On 08/08/17 21:16, Jonas Pfefferle wrote:
> DMA window size needs to be big enough to span all memory segment's
> physical addresses. We do not need multiple levels of IOMMU tables
> as we already span ~70TB of physical memory with 16MB hugepages.
>
> Signed-off-by: Jonas Pfefferle <j...@zurich.ibm.com>
Reviewed-by: Alexey Kardashevskiy <a...@ozlabs.ru>
> ---
> v2:
> * roundup to next power 2 function without loop.
>
> v3:
> * Replace roundup_next_pow2 with rte_align64pow2
>
> v4:
> * do not assume ordering of physical addresses of memsegs
>
> lib/librte_eal/linuxapp/eal/eal_vfio.c | 20 +++++++++++++++++---
> 1 file changed, 17 insertions(+), 3 deletions(-)
>
> diff --git a/lib/librte_eal/linuxapp/eal/eal_vfio.c
> b/lib/librte_eal/linuxapp/eal/eal_vfio.c
> index 946df7e..7d5d61d 100644
> --- a/lib/librte_eal/linuxapp/eal/eal_vfio.c
> +++ b/lib/librte_eal/linuxapp/eal/eal_vfio.c
> @@ -759,10 +759,19 @@ vfio_spapr_dma_map(int vfio_container_fd)
> return -1;
> }
>
> - /* calculate window size based on number of hugepages configured */
> - create.window_size = rte_eal_get_physmem_size();
> + /* create DMA window from 0 to max(phys_addr + len) */
> + for (i = 0; i < RTE_MAX_MEMSEG; i++) {
> + if (ms[i].addr == NULL)
> + break;
> +
> + create.window_size = RTE_MAX(create.window_size,
> + ms[i].phys_addr + ms[i].len);
> + }
> +
> + /* sPAPR requires window size to be a power of 2 */
> + create.window_size = rte_align64pow2(create.window_size);
> create.page_shift = __builtin_ctzll(ms->hugepage_sz);
> - create.levels = 2;
> + create.levels = 1;
>
> ret = ioctl(vfio_container_fd, VFIO_IOMMU_SPAPR_TCE_CREATE, &create);
> if (ret) {
> @@ -771,6 +780,11 @@ vfio_spapr_dma_map(int vfio_container_fd)
> return -1;
> }
>
> + if (create.start_addr != 0) {
> + RTE_LOG(ERR, EAL, " DMA window start address != 0\n");
> + return -1;
> + }
> +
> /* map all DPDK segments for DMA. use 1:1 PA to IOVA mapping */
> for (i = 0; i < RTE_MAX_MEMSEG; i++) {
> struct vfio_iommu_type1_dma_map dma_map;
>
--
Alexey
