Re: [PATCH v5 06/12] net: stmmac: Add glue layer for StarFive JH7110 SoC

From: Emil Renner Berthing
Date: Mon Mar 06 2023 - 07:21:55 EST

On Mon, 6 Mar 2023 at 08:16, Guo Samin <samin.guo@xxxxxxxxxxxxxxxx> wrote:
> 在 2023/3/4 0:18:20, Emil Renner Berthing 写道:
> > On Fri, 3 Mar 2023 at 10:01, Samin Guo <samin.guo@xxxxxxxxxxxxxxxx> wrote:
> >> This adds StarFive dwmac driver support on the StarFive JH7110 SoC.
> >>
> >> Co-developed-by: Emil Renner Berthing <kernel@xxxxxxxx>
> >> Signed-off-by: Emil Renner Berthing <kernel@xxxxxxxx>
> >> Signed-off-by: Samin Guo <samin.guo@xxxxxxxxxxxxxxxx>
> >> ---
> >> MAINTAINERS | 1 +
> >> drivers/net/ethernet/stmicro/stmmac/Kconfig | 12 ++
> >> drivers/net/ethernet/stmicro/stmmac/Makefile | 1 +
> >> .../ethernet/stmicro/stmmac/dwmac-starfive.c | 125 ++++++++++++++++++
> >> 4 files changed, 139 insertions(+)
> >> create mode 100644 drivers/net/ethernet/stmicro/stmmac/dwmac-starfive.c
> >>
> >> diff --git a/MAINTAINERS b/MAINTAINERS
> >> index 4e236b7c7fd2..91a4f190c827 100644
> >> --- a/MAINTAINERS
> >> +++ b/MAINTAINERS
> >> @@ -19916,6 +19916,7 @@ STARFIVE DWMAC GLUE LAYER
> >> M: Emil Renner Berthing <kernel@xxxxxxxx>
> >> M: Samin Guo <samin.guo@xxxxxxxxxxxxxxxx>
> >> S: Maintained
> >> +F: Documentation/devicetree/bindings/net/dwmac-starfive.c
> >> F: Documentation/devicetree/bindings/net/starfive,jh7110-dwmac.yaml
> >>
> >> STARFIVE JH71X0 CLOCK DRIVERS
> >> diff --git a/drivers/net/ethernet/stmicro/stmmac/Kconfig b/drivers/net/ethernet/stmicro/stmmac/Kconfig
> >> index f77511fe4e87..47fbccef9d04 100644
> >> --- a/drivers/net/ethernet/stmicro/stmmac/Kconfig
> >> +++ b/drivers/net/ethernet/stmicro/stmmac/Kconfig
> >> @@ -165,6 +165,18 @@ config DWMAC_SOCFPGA
> >> for the stmmac device driver. This driver is used for
> >> arria5 and cyclone5 FPGA SoCs.
> >>
> >> +config DWMAC_STARFIVE
> >> + tristate "StarFive dwmac support"
> >> + depends on OF && (ARCH_STARFIVE || COMPILE_TEST)
> >
> > There is an extra space between "OF" and "&&" here.
> >
> will drop it
> >
> >> + depends on STMMAC_ETH
> >
> > It's not visible in this patch context, but this whole config option
> > is surrounded by "if STMMAC_ETH" and "if STMMAC_PLATFORM", so "depends
> > on STMMAC_ETH" should not be needed.
> >
> will drop it.
> >> + default ARCH_STARFIVE
> >
> > This driver is not required to boot the JH7110, so we should just
> > default to building it as a module. Eg.
> > default m if ARCH_STARFIVE
>
> Yes, this driver is not required to boot the JH7110, but the network is a very basic module,
> it seems that other dwmac-platforms have been compiled into the kernel instead of modules.

Right, but the defconfig should work on as many platforms as possible,
so if we build in every "basic" module for every platform the kernel
will be huge and waste a lot of memory on drivers that will never be
used.

Also even if this driver was built in the gmac0 would still not work
until the driver for the AON CRG is loaded, which also defaults to m
for the same reasons.

