[PATCH 2/2 v1.0]sched: Updating the sched-stat documentation
From: Ciju Rajan K
Date: Mon Jan 17 2011 - 05:54:11 EST
This patch updates the sched-stat documentation.
Signed-off-by: Ciju Rajan K <ciju@xxxxxxxxxxxxxxxxxx>
---
diff -Naurp a/Documentation/scheduler/sched-stats.txt b/Documentation/scheduler/sched-stats.txt
--- a/Documentation/scheduler/sched-stats.txt 2011-01-17 01:07:47.000000000 +0530
+++ b/Documentation/scheduler/sched-stats.txt 2011-01-17 15:32:05.000000000 +0530
@@ -26,119 +26,81 @@ Note that any such script will necessari
reason to change versions is changes in the output format. For those wishing
to write their own scripts, the fields are described here.
+The first two fields of /proc/schedstat indicates the version (current
+version is 16) and jiffies values. The following values are from
+cpu & domain statistics.
+
CPU statistics
--------------
-cpu<N> 1 2 3 4 5 6 7 8 9 10 11 12
+The format is like this:
+
+cpu<N> 1 2 3 4 5 6 7 8
-NOTE: In the sched_yield() statistics, the active queue is considered empty
- if it has only one process in it, since obviously the process calling
- sched_yield() is that process.
-
-First four fields are sched_yield() statistics:
- 1) # of times both the active and the expired queue were empty
- 2) # of times just the active queue was empty
- 3) # of times just the expired queue was empty
- 4) # of times sched_yield() was called
-
-Next three are schedule() statistics:
- 5) # of times we switched to the expired queue and reused it
- 6) # of times schedule() was called
- 7) # of times schedule() left the processor idle
-
-Next two are try_to_wake_up() statistics:
- 8) # of times try_to_wake_up() was called
- 9) # of times try_to_wake_up() was called to wake up the local cpu
-
-Next three are statistics describing scheduling latency:
- 10) sum of all time spent running by tasks on this processor (in jiffies)
- 11) sum of all time spent waiting to run by tasks on this processor (in
- jiffies)
- 12) # of timeslices run on this cpu
+NOTE: In the sched_yield() statistics, the active queue is considered
+ empty if it has only one process in it, since obviously the
+ process calling sched_yield() is that process.
+
+ 1) # of times sched_yield() was called on this CPU
+ 2) # of times scheduler runs on this CPU
+ 3) # of times scheduler picks idle task as next task on this CPU
+ 4) # of times try_to_wake_up() is run on this CPU
+ (Number of times task wakeup is attempted from this CPU)
+ 5) # of times try_to_wake_up() wakes up a task on the same CPU
+ (local wakeup)
+ 6) Time(ns) for which tasks have run on this CPU
+ 7) Time(ns) for which tasks on this CPU's runqueue have waited
+ before getting to run on the CPU
+ 8) # of tasks that have run on this CPU
Domain statistics
-----------------
-One of these is produced per domain for each cpu described. (Note that if
-CONFIG_SMP is not defined, *no* domains are utilized and these lines
-will not appear in the output.)
+One of these is produced per domain for each cpu described.
+(Note that if CONFIG_SMP is not defined, *no* domains are utilized
+ and these lines will not appear in the output.)
-domain<N> <cpumask> 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
+domain<N> <cpumask> 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
The first field is a bit mask indicating what cpus this domain operates over.
-The next 24 are a variety of load_balance() statistics in grouped into types
-of idleness (idle, busy, and newly idle):
+The next 24 are a variety of load_balance() statistics grouped into
+types of idleness (idle, busy, and newly idle). The three idle
+states are:
+
+CPU_NEWLY_IDLE: Load balancer is being run on a CPU which is
+ about to enter IDLE state
+CPU_IDLE: This state is entered after CPU_NEWLY_IDLE
+ state fails to find a new task for this CPU
+CPU_NOT_IDLE: Load balancer is being run on a CPU when it is
+ not in IDLE state (busy times)
+
+There are eight stats available for each of the three idle states:
- 1) # of times in this domain load_balance() was called when the
- cpu was idle
+ 1) # of times in this domain load_balance() was called
2) # of times in this domain load_balance() checked but found
- the load did not require balancing when the cpu was idle
+ the load did not require balancing
3) # of times in this domain load_balance() tried to move one or
- more tasks and failed, when the cpu was idle
+ more tasks and failed
4) sum of imbalances discovered (if any) with each call to
- load_balance() in this domain when the cpu was idle
- 5) # of times in this domain pull_task() was called when the cpu
- was idle
+ load_balance() in this domain
+ 5) # of times in this domain pull_task() was called
6) # of times in this domain pull_task() was called even though
- the target task was cache-hot when idle
+ the target task was cache-hot
7) # of times in this domain load_balance() was called but did
- not find a busier queue while the cpu was idle
- 8) # of times in this domain a busier queue was found while the
- cpu was idle but no busier group was found
-
- 9) # of times in this domain load_balance() was called when the
- cpu was busy
- 10) # of times in this domain load_balance() checked but found the
- load did not require balancing when busy
- 11) # of times in this domain load_balance() tried to move one or
- more tasks and failed, when the cpu was busy
- 12) sum of imbalances discovered (if any) with each call to
- load_balance() in this domain when the cpu was busy
- 13) # of times in this domain pull_task() was called when busy
- 14) # of times in this domain pull_task() was called even though the
- target task was cache-hot when busy
- 15) # of times in this domain load_balance() was called but did not
- find a busier queue while the cpu was busy
- 16) # of times in this domain a busier queue was found while the cpu
- was busy but no busier group was found
-
- 17) # of times in this domain load_balance() was called when the
- cpu was just becoming idle
- 18) # of times in this domain load_balance() checked but found the
- load did not require balancing when the cpu was just becoming idle
- 19) # of times in this domain load_balance() tried to move one or more
- tasks and failed, when the cpu was just becoming idle
- 20) sum of imbalances discovered (if any) with each call to
- load_balance() in this domain when the cpu was just becoming idle
- 21) # of times in this domain pull_task() was called when newly idle
- 22) # of times in this domain pull_task() was called even though the
- target task was cache-hot when just becoming idle
- 23) # of times in this domain load_balance() was called but did not
- find a busier queue while the cpu was just becoming idle
- 24) # of times in this domain a busier queue was found while the cpu
- was just becoming idle but no busier group was found
-
+ not find a busier queue
+ 8) # of times in this domain a busier queue was found but no
+ busier group was found
+
Next three are active_load_balance() statistics:
25) # of times active_load_balance() was called
26) # of times active_load_balance() tried to move a task and failed
27) # of times active_load_balance() successfully moved a task
- Next three are sched_balance_exec() statistics:
- 28) sbe_cnt is not used
- 29) sbe_balanced is not used
- 30) sbe_pushed is not used
-
- Next three are sched_balance_fork() statistics:
- 31) sbf_cnt is not used
- 32) sbf_balanced is not used
- 33) sbf_pushed is not used
-
- Next three are try_to_wake_up() statistics:
- 34) # of times in this domain try_to_wake_up() awoke a task that
+ Next two are try_to_wake_up() statistics:
+ 28) # of times in this domain try_to_wake_up() awoke a task that
last ran on a different cpu in this domain
- 35) # of times in this domain try_to_wake_up() moved a task to the
+ 29) # of times in this domain try_to_wake_up() moved a task to the
waking cpu because it was cache-cold on its own cpu anyway
- 36) # of times in this domain try_to_wake_up() started passive balancing
/proc/<pid>/schedstat
----------------