cpuacct: per-cgroup utime/stime statistics - v6
PATCH 1/1 cpuacct: per-cgroup utime/stime statistics - v6
Here is the next version which fixes the bug pointed by Li Zefan in v5.
We currently have two stats named stime and utime. In future if we decide to have more fine grained stats like time spent in irq, softirq etc, it will be difficult to name them similar to stime and utime. Hence would it make sense to name these stats as system and user so that we could easily add nice, irq, steal, softirq etc in future ?
Regards,
Bharata.
cpuacct: Add stime and utime statistics
Add per-cgroup cpuacct controller statistics like the system and user time consumed by the group of tasks.
Changelog:
v6
- Fixed a bug in the error path of cpuacct_create() (pointed by Li Zefan).
v5
- In cpuacct_stats_show(), use cputime64_to_clock_t() since we are operating on a 64bit variable here.
v4
- Remove comments in cpuacct_update_stats() which explained why rcu_read_lock() was needed (as per Peter Zijlstra's review comments).
- Don't say that percpu_counter_read() is broken in Documentation/cpuacct.txt as per KAMEZAWA Hiroyuki's review comments.
v3
- Fix a small race in the cpuacct hierarchy walk.
v2
- stime and utime now exported in clock_t units instead of msecs.
- Addressed the code review comments from Balbir and Li Zefan.
- Moved to -tip tree.
v1
- Moved the stime/utime accounting to cpuacct controller.
Earlier versions
- http://lkml.org/lkml/2009/2/25/129
--- Documentation/cgroups/cpuacct.txt | 18 +++++++ kernel/sched.c | 87 +++++++++++++++++++++++++++++++++++--- 2 files changed, 99 insertions(+), 6 deletions(-)
Documentation/cgroups/cpuacct.txt +18 -0
@@ -30,3 +30,21 @@ The above steps create a new group g1 an process (bash) into it. CPU time consumed by this bash and its children can be obtained from g1/cpuacct.usage and the same is accumulated in /cgroups/cpuacct.usage also. + +cpuacct.stat file lists a few statistics which further divide the +CPU time obtained by the cgroup into user and system times. Currently +the following statistics are supported: + +utime: Time spent by tasks of the cgroup in user mode. +stime: Time spent by tasks of the cgroup in kernel mode. + +utime and stime are in USER_HZ unit. + +cpuacct controller uses percpu_counter interface to collect utime and +stime. This causes two side effects: + +- It is theoretically possible to see wrong values for stime and utime. + This is because percpu_counter_read() on 32bit systems isn't safe + against concurrent writes. +- It is possible to see slightly outdated values for stime and utime + due to the batch processing nature of percpu_counter.
kernel/sched.c +81 -6
155 lines not on this page · raw message
Quoted 65 lines · Bharata B Rao
Here is the next version which fixes the bug pointed by Li Zefan in v5.
We currently have two stats named stime and utime. In future if we decide to have more fine grained stats like time spent in irq, softirq etc, it will be difficult to name them similar to stime and utime. Hence would it make sense to name these stats as system and user so that we could easily add nice, irq, steal, softirq etc in future ?
Regards,
Bharata.cpuacct: Add stime and utime statistics
Add per-cgroup cpuacct controller statistics like the system and user time consumed by the group of tasks.
Changelog:
v6
- Fixed a bug in the error path ofcpuacct_create()(pointed by Li Zefan).v5
- Incpuacct_stats_show(), usecputime64_to_clock_t()since we are operating on a 64bit variable here.v4
- Remove comments incpuacct_update_stats()which explained whyrcu_read_lock()was needed (as per Peter Zijlstra's review comments).
- Don't say thatpercpu_counter_read()is broken inDocumentation/cpuacct.txtas per KAMEZAWA Hiroyuki's review comments.v3
- Fix a small race in the cpuacct hierarchy walk.v2
- stime and utime now exported inclock_tunits instead of msecs.
- Addressed the code review comments from Balbir and Li Zefan.
