Re: [PATCH 5/5] Input: add wm97xx accelerated driver for AVR32 AT32AP700X microprocessors

[Jiri Slaby <jirislaby@gmail.com>]

On 3.3.2009 19:24, Mark Brown wrote:
> +static irqreturn_t at32_wm97xx_channel_b_interrupt(int irq, void *dev_id)
> +{
> +	struct at32_wm97xx *at32_wm97xx = dev_id;
> +	struct wm97xx *wm = at32_wm97xx->wm;
> +	int status = ac97c_readl(at32_wm97xx, CBSR);
> +	int retval = IRQ_NONE;

Just a nit, irqreturn_t will become enum one day AFAIK. Use irqreturn_t 
which is guaranteed to be the proper type instead. This rather a 
suggestion for followup fixup.

> +static int __init at32_wm97xx_probe(struct platform_device *pdev)
> +{
> +	struct wm97xx *wm = platform_get_drvdata(pdev);
> +	struct at32_wm97xx *at32_wm97xx;
> +	int ret;
> +
> +	at32_wm97xx = kzalloc(sizeof(struct at32_wm97xx), GFP_KERNEL);
> +	if (!at32_wm97xx) {
> +		dev_dbg(&pdev->dev, "out of memory\n");
> +		return -ENOMEM;
> +	}
> +
> +	at32_wm97xx->wm		= wm;
> +	at32_wm97xx->regs	= (void *)AT32_WM97XX_AC97C_IOMEM;
> +	at32_wm97xx->ac97c_irq	= AT32_WM97XX_AC97C_IRQ;
> +	at32_wm97xx->gpio_pen	= at32_gpio_line;
> +	at32_wm97xx->gpio_irq	= gpio_to_irq(at32_wm97xx->gpio_pen);
> +
> +	setup_timer(&at32_wm97xx->pen_timer, at32_wm97xx_pen_timer,
> +			(unsigned long)at32_wm97xx);
> +
> +	ret = request_irq(at32_wm97xx->ac97c_irq,
> +			at32_wm97xx_channel_b_interrupt,
> +			IRQF_SHARED, "at32-wm97xx-ch-b", at32_wm97xx);
> +	if (ret) {
> +		dev_dbg(&pdev->dev, "could not request ac97c irq\n");

at32_wm97xx leak. I didn't notice last time.

> +		return ret;
> +	}
> +
> +	platform_set_drvdata(pdev, at32_wm97xx);
> +
> +	return wm97xx_register_mach_ops(wm,&at32_mach_ops);

On failure, this will still "leak" irq, I suppose this is what you 
referred as "fix later" in 0/5.