linux-kernel - Re: [PATCH 4/5] mm/slub: Limit min

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <91cc8ab-a0f0-2687-df99-10b2267c7a9@google.com>
Date:   Tue, 22 Feb 2022 15:48:16 -0800 (PST)
From:   David Rientjes <rientjes@...gle.com>
To:     Hyeonggon Yoo <42.hyeyoo@...il.com>
cc:     linux-mm@...ck.org, Roman Gushchin <guro@...com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Vlastimil Babka <vbabka@...e.cz>, linux-kernel@...r.kernel.org,
        Joonsoo Kim <iamjoonsoo.kim@....com>,
        Christoph Lameter <cl@...ux.com>,
        Pekka Enberg <penberg@...nel.org>
Subject: Re: [PATCH 4/5] mm/slub: Limit min_partial only in cache creation

On Mon, 21 Feb 2022, Hyeonggon Yoo wrote:

> SLUB sets number of minimum partial slabs for node (min_partial) using
> set_min_partial(). SLUB holds at least min_partial slabs even if they're empty
> to avoid excessive use of page allocator.
> 
> set_min_partial() limits value of min_partial between MIN_PARTIAL and
> MAX_PARTIAL. As set_min_partial() can be called by min_partial_store()
> too, Only limit value of min_partial in kmem_cache_open() so that it
> can be changed to value that a user wants.
> 
> Signed-off-by: Hyeonggon Yoo <42.hyeyoo@...il.com>

I think this makes sense and there is no reason to limit the bounds that 
may be set at runtime with undocumented behavior.

However, since set_min_partial() is only setting the value in the 
kmem_cache, could we remove the helper function entirely and fold it into 
its two callers?

> ---
>  mm/slub.c | 11 ++++++-----
>  1 file changed, 6 insertions(+), 5 deletions(-)
> 
> diff --git a/mm/slub.c b/mm/slub.c
> index 3a4458976ab7..a4964deccb61 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -4002,10 +4002,6 @@ static int init_kmem_cache_nodes(struct kmem_cache *s)
>  
>  static void set_min_partial(struct kmem_cache *s, unsigned long min)
>  {
> -	if (min < MIN_PARTIAL)
> -		min = MIN_PARTIAL;
> -	else if (min > MAX_PARTIAL)
> -		min = MAX_PARTIAL;
>  	s->min_partial = min;
>  }
>  
> @@ -4184,6 +4180,8 @@ static int calculate_sizes(struct kmem_cache *s, int forced_order)
>  
>  static int kmem_cache_open(struct kmem_cache *s, slab_flags_t flags)
>  {
> +	int min_partial;
> +
>  	s->flags = kmem_cache_flags(s->size, flags, s->name);
>  #ifdef CONFIG_SLAB_FREELIST_HARDENED
>  	s->random = get_random_long();
> @@ -4215,7 +4213,10 @@ static int kmem_cache_open(struct kmem_cache *s, slab_flags_t flags)
>  	 * The larger the object size is, the more slabs we want on the partial
>  	 * list to avoid pounding the page allocator excessively.
>  	 */
> -	set_min_partial(s, ilog2(s->size) / 2);
> +	min_partial = min(MAX_PARTIAL, ilog2(s->size) / 2);
> +	min_partial = max(MIN_PARTIAL, min_partial);
> +
> +	set_min_partial(s, min_partial);
>  
>  	set_cpu_partial(s);
>  
> -- 
> 2.33.1
> 
>