> >
> >> + help
> >> + Support for ethernet controllers on StarFive RISC-V SoCs
> >> +
> >> + This selects the StarFive platform specific glue layer support for
> >> + the stmmac device driver. This driver is used for StarFive JH7110
> >> + ethernet controller.
> >> +
> >> config DWMAC_STI
> >> tristate "STi GMAC support"
> >> default ARCH_STI
> >> diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
> >> index 057e4bab5c08..8738fdbb4b2d 100644
> >> --- a/drivers/net/ethernet/stmicro/stmmac/Makefile
> >> +++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
> >> @@ -23,6 +23,7 @@ obj-$(CONFIG_DWMAC_OXNAS) += dwmac-oxnas.o
> >> obj-$(CONFIG_DWMAC_QCOM_ETHQOS) += dwmac-qcom-ethqos.o
> >> obj-$(CONFIG_DWMAC_ROCKCHIP) += dwmac-rk.o
> >> obj-$(CONFIG_DWMAC_SOCFPGA) += dwmac-altr-socfpga.o
> >> +obj-$(CONFIG_DWMAC_STARFIVE) += dwmac-starfive.o
> >> obj-$(CONFIG_DWMAC_STI) += dwmac-sti.o
> >> obj-$(CONFIG_DWMAC_STM32) += dwmac-stm32.o
> >> obj-$(CONFIG_DWMAC_SUNXI) += dwmac-sunxi.o
> >> diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-starfive.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-starfive.c
> >> new file mode 100644
> >> index 000000000000..566378306f67
> >> --- /dev/null
> >> +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-starfive.c
> >> @@ -0,0 +1,125 @@
> >> +// SPDX-License-Identifier: GPL-2.0+
> >> +/*
> >> + * StarFive DWMAC platform driver
> >> + *
> >> + * Copyright (C) 2022 StarFive Technology Co., Ltd.
> >> + * Copyright (C) 2022 Emil Renner Berthing <kernel@xxxxxxxx>
> >
> > Sorry, after looking at my old git branches where this started as a
> > driver for the JH7100 this should really be
> > * Copyright (C) 2021 Emil Renner Berthing <kernel@xxxxxxxx>
> > * Copyright (C) 2022 StarFive Technology Co., Ltd.
> >
> OK, It should be.
> >> + */
> >> +
> >> +#include <linux/of_device.h>
> >> +
> >> +#include "stmmac_platform.h"
> >> +
> >> +struct starfive_dwmac {
> >> + struct device *dev;
> >> + struct clk *clk_tx;
> >> + struct clk *clk_gtx;
> >
> > This pointer is only set, but never read. Please remove it.
> >>
> >> + bool tx_use_rgmii_rxin_clk;
> >> +};
> >> +
> >> +static void starfive_eth_fix_mac_speed(void *priv, unsigned int speed)
> >
> > This should be starfive_dwmac_fix_mac_speed for consistency.
> >
> Sorry，I missed this, will fix next version.
> >> +{
> >> + struct starfive_dwmac *dwmac = priv;
> >> + unsigned long rate;
> >> + int err;
> >> +
> >> + /* Generally, the rgmii_tx clock is provided by the internal clock,
> >> + * which needs to match the corresponding clock frequency according
> >> + * to different speeds. If the rgmii_tx clock is provided by the
> >> + * external rgmii_rxin, there is no need to configure the clock
> >> + * internally, because rgmii_rxin will be adaptively adjusted.
> >> + */
> >> + if (dwmac->tx_use_rgmii_rxin_clk)
> >> + return;
> >
> > If this function is only needed in certain situations, why not just
> > set the plat_dat->fix_mac_speed callback when it is needed?
> >
> Sounds good idea.
> >> + switch (speed) {
> >> + case SPEED_1000:
> >> + rate = 125000000;
> >> + break;
> >> + case SPEED_100:
> >> + rate = 25000000;
> >> + break;
> >> + case SPEED_10:
> >> + rate = 2500000;
> >> + break;
> >> + default:
> >> + dev_err(dwmac->dev, "invalid speed %u\n", speed);
> >> + break;
> >> + }
> >> +
> >> + err = clk_set_rate(dwmac->clk_tx, rate);
> >> + if (err)
> >> + dev_err(dwmac->dev, "failed to set tx rate %lu\n", rate);
> >> +}
> >> +
> >> +static int starfive_dwmac_probe(struct platform_device *pdev)
> >> +{
> >> + struct plat_stmmacenet_data *plat_dat;
> >> + struct stmmac_resources stmmac_res;
> >> + struct starfive_dwmac *dwmac;
> >> + int err;
> >> +
> >> + err = stmmac_get_platform_resources(pdev, &stmmac_res);
> >> + if (err)
> >> + return err;
> >> +
> >> + plat_dat = stmmac_probe_config_dt(pdev, stmmac_res.mac);
> >> + if (IS_ERR(plat_dat)) {
> >> + dev_err(&pdev->dev, "dt configuration failed\n");
> >> + return PTR_ERR(plat_dat);
> >> + }
> >> +
> >> + dwmac = devm_kzalloc(&pdev->dev, sizeof(*dwmac), GFP_KERNEL);
> >> + if (!dwmac)
> >> + return -ENOMEM;
> >> +
> >> + dwmac->clk_tx = devm_clk_get_enabled(&pdev->dev, "tx");
> >> + if (IS_ERR(dwmac->clk_tx))
> >> + return dev_err_probe(&pdev->dev, PTR_ERR(dwmac->clk_tx),
> >> + "error getting tx clock\n");
> >> +
> >> + dwmac->clk_gtx = devm_clk_get_enabled(&pdev->dev, "gtx");
> >> + if (IS_ERR(dwmac->clk_gtx))
> >> + return dev_err_probe(&pdev->dev, PTR_ERR(dwmac->clk_gtx),
> >> + "error getting gtx clock\n");
> >> +
> >> + if (device_property_read_bool(&pdev->dev, "starfive,tx-use-rgmii-clk"))
> >> + dwmac->tx_use_rgmii_rxin_clk = true;
> >> +
> >> + dwmac->dev = &pdev->dev;
> >> + plat_dat->fix_mac_speed = starfive_eth_fix_mac_speed;
> >
> > Eg.:
> > if (!device_property_read_bool(&pdev->dev, "starfive,tx_use_rgmii_clk"))
> > plat_dat->fix_mac_speed = starfive_dwmac_fix_mac_speed;
> >
> Good idea, so we can remove flag 'tx_use_rgmii_rxin_clk' in struct starfive_dwmac.
> >> + plat_dat->init = NULL;

