The memory cgroup subsystem maintains a private ID infrastructure that
is decoupled from the cgroup IDs. This private ID system exists because
some kernel objects (like swap entries and shadow entries in the
workingset code) can outlive the cgroup they were associated with.
The motivation is best described in commit 73f576c04b941 ("mm:
memcontrol: fix cgroup creation failure after many small jobs").

Unfortunately, some in-kernel users (DAMON, LRU gen debugfs interface,
shrinker debugfs) started exposing these private IDs to userspace.
This is problematic because:

1. The private IDs are internal implementation details that could change
2. Userspace already has access to cgroup IDs through the cgroup
   filesystem
3. Using different ID namespaces in different interfaces is confusing

This series cleans up the memcg ID infrastructure by:

1. Explicitly marking the private ID APIs with "private" in their names
   to make it clear they are for internal use only (swap/workingset)

2. Making the public cgroup ID APIs (mem_cgroup_id/mem_cgroup_get_from_id)
   unconditionally available

3. Converting DAMON, LRU gen, and shrinker debugfs interfaces to use
   the public cgroup IDs instead of the private IDs

4. Removing the now-unused wrapper functions and renaming the public
   APIs for clarity

After this series:
- mem_cgroup_private_id() / mem_cgroup_from_private_id() are used for
  internal kernel objects that outlive their cgroup (swap, workingset)
- mem_cgroup_id() / mem_cgroup_get_from_id() return the public cgroup ID
  (from cgroup_id()) for use in userspace-facing interfaces

Note: please apply this series after the patch at
https://lore.kernel.org/20251225002904.139543-1-shakeel.butt@linux.dev/

Shakeel Butt (8):
  memcg: introduce private id API for in-kernel users
  memcg: expose mem_cgroup_ino() and mem_cgroup_get_from_ino()
    unconditionally
  memcg: mem_cgroup_get_from_ino() returns NULL on error
  memcg: use cgroup_id() instead of cgroup_ino() for memcg ID
  mm/damon: use cgroup ID instead of private memcg ID
  mm/vmscan: use cgroup ID instead of private memcg ID in lru_gen
    interface
  memcg: remove unused mem_cgroup_id() and mem_cgroup_from_id()
  memcg: rename mem_cgroup_ino() to mem_cgroup_id()

 include/linux/damon.h      |  4 +--
 include/linux/memcontrol.h | 26 +++++++----------
 mm/damon/core.c            |  7 ++---
 mm/damon/sysfs-schemes.c   |  6 ++--
 mm/list_lru.c              |  2 +-
 mm/memcontrol-v1.c         |  6 ++--
 mm/memcontrol-v1.h         |  4 +--
 mm/memcontrol.c            | 60 ++++++++++++++++++--------------------
 mm/shrinker_debug.c        | 13 +++++----
 mm/vmscan.c                | 17 ++++-------
 mm/workingset.c            |  8 ++---
 11 files changed, 68 insertions(+), 85 deletions(-)

--
2.47.3

On Thu, 25 Dec 2025 15:21:08 -0800 Shakeel Butt <shakeel.butt@linux.dev> wrote:

> The memory cgroup subsystem maintains a private ID infrastructure that
> is decoupled from the cgroup IDs. This private ID system exists because
> some kernel objects (like swap entries and shadow entries in the
> workingset code) can outlive the cgroup they were associated with.
> The motivation is best described in commit 73f576c04b941 ("mm:
> memcontrol: fix cgroup creation failure after many small jobs").
> 
> Unfortunately, some in-kernel users (DAMON, LRU gen debugfs interface,
> shrinker debugfs) started exposing these private IDs to userspace.

Technically speaking, DAMON is not exposing the private IDs to userspace.  It
does use the ids to specify the memory cgroups in kernel space.  But, when it
communicates with the user space, it uses the paths of the cgroups, not the
ids.

> This is problematic because:
> 
> 1. The private IDs are internal implementation details that could change
> 2. Userspace already has access to cgroup IDs through the cgroup
>    filesystem
> 3. Using different ID namespaces in different interfaces is confusing

Though DAMON is not exposing the IDs to the userspace, I agree it is better to
use public id, mainly because DAMON doesn't really care about the
cgroup-outlive objects.  Also, it would allow easier change of the
implementation details and make more consistent kernel-space API usages.


Thanks,
SJ

[...]

