On 24/06/13 14:34, Grant Likely wrote:
> On Thu, 20 Jun 2013 10:26:28 +0100, James Hogan <james.hogan@xxxxxxxxxx> wrote:
>> diff --git a/Documentation/devicetree/bindings/gpio/gpio-tz1090.txt b/Documentation/devicetree/bindings/gpio/gpio-tz1090.txt
>> new file mode 100644
>> index 0000000..e017d4b
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/gpio/gpio-tz1090.txt
>> @@ -0,0 +1,87 @@
>> +ImgTec TZ1090 GPIO Controller
>> +
>> +Required properties:
>> +- compatible: Compatible property value should be "img,tz1090-gpio>".
>
> typo at end of line
Yes, I'll fix in gpio-tz1090-pdc driver bindings too
>> + Bank subnode optional properties:
>> + - gpio-ranges: Mapping to pin controller pins
>
> This is specific to this binding. To avoid namespace colisions, add a
> "img," prefix to the property name.
This property is described in
Documentation/devicetree/bindings/gpio/gpio.txt... (and my examples are
out of date from when the gpio offset cell was added in v3.10). I'll add
a reference to that Document.
>> +/* GPIO chip callbacks */
>> +
>> +static int tz1090_gpio_direction_input(struct gpio_chip *chip,
>> + unsigned int offset)
>> +{
>> + struct tz1090_gpio_bank *bank = to_bank(chip);
>> + tz1090_gpio_set_bit(bank, REG_GPIO_DIR, offset);
>> +
>> + return 0;
>> +}
>> +
>> +static int tz1090_gpio_direction_output(struct gpio_chip *chip,
>> + unsigned int offset, int output_value)
>> +{
>> + struct tz1090_gpio_bank *bank = to_bank(chip);
>> + int lstat;
>> +
>> + __global_lock2(lstat);
>> + _tz1090_gpio_mod_bit(bank, REG_GPIO_DOUT, offset, output_value);
>> + _tz1090_gpio_clear_bit(bank, REG_GPIO_DIR, offset);
>> + __global_unlock2(lstat);
>> +
>> + return 0;
>> +}
>> +
>> +/*
>> + * Return GPIO level
>> + */
>> +static int tz1090_gpio_get(struct gpio_chip *chip, unsigned int offset)
>> +{
>> + struct tz1090_gpio_bank *bank = to_bank(chip);
>> +
>> + return tz1090_gpio_read_bit(bank, REG_GPIO_DIN, offset);
>> +}
>> +
>> +/*
>> + * Set output GPIO level
>> + */
>> +static void tz1090_gpio_set(struct gpio_chip *chip, unsigned int offset,
>> + int output_value)
>> +{
>> + struct tz1090_gpio_bank *bank = to_bank(chip);
>> +
>> + tz1090_gpio_mod_bit(bank, REG_GPIO_DOUT, offset, output_value);
>> +}
>> +
>> +static int tz1090_gpio_request(struct gpio_chip *chip, unsigned int offset)
>> +{
>> + struct tz1090_gpio_bank *bank = to_bank(chip);
>> + int ret;
>> +
>> + ret = pinctrl_request_gpio(chip->base + offset);
>> + if (ret)
>> + return ret;
>> +
>> + tz1090_gpio_set_bit(bank, REG_GPIO_DIR, offset);
>> + tz1090_gpio_set_bit(bank, REG_GPIO_BIT_EN, offset);
>> +
>> + return 0;
>> +}
>
> Is it possible to use the gpio-generic.c hooks for manipulating the
> gpio bits?
Due to the unfortunate necessity to use the __global_lock2 functions
(for atomic accesses between different non-linux threads/cores) I don't
think this is possible.
