Re: [edk2-platforms][PATCH 1/1] Platform/RPi4: Add ACPI entry for Genet network interface

["Pete Batard" <pete@akeo.ie>]

On 2020.02.08 08:52, Ard Biesheuvel wrote:
> On Mon, 3 Feb 2020 at 13:40, Pete Batard <pete@akeo.ie> wrote:
>>
>> The Raspberry Pi 4 platforms uses a Broadcom Genet network interface, for
>> which we need ACPI entries in order to make it usable under Linux.
>>
>> This patch adds these entries, including a max-dma-burst-size DSD attribute
>> aimed at simplifying support for Genet on distros that use older kernels,
>> such as Debian.
>>
>> Note that we ran these settings through someone working for Broadcom, who
>> okayed the proposed values including ownership of max-dma-burst-size (which
>> we expect to also require for Device Tree usage on older kernels, hence the
>> requirement for a designated owner).
>>
>> Signed-off-by: Pete Batard <pete@akeo.ie>
>> ---
>>   Platform/RaspberryPi/RPi4/AcpiTables/Dsdt.asl | 29 ++++++++++++++++++++
>>   1 file changed, 29 insertions(+)
>>
>> diff --git a/Platform/RaspberryPi/RPi4/AcpiTables/Dsdt.asl b/Platform/RaspberryPi/RPi4/AcpiTables/Dsdt.asl
>> index b2f1d3439211..12c3967fa9e1 100644
>> --- a/Platform/RaspberryPi/RPi4/AcpiTables/Dsdt.asl
>> +++ b/Platform/RaspberryPi/RPi4/AcpiTables/Dsdt.asl
>> @@ -267,6 +267,35 @@ DefinitionBlock ("Dsdt.aml", "DSDT", 5, "MSFT", "EDK2", 2)
>>         }
>>       }
>>
>> +    Device (ETH0)
>> +    {
>> +      Name (_HID, "BCM6E4E")
>> +      Name (_CID, "BCM6E4E")
> 
> What is the point of having identical _HID and _CID? Is that a Windows thing?
> 
>> +      Name (_UID, 0x0)
>> +      Name (_CCA, 0x0)
>> +      Method (_STA)
>> +      {
>> +        Return (0xf)
>> +      }
>> +      Method (_CRS, 0x0, Serialized)
>> +      {
>> +        Name (RBUF, ResourceTemplate ()
>> +        {
>> +          Memory32Fixed (ReadWrite, 0xfd580000, 0x10000, )
> 
> We could grab that base address from a PCD as well, no?

Well, we don't seem to be using PCDs in any of the .asl's we have (we 
only do so on .aslc) and, also, that couldn't really be done without the 
previous series having been integrated.

I don't mind going for a PCD here, but do you really want to do that 
just for ETH0? It'll just look weird because it's going to be the only 
.asl address that we'll pick from a PCD.

If we're going to move to using PCDs in .asl's, I'd rather do that 
across the board and craft a patch that addresses all, including 
Uart.asl and so on... Else, it's a bit inconsistent as far as I'm concerned.

>> +          Interrupt (ResourceConsumer, Level, ActiveHigh, Exclusive) { 0xBD }
>> +          Interrupt (ResourceConsumer, Level, ActiveHigh, Exclusive) { 0xBE }
> 
> Why not
> 
> Interrupt (ResourceConsumer, Level, ActiveHigh, Exclusive) { 0xBD, 0xBE }

Likewise: Consistency (be it ill placed or not).

We've simply followed what GPIO had here.

Again, I don't mind making the .asl cleaner, but if we're going to do 
so, I'd rather not be piecemeal about it, and do that in a separate 
patch that harmonises the whole thing.

If you're okay with that approach, and not require a v2 for this, I can 
send a separate patch that cleans up the .asl's and uses PCDs early next 
week.

Regards,

/Pete

> 
> ?
> 
>> +        })
>> +        Return (RBUF)
>> +      }
>> +      Name (_DSD, Package () {
>> +        ToUUID("daffd814-6eba-4d8c-8a91-bc9bbf4aa301"),
>> +        Package () {
>> +          Package () { "brcm,max-dma-burst-size", 0x08 },
>> +          Package () { "phy-mode", "rgmii" },
>> +        }
>> +      })
>> +    }
>> +
>>       // Description: I2C
>>       Device (I2C1)
>>       {
>> --
>> 2.21.0.windows.1
>>