net-sysfs: add backlog len and CPU id to softnet data (7d58e655) · Commits · 戴 / test

Currently the backlog status in not exposed to user-space. Since long backlogs (or simply not empty ones) can be a source of noticeable latency, -RT deployments need some way to inspect it. Additionally, there isn't a direct match between 'softnet_stat' lines and the related CPU - sd for offline CPUs are not dumped - so this patch also includes the CPU id into such entry. Signed-off-by:

Paolo Abeni <pabeni@redhat.com> Signed-off-by:

David S. Miller <davem@davemloft.net>

net/core/net-procfs.c

+13 −2

Original line number	Diff line number	Diff line
		@@ -116,6 +116,12 @@ static int dev_seq_show(struct seq_file seq, void v)
		return 0;
		}

		static u32 softnet_backlog_len(struct softnet_data *sd)
		{
		return skb_queue_len_lockless(&sd->input_pkt_queue) +
		skb_queue_len_lockless(&sd->process_queue);
		}

		static struct softnet_data softnet_get_online(loff_t pos)
		{
		struct softnet_data *sd = NULL;
		@@ -159,12 +165,17 @@ static int softnet_seq_show(struct seq_file seq, void v)
		rcu_read_unlock();
		#endif

		/* the index is the CPU id owing this sd. Since offline CPUs are not
		* displayed, it would be othrwise not trivial for the user-space
		* mapping the data a specific CPU
		*/
		seq_printf(seq,
		"%08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x\n",
		"%08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x %08x\n",
		sd->processed, sd->dropped, sd->time_squeeze, 0,
		0, 0, 0, 0, /* was fastroute */
		0, /* was cpu_collision */
		sd->received_rps, flow_limit_count);
		sd->received_rps, flow_limit_count,
		softnet_backlog_len(sd), (int)seq->index);
		return 0;
		}

Admin message