On Tue, Aug 27, 2019 at 11:24:05AM -0700, Derrick, Jonathan wrote: > On Mon, 2019-08-26 at 17:06 +0200, Christoph Hellwig wrote: > > With a little tweak to the intel-iommu code we should be able to work > > around the VMD mess for the requester IDs without having to create giant > > amounts of boilerplate DMA ops wrapping code. The other advantage of > > this scheme is that we can respect the real DMA masks for the actual > > devices, and I bet it will only be a matter of time until we'll see the > > first DMA challeneged NVMe devices. > > > > The only downside is that we can't offer vmd as a module given that > > intel-iommu calls into it. But the driver only has about 700 lines > > of code, so this should not be a major issue. > If we're going to remove its ability to be a module, and given its > small size, could we make this default =y? > > Otherwise we risk breaking platforms which have it enabled with OSVs > who miss enabling it Can we keep this as a module if we stick the remapping struct device in pci_sysdata instead of going through the vmd driver to get it? Otherwise, very happy to see this dma wrapping go away.
