Skip to content

PS-10481: Fix range optimizer full table scan for IN() with oversized values #5871

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
percona-mhansson wants to merge 1 commit into
base: 8.4
Choose a base branch
from
Open

PS-10481: Fix range optimizer full table scan for IN() with oversized values #5871

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
174 changes: 174 additions & 0 deletions mysql-test/include/oversized_varchar_key.inc
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,174 @@
eval CREATE TABLE t1 ( pk VARCHAR(16) NOT NULL PRIMARY KEY ) CHARSET=$charset COLLATE=$collation;

INSERT INTO t1 VALUES ('abc'), ('def'), ('ghi'), ('jkl'), ('mno');
ANALYZE TABLE t1;

EXPLAIN
SELECT * FROM t1 WHERE pk IN ('abc', 'this_value_is_>16', 'def');
SELECT * FROM t1 WHERE pk IN ('abc', 'this_value_is_>16', 'def');

EXPLAIN
SELECT * FROM t1 WHERE pk IN ('abc', 'this_value_is_>16', 'def');
SELECT * FROM t1 WHERE pk IN ('abc', 'this_value_is_>16', 'def');

EXPLAIN
SELECT * FROM t1 WHERE pk IN ('this_value_is_>16');
SELECT * FROM t1 WHERE pk IN ('this_value_is_>16');

EXPLAIN
SELECT * FROM t1 WHERE pk IN ('this_value_is_>16', 'and_so_is_this_one');
SELECT * FROM t1 WHERE pk IN ('this_value_is_>16', 'and_so_is_this_one');


eval CREATE TABLE t2 ( a VARCHAR(1), KEY (a) ) CHARSET=$charset COLLATE=$collation;

INSERT INTO t2 VALUES (NULL), ('a'), ('b'), ('æ'), ('ß'), ('s');
INSERT INTO t2 SELECT 'x' FROM t2;
INSERT INTO t2 SELECT 'x' FROM t2;

ANALYZE TABLE t2;

--echo # Strings containing soft hyphen (U+00AD) will not match for UTF-16

EXPLAIN
SELECT * FROM t2 WHERE a IN ('ae', concat(_utf8mb4 x'c2ad', 'b'), 'ss');
SELECT * FROM t2 WHERE a IN ('ae', concat(_utf8mb4 x'c2ad', 'b'), 'ss');

EXPLAIN
SELECT * FROM t2 WHERE a IN (concat(_utf8mb4 x'c2ad', 'ae'), 'ss');
SELECT * FROM t2 WHERE a IN (concat(_utf8mb4 x'c2ad', 'ae'), 'ss');


EXPLAIN
SELECT * FROM t2 WHERE a IN (concat(_utf8mb4 x'c2ad', 'ae', _utf8mb4 x'c2ad'), 'ss');
SELECT * FROM t2 WHERE a IN (concat(_utf8mb4 x'c2ad', 'ae', _utf8mb4 x'c2ad'), 'ss');

EXPLAIN
SELECT * FROM t2 WHERE a = 'ae' OR a = concat(_utf8mb4 x'c2ad', 'b') OR a = 'ss';
SELECT * FROM t2 WHERE a = 'ae' OR a = concat(_utf8mb4 x'c2ad', 'b') OR a = 'ss';

--echo # The plan is unstable in this case, however it always uses an index, and results are consistent.
--replace_column 5 X 10 X
EXPLAIN
SELECT * FROM t2 WHERE a IN ('ae', concat(repeat(_utf8mb4 x'c2ad', 2), 'b', repeat(_utf8mb4 x'c2ad', 2)));
--echo # Adding handler statistics to show that execution is also consistent.
FLUSH STATUS;
SELECT * FROM t2 WHERE a IN ('ae', concat(repeat(_utf8mb4 x'c2ad', 2), 'b', repeat(_utf8mb4 x'c2ad', 2)));
SHOW STATUS LIKE 'Handler_read_%';


eval CREATE TABLE t3 ( a VARCHAR(768), KEY(a)) CHARSET=$charset COLLATE=$collation;

INSERT INTO t3 VALUES( repeat('æ', 767) ), ( repeat('æ', 768) );

ANALYZE TABLE t3;

EXPLAIN
SELECT 1 FROM t3 WHERE a IN ( 'a', repeat('æ', 767) );
SELECT 1 FROM t3 WHERE a IN ( 'a', repeat('æ', 767) );

EXPLAIN
SELECT 1 FROM t3 WHERE a IN ( 'a', repeat('ae', 768) );
SELECT 1 FROM t3 WHERE a IN ( 'a', repeat('ae', 768) );

EXPLAIN
SELECT 1 FROM t3 WHERE a IN ( 'a', repeat('ae', 769) );
SELECT 1 FROM t3 WHERE a IN ( 'a', repeat('ae', 769) );


eval CREATE TABLE t4 ( a VARCHAR(100), KEY(a(50))) CHARSET=$charset COLLATE=$collation;

INSERT INTO t4 VALUES ( repeat('æ', 98) ), ( repeat('æ', 99) ), ( repeat('æ', 100) );
INSERT INTO t4 VALUES ( concat(repeat( 'a', 50 ), repeat( 'æ', 50 )) );

ANALYZE TABLE t4;

EXPLAIN
SELECT 1 FROM t4 WHERE a IN ( 'a', repeat('æ', 98) );
SELECT 1 FROM t4 WHERE a IN ( 'a', repeat('æ', 98) );


eval CREATE TABLE t5 ( a VARCHAR(16383), KEY(a(768))) CHARSET=$charset COLLATE=$collation;

INSERT INTO t5 VALUES( repeat('æ', 16383) ), ( repeat('æ', 16382) ), ( repeat('æ', 16381) );

