[PATCH v9] input: keyboard: Add keys driver for the LPC32xx SoC

Dmitry Torokhov dmitry.torokhov at gmail.com
Wed Jul 11 17:52:35 EST 2012


On Wed, Jul 11, 2012 at 09:49:04AM +0200, Roland Stigge wrote:
> On 07/10/2012 10:55 PM, Dmitry Torokhov wrote:
> > Hi Roland,
> > 
> > On Tue, Jul 10, 2012 at 09:35:10PM +0200, Roland Stigge wrote:
> >> This patch adds a driver for the key scan interface of the LPC32xx SoC
> >>
> > 
> > Could of more things that I had in my patch but forgot to specifically
> > call out:
> > 
> >> +
> >> +	/* Configure the key scanner */
> >> +	clk_prepare_enable(kscandat->clk);
> > 
> > This may fail so we should handle errors.
> > 
> >> +	writel(kscandat->deb_clks, LPC32XX_KS_DEB(kscandat->kscan_base));
> >> +	writel(kscandat->scan_delay, LPC32XX_KS_SCAN_CTL(kscandat->kscan_base));
> >> +	writel(LPC32XX_KSCAN_FTST_USE32K_CLK,
> >> +	       LPC32XX_KS_FAST_TST(kscandat->kscan_base));
> >> +	writel(kscandat->matrix_sz,
> >> +	       LPC32XX_KS_MATRIX_DIM(kscandat->kscan_base));
> >> +	writel(1, LPC32XX_KS_IRQ(kscandat->kscan_base));
> >> +	clk_disable_unprepare(kscandat->clk);
> >> +
> >> +	error = request_irq(irq, lpc32xx_kscan_irq, 0, pdev->name, kscandat);
> > 
> > ...
> > 
> >> +
> >> +	free_irq(platform_get_irq(pdev, 0), pdev);
> > 
> > You are requesting IRQ with kscandat as an argument but freeing it with
> > 'pdev' which will fail.
> > 
> >> +
> >> +	if (kscandat->input->users) {
> >> +		/* Enable clock and clear IRQ */
> >> +		clk_prepare_enable(kscandat->clk);
> > 
> > Need to handle errors here as well.
> > 
> > Since I am partial to my rearrangement (basically I started preferring
> > err_<action> style of error labels as they more explanatory than out2 or
> > fail3 style of labels), could you try this version of my patch (on top
> > of your v9 one). Third time is a charm maybe? :)
> 
> Yes, works fine!
> 
> Thanks!
> 
> Acked-by: Roland Stigge <stigge at antcom.de>

Excellent, I'll fold it up into your original v9 patch and queue for
3.6.

Thank you for your patience.

-- 
Dmitry


More information about the devicetree-discuss mailing list