- Moved to -tip tree.v1
- Moved the stime/utime accounting to cpuacct controller.Earlier versions
- http://lkml.org/lkml/2009/2/25/129Signed-off-by Bharata B Rao <[email protected]>Signed-off-by Balaji Rao <[email protected]>Cc Dhaval Giani <[email protected]>Cc Paul Menage <[email protected]>Cc Andrew Morton <[email protected]>Cc Ingo Molnar <[email protected]>Cc KAMEZAWA Hiroyuki <[email protected]>Reviewed-by Li Zefan <[email protected]>Acked-by Peter Zijlstra <[email protected]>Acked-by Balbir Singh <[email protected]>Tested-by Balbir Singh <[email protected]>--- Documentation/cgroups/cpuacct.txt | 18 +++++++ kernel/sched.c | 87 +++++++++++++++++++++++++++++++++++--- 2 files changed, 99 insertions(+), 6 deletions(-)Documentation/cgroups/cpuacct.txt +0
-0@@ -30,3 +30,21 @@ The above steps create a new group g1 an process (bash) into it. CPU time consumed by this bash and its children
Documentation/cgroups/cpuacct.txt The above steps create a new group g1 an · line 40
Quoted 10 lines
Documentation/cgroups/cpuacct.txt +8
-0can be obtained from g1/cpuacct.usage and the same is accumulated in /cgroups/cpuacct.usage also. + +cpuacct.stat file lists a few statistics which further divide the +CPU time obtained by the cgroup into user and system times. Currently +the following statistics are supported: + +utime: Time spent by tasks of the cgroup in user mode. +stime: Time spent by tasks of the cgroup in kernel mode. +
Bharata can we move away from stime, utime (names) to system and user with time being an implicit unit. I would like to see softirq, hardirq, etc times in the future being added.
Signature · 1 line
Balbir
Quoted 73 lines · Balbir Singh
Quoted 71 lines · Bharata B Rao
Here is the next version which fixes the bug pointed by Li Zefan in v5.
We currently have two stats named stime and utime. In future if we decide to have more fine grained stats like time spent in irq, softirq etc, it will be difficult to name them similar to stime and utime. Hence would it make sense to name these stats as system and user so that we could easily add nice, irq, steal, softirq etc in future ?
Regards,
Bharata.cpuacct: Add stime and utime statistics
Add per-cgroup cpuacct controller statistics like the system and user time consumed by the group of tasks.
Changelog:
v6
- Fixed a bug in the error path ofcpuacct_create()(pointed by Li Zefan).v5
- Incpuacct_stats_show(), usecputime64_to_clock_t()since we are operating on a 64bit variable here.v4
- Remove comments incpuacct_update_stats()which explained whyrcu_read_lock()was needed (as per Peter Zijlstra's review comments).
- Don't say thatpercpu_counter_read()is broken inDocumentation/cpuacct.txtas per KAMEZAWA Hiroyuki's review comments.v3
- Fix a small race in the cpuacct hierarchy walk.v2
- stime and utime now exported inclock_tunits instead of msecs.
- Addressed the code review comments from Balbir and Li Zefan.
- Moved to -tip tree.v1
- Moved the stime/utime accounting to cpuacct controller.Earlier versions
- http://lkml.org/lkml/2009/2/25/129Signed-off-by Bharata B Rao <[email protected]>Signed-off-by Balaji Rao <[email protected]>Cc Dhaval Giani <[email protected]>Cc Paul Menage <[email protected]>Cc Andrew Morton <[email protected]>Cc Ingo Molnar <[email protected]>Cc KAMEZAWA Hiroyuki <[email protected]>Reviewed-by Li Zefan <[email protected]>Acked-by Peter Zijlstra <[email protected]>Acked-by Balbir Singh <[email protected]>Tested-by Balbir Singh <[email protected]>--- Documentation/cgroups/cpuacct.txt | 18 +++++++ kernel/sched.c | 87 +++++++++++++++++++++++++++++++++++--- 2 files changed, 99 insertions(+), 6 deletions(-)Documentation/cgroups/cpuacct.txt +4
-0@@ -30,3 +30,21 @@ The above steps create a new group g1 an process (bash) into it. CPU time consumed by this bash and its children can be obtained from g1/cpuacct.usage and the same is accumulated in /cgroups/cpuacct.usage also. + +cpuacct.stat file lists a few statistics which further divide the +CPU time obtained by the cgroup into user and system times. Currently +the following statistics are supported:
Quoted 8 lines
Documentation/cgroups/cpuacct.txt The above steps create a new group g1 an · line 40
Quoted 4 lines
Documentation/cgroups/cpuacct.txt +4
-0+ +utime: Time spent by tasks of the cgroup in user mode. +stime: Time spent by tasks of the cgroup in kernel mode. +Bharata can we move away from stime, utime (names) to system and user with time being an implicit unit. I would like to see softirq, hardirq, etc times in the future being added.
Done. Posted next version with this change.
Regards,
Bharata.