LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
* [PATCH 3/3] memcgoup: allow memory.failcnt to be reset
@ 2008-03-11 10:09 Li Zefan
  2008-03-11 10:16 ` KAMEZAWA Hiroyuki
  0 siblings, 1 reply; 4+ messages in thread
From: Li Zefan @ 2008-03-11 10:09 UTC (permalink / raw)
  To: Andrew Morton
  Cc: Balbir Singh, KAMEZAWA Hiroyuki, Paul Menage, Pavel Emelianov,
	LKML, linux-mm, Linux Containers

Allow memory.failcnt to be reset to 0:

        echo 0 > memory.failcnt

And '0' is the only valid value.

This is useful when testing or observing the memory resource
controller. Without this function, one will have to remember
the previous failcnt to decide whether memory reclaim has
happened *again*.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
---
 Documentation/controllers/memory.txt |    4 +++-
 mm/memcontrol.c                      |   15 +++++++++++++++
 2 files changed, 18 insertions(+), 1 deletions(-)

diff --git a/Documentation/controllers/memory.txt b/Documentation/controllers/memory.txt
index 866b9cd..28f80e3 100644
--- a/Documentation/controllers/memory.txt
+++ b/Documentation/controllers/memory.txt
@@ -194,7 +194,9 @@ this file after a write to guarantee the value committed by the kernel.
 4096
 
 The memory.failcnt field gives the number of times that the cgroup limit was
-exceeded.
+exceeded. It can be reset.
+
+# echo 0 > memory.failcnt
 
 The memory.stat file gives accounting information. Now, the number of
 caches, RSS and Active pages/Inactive pages are shown.
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 6145031..fd26dc2 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -883,6 +883,20 @@ static int mem_force_empty_write(struct cgroup *cont, struct cftype *cft,
 	return ret;
 }
 
+static int mem_failcnt_write(struct cgroup *cont, struct cftype *cft,
+			     u64 val)
+{
+	struct res_counter *counter;
+
+	if (val != 0)
+		return -EINVAL;
+
+	counter = &mem_cgroup_from_cont(cont)->res;
+	res_counter_write_u64(counter, cft->private, 0);
+
+	return 0;
+}
+
 static const struct mem_cgroup_stat_desc {
 	const char *msg;
 	u64 unit;
@@ -934,6 +948,7 @@ static struct cftype mem_cgroup_files[] = {
 	{
 		.name = "failcnt",
 		.private = RES_FAILCNT,
+		.write_u64 = mem_failcnt_write,
 		.read_u64 = mem_cgroup_read,
 	},
 	{
-- 
1.5.4.rc3

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH 3/3] memcgoup: allow memory.failcnt to be reset
  2008-03-11 10:16 ` KAMEZAWA Hiroyuki
@ 2008-03-11 10:14   ` Balbir Singh
  2008-03-13 23:37     ` Li Zefan
  0 siblings, 1 reply; 4+ messages in thread
From: Balbir Singh @ 2008-03-11 10:14 UTC (permalink / raw)
  To: KAMEZAWA Hiroyuki
  Cc: Li Zefan, Andrew Morton, Paul Menage, Pavel Emelianov, LKML,
	linux-mm, Linux Containers

KAMEZAWA Hiroyuki wrote:
> On Tue, 11 Mar 2008 19:09:02 +0900
> Li Zefan <lizf@cn.fujitsu.com> wrote:
> 
>> Allow memory.failcnt to be reset to 0:
>>
>>         echo 0 > memory.failcnt
>>
>> And '0' is the only valid value.
>>
> Can't this be generic resource counter function ?
> 

I was about to suggest a generic cgroup option, since we do reset values even
for the cpu accounting subsystem.

