On Mon, Mar 7, 2022 at 8:29 PM Miaoqian Lin <linmq006@xxxxxxxxx> wrote:
>
> The device_node pointer is returned by of_parse_phandle() with refcount
> incremented. We should use of_node_put() on it when done.
>
> Fixes: 1df82ec46600 ("PCI: imx: Add workaround for e10728, IMX7d PCIe PLL failure")
> Signed-off-by: Miaoqian Lin <linmq006@xxxxxxxxx>
> ---
> drivers/pci/controller/dwc/pci-imx6.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/pci/controller/dwc/pci-imx6.c b/drivers/pci/controller/dwc/pci-imx6.c
> index 6974bd5aa116..bcc338ab6c11 100644
> --- a/drivers/pci/controller/dwc/pci-imx6.c
> +++ b/drivers/pci/controller/dwc/pci-imx6.c
> @@ -1050,6 +1050,7 @@ static int imx6_pcie_probe(struct platform_device *pdev)
> struct resource res;
>
> ret = of_address_to_resource(np, 0, &res);
> + of_node_put(np);
So now you are using the phy within the driver, but not holding a
reference to its node? That is wrong. Though if a struct device was
created for the phy's node, it would hold a reference. There are a
bunch of error paths where the reference should be dropped. But
really, this driver is a big mess with the phy binding and framework
not used and your nodes probably aren't going to be disappearing on
you anyways.
Rob
