Received-SPF: pass (google.com: domain of linux-kernel+bounces-151839-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
From: Jianfeng Wang <jianfeng.w.wang@oracle.com>
To: linux-mm@kvack.org, linux-kernel@vger.kernel.org
Cc: vbabka@suse.cz, cl@linux.com, akpm@linux-foundation.org,
        penberg@kernel.org, rientjes@google.com
Subject: [PATCH v3 2/2] slub: use count_partial_free_approx() in slab_out_of_memory()
Date: Fri, 19 Apr 2024 10:56:11 -0700
Message-ID: <20240419175611.47413-3-jianfeng.w.wang@oracle.com>
In-Reply-To: <20240419175611.47413-1-jianfeng.w.wang@oracle.com>
References: <20240419175611.47413-1-jianfeng.w.wang@oracle.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

slab_out_of_memory() uses count_partial() to get the exact count
of free objects for each node. As it may get called in the slab
allocation path, count_partial_free_approx() can be used to avoid
the risk and overhead of traversing a long partial slab list.

At the same time, show_slab_objects() still uses count_partial().
Thus, slub users can still have the option to access the exact
count of objects via sysfs if the overhead is acceptable to them.

Signed-off-by: Jianfeng Wang <jianfeng.w.wang@oracle.com>
---
 mm/slub.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/slub.c b/mm/slub.c
index 993cbbdd2b6c..fa55699aa21a 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -3292,7 +3292,7 @@ slab_out_of_memory(struct kmem_cache *s, gfp_t gfpflags, int nid)
 		unsigned long nr_objs;
 		unsigned long nr_free;
 
-		nr_free  = count_partial(n, count_free);
+		nr_free  = count_partial_free_approx(n);
 		nr_slabs = node_nr_slabs(n);
 		nr_objs  = node_nr_objs(n);
 
-- 
2.42.1