Re: [PATCH 1/2] x86/virt/tdx: Use PFN directly for mapping guest private memory

[Dave Hansen]

On 3/20/26 10:31, Edgecombe, Rick P wrote:
> On Fri, 2026-03-20 at 12:59 +0000, Kiryl Shutsemau wrote:
>>> I feel like we argued about this before. But it would be more correct to
>>> just drop level and make it 4k only until huge pages? Otherwise we are
>>> tweaking dead behavior.
>> I guess. But you add one more thing on the list to remember when adding
>> huge page support. This kind of stuff is easy to miss.
> I guess I'm fine either way. I'm not sure Dave will like adding dead branches
> though. I suppose huge pages is close enough that we are looking to merge prep
> work anyway.

Can we add something that will BUG_ON() or fail to compile when the huge
page support comes around?

I'd much rather have:

	BUG_ON(level != PG_LEVEL_4K);
	tdx_clflush_pfn(pfn);

That go implementing a level argument to tdx_clflush_pfn() now. Then
nobody has to remember. The list to remember is in the code.