BUG#920255: Wrong result (extra rows) with loosescan and IN subquery

The problem was that LooseScan execution code assumed that tab->key holds 
the index used for looseScan. This is only true when range or full index
scan are used. In case of ref access, the index is in tab->ref.key (and 
tab->index==0 which explains how LooseScan passed tests with ref access: they 
used one index)

Fixed by setting/using loosescan_key, which always the correct index#.

1 files changed, 6 insertions, 0 deletions

diff --git a/sql/sql_select.h b/sql/sql_select.h
index 8b448130eaf..185bf90ea17 100644
--- a/sql/sql_select.h
+++ b/sql/sql_select.h
@@ -379,6 +379,12 @@ typedef struct st_join_table {
   /* Buffer to save index tuple to be able to skip duplicates */
   uchar *loosescan_buf;
   
+  /* 
+    Index used by LooseScan (we store it here separately because ref access
+    stores it in tab->ref.key, while range scan stores it in tab->index, etc)
+  */
+  uint loosescan_key;
+
   /* Length of key tuple (depends on #keyparts used) to store in the above */
   uint loosescan_key_len;