Re: [Rats] use case document updates on Roots of Trust

+1

A Root of Trust not data. The GlobalPlatform definition[1] is explicit that a Root of Trust has a computing engine and executable code that exist within a security boundary. Keys are optional as there are RoT use-cases that do not necessarily require keys (keys and/or other RoT data also exist inside the security boundary if they are present).

I’m not sure that the TCG Glossary[2] calls this requirement out quite so explicitly as GlobalPlatform, but the definition of RoT in the glossary is clear that an RoT performs functions - something that implies a computing engine.

A Root of Trust can perform other tasks than verification (and need not be capable of verification). This applies to all of the definitions I know (NIST, GlobalPlatform, TCG)

Best regards
Jeremy

[1] GlobalPlatform Root of Trust Definitions and Requirements, v1.1<https://globalplatform.org/wp-content/uploads/2018/07/GP_RoT_Definitions_and_Requirements_v1.1_PublicRelease-2018-06-28.pdf> (NB - unlike some GlobalPlatform specifications, this is not behind a click-through)
[2] TCG Glossary V1.1, Rev 1.0<https://trustedcomputinggroup.org/wp-content/uploads/TCG-Glossary-V1.1-Rev-1.0.pdf>

---
Jeremy O’Donoghue                            email: jodonogh@qti.qualcomm.com<mailto:jodonogh@qti.qualcomm.com>
Director, Engineering                        tel:   +44 1252 363189
NFC & Secure Software and Systems

On 15 Sep 2019, at 21:50, Laurence Lundblade <lgl@island-resort.com<mailto:lgl@island-resort.com>> wrote:

-------------------------------------------------------------------------
CAUTION: This email originated from outside of the organization.
-------------------------------------------------------------------------

On Sep 12, 2019, at 3:31 PM, Salz, Rich <rsalz@akamai.com<mailto:rsalz@akamai.com>> wrote:

I do not see a meaningful difference between "trust anchor" and "trust root" and "root(s) of trust."  All of them:
- Are pieces of data (certificate or key is not meaningful)
- Used to verify something such as a certificate or signature
- Are trusted by the application, based on actions that are "out of band" of the application itself

This is not how I understand a root of trust, nor how I think it is generally used in the TCG or TEE worlds. I think a root of trust involves a CPU, memory and SW that actively does something like boot and measure a device. There is usually a boundary around it so that other software on the device, like the high-level OS, can’t corrupt it. When it is doing reporting as in RATS it will have a private key that can do some signing. When it is doing trusted/secure boot, it will also have a public key to verify the software it loads.

LL

_______________________________________________
RATS mailing list
RATS@ietf.org<mailto:RATS@ietf.org>
https://www.ietf.org/mailman/listinfo/rats