Re: [net-next V7 PATCH 4/5] bpf: cpumap add tracepoints

2017-10-14 Thread David Miller
From: Jesper Dangaard Brouer Date: Thu, 12 Oct 2017 14:27:05 +0200 > @@ -355,7 +360,10 @@ struct bpf_cpu_map_entry *__cpu_map_entry_alloc(u32 > qsize, u32 cpu, int map_id) > err = ptr_ring_init(rcpu->queue, qsize, gfp); > if (err) > goto free_queue; > - rcpu->qsize

[net-next V7 PATCH 4/5] bpf: cpumap add tracepoints

2017-10-12 Thread Jesper Dangaard Brouer
This adds two tracepoint to the cpumap. One for the enqueue side trace_xdp_cpumap_enqueue() and one for the kthread dequeue side trace_xdp_cpumap_kthread(). To mitigate the tracepoint overhead, these are invoked during the enqueue/dequeue bulking phases, thus amortizing the cost. The obvious use