RE: [RESEND PATCH v2 02/11] net: dwc_eth_qos: Add option "snps,reset-gpio" phy-rst gpio for stm32

[Patrick DELAUNAY]

Hi David

> From: David Wu <david...@rock-chips.com>
> Sent: mardi 12 mai 2020 11:56
> 
> It can be seen that most of the Socs using STM mac, "snps,reset-gpio"
> gpio is used, adding this option makes reset function more general.
> 
> Signed-off-by: David Wu <david...@rock-chips.com>
> ---
> 
> Changes in v2:
> - Remove the code is not related (Patrice)
> 
>  drivers/net/dwc_eth_qos.c | 32 +++++++++++++++++++++++++++++++-
>  1 file changed, 31 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/dwc_eth_qos.c b/drivers/net/dwc_eth_qos.c index
> 66a02aa80b..92dab678c7 100644
> --- a/drivers/net/dwc_eth_qos.c
> +++ b/drivers/net/dwc_eth_qos.c
> @@ -314,6 +314,7 @@ struct eqos_priv {
>       struct eqos_tegra186_regs *tegra186_regs;
>       struct reset_ctl reset_ctl;
>       struct gpio_desc phy_reset_gpio;
> +     u32 reset_delays[3];
>       struct clk clk_master_bus;
>       struct clk clk_rx;
>       struct clk clk_ptp_ref;
> @@ -739,6 +740,15 @@ static int eqos_start_resets_stm32(struct udevice *dev)
> 
>       debug("%s(dev=%p):\n", __func__, dev);
>       if (dm_gpio_is_valid(&eqos->phy_reset_gpio)) {
> +             ret = dm_gpio_set_value(&eqos->phy_reset_gpio, 0);
> +             if (ret < 0) {
> +                     pr_err("dm_gpio_set_value(phy_reset, deassert) failed:
> %d",
> +                            ret);
> +                     return ret;
> +             }
> +
> +             udelay(eqos->reset_delays[0]);
> +
>               ret = dm_gpio_set_value(&eqos->phy_reset_gpio, 1);
>               if (ret < 0) {
>                       pr_err("dm_gpio_set_value(phy_reset, assert) failed: 
> %d",
> @@ -746,7 +756,7 @@ static int eqos_start_resets_stm32(struct udevice *dev)
>                       return ret;
>               }
> 
> -             udelay(2);
> +             udelay(eqos->reset_delays[1]);
> 
>               ret = dm_gpio_set_value(&eqos->phy_reset_gpio, 0);
>               if (ret < 0) {
> @@ -754,6 +764,8 @@ static int eqos_start_resets_stm32(struct udevice *dev)
>                              ret);
>                       return ret;
>               }
> +
> +             udelay(eqos->reset_delays[2]);
>       }
>       debug("%s: OK\n", __func__);
> 
> @@ -1864,11 +1876,29 @@ static int eqos_probe_resources_stm32(struct
> udevice *dev)
>               if (ret)
>                       pr_warn("gpio_request_by_name(phy reset) not provided
> %d",
>                               ret);
> +             else
> +                     eqos->reset_delays[1] = 2;
> 
>               eqos->phyaddr = ofnode_read_u32_default(phandle_args.node,
>                                                       "reg", -1);
>       }
> 
> +     if (!dm_gpio_is_valid(&eqos->phy_reset_gpio)) {
> +             int reset_flags = GPIOD_IS_OUT;
> +
> +             if (dev_read_bool(dev, "snps,reset-active-low"))
> +                     reset_flags |= GPIOD_ACTIVE_LOW;
> +
> +             ret = gpio_request_by_name(dev, "snps,reset-gpio", 0,
> +                                        &eqos->phy_reset_gpio, reset_flags);
> +             if (ret == 0)
> +                     ret = dev_read_u32_array(dev, "snps,reset-delays-us",
> +                                              eqos->reset_delays, 3);
> +             else
> +                     pr_warn("gpio_request_by_name(snps,reset-gpio) failed:
> %d",
> +                             ret);
> +     }
> +
>       debug("%s: OK\n", __func__);
>       return 0;
> 
> --
> 2.19.1
> 
> 

This obsolete binding isn't expected to be supported in stm32 glue for dwmac
(and it tis the purpose of eqos_stm32_config)

Reference in linux binding
./Documentation/devicetree/bindings/net/stm32-dwmac.txt (the glue)
./Documentation/devicetree/bindings/net/snps,dwmac.yaml

  snps,reset-gpio:
    deprecated: true

  snps,reset-active-low:
    deprecated: true

  snps,reset-delays-us:
    deprecated: true

I expected that gpio reset in future device tree should be managed by only by 
PHY generic binding 
(upstream in progress on Linux side for STM32MP15x), as described in:

Documentation/devicetree/bindings/net/ethernet-phy.yaml

  reset-gpios:
    maxItems: 1
    description:
      The GPIO phandle and specifier for the PHY reset signal.

  reset-assert-us:
    description:
      Delay after the reset was asserted in microseconds. If this
      property is missing the delay will be skipped.

  reset-deassert-us:
    description:
      Delay after the reset was deasserted in microseconds. If
      this property is missing the delay will be skipped.

See alsoU-Boot: doc/device-tree-bindings/net/phy.txt

Something as 

&mac {
        status = "okay";
        pinctrl-0       = <&ethernet_mii>;
        pinctrl-names   = "default";
        phy-mode        = "mii";
        phy-handle      = <&phy1>;
        mdio0 {
                #address-cells = <1>;
                #size-cells = <0>;
                compatible = "snps,dwmac-mdio";
                phy1: ethernet-phy@1 {
                        reg = <1>; 
+                       reset-assert-us = <10000>;
+                       reset-deassert-us = <30000>;
+                       reset-gpios = <&gpioh 15 GPIO_ACTIVE_LOW>;
                };
        };
};

And this binding "reset-gpios" is supported since the commit 5177b31ba6aa
("net: dwc_eth_qos: implement reset-gpios for stm32")

"reset-assert-us" and "reset-deassert-us" can be added if needed.

For witch product / compatible do you want add the support of this obsolete 
binding ?

Regards

Patrick