[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: 2.6.1 CryptoAPI woes

I feel this is kind of over my head ... to boil it down: does it even
make sense to run reiserfs inside a loopback partition?

Thanks, Joh

On Wed, 21 Jan 2004 01:34:25 -0400
Peter Cordes <peter@llama.nslug.ns.ca> wrote:

> On Tue, Jan 20, 2004 at 11:58:41PM -0500, Hubert Chan wrote:
> > >>>>> "Johannes" == Johannes Graumann <graumann@caltech.edu> writes:
> > 
> > [...]
> > 
> > Johannes> And on another note: in
> > Johannes>
> > http://www.mirrors.wiretapped.net/security/cryptography/filesystems/loop-aes/loop-AES.README
> > Johannes> I read the following: "Don't use a journaling file system
> > on Johannes> top of file backed loop device, unless underlying file
> > system Johannes> is journaled and guarantees data=ordered or
> > data=journal." Johannes> Can anybody comment on whether I can use
> > reiserfs on top of my Johannes> loopback?
> > 
> > The comment has nothing to do with whether or not your encrypted
> > filesystem is a journaling filesystem with or without data=ordered. 
>  Actually, it does.
> > It
> > has to do with using a file-backed loop device (versus
> > partition-backed loop device), where the file is sitting on a
> > journaling filesystem.  If your loop device is a partition, or is
> > file-based, but sits on top of a non-journaled filesystem
>  Wait a second;  I think this one doesn't belong in the list of things
>  that
> will be correct.
> > or a journaled filesystem with data=ordered or
> > journaled, then you can use any filesystem without problems.  (Or,
> > at least, you won't (shouldn't) run into any problems other than
> > what you might run into if it were not on a loopback device.)
> > 
> > Basically, if you don't have data=ordered, or data=journaled, any
> > system crash could completely screw up your entire loopback,
> > rendering it completely unusable.  If you don't plan on having any
> > system crashes or hard reboots, I think you can still run a loopback
> > on top a non-data=ordered journaled filesystem fairly safely.
>  No, the point is that journaling file systems depend on stuff being
>  written
> to disk in the order they want, so if something goes wrong at _any_
> moment, they can pick up the pieces.  ext3 with data=writeback, for
> example, only bothers to strictly control the order of metadata.  A
> loopback to a file on such a filesystem will not preserve write
> ordering, so a journaling filesystem on top of it will be making false
> assumptions.  Filesystem metadata (which needs to be ordered) is just
> data on the loopback device. However, if the underlying filesystem
> preserves data ordering, it can satisfy the requirements of the
> journaling filesystem that's on top of it.
>  I'm not sure if you need data=journal on the underlying filesystem
>  for
> data=journal on the loopback filesystem to make sense, but I don't
> think so.
> -- 
> #define X(x,y) x##y
> Peter Cordes ;  e-mail: X(peter@cor , des.ca)
> "The gods confound the man who first found out how to distinguish the
> hours!
>  Confound him, too, who in this place set up a sundial, to cut and
>  hack my day so wretchedly into small pieces!" -- Plautus, 200 BC

Reply to: