Intel RNG still available?

Andy Isaacson adi at hexapodia.org
Wed Apr 9 16:58:09 EDT 2003


On Wed, Apr 09, 2003 at 09:25:26AM -0700, Carl Ellison wrote:
> I tried to buy a new machine with the chip, one or two years ago.  I
> asked my contacts within Intel exactly what I had to ask for -- the
> part numbers you need to have are: E82802Ax or N82802Ax.  I also
> found out that nearly the entire stock of such parts was shipped to
> IBM for their platforms.
> 
> So I called IBM and told the salesman on the phone I wanted to buy a
> machine with one of those parts.  They were doing web/phone sales
> only, so I couldn't go to the store and open the case and look at the
> motherboard.  The salesman told me that he had no way to know what
> parts were on the machine shipped to me.  That part was not
> associated with any feature that showed up on his price sheet.

It sounds to me like going to a car dealer and demanding an automobile
made with steel mined in Minnesota.  Perhaps they exist, but the
salesperson has absolutely no reason to have such detailed information
about the product they sell.

Since OpenBSD has kernel support for the Intel 8xx RNG, searching the
web for dmesg output is informative.  My laptop says

pchb0 at pci0 dev 0 function 0 "Intel 82815 Hub" rev 0x11: rng active, 7Kb/sec

Googling for 'Intel hub "rng active"' found examples of i810, i815,
i820, i840, and i850 hardware that activate the RNG.

It certainly appears that "the entire stock" was not shipped to IBM, as
dozens of other vendors have systems incorporating these chipsets.

The OpenBSD source code claims that the 810, 810e, 815, 820, 840, 850,
and 860 all provide the RNG.  The conspicuous missing entry on the list
is the 845, which appears on all the Intel-chipset Pentium 4
motherboards currently available from my local whitebox shop.

OpenBSD also has code to support the AMD768 RNG, but google does not
show any instances of this being used in the wild.

-andy

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



More information about the cryptography mailing list