It was found that a number of offlined memcgs were not freed because
they were pinned by some charged pages that were present. Even "echo
1 > /proc/sys/vm/drop_caches" wasn't able to free those pages. These
offlined but not freed memcgs tend to increase in number over time with
the side effect that percpu memory consumption as shown in /proc/meminfo
also increases over time.
In order to find out more information about those pages that pin
offlined memcgs, the page_owner feature is extended to dump memory
cgroup information especially whether the cgroup is offlined or not.
Signed-off-by: Waiman Long <[email protected]>
---
mm/page_owner.c | 28 ++++++++++++++++++++++++++++
1 file changed, 28 insertions(+)
diff --git a/mm/page_owner.c b/mm/page_owner.c
index c52ce9d6bc3b..e5d8c642296b 100644
--- a/mm/page_owner.c
+++ b/mm/page_owner.c
@@ -10,6 +10,7 @@
#include <linux/migrate.h>
#include <linux/stackdepot.h>
#include <linux/seq_file.h>
+#include <linux/memcontrol.h>
#include <linux/sched/clock.h>
#include "internal.h"
@@ -339,6 +340,7 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
depot_stack_handle_t handle)
{
int ret = 0, pageblock_mt, page_mt;
+ unsigned long __maybe_unused memcg_data;
char *kbuf;
count = min_t(size_t, count, PAGE_SIZE);
@@ -371,6 +373,32 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
"Page has been migrated, last migrate reason: %s\n",
migrate_reason_names[page_owner->last_migrate_reason]);
+#ifdef CONFIG_MEMCG
+ /*
+ * Look for memcg information and print it out
+ */
+ memcg_data = READ_ONCE(page->memcg_data);
+ if (memcg_data) {
+ struct mem_cgroup *memcg = page_memcg_check(page);
+ bool onlined;
+ char name[80];
+
+ if (memcg_data & MEMCG_DATA_OBJCGS)
+ SNPRINTF(kbuf, count, ret, err, "Slab cache page\n");
+
+ if (!memcg)
+ goto copy_out;
+
+ onlined = (memcg->css.flags & CSS_ONLINE);
+ cgroup_name(memcg->css.cgroup, name, sizeof(name) - 1);
+ SNPRINTF(kbuf, count, ret, err, "Charged %sto %smemcg %s\n",
+ PageMemcgKmem(page) ? "(via objcg) " : "",
+ onlined ? "" : "offlined ", name);
+ }
+
+copy_out:
+#endif
+
SNPRINTF(kbuf, count, ret, err, "\n");
if (copy_to_user(buf, kbuf, ret))
--
2.27.0
On Fri, Jan 28, 2022 at 02:56:42PM -0500, Waiman Long wrote:
> It was found that a number of offlined memcgs were not freed because
> they were pinned by some charged pages that were present. Even "echo
> 1 > /proc/sys/vm/drop_caches" wasn't able to free those pages. These
> offlined but not freed memcgs tend to increase in number over time with
> the side effect that percpu memory consumption as shown in /proc/meminfo
> also increases over time.
>
> In order to find out more information about those pages that pin
> offlined memcgs, the page_owner feature is extended to dump memory
> cgroup information especially whether the cgroup is offlined or not.
