Skip to contentSkip to conversation

suspend: add initcall_debug style timing for suspend/resume

PATCH In review Arjan van de Ven 2 participants 5 messages Dec 6, 2009 – Dec 12, 2009

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 suspend: add initcall_debug style timing for suspend/resume

Quoted 2 lines

From fcd52b97aa02f19baa570e4a62dfdfab0faed976 Mon Sep 17 00:00:00 2001

From: Arjan van de Ven <[email protected]>
Date: Sun, 6 Dec 2009 12:32:28 -0800
Subject: [PATCH] suspend: add initcall_debug style timing for suspend/resume

In order to diagnose overall suspend/resume times, we need basic instrumentation to break down the total time into per device timing, similar to initcall_debug.

This patch adds the basic timing instrumentation, needed for a scritps/bootgraph.pl equivalent or humans. The bootgraph.pl program is still a work in progress, but is far enough along to know that this patch is sufficient.

---
 drivers/base/power/main.c |   29 +++++++++++++++++++++++++++++
 include/linux/init.h      |    2 ++
 2 files changed, 31 insertions(+), 0 deletions(-)
drivers/base/power/main.c +29 -0
index 8aa2443..2ad345c 100644
@@ -25,6 +25,7 @@
 #include <linux/resume-trace.h>
 #include <linux/rwsem.h>
 #include <linux/interrupt.h>
+#include <linux/sched.h>
 
 #include "../base.h"
 #include "power.h"
@@ -172,6 +173,12 @@ static int pm_op(struct device *dev,
 		 pm_message_t state)
 {
 	int error = 0;
+	ktime_t calltime, delta, rettime;
+
+	if (initcall_debug) {
+		printk("calling  %s+ @ %i\n", dev_name(dev), task_pid_nr(current));
+		calltime = ktime_get();
+	}
 
 	switch (state.event) {
 #ifdef CONFIG_SUSPEND
@@ -219,6 +226,14 @@ static int pm_op(struct device *dev,
 	default:
 		error = -EINVAL;
 	}
+
+	if (initcall_debug) {
+		rettime = ktime_get();
+		delta = ktime_sub(rettime, calltime);
+		printk("initcall %s+ returned %d after %Ld usecs\n", dev_name(dev),
+			error, (unsigned long long) ktime_to_ns(delta) >> 10);
+	}
+
 	return error;
 }
 
@@ -236,6 +251,12 @@ static int pm_noirq_op(struct device *dev,
 			pm_message_t state)
 {
 	int error = 0;
+	ktime_t calltime, delta, rettime;
+
+	if (initcall_debug) {
+		printk("calling  %s_i+ @ %i\n", dev_name(dev), task_pid_nr(current));
+		calltime = ktime_get();
+	}
 
 	switch (state.event) {
 #ifdef CONFIG_SUSPEND
@@ -283,6 +304,14 @@ static int pm_noirq_op(struct device *dev,
 	default:
 		error = -EINVAL;
 	}
+
+	if (initcall_debug) {
+		rettime = ktime_get();
+		delta = ktime_sub(rettime, calltime);
+		printk("initcall %s_i+ returned %d after %Ld usecs\n", dev_name(dev),
+			error, (unsigned long long) ktime_to_ns(delta) >> 10);
+	}
+
 	return error;
 }
 
include/linux/init.h +2 -0
index ff8bde5..ab1d31f 100644
@@ -149,6 +149,8 @@ void prepare_namespace(void);
 
 extern void (*late_time_init)(void);
 
+extern int initcall_debug;
+
 #endif
   
 #ifndef MODULE
Signature · 7 lines
1.6.2.5


-- 
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#1 lore raw
Quoted 47 lines · Arjan van de Ven

From fcd52b97aa02f19baa570e4a62dfdfab0faed976 Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <[email protected]>
Date: Sun, 6 Dec 2009 12:32:28 -0800
Subject: [PATCH] suspend: add initcall_debug style timing for suspend/resume

In order to diagnose overall suspend/resume times, we need basic instrumentation to break down the total time into per device timing, similar to initcall_debug.

This patch adds the basic timing instrumentation, needed for a scritps/bootgraph.pl equivalent or humans. The bootgraph.pl program is still a work in progress, but is far enough along to know that this patch is sufficient.

---
 drivers/base/power/main.c |   29 +++++++++++++++++++++++++++++
 include/linux/init.h      |    2 ++
 2 files changed, 31 insertions(+), 0 deletions(-)
drivers/base/power/main.c +7 -0
index 8aa2443..2ad345c 100644
@@ -25,6 +25,7 @@
 #include <linux/resume-trace.h>
 #include <linux/rwsem.h>
 #include <linux/interrupt.h>
+#include <linux/sched.h>
 
 #include "../base.h"
 #include "power.h"
@@ -172,6 +173,12 @@ static int pm_op(struct device *dev,
 		 pm_message_t state)
 {
 	int error = 0;
+	ktime_t calltime, delta, rettime;
+
+	if (initcall_debug) {
+		printk("calling  %s+ @ %i\n", dev_name(dev), task_pid_nr(current));
+		calltime = ktime_get();
+	}
 
 	switch (state.event) {
 #ifdef CONFIG_SUSPEND

drivers/base/power/main.c static int pm_op(struct device *dev, · line 234

Quoted 9 lines
drivers/base/power/main.c +6 -0
@@ -219,6 +226,14 @@ static int pm_op(struct device *dev,
 	default:
 		error = -EINVAL;
 	}
+
+	if (initcall_debug) {
+		rettime = ktime_get();
+		delta = ktime_sub(rettime, calltime);
+		printk("initcall %s+ returned %d after %Ld usecs\n", dev_name(dev),
+			error, (unsigned long long) ktime_to_ns(delta) >> 10);

I've changed that "initcall" to "call". Hopefully that won't break your scripts. ;-)

Rafael

#2 lore raw
Quoted 45 lines · Rafael J. Wysocki
Quoted 44 lines · Arjan van de Ven

From fcd52b97aa02f19baa570e4a62dfdfab0faed976 Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <[email protected]>
Date: Sun, 6 Dec 2009 12:32:28 -0800
Subject: [PATCH] suspend: add initcall_debug style timing for suspend/resume

In order to diagnose overall suspend/resume times, we need basic instrumentation to break down the total time into per device timing, similar to initcall_debug.

