There'd be no work for memcg-aware shrinkers when kernel memory is not
accounted per cgroup, so we can skip allocating per memcg shrinker data.
This saves some memory, avoids holding shrinker_mutex with O(nr_memcgs)
and saves work in shrink_slab_memcg().

Then there are SHRINKER_NONSLAB shrinkers which handle non-kernel memory
so nokmem should not disable their per-memcg behavior. Such shrinkers
(e.g.  deferred_split_shrinker) still need access to per-memcg data (see
also commit 0a432dcbeb32e ("mm: shrinker: make shrinker not depend on
memcg kmem")).

The savings with this patch come on container hosts that create many
superblocks (each with own shrinker) but tracking and processing
per-memcg data is pointless with nokmem (shrink_slab_memcg() is
partially guarded with !memcg_kmem_online already).

The patch uses "boottime" predicate mem_cgroup_kmem_disabled() (not
memcg_kmem_online()) to avoid mistakenly un-MEMCG_AWARE-ing shrinkers
registered before first non-root memcg is mkdir'd.

Signed-off-by: Michal Koutný <mkoutny@suse.com>
---
 mm/shrinker.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/mm/shrinker.c b/mm/shrinker.c
index 4a93fd433689a..7d7302619b7f7 100644
--- a/mm/shrinker.c
+++ b/mm/shrinker.c
@@ -219,6 +219,8 @@ static int shrinker_memcg_alloc(struct shrinker *shrinker)
 
 	if (mem_cgroup_disabled())
 		return -ENOSYS;
+	if (mem_cgroup_kmem_disabled() && !(shrinker->flags & SHRINKER_NONSLAB))
+		return -ENOSYS;
 
 	mutex_lock(&shrinker_mutex);
 	id = idr_alloc(&shrinker_idr, shrinker, 0, 0, GFP_KERNEL);

---
base-commit: cd2e103d57e5615f9bb027d772f93b9efd567224
change-id: 20260225-cgroup-ml-nokmem-shrinker-7da42fbcf8f2

Best regards,
-- 
Michal Koutný <mkoutny@suse.com>

> On Feb 26, 2026, at 02:38, Michal Koutný <mkoutny@suse.com> wrote:
> 
> There'd be no work for memcg-aware shrinkers when kernel memory is not
> accounted per cgroup, so we can skip allocating per memcg shrinker data.
> This saves some memory, avoids holding shrinker_mutex with O(nr_memcgs)
> and saves work in shrink_slab_memcg().
> 
> Then there are SHRINKER_NONSLAB shrinkers which handle non-kernel memory
> so nokmem should not disable their per-memcg behavior. Such shrinkers
> (e.g.  deferred_split_shrinker) still need access to per-memcg data (see
> also commit 0a432dcbeb32e ("mm: shrinker: make shrinker not depend on
> memcg kmem")).
> 
> The savings with this patch come on container hosts that create many
> superblocks (each with own shrinker) but tracking and processing
> per-memcg data is pointless with nokmem (shrink_slab_memcg() is
> partially guarded with !memcg_kmem_online already).
> 
> The patch uses "boottime" predicate mem_cgroup_kmem_disabled() (not
> memcg_kmem_online()) to avoid mistakenly un-MEMCG_AWARE-ing shrinkers
> registered before first non-root memcg is mkdir'd.
> 
> Signed-off-by: Michal Koutný <mkoutny@suse.com>

Reviewed-by: Muchun Song <muchun.song@linux.dev>

Thanks

On 2/26/26 2:38 AM, Michal Koutný wrote:
> There'd be no work for memcg-aware shrinkers when kernel memory is not
> accounted per cgroup, so we can skip allocating per memcg shrinker data.
> This saves some memory, avoids holding shrinker_mutex with O(nr_memcgs)
> and saves work in shrink_slab_memcg().
> 
> Then there are SHRINKER_NONSLAB shrinkers which handle non-kernel memory
> so nokmem should not disable their per-memcg behavior. Such shrinkers
> (e.g.  deferred_split_shrinker) still need access to per-memcg data (see
> also commit 0a432dcbeb32e ("mm: shrinker: make shrinker not depend on
> memcg kmem")).
> 
> The savings with this patch come on container hosts that create many
> superblocks (each with own shrinker) but tracking and processing
> per-memcg data is pointless with nokmem (shrink_slab_memcg() is
> partially guarded with !memcg_kmem_online already).
> 
> The patch uses "boottime" predicate mem_cgroup_kmem_disabled() (not
> memcg_kmem_online()) to avoid mistakenly un-MEMCG_AWARE-ing shrinkers
> registered before first non-root memcg is mkdir'd.
> 
> Signed-off-by: Michal Koutný <mkoutny@suse.com>
> ---
>   mm/shrinker.c | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/mm/shrinker.c b/mm/shrinker.c
> index 4a93fd433689a..7d7302619b7f7 100644
> --- a/mm/shrinker.c
> +++ b/mm/shrinker.c
> @@ -219,6 +219,8 @@ static int shrinker_memcg_alloc(struct shrinker *shrinker)
>   
>   	if (mem_cgroup_disabled())
>   		return -ENOSYS;
> +	if (mem_cgroup_kmem_disabled() && !(shrinker->flags & SHRINKER_NONSLAB))
> +		return -ENOSYS;

Make sense, and can you help update the following comment in
shrinker_alloc() as well:

         /*
	 * The nr_deferred is available on per memcg level for memcg aware
	 * shrinkers, so only allocate nr_deferred in the following cases:
	 *  - non-memcg-aware shrinkers
	 *  - !CONFIG_MEMCG
	 *  - memcg is disabled by kernel command line
	 */

Otherwise:

Acked-by: Qi Zheng <zhengqi.arch@bytedance.com>

Thanks,
Qi

>   
>   	mutex_lock(&shrinker_mutex);
>   	id = idr_alloc(&shrinker_idr, shrinker, 0, 0, GFP_KERNEL);
> 
> ---
> base-commit: cd2e103d57e5615f9bb027d772f93b9efd567224
> change-id: 20260225-cgroup-ml-nokmem-shrinker-7da42fbcf8f2
> 
> Best regards,

Michal Koutný <mkoutny@suse.com> writes:

> There'd be no work for memcg-aware shrinkers when kernel memory is not
> accounted per cgroup, so we can skip allocating per memcg shrinker data.
> This saves some memory, avoids holding shrinker_mutex with O(nr_memcgs)
> and saves work in shrink_slab_memcg().
>
> Then there are SHRINKER_NONSLAB shrinkers which handle non-kernel memory
> so nokmem should not disable their per-memcg behavior. Such shrinkers
> (e.g.  deferred_split_shrinker) still need access to per-memcg data (see
> also commit 0a432dcbeb32e ("mm: shrinker: make shrinker not depend on
> memcg kmem")).
>
> The savings with this patch come on container hosts that create many
> superblocks (each with own shrinker) but tracking and processing
> per-memcg data is pointless with nokmem (shrink_slab_memcg() is
> partially guarded with !memcg_kmem_online already).
>
> The patch uses "boottime" predicate mem_cgroup_kmem_disabled() (not
> memcg_kmem_online()) to avoid mistakenly un-MEMCG_AWARE-ing shrinkers
> registered before first non-root memcg is mkdir'd.
>
> Signed-off-by: Michal Koutný <mkoutny@suse.com>

Reviewed-by: Roman Gushchin <roman.gushchin@linux.dev>

Thanks!