Re: [PATCH 12/12] staging: r8188eu: hal_data_sz is set but never used

[Michael Straube <straube.linux@gmail.com>]

On 12/5/21 15:42, Pavel Skripkin wrote:
> On 12/5/21 16:59, Michael Straube wrote:
>> hal_data_sz in struct adapter is set but never used. Remove it.
>>
>> Signed-off-by: Michael Straube <straube.linux@gmail.com>
>> ---
>>   drivers/staging/r8188eu/hal/usb_halinit.c   | 1 -
>>   drivers/staging/r8188eu/include/drv_types.h | 1 -
>>   2 files changed, 2 deletions(-)
>>
>> diff --git a/drivers/staging/r8188eu/hal/usb_halinit.c 
>> b/drivers/staging/r8188eu/hal/usb_halinit.c
>> index 641aaf299109..da966538596f 100644
>> --- a/drivers/staging/r8188eu/hal/usb_halinit.c
>> +++ b/drivers/staging/r8188eu/hal/usb_halinit.c
>> @@ -1969,5 +1969,4 @@ void rtl8188eu_alloc_haldata(struct adapter *adapt)
>>       adapt->HalData = kzalloc(sizeof(struct hal_data_8188e), 
>> GFP_KERNEL);
>>       if (!adapt->HalData)
>>           DBG_88E("cant not alloc memory for HAL DATA\n");
>> -    adapt->hal_data_sz = sizeof(struct hal_data_8188e);
>>   }
> 
> Not related to your patch, but not returning an error from this function 
>   looks very dangerous to me.

I agree.

> 
> adapt->HalData is used in GET_HAL_DATA() macro all across the driver 
> code and nobody checks if it valid or not. If allocation fails here, 
> than we will likely hit GPF while accessing hal_data fields.
> 
> Maybe we can embed struct hal_data_8188e instead of storing a pointer to 
> it?
> 
> 
> 

We could remove rtl8188eu_alloc_haldata() completely and replace its
usage in the function rtw_usb_if1_init().

--- a/drivers/staging/r8188eu/os_dep/usb_intf.c
+++ b/drivers/staging/r8188eu/os_dep/usb_intf.c
@@ -362,7 +362,9 @@ static struct adapter *rtw_usb_if1_init(struct 
dvobj_priv *dvobj,
         padapter = rtw_netdev_priv(pnetdev);

         /* step 2. allocate HalData */
-       rtl8188eu_alloc_haldata(padapter);
+       padapter->HalData = kzalloc(sizeof(*padapter->HalData), GFP_KERNEL);
+       if (!padapter->HalData)
+               goto handle_dualmac;

This way rtw_drv_init() would return -ENODEV if the allocation fails.
What do you think?

Regards,
Michael