /  

 
 
 

1. What kind of support for dynamically loadable kernel modules

Can someone please fill me in on the current status of dynamically
loadable kernel modules within Linux? The FAQ does not address this
issue at all.

In particular, I don't know if this can be done or not, but I am
interested in writing some kind of dynamically loadable kernel
module that would either trap or preempt certain system calls
and do its own thing. Then, at some point of time, the "module"
would terminate itself and return back to the normal world, and
the system would once again be a normal Linux system.

Is this a reasonable project given the current state of linux?

Any tips or pointers you can provide me would be greatly appreciated.

Thanks very much,
Hayim Hendeles


2. Long filename ( > 14 chars )

3. can't probe ppp module in kernel 2.2.2.5-15 using loadable module

4. RH5.1 module setup

5. Anything like COM in Java?

6. uuups: [mount: fs type devpts not supported by kernel] + [MODULES]

7. fddev fixes due to massive change.

8. fs fs=iso9660 not supported by kernel ->LM 7.2

9. New support for loadable modules in 1.1.85

10. [RFC] __init and friends support for loadable modules

11. Loadable Module support

12. Is loopback-based FS slower than normal FS?