Hi Joerg,
On 5/2/2022 4:24 PM, Joerg Roedel wrote: > Hi Vasant, > > On Fri, Apr 29, 2022 at 08:15:49PM +0530, Vasant Hegde wrote: >> We still need to parse IVHD to find max devices supported by each PCI segment >> (same as the way its doing it today). Hence we need all these variables. > > From what I have seen since a few years the IVRS tables enumerate the > whole PCI segment, up to device ff:1f.7. This results in the maximum > being allocated for all data structures anyway. Therefore we can > probably think about skipping the scan to find the largest bdf and just > assume it is ff:1f.7, saving us all the size-tracking variables? With PCI segment, I think we will have segments with less than ff:1f:7. Hence we need these variables. -Vasant _______________________________________________ iommu mailing list iommu@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/iommu
