[PATCH v2 3/4] ceph: flush mdlog before umounting

[xiubli@xxxxxxxxxx]

From: Xiubo Li <xiubli@xxxxxxxxxx>

Signed-off-by: Xiubo Li <xiubli@xxxxxxxxxx>
---
 fs/ceph/mds_client.c         | 25 +++++++++++++++++++++++++
 fs/ceph/mds_client.h         |  1 +
 fs/ceph/strings.c            |  1 +
 include/linux/ceph/ceph_fs.h |  1 +
 4 files changed, 28 insertions(+)

diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 5c3ec7eb8141..79aa4ce3a388 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -4690,6 +4690,30 @@ static void wait_requests(struct ceph_mds_client *mdsc)
 	dout("wait_requests done\n");
 }
 
+void send_flush_mdlog(struct ceph_mds_session *s)
+{
+	struct ceph_msg *msg;
+
+	/*
+	 * Pre-luminous MDS crashes when it sees an unknown session request
+	 */
+	if (!CEPH_HAVE_FEATURE(s->s_con.peer_features, SERVER_LUMINOUS))
+		return;
+
+	mutex_lock(&s->s_mutex);
+	dout("request mdlog flush to mds%d (%s)s seq %lld\n", s->s_mds,
+	     ceph_session_state_name(s->s_state), s->s_seq);
+	msg = ceph_create_session_msg(CEPH_SESSION_REQUEST_FLUSH_MDLOG,
+				      s->s_seq);
+	if (!msg) {
+		pr_err("failed to request mdlog flush to mds%d (%s) seq %lld\n",
+		       s->s_mds, ceph_session_state_name(s->s_state), s->s_seq);
+	} else {
+		ceph_con_send(&s->s_con, msg);
+	}
+	mutex_unlock(&s->s_mutex);
+}
+
 /*
  * called before mount is ro, and before dentries are torn down.
  * (hmm, does this still race with new lookups?)
@@ -4699,6 +4723,7 @@ void ceph_mdsc_pre_umount(struct ceph_mds_client *mdsc)
 	dout("pre_umount\n");
 	mdsc->stopping = 1;
 
+	ceph_mdsc_iterate_sessions(mdsc, send_flush_mdlog, true);
 	ceph_mdsc_iterate_sessions(mdsc, lock_unlock_session, false);
 	ceph_flush_dirty_caps(mdsc);
 	wait_requests(mdsc);
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index fca2cf427eaf..a7af09257382 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -523,6 +523,7 @@ static inline void ceph_mdsc_put_request(struct ceph_mds_request *req)
 	kref_put(&req->r_kref, ceph_mdsc_release_request);
 }
 
+extern void send_flush_mdlog(struct ceph_mds_session *s);
 extern void ceph_mdsc_iterate_sessions(struct ceph_mds_client *mdsc,
 				       void (*cb)(struct ceph_mds_session *),
 				       bool check_state);
diff --git a/fs/ceph/strings.c b/fs/ceph/strings.c
index 4a79f3632260..573bb9556fb5 100644
--- a/fs/ceph/strings.c
+++ b/fs/ceph/strings.c
@@ -46,6 +46,7 @@ const char *ceph_session_op_name(int op)
 	case CEPH_SESSION_FLUSHMSG_ACK: return "flushmsg_ack";
 	case CEPH_SESSION_FORCE_RO: return "force_ro";
 	case CEPH_SESSION_REJECT: return "reject";
+	case CEPH_SESSION_REQUEST_FLUSH_MDLOG: return "flush_mdlog";
 	}
 	return "???";
 }
diff --git a/include/linux/ceph/ceph_fs.h b/include/linux/ceph/ceph_fs.h
index 57e5bd63fb7a..ae60696fe40b 100644
--- a/include/linux/ceph/ceph_fs.h
+++ b/include/linux/ceph/ceph_fs.h
@@ -300,6 +300,7 @@ enum {
 	CEPH_SESSION_FLUSHMSG_ACK,
 	CEPH_SESSION_FORCE_RO,
 	CEPH_SESSION_REJECT,
+	CEPH_SESSION_REQUEST_FLUSH_MDLOG,
 };
 
 extern const char *ceph_session_op_name(int op);
-- 
2.27.0