Re: Using a generic PciHostBridgeDxe driver for a multi-PCIe-domain platform

[Vladimir Olovyannikov <vladimir.olovyannikov@broadcom.com>]

> -----Original Message-----
> From: Laszlo Ersek [mailto:lersek@redhat.com]
> Sent: May-30-17 11:50 AM
> To: Vladimir Olovyannikov; edk2-devel@lists.01.org
> Cc: Ard Biesheuvel
> Subject: Re: [edk2] Using a generic PciHostBridgeDxe driver for a
> multi-PCIe-
> domain platform
>
> On 05/30/17 20:32, Laszlo Ersek wrote:
> > On 05/30/17 18:23, Vladimir Olovyannikov wrote:
> >> Hi,
> >>
> >> I've started PCIe stack implementation design for an armv8 aarch64
> >> platform.
> >> The platform's PCIe represents several host bridges, and each
> >> hostbridge has one rootbridge.
> >> They do not share any resources between each other.
> >> Looking into the PciHostBridgeDxe implementation I can see that it
> >> supports only one hostbridge, and there is a comment:
> >> // Most systems in the world including complex servers have only one
> >> Host Bridge.
> >>
> >> So in my case should I create my own PciHostBridgeDxe driver
> >> supporting multiple hostbridges and do not use the Industry standard
> driver?
> >> I am very new to it, and will appreciate any help or idea.
> >
> > I think you can use PciHostBridgeDxe on this platform:
> >
> > - Implement a PciHostBridgeLib instance (see
> > <MdeModulePkg/Include/Library/PciHostBridgeLib.h>) that returns
> > PCI_ROOT_BRIDGE objects with different Segment fields, from
> > PciHostBridgeGetRootBridges().
> >
> > - Implement a PciSegmentLib instance (see
> > <MdePkg/Include/Library/PciSegmentLib.h>) that routes the config space
> > addresses, encoded by the PCI_SEGMENT_LIB_ADDRESS() macro,
> according
> > to your platform.
> >
> > PciHostBridgeDxe and PciBusDxe should "just work" atop. To my
> > understanding, PciBusDxe delegates all config space accesses to
> > PciHostBridgeDxe, via EFI_PCI_ROOT_BRIDGE_IO_PROTOCOL. And
> > PciHostBridgeDxe delegates all config space accesses to the platform's
> > PciSegmentLib.
>
> A small addition. Assuming the general case, i.e. when you have a
> different
> number of root bridges on each of several host bridges, you still have to
> number all those root bridges incrementally, in a curious, flat address
> space.
> And that address space is the PcieRoot(N) device path node that is
> supposed
> to start the "DevicePath" member of each PCI_ROOT_BRIDGE object that
> you return from PciHostBridgeGetRootBridges().
>
> You can read about the PcieRoot() devpath node in the UEFI 2.6 spec.
> Basically, you have
>
> ACPI_HID_DEVICE_PATH.Header = <fill in as usual>;
> ACPI_HID_DEVICE_PATH.HID = EFI_PNP_ID (0x0a08);
> ACPI_HID_DEVICE_PATH.UID = <fill in incrementally across host bridges>;
>
> The UID values used in these devpath nodes should preferably match the
> UID values of the corresponding PCI Express Root Bridge objects
> (=PNP0A08) that you expose in your ACPI tables (DSDT and/or SSDT).
Thanks for explanation Laszlo,
In my case every hostbridge has exactly one rootbridge. I will follow your
advice and will create
SegmentPciLib and platform's PciHostBridgeLib.
So basically my PciHostBridgeLib should treat hostbridges as rootbridges
with different segments in this case?
>
> Thanks
> Laszlo

Thank you,
Vladimir