On 12/02/14 08:44, Andrew Lunn wrote:
>>>> +	bat_priv->bat_v.ogm_buff_len = BATADV_OGM2_HLEN;
>>>> +	ogm_buff = kmalloc(bat_priv->bat_v.ogm_buff_len, GFP_ATOMIC);
>>>
>>> Maybe use kzalloc to ensure it is zero?
>>
>> Not really needed because each field is explicitly assigned later. But
>> to be sure we sit on the safe side we can do that: in the end we are not
>> in the fastpath here.
> 
> I know some of your structures have explicit align bytes. It would be
> good to ensure they are zero so that they could be used in the future
> without having a protocol version bump. So i think it is a good
> patterns to follow in general when not on the fast path.

True. _I_ had this problem in the past already :-)
I'll use kzalloc then!

Thanks!


> 
> 	Andrew
> 


-- 
Antonio Quartulli