1996-02-02 - Re: Unix swapfile security issues…

Header Data

From: Steven L Baur <steve@miranova.com>
To: cypherpunks@toad.com
Message Hash: f84a8d0875e2e62542a9d72c0f2c889550ce02f5636bb71fccbcaebd307f8648
Message ID: <m2hgxakbh1.fsf@miranova.com>
Reply To: <199602010730.XAA09785@infinity.c2.org>
UTC Datetime: 1996-02-02 15:07:14 UTC
Raw Date: Fri, 2 Feb 1996 23:07:14 +0800

Raw message

From: Steven L Baur <steve@miranova.com>
Date: Fri, 2 Feb 1996 23:07:14 +0800
To: cypherpunks@toad.com
Subject: Re: Unix swapfile security issues...
In-Reply-To: <199602010730.XAA09785@infinity.c2.org>
Message-ID: <m2hgxakbh1.fsf@miranova.com>
MIME-Version: 1.0
Content-Type: text/plain


>>>>> "Anonymous" == Anonymous  <mixmaster@alpha.c2.org> writes:

Anonymous> I'm working on a unix application where I want to store a
Anonymous> key in memory and don't want it to get written out to a
Anonymous> swap file.  If the key is in any of the application's
Anonymous> memory pages, it could be swapped out at any time, and
Anonymous> potentially left in the swap file when the computer is
Anonymous> turned off.

That's only a problem if physical security doesn't exist at the
console.  No operating system (or monitor) can overcome the lack of
that.

Anonymous> But, what if the program creates a pipe() and writes the
Anonymous> key into it, then reads the key out when necessary?  A pipe
                             ^^^^^ ^^^ ^^^ ^^^

In which case it's in memory and can be paged or swapped.

Anonymous> has a 4K buffer, but that buffer is in the kernel's memory,
Anonymous> not in the application's pages.  Could a kernel buffer get
Anonymous> written out to a swapfile?

Depending on how the kernel is written, bringing down the machine
could result in a dump of kernel memory being written to the swap
device anyway.

-- 
steve@miranova.com baur
Unsolicited commercial e-mail will be proofread for $250/hour.





Thread