Re: SVSM Attestation and vTPM specification additions - v0.60

[Tom Lendacky <thomas.lendacky@amd.com>]

On 1/12/23 09:24, James Bottomley wrote:
> On Thu, 2023-01-12 at 09:13 -0600, Tom Lendacky wrote:
>> On 1/12/23 07:57, James Bottomley wrote:
>>> On Wed, 2023-01-11 at 17:39 +0100, Christophe de Dinechin wrote:
>>> [...]
>>>> p29: Table 14: How can RAX be used as command ordinal and command
>>>> response size if it's already used for call identifier / result
>>>> value?
>>>
>>> I think there's a bug on the first table: in the prototype we use
>>> RAX [in] (not RCX [in]) for the pointer to the request/response
>>> buffer and RAX [out] as the SVSM return code.
>>
>> No, it's not a bug. You'll need to use RCX as the pointer to the
>> request/response buffer because RAX, per calling convention, contains
>> the protocol id and call id.
> 
> Oh, right, I should have looked at the code first before commenting we
> do indeed use RCX for the shared buffer.
> 
> Do you have a timetable for updating the linux-svsm github for the new
> protocol based calling convention?  We'd like to rebase the new vTPM
> code off it when it's available.

It would mean adding support for at least the attestation protocol and 
encryption support to perform that call. I don't have a timetable at the 
moment.

Thanks,
Tom

> 
> James
>