On Fri, Dec 06, 2019 at 01:52:45PM -0800, Armen Baloyan wrote:
> Intel SkyLake-E was successfully tested for p2pdma
> transactions spanning over a host bridge and PCI
> bridge with IOMMU on.
>
> Signed-off-by: Armen Baloyan <abaloyan@xxxxxxxxxx>
Applied with Logan's reviewed-by to pci/p2pdma for v5.6, thanks!
Logan, the commit log for 494d63b0d5d0 ("PCI/P2PDMA: Whitelist some
Intel host bridges") says:
Intel devices do not have good support for P2P requests that span different
host bridges as the transactions will cross the QPI/UPI bus and this does
not perform well.
Therefore, enable support for these devices only if the host bridges match.
Add Intel devices that have been tested and are known to work. There are
likely many others out there that will need to be tested and added.
That suggests it's possible that P2P DMA may actually work but with
poor performance, and that you only want to add host bridges that work
with good performance to the whitelist.
Armen found that it *works*, but I have no idea what the performance
was. Do you care about the performance, or is this a simple "works /
doesn't work" test?
> ---
> drivers/pci/p2pdma.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/pci/p2pdma.c b/drivers/pci/p2pdma.c
> index 79fcb8d8f1b1..9f8e9df8f4ca 100644
> --- a/drivers/pci/p2pdma.c
> +++ b/drivers/pci/p2pdma.c
> @@ -324,6 +324,9 @@ static const struct pci_p2pdma_whitelist_entry {
> /* Intel Xeon E7 v3/Xeon E5 v3/Core i7 */
> {PCI_VENDOR_ID_INTEL, 0x2f00, REQ_SAME_HOST_BRIDGE},
> {PCI_VENDOR_ID_INTEL, 0x2f01, REQ_SAME_HOST_BRIDGE},
> + /* Intel SkyLake-E. */
> + {PCI_VENDOR_ID_INTEL, 0x2030, 0},
> + {PCI_VENDOR_ID_INTEL, 0x2020, 0},
> {}
> };
>
> --
> 2.16.5
>
