Re: [PATCH 1/2] i386/tdx: Fix build on 32-bit host

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

On 6/6/25 11:28, Xiaoyao Li wrote:
> On 6/6/2025 5:19 PM, Philippe Mathieu-Daudé wrote:
>> On 6/6/25 10:49, Xiaoyao Li wrote:
>>> On 6/3/2025 10:53 PM, Paolo Bonzini wrote:
>>>> On 6/3/25 13:26, Philippe Mathieu-Daudé wrote:
>>>>> On 3/6/25 05:04, Xiaoyao Li wrote:
>>>>>> On 6/3/2025 1:31 AM, Cédric Le Goater wrote:
>>>>>>> Use PRI formats where required and fix pointer cast.
>>>>>>
>>>>>> Maybe we can make 32-bit build exclusive with CONFIG_TDX? since 
>>>>>> TDX is not supported on 32-bit host.
>>>>>
>>>>> Yes please!
>>>>
>>>> No objections, but I'm still applying these first to fix the build.
>>>
>>> Can anyone guide how to implement it? Or directly help cook a patch?
>>>
>>> I'm struggling to learn the 32-bit build stuff and create a 32-bit 
>>> environment.
>>
>> -- >8 --
>> diff --git a/hw/i386/Kconfig b/hw/i386/Kconfig
>> index eb65bda6e07..b5970f9a1f3 100644
>> --- a/hw/i386/Kconfig
>> +++ b/hw/i386/Kconfig
>> @@ -13,7 +13,7 @@ config SGX
>>   config TDX
>>       bool
>>       select X86_FW_OVMF
>> -    depends on KVM
>> +    depends on KVM && !I386
>>
>> ---
> 
> CONFIG_X86_64 selects I386 in target/i386/Kconfig, so above change will 
> just leads to CONFIG_TDX always being 0.
> 
> config X86_64
>      bool
>      select I386

Doh. We could expose TARGET_LONG_BITS to Kconfig (see commit 
bae3e3a5c6b, "meson: make target endianneess available to Kconfig"),
but there is likely a clever way.