public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* [RESEND PATCH v2] nohz: fix idle ticks in cpu summary line of /proc/stat
@ 2012-09-10 11:13 Srivatsa S. Bhat
  2012-09-25 18:50 ` Srivatsa S. Bhat
  0 siblings, 1 reply; 3+ messages in thread
From: Srivatsa S. Bhat @ 2012-09-10 11:13 UTC (permalink / raw)
  To: tglx, mhocko; +Cc: schwidefsky, deepthi, linux-kernel, srivatsa.bhat

From: Michal Hocko <mhocko@suse.cz>

Git commit 09a1d34f8535ecf9 "nohz: Make idle/iowait counter update
conditional" introduced a bug in regard to cpu hotplug. The effect is
that the number of idle ticks in the cpu summary line in /proc/stat is
still counting ticks for offline cpus.

Reproduction is easy, just start a workload that keeps all cpus busy,
switch off one or more cpus and then watch the idle field in top.
On a dual-core with one cpu 100% busy and one offline cpu you will get
something like this:

%Cpu(s): 48.7 us,  1.3 sy,  0.0 ni, 50.0 id,  0.0 wa,  0.0 hi,  0.0 si,
%0.0 st

The problem is that an offline cpu still has ts->idle_active == 1.
To fix this we should make sure that the cpu is online when calling
get_cpu_idle_time_us and get_cpu_iowait_time_us.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Reported-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Michal Hocko <mhocko@suse.cz>
[srivatsa.bhat@linux.vnet.ibm.com: Rebased to current mainline]
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
---

Hi Thomas,

This is a resend of the patch posted by Michal at [1]. Martin had explained
the importance of this patch for fixing the bug for x86 case in [2]. (The s390
fix is already upstream, commit id cb85a6ed67e9). Could you kindly consider
taking this fix?

[1]. http://thread.gmane.org/gmane.linux.kernel/1265374/focus=1266457
[2]. http://thread.gmane.org/gmane.linux.kernel/1265374/focus=1276336

 fs/proc/stat.c |   14 ++++++++++----
 1 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index 64c3b31..e296572 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -45,10 +45,13 @@ static cputime64_t get_iowait_time(int cpu)
 
 static u64 get_idle_time(int cpu)
 {
-	u64 idle, idle_time = get_cpu_idle_time_us(cpu, NULL);
+	u64 idle, idle_time = -1ULL;
+
+	if (cpu_online(cpu))
+		idle_time = get_cpu_idle_time_us(cpu, NULL);
 
 	if (idle_time == -1ULL)
-		/* !NO_HZ so we can rely on cpustat.idle */
+		/* !NO_HZ or cpu offline so we can rely on cpustat.idle */
 		idle = kcpustat_cpu(cpu).cpustat[CPUTIME_IDLE];
 	else
 		idle = usecs_to_cputime64(idle_time);
@@ -58,10 +61,13 @@ static u64 get_idle_time(int cpu)
 
 static u64 get_iowait_time(int cpu)
 {
-	u64 iowait, iowait_time = get_cpu_iowait_time_us(cpu, NULL);
+	u64 iowait, iowait_time = -1ULL;
+
+	if (cpu_online(cpu))
+		iowait_time = get_cpu_iowait_time_us(cpu, NULL);
 
 	if (iowait_time == -1ULL)
-		/* !NO_HZ so we can rely on cpustat.iowait */
+		/* !NO_HZ or cpu offline so we can rely on cpustat.iowait */
 		iowait = kcpustat_cpu(cpu).cpustat[CPUTIME_IOWAIT];
 	else
 		iowait = usecs_to_cputime64(iowait_time);


^ permalink raw reply related	[flat|nested] 3+ messages in thread
* [RESEND PATCH v2] nohz: fix idle ticks in cpu summary line of /proc/stat
@ 2012-10-10  6:21 Srivatsa S. Bhat
  0 siblings, 0 replies; 3+ messages in thread
From: Srivatsa S. Bhat @ 2012-10-10  6:21 UTC (permalink / raw)
  To: tglx, akpm
  Cc: mhocko, mingo, schwidefsky, deepthi, linux-kernel, srivatsa.bhat

From: Michal Hocko <mhocko@suse.cz>

Git commit 09a1d34f8535ecf9 "nohz: Make idle/iowait counter update
conditional" introduced a bug in regard to cpu hotplug. The effect is
that the number of idle ticks in the cpu summary line in /proc/stat is
still counting ticks for offline cpus.

Reproduction is easy, just start a workload that keeps all cpus busy,
switch off one or more cpus and then watch the idle field in top.
On a dual-core with one cpu 100% busy and one offline cpu you will get
something like this:

%Cpu(s): 48.7 us,  1.3 sy,  0.0 ni, 50.0 id,  0.0 wa,  0.0 hi,  0.0 si,
%0.0 st

The problem is that an offline cpu still has ts->idle_active == 1.
To fix this we should make sure that the cpu is online when calling
get_cpu_idle_time_us and get_cpu_iowait_time_us.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Reported-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Michal Hocko <mhocko@suse.cz>
[Srivatsa: Rebased to current mainline]
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
---

This is a resend of the patch posted by Michal at [1]. Martin had explained
the importance of this patch for fixing the bug for x86 case in [2]. (The
s390 fix is already upstream, commit id cb85a6ed67e9).

[1]. http://thread.gmane.org/gmane.linux.kernel/1265374/focus=1266457
[2]. http://thread.gmane.org/gmane.linux.kernel/1265374/focus=1276336

 fs/proc/stat.c |   14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index 64c3b31..e296572 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -45,10 +45,13 @@ static cputime64_t get_iowait_time(int cpu)
 
 static u64 get_idle_time(int cpu)
 {
-	u64 idle, idle_time = get_cpu_idle_time_us(cpu, NULL);
+	u64 idle, idle_time = -1ULL;
+
+	if (cpu_online(cpu))
+		idle_time = get_cpu_idle_time_us(cpu, NULL);
 
 	if (idle_time == -1ULL)
-		/* !NO_HZ so we can rely on cpustat.idle */
+		/* !NO_HZ or cpu offline so we can rely on cpustat.idle */
 		idle = kcpustat_cpu(cpu).cpustat[CPUTIME_IDLE];
 	else
 		idle = usecs_to_cputime64(idle_time);
@@ -58,10 +61,13 @@ static u64 get_idle_time(int cpu)
 
 static u64 get_iowait_time(int cpu)
 {
-	u64 iowait, iowait_time = get_cpu_iowait_time_us(cpu, NULL);
+	u64 iowait, iowait_time = -1ULL;
+
+	if (cpu_online(cpu))
+		iowait_time = get_cpu_iowait_time_us(cpu, NULL);
 
 	if (iowait_time == -1ULL)
-		/* !NO_HZ so we can rely on cpustat.iowait */
+		/* !NO_HZ or cpu offline so we can rely on cpustat.iowait */
 		iowait = kcpustat_cpu(cpu).cpustat[CPUTIME_IOWAIT];
 	else
 		iowait = usecs_to_cputime64(iowait_time);


^ permalink raw reply related	[flat|nested] 3+ messages in thread
end of thread, other threads:[~2012-10-10  6:22 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-09-10 11:13 [RESEND PATCH v2] nohz: fix idle ticks in cpu summary line of /proc/stat Srivatsa S. Bhat
2012-09-25 18:50 ` Srivatsa S. Bhat
  -- strict thread matches above, loose matches on Subject: below --
2012-10-10  6:21 Srivatsa S. Bhat

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox