InnoDB: Make CREATE TABLE return error when the minimum row length

exceeds the maximum record size.  (Bug #5682)


innobase/data/data0type.c:
  Remove function dtype_str_needs_mysql_cmp().
  Document dtype_get_at_most_n_mbchars().
  dtype_get_at_most_n_mbchars(): Use mbminlen and mbmaxlen.
innobase/dict/dict0crea.c:
  dict_build_table_def_step(): Reject if minimum row size is too big.
innobase/include/data0type.h:
  Remove dtype_str_needs_mysql_cmp().
  Document dtype_get_at_most_n_mbchars().
  Add dtype_get_min_size().
innobase/include/data0type.ic:
  Add dtype_get_min_size().
innobase/include/row0mysql.h:
  row_mysql_store_col_in_innobase_format(): Add parameter comp,
  as we will only truncate UTF-8 CHAR(n) columns in row_format=compact.
innobase/include/row0mysql.ic:
  row_mysql_store_col_in_innobase_format(): Add parameter comp,
  as we will only truncate UTF-8 CHAR(n) columns in row_format=compact.
innobase/row/row0mysql.c:
  Pass parameter comp to row_mysql_store_col_in_innobase_format().
innobase/row/row0sel.c:
  Pass parameter comp to row_mysql_store_col_in_innobase_format().
  row_sel_field_store_in_mysql_format(): Undo the stripping of
  UTF-8 CHAR(n) columns by padding with spaces.

4 files changed, 111 insertions, 25 deletions

diff --git a/innobase/include/data0type.h b/innobase/include/data0type.h
index 4c327d0b7ed..174665ca1fa 100644
--- a/innobase/include/data0type.h
+++ b/innobase/include/data0type.h
@@ -145,28 +145,22 @@ store the charset-collation number; one byte is left unused, though */
 #define DATA_NEW_ORDER_NULL_TYPE_BUF_SIZE	6
 
 /*************************************************************************
-Checks if a string type has to be compared by the MySQL comparison functions.
-InnoDB internally only handles binary byte string comparisons, as well as
-latin1_swedish_ci strings. For example, UTF-8 strings have to be compared
-by MySQL. */
-
-ibool
-dtype_str_needs_mysql_cmp(
-/*======================*/
-				/* out: TRUE if a string type that requires
-				comparison with MySQL functions */
-	dtype_t*	dtype);	/* in: type struct */
-/*************************************************************************
-For the documentation of this function, see innobase_get_at_most_n_mbchars()
-in ha_innodb.cc. */
+Determine how many bytes the first n characters of the given string occupy.
+If the string is shorter than n characters, returns the number of bytes
+the characters in the string occupy. */
 
 ulint
 dtype_get_at_most_n_mbchars(
 /*========================*/
-	dtype_t*	dtype,
-	ulint		prefix_len,
-	ulint		data_len,
-	const char*	str);
+					/* out: length of the prefix,
+					in bytes */
+	const dtype_t*	dtype,		/* in: data type */
+	ulint		prefix_len,	/* in: length of the requested
+					prefix, in characters, multiplied by
+					dtype_get_mbmaxlen(dtype) */
+	ulint		data_len,	/* in: length of str (in bytes) */
+	const char*	str);		/* in: the string whose prefix
+					length is being determined */
 /*************************************************************************
 Checks if a data main type is a string type. Also a BLOB is considered a
 string type. */
@@ -306,6 +300,14 @@ dtype_get_fixed_size(
 				/* out: fixed size, or 0 */
 	dtype_t*	type);	/* in: type */
 /***************************************************************************
+Returns the minimum size of a data type. */
+UNIV_INLINE
+ulint
+dtype_get_min_size(
+/*===============*/
+				/* out: minimum size */
+	const dtype_t*	type);	/* in: type */
+/***************************************************************************
 Returns a stored SQL NULL size for a type. For fixed length types it is
 the fixed length of the type, otherwise 0. */
 UNIV_INLINE
diff --git a/innobase/include/data0type.ic b/innobase/include/data0type.ic
index 761f7b3da7f..47f35bdce67 100644
--- a/innobase/include/data0type.ic
+++ b/innobase/include/data0type.ic
@@ -314,6 +314,7 @@ dtype_new_read_for_order_and_null_size(
 	dtype_set_mblen(type);
 }
 
+#ifndef UNIV_HOTBACKUP
 /***************************************************************************
 Returns the size of a fixed size data type, 0 if not a fixed size type. */
 UNIV_INLINE
@@ -323,7 +324,6 @@ dtype_get_fixed_size(
 				/* out: fixed size, or 0 */
 	dtype_t*	type)	/* in: type */
 {
-#ifndef UNIV_HOTBACKUP
 	ulint	mtype;
 
 	mtype = dtype_get_mtype(type);
@@ -401,15 +401,66 @@ dtype_get_fixed_size(
 	}
 
 	return(0);
-#else /* UNIV_HOTBACKUP */
-	/* This function depends on MySQL code that is not included in
-	InnoDB Hot Backup builds.  Besides, this function should never
-	be called in InnoDB Hot Backup. */
-	ut_error;
-#endif /* UNIV_HOTBACKUP */
 }
 
