ha_innodb.cc:

Raise maximum column prefix len to 767 bytes, so that MySQL can create a column prefix index of 255 UTF-8 characters (each takes 3 bytes at the maximum); add comments about why innobase_get_at_most_n_mbchars() works ok
dict0mem.h:
  Raise maximum column prefix len to 767 bytes, so that MySQL can create a column prefix index of 255 UTF-8 characters (each takes 3 bytes at the maximum)
row0mysql.c:
  If MySQL tries to create a column prefix index longer that 255 UTF-8 characters, give an error, and drop the table from the InnoDB internal data dictionary. MySQL did not drop the table there in its own error handling.


innobase/row/row0mysql.c:
  If MySQL tries to create a column prefix index longer that 255 UTF-8 characters, give an error, and drop the table from the InnoDB internal data dictionary. MySQL did not drop the table there in its own error handling.
innobase/include/dict0mem.h:
  Raise maximum column prefix len to 767 bytes, so that MySQL can create a column prefix index of 255 UTF-8 characters (each takes 3 bytes at the maximum)
sql/ha_innodb.cc:
  Raise maximum column prefix len to 767 bytes, so that MySQL can create a column prefix index of 255 UTF-8 characters (each takes 3 bytes at the maximum); add comments about why innobase_get_at_most_n_mbchars() works ok

innobase/include/dict0mem.h

@@ -151,7 +151,12 @@ struct dict_col_struct{
 				in some of the functions below */
 };
 
-#define DICT_MAX_COL_PREFIX_LEN	512
+/* DICT_MAX_COL_PREFIX_LEN is measured in bytes. Starting from 4.1.6, we
+define max col prefix len as 3 * 256, so that one can create a column prefix
+index on 256 characters of a TEXT field also in the UTF-8 charset. In that
+charset, a character may take at most 3 bytes. */
+
+#define DICT_MAX_COL_PREFIX_LEN	768
 
 /* Data structure for a field in an index */
 struct dict_field_struct{
@@ -160,9 +165,13 @@ struct dict_field_struct{
 	ulint		order;		/* flags for ordering this field:
 					DICT_DESCEND, ... */
 	ulint		prefix_len;	/* 0 or the length of the column
-					prefix in a MySQL index of type, e.g.,
+					prefix in bytes in a MySQL index of
-					INDEX (textcol(25)); must be smaller
+					type, e.g., INDEX (textcol(25));
-					than DICT_MAX_COL_PREFIX_LEN */
+					must be smaller than
+					DICT_MAX_COL_PREFIX_LEN; NOTE that
+					in the UTF-8 charset, MySQL reserves
+					sets this to 3 * the prefix len in
+					UTF-8 chars */
 };
 
 /* Data structure for an index tree */

innobase/row/row0mysql.c

@@ -1630,6 +1630,8 @@ row_create_index_for_mysql(
 	
 	trx->op_info = "creating index";
 
+	trx_start_if_not_started(trx);
+
 	/* Check that the same column does not appear twice in the index.
 	Starting from 4.0.14, InnoDB should be able to cope with that, but
 	safer not to allow them. */
@@ -1656,9 +1658,16 @@ row_create_index_for_mysql(
 				goto error_handling;
 			}
 		}
-	}
+		
+		/* Check also that prefix_len < DICT_MAX_COL_PREFIX_LEN */
 
-	trx_start_if_not_started(trx);
+		if (dict_index_get_nth_field(index, i)->prefix_len
+						>= DICT_MAX_COL_PREFIX_LEN) {
+			err = DB_TOO_BIG_RECORD;
+
+			goto error_handling;
+		}
+	}
 
 	if (row_mysql_is_recovered_tmp_table(index->table_name)) {
 

sql/ha_innodb.cc

@@ -3525,10 +3525,6 @@ create_index(
 		        prefix_len = 0;
 		}
 
-		if (prefix_len >= DICT_MAX_COL_PREFIX_LEN) {
-			DBUG_RETURN(-1);
-		}
-
 		/* We assume all fields should be sorted in ascending
 		order, hence the '0': */
 
@@ -5333,39 +5329,32 @@ innobase_get_at_most_n_mbchars(
 	/* If the charset is multi-byte, then we must find the length of the
 	first at most n chars in the string. If the string contains less
 	characters than n, then we return the length to the end of the last
-	full character. */
+	character. */
 
 	if (charset->mbmaxlen > 1) {
-/*		ulint	right_value; */
-
 		/* my_charpos() returns the byte length of the first n_chars
-		characters, or the end of the last full character */
+		characters, or a value bigger than the length of str, if
+		there were not enough full characters in str.
+
+		Why does the code below work:
+		Suppose that we are looking for n UTF-8 characters.
+
+		1) If the string is long enough, then the prefix contains at
+		least n complete UTF-8 characters + maybe some extra
+		characters + an incomplete UTF-8 character. No problem in
+		this case. The function returns the pointer to the
+		end of the nth character.
+
+		2) If the string is not long enough, then the string contains
+		the complete value of a column, that is, only complete UTF-8
+		characters, and we can store in the column prefix index the
+		whole string. */
 
 		char_length = my_charpos(charset, str,
 						str + data_len, n_chars);
-	
-		/*################################################*/
-		/* TODO: my_charpos sometimes returns a non-sensical value
-		that is BIGGER than data_len: try to fix this bug partly with
-		these heuristics. This is NOT a complete bug fix! */
-
 		if (char_length > data_len) {
 			char_length = data_len;
 		}		
-		/*################################################*/
-
-/*		printf("data_len %lu, n_chars %lu, char_len %lu\n",
-				data_len, n_chars, char_length);
-		if (data_len < n_chars) {
-			right_value = data_len;
-		} else {
-			right_value = n_chars;
-		}
-
-		if (right_value != char_length) {
-			printf("ERRRRRROOORRRRRRRRRRRR!!!!!!!!!\n");
-		}
-*/
 	} else {
 		if (data_len < prefix_len) {
 			char_length = data_len;