>
> Signed-off-by: Waiman Long <[email protected]>
> ---
> mm/page_owner.c | 28 ++++++++++++++++++++++++++++
> 1 file changed, 28 insertions(+)
>
> diff --git a/mm/page_owner.c b/mm/page_owner.c
> index c52ce9d6bc3b..e5d8c642296b 100644
> --- a/mm/page_owner.c
> +++ b/mm/page_owner.c
> @@ -10,6 +10,7 @@
> #include <linux/migrate.h>
> #include <linux/stackdepot.h>
> #include <linux/seq_file.h>
> +#include <linux/memcontrol.h>
> #include <linux/sched/clock.h>
>
> #include "internal.h"
> @@ -339,6 +340,7 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
> depot_stack_handle_t handle)
> {
> int ret = 0, pageblock_mt, page_mt;
> + unsigned long __maybe_unused memcg_data;
> char *kbuf;
>
> count = min_t(size_t, count, PAGE_SIZE);
> @@ -371,6 +373,32 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
> "Page has been migrated, last migrate reason: %s\n",
> migrate_reason_names[page_owner->last_migrate_reason]);
>
> +#ifdef CONFIG_MEMCG
> + /*
> + * Look for memcg information and print it out
> + */
> + memcg_data = READ_ONCE(page->memcg_data);
> + if (memcg_data) {
> + struct mem_cgroup *memcg = page_memcg_check(page);
> + bool onlined;
> + char name[80];
> +
> + if (memcg_data & MEMCG_DATA_OBJCGS)
> + SNPRINTF(kbuf, count, ret, err, "Slab cache page\n");
> +
> + if (!memcg)
> + goto copy_out;
> +
> + onlined = (memcg->css.flags & CSS_ONLINE);
> + cgroup_name(memcg->css.cgroup, name, sizeof(name) - 1);
> + SNPRINTF(kbuf, count, ret, err, "Charged %sto %smemcg %s\n",
^^^
Extra specifier?
Did this compile without warnings?
Ira
> + PageMemcgKmem(page) ? "(via objcg) " : "",
> + onlined ? "" : "offlined ", name);
> + }
> +
> +copy_out:
> +#endif
> +
> SNPRINTF(kbuf, count, ret, err, "\n");
>
> if (copy_to_user(buf, kbuf, ret))
> --
> 2.27.0
>
>
On 1/28/22 16:22, Ira Weiny wrote:
> On Fri, Jan 28, 2022 at 02:56:42PM -0500, Waiman Long wrote:
>> It was found that a number of offlined memcgs were not freed because
>> they were pinned by some charged pages that were present. Even "echo
>> 1 > /proc/sys/vm/drop_caches" wasn't able to free those pages. These
>> offlined but not freed memcgs tend to increase in number over time with
>> the side effect that percpu memory consumption as shown in /proc/meminfo
>> also increases over time.
>>
>> In order to find out more information about those pages that pin
>> offlined memcgs, the page_owner feature is extended to dump memory
>> cgroup information especially whether the cgroup is offlined or not.
>>
>> Signed-off-by: Waiman Long <[email protected]>
>> ---
>> mm/page_owner.c | 28 ++++++++++++++++++++++++++++
>> 1 file changed, 28 insertions(+)
>>
>> diff --git a/mm/page_owner.c b/mm/page_owner.c
>> index c52ce9d6bc3b..e5d8c642296b 100644
>> --- a/mm/page_owner.c
>> +++ b/mm/page_owner.c
>> @@ -10,6 +10,7 @@
>> #include <linux/migrate.h>
>> #include <linux/stackdepot.h>
>> #include <linux/seq_file.h>
>> +#include <linux/memcontrol.h>
>> #include <linux/sched/clock.h>
>>
>> #include "internal.h"
>> @@ -339,6 +340,7 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
>> depot_stack_handle_t handle)
>> {
>> int ret = 0, pageblock_mt, page_mt;
>> + unsigned long __maybe_unused memcg_data;
>> char *kbuf;
>>
>> count = min_t(size_t, count, PAGE_SIZE);
>> @@ -371,6 +373,32 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
>> "Page has been migrated, last migrate reason: %s\n",
>> migrate_reason_names[page_owner->last_migrate_reason]);
>>
>> +#ifdef CONFIG_MEMCG
>> + /*
>> + * Look for memcg information and print it out
>> + */
>> + memcg_data = READ_ONCE(page->memcg_data);
>> + if (memcg_data) {
>> + struct mem_cgroup *memcg = page_memcg_check(page);
>> + bool onlined;
>> + char name[80];
>> +
>> + if (memcg_data & MEMCG_DATA_OBJCGS)
>> + SNPRINTF(kbuf, count, ret, err, "Slab cache page\n");
>> +
>> + if (!memcg)
>> + goto copy_out;
>> +
>> + onlined = (memcg->css.flags & CSS_ONLINE);
>> + cgroup_name(memcg->css.cgroup, name, sizeof(name) - 1);
>> + SNPRINTF(kbuf, count, ret, err, "Charged %sto %smemcg %s\n",
> ^^^
> Extra specifier?
