/proc/uptime reports idle time by reading the CPUTIME_IDLE field from
the per-cpu kcpustats. However, on NO_HZ systems, idle time is not
continually updated on idle cpus, leading this value to appear
incorrectly small.
/proc/stat performs an accounting update when reading idle time; we can
use the same approach for uptime.
With this patch, /proc/stat and /proc/uptime now agree on idle time.
Additionally, the following shows idle time tick up consistently on an
idle machine:
(while true; do cat /proc/uptime; sleep 1; done) | awk '{print $2-prev; prev=$2}'
Reported-by: Luigi Rizzo <[email protected]>
Signed-off-by: Josh Don <[email protected]>
---
v2:
- Move get_idle_time() from cputime.c back into stat.c
- Use kcpustat_cpu_fetch
fs/proc/stat.c | 4 ++--
fs/proc/uptime.c | 14 +++++++++-----
include/linux/kernel_stat.h | 1 +
3 files changed, 12 insertions(+), 7 deletions(-)
diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index 6561a06ef905..4fb8729a68d4 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -24,7 +24,7 @@
#ifdef arch_idle_time
-static u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
+u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
{
u64 idle;
@@ -46,7 +46,7 @@ static u64 get_iowait_time(struct kernel_cpustat *kcs, int cpu)
#else
-static u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
+u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
{
u64 idle, idle_usecs = -1ULL;
diff --git a/fs/proc/uptime.c b/fs/proc/uptime.c
index 5a1b228964fb..deb99bc9b7e6 100644
--- a/fs/proc/uptime.c
+++ b/fs/proc/uptime.c
@@ -12,18 +12,22 @@ static int uptime_proc_show(struct seq_file *m, void *v)
{
struct timespec64 uptime;
struct timespec64 idle;
- u64 nsec;
+ u64 idle_nsec;
u32 rem;
int i;
- nsec = 0;
- for_each_possible_cpu(i)
- nsec += (__force u64) kcpustat_cpu(i).cpustat[CPUTIME_IDLE];
+ idle_nsec = 0;
+ for_each_possible_cpu(i) {
+ struct kernel_cpustat kcs;
+
+ kcpustat_cpu_fetch(&kcs, i);
+ idle_nsec += get_idle_time(&kcs, i);
+ }
ktime_get_boottime_ts64(&uptime);
timens_add_boottime(&uptime);
- idle.tv_sec = div_u64_rem(nsec, NSEC_PER_SEC, &rem);
+ idle.tv_sec = div_u64_rem(idle_nsec, NSEC_PER_SEC, &rem);
idle.tv_nsec = rem;
seq_printf(m, "%lu.%02lu %lu.%02lu\n",
(unsigned long) uptime.tv_sec,
diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
index 44ae1a7eb9e3..69ae6b278464 100644
--- a/include/linux/kernel_stat.h
+++ b/include/linux/kernel_stat.h
@@ -102,6 +102,7 @@ extern void account_system_index_time(struct task_struct *, u64,
enum cpu_usage_stat);
extern void account_steal_time(u64);
extern void account_idle_time(u64);
+extern u64 get_idle_time(struct kernel_cpustat *kcs, int cpu);
#ifdef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
static inline void account_process_tick(struct task_struct *tsk, int user)
--
2.33.0.259.gc128427fd7-goog
On Fri, Aug 27, 2021 at 9:55 AM Josh Don <[email protected]> wrote:
>
> /proc/uptime reports idle time by reading the CPUTIME_IDLE field from
> the per-cpu kcpustats. However, on NO_HZ systems, idle time is not
> continually updated on idle cpus, leading this value to appear
> incorrectly small.
>
> /proc/stat performs an accounting update when reading idle time; we can
> use the same approach for uptime.
>
> With this patch, /proc/stat and /proc/uptime now agree on idle time.
