Skip to contentSkip to conversation

memcg usage_in_bytes does not account file mapped and slab memory

RFC In review Anton Vorontsov 2 participants 5 messages Mar 2, 2012 – Apr 23, 2012

Subsystems (3)
Memory management cgroup memcg
lore
Share & tools
lore

Anton Vorontsov

author
#0 lore raw

RFC 1/1 memcg usage_in_bytes does not account file mapped and slab memory

... and thus is useless for low memory notifications.

Hi all!

While working on userspace low memory killer daemon (a supposed substitution for the kernel low memory killer, i.e. drivers/staging/android/lowmemorykiller.c), I noticed that current cgroups memory notifications aren't suitable for such a daemon.

Suppose we want to install a notification when free memory drops below 8 MB. Logically (taking memory hotplug aside), using current usage_in_bytes notifications we would install an event on 'total_ram - 8MB' threshold.

But as usage_in_bytes doesn't account file mapped memory and memory used by kernel slab, the formula won't work.

Currently I use the following patch that makes things going:

mm/memcontrol.c +3 -0
index 228d646..c8abdc5 100644
@@ -3812,6 +3812,9 @@ static inline u64 mem_cgroup_usage(struct mem_cgroup *memcg, bool swap)
 
        val = mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_CACHE);
        val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_RSS);
+       val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_FILE_MAPPED);
+       val += global_page_state(NR_SLAB_RECLAIMABLE);
+       val += global_page_state(NR_SLAB_UNRECLAIMABLE);

But here are some questions:

1. Is there any particular reason we don't currently account file mapped memory in usage_in_bytes?

To me, MEM_CGROUP_STAT_FILE_MAPPED hunk seems logical even if we don't use it for lowmemory notifications.

Plus, it seems that FILE_MAPPED _is_ accounted for the non-root cgroups, so I guess it's clearly a bug for the root memcg?

2. As for NR_SLAB_RECLAIMABLE and NR_SLAB_UNRECLAIMABLE, it seems that these numbers are only applicable for the root memcg. I'm not sure that usage_in_bytes semantics should actually account these, but I tend to think that we should.

All in all, not accounting both 1. and 2. looks like bugs to me.

But if for some reason we don't want to change usage_in_bytes, should I just go ahead and implement a new cftype (say free_in_bytes), which would account free memory as total_ram - cache - rss - mapped - slab, with ability to install notifiers? That way we would also could solve memory hotplug issue in the kernel, so that userland won't need to bother with reinstalling lowmemory notifiers when memory added/removed.

Thanks!

Signature · 2 lines
Anton Vorontsov
Email: [email protected]
#1 lore raw
Quoted 33 lines · On Fri, 2 Mar 2012 20:27:53 +0400 Anton Vorontsov <[email protected]>…

... and thus is useless for low memory notifications.

Hi all!

While working on userspace low memory killer daemon (a supposed substitution for the kernel low memory killer, i.e. drivers/staging/android/lowmemorykiller.c), I noticed that current cgroups memory notifications aren't suitable for such a daemon.

Suppose we want to install a notification when free memory drops below 8 MB. Logically (taking memory hotplug aside), using current usage_in_bytes notifications we would install an event on 'total_ram - 8MB' threshold.

But as usage_in_bytes doesn't account file mapped memory and memory used by kernel slab, the formula won't work.

Currently I use the following patch that makes things going:

mm/memcontrol.c +3 -0
index 228d646..c8abdc5 100644
@@ -3812,6 +3812,9 @@ static inline u64 mem_cgroup_usage(struct mem_cgroup *memcg, bool swap)
 
        val = mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_CACHE);
        val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_RSS);
+       val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_FILE_MAPPED);
+       val += global_page_state(NR_SLAB_RECLAIMABLE);
+       val += global_page_state(NR_SLAB_UNRECLAIMABLE);

But here are some questions:

Quoted 3 lines

1. Is there any particular reason we don't currently account file mapped memory in usage_in_bytes?

CACHE includes all file caches. Why do you think FILE_MAPPED is not included in CACHE ?

Quoted 11 lines

To me, MEM_CGROUP_STAT_FILE_MAPPED hunk seems logical even if we don't use it for lowmemory notifications.

Plus, it seems that FILE_MAPPED _is_ accounted for the non-root cgroups, so I guess it's clearly a bug for the root memcg?

