GFS2: dlm based recovery coordination

This new method of managing recovery is an alternative to
the previous approach of using the userland gfs_controld.

- use dlm slot numbers to assign journal id's
- use dlm recovery callbacks to initiate journal recovery
- use a dlm lock to determine the first node to mount fs
- use a dlm lock to track journals that need recovery

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

fs/gfs2/glock.c

@@ -1353,7 +1353,7 @@ void gfs2_glock_complete(struct gfs2_glock *gl, int ret)
 	spin_lock(&gl->gl_spin);
 	gl->gl_reply = ret;
 
-	if (unlikely(test_bit(DFL_BLOCK_LOCKS, &ls->ls_flags))) {
+	if (unlikely(test_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags))) {
 		if (gfs2_should_freeze(gl)) {
 			set_bit(GLF_FROZEN, &gl->gl_flags);
 			spin_unlock(&gl->gl_spin);

fs/gfs2/glock.h

@@ -121,8 +121,11 @@ enum {
 
 struct lm_lockops {
 	const char *lm_proto_name;
-	int (*lm_mount) (struct gfs2_sbd *sdp, const char *fsname);
- 	void (*lm_unmount) (struct gfs2_sbd *sdp);
+	int (*lm_mount) (struct gfs2_sbd *sdp, const char *table);
+	void (*lm_first_done) (struct gfs2_sbd *sdp);
+	void (*lm_recovery_result) (struct gfs2_sbd *sdp, unsigned int jid,
+				    unsigned int result);
+	void (*lm_unmount) (struct gfs2_sbd *sdp);
 	void (*lm_withdraw) (struct gfs2_sbd *sdp);
 	void (*lm_put_lock) (struct gfs2_glock *gl);
 	int (*lm_lock) (struct gfs2_glock *gl, unsigned int req_state,

fs/gfs2/incore.h

@@ -139,8 +139,45 @@ struct gfs2_bufdata {
 #define GDLM_STRNAME_BYTES	25
 #define GDLM_LVB_SIZE		32
 
+/*
+ * ls_recover_flags:
+ *
+ * DFL_BLOCK_LOCKS: dlm is in recovery and will grant locks that had been
+ * held by failed nodes whose journals need recovery.  Those locks should
+ * only be used for journal recovery until the journal recovery is done.
+ * This is set by the dlm recover_prep callback and cleared by the
+ * gfs2_control thread when journal recovery is complete.  To avoid
+ * races between recover_prep setting and gfs2_control clearing, recover_spin
+ * is held while changing this bit and reading/writing recover_block
+ * and recover_start.
+ *
+ * DFL_NO_DLM_OPS: dlm lockspace ops/callbacks are not being used.
+ *
+ * DFL_FIRST_MOUNT: this node is the first to mount this fs and is doing
+ * recovery of all journals before allowing other nodes to mount the fs.
+ * This is cleared when FIRST_MOUNT_DONE is set.
+ *
+ * DFL_FIRST_MOUNT_DONE: this node was the first mounter, and has finished
+ * recovery of all journals, and now allows other nodes to mount the fs.
+ *
+ * DFL_MOUNT_DONE: gdlm_mount has completed successfully and cleared
+ * BLOCK_LOCKS for the first time.  The gfs2_control thread should now
+ * control clearing BLOCK_LOCKS for further recoveries.
+ *
+ * DFL_UNMOUNT: gdlm_unmount sets to keep sdp off gfs2_control_wq.
+ *
+ * DFL_DLM_RECOVERY: set while dlm is in recovery, between recover_prep()
+ * and recover_done(), i.e. set while recover_block == recover_start.
+ */
+
 enum {
 	DFL_BLOCK_LOCKS		= 0,
+	DFL_NO_DLM_OPS		= 1,
+	DFL_FIRST_MOUNT		= 2,
+	DFL_FIRST_MOUNT_DONE	= 3,
+	DFL_MOUNT_DONE		= 4,
+	DFL_UNMOUNT		= 5,
+	DFL_DLM_RECOVERY	= 6,
 };
 
 struct lm_lockname {
@@ -499,14 +536,26 @@ struct gfs2_sb_host {
 struct lm_lockstruct {
 	int ls_jid;
 	unsigned int ls_first;
-	unsigned int ls_first_done;
 	unsigned int ls_nodir;
 	const struct lm_lockops *ls_ops;
-	unsigned long ls_flags;
 	dlm_lockspace_t *ls_dlm;
 
