Bug #44372 bug 22125 affects handler::index_next_same in 5.1

Submitted: 20 Apr 2009 20:17 Modified: 31 Jul 2009 8:19
Reporter: Zardosht Kasheff (OCA) Email Updates:
Status: Documenting Impact on me:
None 
Category:MySQL Server Severity:S3 (Non-critical)
Version:5.1.30, 5.1 bzr OS:Any
Assigned to: CPU Architecture:Any

[20 Apr 2009 20:17] Zardosht Kasheff
Description:
running falcon_bug_22125 on a storage engine that does NOT implement ::index_next_same (and as a result uses handler::index_next_same), the error described in 22125 is reproduced.

How to repeat:
run falcon_bug_22125 on 5.1.30 using a storage engine that does NOT implement index_next_same on 5.1.30
[31 Jul 2009 8:19] Sveta Smirnova
Thank you for the report.

Verified as described.

To repeat apply patch following to 5.1 tree:

$bzr diff
=== modified file 'storage/myisam/ha_myisam.cc'
--- storage/myisam/ha_myisam.cc 2009-07-08 12:11:34 +0000
+++ storage/myisam/ha_myisam.cc 2009-07-31 08:07:33 +0000
@@ -1724,7 +1724,7 @@
   return error;
 }
 
-int ha_myisam::index_next_same(uchar *buf,
+int ha_myisam::index_next_same1(uchar *buf,
                               const uchar *key __attribute__((unused)),
                               uint length __attribute__((unused)))
 {

=== modified file 'storage/myisam/ha_myisam.h'
--- storage/myisam/ha_myisam.h  2009-07-08 12:11:34 +0000
+++ storage/myisam/ha_myisam.h  2009-07-31 08:07:20 +0000
@@ -75,7 +75,7 @@
   int index_prev(uchar * buf);
   int index_first(uchar * buf);
   int index_last(uchar * buf);
-  int index_next_same(uchar *buf, const uchar *key, uint keylen);
+  int index_next_same1(uchar *buf, const uchar *key, uint keylen);
   int ft_init()
   {
     if (!ft_handler)

Then modify test as following:

$cat t/falcon_bug_22125.test 
#
# Bug #22125: Falcon: Double precision searches fail if index exists
#
--echo *** Bug #22125 ***

# ----------------------------------------------------- #
# --- Initialisation                                --- #
# ----------------------------------------------------- #
#let $engine = 'Falcon';
let $engine = 'MyISAM';
eval SET @@storage_engine = $engine;

--disable_warnings
DROP TABLE IF EXISTS t1;
--enable_warnings

CREATE TABLE t1 (
  f1 double,
  f2 char(5),
  f3 double,
  f4 int
);

INSERT INTO t1 VALUES (0,'zero',0,1);
INSERT INTO t1 VALUES (1.7976931348623157E+308,'max',0,3);
INSERT INTO t1 VALUES (1.7976931348623157E+308,'min',0,4);

UPDATE t1 SET f1 = 1 / f1 WHERE f2 = 'min';

INSERT INTO t1 VALUES (1.7976931348623157E+308,'nan',0,5);
INSERT INTO t1 VALUES (1.7976931348623157E+308,'inf',0,6);
INSERT INTO t1 SELECT -f1,concat('-',f2),f3,-f4 FROM t1;

UPDATE t1 SET f3 = -f3 WHERE f2 = '-nan';

CREATE INDEX it1 on t1 (f1);

# ----------------------------------------------------- #
# --- Test                                          --- #
# ----------------------------------------------------- #
SELECT * FROM t1 ORDER BY f1,f2,f3,f4;
SELECT count(*) FROM t1 where f1 = 0;
SELECT count(*) FROM t1 where f1 < 1;

# Without index we get correct count.
DROP INDEX it1 ON t1;
SELECT count(*) FROM t1 where f1 = 0;
SELECT count(*) FROM t1 where f1 < 1;

# ----------------------------------------------------- #
# --- Check                                         --- #
# ----------------------------------------------------- #
SELECT count(*) FROM t1;

# ----------------------------------------------------- #
# --- Final cleanup                                 --- #
# ----------------------------------------------------- #
DROP TABLE t1;

Get -0 and wrong count in result
[13 Oct 9:36] Knut Anders Hatlen
https://github.com/mysql/mysql-server/pull/788 seems to fix this bug too.

See also bug#67978.
[26 Oct 9:48] MySQL Admin
Posted by developer: Bug status updated to 'Documenting'

Portions of this website are copyright © 2001, 2002 The PHP Group

Page generated in 0.056 sec. using MySQL 8.4.7-u6-cloud

Timestamp references displayed by the system are UTC.
Content reproduced on this site is the property of the respective copyright holders.
It is not reviewed in advance by Oracle and does not necessarily represent the opinion of Oracle or any other party.