TrenchBoot / documentation

Creative Commons Attribution 4.0 International
39 stars 8 forks source link

Specification updates and fixes #23

Open krystian-hebel opened 1 year ago

krystian-hebel commented 1 year ago

This issue is made to gather all small fixes, improvements and other TODOs in one place before adding them to next specification revision. It will also be a place for conversation about proposed changes. After enough changes are listed (or a change significant enough to bump specification revision) we can push them all in one PR to keep version in source file synchronized with produced PDFs.

I (or anyone with permissions to edit the issue message) will periodically update the list below to have all the changes in one place. Feel free to put your suggestions in the comment.


krystian-hebel commented 9 months ago

About a half of currently listed issues was addressed in https://github.com/TrenchBoot/documentation/pull/25. The rest is planned to land in the specification together with changes specific to AMD processors.

krystian-hebel commented 8 months ago

@SergiiDmytruk I think there is no off-by-one error in "TPM Extend Operation". E_0 describes extend operation for Obj_0 and E_n - for Obj_n, not Obj_(n-1). Can we remove that from the list above?

I'm also not sure about using i for index in "Measuring the Policy". Is this just to differentiate those from names used in "TPM Extend Operation"?

SergiiDmytruk commented 8 months ago

I think there is no off-by-one error in "TPM Extend Operation". E_0 describes extend operation for Obj_0 and E_n - for Obj_n, not Obj_(n-1). Can we remove that from the list above?

I meant that the way it's written there are (n + 1) objects in the range [0;n] (both ends included), instead of n of them ([0;n) or [1; n]). Same in "Measuring the policy".

I'm also not sure about using i for index in "Measuring the Policy". Is this just to differentiate those from names used in "TPM Extend Operation"?

If there is only Entry_n = ... and Entry_0, ..., Entry_n, it looks like definitions of Entry_0 through Entry_{n-1} are missing. i usually suggests iteration, n boundary.

krystian-hebel commented 8 months ago

If there is only Entry_n = ... and Entry_0, ..., Entry_n, it looks like definitions of Entry_0 through Entry_{n-1} are missing.

But for the policy it isn't defined through Entry_{n-1}

SergiiDmytruk commented 8 months ago

But for the policy it isn't defined through Entry_{n-1}

How come? Isn't M_policy a result of extending 0 and Entry_0, then extending result with Entry_1, etc. up to the last entry?

krystian-hebel commented 8 months ago

Isn't M_policy a result of extending 0 and Entry_0, then extending result with Entry_1, etc. up to the last entry?

Hmm, I understood it as concatenation of all entries (subset of their fields), but there is a different symbol for concatenation, so maybe you're right. @dpsmith @rossphilipson what are your opinions on this?

SergiiDmytruk commented 8 months ago

Hmm, I understood it as concatenation of all entries (subset of their fields), but there is a different symbol for concatenation, so maybe you're right.

The reason for my interpretation is definition of E() in 5.1 which is recursive when applied to an array of objects.