Re: [OT] Trusted hardware

[Udo A. Steinberg]

On Mon, 10 Oct 2005 15:30:50 -0400 Jonathan S. Shapiro (JSS) wrote:

JSS> I mean "mechanisms by which user mode code can enter supervisor mode."

Sure, the instruction is called SYSENTER and is well-documented ;)

JSS> No, I cannot give examples. Intel is very careful about the NDAs on this
JSS> information, and after watching the difficulties that Robert Collins had
JSS> to deal with, I have always refused to sign these NDAs. And so, of
JSS> course, if I now disclosed any of this I would be subject to prosecution
JSS> and prison for information theft.

Those days are long past. Both Intel and AMD are regularly publishing the
known errata of their chips.

http://download.intel.com/design/mobile/SPECUPDT/30220913.pdf
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/25759.pdf

There are enough errata to make kernel developers cringe, but privilege
escalation of user-mode code is not among any of them.

-Udo.

pgpx55fs2KTrx.pgp
Description: PGP signature