Re: [PATCH v7 09/31] OvmfPkg/SecMain: register GHCB gpa for the SEV-SNP guest

["Brijesh Singh" <brijesh.singh@amd.com>]

On 9/15/21 12:08 PM, Erdem Aktas wrote:
> On Mon, Sep 13, 2021 at 9:20 PM Brijesh Singh <brijesh.singh@amd.com> wrote:
>> +*/
>> +STATIC
>> +VOID
>> +SevSnpGhcbRegister (
>> +  UINTN   Address
>> +  )
>> +{
>> +  MSR_SEV_ES_GHCB_REGISTER  Msr;
>> +  MSR_SEV_ES_GHCB_REGISTER  CurrentMsr;
>> +  EFI_PHYSICAL_ADDRESS      GuestFrameNumber;
>> +
>> +  GuestFrameNumber = Address >> EFI_PAGE_SHIFT;
>> +
>> +  //
>> +  // Save the current MSR Value
>> +  //
>> +  CurrentMsr.GhcbPhysicalAddress = AsmReadMsr64 (MSR_SEV_ES_GHCB);
> 
> We are backing the current MSR value but when was it initialized
> before ? Also is not this function supposed to set the Address as the
> GHCB address? If it is, do we care about the old value?
> 

Good point, there is no reason to read and restore the old GHCB, I will 
remove it in next version. The function does not set this as a GHCB 
address, it send request to hypervisor saying that it would like to use 
this address. If hypervisor is not okay with the address then it may 
recommend something else. We don't support working with the hypervisor 
preferred address. Setting the GHCB address code is common between Snp 
and Es but checking with hypervisor whether its okay to use is new in 
the GHCBv2 and is SNP specific.


> 
>> +  // Restore the MSR
>> +  //
>> +  AsmWriteMsr64 (MSR_SEV_ES_GHCB, CurrentMsr.GhcbPhysicalAddress);
> 
> Why are we restoring the old value? I may have misunderstood but I
> thought this function will set Address as the new GHCB address?
> 
> Thanks
> -Erdem
>