Re: [PATCH v6 5/6] PCI: fu740: Add SiFive FU740 PCIe host controller driver

[Bjorn Helgaas <helgaas@kernel.org>]

On Tue, May 04, 2021 at 06:59:39PM +0800, Greentime Hu wrote:
> From: Paul Walmsley <paul.walmsley@sifive.com>
> 
> Add driver for the SiFive FU740 PCIe host controller.
> This controller is based on the DesignWare PCIe core.
> 
> Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>
> Co-developed-by: Henry Styles <hes@sifive.com>
> Signed-off-by: Henry Styles <hes@sifive.com>
> Co-developed-by: Erik Danie <erik.danie@sifive.com>
> Signed-off-by: Erik Danie <erik.danie@sifive.com>
> Co-developed-by: Greentime Hu <greentime.hu@sifive.com>
> Signed-off-by: Greentime Hu <greentime.hu@sifive.com>
> ---
>  drivers/pci/controller/dwc/Kconfig      |  10 +
>  drivers/pci/controller/dwc/Makefile     |   1 +
>  drivers/pci/controller/dwc/pcie-fu740.c | 309 ++++++++++++++++++++++++
>  3 files changed, 320 insertions(+)
>  create mode 100644 drivers/pci/controller/dwc/pcie-fu740.c
> 
> diff --git a/drivers/pci/controller/dwc/Kconfig b/drivers/pci/controller/dwc/Kconfig
> index 22c5529e9a65..255d43b1661b 100644
> --- a/drivers/pci/controller/dwc/Kconfig
> +++ b/drivers/pci/controller/dwc/Kconfig
> @@ -318,4 +318,14 @@ config PCIE_AL
>  	  required only for DT-based platforms. ACPI platforms with the
>  	  Annapurna Labs PCIe controller don't need to enable this.
>  
> +config PCIE_FU740
> +	bool "SiFive FU740 PCIe host controller"
> +	depends on PCI_MSI_IRQ_DOMAIN
> +	depends on SOC_SIFIVE || COMPILE_TEST
> +	depends on GPIOLIB

1) I'm a little disappointed that I reported the build issue 6 days
   ago when we were already in the merge window, and it's taken until
   now to make some progress.

2) I would prefer not to depend on GPIOLIB because it reduces
   compile-test coverage.  For example, the x86_64 defconfig does not
   enable GPIOLIB, so one must manually enable it to even be able to
   enable PCIE_FU740.

   Many other PCI controller drivers use GPIO, but no others depend on
   GPIOLIB, so I infer that in the !GPIOLIB case, gpio/consumer.h
   provides the stubs required for compile testing.

   We could have a conversation about whether it's better to
   explicitly depend on GPIOLIB here, or whether building a working
   FU740 driver implicitly depends on GPIOLIB being selected
   elsewhere.  That implicit dependency *is* a little obscure, but I
   think that's what other drivers currently do, and I'd like to do
   this consistently unless there's a good reason otherwise.

   Here are some examples of other drivers:

   dwc/pci-dra7xx.c:
     config PCI_DRA7XX_HOST
       depends on SOC_DRA7XX || COMPILE_TEST

     config SOC_DRA7XX
       select ARCH_OMAP2PLUS

     config ARCH_OMAP2PLUS
       select GPIOLIB

   dwc/pci-meson.c:
     config PCI_MESON
       # doesn't, but probably *should* depend on "ARCH_MESON || COMPILE_TEST"

     menuconfig ARCH_MESON
       select GPIOLIB

   dwc/pcie-qcom.c:
     config PCIE_QCOM
       depends on OF && (ARCH_QCOM || COMPILE_TEST)

     config ARCH_QCOM
       select GPIOLIB

   pcie-rockchip.c:
     config PCIE_ROCKCHIP_HOST
       depends on ARCH_ROCKCHIP || COMPILE_TEST

     config ARCH_ROCKCHIP
       select GPIOLIB

> +	select PCIE_DW_HOST
> +	help
> +	  Say Y here if you want PCIe controller support for the SiFive
> +	  FU740.
> +
>  endmenu