> Additionally, the following shows idle time tick up consistently on an
> idle machine:
> (while true; do cat /proc/uptime; sleep 1; done) | awk '{print $2-prev; prev=$2}'
>
> Reported-by: Luigi Rizzo <[email protected]>
> Signed-off-by: Josh Don <[email protected]>
> ---
>
SGTM thanks !
Reviewed-by: Eric Dumazet <[email protected]>
The following commit has been merged into the sched/core branch of tip:
Commit-ID: 1f74f9ea5a0cc9147f5bda4b5715ccb5cc34605d
Gitweb: https://git.kernel.org/tip/1f74f9ea5a0cc9147f5bda4b5715ccb5cc34605d
Author: Josh Don <[email protected]>
AuthorDate: Fri, 27 Aug 2021 09:54:38 -07:00
Committer: Peter Zijlstra <[email protected]>
CommitterDate: Thu, 09 Sep 2021 11:27:31 +02:00
fs/proc/uptime.c: Fix idle time reporting in /proc/uptime
/proc/uptime reports idle time by reading the CPUTIME_IDLE field from
the per-cpu kcpustats. However, on NO_HZ systems, idle time is not
continually updated on idle cpus, leading this value to appear
incorrectly small.
/proc/stat performs an accounting update when reading idle time; we
can use the same approach for uptime.
With this patch, /proc/stat and /proc/uptime now agree on idle time.
Additionally, the following shows idle time tick up consistently on an
idle machine:
(while true; do cat /proc/uptime; sleep 1; done) | awk '{print $2-prev; prev=$2}'
Reported-by: Luigi Rizzo <[email protected]>
Signed-off-by: Josh Don <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Eric Dumazet <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
fs/proc/stat.c | 4 ++--
fs/proc/uptime.c | 14 +++++++++-----
include/linux/kernel_stat.h | 1 +
3 files changed, 12 insertions(+), 7 deletions(-)
diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index 6561a06..4fb8729 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -24,7 +24,7 @@
#ifdef arch_idle_time
-static u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
+u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
{
u64 idle;
@@ -46,7 +46,7 @@ static u64 get_iowait_time(struct kernel_cpustat *kcs, int cpu)
#else
-static u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
+u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
{
u64 idle, idle_usecs = -1ULL;
diff --git a/fs/proc/uptime.c b/fs/proc/uptime.c
index 5a1b228..deb99bc 100644
--- a/fs/proc/uptime.c
+++ b/fs/proc/uptime.c
@@ -12,18 +12,22 @@ static int uptime_proc_show(struct seq_file *m, void *v)
{
struct timespec64 uptime;
struct timespec64 idle;
- u64 nsec;
+ u64 idle_nsec;
u32 rem;
int i;
- nsec = 0;
- for_each_possible_cpu(i)
- nsec += (__force u64) kcpustat_cpu(i).cpustat[CPUTIME_IDLE];
+ idle_nsec = 0;
+ for_each_possible_cpu(i) {
+ struct kernel_cpustat kcs;
+
+ kcpustat_cpu_fetch(&kcs, i);
+ idle_nsec += get_idle_time(&kcs, i);
+ }
ktime_get_boottime_ts64(&uptime);
timens_add_boottime(&uptime);
- idle.tv_sec = div_u64_rem(nsec, NSEC_PER_SEC, &rem);
+ idle.tv_sec = div_u64_rem(idle_nsec, NSEC_PER_SEC, &rem);
idle.tv_nsec = rem;
seq_printf(m, "%lu.%02lu %lu.%02lu\n",
(unsigned long) uptime.tv_sec,
diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
index 44ae1a7..69ae6b2 100644
--- a/include/linux/kernel_stat.h
+++ b/include/linux/kernel_stat.h
@@ -102,6 +102,7 @@ extern void account_system_index_time(struct task_struct *, u64,
enum cpu_usage_stat);
extern void account_steal_time(u64);
extern void account_idle_time(u64);
+extern u64 get_idle_time(struct kernel_cpustat *kcs, int cpu);
#ifdef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
static inline void account_process_tick(struct task_struct *tsk, int user)
The following commit has been merged into the sched/core branch of tip:
Commit-ID: a130e8fbc7de796eb6e680724d87f4737a26d0ac
Gitweb: https://git.kernel.org/tip/a130e8fbc7de796eb6e680724d87f4737a26d0ac
Author: Josh Don <[email protected]>
AuthorDate: Fri, 27 Aug 2021 09:54:38 -07:00
Committer: Peter Zijlstra <[email protected]>
CommitterDate: Tue, 05 Oct 2021 15:51:35 +02:00
fs/proc/uptime.c: Fix idle time reporting in /proc/uptime
/proc/uptime reports idle time by reading the CPUTIME_IDLE field from
the per-cpu kcpustats. However, on NO_HZ systems, idle time is not
continually updated on idle cpus, leading this value to appear
incorrectly small.
