MDEV-25128 Wrong result from join with materialized semi-join and

           splittable derived

If one of joined tables of the processed query is a materialized derived
table (or view or CTE) with GROUP BY clause then under some conditions it
can be subject to split optimization. With this optimization new equalities
are injected into the WHERE condition of the SELECT that specifies this
derived table. The injected equalities are generated for all join orders
with which the split optimization can employed. After the best join order
has been chosen only certain of this equalities are really needed. The
others can be safely removed. If it's not done and some of injected
equalities involve expressions over semi-joins with look-up access then
the query may return a wrong result set.
This patch effectively removes equalities injected for split optimization
that are needed only at the optimization stage and not needed for execution.

Approved by serg@mariadb.com

1 files changed, 42 insertions, 6 deletions

diff --git a/sql/opt_split.cc b/sql/opt_split.cc
index fd7836f55cc..c3a2d03a93b 100644
--- a/sql/opt_split.cc
+++ b/sql/opt_split.cc
@@ -236,6 +236,8 @@ public:
   SplM_field_info *spl_fields;
   /* The number of elements in the above list */
   uint spl_field_cnt;
+  /* The list of equalities injected into WHERE for split optimization */
+  List<Item> inj_cond_list;
   /* Contains the structures to generate all KEYUSEs for pushable equalities */
   List<KEY_FIELD> added_key_fields;
   /* The cache of evaluated execution plans for 'join' with pushed equalities */
@@ -1047,22 +1049,22 @@ SplM_plan_info * JOIN_TAB::choose_best_splitting(double record_count,
 bool JOIN::inject_best_splitting_cond(table_map remaining_tables)
 {
   Item *inj_cond= 0;
-  List<Item> inj_cond_list;
+  List<Item> *inj_cond_list= &spl_opt_info->inj_cond_list;
   List_iterator<KEY_FIELD> li(spl_opt_info->added_key_fields);
   KEY_FIELD *added_key_field;
   while ((added_key_field= li++))
   {
     if (remaining_tables & added_key_field->val->used_tables())
       continue;
-    if (inj_cond_list.push_back(added_key_field->cond, thd->mem_root))
+    if (inj_cond_list->push_back(added_key_field->cond, thd->mem_root))
       return true;
   }
-  DBUG_ASSERT(inj_cond_list.elements);
-  switch (inj_cond_list.elements) {
+  DBUG_ASSERT(inj_cond_list->elements);
+  switch (inj_cond_list->elements) {
   case 1:
-    inj_cond= inj_cond_list.head(); break;
+    inj_cond= inj_cond_list->head(); break;
   default:
-    inj_cond= new (thd->mem_root) Item_cond_and(thd, inj_cond_list);
+    inj_cond= new (thd->mem_root) Item_cond_and(thd, *inj_cond_list);
     if (!inj_cond)
       return true;
   }
@@ -1082,6 +1084,40 @@ bool JOIN::inject_best_splitting_cond(table_map remaining_tables)
 
 /**
   @brief
+    Test if equality is injected for split optimization
+
+  @param
+    eq_item   equality to to test
+
+  @retval
+    true    eq_item is equality injected for split optimization
+    false   otherwise
+*/
+
+bool is_eq_cond_injected_for_split_opt(Item_func_eq *eq_item)
+{
+  Item *left_item= eq_item->arguments()[0]->real_item();
+  if (left_item->type() != Item::FIELD_ITEM)
+    return false;
+  Field *field= ((Item_field *) left_item)->field;
+  if (!field->table->reginfo.join_tab)
+    return false;
+  JOIN *join= field->table->reginfo.join_tab->join;
+  if (!join->spl_opt_info)
+    return false;
+  List_iterator_fast<Item> li(join->spl_opt_info->inj_cond_list);
+  Item *item;
+  while ((item= li++))
+  {
+    if (item == eq_item)
+        return true;
+  }
+  return false;
+}
+
+
+/**
+  @brief
     Fix the splitting chosen for a splittable table in the final query plan
 
   @param