Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Wang Yugui <wangyugui@e16-tech.com>
To:     linux-nfs@vger.kernel.org
Cc:     Wang Yugui <wangyugui@e16-tech.com>
Subject: [PATCH v2] nfsd: serialize filecache garbage collector
Date:   Tue, 31 May 2022 18:34:27 +0800
Message-Id: <20220531103427.47769-1-wangyugui@e16-tech.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

When many(>NFSD_FILE_LRU_THRESHOLD) files are kept as OPEN, such as
xfstests generic/531, nfsd proceses are in CPU high-load state,
and nfsd_file_gc(nfsd filecache garbage collector) waste many CPU times.

concurrency nfsd_file_gc() is almost meaningless, so serialize it.

Signed-off-by: Wang Yugui <wangyugui@e16-tech.com>
---
Changes since v1:
- add static to 'atomic_t nfsd_file_gc_running'.
  thanks for kernel test robot <lkp@intel.com>

 fs/nfsd/filecache.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
index f172412447f5..28a8f8d6d235 100644
--- a/fs/nfsd/filecache.c
+++ b/fs/nfsd/filecache.c
@@ -471,10 +471,15 @@ nfsd_file_lru_walk_list(struct shrink_control *sc)
 	return ret;
 }
 
+/* concurrency nfsd_file_gc() is almost meaningless, so serialize it. */
+static atomic_t nfsd_file_gc_running = ATOMIC_INIT(0);
 static void
 nfsd_file_gc(void)
 {
-	nfsd_file_lru_walk_list(NULL);
+	if(atomic_cmpxchg(&nfsd_file_gc_running, 0, 1) == 0) {
+		nfsd_file_lru_walk_list(NULL);
+		atomic_set(&nfsd_file_gc_running, 0);
+	}
 }
 
 static void
-- 
2.36.1