[dm-devel] [PATCH] dm thin: commit metadata before creating metadata snapshot

[Mike Snitzer]

From: Joe Thornber <ejt at redhat.com>

If metadata is changing rapidly userland sometimes gets a corrupt
metadata block.

The checksums mean userland spots this corruption immediately so there's
no risk of acting on incorrect data.  No corruption exists from the
kernel's point of view, and thin_check passes after pool shutdown.

I believe this is to do with shared blocks at the first level of the
{device, mapping} btree.  Prior to the metadata-snap support no sharing
at this level was possible.  The problem goes away if we commit before
we take the metadata snapshot (which is a sensible thing to do anyway).

Signed-off-by: Joe Thornber <ejt at redhat.com>
Signed-off-by: Mike Snitzer <snitzer at redhat.com>
---
 drivers/md/dm-thin.c |    7 +++++++
 1 files changed, 7 insertions(+), 0 deletions(-)

diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 54ae5c3..7dd25ff 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -2296,6 +2296,13 @@ static int process_reserve_metadata_snap_mesg(unsigned argc, char **argv, struct
 	if (r)
 		return r;
 
+	r = dm_pool_commit_metadata(pool->pmd);
+	if (r) {
+		DMERR("%s: dm_pool_commit_metadata() failed, error = %d",
+		      __func__, r);
+		return r;
+	}
+
 	r = dm_pool_reserve_metadata_snap(pool->pmd);
 	if (r)
 		DMWARN("reserve_metadata_snap message failed.");
-- 
1.7.4.4