Re: netlink callback and socket buffer.

[Pablo Neira Ayuso <pablo@netfilter.org>]

Giacomo wrote:
> Good evening to everybody.
> 
> In recent kernels (>=2.6.24) netlink_kernel_create() callback function
> has the following signature:
> 
> 
> (void *) function(struct sk_buff* skb);
> 
> while before it was
> 
> (void  *) function(struct sock *skctrl, int len)
> 
> In the latter, i used to call skb_dequeue, obtain the skb and then,
> once data was extracted, I used
> to kfree_skb(skb); where skb is a sk_buff* obtained by skb_dequeue();
> 
> Now (with the (void *) function(struct sk_buff* skb)  callback) I get
> data from skb but DO NOT HAVE
> TO kfree_skb() it...
> 
> Is it true?  Is it automatically freed by the kernel once the callback returns?

Right, look at netlink_unicast_kernel() in af_netlink.c

-- 
"Los honestos son inadaptados sociales" -- Les Luthiers