From bb81de5b554aa727f571fb5f74e42f2ab26cfb96 Mon Sep 17 00:00:00 2001
From: Igor Fedotov <ifedotov@suse.com>
Date: Fri, 17 Apr 2020 19:26:29 +0300
Subject: [PATCH] os/bluestore: invoke _prepare_ondisk_format_super as the last
 op

Fixes: https://tracker.ceph.com/issues/45133

Signed-off-by: Igor Fedotov <ifedotov@suse.com>
---
 src/os/bluestore/BlueStore.cc | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)

diff --git a/src/os/bluestore/BlueStore.cc b/src/os/bluestore/BlueStore.cc
index de18468d474526..6f6273d3a873a3 100644
--- a/src/os/bluestore/BlueStore.cc
+++ b/src/os/bluestore/BlueStore.cc
@@ -11084,6 +11084,9 @@ int BlueStore::_upgrade_super()
 	t->rmkey(PREFIX_SUPER, "min_min_alloc_size");
       }
       ondisk_format = 2;
+
+      // make sure upgrade is performed within a single transaction
+      // hence preparing new super record here other than at the end
       _prepare_ondisk_format_super(t);
       int r = db->submit_transaction_sync(t);
       ceph_assert(r == 0);
@@ -11096,19 +11099,23 @@ int BlueStore::_upgrade_super()
       // - super: added per_pool_omap key, which indicates that *all* objects
       //   are using the new prefix and key format
       ondisk_format = 3;
-      KeyValueDB::Transaction t = db->get_transaction();
-      _prepare_ondisk_format_super(t);
-      int r = db->submit_transaction_sync(t);
-      ceph_assert(r == 0);
     }
     if (ondisk_format == 3) {
       // changes:
       // - FreelistManager keeps meta within bdev label
       int r = _write_out_fm_meta(0);
       ceph_assert(r == 0);
-
       ondisk_format = 4;
     }
+    // This to be the last operation
+    // Upgrade to format 2 is completed at this point,
+    // perform additional upgrade is needed
+    if (ondisk_format > 2) {
+      KeyValueDB::Transaction t = db->get_transaction();
+      _prepare_ondisk_format_super(t);
+      int r = db->submit_transaction_sync(t);
+      ceph_assert(r == 0);
+    }
   }
   // done
   dout(1) << __func__ << " done" << dendl;