SUNRPC: provide a mechanism for collecting stats in the RPC client

Add a simple mechanism for collecting stats in the RPC client.  Stats are
tabulated during xprt_release.  Note that per_cpu shenanigans are not
required here because the RPC client already serializes on the transport
write lock.

Test plan:
Compile kernel with CONFIG_NFS enabled.  Basic performance regression
testing with high-speed networking and high performance server.

Signed-off-by: Chuck Lever <cel@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

1 files changed, 105 insertions, 0 deletions

diff --git a/net/sunrpc/stats.c b/net/sunrpc/stats.c
index 4979f226e28..24ac7163b9c 100644
--- a/net/sunrpc/stats.c
+++ b/net/sunrpc/stats.c
@@ -21,6 +21,7 @@
 #include <linux/seq_file.h>
 #include <linux/sunrpc/clnt.h>
 #include <linux/sunrpc/svcsock.h>
+#include <linux/sunrpc/metrics.h>
 
 #define RPCDBG_FACILITY	RPCDBG_MISC
 
@@ -106,6 +107,110 @@ void svc_seq_show(struct seq_file *seq, const struct svc_stat *statp) {
 	}
 }
 
+/**
+ * rpc_alloc_iostats - allocate an rpc_iostats structure
+ * @clnt: RPC program, version, and xprt
+ *
+ */
+struct rpc_iostats *rpc_alloc_iostats(struct rpc_clnt *clnt)
+{
+	unsigned int ops = clnt->cl_maxproc;
+	size_t size = ops * sizeof(struct rpc_iostats);
+	struct rpc_iostats *new;
+
+	new = kmalloc(size, GFP_KERNEL);
+	if (new)
+		memset(new, 0 , size);
+	return new;
+}
+EXPORT_SYMBOL(rpc_alloc_iostats);
+
+/**
+ * rpc_free_iostats - release an rpc_iostats structure
+ * @stats: doomed rpc_iostats structure
+ *
+ */
+void rpc_free_iostats(struct rpc_iostats *stats)
+{
+	kfree(stats);
+}
+EXPORT_SYMBOL(rpc_free_iostats);
+
+/**
+ * rpc_count_iostats - tally up per-task stats
+ * @task: completed rpc_task
+ *
+ * Relies on the caller for serialization.
+ */
+void rpc_count_iostats(struct rpc_task *task)
+{
+	struct rpc_rqst *req = task->tk_rqstp;
+	struct rpc_iostats *stats = task->tk_client->cl_metrics;
+	struct rpc_iostats *op_metrics;
+	long rtt, execute, queue;
+
+	if (!stats || !req)
+		return;
+	op_metrics = &stats[task->tk_msg.rpc_proc->p_proc];
+
+	op_metrics->om_ops++;
+	op_metrics->om_ntrans += req->rq_ntrans;
+	op_metrics->om_timeouts += task->tk_timeouts;
+
+	op_metrics->om_bytes_sent += task->tk_bytes_sent;
+	op_metrics->om_bytes_recv += req->rq_received;
+
+	queue = (long)req->rq_xtime - task->tk_start;
+	if (queue < 0)
+		queue = -queue;
+	op_metrics->om_queue += queue;
+
+	rtt = task->tk_rtt;
+	if (rtt < 0)
+		rtt = -rtt;
+	op_metrics->om_rtt += rtt;
+
+	execute = (long)jiffies - task->tk_start;
+	if (execute < 0)
+		execute = -execute;
+	op_metrics->om_execute += execute;
+}
+
+#define MILLISECS_PER_JIFFY	(1000UL / HZ)
+
+void rpc_print_iostats(struct seq_file *seq, struct rpc_clnt *clnt)
+{
+	struct rpc_iostats *stats = clnt->cl_metrics;
+	struct rpc_xprt *xprt = clnt->cl_xprt;
+	unsigned int op, maxproc = clnt->cl_maxproc;
+
+	if (!stats)
+		return;
+
+	seq_printf(seq, "\tRPC iostats version: %s  ", RPC_IOSTATS_VERS);
+	seq_printf(seq, "p/v: %u/%u (%s)\n",
+			clnt->cl_prog, clnt->cl_vers, clnt->cl_protname);
+
+	if (xprt)
+		xprt->ops->print_stats(xprt, seq);
+
+	seq_printf(seq, "\tper-op statistics\n");
+	for (op = 0; op < maxproc; op++) {
+		struct rpc_iostats *metrics = &stats[op];
+		seq_printf(seq, "%12u: ", op);
+		seq_printf(seq, "%lu %lu %lu %Lu %Lu %Lu %Lu %Lu\n",
+				metrics->om_ops,
+				metrics->om_ntrans,
+				metrics->om_timeouts,
+				metrics->om_bytes_sent,
+				metrics->om_bytes_recv,
+				metrics->om_queue * MILLISECS_PER_JIFFY,
+				metrics->om_rtt * MILLISECS_PER_JIFFY,
+				metrics->om_execute * MILLISECS_PER_JIFFY);
+	}
+}
+EXPORT_SYMBOL(rpc_print_iostats);
+
 /*
  * Register/unregister RPC proc files
  */