Merge branch 10.2 into 10.3

Author: abarkov

mysql-test/main/ctype_utf32.result

@@ -2890,5 +2890,28 @@ HEX(c1)
 0000006100000063
 DROP TABLE t1;
 #
+# MDEV-23210 Assertion `(length % 4) == 0' failed in my_lengthsp_utf32 on ALTER TABLE, SELECT and INSERT
+#
+CREATE TABLE t1 (a CHAR(1));
+SET COLLATION_CONNECTION=utf32_general_ci, CHARACTER_SET_CLIENT=binary;
+ALTER TABLE t1 CHANGE a a ENUM('a','a') CHARACTER SET utf32;
+ERROR HY000: Column 'a' has duplicated value 'a' in ENUM
+ALTER TABLE t1 CHANGE a a ENUM('aaa') CHARACTER SET utf32;
+ERROR HY000: Invalid utf32 character string: '\x00aaa'
+ALTER TABLE t1 CHANGE a a ENUM('aa') CHARACTER SET utf32;
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` enum('慡') CHARACTER SET utf32 DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+ALTER TABLE t1 CHANGE a a ENUM('a','b') CHARACTER SET utf32;
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` enum('a','b') CHARACTER SET utf32 DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+DROP TABLE t1;
+SET NAMES utf8;
+#
 # End of 10.2 tests
 #

mysql-test/main/ctype_utf32.test

@@ -1048,6 +1048,25 @@ INSERT INTO t1 (c1) VALUES (1),(2),(3);
 SELECT HEX(c1) FROM t1 ORDER BY c1;
 DROP TABLE t1;
 
+
+--echo #
+--echo # MDEV-23210 Assertion `(length % 4) == 0' failed in my_lengthsp_utf32 on ALTER TABLE, SELECT and INSERT
+--echo #
+
+CREATE TABLE t1 (a CHAR(1));
+SET COLLATION_CONNECTION=utf32_general_ci, CHARACTER_SET_CLIENT=binary;
+--error ER_DUPLICATED_VALUE_IN_TYPE
+ALTER TABLE t1 CHANGE a a ENUM('a','a') CHARACTER SET utf32;
+--error ER_INVALID_CHARACTER_STRING
+ALTER TABLE t1 CHANGE a a ENUM('aaa') CHARACTER SET utf32;
+ALTER TABLE t1 CHANGE a a ENUM('aa') CHARACTER SET utf32;
+SHOW CREATE TABLE t1;
+ALTER TABLE t1 CHANGE a a ENUM('a','b') CHARACTER SET utf32;
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+SET NAMES utf8;
+
+
 --echo #
 --echo # End of 10.2 tests
 --echo #

mysql-test/main/ctype_utf32_uca.result

@@ -7941,5 +7941,20 @@ EXECUTE s;
 DEALLOCATE PREPARE s;
 SET NAMES utf8;
 #
+# MDEV-23210 Assertion `(length % 4) == 0' failed in my_lengthsp_utf32 on ALTER TABLE, SELECT and INSERT
+#
+CREATE TABLE t1 (a CHAR(1));
+SET COLLATION_CONNECTION=utf32_myanmar_ci, CHARACTER_SET_CLIENT=binary;
+ALTER TABLE t1 CHANGE a a ENUM('a','a') CHARACTER SET utf32;
+ERROR HY000: Column 'a' has duplicated value 'a' in ENUM
+ALTER TABLE t1 CHANGE a a ENUM('a','b') CHARACTER SET utf32;
+SHOW CREATE TABLE t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `a` enum('a','b') CHARACTER SET utf32 DEFAULT NULL
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+DROP TABLE t1;
+SET NAMES utf8;
+#
 # End of 10.2 tests
 #

mysql-test/main/ctype_utf32_uca.test

@@ -290,6 +290,19 @@ EXECUTE s;
 DEALLOCATE PREPARE s;
 SET NAMES utf8;
 
