Re: [PATCH v10 03/11] x86/cpufeatures: Add TDX Guest CPU feature

[Sathyanarayanan Kuppuswamy <sathyanarayanan.kuppuwamy@linux.intel.com>]

On 10/13/21 4:02 PM, Thomas Gleixner wrote:
> On Wed, Oct 13 2021 at 15:28, Sathyanarayanan Kuppuswamy wrote:
>> On 10/13/21 2:37 PM, Borislav Petkov wrote:
>>> On Wed, Oct 13, 2021 at 11:25:35PM +0200, Thomas Gleixner wrote:
>>>> So this ends up in doing:
>>>>
>>>>      use();
>>>>      init();
>>>>
>>>> Can you spot what's wrong with that?
>>>>
>>>> That's a clear violation of common sense and is simply not going to
>>>> happen. Why? If you think about deep defensive programming then use()
>>>> will look like this:
>>>>
>>>> use()
>>>> {
>>>>           assert(initialized);
>>>> }
>>>>
>>>> which is not something made up. It's a fundamental principle of
>>>> programming and some languages enforce that for very good reasons.
>>>>
>>>> Just because it can be done in C is no justification.
>>> Oh, I heartily agree.
>>>    
>>>> What's wrong with:
>>>>
>>>> x86_64_start_kernel()
>>>>
>>>>       tdx_early_init();
>>>>
>>>>       copy_bootdata();
>>>>       
>>>>       tdx_late_init();
>>>>
>>>> Absolutely nothing. It's clear, simple and well defined.
>>> I like simple more than anyone, so sure, I'd prefer that a lot more.
>>>
>>> And so the options parsing would need to happen early using, say,
>>> cmdline_find_option() or so, like sme_enable() does.
>> Since in tdx_early_init() all we are going to do is to initialize
>> "tdx_guest_detected" using cpuid call, shall we name it
>> tdx_guest_cpuid_init()? (similar to sme_enable call in AMD)
> How is that similar?
>
> Just chose a name which makes sense in the overall scheme. I surely care
> about naming convetions, but what I care more about is correctness.
>
> Whether it ends up being named
>
>          tdx_enable() - to match the SME muck
>
> or
>
>          tdx_detect()
>
> or whatever makes sense does not really matter. As long as it makes
> sense. That's bikeshed painting realm.
>
> Coming back to your suggestion 'tdx_guest_cpuid_init()'. Just sit back
> and think about what that name says:
>
>      tdx_guest_cpuid_init()
>
> For the uniformed reader this says:
>
>      If tdx guest then initialize CPUID
>
> which is obviously not what you want to express, right?
>
> So, naming matters but you are free to chose something which makes
> sense.

Makes sense. I agree tdx_guest_cpuid_init() name is bit confusing.
I will use tdx_detect as you have mentioned.

>
> Thanks,
>
>          tglx

-- 
Sathyanarayanan Kuppuswamy
Linux Kernel Developer