Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.35 as permitted sender) client-ip=23.128.96.35;
From:   Jingbo Xu <jefflexu@linux.alibaba.com>
To:     tj@kernel.org, guro@fb.com
Cc:     lizefan.x@bytedance.com, hannes@cmpxchg.org,
        cgroups@vger.kernel.org, jack@suse.cz,
        linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        viro@zeniv.linux.org.uk, brauner@kernel.org, willy@infradead.org,
        joseph.qi@linux.alibaba.com
Subject: [PATCH] writeback, cgroup: switch inodes with dirty timestamps to release dying cgwbs
Date:   Wed, 11 Oct 2023 16:42:28 +0800
Message-Id: <20231011084228.77615-1-jefflexu@linux.alibaba.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

The cgwb cleanup routine will try to release the dying cgwb by switching
the attached inodes.  It fetches the attached inodes from wb->b_attached
list, omitting the fact that inodes only with dirty timestamps reside in
wb->b_dirty_time list, which is the case when lazytime is enabled.  This
causes enormous zombie memory cgroup when lazytime is enabled, as inodes
with dirty timestamps can not be switched to a live cgwb for a long time.

It is reasonable not to switch cgwb for inodes with dirty data, as
otherwise it may break the bandwidth restrictions.  However since the
writeback of inode metadata is not accounted, let's also switch inodes
with dirty timestamps to avoid zombie memory and block cgroups when
laztytime is enabled.

Fixs: c22d70a162d3 ("writeback, cgroup: release dying cgwbs by switching attached inodes")
Signed-off-by: Jingbo Xu <jefflexu@linux.alibaba.com>
---
This issue is reported on our production environment (5.10 kernel with
the dying cgwbs optimization[1] backported, and ext4 mounted with "-o
relatime,lazytime"), while I can also reproduce it on the latest
mainline kernel.

[1] c22d70a162d3 ("writeback, cgroup: release dying cgwbs by switching
attached inodes")
---
 fs/fs-writeback.c | 33 +++++++++++++++++++++------------
 1 file changed, 21 insertions(+), 12 deletions(-)

diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index c1af01b2c42d..89125760e4ad 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -613,6 +613,24 @@ static void inode_switch_wbs(struct inode *inode, int new_wb_id)
 	kfree(isw);
 }
 
+static bool isw_prepare_wbs_switch(struct inode_switch_wbs_context *isw,
+				   struct list_head *list, int *nr)
+{
+	struct inode *inode;
+
+	list_for_each_entry(inode, list, i_io_list) {
+		if (!inode_prepare_wbs_switch(inode, isw->new_wb))
+			continue;
+
+		isw->inodes[*nr] = inode;
+		(*nr)++;
+
+		if (*nr >= WB_MAX_INODES_PER_ISW - 1)
+			return true;
+	}
+	return false;
+}
+
 /**
  * cleanup_offline_cgwb - detach associated inodes
  * @wb: target wb
@@ -625,7 +643,6 @@ bool cleanup_offline_cgwb(struct bdi_writeback *wb)
 {
 	struct cgroup_subsys_state *memcg_css;
 	struct inode_switch_wbs_context *isw;
-	struct inode *inode;
 	int nr;
 	bool restart = false;
 
@@ -647,17 +664,9 @@ bool cleanup_offline_cgwb(struct bdi_writeback *wb)
 
 	nr = 0;
 	spin_lock(&wb->list_lock);
-	list_for_each_entry(inode, &wb->b_attached, i_io_list) {
-		if (!inode_prepare_wbs_switch(inode, isw->new_wb))
-			continue;
-
-		isw->inodes[nr++] = inode;
-
-		if (nr >= WB_MAX_INODES_PER_ISW - 1) {
-			restart = true;
-			break;
-		}
-	}
+	restart = isw_prepare_wbs_switch(isw, &wb->b_attached, &nr);
+	if (!restart)
+		restart = isw_prepare_wbs_switch(isw, &wb->b_dirty_time, &nr);
 	spin_unlock(&wb->list_lock);
 
 	/* no attached inodes? bail out */
-- 
2.19.1.6.gb485710b