Re: [PATCH] x86: Fix AMD_SVM and INTEL_VMX dependency

["Orzel, Michal" <michal.orzel@xxxxxxx>]

On 01/09/2025 13:19, Jan Beulich wrote:
> On 01.09.2025 12:43, Michal Orzel wrote:
>> Commit e3ed540f2e9f was meant to make AMD_SVM dependent on AMD and
>> INTEL_VMX on INTEL. Such dependency should be done using 'depends on'
>> and not 'if' next to prompt that deals only with the visibility of the
>> given Kconfig option. This makes it impossible to e.g. disable INTEL_VMX
>> when INTEL is disabled (option is hidden).
> 
> Hmm, yes, just that ...
> 
>> --- a/xen/arch/x86/hvm/Kconfig
>> +++ b/xen/arch/x86/hvm/Kconfig
>> @@ -16,7 +16,8 @@ menuconfig HVM
>>  if HVM
>>  
>>  config AMD_SVM
>> -    bool "AMD-V" if AMD && EXPERT
>> +    bool "AMD-V" if EXPERT
>> +    depends on AMD
>>      default y
>>      help
>>        Enables virtual machine extensions on platforms that implement the
>> @@ -25,7 +26,8 @@ config AMD_SVM
>>        If in doubt, say Y.
>>  
>>  config INTEL_VMX
>> -    bool "Intel VT-x" if INTEL && EXPERT
>> +    bool "Intel VT-x" if EXPERT
>> +    depends on INTEL
>>      default y
>>      select ARCH_VCPU_IOREQ_COMPLETION
>>      help
> 
> ... now it becomes impossible to _enable_ INTEL_VMX when INTEL is disabled,
> yet which may be of interest if you target some other vendor's VMX
> implementation. Perhaps really we should have
> 
> config INTEL_VMX
>       bool "Intel VT-x" if EXPERT
>       default INTEL
I did not think such configuration makes sense (VMX without INTEL)
which is in line with the last sentence from the mentioned commit.
I'm ok either way.

~Michal