BUG#35478: sort_union() returns bad data when sort_buffer_size is hit

- In QUICK_INDEX_MERGE_SELECT::read_keys_and_merge: when we got table->sort from Unique,
  tell init_read_record() not to use rr_from_cache() because a) rowids are already sorted
  and b) it might be that the the data is used by filesort(), which will need record rowids
  (which rr_from_cache() cannot provide).
- Fully de-initialize the table->sort read in QUICK_INDEX_MERGE_SELECT::get_next(). This fixes BUG#35477.
(bk trigger: file as fix for BUG#35478).

sql/filesort.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - make find_all_keys() use quick->get_next() instead of init_read_record(r)/r.read_record() calls
  - added dbug printout
sql/mysql_priv.h:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added parameter to init_read_record
sql/opt_range.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - In QUICK_INDEX_MERGE_SELECT::read_keys_and_merge: when we got table->sort from Unique,
    tell init_read_record() not to use rr_from_cache() because a) rowids are already sorted
    and b) it might be that the the data is used by filesort(), which will need record rowids
    (which rr_from_cache() cannot provide).
  - Fully de-initialize the table->sort read in QUICK_INDEX_MERGE_SELECT::get_next().
sql/records.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added disable_rr_cache parameter to init_read_record
  - Added comment
sql/sql_acl.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added parameter to init_read_record
sql/sql_delete.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added parameter to init_read_record
sql/sql_help.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added parameter to init_read_record
sql/sql_select.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added parameter to init_read_record
sql/sql_table.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added parameter to init_read_record
sql/sql_udf.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added parameter to init_read_record
sql/sql_update.cc:
  BUG#35478: sort_union() returns bad data when sort_buffer_size is hit
  - Added parameter to init_read_record

1 files changed, 10 insertions, 5 deletions

diff --git a/sql/opt_range.cc b/sql/opt_range.cc
index 17a4701b515..95cda371673 100644
--- a/sql/opt_range.cc
+++ b/sql/opt_range.cc
@@ -6494,7 +6494,7 @@ int QUICK_INDEX_MERGE_SELECT::read_keys_and_merge()
   QUICK_RANGE_SELECT* cur_quick;
   int result;
   Unique *unique;
-  DBUG_ENTER("QUICK_INDEX_MERGE_SELECT::prepare_unique");
+  DBUG_ENTER("QUICK_INDEX_MERGE_SELECT::read_keys_and_merge");
 
   /* We're going to just read rowids. */
   if (head->file->extra(HA_EXTRA_KEYREAD))
@@ -6565,13 +6565,17 @@ int QUICK_INDEX_MERGE_SELECT::read_keys_and_merge()
 
   }
 
-  /* ok, all row ids are in Unique */
+  /*
+    Ok all rowids are in the Unique now. The next call will initialize
+    head->sort structure so it can be used to iterate through the rowids
+    sequence.
+  */
   result= unique->get(head);
   delete unique;
   doing_pk_scan= FALSE;
-  /* start table scan */
-  init_read_record(&read_record, thd, head, (SQL_SELECT*) 0, 1, 1);
-  /* index_merge currently doesn't support "using index" at all */
+
+  /* Start the rnd_pos() scan. */
+  init_read_record(&read_record, thd, head, (SQL_SELECT*) 0, 1 , 1, TRUE);
   head->file->extra(HA_EXTRA_NO_KEYREAD);
 
   DBUG_RETURN(result);
@@ -6601,6 +6605,7 @@ int QUICK_INDEX_MERGE_SELECT::get_next()
   {
     result= HA_ERR_END_OF_FILE;
     end_read_record(&read_record);
+    free_io_cache(head);
     /* All rows from Unique have been retrieved, do a clustered PK scan */
     if (pk_quick_select)
     {