+--echo #
+--echo # MDEV-23210 Assertion `(length % 4) == 0' failed in my_lengthsp_utf32 on ALTER TABLE, SELECT and INSERT
+--echo #
+
+CREATE TABLE t1 (a CHAR(1));
+SET COLLATION_CONNECTION=utf32_myanmar_ci, CHARACTER_SET_CLIENT=binary;
+--error ER_DUPLICATED_VALUE_IN_TYPE
+ALTER TABLE t1 CHANGE a a ENUM('a','a') CHARACTER SET utf32;
+ALTER TABLE t1 CHANGE a a ENUM('a','b') CHARACTER SET utf32;
+SHOW CREATE TABLE t1;
+DROP TABLE t1;
+SET NAMES utf8;
+
 
 --echo #
 --echo # End of 10.2 tests

mysql-test/suite/gcol/inc/gcol_supported_sql_funcs_main.inc

@@ -12,6 +12,9 @@
 # Change Date:                                                                 #
 # Change:                                                                      #
 ################################################################################
+
+--source include/have_des.inc
+
 set time_zone="+03:00";
 --echo #
 --echo # NUMERIC FUNCTIONS

sql/sql_class.cc

@@ -2306,6 +2306,58 @@ bool THD::convert_string(LEX_STRING *to, CHARSET_INFO *to_cs,
 }
 
 
+/*
+  Reinterpret a binary string to a character string
+
+  @param[OUT] to    The result will be written here,
+                    either the original string as is,
+                    or a newly alloced fixed string with
+                    some zero bytes prepended.
+  @param cs         The destination character set
+  @param str        The binary string
+  @param length     The length of the binary string
+
+  @return           false on success
+  @return           true on error
+*/
+
+bool THD::reinterpret_string_from_binary(LEX_CSTRING *to, CHARSET_INFO *cs,
+                                         const char *str, size_t length)
+{
+  /*
+    When reinterpreting from binary to tricky character sets like
+    UCS2, UTF16, UTF32, we may need to prepend some zero bytes.
+    This is possible in scenarios like this:
+      SET COLLATION_CONNECTION=utf32_general_ci, CHARACTER_SET_CLIENT=binary;
+    This code is similar to String::copy_aligned().
+  */
+  size_t incomplete= length % cs->mbminlen; // Bytes in an incomplete character
+  if (incomplete)
+  {
+    size_t zeros= cs->mbminlen - incomplete;
+    size_t aligned_length= zeros + length;
+    char *dst= (char*) alloc(aligned_length + 1);
+    if (!dst)
+    {
+      to->str= NULL; // Safety
+      to->length= 0;
+      return true;
+    }
+    bzero(dst, zeros);
+    memcpy(dst + zeros, str, length);
+    dst[aligned_length]= '\0';
+    to->str= dst;
+    to->length= aligned_length;
+  }
+  else
+  {
+    to->str= str;
+    to->length= length;
+  }
+  return check_string_for_wellformedness(to->str, to->length, cs);
+}
+
+
 /*
   Convert a string between two character sets.
   dstcs and srccs cannot be &my_charset_bin.

sql/sql_class.h

@@ -3717,6 +3717,8 @@ class THD :public Statement,
   bool convert_string(LEX_STRING *to, CHARSET_INFO *to_cs,
 		      const char *from, size_t from_length,
 		      CHARSET_INFO *from_cs);
+  bool reinterpret_string_from_binary(LEX_CSTRING *to, CHARSET_INFO *to_cs,
+                                      const char *from, size_t from_length);
   bool convert_string(LEX_CSTRING *to, CHARSET_INFO *to_cs,
                       const char *from, size_t from_length,
                       CHARSET_INFO *from_cs)
@@ -3733,6 +3735,8 @@ class THD :public Statement,
   {
     if (!simple_copy_is_possible)
       return unlikely(convert_string(to, tocs, from->str, from->length, fromcs));
+    if (fromcs == &my_charset_bin)
+      return reinterpret_string_from_binary(to, tocs, from->str, from->length);
     *to= *from;
     return false;
   }