2. As for NR_SLAB_RECLAIMABLE and NR_SLAB_UNRECLAIMABLE, it seems that these numbers are only applicable for the root memcg. I'm not sure that usage_in_bytes semantics should actually account these, but I tend to think that we should.

Now, SLAB is not accounted by memcg at all.
See memifo if necessary.

Quoted 2 lines

All in all, not accounting both 1. and 2. looks like bugs to me.

It's spec. not bug. If you want to see slab status in memcg's file, Please add kernel memory accounting feature. There has been already 2 proposals. Check them and comment.

Thanks,
-Kame

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author
#2 lore raw

RFC PATCH 1/1 memcg: MEMCG_NR_FILE_MAPPED should update _STAT_CACHE as well

...otherwise the we're getting the wrong numbers in usage_in_bytes.

On Mon, Mar 05, 2012 at 09:19:34AM +0900, KAMEZAWA Hiroyuki wrote:
[...]

Quoted 13 lines
Quoted 13 lines
mm/memcontrol.c +1 -0
index 228d646..c8abdc5 100644
@@ -3812,6 +3812,9 @@ static inline u64 mem_cgroup_usage(struct mem_cgroup *memcg, bool swap)
 
        val = mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_CACHE);
        val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_RSS);
+       val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_FILE_MAPPED);

1. Is there any particular reason we don't currently account file mapped memory in usage_in_bytes?

Quoted 7 lines
Quoted 5 lines

To me, MEM_CGROUP_STAT_FILE_MAPPED hunk seems logical even if we don't use it for lowmemory notifications.

Plus, it seems that FILE_MAPPED _is_ accounted for the non-root cgroups, so I guess it's clearly a bug for the root memcg?

CACHE includes all file caches. Why do you think FILE_MAPPED is not included in CACHE ?

There were tons of changes in the memcg lately, but I believe the issue is still there.

For example, looking into this code flow:

-> page_add_file_rmap() (mm/rmap.c)
-> mem_cgroup_inc_page_stat(page, MEMCG_NR_FILE_MAPPED) (include/linux/memcontrol.h)
-> void mem_cgroup_update_page_stat(page, MEMCG_NR_FILE_MAPPED, 1) (mm/memcontrol.c)

And then:

void mem_cgroup_update_page_stat(struct page *page,
enum mem_cgroup_page_stat_item idx, int val)

{
        ...
        switch (idx) {
        case MEMCG_NR_FILE_MAPPED:
                idx = MEM_CGROUP_STAT_FILE_MAPPED;
                break;
        default:
                BUG();
        }

        this_cpu_add(memcg->stat->count[idx], val);
        ...
}

So, clearly, this function only bothers updating _FILE_MAPPED only, leaving _CACHE alone.

If you're saying that _CACHE meant to include _FILE_MAPPED, then I guess the patch down below would be a proper fix then... Otherwise we need to be consistent on stats reporting, and either fall-back to my original fix (in mem_cgroup_usage()), or think about doing it some other way...

Signed-off-by Anton Vorontsov <[email protected]>

---

The patch is against current -next.

Thanks,

 mm/memcontrol.c |    2 ++
 1 file changed, 2 insertions(+)
mm/memcontrol.c +2 -0
index 884e936..760ecf5 100644
@@ -1958,6 +1958,8 @@ void mem_cgroup_update_page_stat(struct page *page,
 
 	switch (idx) {
 	case MEMCG_NR_FILE_MAPPED:
+		idx = MEM_CGROUP_STAT_CACHE;
+		this_cpu_add(memcg->stat->count[idx], val);
 		idx = MEM_CGROUP_STAT_FILE_MAPPED;
 		break;
 	default:
Signature · 1 line
1.7.9.2
NAK
#3 lore raw

Re: [PATCH RFC] memcg: MEMCG_NR_FILE_MAPPED should update _STAT_CACHE as well

Quoted 52 lines · Anton Vorontsov

...otherwise the we're getting the wrong numbers in usage_in_bytes.

On Mon, Mar 05, 2012 at 09:19:34AM +0900, KAMEZAWA Hiroyuki wrote:
[...]

Quoted 13 lines
Quoted 13 lines
mm/memcontrol.c +1 -0
index 228d646..c8abdc5 100644
@@ -3812,6 +3812,9 @@ static inline u64 mem_cgroup_usage(struct mem_cgroup *memcg, bool swap)
 
        val = mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_CACHE);
        val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_RSS);
+       val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_FILE_MAPPED);

1. Is there any particular reason we don't currently account file mapped memory in usage_in_bytes?

Quoted 7 lines
Quoted 5 lines

To me, MEM_CGROUP_STAT_FILE_MAPPED hunk seems logical even if we don't use it for lowmemory notifications.

