PGP "master keys"

[Travis H.]

On 29 Apr 2006 02:00:18 -0000, StealthMonger
<StealthMonger at nym.alias.net> wrote:
> Interesting epilog: theregister has apparently now edited out all
> mention of master keys.

They probably had their misunderstanding pointed out to them by
countless people by now.

But... did anyone else note the phrasing of the qualification Redmond
ostensibly used?

``BitLocker has landed Redmond in some hot water over its insistence
that there are no back doors for law enforcement.''

On first reading, one might assume they meant no back doors except for
the overt corporate ADK, but that is not in fact what they said.

Does anyone have any experience with disk or filesystem encryption,
especially with regard to unclean shutdowns and power failures? 
Normal file systems are designed to fail in ways that are easy to
clean up with fsck, but when you start to throw encryption into the
mix, it seems like you can easily end up with something unrecoverable.
 Even without encryption I've seen apparent bugs in ext2fs on SMP
machines that lead to sectors of nulls placed in files that were being
written around the time the system crashed.

Personally, I was playing with disk encryption on my system, shut down
the system and something was holding file descriptors open... the
system tried to kill everything three times, and then gave up and
rebooted.  As a consequence, I had my first unrecoverable data loss
since I started keeping track (probably 1992 or so), since I had not
backed up the data (the file system was too large for my backup
device).

Lesson learned!  Now I do a nightly rsync to a partition that is only
briefly mounted.  Not as good as backup tapes, but it'll do for now.

Are there any good solutions to the problem where a key isn't used
frequently enough to stay in human memory, yet needs to be present in
certain rare circumstances?  Even with PGP keys... I've forgotten some
of mine.  Print it out and put it in a safety deposit box?  I wonder
if the typical corporate escrow key is exercised enough to avoid
needing to write it down.

IMHO interaction with human factors and imperfect hardware/software
are understudied relative to their importance in actually having a
functional robust real-world system.  How complex can passwords be
before users start to write them down?  How many times does it take to
memorize a passphrase?  How frequently must one use it in order to
retain it?
--
"Curiousity killed the cat, but for a while I was a suspect" -- Steven Wright
Security Guru for Hire http://www.lightconsulting.com/~travis/ -><-
GPG fingerprint: 9D3F 395A DAC5 5CCC 9066  151D 0A6B 4098 0C55 1484

---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at metzdowd.com