Re: [RFC v2-fix-v4 1/1] x86/tdx: Skip WBINVD instruction for TDX guest

From: Kuppuswamy, Sathyanarayanan
Date: Tue Jun 08 2021 - 23:56:45 EST

On 6/8/21 8:40 PM, Dan Williams wrote:
..."KVM gets away with it" is not a justification that TDX can stand
on otherwise we would not be here fixing up ACPICA properly.

How about:

"TDX guests use standard ACPI mechanisms to signal sleep state entry
(including reboot) to the host. The ACPI specification mandates WBINVD
on any sleep state entry with the expectation that the platform is
only responsible for maintaining the state of memory over sleep
states, not preserving dirty data in any CPU caches. ACPI cache
flushing requirements pre-date the advent of virtualization. Given TDX
guest sleep state entry does not affect any host power rails it is not
required to flush caches. The host is responsible for maintaining
cache state over its own bare metal sleep state transitions that
power-off the cache. If the host fails to manage caches over its sleep
state transitions the guest..."

I don't know how to finish the last sentence. What does TDX do if it
is resumed after host suspend and the host somehow arranged for dirty
TDX lines to be lost.

TDX guest does not support S3. It will be disabled in ACPI tables. It
is a TDX firmware spec requirement. Please check the following spec,
sec 2.1

In TDX guest, we encounter cache flushes only in shutdown and reboot path.
So there is no resume path.

Will that be noticed by TDX integrity
mechanisms? I did not immediately find an answer to this with a brief
look at the specs.

Sathyanarayanan Kuppuswamy
Linux Kernel Developer