From: Prasad Sodagudi <psodagud@codeaurora.org>
To: rostedt@goodmis.org, tglx@linutronix.de, qais.yousef@arm.com,
peterz@infradead.org, mingo@kernel.org, cai@lca.pw,
tyhicks@canonical.com, arnd@arndb.de
Cc: rameezmustafa@codeaurora.org, linux-kernel@vger.kernel.org,
Prasad Sodagudi <psodagud@codeaurora.org>
Subject: [PATCH 1/2] cpu/hotplug: Add cpuhp_latency trace event
Date: Wed, 23 Sep 2020 16:37:45 -0700 [thread overview]
Message-ID: <1600904266-102397-2-git-send-email-psodagud@codeaurora.org> (raw)
In-Reply-To: <1600904266-102397-1-git-send-email-psodagud@codeaurora.org>
Add ftrace event trace_cpuhp_latency to track cpu
hotplug latency. It helps to track the hotplug latency
impact by firmware changes and kernel cpu hotplug callbacks.
Signed-off-by: Prasad Sodagudi <psodagud@codeaurora.org>
---
include/trace/events/cpuhp.h | 29 +++++++++++++++++++++++++++++
kernel/cpu.c | 9 +++++++++
2 files changed, 38 insertions(+)
diff --git a/include/trace/events/cpuhp.h b/include/trace/events/cpuhp.h
index ad16f77..c871850 100644
--- a/include/trace/events/cpuhp.h
+++ b/include/trace/events/cpuhp.h
@@ -6,6 +6,7 @@
#define _TRACE_CPUHP_H
#include <linux/tracepoint.h>
+#include <linux/sched/clock.h>
TRACE_EVENT(cpuhp_enter,
@@ -89,6 +90,34 @@ TRACE_EVENT(cpuhp_exit,
__entry->cpu, __entry->state, __entry->idx, __entry->ret)
);
+TRACE_EVENT(cpuhp_latency,
+
+ TP_PROTO(unsigned int cpu, unsigned int state,
+ u64 start_time, int ret),
+
+ TP_ARGS(cpu, state, start_time, ret),
+
+ TP_STRUCT__entry(
+ __field(unsigned int, cpu)
+ __field(unsigned int, state)
+ __field(u64, time)
+ __field(int, ret)
+ ),
+
+ TP_fast_assign(
+ __entry->cpu = cpu;
+ __entry->state = state;
+ __entry->time = div64_u64(sched_clock() - start_time, 1000);
+ __entry->ret = ret;
+ ),
+
+ TP_printk(" cpu:%d state:%s latency:%llu USEC ret: %d",
+ __entry->cpu, __entry->state ? "online" : "offline",
+ __entry->time, __entry->ret)
+);
+
+
+
#endif
/* This part must be outside protection */
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 6ff2578..68b3740 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -36,6 +36,7 @@
#include <trace/events/power.h>
#define CREATE_TRACE_POINTS
#include <trace/events/cpuhp.h>
+#include <linux/sched/clock.h>
#include "smpboot.h"
@@ -994,6 +995,7 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen,
{
struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
int prev_state, ret = 0;
+ u64 start_time = 0;
if (num_online_cpus() == 1)
return -EBUSY;
@@ -1002,6 +1004,8 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen,
return -EINVAL;
cpus_write_lock();
+ if (trace_cpuhp_latency_enabled())
+ start_time = sched_clock();
cpuhp_tasks_frozen = tasks_frozen;
@@ -1040,6 +1044,7 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen,
}
out:
+ trace_cpuhp_latency(cpu, 0, start_time, ret);
cpus_write_unlock();
/*
* Do post unplug cleanup. This is still protected against
@@ -1192,8 +1197,11 @@ static int _cpu_up(unsigned int cpu, int tasks_frozen, enum cpuhp_state target)
struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu);
struct task_struct *idle;
int ret = 0;
+ u64 start_time = 0;
cpus_write_lock();
+ if (trace_cpuhp_latency_enabled())
+ start_time = sched_clock();
if (!cpu_present(cpu)) {
ret = -EINVAL;
@@ -1241,6 +1249,7 @@ static int _cpu_up(unsigned int cpu, int tasks_frozen, enum cpuhp_state target)
target = min((int)target, CPUHP_BRINGUP_CPU);
ret = cpuhp_up_callbacks(cpu, st, target);
out:
+ trace_cpuhp_latency(cpu, 1, start_time, ret);
cpus_write_unlock();
arch_smt_update();
return ret;
--
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project
next prev parent reply other threads:[~2020-09-23 23:38 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-09-23 23:37 [PATCH 0/2] measure latency of cpu hotplug path Prasad Sodagudi
2020-09-23 23:37 ` Prasad Sodagudi [this message]
2020-09-28 14:58 ` [PATCH 1/2] cpu/hotplug: Add cpuhp_latency trace event Thomas Gleixner
2020-09-23 23:37 ` [PATCH 2/2] cpu-hotplug: Always use real time scheduling when hotplugging a CPU Prasad Sodagudi
2020-09-24 8:34 ` [PATCH 0/2] measure latency of cpu hotplug path peterz
2020-09-24 14:58 ` Steven Rostedt
2020-09-28 2:41 ` psodagud
2020-09-28 7:40 ` Peter Zijlstra
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1600904266-102397-2-git-send-email-psodagud@codeaurora.org \
--to=psodagud@codeaurora.org \
--cc=arnd@arndb.de \
--cc=cai@lca.pw \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=peterz@infradead.org \
--cc=qais.yousef@arm.com \
--cc=rameezmustafa@codeaurora.org \
--cc=rostedt@goodmis.org \
--cc=tglx@linutronix.de \
--cc=tyhicks@canonical.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).