wt., 12 lis 2019 o 15:10 Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> napisał(a):
>
> On Tue, Nov 12, 2019 at 11:11:35AM +0100, Bartosz Golaszewski wrote:
> > pt., 8 lis 2019 o 16:34 Thierry Reding <thierry.reding@xxxxxxxxx> napisał(a):
> > >
> > > From: Thierry Reding <treding@xxxxxxxxxx>
> > >
> > > The controls for the GG port on Tegra194 resides in the power partition
> > > of the C5 PCIe controller and its interrupt route mapping can therefore
> > > not be programmed by early boot firmware along with that of the other
> > > ports.
> > >
> > > Detect this generically by looking at which controls have already been
> > > locked down using the security registers and fill in default values for
> > > controls that are unlocked.
> > >
> > > Signed-off-by: Thierry Reding <treding@xxxxxxxxxx>
> > > ---
> > > drivers/gpio/gpio-tegra186.c | 46 ++++++++++++++++++++++++++++++++++++
> > > 1 file changed, 46 insertions(+)
> > >
> > > diff --git a/drivers/gpio/gpio-tegra186.c b/drivers/gpio/gpio-tegra186.c
> > > index 32714cefdfde..404ebb82bfa2 100644
> > > --- a/drivers/gpio/gpio-tegra186.c
> > > +++ b/drivers/gpio/gpio-tegra186.c
> > > @@ -15,6 +15,14 @@
> > > #include <dt-bindings/gpio/tegra186-gpio.h>
> > > #include <dt-bindings/gpio/tegra194-gpio.h>
> > >
> > > +/* security registers */
> > > +#define TEGRA186_GPIO_CTL_SCR 0x0c
> > > +#define TEGRA186_GPIO_CTL_SCR_SEC_WEN BIT(28)
> > > +#define TEGRA186_GPIO_CTL_SCR_SEC_REN BIT(27)
> > > +
> > > +#define TEGRA186_GPIO_INT_ROUTE_MAPPING(p, x) (0x14 + (p) * 0x20 + (x) * 4)
> > > +
> > > +/* control registers */
> > > #define TEGRA186_GPIO_ENABLE_CONFIG 0x00
> > > #define TEGRA186_GPIO_ENABLE_CONFIG_ENABLE BIT(0)
> > > #define TEGRA186_GPIO_ENABLE_CONFIG_OUT BIT(1)
> > > @@ -64,6 +72,7 @@ struct tegra_gpio {
> > >
> > > const struct tegra_gpio_soc *soc;
> > >
> > > + void __iomem *secure;
> > > void __iomem *base;
> > > };
> > >
> > > @@ -449,6 +458,37 @@ static const struct of_device_id tegra186_pmc_of_match[] = {
> > > { /* sentinel */ }
> > > };
> > >
> > > +static void tegra186_gpio_init_route_mapping(struct tegra_gpio *gpio)
> > > +{
> > > + unsigned int i, j;
> > > + u32 value;
> > > +
> > > + for (i = 0; i < gpio->soc->num_ports; i++) {
> > > + const struct tegra_gpio_port *port = &gpio->soc->ports[i];
> > > + unsigned int offset, p = port->port;
> > > + void __iomem *base;
> > > +
> > > + base = gpio->secure + port->bank * 0x1000 + 0x800;
> > > +
> > > + value = readl(base + TEGRA186_GPIO_CTL_SCR);
> > > +
> > > + /*
> > > + * For controllers that haven't been locked down yet, make
> > > + * sure to program the default interrupt route mapping.
> > > + */
> > > + if ((value & TEGRA186_GPIO_CTL_SCR_SEC_REN) == 0 &&
> > > + (value & TEGRA186_GPIO_CTL_SCR_SEC_WEN) == 0) {
> > > + for (j = 0; j < 8; j++) {
> > > + offset = TEGRA186_GPIO_INT_ROUTE_MAPPING(p, j);
> > > +
> > > + value = readl(base + offset);
> > > + value = BIT(port->pins) - 1;
> > > + writel(value, base + offset);
> > > + }
> > > + }
> > > + }
> > > +}
> > > +
> > > static int tegra186_gpio_probe(struct platform_device *pdev)
> > > {
> > > unsigned int i, j, offset;
> > > @@ -464,6 +504,10 @@ static int tegra186_gpio_probe(struct platform_device *pdev)
> > >
> > > gpio->soc = of_device_get_match_data(&pdev->dev);
> > >
> > > + gpio->secure = devm_platform_ioremap_resource_byname(pdev, "security");
> > > + if (IS_ERR(gpio->secure))
> > > + return PTR_ERR(gpio->secure);
> > > +
> > > gpio->base = devm_platform_ioremap_resource_byname(pdev, "gpio");
> > > if (IS_ERR(gpio->base))
> > > return PTR_ERR(gpio->base);
> > > @@ -558,6 +602,8 @@ static int tegra186_gpio_probe(struct platform_device *pdev)
> > > return -EPROBE_DEFER;
> > > }
> > >
> > > + tegra186_gpio_init_route_mapping(gpio);
> > > +
> > > irq->map = devm_kcalloc(&pdev->dev, gpio->gpio.ngpio,
> > > sizeof(*irq->map), GFP_KERNEL);
> > > if (!irq->map)
> > > --
> > > 2.23.0
> > >
> >
> > This doesn't apply without a patch that went through Greg's driver-core tree.
> >
> > Greg: can you provide us with an immutable branch which we can merge
> > into the gpio for-next tree?
>
> A branch for/called what?
>
> You can always pull from my driver-core-next branch, I never rebase it.
>
That works for me - thanks!
Bart
> thanks,
>
> greg k-h
