Re: [PATCH 1/2] Input: DaVinci Keypad Driver

[Miguel Aguilar <miguel.aguilar@ridgerun.com>]

Dmitry Torokhov wrote:
> On Wed, Sep 23, 2009 at 11:25:59AM -0600, Miguel Aguilar wrote:
>>>>>> Please kopy keymap into the davinci_kp stucture and use it so that
>>>>>> platform data is never changed and can be declared const.
>>>>> Do you mean something like this?
>>>>>
>>>>> struct davinci_kp {
>>>>>     ...
>>>>>     const int    *keymap;
>>>>>     ...
>>>>> };
>>>>>
>>>> More like:
>>>>
>>>> struct davinci_kp {
>>>>     ...
>>>>     unsgned char keymap[];
>>>>     ...
>>>> };
>>>>
>>> [MA] Ok.
>> [MA] Why usigned char with no pointer and not u32 as most of the keypad 
>> driver as defined?
> 
> Sorry, meant to say "unsigned short keymap[...]", we not going to have
> more than 64K keycodes. You need to fill the array dimension, if the
> size is not known before hand (and if upper bound is too jigh to always
> allocate max) then you'll have to allocate it separately. Hm, well, if
> you make it the last element of the davinci_kp structure you can alwqays
> allocate the needed amount of memory, like this:
> 
> struct davinci_kp {
> 	...
> 	...
> 	unsigned short keymap[];
> };
> 
> ...
> 
> kp = kzalloc(sizeof(struct davinci_kp) +
> 			sizeof(unsigned short) * pdata->keymap_size,
> 	     GFP_KERNEL);
> 

Why do I need to use keymap[]; instead of *keymap? If I use keypad[] how is the 
proper way to assign the platform keymap to the private keymap of the driver?