[PATCH RESEND] Input: omap-keypad: dynamically handle register offsets

[Dmitry Torokhov]

On Tue, Apr 10, 2012 at 11:39:43AM -0500, H Hartley Sweeten wrote:
> On Tuesday, April 10, 2012 9:24 AM, Dmitry Torokhov wrote:
> > On Tue, Apr 03, 2012 at 10:52:26AM +0530, Sourav Poddar wrote:
> >> From: G, Manjunath Kondaiah <manjugk at ti.com>
> >> 
> >> Keypad controller register offsets are different for omap4
> >> and omap5. Handle these offsets through static mapping and
> >> assign these mappings during run time.
> >> 
> >
> > In addition to Felipe's comments.
> >
> >> @@ -76,11 +81,66 @@ struct omap4_keypad {
> >>  
> >>  	unsigned int rows;
> >>  	unsigned int cols;
> >> +	unsigned int revision;
> >> +	u32 irqstatus;
> >> +	u32 irqenable;
> >
> >	u32 reg_offset;
> >
> > and you probably won't need revision field.
> >
> >>  	unsigned int row_shift;
> >>  	unsigned char key_state[8];
> >>  	unsigned short keymap[];
> >>  };
> >>  
> >> +static int kbd_readl(struct omap4_keypad *keypad_data, u32 offset)
> >> +{
> >> +	if (keypad_data->revision == KBD_REVISION_OMAP4)
> >> +		return __raw_readl(keypad_data->base + offset);
> 
> keypad_data->base is an ioremap'ed address. Shouldn't all the __raw_{read,write}l be
> {read,write}l instead?

Hmm, does it need endian conversion?

-- 
Dmitry