Re: [PATCH] clk: fix member type of struct clk_hw_onecell_data

From: Masahiro Yamada
Date: Tue Apr 26 2016 - 02:19:22 EST


Hi Stephen,


2016-04-26 7:27 GMT+09:00 Stephen Boyd <sboyd@xxxxxxxxxxxxxx>:
> On 04/25, Masahiro Yamada wrote:
>> We cannot assign any value to an array type variable. So,
>>
>> hw_data->hws = kcalloc(hw_data->num, sizeof(struct clk_hw *),
>> GFP_KERNEL);
>>
>> fails with "invalid use of flexible array member" error.
>
> That's good. We don't want assignment to the hws member of this
> structure to happen from an allocation like kcalloc.
>
>>
>> There are two ways to fix this issue.
>>
>> [1] Make it a double-pointer
>> struct clk_hw_onecell_data {
>> size_t num;
>> struct clk_hw **hws;
>> };
>>
>> This works as struct clk_onecell_data does.
>
> True we could go back to the old style with two allocations.
>
>>
>> [2] Make it a zero-sized array
>> struct clk_hw_onecell_data {
>> size_t num;
>> struct clk_hw *hws[0];
>> };
>>
>> This allows one-shot memory allocation like this:
>>
>> hw_data = kmalloc(sizeof(*hw_data) + clk_num * sizeof(struct clk_hw *),
>> GFP_KERNEL);
>>
>
> Good, that was possible before this patch wasn't it?
>
>> This commit adopts [2] because it looks like Stephen's intention
>> (he moved hws[] to the bottom of struct clk_hw_onecell_data).
>>
>> Fixes: 0861e5b8cf80 ("clk: Add clk_hw OF clk providers")
>> Signed-off-by: Masahiro Yamada <yamada.masahiro@xxxxxxxxxxxxx>
>> ---
>
>>
>> include/linux/clk-provider.h | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/include/linux/clk-provider.h b/include/linux/clk-provider.h
>> index fd2ccd5..1850e25 100644
>> --- a/include/linux/clk-provider.h
>> +++ b/include/linux/clk-provider.h
>> @@ -769,7 +769,7 @@ struct clk_onecell_data {
>>
>> struct clk_hw_onecell_data {
>> size_t num;
>> - struct clk_hw *hws[];
>> + struct clk_hw *hws[0];
>
> I'm totally lost now. Isn't a flex array with [] or with [0] the
> same?

I was misunderstanding.

You are right. They are the same.


> The latter being a GCC extension while the former being a
> C99 standard?


Please ignore this patch. Sorry for noise.




--
Best Regards
Masahiro Yamada