Plus, it seems that FILE_MAPPED _is_ accounted for the non-root cgroups, so I guess it's clearly a bug for the root memcg?

CACHE includes all file caches. Why do you think FILE_MAPPED is not included in CACHE ?

There were tons of changes in the memcg lately, but I believe the issue is still there.

For example, looking into this code flow:

-> page_add_file_rmap() (mm/rmap.c)
-> mem_cgroup_inc_page_stat(page, MEMCG_NR_FILE_MAPPED) (include/linux/memcontrol.h)
-> void mem_cgroup_update_page_stat(page, MEMCG_NR_FILE_MAPPED, 1) (mm/memcontrol.c)

And then:

void mem_cgroup_update_page_stat(struct page *page,
enum mem_cgroup_page_stat_item idx, int val)

{
        ...
        switch (idx) {
        case MEMCG_NR_FILE_MAPPED:
                idx = MEM_CGROUP_STAT_FILE_MAPPED;
                break;
        default:
                BUG();
        }

        this_cpu_add(memcg->stat->count[idx], val);
        ...
}
Quoted 9 lines

So, clearly, this function only bothers updating _FILE_MAPPED only, leaving _CACHE alone.

If you're saying that _CACHE meant to include _FILE_MAPPED, then I guess the patch down below would be a proper fix then... Otherwise we need to be consistent on stats reporting, and either fall-back to my original fix (in mem_cgroup_usage()), or think about doing it some other way...

NACK.
CACHE is updated at charge()/uncharge()...inserting/removing page cache to radix-tree.

Thanks,
-Kame

Quoted 15 lines
Signed-off-by Anton Vorontsov <[email protected]>

---

The patch is against current -next.

Thanks,

 mm/memcontrol.c |    2 ++
 1 file changed, 2 insertions(+)
mm/memcontrol.c +0 -0
index 884e936..760ecf5 100644

mm/memcontrol.c void mem_cgroup_update_page_stat(struct page *page, · line 1965

Quoted 8 lines
mm/memcontrol.c +2 -0
@@ -1958,6 +1958,8 @@ void mem_cgroup_update_page_stat(struct page *page,
 
 	switch (idx) {
 	case MEMCG_NR_FILE_MAPPED:
+		idx = MEM_CGROUP_STAT_CACHE;
+		this_cpu_add(memcg->stat->count[idx], val);
 		idx = MEM_CGROUP_STAT_FILE_MAPPED;
 		break;
 	default:

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

author
#4 lore raw

Re: [PATCH RFC] memcg: MEMCG_NR_FILE_MAPPED should update _STAT_CACHE as well

On Mon, Apr 23, 2012 at 05:35:12PM +0900, KAMEZAWA Hiroyuki wrote:
[...]

Quoted 26 lines
Quoted 26 lines

For example, looking into this code flow:

-> page_add_file_rmap() (mm/rmap.c)
-> mem_cgroup_inc_page_stat(page, MEMCG_NR_FILE_MAPPED) (include/linux/memcontrol.h)
-> void mem_cgroup_update_page_stat(page, MEMCG_NR_FILE_MAPPED, 1) (mm/memcontrol.c)

And then:

void mem_cgroup_update_page_stat(struct page *page,
enum mem_cgroup_page_stat_item idx, int val)

{
        ...
        switch (idx) {
        case MEMCG_NR_FILE_MAPPED:
                idx = MEM_CGROUP_STAT_FILE_MAPPED;
                break;
        default:
                BUG();
        }

        this_cpu_add(memcg->stat->count[idx], val);
        ...
}

So, clearly, this function only bothers updating _FILE_MAPPED only, leaving _CACHE alone.

[...]

Quoted 3 lines

NACK.
CACHE is updated at charge()/uncharge()...inserting/removing page cache to radix-tree.

Interesting; true, we have charge/uncharge in __do_fault()/do_wp_page and friends. So, we seem to update FILE_MAPPED in the rmap via cgroup_dec/inc_page_stat, and CACHE is updated via charge/uncharge. Hm.

The code in memory.c is full of if/else ifs, and I wonder if there's some discrepancy in there, but briefly looking it looks fine. The code looks correct indeed, but I'm getting the wrong stats. :-/

Thanks,

Signature · 9 lines
Anton Vorontsov
Email: [email protected]

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>