Re: Tempfile vulnerabilities

[deraadt () CVS OPENBSD ORG (Theo de Raadt)]

The terrible /tmp race handling aside...

I suppose then that anyone who attacks a machine which relies on
/dev/random -- a world readable device -- should do the following:

        cat /dev/random > /dev/null &

Crypto software which uses those devices should be doing some kind of
checking to make sure that they are getting at least good entropy.  I
suppose I could even argue that the random devices should make it easy
for customer software to determine that entropy is low.

> On Mon, 31 Jan 2000, Grant Taylor wrote:
>
> >        open RAN, "/dev/random" || die;
> >        read(RAN,$foo,16);
> >        close RAN;
> >        $file = '/tmp/autobuse' . unpack('H16',$foo);
>
> Please, never use /dev/random or /dev/urandom for such purposes.
>
> Aside the fact, that it does not help much in what you want to achieve
> it is a desaster to system performance because it empties the system's
> entropy pool and wastes precious entropy for unneeded things.
>
> Crypto software _really_ needs these random numbers.
>
>
> --
> Werner Koch at guug.de           www.gnupg.org           keyid 621CC013
>
>      Boycott Amazon!  -  http://www.gnu.org/philosophy/amazon.html