Re: [PATCH v2 0/6] RPi4: Enable ACPI PCIe conduit

["Jeremy Linton" <jeremy.linton@arm.com>]

Hi,

On 8/17/21 10:09 AM, Ard Biesheuvel wrote:
> On Tue, 10 Aug 2021 at 05:36, Jeremy Linton <jeremy.linton@arm.com> wrote:
>>
>> A new Arm standard DEN0115A specifies how
>> platforms that don't have standard ECAM can
>> use the firmware to handle config read/write
>> operations. This is mostly implemented in TFA
>> but UEFI needs to assure that there is a
>> description of the root complex in the ACPI
>> namespace.
>>
>> This set adds that description based on
>> a new menu item which toggles between XHCI
>> platform description and PCIe via a BDS
>> menu selection on the RPi4. The CM4 is really
>> the platform that needs this as it has a PCIe
>> slot. On that platform PCIe is enabled by default.
>>
>> v1->v2:
>>      Use global shared interrupts in PCI PRT
>>          which is a pretty significant
>>          simplification.
>>      Modify bus max to use the secondary side
>>          of the root port for enforcing device
>>          limits
>>      Various other AML cleanups per Ard (drop
>>          redundant _DMA, bump UID to make it
>>          unique, etc)
> 
> I still see a _DMA method on the PCIE node, and the UID you chose is
> already used elsewhere.

Yah, thanks for keeping me honest here, I'm not really sure what 
happened with the _UID, but I would have probably killed windows again. 
<sigh>

OTOH, picking off the pcie bridge specific _DMA causes, failures, which 
don't really make sense either, except that the ACPI spec explicitly 
notes the case of a PCIe bridge with a DMA constrain needs to have it 
defined on the device responsible for the translation/limitation.

In the end really, the container is unnecessarily, so removing it solves 
both the problem of all these duplicate containers as well as the 
duplicate DMA producer.

I will repost RSN.

> 
>>      Break link status move into its own patch
>>      MADT->MCFG typos in various comments
>>      Commit message tweaking
>>
>> Jeremy Linton (6):
>>    Platform/RaspberryPi: Add XHCI/PCI selection menu
>>    Platform/RaspberryPi: Break XHCI into its own SSDT
>>    Platform/RaspberryPi: Add PCIe SSDT
>>    Silicon/Broadcom/Bcm27xx: Relax PCIe device restriction
>>    Silicon/Broadcom/Bcm27xx: Move linkup check into the cfg accessor
>>    Platform/RaspberryPi: Enable NVMe boot on CM4
>>
>>   Platform/RaspberryPi/AcpiTables/AcpiTables.inf     |   4 +
>>   Platform/RaspberryPi/AcpiTables/Dsdt.asl           |   3 -
>>   Platform/RaspberryPi/AcpiTables/Pci.asl            | 209 +++++++++++++++++++++
>>   Platform/RaspberryPi/AcpiTables/Xhci.asl           |  35 ++--
>>   Platform/RaspberryPi/Drivers/ConfigDxe/ConfigDxe.c |  56 ++++++
>>   .../RaspberryPi/Drivers/ConfigDxe/ConfigDxe.inf    |   1 +
>>   .../RaspberryPi/Drivers/ConfigDxe/ConfigDxeHii.uni |   5 +
>>   .../RaspberryPi/Drivers/ConfigDxe/ConfigDxeHii.vfr |  17 ++
>>   Platform/RaspberryPi/Include/ConfigVars.h          |   4 +
>>   Platform/RaspberryPi/RPi3/RPi3.dsc                 |   6 +
>>   Platform/RaspberryPi/RPi4/RPi4.dsc                 |  13 ++
>>   Platform/RaspberryPi/RPi4/RPi4.fdf                 |   5 +
>>   Platform/RaspberryPi/RaspberryPi.dec               |   1 +
>>   .../Bcm2711PciHostBridgeLibConstructor.c           |   5 -
>>   .../Library/Bcm2711PciSegmentLib/PciSegmentLib.c   |  28 ++-
>>   15 files changed, 364 insertions(+), 28 deletions(-)
>>   create mode 100644 Platform/RaspberryPi/AcpiTables/Pci.asl
>>
>> --
>> 2.13.7
>>