Re: [PATCH 4/5] PCI: Add quirk for multifunction devices of LS7A

[Bjorn Helgaas <helgaas@kernel.org>]

On Fri, May 14, 2021 at 04:00:24PM +0800, Huacai Chen wrote:
> From: Jianmin Lv <lvjianmin@loongson.cn>
> 
> In LS7A, multifunction device use same pci PIN and different
> irq for different function, so fix it for standard pci PIN
> usage.

Apparently the defect here is that the Interrupt Pin register reports
the wrong INTx values?

Will this be fixed in future hardware so we don't have to add more
devices to the quirk?

> Signed-off-by: Jianmin Lv <lvjianmin@loongson.cn>
> Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
> ---
>  drivers/pci/quirks.c | 17 +++++++++++++++++
>  1 file changed, 17 insertions(+)
> 
> diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> index 10b3b2057940..6ab4b3bba36b 100644
> --- a/drivers/pci/quirks.c
> +++ b/drivers/pci/quirks.c
> @@ -242,6 +242,23 @@ DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_LOONGSON,
>  DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_LOONGSON,
>  			DEV_LS7A_LPC, loongson_system_bus_quirk);
>  
> +static void loongson_pci_pin_quirk(struct pci_dev *dev)
> +{
> +	static const struct pci_device_id devids[] = {
> +		{ PCI_VDEVICE(LOONGSON, DEV_PCIE_PORT_0) },
> +		{ PCI_VDEVICE(LOONGSON, DEV_PCIE_PORT_1) },
> +		{ PCI_VDEVICE(LOONGSON, DEV_PCIE_PORT_2) },
> +		{ 0, },
> +	};
> +
> +	if (pci_match_id(devids, dev)) {
> +		u8 fun = dev->devfn & 7;

Use PCI_FUNC().

> +		dev->pin = 1 + (fun & 3);
> +	}
> +}
> +DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_LOONGSON, PCI_ANY_ID, loongson_pci_pin_quirk);

The usual pattern is to list each device here instead of using
pci_match_id() in the quirk, e.g.,

  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_LOONGSON, 0x7a09, loongson_pci_pin_quirk);
  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_LOONGSON, 0x7a19, loongson_pci_pin_quirk);
  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_LOONGSON, 0x7a29, loongson_pci_pin_quirk);

>  static void loongson_mrrs_quirk(struct pci_dev *dev)
>  {
>  	struct pci_bus *bus = dev->bus;
> -- 
> 2.27.0
>