Re: [PATCH 3/6] iommu: kconfig: Move IOMMU_IOPF into INTEL_IOMMU

From: Joel Granados
Date: Mon Sep 16 2024 - 08:26:54 EST


On Sun, Sep 15, 2024 at 10:37:34AM -0300, Jason Gunthorpe wrote:
> On Wed, Sep 11, 2024 at 09:40:59AM +0200, Joel Granados wrote:
> > On Sat, Sep 07, 2024 at 09:20:19AM -0300, Jason Gunthorpe wrote:
> > > On Wed, Sep 04, 2024 at 03:17:14PM +0200, Joel Granados via B4 Relay wrote:
> > > > @@ -51,7 +52,6 @@ config INTEL_IOMMU_SVM
> > > > depends on X86_64
> > > > select MMU_NOTIFIER
> > > > select IOMMU_SVA
> > > > - select IOMMU_IOPF
> > >

Will answer this based on https://lore.kernel.org/20240913-jag-iopfv8-v2-0-dea01c2343bc@xxxxxxxxxxx
as there are small differences with the current version of this patch.

> > > Does patch 1 still compile if INTEL_IOMMU_SVM=n?
> > It does for me. With and without INTEL_IOMMU_SVM. Is there a config that
> > is not compiling for you? You can send that you to me and I can give it
> > a try, to see if I get the same issue.
>
> Okay, it compiles, but it is broken right? This hunk just can't be
> correct, SVM needs IOPF to work

AFAIK Yes: SVM *does* need IOPF to work. But IOMMU_IOPF *will* be
selected if INTEL_IOMMU_SVM is selected because INTEL_IOMMU_SVM is
covered by the `if INTEL_IOMMU` condition. Its the same way that the
relation between INTEL_IOMMU_SVM and PCI_PASID is handled: PCI_PASID
will be set if INTEL_IOMMU_SVM is set.

Please let me know if I'm misreading or have missed something.

Best

--

Joel Granados