From: Joe Thornber <ejt@redhat.com>

Generate a dm event when the amount of remaining thin pool metadata
space falls below a certain level.

The threshold is taken to be a quarter of the size of the metadata
device with a minimum threshold of 4MB.

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

---
 drivers/md/dm-thin-metadata.c |   14 ++++++++++++++
 drivers/md/dm-thin-metadata.h |    6 ++++++
 drivers/md/dm-thin.c          |   38 ++++++++++++++++++++++++++++++++++++++
 3 files changed, 58 insertions(+)

Index: linux/drivers/md/dm-thin-metadata.c
===================================================================
--- linux.orig/drivers/md/dm-thin-metadata.c
+++ linux/drivers/md/dm-thin-metadata.c
@@ -1696,3 +1696,17 @@ void dm_pool_metadata_read_only(struct d
 	dm_bm_set_read_only(pmd->bm);
 	up_write(&pmd->root_lock);
 }
+
+int dm_pool_register_metadata_threshold(struct dm_pool_metadata *pmd,
+					dm_block_t threshold,
+					dm_sm_threshold_fn fn,
+					void *context)
+{
+	int r;
+
+	down_write(&pmd->root_lock);
+	r = dm_sm_register_threshold_callback(pmd->metadata_sm, threshold, fn, context);
+	up_write(&pmd->root_lock);
+
+	return r;
+}
Index: linux/drivers/md/dm-thin-metadata.h
===================================================================
--- linux.orig/drivers/md/dm-thin-metadata.h
+++ linux/drivers/md/dm-thin-metadata.h
@@ -8,6 +8,7 @@
 #define DM_THIN_METADATA_H
 
 #include "persistent-data/dm-block-manager.h"
+#include "persistent-data/dm-space-map.h"
 
 #define THIN_METADATA_BLOCK_SIZE 4096
 
@@ -193,6 +194,11 @@ int dm_pool_resize_metadata_dev(struct d
  */
 void dm_pool_metadata_read_only(struct dm_pool_metadata *pmd);
 
+int dm_pool_register_metadata_threshold(struct dm_pool_metadata *pmd,
+					dm_block_t threshold,
+					dm_sm_threshold_fn fn,
+					void *context);
+
 /*----------------------------------------------------------------*/
 
 #endif
Index: linux/drivers/md/dm-thin.c
===================================================================
--- linux.orig/drivers/md/dm-thin.c
+++ linux/drivers/md/dm-thin.c
@@ -1281,6 +1281,10 @@ static void process_bio_fail(struct thin
 	bio_io_error(bio);
 }
 
+/*
+ * FIXME: should we also commit due to size of transaction, measured in
+ * metadata blocks?
+ */
 static int need_commit_due_to_time(struct pool *pool)
 {
 	return jiffies < pool->last_commit_jiffies ||
@@ -1909,6 +1913,16 @@ static int parse_pool_features(struct dm
 	return r;
 }
 
+static void metadata_low_callback(void *context)
+{
+	struct pool *pool = context;
+
+	DMWARN("%s: reached low water mark for metadata device: sending event.",
+	       dm_device_name(pool->pool_md));
+
+	dm_table_event(pool->ti->table);
+}
+
 static sector_t get_metadata_dev_size(struct block_device *bdev)
 {
 	sector_t metadata_dev_size = i_size_read(bdev->bd_inode) >> SECTOR_SHIFT;
@@ -1933,6 +1947,23 @@ static dm_block_t get_metadata_dev_size_
 }
 
 /*
+ * When a metadata threshold is crossed a dm event is triggered, and
+ * userland should respond by growing the metadata device.  We could let
+ * userland set the threshold, like we do with the data threshold, but I'm
+ * not sure they know enough to do this well.
+ */
+static dm_block_t calc_metadata_threshold(struct pool_c *pt)
+{
+	/*
+	 * 4M is ample for all ops with the possible exception of thin
+	 * device deletion which is harmless if it fails (just retry the
+	 * delete after you've grown the device).
+	 */
+	dm_block_t quarter = get_metadata_dev_size_in_blocks(pt->metadata_dev->bdev) / 4;
+	return min((dm_block_t)1024ULL /* 4M */, quarter);
+}
+
+/*
  * thin-pool <metadata dev> <data dev>
  *	     <data block size (sectors)>
  *	     <low water mark (blocks)>
@@ -2065,6 +2096,13 @@ static int pool_ctr(struct dm_target *ti
 	}
 	ti->private = pt;
 
+	r = dm_pool_register_metadata_threshold(pt->pool->pmd,
+						calc_metadata_threshold(pt),
+						metadata_low_callback,
+						pool);
+	if (r)
+		goto out_free_pt;
+
 	pt->callbacks.congested_fn = pool_is_congested;
 	dm_table_add_target_callbacks(ti->table, &pt->callbacks);