[PATCH] net: stmmac: dwmac-qcom-ethqos: Enable RGMII functional clock on resume

[Bhupesh Sharma]

+Cc: stable tree as I think this is an important fix for stmmac
dwmac-qcom-ethernet driver and affects ethernet functionality on QCOM
boards which use this driver.

More below..

On Wed, 23 Mar 2022 at 09:01, Bjorn Andersson
<bjorn.andersson at linaro.org> wrote:
>
> When the Qualcomm ethqos driver is properly described in its associated
> GDSC power-domain, the hardware will be powered down and loose its state
> between qcom_ethqos_probe() and stmmac_init_dma_engine().
>
> The result of this is that the functional clock from the RGMII IO macro
> is no longer provides and the DMA software reset in dwmac4_dma_reset()
> will time out, due to lacking clock signal.
>
> Re-enable the functional clock, as part of the Qualcomm specific clock
> enablement sequence to avoid this problem.
>
> The final clock configuration will be adjusted by ethqos_fix_mac_speed()
> once the link is being brought up.
>
> Fixes: a7c30e62d4b8 ("net: stmmac: Add driver for Qualcomm ethqos")
> Signed-off-by: Bjorn Andersson <bjorn.andersson at linaro.org>
> ---
>  drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c | 7 +++++++
>  1 file changed, 7 insertions(+)
>
> diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
> index 0cc28c79cc61..835caa15d55f 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-qcom-ethqos.c
> @@ -487,6 +487,13 @@ static int ethqos_clks_config(void *priv, bool enabled)
>                         dev_err(&ethqos->pdev->dev, "rgmii_clk enable failed\n");
>                         return ret;
>                 }
> +
> +               /* Enable functional clock to prevent DMA reset to timeout due
> +                * to lacking PHY clock after the hardware block has been power
> +                * cycled. The actual configuration will be adjusted once
> +                * ethqos_fix_mac_speed() is invoked.
> +                */
> +               ethqos_set_func_clk_en(ethqos);
>         } else {
>                 clk_disable_unprepare(ethqos->rgmii_clk);
>         }
> --
> 2.33.1

Thanks for the catch, Bjorn. I tested this on the SA8155p-ADP board
and the eth interface can be moved from 'on' to 'off' state and
vice-versa properly after this change and we no longer need the EMAC
GDSC quirk, so:

Tested-and-Reviewed-by: Bhupesh Sharma <bhupesh.sharma at linaro.org>

Regards.