Re: [PATCH RFC PREVIEW 2/6] iommu: Make IOMMU_IOPF selectable in Kconfig

[Joel Granados]

On Mon, Aug 26, 2024 at 11:05:58AM -0300, Jason Gunthorpe wrote:
> On Mon, Aug 26, 2024 at 01:40:28PM +0200, Klaus Jensen wrote:
> > From: Joel Granados <j.granados@xxxxxxxxxxx>
> > 
> > IOMMU_IOPF is no longer selectable through INTEL_IOMMU_SVM effectively
> > severing their relation and allowing them to be used independently.
> > 
> > Signed-off-by: Joel Granados <j.granados@xxxxxxxxxxx>
> > ---
> >  drivers/iommu/Kconfig       | 2 +-
> >  drivers/iommu/intel/Kconfig | 1 -
> >  2 files changed, 1 insertion(+), 2 deletions(-)
> > 
> > diff --git a/drivers/iommu/Kconfig b/drivers/iommu/Kconfig
> > index a82f10054aec..d3ee8a0ad4a6 100644
> > --- a/drivers/iommu/Kconfig
> > +++ b/drivers/iommu/Kconfig
> > @@ -164,7 +164,7 @@ config IOMMU_SVA
> >  	bool
> >  
> >  config IOMMU_IOPF
> > -	bool
> > +	bool "Enable IO page fault in IOMMU"
> 
> 
> Currently IOMMU_IOPF indicates that the driver wants to consume the
> library functions around IOPF, it is not a user selectable because any
> driver that links to those functions should have them working. If you
> want to make the core driver use them then the select should be moved
> from the SVM sub config to the core driver config.
This is in line with the feedback that I got from Lu. I'll put
IOMMU_IOPF under INTEL_IOMMU and drop this commit from the set.

Thx.

> 
> If you want to change IOMMU_IOPF to a user configurable then it should
> have a full help text and all the kconfig places touching it should be
> turned into
> 
>   depends IOMMU_IOPF
> 
> Ie you can't enable any drivers SVA kconfig without also picking
> IOMMU_IOPF.
> 
> This is doing neither fully.. Pick one :)
> 
> Jason

-- 

Joel Granados