summaryrefslogtreecommitdiffstats
path: root/net/sunrpc
diff options
context:
space:
mode:
authorChuck Lever <chuck.lever@oracle.com>2010-05-07 19:34:47 +0200
committerTrond Myklebust <Trond.Myklebust@netapp.com>2010-05-14 21:09:33 +0200
commitff8399709e41bf72b4cb145612a0f9a9f7283c83 (patch)
tree5577163d56e30020d907245f00dcef17730f037b /net/sunrpc
parentktime: introduce ktime_to_ms() (diff)
downloadlinux-ff8399709e41bf72b4cb145612a0f9a9f7283c83.tar.xz
linux-ff8399709e41bf72b4cb145612a0f9a9f7283c83.zip
SUNRPC: Replace jiffies-based metrics with ktime-based metrics
Currently RPC performance metrics that tabulate elapsed time use jiffies time values. This is problematic on systems that use slow jiffies (for instance 100HZ systems built for paravirtualized environments). It is also a problem for computing precise latency statistics for advanced network transports, such as InfiniBand, that can have round-trip latencies significanly faster than a single clock tick. For the RPC client, adopt the high resolution time stamp mechanism already used by the network layer and blktrace: ktime. We use ktime format time stamps for all internal computations, and convert to milliseconds for presentation. As a result, we need only addition operations in the performance critical paths; multiply/divide is required only for presentation. We could report RTT metrics in microseconds. In fact the mountstats format is versioned to accomodate exactly this kind of interface improvement. For now, however, we'll stay with millisecond precision for presentation to maintain backwards compatibility with the handful of currently deployed user space tools. At a later point, we'll move to an API such as BDI_STATS where a finer timestamp precision can be reported. Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Diffstat (limited to 'net/sunrpc')
-rw-r--r--net/sunrpc/sched.c2
-rw-r--r--net/sunrpc/stats.c27
-rw-r--r--net/sunrpc/xprt.c8
3 files changed, 15 insertions, 22 deletions
diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c
index c8979ce5d88a..aa7b07ef5d55 100644
--- a/net/sunrpc/sched.c
+++ b/net/sunrpc/sched.c
@@ -834,7 +834,7 @@ static void rpc_init_task(struct rpc_task *task, const struct rpc_task_setup *ta
}
/* starting timestamp */
- task->tk_start = jiffies;
+ task->tk_start = ktime_get();
dprintk("RPC: new task initialized, procpid %u\n",
task_pid_nr(current));
diff --git a/net/sunrpc/stats.c b/net/sunrpc/stats.c
index 5785d2037f45..aacd95f0dce5 100644
--- a/net/sunrpc/stats.c
+++ b/net/sunrpc/stats.c
@@ -144,7 +144,7 @@ void rpc_count_iostats(struct rpc_task *task)
struct rpc_rqst *req = task->tk_rqstp;
struct rpc_iostats *stats;
struct rpc_iostats *op_metrics;
- long rtt, execute, queue;
+ ktime_t delta;
if (!task->tk_client || !task->tk_client->cl_metrics || !req)
return;
@@ -159,20 +159,13 @@ void rpc_count_iostats(struct rpc_task *task)
op_metrics->om_bytes_sent += task->tk_bytes_sent;
op_metrics->om_bytes_recv += req->rq_reply_bytes_recvd;
- queue = (long)req->rq_xtime - task->tk_start;
- if (queue < 0)
- queue = -queue;
- op_metrics->om_queue += queue;
+ delta = ktime_sub(req->rq_xtime, task->tk_start);
+ op_metrics->om_queue = ktime_add(op_metrics->om_queue, delta);
- rtt = task->tk_rtt;
- if (rtt < 0)
- rtt = -rtt;
- op_metrics->om_rtt += rtt;
+ op_metrics->om_rtt = ktime_add(op_metrics->om_rtt, task->tk_rtt);
- execute = (long)jiffies - task->tk_start;
- if (execute < 0)
- execute = -execute;
- op_metrics->om_execute += execute;
+ delta = ktime_sub(ktime_get(), task->tk_start);
+ op_metrics->om_execute = ktime_add(op_metrics->om_execute, delta);
}
static void _print_name(struct seq_file *seq, unsigned int op,
@@ -186,8 +179,6 @@ static void _print_name(struct seq_file *seq, unsigned int op,
seq_printf(seq, "\t%12u: ", op);
}
-#define MILLISECS_PER_JIFFY (1000 / HZ)
-
void rpc_print_iostats(struct seq_file *seq, struct rpc_clnt *clnt)
{
struct rpc_iostats *stats = clnt->cl_metrics;
@@ -214,9 +205,9 @@ void rpc_print_iostats(struct seq_file *seq, struct rpc_clnt *clnt)
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);
+ ktime_to_ms(metrics->om_queue),
+ ktime_to_ms(metrics->om_rtt),
+ ktime_to_ms(metrics->om_execute));
}
}
EXPORT_SYMBOL_GPL(rpc_print_iostats);
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index 698c62712294..8986b1b82862 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -43,6 +43,7 @@
#include <linux/interrupt.h>
#include <linux/workqueue.h>
#include <linux/net.h>
+#include <linux/ktime.h>
#include <linux/sunrpc/clnt.h>
#include <linux/sunrpc/metrics.h>
@@ -779,10 +780,11 @@ static void xprt_update_rtt(struct rpc_task *task)
struct rpc_rqst *req = task->tk_rqstp;
struct rpc_rtt *rtt = task->tk_client->cl_rtt;
unsigned timer = task->tk_msg.rpc_proc->p_timer;
+ long m = usecs_to_jiffies(ktime_to_us(task->tk_rtt));
if (timer) {
if (req->rq_ntrans == 1)
- rpc_update_rtt(rtt, timer, task->tk_rtt);
+ rpc_update_rtt(rtt, timer, m);
rpc_set_timeo(rtt, timer, req->rq_ntrans - 1);
}
}
@@ -803,7 +805,7 @@ void xprt_complete_rqst(struct rpc_task *task, int copied)
task->tk_pid, ntohl(req->rq_xid), copied);
xprt->stat.recvs++;
- task->tk_rtt = (long)jiffies - req->rq_xtime;
+ task->tk_rtt = ktime_sub(ktime_get(), req->rq_xtime);
if (xprt->ops->timer != NULL)
xprt_update_rtt(task);
@@ -904,7 +906,7 @@ void xprt_transmit(struct rpc_task *task)
return;
req->rq_connect_cookie = xprt->connect_cookie;
- req->rq_xtime = jiffies;
+ req->rq_xtime = ktime_get();
status = xprt->ops->send_request(task);
if (status != 0) {
task->tk_status = status;