Here is a suggestion for how to appraoch the TCPA spec based on the parts I have found to be relatively good explanations. The spec is available from http://www.trustedcomputing.org/docs/main%20v1_1b.pdf. First read the first few pages up to page 7. This provides an overview and a block diagram, although at this point not all the terms will be familiar. One hint: the "root of trust for measurement" is the set of hardware which has to be working for the boot measurement process to work: the CPU, the TPM, the part of the BIOS that deals with measurements, the motherboard, the secure connections of the chips to the motherboard. If all this stuff is OK then the measurements will be accurate. (Measurements basically are hashes of code and of machine configuration status.) The "root of trust for reporting" is the endorsement key, or more fundamentally the cert on the endorsement key. The cert is issued by the manufacturer, AKA the "TPM Entity" or TPME. That's what makes other people believe your attestations. And the "root of trust for storage" is the storage root key, described in the section on protected storage. There are a few more pages of introduction which aren't too clear, then a long section of data structures which should be skipped until you need to reference them. This brings you to page 97, authorization and ownership. I haven't really studied this part. Probably just read this one page to get an idea of what is involved. I still need to learn more about this. I'd skip on to pages 136-137, on the measurement process and the PCRs which hold the results of the measurement. Then I'd read pages 145-150 on protected storage. This part is pretty well written. It is a reasonably self contained part of the TPM functionality. You just need to know a little bit about the PCRs from the earlier section to understand how data is locked to the specific program which is running. Then I'd read page 261 on the endorsement key, and then 267-269 on how it is used to create a pseudonymous identity. This is the part about communicating with the Privacy CA. BTW an expert told me he has concerns about possible security loopholes in this protocol, but he is communicating with TCPA about them. I think if you just read these selections, about 15 pages, you will have a much better idea of how the spec works. Then you can read some of the specific API descriptions to see more details about the functionality. There is also a glossary at the end which can be helpful for some (but not all) of the terminology. There is another spec, http://www.trustedcomputing.org/docs/TCPA_PCSpecificSpecification_v100.pdf that describes the specific register and trap binding for implementing the TCPA API on Intel PCs. It is much shorter but it is pretty incomprehensible until you have at least read the basics of the main spec. --------------------------------------------------------------------- The Cryptography Mailing List Unsubscribe by sending "unsubscribe cryptography" to majordomo@wasabisystems.com
On Mon, 5 Aug 2002, AARG! Anonymous wrote:
Here is a suggestion for how to appraoch the TCPA spec based on the parts I have found to be relatively good explanations. The spec is available
I'd wish you wouldn't lay on the exegesis of the Sacred Document quite so heavily. Reason: your prophet stinks. Please spare your's and our neurons for more worthwhile endeavours. If the stinking thing ever comes through you can always burn midnight oil in efforts on how to misuse is in the best possible way. Meanwhile, the amount of floorspace so far allotted is way overblown.
participants (2)
-
AARG!Anonymous
-
Eugen Leitl