1996-11-15 - Mounting Crypted directories on Multiuser Machines.

Header Data

From: snow <snow@smoke.suba.com>
To: cypherpunks@toad.com
Message Hash: 3e927e45658261223bdef1ceb7d3489ba11a470e2a87884903853860f8f321d1
Message ID: <199611150654.AAA01811@smoke.suba.com>
Reply To: N/A
UTC Datetime: 1996-11-15 06:37:51 UTC
Raw Date: Thu, 14 Nov 1996 22:37:51 -0800 (PST)

Raw message

From: snow <snow@smoke.suba.com>
Date: Thu, 14 Nov 1996 22:37:51 -0800 (PST)
To: cypherpunks@toad.com
Subject: Mounting Crypted directories on Multiuser Machines.
Message-ID: <199611150654.AAA01811@smoke.suba.com>
MIME-Version: 1.0
Content-Type: text/plain


 
 I was wondering about something: 
 
       Say you have a unix (or other multi-user OS) box, how hard would 
 it be for someone who was good at programming to modify Blaze's CFS to 
 allow the following:
 
       Every user has a directory:
 
      /home/usr1/
           /usr2/
           /usr3/
 
       and inside each directory they have:
 
       /home/usr1/html/
                 /files/
                 /.login    
                 /.usr1crypt
 
      such that /.usr1crypt gets mounted at _login_ time as a crypted file
 system under /files.
 
      The way I envision this is that one would log in (either from the console,
 or via ssh ideally) and be presented with the option of mounting said directory
 and asked for a passphrase, then the directory gets mounted. 
 
      I took a look at Blaze's CFS, but he mentions that it is really only for 
 a single user system, and well <looks sheepish I can't get it to compile 
 on my machine, so I can't really play with it on my end. 
 
      I would think that this would be fairly difficult, otherwise it would 
 have already been done right? 
 
      Or am I missing something more basic? 
 
      It would seem that running something like this would do 2 things.
   
      1) It would be much more difficult to prove that a service provider 
         knew what files a user was keeping lying around because unless
         the user was logged in, not even the Sysadmin could "peek" at the
         files.
      2) Provide the user with greater privacy. Users could keep PGP keys 
         on the system without much risk, and as long as access was either
         thru the console, or thru something like ssh, you should be rather
         safe. 
  
       Is anyone working toward something like this? I kinda got the idea that
 CFS was more designed and intended for single-user-at-a-time systems, but the
 application I had in mind was more of a (old) C2-type organization. 
 
 
 Petro, Christopher C.
 petro@suba.com <prefered for any non-list stuff
 snow@smoke.suba.com





Thread