[PATCH 17/20] ARM: Exynos: Add device tree support for gpio wakeup interrupt controller
Grant Likely
grant.likely at secretlab.ca
Tue May 15 12:35:21 EDT 2012
On Mon, 30 Apr 2012 12:14:27 -0700, Thomas Abraham <thomas.abraham at linaro.org> wrote:
> Add device tree support for gpio wakeup source interrupt controller
> for Exynos platforms.
>
> Cc: Rob Herring <rob.herring at calxeda.com>
> Cc: Grant Likely <grant.likely at secretlab.ca>
> Signed-off-by: Thomas Abraham <thomas.abraham at linaro.org>
> ---
> arch/arm/mach-exynos/common.c | 92 +++++++++++++++++++++++++++++++++++------
> 1 files changed, 79 insertions(+), 13 deletions(-)
>
> diff --git a/arch/arm/mach-exynos/common.c b/arch/arm/mach-exynos/common.c
> index 544f8b5..074508f 100644
> --- a/arch/arm/mach-exynos/common.c
> +++ b/arch/arm/mach-exynos/common.c
> @@ -64,7 +64,8 @@ static void exynos4_init_clocks(int xtal);
> static void exynos5_init_clocks(int xtal);
> static void exynos_init_uarts(struct s3c2410_uartcfg *cfg, int no);
> static int exynos_init(void);
> -static int exynos_init_irq_eint(void);
> +static int exynos_init_irq_eint(struct device_node *np,
> + struct device_node *parent);
>
> static struct cpu_table cpu_ids[] __initdata = {
> {
> @@ -589,6 +590,8 @@ static const struct of_device_id exynos4_dt_irq_match[] = {
> { .compatible = "arm,cortex-a9-gic", .data = gic_of_init, },
> { .compatible = "samsung,exynos4210-combiner",
> .data = combiner_of_init, },
> + { .compatible = "samsung,exynos5210-wakeup-eint-map",
> + .data = exynos_init_irq_eint, },
> {},
> };
> #endif
> @@ -606,8 +609,10 @@ void __init exynos4_init_irq(void)
> of_irq_init(exynos4_dt_irq_match);
> #endif
>
> - if (!of_have_populated_dt())
> + if (!of_have_populated_dt()) {
> combiner_init(S5P_VA_COMBINER_BASE, NULL);
> + exynos_init_irq_eint(NULL, NULL);
> + }
>
> /*
> * The parameters of s5p_init_irq() are for VIC init.
> @@ -615,7 +620,6 @@ void __init exynos4_init_irq(void)
> * uses GIC instead of VIC.
> */
> s5p_init_irq(NULL, 0);
> - exynos_init_irq_eint();
> }
>
> void __init exynos5_init_irq(void)
> @@ -629,7 +633,6 @@ void __init exynos5_init_irq(void)
> * uses GIC instead of VIC.
> */
> s5p_init_irq(NULL, 0);
> - exynos_init_irq_eint();
> }
>
> struct bus_type exynos_subsys = {
> @@ -1002,17 +1005,72 @@ static int exynos_eint_irq_domain_map(struct irq_domain *d, unsigned int irq,
> return 0;
> }
>
> +#ifdef CONFIG_OF
> +static int exynos_eint_irq_domain_xlate(struct irq_domain *d,
> + struct device_node *controller, const u32 *intspec,
> + unsigned int intsize, unsigned long *out_hwirq,
> + unsigned int *out_type)
> +{
> + if (d->of_node != controller)
> + return -EINVAL;
> + if (intsize < 2)
> + return -EINVAL;
> + *out_hwirq = intspec[0];
> +
> + switch (intspec[1]) {
> + case S5P_IRQ_TYPE_LEVEL_LOW:
> + *out_type = IRQ_TYPE_LEVEL_LOW;
> + break;
> + case S5P_IRQ_TYPE_LEVEL_HIGH:
> + *out_type = IRQ_TYPE_LEVEL_HIGH;
> + break;
> + case S5P_IRQ_TYPE_EDGE_FALLING:
> + *out_type = IRQ_TYPE_EDGE_FALLING;
> + break;
> + case S5P_IRQ_TYPE_EDGE_RISING:
> + *out_type = IRQ_TYPE_EDGE_RISING;
> + break;
> + case S5P_IRQ_TYPE_EDGE_BOTH:
> + *out_type = IRQ_TYPE_EDGE_BOTH;
> + break;
> + };
> +
> + return 0;
I'm recommending that new irq bindings use the Linux IRQ_TYPE_* values
just because it provides some commonality between irq controllers.
