Re: [RFC PATCH 00/20] TD-Preserving updates

[Chao Gao <chao.gao@intel.com>]

On Fri, May 23, 2025 at 02:52:23AM -0700, Chao Gao wrote:
>Hi Reviewers,
>
>This series adds support for runtime TDX module updates that preserve
>running TDX guests (a.k.a, TD-Preserving updates). The goal is to gather
>feedback on the feature design. Please pay attention to the following items:
>
>1. TD-Preserving updates are done in stop_machine() context. it copy-pastes
>   part of multi_cpu_stop() to guarantee step-locked progress on all CPUs.
>   But, there are a few differences between them. I am wondering whether
>   these differences have reached a point where abstracting a common
>   function might do more harm than good. See more details in patch 10.
>
>2. P-SEAMLDR seamcalls (specificially SEAMRET from P-SEAMLDR) clear current
>   VMCS pointers, which may disrupt KVM. To prevent VMX instructions in IRQ
>   context from encountering NULL current-VMCS pointers, P-SEAMLDR
>   seamcalls are called with IRQ disabled. I'm uncertain if NMIs could
>   cause a problem, but I believe they won't. See more information in patch 3.
>
>3. Two helpers, cpu_vmcs_load() and cpu_vmcs_store(), are added in patch 3
>   to save and restore the current VMCS. KVM has a variant of cpu_vmcs_load(),
>   i.e., vmcs_load(). Extracting KVM's version would cause a lot of code
>   churn, and I don't think that can be justified for reducing ~16 LoC
>   duplication. Please let me know if you disagree.

Gentle ping!

There are three open issues: one regarding stop_machine() and two related to
interactions with KVM.

Sean and Paul, do you have any preferences or insights on these matters?