Re: [edk2-devel] [PATCH RFC v2 11/28] OvmfPkg: Reserve Secrets page in MEMFD

["Laszlo Ersek" <lersek@redhat.com>]

Hi Dov,

On 05/06/21 12:57, Dov Murik wrote:
> 
> 
> On 05/05/2021 22:33, Laszlo Ersek wrote:
>> On 05/05/21 15:11, Brijesh Singh wrote:
>>>
>>> On 5/5/21 1:42 AM, Dov Murik wrote:
>>>> [+cc: Tobin]
>>>>
>>>> Hi Brijesh,
>>>>
>>>> On 30/04/2021 14:51, Brijesh Singh wrote:
>>>>> BZ: https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fbugzilla.tianocore.org%2Fshow_bug.cgi%3Fid%3D3275&data=04%7C01%7Cbrijesh.singh%40amd.com%7C93168c94eb6d44ed08e608d90f910426%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637557937779907471%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=nLpmk3G%2BmXcZrzXxCmO3M9EDPiLRnP1IUmPqRQNbBuU%3D&reserved=0
>>>>>
>>>>> When AMD SEV is enabled in the guest VM, a hypervisor need to insert a
>>>>> secrets page.
>>>>>
>>>>> When SEV-SNP is enabled, the secrets page contains the VM platform
>>>>> communication keys. The guest BIOS and OS can use this key to communicate
>>>>> with the SEV firmware to get attesation report. See the SEV-SNP firmware
>>>>> spec for more details for the content of the secrets page.
>>>>>
>>>>> When SEV and SEV-ES is enabled, the secrets page contains the information
>>>>> provided by the guest owner after the attestation. See the SEV
>>>>> LAUNCH_SECRET command for more details.
>>>>>
>>>>> Cc: James Bottomley <jejb@linux.ibm.com>
>>>>> Cc: Min Xu <min.m.xu@intel.com>
>>>>> Cc: Jiewen Yao <jiewen.yao@intel.com>
>>>>> Cc: Tom Lendacky <thomas.lendacky@amd.com>
>>>>> Cc: Jordan Justen <jordan.l.justen@intel.com>
>>>>> Cc: Ard Biesheuvel <ardb+tianocore@kernel.org>
>>>>> Cc: Laszlo Ersek <lersek@redhat.com>
>>>>> Cc: Erdem Aktas <erdemaktas@google.com>
>>>>> Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
>>>>> ---
>>>>>  OvmfPkg/AmdSev/SecretPei/SecretPei.c   | 16 +++++++++++++++-
>>>>>  OvmfPkg/AmdSev/SecretPei/SecretPei.inf |  1 +
>>>>>  OvmfPkg/OvmfPkgX64.dsc                 |  2 ++
>>>>>  OvmfPkg/OvmfPkgX64.fdf                 |  5 +++++
>>>>>  4 files changed, 23 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/OvmfPkg/AmdSev/SecretPei/SecretPei.c b/OvmfPkg/AmdSev/SecretPei/SecretPei.c
>>>>> index ad491515dd..92836c562c 100644
>>>>> --- a/OvmfPkg/AmdSev/SecretPei/SecretPei.c
>>>>> +++ b/OvmfPkg/AmdSev/SecretPei/SecretPei.c
>>>>> @@ -7,6 +7,7 @@
>>>>>  #include <PiPei.h>
>>>>>  #include <Library/HobLib.h>
>>>>>  #include <Library/PcdLib.h>
>>>>> +#include <Library/MemEncryptSevLib.h>
>>>>>
>>>>>  EFI_STATUS
>>>>>  EFIAPI
>>>>> @@ -15,10 +16,23 @@ InitializeSecretPei (
>>>>>    IN CONST EFI_PEI_SERVICES     **PeiServices
>>>>>    )
>>>>>  {
>>>>> +  UINTN   Type;
>>>>> +
>>>>> +  //
>>>>> +  // The secret page should be mapped encrypted by the guest OS and must not
>>>>> +  // be treated as a system RAM. Mark it as ACPI NVS so that guest OS maps it
>>>>> +  // encrypted.
>>>>> +  //
>>>>> +  if (MemEncryptSevSnpIsEnabled ()) {
>>>>> +    Type = EfiACPIMemoryNVS;
>>>>> +  } else {
>>>>> +    Type = EfiBootServicesData;
>>>>> +  }
>>>>> +