 /***************************************************************************
+Returns the size of a fixed size data type, 0 if not a fixed size type. */
+UNIV_INLINE
+ulint
+dtype_get_min_size(
+/*===============*/
+				/* out: minimum size */
+	const dtype_t*	type)	/* in: type */
+{
+	switch (type->mtype) {
+	case DATA_SYS:
+#ifdef UNIV_DEBUG
+			switch (type->prtype & DATA_MYSQL_TYPE_MASK) {
+			default:
+				ut_ad(0);
+				return(0);
+			case DATA_ROW_ID:
+				ut_ad(type->len == DATA_ROW_ID_LEN);
+				break;
+			case DATA_TRX_ID:
+				ut_ad(type->len == DATA_TRX_ID_LEN);
+				break;
+			case DATA_ROLL_PTR:
+				ut_ad(type->len == DATA_ROLL_PTR_LEN);
+				break;
+			case DATA_MIX_ID:
+				ut_ad(type->len == DATA_MIX_ID_LEN);
+				break;
+			}
+#endif /* UNIV_DEBUG */
+	case DATA_CHAR:
+	case DATA_FIXBINARY:
+	case DATA_INT:
+	case DATA_FLOAT:
+	case DATA_DOUBLE:
+	case DATA_MYSQL:
+			if ((type->prtype & DATA_BINARY_TYPE)
+					|| type->mbminlen == type->mbmaxlen) {
+				return(type->len);
+			}
+			/* this is a variable-length character set */
+			ut_a(type->mbminlen > 0);
+			ut_a(type->mbmaxlen > type->mbminlen);
+			return(type->len * type->mbminlen / type->mbmaxlen);
+	case DATA_VARCHAR:
+	case DATA_BINARY:
+	case DATA_DECIMAL:
+	case DATA_VARMYSQL:
+	case DATA_BLOB:
+			return(0); 
+	default:	ut_error;
+	}
+
+	return(0);
+}
+#endif /* !UNIV_HOTBACKUP */
+
+/***************************************************************************
 Returns a stored SQL NULL size for a type. For fixed length types it is
 the fixed length of the type, otherwise 0. */
 UNIV_INLINE
diff --git a/innobase/include/row0mysql.h b/innobase/include/row0mysql.h
index 7ffa6ebf87d..48a9d9bc941 100644
--- a/innobase/include/row0mysql.h
+++ b/innobase/include/row0mysql.h
@@ -99,6 +99,7 @@ row_mysql_store_col_in_innobase_format(
 					as dfield is used! */
 	ulint		col_len,	/* in: MySQL column length */
 	ulint		type,		/* in: data type */
+	bool		comp,		/* in: TRUE=compact format */
 	ulint		is_unsigned);	/* in: != 0 if unsigned integer type */
 /********************************************************************
 Handles user errors and lock waits detected by the database engine. */
diff --git a/innobase/include/row0mysql.ic b/innobase/include/row0mysql.ic
index 1f5d0b0c990..910546e298c 100644
--- a/innobase/include/row0mysql.ic
+++ b/innobase/include/row0mysql.ic
@@ -67,6 +67,7 @@ row_mysql_store_col_in_innobase_format(
 					as dfield is used! */
 	ulint		col_len,	/* in: MySQL column length */
 	ulint		type,		/* in: data type */
+	bool		comp,		/* in: TRUE=compact format */
 	ulint		is_unsigned)	/* in: != 0 if unsigned integer type */
 {
 	byte*	ptr 	= mysql_data;
@@ -113,6 +114,37 @@ row_mysql_store_col_in_innobase_format(
 				col_len--;
 			}
 		}
+	} else if (comp && type == DATA_MYSQL
+			&& dtype_get_mbminlen(dfield_get_type(dfield)) == 1
+			&& dtype_get_mbmaxlen(dfield_get_type(dfield)) > 1) {
+		/* We assume that this CHAR field is encoded in a
+		variable-length character set where spaces have
+		1:1 correspondence to 0x20 bytes, such as UTF-8.
+
+		Consider a CHAR(n) field, a field of n characters.
+		It will contain between n*mbminlen and n*mbmaxlen bytes.
+		We will try to truncate it to n bytes by stripping
+		space padding.  If the field contains single-byte
+		characters only, it will be truncated to n characters.
+		Consider a CHAR(5) field containing the string ".a   "
+		where "." denotes a 3-byte character represented by
+		the bytes "$%&".  After our stripping, the string will
+		be stored as "$%&a " (5 bytes).  The string ".abc "
+		will be stored as "$%&abc" (6 bytes).
+
+		The space padding will be restored in row0sel.c, function
+		row_sel_field_store_in_mysql_format(). */
+
+		ulint		n_chars;
+		dtype_t*	dtype = dfield_get_type(dfield);
+
+		ut_a(!(dtype_get_len(dtype) % dtype_get_mbmaxlen(dtype)));
+		n_chars = dtype_get_len(dtype) / dtype_get_mbmaxlen(dtype);
+
+		/* Strip space padding. */
+		while (col_len > n_chars && ptr[col_len - 1] == 0x20) {
+			col_len--;
+		}
 	} else if (type == DATA_BLOB) {
 		ptr = row_mysql_read_blob_ref(&col_len, mysql_data, col_len);
 	}