dm: bind new table before destroying old

When replacing a mapped device's table during a 'resume', delay the
destruction of the old table until the new one is successfully in place.

This will make it easier for a later patch to transfer internal state
information from the old table to the new one (something we do not currently
support) while giving us more options for reversion if a later part
of the operation fails.

Devices are always in the suspended state during dm_swap_table().
This patch reinforces the requirement that all I/O must have been
flushed from the table targets while in this state (including any in
workqueues).  In the case of 'noflush' suspending, unprocessed
I/O should have been 'pushed back' to the dm core prior to this point,
for resubmission after the new table is in place.

Signed-off-by: Alasdair G Kergon <agk@redhat.com>

2 files changed, 14 insertions, 5 deletions

diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
index 1a6cb3c7822e..3162b9c3c3f2 100644
--- a/drivers/md/dm-table.c
+++ b/drivers/md/dm-table.c
@@ -237,6 +237,9 @@ void dm_table_destroy(struct dm_table *t)
 {
         unsigned int i;
 
+        if (!t)
+                return;
+
         while (atomic_read(&t->holders))
                 msleep(1);
         smp_mb();
diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 16f759fe04fe..255b75c61544 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -2077,19 +2077,23 @@ static int __bind(struct mapped_device *md, struct dm_table *t,
         return 0;
 }
 
-static void __unbind(struct mapped_device *md)
+/*
+ * Returns unbound table for the caller to free.
+ */
+static struct dm_table *__unbind(struct mapped_device *md)
 {
         struct dm_table *map = md->map;
         unsigned long flags;
 
         if (!map)
-                return;
+                return NULL;
 
         dm_table_event_callback(map, NULL, NULL);
         write_lock_irqsave(&md->map_lock, flags);
         md->map = NULL;
         write_unlock_irqrestore(&md->map_lock, flags);
-        dm_table_destroy(map);
+
+        return map;
 }
 
 /*
@@ -2182,7 +2186,7 @@ void dm_put(struct mapped_device *md)
                 }
                 dm_sysfs_exit(md);
                 dm_table_put(map);
-                __unbind(md);
+                dm_table_destroy(__unbind(md));
                 free_dev(md);
         }
 }
@@ -2368,6 +2372,7 @@ static void dm_rq_barrier_work(struct work_struct *work)
  */
 int dm_swap_table(struct mapped_device *md, struct dm_table *table)
 {
+        struct dm_table *map;
         struct queue_limits limits;
         int r = -EINVAL;
 
@@ -2388,8 +2393,9 @@ int dm_swap_table(struct mapped_device *md, struct dm_table *table)
                 goto out;
         }
 
-        __unbind(md);
+        map = __unbind(md);
         r = __bind(md, table, &limits);
+        dm_table_destroy(map);
 
 out:
         mutex_unlock(&md->suspend_lock);