Re: [PATCH] staging: rtl8192u: initialize array in C compliant way

Re: [PATCH] staging: rtl8192u: initialize array in C compliant way

[Dan Carpenter]

On Tue, May 06, 2014 at 10:22:08AM +0200, Martin Kepplinger wrote:
> Don't list elements to initialize. Remaining elements of a partly
> initialized array are set to zero. Sparse complained here.
> 
> Signed-off-by: Martin Kepplinger <martink@posteo.de>
> ---
>  drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c b/drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c
> index 426f223..c96dbab 100644
> --- a/drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c
> +++ b/drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c
> @@ -241,7 +241,7 @@ static PTS_COMMON_INFO SearchAdmitTRStream(struct ieee80211_device *ieee,
>  {
>  	//DIRECTION_VALUE	dir;
>  	u8	dir;
> -	bool				search_dir[4] = {0, 0, 0, 0};
> +	bool				search_dir[4] = {0};

That's weird.  The original code is valid but it generates a sparse
warning.

drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c:244:58: warning: Initializer entry defined twice
drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c:244:61:   also defined here

It has something to do with "_Bool" types.  Changing it to "int" will
also make the warning disappear.  I've CC'd the sparse list to see if
anyone knows what's happening.

regards,
dan carpenter

Re: [PATCH] staging: rtl8192u: initialize array in C compliant way

[Christopher Li]

On Tue, May 6, 2014 at 1:47 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>> -     bool                            search_dir[4] = {0, 0, 0, 0};
>> +     bool                            search_dir[4] = {0};
>
> That's weird.  The original code is valid but it generates a sparse
> warning.
>
> drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c:244:58: warning: Initializer entry defined twice
> drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c:244:61:   also defined here
>
> It has something to do with "_Bool" types.  Changing it to "int" will
> also make the warning disappear.  I've CC'd the sparse list to see if
> anyone knows what's happening.

This is easy reproducible with a small test case. It is a sparse bug for sure.

Let me take a look.

Chris

Re: [PATCH] staging: rtl8192u: initialize array in C compliant way

[Christopher Li]

On Tue, May 6, 2014 at 1:47 AM, Dan Carpenter <dan.carpenter@oracle.com> wrote:
>> diff --git a/drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c b/drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c
>> index 426f223..c96dbab 100644
>> --- a/drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c
>> +++ b/drivers/staging/rtl8192u/ieee80211/rtl819x_TSProc.c
>> @@ -241,7 +241,7 @@ static PTS_COMMON_INFO SearchAdmitTRStream(struct ieee80211_device *ieee,
>>  {
>>       //DIRECTION_VALUE       dir;
>>       u8      dir;
>> -     bool                            search_dir[4] = {0, 0, 0, 0};
>> +     bool                            search_dir[4] = {0};
>
> That's weird.  The original code is valid but it generates a sparse
> warning.

Sorry for the late reply. I see what is going on already. The bits_to_bytes()
function is buggy for bool type. Bool is the only type has bits less than 8.
It truncates the bool byte size to 0. As a result, the array layout function
convert_index() always set the expr->init_offset to 0 for bool type:

"e->init_offset = from * bits_to_bytes(e->ctype->bit_size);"

Chris