Re: [PATCH v8 06/11] x86/traps: Add #VE support for TDX guest

[Dave Hansen <dave.hansen@intel.com>]

On 10/7/21 10:06 AM, Borislav Petkov wrote:
> On Mon, Oct 04, 2021 at 07:52:00PM -0700, Kuppuswamy Sathyanarayanan wrote:
>> +unsigned long tdx_get_ve_info(struct ve_info *ve)
>> +{
>> +	struct tdx_module_output out = {0};
>> +	u64 ret;
>> +
>> +	/*
>> +	 * NMIs and machine checks are suppressed. Before this point any
>> +	 * #VE is fatal. After this point (TDGETVEINFO call), NMIs and
>> +	 * additional #VEs are permitted (but it is expected not to
>> +	 * happen unless kernel panics).
>> +	 */
>> +	ret = __tdx_module_call(TDGETVEINFO, 0, 0, 0, 0, &out);
> Same question as before - why do you need to clear this @out thing above
> when __tdx_module_call() will overwrite it?
> 
> What you should do instead is check that @ve pointer which you get
> passed in - it might be NULL.

Hi Borislav,

That ve_info really is specific to handling a fault.  There's only one
call site:

> +DEFINE_IDTENTRY(exc_virtualization_exception)
> +{
> +	struct ve_info ve;
...
> +	ret = tdx_get_ve_info(&ve);

I think it's equivalent to something like a 'pt_regs' or 'stack_info'
that we pass around in other exception handlers.  It's always stack
allocated.  It's never dynamically allocated and NULL is never passed
for some other semantic reason.