Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
5.5, 10.0, 10.1, 10.2, 10.3, 10.4
-
None
Description
DROP TABLE IF EXISTS t1; |
CREATE TABLE t1 ( |
id TINYINT NOT NULL AUTO_INCREMENT, |
name CHAR(30) NOT NULL, |
PRIMARY KEY (id) |
) ENGINE=InnoDB;
|
INSERT INTO t1 (name) VALUES ('dog'); |
UPDATE t1 SET id=-1 WHERE id=1; |
INSERT INTO t1 (name) VALUES ('cat'); |
SELECT * FROM t1; |
DROP TABLE t1; |
+----+------+
|
| id | name |
|
+----+------+
|
| -1 | dog |
|
| 2 | cat |
|
+----+------+
|
If I now change ENGINE to HEAP, MyISAM or ARIA:
DROP TABLE IF EXISTS t1; |
CREATE TABLE t1 ( |
id TINYINT NOT NULL AUTO_INCREMENT, |
name CHAR(30) NOT NULL, |
PRIMARY KEY (id) |
) ENGINE=MyISAM;
|
INSERT INTO t1 (name) VALUES ('dog'); |
UPDATE t1 SET id=-1 WHERE id=1; |
INSERT INTO t1 (name) VALUES ('cat'); |
SELECT * FROM t1; |
DROP TABLE t1; |
the script fails in the last INSERT with this error:
ERROR 167 (22003): Out of range value for column 'id' at row 1
|
Looks wrong. All engines should insert two rows without errors, like InnoDB does.
Note, if I run the same scripts on other platforms (e.g. x86-64), all engines work as expected, like InnoDB.
The problem resides in these code fragments in HEAP, MyISAM and ARIA implementations respectively:
void heap_update_auto_increment(HP_INFO *info, const uchar *record) |
{
|
..
|
case HA_KEYTYPE_INT8: |
s_value= (longlong) *(char*)key; |
break; |
..
|
}
|
ulonglong retrieve_auto_increment(MI_INFO *info,const uchar *record) |
{
|
..
|
case HA_KEYTYPE_INT8: |
s_value= (longlong) *(char*)key; |
break; |
..
|
}
|
ulonglong ma_retrieve_auto_increment(const uchar *key, uint8 key_type) |
{
|
..
|
case HA_KEYTYPE_INT8: |
s_value= (longlong) *(const char*)key; |
break; |
 |
..
|
}
|
"char" is unsigned by default on PPC64.
The cast operators should include the "signed" keyword, i.e:
case HA_KEYTYPE_INT8: |
s_value= (longlong) *(const signed char*)key; |
break; |
Attachments
Issue Links
- relates to
-
-
- Closed
-
-
-
- Closed
-