1
0
Fork 0

[PATCH] sched: update Documentation/sched-stats.txt

While learning about schedstats I found that the documentation in the tree
is old.  I updated it and found some interesting stuff like schedstats
version 14 is the same as version and version 13 never saw a kernel
release!  Also there are 6 fields in the current schedstats that are not
used anymore.  Nick had made them irrelevant in commit
476d139c21 but never removed them.

Thanks to Rick's perl script who I borrowed some of the updated descriptions
from.

Signed-off-by: Joachim Deguara <joachim.deguara@amd.com>
Acked-by: Nick Piggin <npiggin@suse.de>
Cc: Rick Lindsley <ricklind@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
hifive-unleashed-5.1
Joachim Deguara 2007-07-26 13:40:43 +02:00 committed by Ingo Molnar
parent f337346193
commit b762f3ffb7
1 changed files with 94 additions and 91 deletions

View File

@ -1,10 +1,11 @@
Version 10 of schedstats includes support for sched_domains, which
hit the mainline kernel in 2.6.7. Some counters make more sense to be
per-runqueue; other to be per-domain. Note that domains (and their associated
information) will only be pertinent and available on machines utilizing
CONFIG_SMP.
Version 14 of schedstats includes support for sched_domains, which hit the
mainline kernel in 2.6.20 although it is identical to the stats from version
12 which was in the kernel from 2.6.13-2.6.19 (version 13 never saw a kernel
release). Some counters make more sense to be per-runqueue; other to be
per-domain. Note that domains (and their associated information) will only
be pertinent and available on machines utilizing CONFIG_SMP.
In version 10 of schedstat, there is at least one level of domain
In version 14 of schedstat, there is at least one level of domain
statistics for each cpu listed, and there may well be more than one
domain. Domains have no particular names in this implementation, but
the highest numbered one typically arbitrates balancing across all the
@ -27,7 +28,7 @@ to write their own scripts, the fields are described here.
CPU statistics
--------------
cpu<N> 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
cpu<N> 1 2 3 4 5 6 7 8 9 10 11 12
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
@ -39,48 +40,20 @@ First four fields are sched_yield() statistics:
3) # of times just the expired queue was empty
4) # of times sched_yield() was called
Next four are schedule() statistics:
5) # of times the active queue had at least one other process on it
6) # of times we switched to the expired queue and reused it
7) # of times schedule() was called
8) # of times schedule() left the processor idle
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 four are active_load_balance() statistics:
9) # of times active_load_balance() was called
10) # of times active_load_balance() caused this cpu to gain a task
11) # of times active_load_balance() caused this cpu to lose a task
12) # of times active_load_balance() tried to move a task and failed
Next three are try_to_wake_up() statistics:
13) # of times try_to_wake_up() was called
14) # of times try_to_wake_up() successfully moved the awakening task
15) # of times try_to_wake_up() attempted to move the awakening task
Next two are wake_up_new_task() statistics:
16) # of times wake_up_new_task() was called
17) # of times wake_up_new_task() successfully moved the new task
Next one is a sched_migrate_task() statistic:
18) # of times sched_migrate_task() was called
Next one is a sched_balance_exec() statistic:
19) # of times sched_balance_exec() was called
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:
20) sum of all time spent running by tasks on this processor (in ms)
21) sum of all time spent waiting to run by tasks on this processor (in ms)
22) # of tasks (not necessarily unique) given to the processor
The last six are statistics dealing with pull_task():
23) # of times pull_task() moved a task to this cpu when newly idle
24) # of times pull_task() stole a task from this cpu when another cpu
was newly idle
25) # of times pull_task() moved a task to this cpu when idle
26) # of times pull_task() stole a task from this cpu when another cpu
was idle
27) # of times pull_task() moved a task to this cpu when busy
28) # of times pull_task() stole a task from this cpu when another cpu
was busy
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
Domain statistics
@ -89,65 +62,95 @@ 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> 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
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
The first field is a bit mask indicating what cpus this domain operates over.
The next fifteen are a variety of load_balance() statistics:
The next 24 are a variety of load_balance() statistics in grouped into types
of idleness (idle, busy, and newly idle):
1) # of times in this domain load_balance() was called when the cpu
was idle
2) # of times in this domain load_balance() was called when the cpu
was busy
3) # of times in this domain load_balance() was called when the cpu
was just becoming idle
4) # of times in this domain load_balance() tried to move one or more
tasks and failed, when the cpu was idle
5) # of times in this domain load_balance() tried to move one or more
tasks and failed, when the cpu was busy
6) # of times in this domain load_balance() tried to move one or more
tasks and failed, when the cpu was just becoming idle
7) sum of imbalances discovered (if any) with each call to
load_balance() in this domain when the cpu was idle
8) sum of imbalances discovered (if any) with each call to
load_balance() in this domain when the cpu was busy
9) sum of imbalances discovered (if any) with each call to
load_balance() in this domain when the cpu was just becoming idle
10) # of times in this domain load_balance() was called but did not find
a busier queue while the cpu was idle
11) # of times in this domain load_balance() was called but did not find
a busier queue while the cpu was busy
12) # of times in this domain load_balance() was called but did not find
a busier queue while the cpu was just becoming idle
13) # of times in this domain a busier queue was found while the cpu was
idle but no busier group was found
14) # of times in this domain a busier queue was found while the cpu was
busy but no busier group was found
15) # of times in this domain a busier queue was found while the cpu was
just becoming idle but no busier group was found
1) # of times in this domain load_balance() was called when the
cpu was idle
2) # of times in this domain load_balance() checked but found
the load did not require balancing when the cpu was idle
3) # of times in this domain load_balance() tried to move one or
more tasks and failed, when the cpu was idle
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
6) # of times in this domain pull_task() was called even though
the target task was cache-hot when idle
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
Next two are sched_balance_exec() statistics:
17) # of times in this domain sched_balance_exec() successfully pushed
a task to a new cpu
18) # of times in this domain sched_balance_exec() tried but failed to
push a task to a new cpu
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
Next two are try_to_wake_up() statistics:
19) # of times in this domain try_to_wake_up() tried to move a task based
on affinity and cache warmth
20) # of times in this domain try_to_wake_up() tried to move a task based
on load balancing
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
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
last ran on a different cpu in this domain
35) # 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
----------------
schedstats also adds a new /proc/<pid/schedstat file to include some of
the same information on a per-process level. There are three fields in
this file correlating to fields 20, 21, and 22 in the CPU fields, but
they only apply for that process.
this file correlating for that process to:
1) time spent on the cpu
2) time spent waiting on a runqueue
3) # of timeslices run on this cpu
A program could be easily written to make use of these extra fields to
report on how well a particular process or set of processes is faring
under the scheduler's policies. A simple version of such a program is
available at
http://eaglet.rain.com/rick/linux/schedstat/v10/latency.c
http://eaglet.rain.com/rick/linux/schedstat/v12/latency.c