Re: [PATCH v2 0/5] TDX host: kexec/kdump support

[Tom Lendacky <thomas.lendacky@amd.com>]

On 5/10/25 06:20, Kai Huang wrote:
> Hi Dave,
> 
> This series is the latest attempt to support kexec on TDX host following
> your suggestion to use a percpu boolean to control WBINVD during kexec.
> I appreciate if you can help to review.
> 
> The last attempt was to have one patch to make TDX and kexec mutually
> exclusive at runtime while allowing them to be both enabled in Kconfig,
> but it turns out to be overkill.  Dave proposed another option of using
> a percpu boolean to track the need for flushing:
> 
> https://lore.kernel.org/lkml/20250416230259.97989-1-kai.huang@intel.com/
> 
> One advantage of the percpu boolean is for TDX we could do optimization
> to do:
> 
> 	wbinvd();
> 	percpu(boolean) = 0;
> 
> for all CPUs at early stage to avoid having to do WBINVD in
> stop_this_cpu() at kexec time.  I made a RFC patch to show the idea:
> 
> https://github.com/intel/tdx/commit/d9f0123b1d63ba24f472da6971181939ce53d2e3
> 
> I'll also reply this RFC patch to this coverletter in case anyone wants
> to have a discussion.  Nevertheless, my exception is this series can be
> merged first w/o the RFC patch.  We can follow up the optimizations
> later.
> 
> This series is tagged v2, since it's a closer follow on to the RFC
> patchset, which was posted before the single patch:
> 
> https://lore.kernel.org/all/cover.1741778537.git.kai.huang@intel.com/
> 
> This series is based on today's tip/master.

I'm on PTO for the next few days, so I'll try to review/test the SME
related changes when I return.

Thanks,
Tom

>