Date: Fri, 8 Nov 2013 17:33:09 -0800 From: John-Mark Gurney <jmg@funkthat.com> To: Bruno =?iso-8859-1?Q?Lauz=E9?= <brunolauze@msn.com> Cc: "freebsd-geom@freebsd.org" <freebsd-geom@freebsd.org> Subject: Re: GELI Passphrase Providers Message-ID: <20131109013309.GK2279@funkthat.com> In-Reply-To: <BLU179-W5B844903E3EE0ABA8CC92C6F20@phx.gbl> References: <BLU179-W5B844903E3EE0ABA8CC92C6F20@phx.gbl>
next in thread | previous in thread | raw e-mail | index | archive | help
Bruno Lauz wrote this message on Fri, Nov 08, 2013 at 18:03 -0500: > Right now, there's only "cngets" used to provide passphrase for GELI disk encryption. > In the future, considering embedded solutiona, and cloud data centers, co-location, etc.., > would different geli passphrase providers be planned? > > > One thing that I dream of (for embedded projects): > > > While prompting the passphrase on the console, have some settings in loader.conf to > provide an iface, ip, netmask gateway to mount and implement a Single Packet Authorization mechanism with IPSec. We already have some of this via DHCP/BOOTP kernel for net booting and locating root FS, so it might be easier than having to create all of the infrastructure yourself... It is an insteresting idea... Though if http://www.linuxjournal.com/article/9565?page=0,2 properly describes their crypto, i'm not confortable with it... They should have used an authenticated encryption mode like AES-GCM, AES-EAX or Encrypt and then append/prepend an HMAC, or one of the others... One of the issues w/ decrypt then verify is that you now can cause the destination to decrypt arbitrary data... If you have a side channel (SPA/DPA and related attacks) on the destination (maybe it's an embedded system), you could extrac the key... > The impossibility to be on-site to enter passphrase prevent disk encryption for multiple scenarios, and in my humble opinion, those are the same scenarios where encryption is mandatory like embedded Device in the wild, co-location, Off-site servers... even bhyve... > > Of course, I know IPMI or KVM solutions are possible, just wandering if we oversee any solutions without those required. > > Any opinions? This is interesting as I'm trying to figure out how to deal w/ systems where you have many encrypted disks (say an array using ZFS) and how to get all them decrypted w/o having to enter the passphrase n times... Right now I use a separate zfs key store that has a bunch of key files which I then use w/o passphrased on the array... It works, but isn't the best solution... There has been talk about teaching geli to attach multiple disks using the same passphrase, but I haven't evaulated how well this works, and if it would work well for geli devices that ask for passphrases on boot.. P.S. Apparently a lot more people are using geli that I suspected. I'm happy that this is happening, but we have a lot of work to make it more usable. Thanks! -- John-Mark Gurney Voice: +1 415 225 5579 "All that I will do, has been done, All that I have, has not."
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20131109013309.GK2279>