Re: [PATCH v8 5/5] x86/tdx: Add Quote generation support

[Erdem Aktas <erdemaktas@google.com>]

I apologize for bumping an old thread up and getting in the discussion
this late.

Bumping this thread because as Chong mentioned below, it is essential
for us to have TDG.VP.VMCALL<GetQuote> interface as we do not
enable/support virtio in many use cases.
Coming from the  "Add TDX Guest Attestation support patch set [1]"
which does not have the Quote support as per Dave's review suggestion,
I am looking for what our options are here.

@Hansen, Dave  : as far as I understand your concerns were:

>>Do we *REALLY* need specific driver functionality for this? For
>>instance, is there no existing virtio device that can send blobs back and forth?

As Chong and I said, we have to have a vsock alternative solution if
we want to provide TDX based products for our customers. Otherwise it
will really limit our ability to adopt this technology.

>> You're basically saying that every confidential computing technology
>> should have its own host user <-> guest kernel <-> guest user ABI.
>> That's insanity.  If we do this, we need *one* interface that says "talk
>> to the hypervisor" that's common for all hypervisors and hardware
>> vendors, or at least more than *one*.

This is already happening, right? AMD has GHCB ABI while intel has
GHCI. Also there is already a guest driver and IOCTL interface for
TDREPORT, so would it be really that insane to have another IOCTL
interface for GetQuote?

I would like to get your opinion on how we can progress on this?

-Erdem

[1] https://lore.kernel.org/lkml/20221116223820.819090-1-sathyanarayanan.kuppuswamy@linux.intel.com/t/#m341e6475982318d74be160c09a1856ad87e184cf

> From: Chong Cai <chongc@google.com>
> Date: Fri, Jan 6, 2023 at 2:02 PM
> Subject: Re: [PATCH v8 5/5] x86/tdx: Add Quote generation support
> To: Nakajima, Jun <jun.nakajima@intel.com>
> Cc: <ak@linux.intel.com>, <bp@alien8.de>, <dave.hansen@intel.com>, <dave.hansen@linux.intel.com>, <hpa@zytor.com>, <isaku.yamahata@gmail.com>, <kai.huang@intel.com>, <khalid.elmously@canonical.com>, <kirill.shutemov@linux.intel.com>, <linux-kernel@vger.kernel.org>, <marcelo.cerri@canonical.com>, <mingo@redhat.com>, <philip.cox@canonical.com>, <sathyanarayanan.kuppuswamy@linux.intel.com>, <tglx@linutronix.de>, <tim.gardner@canonical.com>, <tony.luck@intel.com>, <wander@redhat.com>, <x86@kernel.org>
>
>
> Hi everyone,
>
>
> Sorry for being late in the discussion.
>
>
> We just want to add that we have use cases where virtio/vsock is not supported at all; therefore, getquote TDVMCALL support is required for us to adopt this technology.
>
>
> Is there anything we can help here?
>
>


On Sun, Nov 20, 2022 at 6:26 PM Dave Hansen <dave.hansen@intel.com> wrote:
>
> On 11/20/22 18:04, Guorui Yu wrote:
> > Kindly ping for any comments here?
>
> No comments from me!
>
> Seriously, this is from June.  I honestly have no idea what this patch
> is for or why you're asking about it.  If it were  important, I would
> have expected a new post with a new cover letter telling me some time in
> the past 5-6 months.
>
> Oh, and btw, I haven't seen you reviewing any other x86 submissions
> during that past 5-6 months either.  I'm tend to be more likely to bump
> things up near the front of the queue for folks that are helping to keep
> the queue short and manageable.