>
> Did this compile without warnings?
Yes, there was no warning.
Cheers,
Longmna
> Ira
>
>> + PageMemcgKmem(page) ? "(via objcg) " : "",
>> + onlined ? "" : "offlined ", name);
>> + }
>> +
>> +copy_out:
>> +#endif
>> +
>> SNPRINTF(kbuf, count, ret, err, "\n");
>>
>> if (copy_to_user(buf, kbuf, ret))
>> --
>> 2.27.0
>>
>>
On Fri, Jan 28, 2022 at 04:31:07PM -0500, Waiman Long wrote:
> On 1/28/22 16:22, Ira Weiny wrote:
> > On Fri, Jan 28, 2022 at 02:56:42PM -0500, Waiman Long wrote:
> > > It was found that a number of offlined memcgs were not freed because
> > > they were pinned by some charged pages that were present. Even "echo
> > > 1 > /proc/sys/vm/drop_caches" wasn't able to free those pages. These
> > > offlined but not freed memcgs tend to increase in number over time with
> > > the side effect that percpu memory consumption as shown in /proc/meminfo
> > > also increases over time.
> > >
> > > In order to find out more information about those pages that pin
> > > offlined memcgs, the page_owner feature is extended to dump memory
> > > cgroup information especially whether the cgroup is offlined or not.
> > >
> > > Signed-off-by: Waiman Long <[email protected]>
> > > ---
> > > mm/page_owner.c | 28 ++++++++++++++++++++++++++++
> > > 1 file changed, 28 insertions(+)
> > >
> > > diff --git a/mm/page_owner.c b/mm/page_owner.c
> > > index c52ce9d6bc3b..e5d8c642296b 100644
> > > --- a/mm/page_owner.c
> > > +++ b/mm/page_owner.c
> > > @@ -10,6 +10,7 @@
> > > #include <linux/migrate.h>
> > > #include <linux/stackdepot.h>
> > > #include <linux/seq_file.h>
> > > +#include <linux/memcontrol.h>
> > > #include <linux/sched/clock.h>
> > > #include "internal.h"
> > > @@ -339,6 +340,7 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
> > > depot_stack_handle_t handle)
> > > {
> > > int ret = 0, pageblock_mt, page_mt;
> > > + unsigned long __maybe_unused memcg_data;
> > > char *kbuf;
> > > count = min_t(size_t, count, PAGE_SIZE);
> > > @@ -371,6 +373,32 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
> > > "Page has been migrated, last migrate reason: %s\n",
> > > migrate_reason_names[page_owner->last_migrate_reason]);
> > > +#ifdef CONFIG_MEMCG
> > > + /*
> > > + * Look for memcg information and print it out
> > > + */
> > > + memcg_data = READ_ONCE(page->memcg_data);
> > > + if (memcg_data) {
> > > + struct mem_cgroup *memcg = page_memcg_check(page);
> > > + bool onlined;
> > > + char name[80];
> > > +
> > > + if (memcg_data & MEMCG_DATA_OBJCGS)
> > > + SNPRINTF(kbuf, count, ret, err, "Slab cache page\n");
> > > +
> > > + if (!memcg)
> > > + goto copy_out;
> > > +
> > > + onlined = (memcg->css.flags & CSS_ONLINE);
> > > + cgroup_name(memcg->css.cgroup, name, sizeof(name) - 1);
> > > + SNPRINTF(kbuf, count, ret, err, "Charged %sto %smemcg %s\n",
> > ^^^
> > Extra specifier?
> >
> > Did this compile without warnings?
>
> Yes, there was no warning.
But isn't that an extra specifier?
