On 16.06.2025 12:37, Stewart Hildebrand wrote:
> On 6/16/25 02:42, Jan Beulich wrote:
>> On 13.06.2025 17:17, Stewart Hildebrand wrote:
>>> --- a/xen/arch/arm/Kconfig
>>> +++ b/xen/arch/arm/Kconfig
>>> @@ -8,6 +8,8 @@ config ARM_64
>>> depends on !ARM_32
>>> select 64BIT
>>> select HAS_FAST_MULTIPLY
>>> + select HAS_VPCI_GUEST_SUPPORT if PCI_PASSTHROUGH
>>> + select HAS_PASSTHROUGH if PCI_PASSTHROUGH
>>
>> Seeing this, I like this as little as I liked ...
>>
>>> @@ -258,6 +260,12 @@ config PARTIAL_EMULATION
>>>
>>> source "arch/arm/firmware/Kconfig"
>>>
>>> +config PCI_PASSTHROUGH
>>> + bool "PCI passthrough" if EXPERT
>>> + depends on ARM_64
>>
>> ... the form with the select-s put here. I'll (obviously) leave it to the
>> Arm maintainers to judge, but my recommendation would be to simply drop
>> this patch. As per the description it's merely "make it easier ...",
>> which imo doesn't warrant such an abuse of HAS_*.
>
> "easier" was a poor choice of word. "possible" is more accurate. This
> patch addresses a real issue: currently the PCI and vPCI bits can't be
> built for Arm, allowing build issues to go unnoticed. E.g. see
> 4ce671963eb1 ("xen/arm: fix build with HAS_PCI").
Which gets us back to the question of whether to use "depends on
HAS_PASSTHROUGH" (I think yes then) and where to put the remaining select
(might then better move back to the new option).
Jan
