Hi,
I massaged the RFC patches a little longer to integrate all the feedback received so far. They are ready to be considered 'real' patches now.
The 'dynamically load routing kernel module' feature has not found its way into my patchset. At this point I still see no benefit and only complications. The mentioned reasons (kernel folks might not like it / helps to better abstract) can be addressed in a different way (for example asking David if he is going to accept it or not). The only reason seems to be "because we can". That's not enough for me. This does not mean we can't add this feature at a later point in time when we have the feeling it actually brings some benefits.
Here the changelog:
* sysfs documentation added * changed batman iv function prefix to bat_iv * renamed struct bat_algo to struct bat_algo_ops * all bat_algo_ops callbacks are mandatory for now * added struct bat_algo_ops documentation * introduced bat_algo.h for the routing algo init calls * bat_algo_ops->name became a pointer * removed all (unnecessary) locking from bat_algo_*
Cheers, Marek