This patch adds the basic timing instrumentation, needed for a scritps/bootgraph.pl equivalent or humans. The bootgraph.pl program is still a work in progress, but is far enough along to know that this patch is sufficient.

---
 drivers/base/power/main.c |   29 +++++++++++++++++++++++++++++
 include/linux/init.h      |    2 ++
 2 files changed, 31 insertions(+), 0 deletions(-)
drivers/base/power/main.c +7 -0
index 8aa2443..2ad345c 100644
@@ -25,6 +25,7 @@
 #include <linux/resume-trace.h>
 #include <linux/rwsem.h>
 #include <linux/interrupt.h>
+#include <linux/sched.h>
 
 #include "../base.h"
 #include "power.h"
@@ -172,6 +173,12 @@ static int pm_op(struct device *dev,
 		 pm_message_t state)
 {
 	int error = 0;
+	ktime_t calltime, delta, rettime;
+
+	if (initcall_debug) {
+		printk("calling  %s+ @ %i\n", dev_name(dev), task_pid_nr(current));
+		calltime = ktime_get();
+	}
 
Quoted 14 lines

drivers/base/power/main.c static int pm_op(struct device *dev, · line 234

Quoted 12 lines
drivers/base/power/main.c +6 -0
 	switch (state.event) {
 #ifdef CONFIG_SUSPEND
@@ -219,6 +226,14 @@ static int pm_op(struct device *dev,
 	default:
 		error = -EINVAL;
 	}
+
+	if (initcall_debug) {
+		rettime = ktime_get();
+		delta = ktime_sub(rettime, calltime);
+		printk("initcall %s+ returned %d after %Ld usecs\n", dev_name(dev),
+			error, (unsigned long long) ktime_to_ns(delta) >> 10);

I've changed that "initcall" to "call". Hopefully that won't break your scripts. ;-)

In addition to that I've changed printk to pr_infor everywhere in the patch, is that fine with you?

Rafael

#3 lore raw
Quoted 61 lines · Rafael J. Wysocki
Quoted 45 lines · Rafael J. Wysocki
Quoted 44 lines · Arjan van de Ven

From fcd52b97aa02f19baa570e4a62dfdfab0faed976 Mon Sep 17 00:00:00 2001
From: Arjan van de Ven <[email protected]>
Date: Sun, 6 Dec 2009 12:32:28 -0800
Subject: [PATCH] suspend: add initcall_debug style timing for suspend/resume

In order to diagnose overall suspend/resume times, we need basic instrumentation to break down the total time into per device timing, similar to initcall_debug.

This patch adds the basic timing instrumentation, needed for a scritps/bootgraph.pl equivalent or humans. The bootgraph.pl program is still a work in progress, but is far enough along to know that this patch is sufficient.

---
 drivers/base/power/main.c |   29 +++++++++++++++++++++++++++++
 include/linux/init.h      |    2 ++
 2 files changed, 31 insertions(+), 0 deletions(-)
drivers/base/power/main.c +7 -0
index 8aa2443..2ad345c 100644
@@ -25,6 +25,7 @@
 #include <linux/resume-trace.h>
 #include <linux/rwsem.h>
 #include <linux/interrupt.h>
+#include <linux/sched.h>
 
 #include "../base.h"
 #include "power.h"
@@ -172,6 +173,12 @@ static int pm_op(struct device *dev,
 		 pm_message_t state)
 {
 	int error = 0;
+	ktime_t calltime, delta, rettime;
+
+	if (initcall_debug) {
+		printk("calling  %s+ @ %i\n", dev_name(dev), task_pid_nr(current));
+		calltime = ktime_get();
+	}
 
Quoted 14 lines

drivers/base/power/main.c static int pm_op(struct device *dev, · line 234

Quoted 12 lines
drivers/base/power/main.c +6 -0
 	switch (state.event) {
 #ifdef CONFIG_SUSPEND
@@ -219,6 +226,14 @@ static int pm_op(struct device *dev,
 	default:
 		error = -EINVAL;
 	}
+
+	if (initcall_debug) {
+		rettime = ktime_get();
+		delta = ktime_sub(rettime, calltime);
+		printk("initcall %s+ returned %d after %Ld usecs\n", dev_name(dev),
+			error, (unsigned long long) ktime_to_ns(delta) >> 10);

I've changed that "initcall" to "call". Hopefully that won't break your scripts. ;-)

Quoted 1 line

In addition to that I've changed printk to pr_infor

Argh. pr_info, that is.

Rafael

#4 lore raw
In reply to Rafael J. Wysocki
Quoted 2 lines · On Sat, 12 Dec 2009 19:27:55 +0100 "Rafael J. Wysocki" <[email protected]> wrote:

I've changed that "initcall" to "call". Hopefully that won't break your scripts. ;-)

it will, but I can adjust them no big deal

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org