-	int ls_recover_jid_done;
-	int ls_recover_jid_status;
+	int ls_recover_jid_done;   /* These two are deprecated, */
+	int ls_recover_jid_status; /* used previously by gfs_controld */
+
+	struct dlm_lksb ls_mounted_lksb; /* mounted_lock */
+	struct dlm_lksb ls_control_lksb; /* control_lock */
+	char ls_control_lvb[GDLM_LVB_SIZE]; /* control_lock lvb */
+	struct completion ls_sync_wait; /* {control,mounted}_{lock,unlock} */
+
+	spinlock_t ls_recover_spin; /* protects following fields */
+	unsigned long ls_recover_flags; /* DFL_ */
+	uint32_t ls_recover_mount; /* gen in first recover_done cb */
+	uint32_t ls_recover_start; /* gen in last recover_done cb */
+	uint32_t ls_recover_block; /* copy recover_start in last recover_prep */
+	uint32_t ls_recover_size; /* size of recover_submit, recover_result */
+	uint32_t *ls_recover_submit; /* gen in last recover_slot cb per jid */
+	uint32_t *ls_recover_result; /* result of last jid recovery */
 };
 
 struct gfs2_sbd {
@@ -544,6 +593,7 @@ struct gfs2_sbd {
 	wait_queue_head_t sd_glock_wait;
 	atomic_t sd_glock_disposal;
 	struct completion sd_locking_init;
+	struct delayed_work sd_control_work;
 
 	/* Inode Stuff */
 

fs/gfs2/main.c

@@ -28,6 +28,8 @@
 #include "recovery.h"
 #include "dir.h"
 
+struct workqueue_struct *gfs2_control_wq;
+
 static struct shrinker qd_shrinker = {
 	.shrink = gfs2_shrink_qd_memory,
 	.seeks = DEFAULT_SEEKS,
@@ -146,12 +148,19 @@ static int __init init_gfs2_fs(void)
 	if (!gfs_recovery_wq)
 		goto fail_wq;
 
+	gfs2_control_wq = alloc_workqueue("gfs2_control",
+			       WQ_NON_REENTRANT | WQ_UNBOUND | WQ_FREEZABLE, 0);
+	if (!gfs2_control_wq)
+		goto fail_control;
+
 	gfs2_register_debugfs();
 
 	printk("GFS2 installed\n");
 
 	return 0;
 
+fail_control:
+	destroy_workqueue(gfs_recovery_wq);
 fail_wq:
 	unregister_filesystem(&gfs2meta_fs_type);
 fail_unregister:
@@ -195,6 +204,7 @@ static void __exit exit_gfs2_fs(void)
 	unregister_filesystem(&gfs2_fs_type);
 	unregister_filesystem(&gfs2meta_fs_type);
 	destroy_workqueue(gfs_recovery_wq);
+	destroy_workqueue(gfs2_control_wq);
 
 	rcu_barrier();
 

fs/gfs2/ops_fstype.c

@@ -562,8 +562,12 @@ static void gfs2_others_may_mount(struct gfs2_sbd *sdp)
 {
 	char *message = "FIRSTMOUNT=Done";
 	char *envp[] = { message, NULL };
-	struct lm_lockstruct *ls = &sdp->sd_lockstruct;
-	ls->ls_first_done = 1;
+
+	fs_info(sdp, "first mount done, others may mount\n");
+
+	if (sdp->sd_lockstruct.ls_ops->lm_first_done)
+		sdp->sd_lockstruct.ls_ops->lm_first_done(sdp);
+
 	kobject_uevent_env(&sdp->sd_kobj, KOBJ_CHANGE, envp);
 }
 
@@ -944,7 +948,6 @@ static int gfs2_lm_mount(struct gfs2_sbd *sdp, int silent)
 	struct gfs2_args *args = &sdp->sd_args;
 	const char *proto = sdp->sd_proto_name;
 	const char *table = sdp->sd_table_name;
-	const char *fsname;
 	char *o, *options;
 	int ret;
 
@@ -1004,21 +1007,12 @@ hostdata_error:
 		}
 	}
 
-	if (sdp->sd_args.ar_spectator)
-		snprintf(sdp->sd_fsname, GFS2_FSNAME_LEN, "%s.s", table);
-	else
-		snprintf(sdp->sd_fsname, GFS2_FSNAME_LEN, "%s.%u", table,
-			 sdp->sd_lockstruct.ls_jid);
-
-	fsname = strchr(table, ':');
-	if (fsname)
-		fsname++;
 	if (lm->lm_mount == NULL) {
 		fs_info(sdp, "Now mounting FS...\n");
 		complete_all(&sdp->sd_locking_init);
 		return 0;
 	}
-	ret = lm->lm_mount(sdp, fsname);
+	ret = lm->lm_mount(sdp, table);
 	if (ret == 0)
 		fs_info(sdp, "Joined cluster. Now mounting FS...\n");
 	complete_all(&sdp->sd_locking_init);
@@ -1124,6 +1118,8 @@ static int fill_super(struct super_block *sb, struct gfs2_args *args, int silent
 	if (error)
 		goto fail;
 
+	snprintf(sdp->sd_fsname, GFS2_FSNAME_LEN, "%s", sdp->sd_table_name);
+
 	gfs2_create_debugfs_file(sdp);
 
 	error = gfs2_sys_fs_add(sdp);
@@ -1160,6 +1156,13 @@ static int fill_super(struct super_block *sb, struct gfs2_args *args, int silent
 		goto fail_sb;
 	}
 