Ira
>
> Cheers,
> Longmna
>
> > Ira
> >
> > > + PageMemcgKmem(page) ? "(via objcg) " : "",
> > > + onlined ? "" : "offlined ", name);
> > > + }
> > > +
> > > +copy_out:
> > > +#endif
> > > +
> > > SNPRINTF(kbuf, count, ret, err, "\n");
> > > if (copy_to_user(buf, kbuf, ret))
> > > --
> > > 2.27.0
> > >
> > >
>
On 1/28/22 16:48, Ira Weiny wrote:
> On Fri, Jan 28, 2022 at 04:31:07PM -0500, Waiman Long wrote:
>> On 1/28/22 16:22, Ira Weiny wrote:
>>> On Fri, Jan 28, 2022 at 02:56:42PM -0500, Waiman Long wrote:
>>>> It was found that a number of offlined memcgs were not freed because
>>>> they were pinned by some charged pages that were present. Even "echo
>>>> 1 > /proc/sys/vm/drop_caches" wasn't able to free those pages. These
>>>> offlined but not freed memcgs tend to increase in number over time with
>>>> the side effect that percpu memory consumption as shown in /proc/meminfo
>>>> also increases over time.
>>>>
>>>> In order to find out more information about those pages that pin
>>>> offlined memcgs, the page_owner feature is extended to dump memory
>>>> cgroup information especially whether the cgroup is offlined or not.
>>>>
>>>> Signed-off-by: Waiman Long <[email protected]>
>>>> ---
>>>> mm/page_owner.c | 28 ++++++++++++++++++++++++++++
>>>> 1 file changed, 28 insertions(+)
>>>>
>>>> diff --git a/mm/page_owner.c b/mm/page_owner.c
>>>> index c52ce9d6bc3b..e5d8c642296b 100644
>>>> --- a/mm/page_owner.c
>>>> +++ b/mm/page_owner.c
>>>> @@ -10,6 +10,7 @@
>>>> #include <linux/migrate.h>
>>>> #include <linux/stackdepot.h>
>>>> #include <linux/seq_file.h>
>>>> +#include <linux/memcontrol.h>
>>>> #include <linux/sched/clock.h>
>>>> #include "internal.h"
>>>> @@ -339,6 +340,7 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
>>>> depot_stack_handle_t handle)
>>>> {
>>>> int ret = 0, pageblock_mt, page_mt;
>>>> + unsigned long __maybe_unused memcg_data;
>>>> char *kbuf;
>>>> count = min_t(size_t, count, PAGE_SIZE);
>>>> @@ -371,6 +373,32 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
>>>> "Page has been migrated, last migrate reason: %s\n",
>>>> migrate_reason_names[page_owner->last_migrate_reason]);
>>>> +#ifdef CONFIG_MEMCG
>>>> + /*
>>>> + * Look for memcg information and print it out
>>>> + */
>>>> + memcg_data = READ_ONCE(page->memcg_data);
>>>> + if (memcg_data) {
>>>> + struct mem_cgroup *memcg = page_memcg_check(page);
>>>> + bool onlined;
>>>> + char name[80];
>>>> +
>>>> + if (memcg_data & MEMCG_DATA_OBJCGS)
>>>> + SNPRINTF(kbuf, count, ret, err, "Slab cache page\n");
>>>> +
>>>> + if (!memcg)
>>>> + goto copy_out;
>>>> +
>>>> + onlined = (memcg->css.flags & CSS_ONLINE);
>>>> + cgroup_name(memcg->css.cgroup, name, sizeof(name) - 1);
>>>> + SNPRINTF(kbuf, count, ret, err, "Charged %sto %smemcg %s\n",
>>> ^^^
>>> Extra specifier?
>>>
>>> Did this compile without warnings?
>> Yes, there was no warning.
> But isn't that an extra specifier?