>>>> Would it make sense to always use EfiACPIMemoryNVS for the injected secret area, even for regular SEV (non-SNP)?
>>>
>>> Ideally yes. Maybe James had some reasons for choosing the
>>> EfiBootServicesData. If I had to guess, it was mainly because there no
>>> guest kernel support which consumes the SEV secrets page.
>>
>> git-blame fingers commit bff2811c6d99 ("OvmfPkg/AmdSev: assign and
>> reserve the Sev Secret area", 2020-12-14).
>>
>> Commit bff2811c6d99 makes it clear that the area in question lives in MEMFD.
>>
>> We're populating the area in the PEI phase. We don't want anything in
>> DXE to overwrite it.
>>
>> Once the bootloader (and/or perhaps the kernel's EFI stub) fetched the
>> secret from that particular location, there is no need to prevent later
>> parts of the OS (the actual kernel) from repurposing that area. That's
>> why EfiBootServicesData was used.
>>
> 
> The first use of the secret area was to hold the guest luks disk passphrase; this is used in the grub-inside-OVMF (AmdSev package), and there was no need to keep that page around for the guest kernel.
> 
> The reason I'm raising this whole point is that we're working now on guest-kernel support for reading secrets from that injected page (for plain SEV).  We considered either (a) modifying the secrets page memory type to reserved here, or (b) add code to the kernel EFI stub that would copy this page somewhere else for kernel's later use (which seems more work and not sure what's the advantage).
> 
> Option (b) seems harder and more fragile, and I'm not sure if there are any advantages (though I'm definitely not an expert in that area).
> 
> 
> 
> 
>>> Since the
>>> memory is not marked ACPI NVS, so it can be used as a system RAM after
>>> the ExitBootServices is called in the kernel.
>>
>> Yes.
>>
>> I don't think AcpiNVS would be a good fit. Linux saves and restores
>> AcpiNVS areas upon S3 suspend/resume. Regardless of whether S3 works, or
>> will work, in SEV* guests, if we don't want the guest kernel to touch
>> that area *at all*, Reserved is a better type.
> 
> Thanks for this clarification.
> 
>>
>> Please refer to "Table 7-6 Memory Type Usage after ExitBootServices()"
>> in the UEFI spec (v2.9).
>>
>>>
>>> I am fine with using ACPI NVS for both SEV and SEV-SNP. I was not able
>>> to build and run AmdSev package in my setup, can you submit a prepatch
>>> to change the memory type and verify that it works ?
>>
>> NB: I've not yet reached this patch in my own review of the series, so
>> I'm likely missing some context. I do have a thought -- under SEV-SNP,
>> the secrets page apparently needs different (stronger) protection from
>> the host as under plain SEV. I don't think that hiding the different
>> protection requirements behind a single common memory type is helpful.
>> Not to mention the wasted memory in the plain SEV case -- it's not a lot
>> of memory, mind you, but the principle matters.
>>
> 
> Like I said above, we have plans to have this small amount of memory available also to the guest OS; so maybe that shouldn't be the driving force in the decision here.

