ha_innodb.cc:

  Fix bug: MySQL could erroneously return Empty set if InnoDB estimated index range size to 0 records though the range was not empty; MySQL also failed to do the next-key locking in the case of an empty index range


sql/ha_innodb.cc:
  Fix bug: MySQL could erroneously return Empty set if InnoDB estimated index range size to 0 records though the range was not empty; MySQL also failed to do the next-key locking in the case of an empty index range

1 files changed, 10 insertions, 0 deletions

diff --git a/sql/ha_innodb.cc b/sql/ha_innodb.cc
index 0af1fb7b50a..ea8eaf6653e 100644
--- a/sql/ha_innodb.cc
+++ b/sql/ha_innodb.cc
@@ -3486,6 +3486,16 @@ ha_innobase::records_in_range(
 
 	prebuilt->trx->op_info = (char*)"";
 
+	/* The MySQL optimizer seems to believe an estimate of 0 rows is
+	always accurate and may return the result 'Empty set' based on that.
+	The accuracy is not guaranteed, and even if it were, for a locking
+	read we should anyway perform the search to set the next-key lock.
+	Add 1 to the value to make sure MySQL does not make the assumption! */
+
+	if (n_rows == 0) {
+	        n_rows = 1;
+	}
+
 	DBUG_RETURN((ha_rows) n_rows);
 }