There are 3 arguments to the format string that match the 3 "%s" in it:
1) PageMemcgKmem(page) ? "(via objcg) " : ""
2) onlined ? "" : "offlined
3) name
Cheers,
Longman
On Fri, Jan 28, 2022 at 10:35:22PM -0500, Waiman Long wrote:
> On 1/28/22 16:48, Ira Weiny wrote:
> > On Fri, Jan 28, 2022 at 04:31:07PM -0500, Waiman Long wrote:
> > > On 1/28/22 16:22, Ira Weiny wrote:
> > > > On Fri, Jan 28, 2022 at 02:56:42PM -0500, Waiman Long wrote:
> > > > > It was found that a number of offlined memcgs were not freed because
> > > > > they were pinned by some charged pages that were present. Even "echo
> > > > > 1 > /proc/sys/vm/drop_caches" wasn't able to free those pages. These
> > > > > offlined but not freed memcgs tend to increase in number over time with
> > > > > the side effect that percpu memory consumption as shown in /proc/meminfo
> > > > > also increases over time.
> > > > >
> > > > > In order to find out more information about those pages that pin
> > > > > offlined memcgs, the page_owner feature is extended to dump memory
> > > > > cgroup information especially whether the cgroup is offlined or not.
> > > > >
> > > > > Signed-off-by: Waiman Long <[email protected]>
> > > > > ---
> > > > > mm/page_owner.c | 28 ++++++++++++++++++++++++++++
> > > > > 1 file changed, 28 insertions(+)
> > > > >
> > > > > diff --git a/mm/page_owner.c b/mm/page_owner.c
> > > > > index c52ce9d6bc3b..e5d8c642296b 100644
> > > > > --- a/mm/page_owner.c
> > > > > +++ b/mm/page_owner.c
> > > > > @@ -10,6 +10,7 @@
> > > > > #include <linux/migrate.h>
> > > > > #include <linux/stackdepot.h>
> > > > > #include <linux/seq_file.h>
> > > > > +#include <linux/memcontrol.h>
> > > > > #include <linux/sched/clock.h>
> > > > > #include "internal.h"
> > > > > @@ -339,6 +340,7 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
> > > > > depot_stack_handle_t handle)
> > > > > {
> > > > > int ret = 0, pageblock_mt, page_mt;
> > > > > + unsigned long __maybe_unused memcg_data;
> > > > > char *kbuf;
> > > > > count = min_t(size_t, count, PAGE_SIZE);
> > > > > @@ -371,6 +373,32 @@ print_page_owner(char __user *buf, size_t count, unsigned long pfn,
> > > > > "Page has been migrated, last migrate reason: %s\n",
> > > > > migrate_reason_names[page_owner->last_migrate_reason]);
> > > > > +#ifdef CONFIG_MEMCG
> > > > > + /*
> > > > > + * Look for memcg information and print it out
> > > > > + */
> > > > > + memcg_data = READ_ONCE(page->memcg_data);
> > > > > + if (memcg_data) {
> > > > > + struct mem_cgroup *memcg = page_memcg_check(page);
> > > > > + bool onlined;
> > > > > + char name[80];
> > > > > +
> > > > > + if (memcg_data & MEMCG_DATA_OBJCGS)
> > > > > + SNPRINTF(kbuf, count, ret, err, "Slab cache page\n");
> > > > > +
> > > > > + if (!memcg)
> > > > > + goto copy_out;
> > > > > +
> > > > > + onlined = (memcg->css.flags & CSS_ONLINE);
> > > > > + cgroup_name(memcg->css.cgroup, name, sizeof(name) - 1);
> > > > > + SNPRINTF(kbuf, count, ret, err, "Charged %sto %smemcg %s\n",
> > > > ^^^
> > > > Extra specifier?
> > > >
> > > > Did this compile without warnings?
> > > Yes, there was no warning.
> > But isn't that an extra specifier?
>
> There are 3 arguments to the format string that match the 3 "%s" in it:
>
> 1) PageMemcgKmem(page) ? "(via objcg) " : ""
> 2) onlined ? "" : "offlined
> 3) name
My apologies. My parsing of the ? statements was off. FWIW putting
', name'
on the next line would make it more clear... But I see now...
Sorry,
Ira
>
> Cheers,
> Longman
>