MDEV-7895 - HANDLER READ doesn't upgrade metadata lock from S to SRbb-mdev7895

Change code for HANDLER READ statements to upgrade S metadata lock to
SR metadata lock for the duration of read. This allows us properly
isolate HANDLER READ from LOCK TABLES WRITE and makes metadata locking
for these statements consistent with locking for other DML.

HANDLER-related tests had to be adjusted to take into account that
HANDLER READ will wait for and acquire SR lock.

3 files changed, 86 insertions, 59 deletions

diff --git a/sql/lock.cc b/sql/lock.cc
index c241e635e6b..ad0aedbee40 100644
--- a/sql/lock.cc
+++ b/sql/lock.cc
@@ -165,18 +165,12 @@ lock_tables_check(THD *thd, TABLE **tables, uint count, uint flags)
       write we must own metadata lock of MDL_SHARED_WRITE or stronger
       type. For table to be locked for read we must own metadata lock
       of MDL_SHARED_READ or stronger type).
-      The only exception are HANDLER statements which are allowed to
-      lock table for read while having only MDL_SHARED lock on it.
     */
     DBUG_ASSERT(t->s->tmp_table ||
                 thd->mdl_context.is_lock_owner(MDL_key::TABLE,
                                  t->s->db.str, t->s->table_name.str,
                                  t->reginfo.lock_type >= TL_WRITE_ALLOW_WRITE ?
-                                 MDL_SHARED_WRITE : MDL_SHARED_READ) ||
-                (t->open_by_handler &&
-                 thd->mdl_context.is_lock_owner(MDL_key::TABLE,
-                                  t->s->db.str, t->s->table_name.str,
-                                  MDL_SHARED)));
+                                 MDL_SHARED_WRITE : MDL_SHARED_READ));
 
     /*
       Prevent modifications to base tables if READ_ONLY is activated.
diff --git a/sql/sql_class.h b/sql/sql_class.h
index 73637f8d8eb..6b64f710176 100644
--- a/sql/sql_class.h
+++ b/sql/sql_class.h
@@ -1535,6 +1535,41 @@ private:
 
 
 /**
+  Internal error handler to process an error from MDL_context::upgrade_lock()
+  and mysql_lock_tables(). Used by implementations of HANDLER READ and
+  LOCK TABLES LOCAL.
+*/
+
+class MDL_deadlock_and_lock_abort_error_handler: public Internal_error_handler
+{
+public:
+  /**
+    Handle an error from MDL_context::upgrade_lock() and mysql_lock_tables().
+    Ignore ER_LOCK_ABORTED and ER_LOCK_DEADLOCK errors.
+  */
+  virtual
+  bool handle_condition(THD *thd,
+                        uint sql_errno,
+                        const char *sqlstate,
+                        Sql_condition::enum_warning_level level,
+                        const char* msg,
+                        Sql_condition **cond_hdl)
+  {
+    *cond_hdl= NULL;
+    if (sql_errno == ER_LOCK_ABORTED || sql_errno == ER_LOCK_DEADLOCK)
+      m_need_reopen= true;
+
+    return m_need_reopen;
+  }
+
+  bool need_reopen() const { return m_need_reopen; };
+  void init() { m_need_reopen= false; };
+private:
+  bool m_need_reopen;
+};
+
+
+/**
   Tables that were locked with LOCK TABLES statement.
 
   Encapsulates a list of TABLE_LIST instances for tables
diff --git a/sql/sql_handler.cc b/sql/sql_handler.cc
index 7dcc6fa0e95..e117aa593b5 100644
--- a/sql/sql_handler.cc
+++ b/sql/sql_handler.cc
@@ -488,56 +488,6 @@ bool mysql_ha_close(THD *thd, TABLE_LIST *tables)
 
 
 /**
-  A helper class to process an error from mysql_lock_tables().
-  HANDLER READ statement's attempt to lock the subject table
-  may get aborted if there is a pending DDL. In that case
-  we close the table, reopen it, and try to read again.
-  This is implicit and obscure, since HANDLER position
-  is lost in the process, but it's the legacy server
-  behaviour we should preserve.
-*/
-
-class Sql_handler_lock_error_handler: public Internal_error_handler
-{
-public:
-  virtual
-  bool handle_condition(THD *thd,
-                        uint sql_errno,
-                        const char *sqlstate,
-                        Sql_condition::enum_warning_level level,
-                        const char* msg,
-                        Sql_condition **cond_hdl);
-
-  bool need_reopen() const { return m_need_reopen; };
-  void init() { m_need_reopen= FALSE; };
-private:
-  bool m_need_reopen;
-};
-
-
-/**
-  Handle an error from mysql_lock_tables().
-  Ignore ER_LOCK_ABORTED errors.
-*/
-
-bool
-Sql_handler_lock_error_handler::
-handle_condition(THD *thd,
-                 uint sql_errno,
-                 const char *sqlstate,
-                 Sql_condition::enum_warning_level level,
-                 const char* msg,
-                 Sql_condition **cond_hdl)
-{
-  *cond_hdl= NULL;
-  if (sql_errno == ER_LOCK_ABORTED)
-    m_need_reopen= TRUE;
-
-  return m_need_reopen;
-}
-
-
-/**
    Finds an open HANDLER table.
 
