BUG#50351 - ft_min_word_len=2 Causes query to hang

Performing fulltext prefix search (a word with truncation
operator) may cause a dead-loop. ft_min_word_len value
doesn't matter actually.

The problem was introduced along with "smarter index merge"
optimization.

mysql-test/r/fulltext.result:
  A test case for BUG#50351.
mysql-test/t/fulltext.test:
  A test case for BUG#50351.
storage/myisam/ft_boolean_search.c:
  When going up to first-level tree, we need to restore docid[0],
  so it informs fulltext index merge not to enter this second-level
  tree again (avoiding dead-loop).

1 files changed, 11 insertions, 1 deletions

diff --git a/storage/myisam/ft_boolean_search.c b/storage/myisam/ft_boolean_search.c
index 492261c5efc..52ad6b11aa1 100644
--- a/storage/myisam/ft_boolean_search.c
+++ b/storage/myisam/ft_boolean_search.c
@@ -437,8 +437,18 @@ static int _ft2_search(FTB *ftb, FTB_WORD *ftbw, my_bool init_search)
         return 0;
     }
 
-    /* going up to the first-level tree to continue search there */
+    /*
+      Going up to the first-level tree to continue search there.
+      Only done when performing prefix search.
+
+      Key buffer data pointer as well as docid[0] may be smaller
+      than values we got while searching first-level tree. Thus
+      they must be restored to original values to avoid dead-loop,
+      when subsequent search for a bigger value eventually ends up
+      in this same second-level tree.
+    */
     _mi_dpointer(info, (uchar*) (lastkey_buf+HA_FT_WLEN), ftbw->key_root);
+    ftbw->docid[0]= ftbw->key_root;
     ftbw->key_root=info->s->state.key_root[ftb->keynr];
     ftbw->keyinfo=info->s->keyinfo+ftb->keynr;
     ftbw->off=0;