Using the S5P_IRQ_TYPE_ values doesn't make much sense when it only
gets immediately converted to another set of arbitrary values.
> +}
> +#else
> +static int exynos_eint_irq_domain_xlate(struct irq_domain *d,
> + struct device_node *controller, const u32 *intspec,
> + unsigned int intsize, unsigned long *out_hwirq,
> + unsigned int *out_type)
> +{
> + return -EINVAL;
> +}
> +#endif
> +
> static struct irq_domain_ops exynos_eint_irq_domain_ops = {
> + .xlate = exynos_eint_irq_domain_xlate,
> .map = exynos_eint_irq_domain_map,
> };
>
> -static int __init exynos_init_irq_eint(void)
> +static int __init exynos_init_irq_eint(struct device_node *eint_np,
> + struct device_node *parent)
> {
> - int irq, *src_int, irq_base;
> + int irq, *src_int, irq_base, irq_eint;
> unsigned int paddr;
> + static unsigned int retry = 0;
> + static struct device_node *np;
>
> - paddr = soc_is_exynos5250() ? EXYNOS5_PA_GPIO1 : EXYNOS4_PA_GPIO2;
> - exynos_eint_base = ioremap(paddr, SZ_4K);
> + if (retry)
> + goto retry_init;
> +
> + if (!eint_np) {
> + paddr = soc_is_exynos5250() ? EXYNOS5_PA_GPIO1 :
> + EXYNOS4_PA_GPIO2;
> + exynos_eint_base = ioremap(paddr, SZ_4K);
> + } else {
> + np = of_get_parent(eint_np);
> + exynos_eint_base = of_iomap(np, 0);
> + }
> if (!exynos_eint_base) {
> pr_err("unable to ioremap for EINT base address\n");
> return -ENXIO;
> @@ -1025,21 +1083,29 @@ static int __init exynos_init_irq_eint(void)
> "linux irq base\n", __func__, irq_base);
> }
>
> - irq_domain = irq_domain_add_legacy(NULL, EXYNOS_EINT_NR, irq_base, 0,
> + irq_domain = irq_domain_add_legacy(np, EXYNOS_EINT_NR, irq_base, 0,
> &exynos_eint_irq_domain_ops, NULL);
> if (WARN_ON(!irq_domain)) {
> pr_warning("%s: irq domain init failed\n", __func__);
> return 0;
> }
>
> - irq_set_chained_handler(EXYNOS_IRQ_EINT16_31, exynos_irq_demux_eint16_31);
> + irq_eint = eint_np ? irq_of_parse_and_map(np, 16) : EXYNOS_IRQ_EINT16_31;
> + irq_set_chained_handler(irq_eint, exynos_irq_demux_eint16_31);
>
> - for (irq = 0 ; irq <= 15; irq++) {
> +retry_init:
> + for (irq = 0; irq <= 15; irq++) {
> eint0_15_data[irq] = irq;
> src_int = soc_is_exynos5250() ? exynos5_eint0_15_src_int :
> exynos4_eint0_15_src_int;
> - irq_set_handler_data(src_int[irq], &eint0_15_data[irq]);
> - irq_set_chained_handler(src_int[irq], exynos_irq_eint0_15);
> + irq_eint = eint_np ? irq_of_parse_and_map(np, irq) : src_int[irq];
> + if (!irq_eint) {
> + of_node_put(np);
> + retry = 1;
> + return -EAGAIN;
> + }
> + irq_set_handler_data(irq_eint, &eint0_15_data[irq]);
> + irq_set_chained_handler(irq_eint, exynos_irq_eint0_15);
After switching to irq_domain, all of the contents of this loop should
be in the .map hook.
g.
More information about the linux-arm-kernel
mailing list