On Tue, 2013-05-28 at 12:38 -0700, Stephen Hemminger wrote: > On Tue, 28 May 2013 12:40:20 -0600 > Alex Williamson <alex.williamson@xxxxxxxxxx> wrote: > > > Some PCIe-to-PCI bridges are not fully compliant with the PCIe spec > > and do not include a PCIe capability. pci_is_pcie() is not useful on > > these devices, making it difficult to determine where we transition > > from a legacy PCI bus to a PCIe link. PCI-core doesn't want a quirked > > bridge test for fear of confusion that a PCIe capability would be > > expected, so implement it in IOMMU code. > > > > Signed-off-by: Alex Williamson <alex.williamson@xxxxxxxxxx> > > This won't work for me because I run without IOMMU. > The overhead of IOMMU setup per-packet significantly impacts network performance. You were one of the reporters of this issue: http://www.spinics.net/lists/linux-pci/msg19579.html Does this not work for you or are you saying this is not relevant to you because you no longer attempt to enable the IOMMU? Thanks, Alex -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