+	if (sdp->sd_args.ar_spectator)
+		snprintf(sdp->sd_fsname, GFS2_FSNAME_LEN, "%s.s",
+			 sdp->sd_table_name);
+	else
+		snprintf(sdp->sd_fsname, GFS2_FSNAME_LEN, "%s.%u",
+			 sdp->sd_table_name, sdp->sd_lockstruct.ls_jid);
+
 	error = init_inodes(sdp, DO);
 	if (error)
 		goto fail_sb;

fs/gfs2/recovery.c

@@ -436,12 +436,16 @@ static void gfs2_recovery_done(struct gfs2_sbd *sdp, unsigned int jid,
 	char env_status[20];
 	char *envp[] = { env_jid, env_status, NULL };
 	struct lm_lockstruct *ls = &sdp->sd_lockstruct;
+
         ls->ls_recover_jid_done = jid;
         ls->ls_recover_jid_status = message;
 	sprintf(env_jid, "JID=%d", jid);
 	sprintf(env_status, "RECOVERY=%s",
 		message == LM_RD_SUCCESS ? "Done" : "Failed");
         kobject_uevent_env(&sdp->sd_kobj, KOBJ_CHANGE, envp);
+
+	if (sdp->sd_lockstruct.ls_ops->lm_recovery_result)
+		sdp->sd_lockstruct.ls_ops->lm_recovery_result(sdp, jid, message);
 }
 
 void gfs2_recover_func(struct work_struct *work)

fs/gfs2/sys.c

@@ -298,7 +298,7 @@ static ssize_t block_show(struct gfs2_sbd *sdp, char *buf)
 	ssize_t ret;
 	int val = 0;
 
-	if (test_bit(DFL_BLOCK_LOCKS, &ls->ls_flags))
+	if (test_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags))
 		val = 1;
 	ret = sprintf(buf, "%d\n", val);
 	return ret;
@@ -313,9 +313,9 @@ static ssize_t block_store(struct gfs2_sbd *sdp, const char *buf, size_t len)
 	val = simple_strtol(buf, NULL, 0);
 
 	if (val == 1)
-		set_bit(DFL_BLOCK_LOCKS, &ls->ls_flags);
+		set_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
 	else if (val == 0) {
-		clear_bit(DFL_BLOCK_LOCKS, &ls->ls_flags);
+		clear_bit(DFL_BLOCK_LOCKS, &ls->ls_recover_flags);
 		smp_mb__after_clear_bit();
 		gfs2_glock_thaw(sdp);
 	} else {
@@ -350,8 +350,8 @@ static ssize_t lkfirst_store(struct gfs2_sbd *sdp, const char *buf, size_t len)
 		goto out;
 	if (sdp->sd_lockstruct.ls_ops->lm_mount == NULL)
 		goto out;
-        sdp->sd_lockstruct.ls_first = first;
-        rv = 0;
+	sdp->sd_lockstruct.ls_first = first;
+	rv = 0;
 out:
         spin_unlock(&sdp->sd_jindex_spin);
         return rv ? rv : len;
@@ -360,19 +360,14 @@ out:
 static ssize_t first_done_show(struct gfs2_sbd *sdp, char *buf)
 {
 	struct lm_lockstruct *ls = &sdp->sd_lockstruct;
-	return sprintf(buf, "%d\n", ls->ls_first_done);
+	return sprintf(buf, "%d\n", !!test_bit(DFL_FIRST_MOUNT_DONE, &ls->ls_recover_flags));
 }
 
-static ssize_t recover_store(struct gfs2_sbd *sdp, const char *buf, size_t len)
+int gfs2_recover_set(struct gfs2_sbd *sdp, unsigned jid)
 {
-	unsigned jid;
 	struct gfs2_jdesc *jd;
 	int rv;
 
-	rv = sscanf(buf, "%u", &jid);
-	if (rv != 1)
-		return -EINVAL;
-
 	rv = -ESHUTDOWN;
 	spin_lock(&sdp->sd_jindex_spin);
 	if (test_bit(SDF_NORECOVERY, &sdp->sd_flags))
@@ -389,6 +384,20 @@ static ssize_t recover_store(struct gfs2_sbd *sdp, const char *buf, size_t len)
 	}
 out:
 	spin_unlock(&sdp->sd_jindex_spin);
+	return rv;
+}
+
+static ssize_t recover_store(struct gfs2_sbd *sdp, const char *buf, size_t len)
+{
+	unsigned jid;
+	int rv;
+
+	rv = sscanf(buf, "%u", &jid);
+	if (rv != 1)
+		return -EINVAL;
+
+	rv = gfs2_recover_set(sdp, jid);
+
 	return rv ? rv : len;
 }
 

fs/gfs2/sys.h

@@ -19,5 +19,7 @@ void gfs2_sys_fs_del(struct gfs2_sbd *sdp);
 int gfs2_sys_init(void);
 void gfs2_sys_uninit(void);
 
+int gfs2_recover_set(struct gfs2_sbd *sdp, unsigned jid);
+
 #endif /* __SYS_DOT_H__ */
 

include/linux/gfs2_ondisk.h

@@ -22,6 +22,8 @@
 #define GFS2_LIVE_LOCK		1
 #define GFS2_TRANS_LOCK		2
 #define GFS2_RENAME_LOCK	3
+#define GFS2_CONTROL_LOCK	4
+#define GFS2_MOUNTED_LOCK	5
 
 /* Format numbers for various metadata types */