Re: [PATCH v2 5/5] vfio/pci: Latch all module parameters per device
From: fengchengwen
Date: Sun Jun 14 2026 - 21:00:33 EST
On 6/12/2026 10:11 PM, Alex Williamson wrote:
> On Fri, 12 Jun 2026 16:40:37 +0800
> fengchengwen <fengchengwen@xxxxxxxxxx> wrote:
>>
>> Since nointxmask and disable_idle_d3 already have S_IWUSR and share the same
>> "latch per device at init" behavior, it would be better to also add S_IWUSR
>> to disable_vga for permission and behavior consistency.
>>
>> Runtime writes will only affect future devices, existing devices keep their
>> latched value, so no extra risk.
>
> The difference is that notintxmask and disable_idle_d3 are typically
> diagnostic measures while disable_vga is a feature mask. disable_vga
> has never supported runtime modification. So, while I agree it's
> possible, this would expand the scope of this series from bug fixes and
> behavior restoration into feature enablement, which should have some
> actual use case in mind. Therefore I don't intend to pursue it in this
> series. Thanks,
Understood and agreed. Thanks for clarifying the scope.
>
> Alex
>