Re: [PATCH v2 3/4] iommu: Redefine IOMMU_CAP_CACHE_COHERENCY as the cap flag for IOMMU_CACHE

[Jason Gunthorpe <jgg@nvidia.com>]

On Fri, Apr 08, 2022 at 08:21:55AM +0000, Tian, Kevin wrote:
> (CC Jason Wang)
> 
> > From: Jason Gunthorpe
> > Sent: Thursday, April 7, 2022 11:24 PM
> > 
> > While the comment was correct that this flag was intended to convey the
> > block no-snoop support in the IOMMU, it has become widely implemented
> > and
> > used to mean the IOMMU supports IOMMU_CACHE as a map flag. Only the
> > Intel
> > driver was different.
> > 
> > Now that the Intel driver is using enforce_cache_coherency() update the
> > comment to make it clear that IOMMU_CAP_CACHE_COHERENCY is only
> > about
> > IOMMU_CACHE.  Fix the Intel driver to return true since IOMMU_CACHE
> > always
> > works.
> > 
> > The two places that test this flag, usnic and vdpa, are both assigning
> > userspace pages to a driver controlled iommu_domain and require
> > IOMMU_CACHE behavior as they offer no way for userspace to synchronize
> > caches.
> > 
> > Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
> 
> Reviewed-by: Kevin Tian <kevin.tian@intel.com>
> 
> btw the comment about vsnic and vdpa matches my thought. But
> a recent change in Qemu [1] possibly wants confirmation from
> Jason Wang.
> 
> [1] https://lore.kernel.org/all/20220304133556.233983-20-mst@redhat.com/

That patch seems to have run into the confusion this series is
addressing.

VFIO_DMA_CC_IOMMU and snoop control is absolutely not needed by
VDPA. We expect the VDPA kernel driver to be well behaved and not
cause its device to generate no-snoop TLPs.

VDPA needs IOMMU_CACHE only.

Jason