Bug#32705 - myisam corruption: Key in wrong position

            at page 1024 with ucs2_bin

Inserting strings with a common prefix into a table with
characterset UCS2 corrupted the table.

An efficient search method was used, which compares end space
with ASCII blank. This doesn't work for character sets like UCS2,
which do not encode blank like ASCII does.

Use the less efficient search method _mi_seq_search()
for charsets with mbminlen > 1.


myisam/mi_open.c:
  Bug#32705 - myisam corruption: Key in wrong position
              at page 1024 with ucs2_bin
  Use _mi_seq_search() for charsets with mbminlen > 1.
mysql-test/r/myisam.result:
  Bug#32705 - myisam corruption: Key in wrong position
              at page 1024 with ucs2_bin
  Added test result.
mysql-test/t/myisam.test:
  Bug#32705 - myisam corruption: Key in wrong position
              at page 1024 with ucs2_bin
  Added test.

1 files changed, 10 insertions, 1 deletions

diff --git a/myisam/mi_open.c b/myisam/mi_open.c
index ec169ac8785..5314d6a9a6c 100644
--- a/myisam/mi_open.c
+++ b/myisam/mi_open.c
@@ -791,8 +791,17 @@ static void setup_key_functions(register MI_KEYDEF *keyinfo)
     keyinfo->get_key= _mi_get_pack_key;
     if (keyinfo->seg[0].flag & HA_PACK_KEY)
     {						/* Prefix compression */
+      /*
+        _mi_prefix_search() compares end-space against ASCII blank (' ').
+        It cannot be used for character sets, that do not encode the
+        blank character like ASCII does. UCS2 is an example. All
+        character sets with a fixed width > 1 or a mimimum width > 1
+        cannot represent blank like ASCII does. In these cases we have
+        to use _mi_seq_search() for the search.
+      */
       if (!keyinfo->seg->charset || use_strnxfrm(keyinfo->seg->charset) ||
-          (keyinfo->seg->flag & HA_NULL_PART))
+          (keyinfo->seg->flag & HA_NULL_PART) ||
+          (keyinfo->seg->charset->mbminlen > 1))
         keyinfo->bin_search=_mi_seq_search;
       else
         keyinfo->bin_search=_mi_prefix_search;