On 7/22/15 1:09 AM, Kaixu Xia wrote:
We want to traverse the map elements and make use of the map value one by one. So add new function map->ops->map_traverse_elem() to traverse map elements.
...
@@ -26,6 +28,7 @@ struct bpf_map_ops { void *(*map_lookup_elem)(struct bpf_map *map, void *key); int (*map_update_elem)(struct bpf_map *map, void *key, void *value, u64 flags); int (*map_delete_elem)(struct bpf_map *map, void *key); + int (*map_traverse_elem)(bpf_map_traverse_callback func, struct bpf_map *map);
I think 'traverse' is too error prone. Better to add 'map_delete_all_elem' callback. Then convert bpf_prog_array_map_clear() into such callback and similar callback from perf_event_array. Then call it with from bpf_map_release() and drop if (map->map_type == PROG_ARRAY) there. Just unconditionally map->ops->map_delete_all_elem(). Though looking at patch 3. I don't see why you need that. prog_array has to do it due to potential circular dependency between prog_fd of a program that uses a prog_array_fd and stores prog_fd inside that prog_array. In case of perf_event_array no such thing possible. Just release them as part of map_free. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/