cvs commit: src/sbin/geom/class/eli geom_eli.c

Pawel Jakub Dawidek pjd at
Tue Jun 6 07:17:36 UTC 2006

On Mon, Jun 05, 2006 at 06:32:48PM -0700, Nate Lawson wrote:
+> Pawel Jakub Dawidek wrote:
+> >pjd         2006-06-05 21:40:54 UTC
+> >  FreeBSD src repository
+> >  Modified files:
+> >    sbin/geom/class/eli  geom_eli.c   Log:
+> >  Userland bits of geli(8) data authentication.
+> >  Now, encryption algorithm is given using '-e' option, not '-a'.
+> >  The '-a' option is now used to specify authentication algorithm.
+> >    Supported by:   Wheel Sp. z o.o. (
+> >    Revision  Changes    Path
+> >  1.11      +29 -15    src/sbin/geom/class/eli/geom_eli.c
+> Excellent!  One of my longstanding complaints has been that no block encryption software supported integrity, only privacy.

The problem is that it was not easy to make it reliable, ie. to be sure
that storing both data and HMAC is atomic operation, so user won't get
false postitives on system crash or power failure.
But I found a way to do it, so here it is:)
If you are interested how it is done, I tried to describe it at the
beginning of g_eli_integrity.c.
(I need to write a paper about GELI someday...)

+> As far as the flag change goes, won't this make it difficult to MFC this new feature later?

One will get an error if it tries to specify encryption algorithm with
'-a' flag, so nothing bad will happen.
I handle metadata backward compatibility, so we are safe here.

If needed I can eventually accept encryption algorithm specified with
'-a' flag and print a warning.

The bigger problem is that to MFC geli(8) authentication, I need to MFC
my recent opencrypto work, which I'd like to be well tested first.

Pawel Jakub Dawidek             
pjd at                 
FreeBSD committer                         Am I Evil? Yes, I Am!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url :

More information about the cvs-src mailing list