[PATCH v2 1/4] pci: iProc: define Broadcom iProc PCIe binding

[Ray Jui]

On 12/12/2014 4:14 AM, Arnd Bergmann wrote:
> On Thursday 11 December 2014 18:36:54 Ray Jui wrote:
>> index 0000000..040bc0f
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/pci/brcm,iproc-pcie.txt
>> @@ -0,0 +1,74 @@
>> +* Broadcom iProc PCIe controller
>> +
>> +Required properties:
>> +- compatible: Must be "brcm,iproc-pcie"
>> +- reg: base address and length of the PCIe controller and the MDIO interface
>> +  that controls the PCIe PHY
>> +- #interrupt-cells: set to <1>
>> +- interrupts: interrupt IDs
>
> How many, and what are they?
>
Different iProc SoCs might have different number of interrupts. I'll 
elaborate more on the next patch.

>> +- interrupt-map-mask and interrupt-map, standard PCI properties to define the
>> +  mapping of the PCIe interface to interrupt numbers
>> +- bus-range: PCI bus numbers covered
>> +- #address-cells: set to <3>
>> +- #size-cells: set to <2>
>> +- device_type: set to "pci"
>> +- ranges: ranges for the PCI memory and I/O regions
>> +- phy-addr: MDC/MDIO adddress of the PCIe PHY
>
> It looks like the phy controller is separate from the PCI controller,
> and you even list the same register range for both PHYs. Better make
> that a separate driver and put the phy address into the "phys" reference.
>
Okay. In this case, I need to create a separate PHY driver under the 
drivers/phy directory and have the PCIe host driver reference it through 
the standard PHY API.

>> +- have-msi-inten-reg: Required for legacy iProc PCIe controllers that need the
>> +  MSI interrupt enable register to be set explicitly
>> +
>> +The Broadcom iProc PCie driver adapts the multi-domain structure, i.e., each
>> +interface has its own domain and therefore has its own device node
>> +Example:
>> +
>> +SoC specific DT Entry:
>> +
>> +       pcie0: pcie at 18012000 {
>> +               compatible = "brcm,iproc-pcie";
>> +               reg = <0x18012000 0x1000>,
>> +                       <0x18002000 0x1000>;
>
> I guess the addresses should be relative to the BCMA bus, and this node
> get moved under that. Please see Hauke's patch series, we've discussed
> this in great length already.
>

As Arend van Spriel pointed out in the previous discussion:

BCMA core is the bus driver for discoverable ARM AXI interconnect. Apart 
from that it also provides drivers for some cores. For the chips to be 
discoverable it needs additional IP logic.

Not all iProc family of SoCs have the additional IP logic and for those 
which don't, they cannot use the BCMA bus.

>> +               #interrupt-cells = <1>;
>> +               interrupts = <GIC_SPI 96 IRQ_TYPE_NONE>,
>> +                            <GIC_SPI 97 IRQ_TYPE_NONE>,
>> +                            <GIC_SPI 98 IRQ_TYPE_NONE>,
>> +                            <GIC_SPI 99 IRQ_TYPE_NONE>,
>> +                            <GIC_SPI 100 IRQ_TYPE_NONE>,
>> +                            <GIC_SPI 101 IRQ_TYPE_NONE>;
>
>
>
>> +               interrupt-map-mask = <0 0 0 0>;
>> +               interrupt-map = <0 0 0 0 &gic GIC_SPI 100 IRQ_TYPE_NONE>;
>
> This interrupt is also listed in the "interrupts" above, which is
> probably a mistake, unless the IRQ line is shared between all PCI
> devices and the PCI host itself.
>
interrupts are for MSI interrupt support and interrupt-map is for legacy 
INTx support. To my best knowledge, MSI and INTx cannot be used at the 
same time. "nvidia,tegra20-pcie.txt" and "rcar-pci.txt" have similar 
configurations.

>> +               bus-range = <0x00 0xFF>;
>> +
>> +               #address-cells = >;
>> +               #size-cells = <2>;
>> +               device_type = "pci";
>> +               ranges = <0x81000000 0 0          0x28000000 0 0x00010000   /* downstream I/O */
>> +                         0x82000000 0 0x20000000 0x20000000 0 0x04000000>; /* non-prefetchable memory */
>> +               phy-addr = <5>;
>> +       };
>>
>
> 	Arnd
>