ANALYZE TABLE t5;

EXPLAIN
SELECT 1 FROM t5 WHERE a IN ( 'a', repeat('æ', 16381) );
SELECT 1 FROM t5 WHERE a IN ( 'a', repeat('æ', 16381) );

EXPLAIN
SELECT 1 FROM t5 WHERE a IN ( 'a', repeat('æ', 16382) );
SELECT 1 FROM t5 WHERE a IN ( 'a', repeat('æ', 16382) );

EXPLAIN
SELECT 1 FROM t5 WHERE a IN ( 'a', repeat('æ', 16383) );
SELECT 1 FROM t5 WHERE a IN ( 'a', repeat('æ', 16383) );


eval CREATE TABLE t6 ( a VARCHAR(6), KEY(a(3)) ) CHARSET=$charset COLLATE=$collation;

INSERT INTO t6 VALUES ('aaaaaa'), ('æaaaaa'), ('ææaaaa'), ('æææaaa'), ('ææææaa'), ('æææææa'), ('ææææææ'),
('😊aaaaa'), ('😊æaaaa'), ('😊😊aaaa'), ('😊😊æaaa'), ('😊😊😊aaa');

ANALYZE TABLE t6;

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', 'aaaaaa' );
SELECT * FROM t6 WHERE a IN ( '', 'aaaaaa' );

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', 'æaaaaa' );
SELECT * FROM t6 WHERE a IN ( '', 'æaaaaa' );

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', 'ææaaaa' );
SELECT * FROM t6 WHERE a IN ( '', 'ææaaaa' );

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', 'æææaaa' );
SELECT * FROM t6 WHERE a IN ( '', 'æææaaa' );

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', 'ææææaa' );
SELECT * FROM t6 WHERE a IN ( '', 'ææææaa' );

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', 'æææææa' );
SELECT * FROM t6 WHERE a IN ( '', 'æææææa' );

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', 'ææææææ' );
SELECT * FROM t6 WHERE a IN ( '', 'ææææææ' );

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', '😊aaaaa');
SELECT * FROM t6 WHERE a IN ( '', '😊aaaaa');

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', '😊æaaaa');
SELECT * FROM t6 WHERE a IN ( '', '😊æaaaa');

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', '😊😊aaaa');
SELECT * FROM t6 WHERE a IN ( '', '😊😊aaaa');

EXPLAIN
SELECT * FROM t6 WHERE a IN ( '', '😊😊æaaa');
SELECT * FROM t6 WHERE a IN ( '', '😊😊😊aaa');

EXPLAIN
SELECT * FROM t6 WHERE a IN ('aaaaaa', 'æaaaaa', 'ææaaaa', 'æææaaa', 'ææææaa', 'æææææa', 'ææææææ');
SELECT * FROM t6 WHERE a IN ('aaaaaa', 'æaaaaa', 'ææaaaa', 'æææaaa', 'ææææaa', 'æææææa', 'ææææææ');

EXPLAIN
SELECT * FROM t6 WHERE a IN
('aaaaaa', 'æaaaaa', 'ææaaaa', 'æææaaa', 'ææææaa', 'æææææa', 'ææææææ',
'😊aaaaa', '😊æaaaa', '😊😊aaaa', '😊😊æaaa', '😊😊😊aaa');
SELECT * FROM t6 WHERE a IN
('aaaaaa', 'æaaaaa', 'ææaaaa', 'æææaaa', 'ææææaa', 'æææææa', 'ææææææ',
'😊aaaaa', '😊æaaaa', '😊😊aaaa', '😊😊æaaa', '😊😊😊aaa');


DROP TABLE t1, t2, t3, t4, t5, t6;
Copy link
Copy Markdown
Contributor

@catalinbp catalinbp Apr 10, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Include this

SET @old_max_allowed_packet  = @@global.max_allowed_packet
SET @old_max_allowed_packet= @@global.max_allowed_packet;
SET @@global.max_allowed_packet = 1024 * 1024 + 1024;
CREATE TABLE t (x VARCHAR(1), KEY (x));
INSERT INTO t VALUES ('a'), ('b');

--echo # U+00AD is SOFT HYPHEN, which is ignored in the default collation, so
--echo # large number of SOFT HYPHEN characters followed by 'a' should match 'a'.
SET @soft_hyphen = CAST(_utf16 x'00AD' AS CHAR CHARSET utf8mb4);
EXPLAIN FORMAT=TREE SELECT * FROM t  IGNORE KEY (x) WHERE x <= CONCAT(REPEAT(@soft_hyphen, 1024 * 1024 + 1), 'a');
SELECT * FROM t  IGNORE KEY (x) WHERE x <= CONCAT(REPEAT(@soft_hyphen, 1024 * 1024 + 1), 'a');
EXPLAIN FORMAT=TREE SELECT * FROM t  WHERE x <= CONCAT(REPEAT(@soft_hyphen, 1024 * 1024 + 1024 + 1), 'a');
SELECT * FROM t WHERE x <= CONCAT(REPEAT(@soft_hyphen, 1024 * 1024 + 1), 'a');
EXPLAIN FORMAT=TREE SELECT * FROM t  FORCE KEY (x) WHERE x <= CONCAT(REPEAT(@soft_hyphen, 1024 * 1024 + 1024 + 1), 'a');
SELECT * FROM t  FORCE KEY (x) WHERE x <= CONCAT(REPEAT(@soft_hyphen, 1024 * 1024 + 1), 'a');
DROP TABLE t;

SET @@global.max_allowed_packet = @old_max_allowed_packet;

I see different results between table scan and index range scan.

Loading