md: manage redundancy group in sysfs when changing level.

Some levels expect the 'redundancy group' to be present,
others don't.
So when we change level of an array we might need to
add or remove this group.

This requires fixing up the current practice of overloading ->private
to indicate (when ->pers == NULL) that something needs to be removed.
So create a new ->to_remove to fill that role.

When changing levels, we may need to add or remove attributes.  When
changing RAID5 -> RAID6, we both add and remove the same thing.  It is
important to catch this and optimise it out as the removal is delayed
until a lock is released, so trying to add immediately would cause
problems.


Cc: stable@kernel.org
Signed-off-by: NeilBrown <neilb@suse.de>

3 files changed, 38 insertions, 13 deletions

diff --git a/drivers/md/md.c b/drivers/md/md.c
index edf777f..e8d2388 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -509,9 +509,9 @@ static inline int mddev_trylock(mddev_t * mddev)
 
 static struct attribute_group md_redundancy_group;
 
-static inline void mddev_unlock(mddev_t * mddev)
+static void mddev_unlock(mddev_t * mddev)
 {
-	if (mddev->pers == NULL && mddev->private) {
+	if (mddev->to_remove) {
 		/* These cannot be removed under reconfig_mutex as
 		 * an access to the files will try to take reconfig_mutex
 		 * while holding the file unremovable, which leads to
@@ -520,16 +520,20 @@ static inline void mddev_unlock(mddev_t * mddev)
 		 * it while holding reconfig_mutex, and md_run can
 		 * use it to wait for the remove to complete.
 		 */
+		struct attribute_group *to_remove = mddev->to_remove;
+		mddev->to_remove = NULL;
 		mutex_lock(&mddev->open_mutex);
 		mutex_unlock(&mddev->reconfig_mutex);
 
-		sysfs_remove_group(&mddev->kobj, &md_redundancy_group);
-		if (mddev->private != (void*)1)
-			sysfs_remove_group(&mddev->kobj, mddev->private);
-		if (mddev->sysfs_action)
-			sysfs_put(mddev->sysfs_action);
-		mddev->sysfs_action = NULL;
-		mddev->private = NULL;
+		if (to_remove != &md_redundancy_group)
+			sysfs_remove_group(&mddev->kobj, to_remove);
+		if (mddev->pers == NULL ||
+		    mddev->pers->sync_request == NULL) {
+			sysfs_remove_group(&mddev->kobj, &md_redundancy_group);
+			if (mddev->sysfs_action)
+				sysfs_put(mddev->sysfs_action);
+			mddev->sysfs_action = NULL;
+		}
 		mutex_unlock(&mddev->open_mutex);
 	} else
 		mutex_unlock(&mddev->reconfig_mutex);
@@ -2996,6 +3000,23 @@ level_store(mddev_t *mddev, const char *buf, size_t len)
 	/* Looks like we have a winner */
 	mddev_suspend(mddev);
 	mddev->pers->stop(mddev);
+	
+	if (mddev->pers->sync_request == NULL &&
+	    pers->sync_request != NULL) {
+		/* need to add the md_redundancy_group */
+		if (sysfs_create_group(&mddev->kobj, &md_redundancy_group))
+			printk(KERN_WARNING
+			       "md: cannot register extra attributes for %s\n",
+			       mdname(mddev));
+		mddev->sysfs_action = sysfs_get_dirent(mddev->kobj.sd, "sync_action");
+	}		
+	if (mddev->pers->sync_request != NULL &&
+	    pers->sync_request == NULL) {
+		/* need to remove the md_redundancy_group */
+		if (mddev->to_remove == NULL)
+			mddev->to_remove = &md_redundancy_group;
+	}
+
 	module_put(mddev->pers->owner);
 	/* Invalidate devices that are now superfluous */
 	list_for_each_entry(rdev, &mddev->disks, same_set)
@@ -4550,8 +4571,8 @@ static int do_md_stop(mddev_t * mddev, int mode, int is_open)
 			mddev->queue->unplug_fn = NULL;
 			mddev->queue->backing_dev_info.congested_fn = NULL;
 			module_put(mddev->pers->owner);
-			if (mddev->pers->sync_request && mddev->private == NULL)
-				mddev->private = (void*)1;
+			if (mddev->pers->sync_request && mddev->to_remove == NULL)
+				mddev->to_remove = &md_redundancy_group;
 			mddev->pers = NULL;
 			/* tell userspace to handle 'inactive' */
 			sysfs_notify_dirent(mddev->sysfs_state);
diff --git a/drivers/md/md.h b/drivers/md/md.h
index 8e4c75c..722f5df 100644
--- a/drivers/md/md.h
+++ b/drivers/md/md.h
@@ -305,6 +305,7 @@ struct mddev_s
 	atomic_t 			max_corr_read_errors; /* max read retries */
 	struct list_head		all_mddevs;
 
+	struct attribute_group		*to_remove;
 	/* Generic barrier handling.
 	 * If there is a pending barrier request, all other
 	 * writes are blocked while the devices are flushed.
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 70ffbd0..a361398 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -5090,7 +5090,9 @@ static int run(mddev_t *mddev)
 	}
 
 	/* Ok, everything is just fine now */
-	if (sysfs_create_group(&mddev->kobj, &raid5_attrs_group))
+	if (mddev->to_remove == &raid5_attrs_group)
+		mddev->to_remove = NULL;
+	else if (sysfs_create_group(&mddev->kobj, &raid5_attrs_group))
 		printk(KERN_WARNING
 		       "raid5: failed to create sysfs attributes for %s\n",
 		       mdname(mddev));
@@ -5137,7 +5139,8 @@ static int stop(mddev_t *mddev)
 	mddev->queue->backing_dev_info.congested_fn = NULL;
 	blk_sync_queue(mddev->queue); /* the unplug fn references 'conf'*/
 	free_conf(conf);
-	mddev->private = &raid5_attrs_group;
+	mddev->private = NULL;
+	mddev->to_remove = &raid5_attrs_group;
 	return 0;
 }