2024-03-11 13:27:48

by Xiongwei Song

[permalink] [raw]
Subject: [PATCH] mm/slub: Simplify get_partial_node()

From: Xiongwei Song <[email protected]>

Remove the check of !kmem_cache_has_cpu_partial() because it is always
false, we've known this by calling kmem_cache_debug() before calling
remove_partial(), so we can remove the check.

Meanwhile, redo filling cpu partial and add comment to improve the
readability.

Signed-off-by: Xiongwei Song <[email protected]>
---
mm/slub.c | 22 ++++++++++++----------
1 file changed, 12 insertions(+), 10 deletions(-)

diff --git a/mm/slub.c b/mm/slub.c
index a3ab096c38c0..62388f2a0ac7 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -2620,19 +2620,21 @@ static struct slab *get_partial_node(struct kmem_cache *s,
if (!partial) {
partial = slab;
stat(s, ALLOC_FROM_PARTIAL);
- } else {
+
+ /* Fill cpu partial if needed from next iteration, or break */
+ if (IS_ENABLED(CONFIG_SLUB_CPU_PARTIAL))
+ continue;
+ else
+ break;
+ }
+
+ if (IS_ENABLED(CONFIG_SLUB_CPU_PARTIAL)) {
put_cpu_partial(s, slab, 0);
stat(s, CPU_PARTIAL_NODE);
- partial_slabs++;
- }
-#ifdef CONFIG_SLUB_CPU_PARTIAL
- if (!kmem_cache_has_cpu_partial(s)
- || partial_slabs > s->cpu_partial_slabs / 2)
- break;
-#else
- break;
-#endif

+ if (++partial_slabs > s->cpu_partial_slabs/2)
+ break;
+ }
}
spin_unlock_irqrestore(&n->list_lock, flags);
return partial;
--
2.34.1



2024-03-12 06:23:12

by Chengming Zhou

[permalink] [raw]
Subject: Re: [PATCH] mm/slub: Simplify get_partial_node()

On 2024/3/11 21:27, [email protected] wrote:
> From: Xiongwei Song <[email protected]>
>
> Remove the check of !kmem_cache_has_cpu_partial() because it is always
> false, we've known this by calling kmem_cache_debug() before calling
> remove_partial(), so we can remove the check.

This is correct.

>
> Meanwhile, redo filling cpu partial and add comment to improve the
> readability.
>
> Signed-off-by: Xiongwei Song <[email protected]>
> ---
> mm/slub.c | 22 ++++++++++++----------
> 1 file changed, 12 insertions(+), 10 deletions(-)
>
> diff --git a/mm/slub.c b/mm/slub.c
> index a3ab096c38c0..62388f2a0ac7 100644
> --- a/mm/slub.c
> +++ b/mm/slub.c
> @@ -2620,19 +2620,21 @@ static struct slab *get_partial_node(struct kmem_cache *s,
> if (!partial) {
> partial = slab;
> stat(s, ALLOC_FROM_PARTIAL);
> - } else {
> +
> + /* Fill cpu partial if needed from next iteration, or break */
> + if (IS_ENABLED(CONFIG_SLUB_CPU_PARTIAL))
> + continue;
> + else
> + break;
> + }
> +
> + if (IS_ENABLED(CONFIG_SLUB_CPU_PARTIAL)) {

But this won't work since "s->cpu_partial_slabs" is defined under CONFIG_SLUB_CPU_PARTIAL,
you would get compiler error if building without CONFIG_SLUB_CPU_PARTIAL.

Thanks.

> put_cpu_partial(s, slab, 0);
> stat(s, CPU_PARTIAL_NODE);
> - partial_slabs++;
> - }
> -#ifdef CONFIG_SLUB_CPU_PARTIAL
> - if (!kmem_cache_has_cpu_partial(s)
> - || partial_slabs > s->cpu_partial_slabs / 2)
> - break;
> -#else
> - break;
> -#endif
>
> + if (++partial_slabs > s->cpu_partial_slabs/2)
> + break;
> + }
> }
> spin_unlock_irqrestore(&n->list_lock, flags);
> return partial;

2024-03-12 07:47:00

by Song, Xiongwei

[permalink] [raw]
Subject: RE: [PATCH] mm/slub: Simplify get_partial_node()

> On 2024/3/11 21:27, [email protected] wrote:
> > From: Xiongwei Song <[email protected]>
> >
> > Remove the check of !kmem_cache_has_cpu_partial() because it is always
> > false, we've known this by calling kmem_cache_debug() before calling
> > remove_partial(), so we can remove the check.
>
> This is correct.
>
> >
> > Meanwhile, redo filling cpu partial and add comment to improve the
> > readability.
> >
> > Signed-off-by: Xiongwei Song <[email protected]>
> > ---
> > mm/slub.c | 22 ++++++++++++----------
> > 1 file changed, 12 insertions(+), 10 deletions(-)
> >
> > diff --git a/mm/slub.c b/mm/slub.c
> > index a3ab096c38c0..62388f2a0ac7 100644
> > --- a/mm/slub.c
> > +++ b/mm/slub.c
> > @@ -2620,19 +2620,21 @@ static struct slab *get_partial_node(struct
> kmem_cache *s,
> > if (!partial) {
> > partial = slab;
> > stat(s, ALLOC_FROM_PARTIAL);
> > - } else {
> > +
> > + /* Fill cpu partial if needed from next iteration, or break */
> > + if (IS_ENABLED(CONFIG_SLUB_CPU_PARTIAL))
> > + continue;
> > + else
> > + break;
> > + }
> > +
> > + if (IS_ENABLED(CONFIG_SLUB_CPU_PARTIAL)) {
>
> But this won't work since "s->cpu_partial_slabs" is defined under
> CONFIG_SLUB_CPU_PARTIAL,
> you would get compiler error if building without CONFIG_SLUB_CPU_PARTIAL.

Yes, maybe we can use "#if IS_ENABLED(CONFIG_SLUB_CPU_PARTIAL)" instead.

Regards,
Xiongwei

>
> Thanks.
>
> > put_cpu_partial(s, slab, 0);
> > stat(s, CPU_PARTIAL_NODE);
> > - partial_slabs++;
> > - }
> > -#ifdef CONFIG_SLUB_CPU_PARTIAL
> > - if (!kmem_cache_has_cpu_partial(s)
> > - || partial_slabs > s->cpu_partial_slabs / 2)
> > - break;
> > -#else
> > - break;
> > -#endif
> >
> > + if (++partial_slabs > s->cpu_partial_slabs/2)
> > + break;
> > + }
> > }
> > spin_unlock_irqrestore(&n->list_lock, flags);
> > return partial;