[Ocfs2-tools-devel] [PATCH v2 2/3] ocfs2: fix cluster unlocked if journal is replayed

[Joseph Qi]

Currently maybe_replay_journals will close and open fs again if journal
is replayed. And dlm and cluster info will be lost in this case, which
will lead to cluster unlocked. So we have to unlock cluster before close
fs and relock it again after open it.

Signed-off-by: Joseph Qi <joseph.qi at huawei.com>
Signed-off-by: Yiwen Jiang <jiangyiwen at huawei.com>
Reviewed-by: Jun Piao <piaojun at huawei.com>
---
 fsck.ocfs2/fsck.c | 117 ++++++++++++++++++++++++++++++++++--------------------
 1 file changed, 75 insertions(+), 42 deletions(-)

diff --git a/fsck.ocfs2/fsck.c b/fsck.ocfs2/fsck.c
index f71fb21..7886f32 100644
--- a/fsck.ocfs2/fsck.c
+++ b/fsck.ocfs2/fsck.c
@@ -79,6 +79,8 @@ static o2fsck_state _ost;
 static int cluster_locked = 0;

 static void mark_magical_clusters(o2fsck_state *ost);
+static errcode_t fsck_lock_fs(o2fsck_state *ost);
+static void fsck_unlock_fs(o2fsck_state *ost);

 static void handle_signal(int sig)
 {
@@ -518,6 +520,8 @@ static errcode_t maybe_replay_journals(o2fsck_state *ost, char *filename,
 	if (!replayed)
 		goto out;

+	fsck_unlock_fs(ost);
+
 	ret = ocfs2_close(ost->ost_fs);
 	if (ret) {
 		com_err(whoami, ret, "while closing \"%s\"", filename);
@@ -525,6 +529,13 @@ static errcode_t maybe_replay_journals(o2fsck_state *ost, char *filename,
 	}

 	ret = open_and_check(ost, filename, open_flags, blkno, blksize);
+	if (ret) {
+		com_err(whoami, ret, "while opening and checking \"%s\"",
+				filename);
+		goto out;
+	}
+
+	ret = fsck_lock_fs(ost);
 out:
 	return ret;
 }
@@ -676,6 +687,67 @@ bail:
 	return ret;
 }

+static errcode_t fsck_lock_fs(o2fsck_state *ost)
+{
+	errcode_t ret;
+
+	ret = o2cb_init();
+	if (ret) {
+		com_err(whoami, ret, "while initializing the cluster");
+		goto bail;
+	}
+
+	block_signals(SIG_BLOCK);
+	ret = ocfs2_initialize_dlm(ost->ost_fs, whoami);
+	if (ret == O2CB_ET_INVALID_STACK_NAME ||
+			ret == O2CB_ET_INVALID_CLUSTER_NAME ||
+			ret == O2CB_ET_INVALID_HEARTBEAT_MODE) {
+		block_signals(SIG_UNBLOCK);
+		ret = recover_cluster_info(ost);
+		if (ret) {
+			com_err(whoami, ret,
+					"while recovering cluster information");
+			goto bail;
+		}
+		block_signals(SIG_BLOCK);
+		ret = ocfs2_initialize_dlm(ost->ost_fs, whoami);
+	}
+	if (ret) {
+		block_signals(SIG_UNBLOCK);
+		com_err(whoami, ret, "while initializing the DLM");
+		goto bail;
+	}
+
+	ret = ocfs2_lock_down_cluster(ost->ost_fs);
+	if (ret) {
+		if (ost->ost_fs->fs_dlm_ctxt)
+			ocfs2_shutdown_dlm(ost->ost_fs, whoami);
+		block_signals(SIG_UNBLOCK);
+		com_err(whoami, ret, "while locking down the cluster");
+		goto bail;
+	}
+	cluster_locked = 1;
+	block_signals(SIG_UNBLOCK);
+
+bail:
+	return ret;
+}
+
+static void fsck_unlock_fs(o2fsck_state *ost)
+{
+	if (!ost->ost_fs)
+		return;
+
+	block_signals(SIG_BLOCK);
+	if (ost->ost_fs->fs_dlm_ctxt)
+		ocfs2_release_cluster(ost->ost_fs);
+	cluster_locked = 0;
+
+	if (ost->ost_fs->fs_dlm_ctxt)
+		ocfs2_shutdown_dlm(ost->ost_fs, whoami);
+	block_signals(SIG_UNBLOCK);
+}
+
 int main(int argc, char **argv)
 {
 	char *filename;
@@ -902,41 +974,11 @@ int main(int argc, char **argv)

 	if (open_flags & OCFS2_FLAG_RW && !ost->ost_skip_o2cb &&
 	    !ocfs2_mount_local(ost->ost_fs)) {
-		ret = o2cb_init();
+		ret = fsck_lock_fs(ost);
 		if (ret) {
-			com_err(whoami, ret, "while initializing the cluster");
-			goto close;
-		}
-
-		block_signals(SIG_BLOCK);
-		ret = ocfs2_initialize_dlm(ost->ost_fs, whoami);
-		if (ret == O2CB_ET_INVALID_STACK_NAME ||
-		    ret == O2CB_ET_INVALID_CLUSTER_NAME ||
-		    ret == O2CB_ET_INVALID_HEARTBEAT_MODE) {
-			block_signals(SIG_UNBLOCK);
-			ret = recover_cluster_info(ost);
-			if (ret) {
-				com_err(whoami, ret,
-					"while recovering cluster information");
-				goto close;
-			}
-			block_signals(SIG_BLOCK);
-			ret = ocfs2_initialize_dlm(ost->ost_fs, whoami);
-		}
-		if (ret) {
-			block_signals(SIG_UNBLOCK);
-			com_err(whoami, ret, "while initializing the DLM");
-			goto close;
-		}
-
-		ret = ocfs2_lock_down_cluster(ost->ost_fs);
-		if (ret) {
-			block_signals(SIG_UNBLOCK);
-			com_err(whoami, ret, "while locking down the cluster");
+			fsck_mask |= FSCK_ERROR;
 			goto close;
 		}
-		cluster_locked = 1;
-		block_signals(SIG_UNBLOCK);
 	}

 	printf("Checking OCFS2 filesystem in %s:\n", filename);
@@ -1086,18 +1128,9 @@ clear_dirty_flag:
 	}

 unlock:
-	block_signals(SIG_BLOCK);
-	if (ost->ost_fs->fs_dlm_ctxt)
-		ocfs2_release_cluster(ost->ost_fs);
-	cluster_locked = 0;
-	block_signals(SIG_UNBLOCK);
+	fsck_unlock_fs(ost);

 close:
-	block_signals(SIG_BLOCK);
-	if (ost->ost_fs->fs_dlm_ctxt)
-		ocfs2_shutdown_dlm(ost->ost_fs, whoami);
-	block_signals(SIG_UNBLOCK);
-
 	if (ost->ost_fs) {
 		ret = ocfs2_close(ost->ost_fs);
 		if (ret) {
-- 
1.8.4.3