Re: Intel Pentium Bug

[aleph1 () dfw net (Aleph One)]

On Sat, 8 Nov 1997, Daniele Orlandi wrote:

> If the CPU microcode is really updateable, it's realistic to think that:
>
> 1- Upload requires to be done only when running in ring-0, just like all the
>    other protected instructions. No user program will be able to use this
>    feature.

As you (and Eric) point out this is correct. But at the time we where
thinking Windows 95 viruses so running in ring-0 is not an issue. Of
curse under Unix you should have to subvert the root account and from
there install something like a loadable module (similarly under NT).
But it happens all to often.

> 2- New microcode will not be permanent, otherwise, if you, by mistake,
>    put garbage in your CPU, you'll not have to throw the CPU in the trash.

Good point. But do they have enough memory to store the original copy of
the microcode and the update? Unlikely, that memory is highly expensive.

As other have pointed out you may need to use a device to reprogram the
chip (makes sense). This would allow intel and OEMs to fix the parts on
their inventory and to trade parts for customers concerned. Of curse with
a problem of this magnitude _EVERYONE_ that runs a multiuser system will
want to upgrade. Thats a lot of people. And a software fix would make it
easy to distribute.

There must be some quite some activity tonight at Intel.

> Bye...
>
> --
>  Daniele
>
> --------------------------------------------------------------------
>  Daniele Orlandi - Utility Line Italia - http://www.xt.net
>  Via Mezzera 29/A - 20030 Seveso (MI) - Italy - FAX: ++39(362)540153
> --------------------------------------------------------------------

Aleph One / aleph1 () dfw net
http://underground.org/
KeyID 1024/948FD6B5
Fingerprint EE C9 E8 AA CB AF 09 61  8C 39 EA 47 A8 6A B8 01