lkml.org 
[lkml]   [2011]   [Jan]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 2/2 v1.0]sched: Updating the sched-stat documentation
Hi Ciju,

(2011/01/18 15:04), Ciju Rajan K wrote:
> sched: Updating the sched-stat documentation
>
> Some of the unused fields are removed from /proc/schedstat.
> This is the documentation changes reflecting the same.
>
> Signed-off-by: Ciju Rajan K<ciju@linux.vnet.ibm.com>
> ---
>
> 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 & 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.

There are no active/expired queue any more.

> +
> + 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:

It's more helpful iffiled 1-8 are for CPU_NEWLY_IDLE, 9-15 are for
CPU_IDLE, and 16-24 is for CPU_NOT_IDLE. Current description doesn't
say whether

1) lb_count[CPU_NEWLY_IDLE],
2) lb_balanced[CPU_NEWLY_IDLE],
3) lb_failed[CPU_NEWLY_IDLE],
...

or

1) lb_count[CPU_NEWLY_IDLE]
2) lb_count[CPU_IDLE]
3) lb_count[CPU_NOT_IDLE]
...

Thanks,
Satoru

>
> - 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
> ----------------
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
>




\
 
 \ /
  Last update: 2011-01-19 08:15    [W:0.601 / U:0.448 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site