Re: [PATCH 1/3] RTC:s35390a: Add Alarm interrupt support

[Felipe Balbi]

On Tue, Aug 17, 2010 at 10:48:39AM +0200, ext hvaibhav@xxxxxx wrote:
> From: Vaibhav Hiremath <hvaibhav@xxxxxx>

you need a description here.

> Signed-off-by: Vaibhav Hiremath <hvaibhav@xxxxxx>

[snip]

> +static void s35390a_work(struct work_struct *work)
> +{
> +	struct s35390a *s35390a;
> +	struct i2c_client *client;
> +	char buf[1];
> +
> +	s35390a = container_of(work, struct s35390a, work);
> +	if (!s35390a)
> +		return;

container_of() will never return NULL. You can remove this check. You 
won't need this, actually after converting to threaded_irq, see below.

> +static irqreturn_t s35390a_irq(int irq, void *client)
> +{
> +	struct s35390a *s35390a;

all the other drivers will do:

static irqreturn_t s35390a_irq(int irq, void *_s35390a)
{
	struct s35390a	*s35390a = _s35390a

	[...]

> +	if (!client)
> +		return IRQ_HANDLED;

if client is NULL, you should let this oops.

> +	schedule_work(&s35390a->work);

please don't use workqueue. Use threaded IRQ.

> @@ -261,15 +447,30 @@ static int s35390a_probe(struct i2c_client *client,
> 	if (s35390a_get_datetime(client, &tm) < 0)
> 		dev_warn(&client->dev, "clock needs to be set\n");
>
> +	INIT_WORK(&s35390a->work, s35390a_work);
> +
> +	if (client->irq > 0) {

irq 0 is a valid number.

> +		err = request_irq(client->irq, s35390a_irq, IRQF_TRIGGER_LOW,
> +				client->name, client);

instead of the i2c client, you can pass s35390. Also use 
request_threaded_irq();

--
balbi

DefectiveByDesign.org
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/