What you describe (= runtime guest OS access) seems to justify changing
the memory type of this allocation. However, that update doesn't look
tied to SEV-SNP -- it's basically a "change of use case" (change of
purpose) under plain SEV too. I think that deserves a separate
(stand-alone) patch; maybe even a separate TianoCore BZ ticket.

Thanks
Laszlo



> 
> -Dov
> 
>  
> 
>> So ATM I would like to keep this patch in the SEV-SNP series, and to
>> preserve the different memory types between SEV and SEV-SNP.
>>
>> Thanks
>> Laszlo
>>
>>
>>
>>
>>>
>>>>
>>>> -Dov
>>>>
>>>>
>>>>
>>>>>    BuildMemoryAllocationHob (
>>>>>      PcdGet32 (PcdSevLaunchSecretBase),
>>>>>      PcdGet32 (PcdSevLaunchSecretSize),
>>>>> -    EfiBootServicesData
>>>>> +    Type
>>>>>      );
>>>>>
>>>>>    return EFI_SUCCESS;
>>>>> diff --git a/OvmfPkg/AmdSev/SecretPei/SecretPei.inf b/OvmfPkg/AmdSev/SecretPei/SecretPei.inf
>>>>> index 08be156c4b..9265f8adee 100644
>>>>> --- a/OvmfPkg/AmdSev/SecretPei/SecretPei.inf
>>>>> +++ b/OvmfPkg/AmdSev/SecretPei/SecretPei.inf
>>>>> @@ -26,6 +26,7 @@
>>>>>    HobLib
>>>>>    PeimEntryPoint
>>>>>    PcdLib
>>>>> +  MemEncryptSevLib
>>>>>
>>>>>  [FixedPcd]
>>>>>    gUefiOvmfPkgTokenSpaceGuid.PcdSevLaunchSecretBase
>>>>> diff --git a/OvmfPkg/OvmfPkgX64.dsc b/OvmfPkg/OvmfPkgX64.dsc
>>>>> index a7d747f6b4..593c0e69f6 100644
>>>>> --- a/OvmfPkg/OvmfPkgX64.dsc
>>>>> +++ b/OvmfPkg/OvmfPkgX64.dsc
>>>>> @@ -716,6 +716,7 @@
>>>>>    OvmfPkg/SmmAccess/SmmAccessPei.inf
>>>>>  !endif
>>>>>    UefiCpuPkg/CpuMpPei/CpuMpPei.inf
>>>>> +  OvmfPkg/AmdSev/SecretPei/SecretPei.inf
>>>>>
>>>>>  !if $(TPM_ENABLE) == TRUE
>>>>>    OvmfPkg/Tcg/Tcg2Config/Tcg2ConfigPei.inf
>>>>> @@ -965,6 +966,7 @@
>>>>>    OvmfPkg/PlatformDxe/Platform.inf
>>>>>    OvmfPkg/AmdSevDxe/AmdSevDxe.inf
>>>>>    OvmfPkg/IoMmuDxe/IoMmuDxe.inf
>>>>> +  OvmfPkg/AmdSev/SecretDxe/SecretDxe.inf
>>>>>
>>>>>  !if $(SMM_REQUIRE) == TRUE
>>>>>    OvmfPkg/SmmAccess/SmmAccess2Dxe.inf
>>>>> diff --git a/OvmfPkg/OvmfPkgX64.fdf b/OvmfPkg/OvmfPkgX64.fdf
>>>>> index d519f85328..b04175f77c 100644
>>>>> --- a/OvmfPkg/OvmfPkgX64.fdf
>>>>> +++ b/OvmfPkg/OvmfPkgX64.fdf
>>>>> @@ -88,6 +88,9 @@ gUefiCpuPkgTokenSpaceGuid.PcdSevEsWorkAreaBase|gUefiCpuPkgTokenSpaceGuid.PcdSevE
>>>>>  0x00C000|0x001000
>>>>>  gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecGhcbBackupBase|gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecGhcbBackupSize
>>>>>
>>>>> +0x00D000|0x001000
>>>>> +gUefiOvmfPkgTokenSpaceGuid.PcdSevLaunchSecretBase|gUefiOvmfPkgTokenSpaceGuid.PcdSevLaunchSecretSize
>>>>> +
>>>>>  0x010000|0x010000
>>>>>  gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecPeiTempRamBase|gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecPeiTempRamSize
>>>>>
>>>>> @@ -178,6 +181,7 @@ INF  OvmfPkg/Tcg/Tcg2Config/Tcg2ConfigPei.inf
>>>>>  INF  SecurityPkg/Tcg/TcgPei/TcgPei.inf
>>>>>  INF  SecurityPkg/Tcg/Tcg2Pei/Tcg2Pei.inf
>>>>>  !endif
>>>>> +INF  OvmfPkg/AmdSev/SecretPei/SecretPei.inf
>>>>>
>>>>>  ################################################################################
>>>>>
>>>>> @@ -313,6 +317,7 @@ INF  OvmfPkg/LinuxInitrdDynamicShellCommand/LinuxInitrdDynamicShellCommand.inf
>>>>>  INF  ShellPkg/Application/Shell/Shell.inf
>>>>>
>>>>>  INF MdeModulePkg/Logo/LogoDxe.inf
>>>>> +INF OvmfPkg/AmdSev/SecretDxe/SecretDxe.inf
>>>>>
>>>>>  #
>>>>>  # Network modules
>>>>>
>>>
>>>
>>> 
>>>
>>>
>>
>