    @params name		Name of handler to open
@@ -733,7 +683,8 @@ bool mysql_ha_read(THD *thd, TABLE_LIST *tables,
   int           error, keyno;
   uint          num_rows;
   uchar		*UNINIT_VAR(key);
-  Sql_handler_lock_error_handler sql_handler_lock_error;
+  MDL_deadlock_and_lock_abort_error_handler sql_handler_lock_error;
+  MDL_savepoint mdl_savepoint;
   DBUG_ENTER("mysql_ha_read");
   DBUG_PRINT("enter",("'%s'.'%s' as '%s'",
                       tables->db, tables->table_name, tables->alias));
@@ -752,6 +703,48 @@ retry:
   tables->table= table;                         // This is used by fix_fields
   table->pos_in_table_list= tables;
 
+  sql_handler_lock_error.init();
+
+  /*
+    For non-temporary tables we need to acquire SR lock in order to ensure
+    that HANDLER READ is blocked by LOCK TABLES WRITE in other connections
+    for storage engines which don't use THR_LOCK locks (e.g. InnoDB).
+
+    To simplify clean-up code we take MDL_savepoint even for temporary tables.
+  */
+  mdl_savepoint= thd->mdl_context.mdl_savepoint();
+
+  if (table->s->tmp_table == NO_TMP_TABLE)
+  {
+    MDL_request read_request;
+
+    read_request.init(&handler->mdl_request.key, MDL_SHARED_READ,
+                      MDL_TRANSACTION);
+
+    thd->push_internal_handler(&sql_handler_lock_error);
+
+    error= thd->mdl_context.acquire_lock(&read_request,
+                                         thd->variables.lock_wait_timeout);
+    thd->pop_internal_handler();
+
+    if (sql_handler_lock_error.need_reopen())
+    {
+      /*
+        HANDLER READ statement's attempt to upgrade lock on the subject table
+        may get aborted if there is a pending DDL. In that case we close the
+        table, reopen it, and try to read again.
+        This is implicit and obscure, since HANDLER position is lost in the
+        process, but it's the legacy server behaviour we should preserve.
+      */
+      DBUG_ASSERT(error && !thd->is_error());
+      mysql_ha_close_table(handler);
+      goto retry;
+    }
+
+    if (error)
+      goto err0;
+  }
+
   if (handler->lock->lock_count > 0)
   {
     int lock_error;
@@ -768,7 +761,7 @@ retry:
     */
     thd->set_open_tables(table);
 
-    sql_handler_lock_error.init();
+    /* Re-use Sql_handler_lock_error instance which was initialized earlier. */
     thd->push_internal_handler(&sql_handler_lock_error);
 
     lock_error= mysql_lock_tables(thd, handler->lock,
@@ -797,6 +790,7 @@ retry:
       */
       DBUG_ASSERT(! thd->transaction_rollback_request);
       trans_rollback_stmt(thd);
+      thd->mdl_context.rollback_to_savepoint(mdl_savepoint);
       mysql_ha_close_table(handler);
       if (thd->stmt_arena->is_stmt_execute())
       {
@@ -812,7 +806,10 @@ retry:
     }
 
     if (lock_error)
+    {
+      thd->mdl_context.rollback_to_savepoint(mdl_savepoint);
       goto err0; // mysql_lock_tables() printed error message already
+    }
   }
 
   if (mysql_ha_fix_cond_and_key(handler, mode, keyname, key_expr, cond, 0))
@@ -957,6 +954,7 @@ ok:
   */
   trans_commit_stmt(thd);
   mysql_unlock_tables(thd, handler->lock, 0);
+  thd->mdl_context.rollback_to_savepoint(mdl_savepoint);
   my_eof(thd);
   DBUG_PRINT("exit",("OK"));
   DBUG_RETURN(FALSE);