Bug #26021 - valgrind reports error regarding handle_trailing_share and client thread share usage

- add ndb_share connect_count to decide if share can be reused to setup replication


sql/ha_ndbcluster.cc:
  add additional logging info
  remove inline real_free_share (not needed) and confuses valgrind printout
sql/ha_ndbcluster.h:
  add connect_count to ndb_share to enable checking for version of share
sql/ha_ndbcluster_binlog.cc:
  use ndb_share connect_count to decide if share can be reused to setup replication
  - share will be created early if table is accessed prior to handler setting up binlog stuff, see bug 26021
sql/ha_ndbcluster_binlog.h:
  remove inline real_free_share (not needed) and confuses valgrind printout

4 files changed, 34 insertions, 17 deletions

diff --git a/sql/ha_ndbcluster.cc b/sql/ha_ndbcluster.cc
index 21697be83aa..30515564a60 100644
--- a/sql/ha_ndbcluster.cc
+++ b/sql/ha_ndbcluster.cc
@@ -6837,7 +6837,7 @@ static int ndbcluster_end(handlerton *hton, ha_panic_function type)
       fprintf(stderr, "NDB: table share %s with use_count %d not freed\n",
               share->key, share->use_count);
 #endif
-      real_free_share(&share);
+      ndbcluster_real_free_share(&share);
     }
     pthread_mutex_unlock(&ndbcluster_mutex);
   }
@@ -7449,14 +7449,20 @@ int handle_trailing_share(NDB_SHARE *share)
   bzero((char*) &table_list,sizeof(table_list));
   table_list.db= share->db;
   table_list.alias= table_list.table_name= share->table_name;
+  safe_mutex_assert_owner(&LOCK_open);
   close_cached_tables(thd, 0, &table_list, TRUE);
 
   pthread_mutex_lock(&ndbcluster_mutex);
   if (!--share->use_count)
   {
-    DBUG_PRINT("info", ("NDB_SHARE: close_cashed_tables %s freed share.",
-               share->key)); 
-    real_free_share(&share);
+    if (ndb_extra_logging)
+      sql_print_information("NDB_SHARE: trailing share %s(connect_count: %u) "
+                            "released by close_cached_tables at "
+                            "connect_count: %u",
+                            share->key,
+                            share->connect_count,
+                            g_ndb_cluster_connection->get_connect_count());
+    ndbcluster_real_free_share(&share);
     DBUG_RETURN(0);
   }
 
@@ -7466,10 +7472,14 @@ int handle_trailing_share(NDB_SHARE *share)
   */
   if (share->state != NSS_DROPPED && !--share->use_count)
   {
-    DBUG_PRINT("info", ("NDB_SHARE: %s already exists, "
-                        "use_count=%d  state != NSS_DROPPED.",
-                        share->key, share->use_count)); 
-    real_free_share(&share);
+    if (ndb_extra_logging)
+      sql_print_information("NDB_SHARE: trailing share %s(connect_count: %u) "
+                            "released after NSS_DROPPED check "
+                            "at connect_count: %u",
+                            share->key,
+                            share->connect_count,
+                            g_ndb_cluster_connection->get_connect_count());
+    ndbcluster_real_free_share(&share);
     DBUG_RETURN(0);
   }
   DBUG_PRINT("error", ("NDB_SHARE: %s already exists  use_count=%d.",
@@ -7736,7 +7746,7 @@ void ndbcluster_free_share(NDB_SHARE **share, bool have_lock)
     (*share)->util_lock= 0;
   if (!--(*share)->use_count)
   {
-    real_free_share(share);
+    ndbcluster_real_free_share(share);
   }
   else
   {
diff --git a/sql/ha_ndbcluster.h b/sql/ha_ndbcluster.h
index 5b6900766b6..63665fde0f8 100644
--- a/sql/ha_ndbcluster.h
+++ b/sql/ha_ndbcluster.h
@@ -108,6 +108,7 @@ typedef struct st_ndbcluster_share {
   char *table_name;
   Ndb::TupleIdRange tuple_id_range;
 #ifdef HAVE_NDB_BINLOG
+  uint32 connect_count;
   uint32 flags;
   NdbEventOperation *op;
   NdbEventOperation *op_old; // for rename table
diff --git a/sql/ha_ndbcluster_binlog.cc b/sql/ha_ndbcluster_binlog.cc
index e48b92ca808..5a0089e59ac 100644
--- a/sql/ha_ndbcluster_binlog.cc
+++ b/sql/ha_ndbcluster_binlog.cc
@@ -362,6 +362,8 @@ void ndbcluster_binlog_init_share(NDB_SHARE *share, TABLE *_table)
   int do_event_op= ndb_binlog_running;
   DBUG_ENTER("ndbcluster_binlog_init_share");
 
+  share->connect_count= g_ndb_cluster_connection->get_connect_count();
+
   share->op= 0;
   share->table= 0;
 
@@ -605,7 +607,7 @@ static int ndbcluster_binlog_end(THD *thd)
                    ("table->s->db.table_name: %s.%s",
                     share->table->s->db.str, share->table->s->table_name.str));
       if (share->state != NSS_DROPPED && !--share->use_count)
-        real_free_share(&share);
+        ndbcluster_real_free_share(&share);
       else
       {
         DBUG_PRINT("share",
@@ -2443,11 +2445,20 @@ int ndbcluster_create_binlog_setup(Ndb *ndb, const char *key,
       pthread_mutex_unlock(&ndbcluster_mutex);
       DBUG_RETURN(1);
     }
-    handle_trailing_share(share);
+    if (share->connect_count != 
+        g_ndb_cluster_connection->get_connect_count())
+    {
+      handle_trailing_share(share);
+      share= NULL;
+    }
   }
 
   /* Create share which is needed to hold replication information */
-  if (!(share= get_share(key, 0, TRUE, TRUE)))
+  if (share)
+  {
+    ++share->use_count;
+  }
+  else if (!(share= get_share(key, 0, TRUE, TRUE)))
   {
     sql_print_error("NDB Binlog: "
                     "allocating table share for %s failed", key);
diff --git a/sql/ha_ndbcluster_binlog.h b/sql/ha_ndbcluster_binlog.h
index 44183c6de9d..00fc689f061 100644
--- a/sql/ha_ndbcluster_binlog.h
+++ b/sql/ha_ndbcluster_binlog.h
@@ -208,11 +208,6 @@ inline void free_share(NDB_SHARE **share, bool have_lock= FALSE)
   ndbcluster_free_share(share, have_lock);
 }
 
-inline void real_free_share(NDB_SHARE **share)
-{
-  ndbcluster_real_free_share(share);
-}
-
 inline
 Thd_ndb *
 get_thd_ndb(THD *thd) { return (Thd_ndb *) thd->ha_data[ndbcluster_hton->slot]; }