-- 
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH 3/3] memcgoup: allow memory.failcnt to be reset
  2008-03-11 10:09 [PATCH 3/3] memcgoup: allow memory.failcnt to be reset Li Zefan
@ 2008-03-11 10:16 ` KAMEZAWA Hiroyuki
  2008-03-11 10:14   ` Balbir Singh
  0 siblings, 1 reply; 4+ messages in thread
From: KAMEZAWA Hiroyuki @ 2008-03-11 10:16 UTC (permalink / raw)
  To: Li Zefan
  Cc: Andrew Morton, Balbir Singh, Paul Menage, Pavel Emelianov, LKML,
	linux-mm, Linux Containers

On Tue, 11 Mar 2008 19:09:02 +0900
Li Zefan <lizf@cn.fujitsu.com> wrote:

> Allow memory.failcnt to be reset to 0:
> 
>         echo 0 > memory.failcnt
> 
> And '0' is the only valid value.
> 
Can't this be generic resource counter function ?

Thanks,
-Kame


> This is useful when testing or observing the memory resource
> controller. Without this function, one will have to remember
> the previous failcnt to decide whether memory reclaim has
> happened *again*.
> 
> Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
> ---
>  Documentation/controllers/memory.txt |    4 +++-
>  mm/memcontrol.c                      |   15 +++++++++++++++
>  2 files changed, 18 insertions(+), 1 deletions(-)
> 
> diff --git a/Documentation/controllers/memory.txt b/Documentation/controllers/memory.txt
> index 866b9cd..28f80e3 100644
> --- a/Documentation/controllers/memory.txt
> +++ b/Documentation/controllers/memory.txt
> @@ -194,7 +194,9 @@ this file after a write to guarantee the value committed by the kernel.
>  4096
>  
>  The memory.failcnt field gives the number of times that the cgroup limit was
> -exceeded.
> +exceeded. It can be reset.
> +
> +# echo 0 > memory.failcnt
>  
>  The memory.stat file gives accounting information. Now, the number of
>  caches, RSS and Active pages/Inactive pages are shown.
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 6145031..fd26dc2 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -883,6 +883,20 @@ static int mem_force_empty_write(struct cgroup *cont, struct cftype *cft,
>  	return ret;
>  }
>  
> +static int mem_failcnt_write(struct cgroup *cont, struct cftype *cft,
> +			     u64 val)
> +{
> +	struct res_counter *counter;
> +
> +	if (val != 0)
> +		return -EINVAL;
> +
> +	counter = &mem_cgroup_from_cont(cont)->res;
> +	res_counter_write_u64(counter, cft->private, 0);
> +
> +	return 0;
> +}
> +
>  static const struct mem_cgroup_stat_desc {
>  	const char *msg;
>  	u64 unit;
> @@ -934,6 +948,7 @@ static struct cftype mem_cgroup_files[] = {
>  	{
>  		.name = "failcnt",
>  		.private = RES_FAILCNT,
> +		.write_u64 = mem_failcnt_write,
>  		.read_u64 = mem_cgroup_read,
>  	},
>  	{
> -- 
> 1.5.4.rc3
> 


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH 3/3] memcgoup: allow memory.failcnt to be reset
  2008-03-11 10:14   ` Balbir Singh
@ 2008-03-13 23:37     ` Li Zefan
  0 siblings, 0 replies; 4+ messages in thread
From: Li Zefan @ 2008-03-13 23:37 UTC (permalink / raw)
  To: balbir
  Cc: KAMEZAWA Hiroyuki, Andrew Morton, Paul Menage, Pavel Emelianov,
	LKML, linux-mm, Linux Containers

Balbir Singh wrote:
> KAMEZAWA Hiroyuki wrote:
>> On Tue, 11 Mar 2008 19:09:02 +0900
>> Li Zefan <lizf@cn.fujitsu.com> wrote:
>>
>>> Allow memory.failcnt to be reset to 0:
>>>
>>>         echo 0 > memory.failcnt
>>>
>>> And '0' is the only valid value.
>>>
>> Can't this be generic resource counter function ?
>>
> 
> I was about to suggest a generic cgroup option, since we do reset values even
> for the cpu accounting subsystem.
> 

It won't help. You still have to write the write function, and you have to call
some res_counter routines to reset the value, and maybe also do some other
work.


^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2008-03-13 23:43 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2008-03-11 10:09 [PATCH 3/3] memcgoup: allow memory.failcnt to be reset Li Zefan
2008-03-11 10:16 ` KAMEZAWA Hiroyuki
2008-03-11 10:14   ` Balbir Singh
2008-03-13 23:37     ` Li Zefan

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).