On Tue, 2019-03-19 at 16:41 +-0200, Maxim Levitsky wrote: +AD4 +ACo All guest memory is mapped into the physical nvme device +AD4 but not 1:1 as vfio-pci would do this. +AD4 This allows very efficient DMA. +AD4 To support this, patch 2 adds ability for a mdev device to listen on +AD4 guest's memory map events. +AD4 Any such memory is immediately pinned and then DMA mapped. +AD4 (Support for fabric drivers where this is not possible exits too, +AD4 in which case the fabric driver will do its own DMA mapping) Does this mean that all guest memory is pinned all the time? If so, are you sure that's acceptable? Additionally, what is the performance overhead of the IOMMU notifier added by patch 8/9? How often was that notifier called per second in your tests and how much time was spent per call in the notifier callbacks? Thanks, Bart.
