- Measure streaming vs uncached performance?
- Prefer cached memory for streaming DMA to userspace?  https://aelseb.wordpress.com/2015/04/11/contiguous-memory-on-arm-and-cache-coherency/
- Check IOMMU existence and use vmalloc instead of kmalloc for dma_map_*()?
- Invalidate caches? dma_sync_single_*() not used.