1995-01-05 - Re: Are 2048-bit pgp keys really secure ?

Header Data

From: danisch@ira.uka.de (Hadmut Danisch)
To: cypherpunks@toad.com
Message Hash: b23f64ba0e2093f778f801f07c22f6f460bba3d194a85f73184acf487481611f
Message ID: <9501052056.AA24402@elysion.iaks.ira.uka.de>
Reply To: N/A
UTC Datetime: 1995-01-05 20:56:32 UTC
Raw Date: Thu, 5 Jan 95 12:56:32 PST

Raw message

From: danisch@ira.uka.de (Hadmut Danisch)
Date: Thu, 5 Jan 95 12:56:32 PST
To: cypherpunks@toad.com
Subject: Re: Are 2048-bit pgp keys really secure ?
Message-ID: <9501052056.AA24402@elysion.iaks.ira.uka.de>
MIME-Version: 1.0
Content-Type: text/plain


> A somewhat disturbing trend has appeared in the low-end cost-sensitive PC
> SIMM market.  Some supposedly 9-bit SIMMs are actually 8-bit SIMMs plus
> a parity generator.  This means that the parity checking is essentially
> subverted, because the parity bit is generated from the stored contents
> of memory at read time, rather than the stored contents when it was
> written to.  As such, NO bit errors are detected.

So why not do a cheap trick:

After the small primes check calculate a CRC checksum over the number.
Then do the primality check. If it is a prime, store it together with
the CRC. The CRC can be checked for every use of the number.

(PGP encrypts the secret key and therefore it generates a CRC for the
encrypted packet. But this CRC is generate after the primality check.)


We already had some SIMM modules with bit errors. They were detected by
a parity check. If cheap pc simms don't have a real parity bit, the probability
of having such a bug isn't as low as 10^-40. If cheap parity-less simms
are available they get sold (cheap and expensive). 


Hadmut

BTW: Some weeks ago they found motherbords with falsified cache rams:
They had just the normal DIL-ICs and the normal pins, but the plastic of
the ICs didn't contain a chip. The BIOS was modified to give out a message
about a good cache ram check at boot time.





Thread