/proc/stat performs an accounting update when reading idle time; we
can use the same approach for uptime.
With this patch, /proc/stat and /proc/uptime now agree on idle time.
Additionally, the following shows idle time tick up consistently on an
idle machine:
(while true; do cat /proc/uptime; sleep 1; done) | awk '{print $2-prev; prev=$2}'
Reported-by: Luigi Rizzo <[email protected]>
Signed-off-by: Josh Don <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Eric Dumazet <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
---
fs/proc/stat.c | 4 ++--
fs/proc/uptime.c | 14 +++++++++-----
include/linux/kernel_stat.h | 1 +
3 files changed, 12 insertions(+), 7 deletions(-)
diff --git a/fs/proc/stat.c b/fs/proc/stat.c
index 6561a06..4fb8729 100644
--- a/fs/proc/stat.c
+++ b/fs/proc/stat.c
@@ -24,7 +24,7 @@
#ifdef arch_idle_time
-static u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
+u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
{
u64 idle;
@@ -46,7 +46,7 @@ static u64 get_iowait_time(struct kernel_cpustat *kcs, int cpu)
#else
-static u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
+u64 get_idle_time(struct kernel_cpustat *kcs, int cpu)
{
u64 idle, idle_usecs = -1ULL;
diff --git a/fs/proc/uptime.c b/fs/proc/uptime.c
index 5a1b228..deb99bc 100644
--- a/fs/proc/uptime.c
+++ b/fs/proc/uptime.c
@@ -12,18 +12,22 @@ static int uptime_proc_show(struct seq_file *m, void *v)
{
struct timespec64 uptime;
struct timespec64 idle;
- u64 nsec;
+ u64 idle_nsec;
u32 rem;
int i;
- nsec = 0;
- for_each_possible_cpu(i)
- nsec += (__force u64) kcpustat_cpu(i).cpustat[CPUTIME_IDLE];
+ idle_nsec = 0;
+ for_each_possible_cpu(i) {
+ struct kernel_cpustat kcs;
+
+ kcpustat_cpu_fetch(&kcs, i);
+ idle_nsec += get_idle_time(&kcs, i);
+ }
ktime_get_boottime_ts64(&uptime);
timens_add_boottime(&uptime);
- idle.tv_sec = div_u64_rem(nsec, NSEC_PER_SEC, &rem);
+ idle.tv_sec = div_u64_rem(idle_nsec, NSEC_PER_SEC, &rem);
idle.tv_nsec = rem;
seq_printf(m, "%lu.%02lu %lu.%02lu\n",
(unsigned long) uptime.tv_sec,
diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h
index 44ae1a7..69ae6b2 100644
--- a/include/linux/kernel_stat.h
+++ b/include/linux/kernel_stat.h
@@ -102,6 +102,7 @@ extern void account_system_index_time(struct task_struct *, u64,
enum cpu_usage_stat);
extern void account_steal_time(u64);
extern void account_idle_time(u64);
+extern u64 get_idle_time(struct kernel_cpustat *kcs, int cpu);
#ifdef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
static inline void account_process_tick(struct task_struct *tsk, int user)