On Friday 24 July 2020 15:29:30 Pali Rohár wrote:
> Reset gpio is optional and it does not have to be defined for all boards.
>
> So in mvebu_pcie_powerdown() like in mvebu_pcie_powerup() check that reset
> gpio is defined prior usage to prevent NULL pointer dereference.
>
> Signed-off-by: Pali Rohár <pali@xxxxxxxxxx>
> ---
> drivers/pci/controller/pci-mvebu.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/pci/controller/pci-mvebu.c b/drivers/pci/controller/pci-mvebu.c
> index 153a64676bc9..58607cbe84c8 100644
> --- a/drivers/pci/controller/pci-mvebu.c
> +++ b/drivers/pci/controller/pci-mvebu.c
> @@ -947,7 +947,8 @@ static int mvebu_pcie_powerup(struct mvebu_pcie_port *port)
> */
> static void mvebu_pcie_powerdown(struct mvebu_pcie_port *port)
> {
> - gpiod_set_value_cansleep(port->reset_gpio, 1);
> + if (port->reset_gpio)
> + gpiod_set_value_cansleep(port->reset_gpio, 1);
Please drop this patch. I have realized that gpiod_set_value_cansleep()
calls VALIDATE_DESC_VOID() macro which returns from current running
function if passed pointer is NULL. So this patch is not needed as
gpiod_set_value_cansleep() may be called with NULL pointer.
>
> clk_disable_unprepare(port->clk);
> }
> --
> 2.20.1
>
