On Thu, Aug 9, 2018 at 11:43 AM, Bjorn Helgaas <helgaas@xxxxxxxxxx> wrote: > [+cc David, Logan, Alex, iommu list] > > On Thu, Aug 09, 2018 at 11:14:13AM -0700, Eric Pilmore wrote: >> Didn't get any response on the IRC channel so trying here. >> >> Was wondering if anybody here has used IOAT DMA engines with an >> IOMMU turned on (Xeon based system)? My specific question is really >> whether it is possible to DMA (w/IOAT) to a PCI BAR address as the >> destination without having to map that address to the IOVA space of >> the DMA engine first (assuming the IOMMU is on)? > > So is this a peer-to-peer DMA scenario? You mention DMA, which would > be a transaction initiated by a PCI device, to a PCI BAR address, so > it doesn't sound like system memory is involved. No, not peer-to-peer. This is from system memory (e.g. SKB buffer which has had an IOMMU mapping created) to a PCI BAR address. > > I copied some folks who know a lot more about this than I do. Thank you! > >> I am encountering issues where I see PTE Errors reported from DMAR >> in this scenario, but I do not if I use a different DMA engine >> that's sitting downstream off the PCI tree. I'm wondering if the >> IOAT DMA failure is some artifact of these engines sitting behind >> the Host Bridge. >> >> Thanks in advance! >> Eric
