On 22/06/15 09:13, Sven Eckelmann wrote:
> -	gw_divisor = BATADV_TQ_LOCAL_WINDOW_SIZE * BATADV_TQ_LOCAL_WINDOW_SIZE;
> -	gw_divisor *= 64;

Marek can you comment on why this constant was used in this computation?
To me it looks like we only needed a "good enough number" and not a real
reference to the local window size.

The remaining change looks good to me.

If you merge this patch I'll then squash it with the uint64_t change
that we already have and I will re-send the pull request to David.

Thanks,

> -
>  	rcu_read_lock();
>  	hlist_for_each_entry_rcu(gw_node, &bat_priv->gw.list, list) {
>  		if (gw_node->deleted)
> @@ -167,7 +163,7 @@ batadv_gw_get_best_gw_node(struct batadv_priv *bat_priv)
>  			tmp_gw_factor = tq_avg * tq_avg;
>  			tmp_gw_factor *= gw_node->bandwidth_down;
>  			tmp_gw_factor *= 100 * 100;
> -			tmp_gw_factor /= gw_divisor;
> +			tmp_gw_factor >>= 18;
>  
>  			if ((tmp_gw_factor > max_gw_factor) ||
>  			    ((tmp_gw_factor == max_gw_factor) &&
> 

-- 
Antonio Quartulli