On Thu 25-12-25 15:21:08, Shakeel Butt wrote:
> The memory cgroup subsystem maintains a private ID infrastructure that
> is decoupled from the cgroup IDs. This private ID system exists because
> some kernel objects (like swap entries and shadow entries in the
> workingset code) can outlive the cgroup they were associated with.
> The motivation is best described in commit 73f576c04b941 ("mm:
> memcontrol: fix cgroup creation failure after many small jobs").
> 
> Unfortunately, some in-kernel users (DAMON, LRU gen debugfs interface,
> shrinker debugfs) started exposing these private IDs to userspace.
> This is problematic because:
> 
> 1. The private IDs are internal implementation details that could change
> 2. Userspace already has access to cgroup IDs through the cgroup
>    filesystem
> 3. Using different ID namespaces in different interfaces is confusing
> 
> This series cleans up the memcg ID infrastructure by:
> 
> 1. Explicitly marking the private ID APIs with "private" in their names
>    to make it clear they are for internal use only (swap/workingset)
> 
> 2. Making the public cgroup ID APIs (mem_cgroup_id/mem_cgroup_get_from_id)
>    unconditionally available
> 
> 3. Converting DAMON, LRU gen, and shrinker debugfs interfaces to use
>    the public cgroup IDs instead of the private IDs
> 
> 4. Removing the now-unused wrapper functions and renaming the public
>    APIs for clarity
> 
> After this series:
> - mem_cgroup_private_id() / mem_cgroup_from_private_id() are used for
>   internal kernel objects that outlive their cgroup (swap, workingset)
> - mem_cgroup_id() / mem_cgroup_get_from_id() return the public cgroup ID
>   (from cgroup_id()) for use in userspace-facing interfaces
> 
> Note: please apply this series after the patch at
> https://lore.kernel.org/20251225002904.139543-1-shakeel.butt@linux.dev/

Makes sense to me. Originally I was not supper happy about the private
interface as this should be really private to memcg proper but then I
have noticed the lru code needs this outside and dealing with that would
be quite messy so an explicit name is probably better in the end.

Feel free to add
Acked-by: Michal Hocko <mhocko@suse.com>

Thanks!

> 
> Shakeel Butt (8):
>   memcg: introduce private id API for in-kernel users
>   memcg: expose mem_cgroup_ino() and mem_cgroup_get_from_ino()
>     unconditionally
>   memcg: mem_cgroup_get_from_ino() returns NULL on error
>   memcg: use cgroup_id() instead of cgroup_ino() for memcg ID
>   mm/damon: use cgroup ID instead of private memcg ID
>   mm/vmscan: use cgroup ID instead of private memcg ID in lru_gen
>     interface
>   memcg: remove unused mem_cgroup_id() and mem_cgroup_from_id()
>   memcg: rename mem_cgroup_ino() to mem_cgroup_id()
> 
>  include/linux/damon.h      |  4 +--
>  include/linux/memcontrol.h | 26 +++++++----------
>  mm/damon/core.c            |  7 ++---
>  mm/damon/sysfs-schemes.c   |  6 ++--
>  mm/list_lru.c              |  2 +-
>  mm/memcontrol-v1.c         |  6 ++--
>  mm/memcontrol-v1.h         |  4 +--
>  mm/memcontrol.c            | 60 ++++++++++++++++++--------------------
>  mm/shrinker_debug.c        | 13 +++++----
>  mm/vmscan.c                | 17 ++++-------
>  mm/workingset.c            |  8 ++---
>  11 files changed, 68 insertions(+), 85 deletions(-)
> 
> --
> 2.47.3

-- 
Michal Hocko
SUSE Labs

On Tue, 6 Jan 2026 11:25:59 +0100 Michal Hocko <mhocko@suse.com> wrote:

> > Note: please apply this series after the patch at
> > https://lore.kernel.org/20251225002904.139543-1-shakeel.butt@linux.dev/

OK, that's in mm-hotfixes-unstable.

> Makes sense to me. Originally I was not supper happy about the private
> interface as this should be really private to memcg proper but then I
> have noticed the lru code needs this outside and dealing with that would
> be quite messy so an explicit name is probably better in the end.
> 
> Feel free to add
> Acked-by: Michal Hocko <mhocko@suse.com>

Great, thanks, I'll add it.  The series in somewhat old (Dec 25) but
nothing seems to have changed.