>> +/* IRQ chip handlers */
>> +
>> +/* Get TZ1090 GPIO chip from irq data provided to generic IRQ callbacks */
>> +static inline struct tz1090_gpio_bank *irqd_to_gpio_bank(struct irq_data *data)
>> +{
>> + return (struct tz1090_gpio_bank *)data->domain->host_data;
>> +}
>> +
>> +static void tz1090_gpio_irq_clear(struct tz1090_gpio_bank *bank,
>> + unsigned int offset)
>> +{
>> + tz1090_gpio_clear_bit(bank, REG_GPIO_IRQ_STS, offset);
>> +}
>> +
>> +static void tz1090_gpio_irq_enable(struct tz1090_gpio_bank *bank,
>> + unsigned int offset, bool enable)
>> +{
>> + tz1090_gpio_mod_bit(bank, REG_GPIO_IRQ_EN, offset, enable);
>> +}
>> +
>> +static void tz1090_gpio_irq_polarity(struct tz1090_gpio_bank *bank,
>> + unsigned int offset, unsigned int polarity)
>> +{
>> + tz1090_gpio_mod_bit(bank, REG_GPIO_IRQ_PLRT, offset, polarity);
>> +}
>> +
>> +static int tz1090_gpio_valid_handler(struct irq_desc *desc)
>> +{
>> + return desc->handle_irq == handle_level_irq ||
>> + desc->handle_irq == handle_edge_irq;
>> +}
>> +
>> +static void tz1090_gpio_irq_type(struct tz1090_gpio_bank *bank,
>> + unsigned int offset, unsigned int type)
>> +{
>> + tz1090_gpio_mod_bit(bank, REG_GPIO_IRQ_TYPE, offset, type);
>> +}
>> +
>> +/* set polarity to trigger on next edge, whether rising or falling */
>> +static void tz1090_gpio_irq_next_edge(struct tz1090_gpio_bank *bank,
>> + unsigned int offset)
>> +{
>> + unsigned int value_p, value_i;
>> + int lstat;
>> +
>> + /*
>> + * Set the GPIO's interrupt polarity to the opposite of the current
>> + * input value so that the next edge triggers an interrupt.
>> + */
>> + __global_lock2(lstat);
>> + value_i = ~tz1090_gpio_read(bank, REG_GPIO_DIN);
>> + value_p = tz1090_gpio_read(bank, REG_GPIO_IRQ_PLRT);
>> + value_p &= ~BIT(offset);
>> + value_p |= value_i & BIT(offset);
>> + tz1090_gpio_write(bank, REG_GPIO_IRQ_PLRT, value_p);
>> + __global_unlock2(lstat);
>> +}
>> +
>> +static void gpio_ack_irq(struct irq_data *data)
>> +{
>> + struct tz1090_gpio_bank *bank = irqd_to_gpio_bank(data);
>> +
>> + tz1090_gpio_irq_clear(bank, data->hwirq);
>> +}
>> +
>> +static void gpio_mask_irq(struct irq_data *data)
>> +{
>> + struct tz1090_gpio_bank *bank = irqd_to_gpio_bank(data);
>> +
>> + tz1090_gpio_irq_enable(bank, data->hwirq, false);
>> +}
>> +
>> +static void gpio_unmask_irq(struct irq_data *data)
>> +{
>> + struct tz1090_gpio_bank *bank = irqd_to_gpio_bank(data);
>> +
>> + tz1090_gpio_irq_enable(bank, data->hwirq, true);
>> +}
>
> Similarly, can this driver use the generic irq chip to eliminate the
> above hooks?
hmm, I could probably get away with it for irq callbacks since a bank's
IRQ cannot be shared with non-Linux threads/cores.
>
> [...]
>> +
>> +static void tz1090_gpio_register_banks(struct tz1090_gpio *priv)
>> +{
>> + struct device_node *np = priv->dev->of_node;
>> + struct device_node *node;
>> +
>> + for_each_available_child_of_node(np, node) {
>> + struct tz1090_gpio_bank_info info;
>> + const __be32 *addr;
>> + int len, ret;
>> +
>> + addr = of_get_property(node, "reg", &len);
>> + if (!addr || (len < sizeof(int))) {
>> + dev_err(priv->dev, "invalid reg on %s\n",
>> + node->full_name);
>> + continue;
>> + }
>
> Use of_property_read_u32(). It's safer and does the be32 conversion for you.
will do.
Thanks for the review.
Cheers
James
--
To unsubscribe from this list: send the line "unsubscribe linux-doc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
