On Wed, Sep 5, 2018 at 7:29 AM Jean-Jacques Hiblot jjhiblot@ti.com wrote:
Adam,
On 05/09/2018 11:35, Adam Ford wrote:
The gpio driver doesn't current support knowing whether or not GPIO is active low or high. It simply returns the value. The side effect of this, is that the MMC routines which check the status card detect or write protect must use a u-boot specific dts modification to enable inverting the pin values when GPIO_ACTIVE_LOW is used. This allows the invert option to be removed
Signed-off-by: Adam Ford aford173@gmail.com
diff --git a/drivers/gpio/omap_gpio.c b/drivers/gpio/omap_gpio.c index 555eba2662..0ecd2f374a 100644 --- a/drivers/gpio/omap_gpio.c +++ b/drivers/gpio/omap_gpio.c @@ -24,6 +24,7 @@ #include <asm/io.h> #include <linux/errno.h> #include <malloc.h> +#include <dt-bindings/gpio/gpio.h>
DECLARE_GLOBAL_DATA_PTR;
@@ -275,12 +276,25 @@ static int omap_gpio_get_function(struct udevice *dev, unsigned offset) return GPIOF_INPUT; }
+static int omap_gpio_xlate(struct udevice *dev, struct gpio_desc *desc,
struct ofnode_phandle_args *args)+{
desc->offset = args->args[0];if (args->args[1] & GPIO_ACTIVE_LOW)desc->flags = GPIOD_ACTIVE_LOW;elsedesc->flags = 0;return 0;+}
Do we need this ? It looks a lot like the default behaviour (see gpio_xlate_offs_flags in gpio-uclass.c).
I don't know. I saw that other gpio drivers had the xlate, so it seemed like it made sense to have it here. I'll revert it, and keep the other stuff and see what happens along with removing the entry from the struct. I don't have the other (non-Logic PD) boards, so I cannot test them.
adam
JJ
Do we need this ? It looks like the default behaviour static const struct dm_gpio_ops gpio_omap_ops = { .direction_input = omap_gpio_direction_input, .direction_output = omap_gpio_direction_output, .get_value = omap_gpio_get_value, .set_value = omap_gpio_set_value, .get_function = omap_gpio_get_function,
.xlate = omap_gpio_xlate,};
static int omap_gpio_probe(struct udevice *dev)