Btw. plat_dat is initialized by kzalloc in stmmac_probe_config_dt and
I can't seem to find anything that sets plat_dat->init, so I think
this is redundant.

> >> + plat_dat->bsp_priv = dwmac;
> >> + plat_dat->dma_cfg->dche = true;
> >> +
> >> + err = stmmac_dvr_probe(&pdev->dev, plat_dat, &stmmac_res);
> >> + if (err) {
> >> + stmmac_remove_config_dt(pdev, plat_dat);
> >> + return err;
> >> + }
> >> +
> >> + return 0;
> >> +}
> >> +
> >> +static const struct of_device_id starfive_dwmac_match[] = {
> >> + { .compatible = "starfive,jh7110-dwmac" },
> >> + { /* sentinel */ }
> >> +};
> >> +MODULE_DEVICE_TABLE(of, starfive_dwmac_match);
> >> +
> >> +static struct platform_driver starfive_dwmac_driver = {
> >> + .probe = starfive_dwmac_probe,
> >> + .remove = stmmac_pltfr_remove,
> >> + .driver = {
> >> + .name = "starfive-dwmac",
> >> + .pm = &stmmac_pltfr_pm_ops,
> >> + .of_match_table = starfive_dwmac_match,
> >> + },
> >> +};
> >> +module_platform_driver(starfive_dwmac_driver);
> >> +
> >> +MODULE_LICENSE("GPL");
> >> +MODULE_DESCRIPTION("StarFive DWMAC platform driver");
> >> +MODULE_AUTHOR("Emil Renner Berthing <kernel@xxxxxxxx>");
> >> +MODULE_AUTHOR("Samin Guo <samin.guo@xxxxxxxxxxxxxxxx>");
> >> --
> >> 2.17.1
> >>
> >>
> >> _______________________________________________
> >> linux-riscv mailing list
> >> linux-riscv@xxxxxxxxxxxxxxxxxxx
> >> http://lists.infradead.org/mailman/listinfo/linux-riscv
> Best regards,
> Samin
>
> --
> Best regards,
> Samin