Re: [PATCH RFC v2 03/20] mm/slab: make caches with sheaves mergeable

[Suren Baghdasaryan]

On Mon, Jan 12, 2026 at 3:17 PM Vlastimil Babka <vbabka@xxxxxxx> wrote:
>
> Before enabling sheaves for all caches (with automatically determined
> capacity), their enablement should no longer prevent merging of caches.
> Limit this merge prevention only to caches that were created with a
> specific sheaf capacity, by adding the SLAB_NO_MERGE flag to them.
>
> Signed-off-by: Vlastimil Babka <vbabka@xxxxxxx>
> ---
>  mm/slab_common.c | 13 +++++++------
>  1 file changed, 7 insertions(+), 6 deletions(-)
>
> diff --git a/mm/slab_common.c b/mm/slab_common.c
> index 52591d9c04f3..54c17dc6d5ec 100644
> --- a/mm/slab_common.c
> +++ b/mm/slab_common.c
> @@ -163,9 +163,6 @@ int slab_unmergeable(struct kmem_cache *s)
>                 return 1;
>  #endif
>
> -       if (s->cpu_sheaves)
> -               return 1;
> -
>         /*
>          * We may have set a slab to be unmergeable during bootstrap.
>          */
> @@ -190,9 +187,6 @@ static struct kmem_cache *find_mergeable(unsigned int size, slab_flags_t flags,
>         if (IS_ENABLED(CONFIG_HARDENED_USERCOPY) && args->usersize)
>                 return NULL;
>
> -       if (args->sheaf_capacity)
> -               return NULL;
> -
>         flags = kmem_cache_flags(flags, name);
>
>         if (flags & SLAB_NEVER_MERGE)
> @@ -337,6 +331,13 @@ struct kmem_cache *__kmem_cache_create_args(const char *name,
>         flags &= ~SLAB_DEBUG_FLAGS;
>  #endif
>
> +       /*
> +        * Caches with specific capacity are special enough. It's simpler to
> +        * make them unmergeable.
> +        */
> +       if (args->sheaf_capacity)
> +               flags |= SLAB_NO_MERGE;

So, this is very subtle and maybe not that important but the comment
for kmem_cache_args.sheaf_capacity claims "When slub_debug is enabled
for the cache, the sheaf_capacity argument is ignored.". With this
change this argument is not completely ignored anymore... It sets
SLAB_NO_MERGE even if slub_debug is enabled, doesn't it?

> +
>         mutex_lock(&slab_mutex);
>
>         err = kmem_cache_sanity_check(name, object_size);
>
> --
> 2.52.0
>