[PATCH v2 2/5] dt-bindings: net: Add Broadcom BCM4377 family PCIe Bluetooth

Sven Peter sven at svenpeter.dev
Mon Sep 19 08:15:54 PDT 2022



On Thu, Sep 15, 2022, at 15:09, Rob Herring wrote:
> On Mon, Sep 12, 2022 at 4:12 PM Rob Herring <robh at kernel.org> wrote:
>>
>> On Thu, Sep 08, 2022 at 01:19:17PM +0200, Krzysztof Kozlowski wrote:
>> > On 07/09/2022 19:09, Sven Peter wrote:
>> > > These chips are combined Wi-Fi/Bluetooth radios which expose a
>> > > PCI subfunction for the Bluetooth part.
>> > > They are found in Apple machines such as the x86 models with the T2
>> > > chip or the arm64 models with the M1 or M2 chips.
>> > >
>> > > Signed-off-by: Sven Peter <sven at svenpeter.dev>
>> > > ---
>
>> > > +examples:
>> > > +  - |
>> > > +    pcie {
>> > > +      #address-cells = <3>;
>> > > +      #size-cells = <2>;
>> > > +
>> > > +      bluetooth at 0,1 {
>> >
>> > The unit address seems to be different than reg.
>>
>> Right, this says dev 0, func 1.
>
> Actually, the reg value of 0x100 is correct. func is bits 8-10. dev
> starts in bit 11.

Yup, if I write the example as

  - |
    pcie at a0000000 {
      #address-cells = <3>;
      #size-cells = <2>;
      reg = <0xa0000000 0x1000000>;
      device_type = "pci";
      ranges = <0x43000000 0x6 0xa0000000 0xa0000000 0x0 0x20000000>;

      bluetooth at 0,1 {
        compatible = "pci14e4,5f69";
        reg = <0x100 0x0 0x0 0x0 0x0>;
        brcm,board-type = "apple,honshu";
        /* To be filled by the bootloader */
        local-bd-address = [00 00 00 00 00 00];
      };
    };


then no warnings appear. If I instead use "bluetooth at 0,2" I get the following
warning:

Warning (pci_device_reg): /example-0/pcie at a0000000/bluetooth at 0,2: PCI unit address format error, expected "0,1"



Sven



More information about the linux-arm-kernel mailing list