Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.6.1
Description
After MDEV-25691, the following modification to a test would cause a test failure:
diff --git a/mysql-test/suite/innodb/t/alter_foreign_crash.test b/mysql-test/suite/innodb/t/alter_foreign_crash.test
|
index e2a6c301bf7..b65dfa175c2 100644
|
--- a/mysql-test/suite/innodb/t/alter_foreign_crash.test
|
+++ b/mysql-test/suite/innodb/t/alter_foreign_crash.test
|
@@ -2,6 +2,7 @@
|
--source include/have_debug_sync.inc
|
# The embedded server does not support restarting.
|
--source include/not_embedded.inc
|
+--source include/have_case_sensitive_file_system.inc
|
|
--echo #
|
--echo # Bug #20476395 DICT_LOAD_FOREIGNS() FAILED IN
|
@@ -10,6 +11,11 @@
|
|
call mtr.add_suppression("InnoDB: Failed to load table");
|
|
+CREATE DATABASE Bug;
|
+CREATE TABLE Bug.parent(a SERIAL) ENGINE=INNODB;
|
+CREATE TABLE Bug.child(a SERIAL, FOREIGN KEY f(a) REFERENCES Bug.parent(a))
|
+ENGINE=INNODB;
|
+
|
create database bug;
|
use bug;
|
|
@@ -18,6 +24,7 @@ create table child(a serial, foreign key fk (a) references parent(a))engine=inno
|
|
insert into parent values(1);
|
insert into child values(1);
|
+drop database Bug;
|
|
connect (con1,localhost,root,,bug);
|
SET DEBUG_SYNC='innodb_rename_table_ready SIGNAL s1 WAIT_FOR s2 EXECUTE 2'; |
CURRENT_TEST: innodb.alter_foreign_crash
|
mysqltest: At line 44: query 'drop table parent' succeeded - should have failed with error ER_ROW_IS_REFERENCED_2 (1451)...
|
The problem is that the primary key of SYS_FOREIGN.ID is in latin1_swedish_ci and not are in a binary collation. Even though table names are in binary collation, the internal name Bug/fk would clash with bug/fk. Hence, the above test creates Buf/f. On DROP DATABASE Bug, also the constraint bug/fk would be removed. The following fixes that:
diff --git a/storage/innobase/handler/ha_innodb.cc b/storage/innobase/handler/ha_innodb.cc
|
index 4a3be1fb1d8..2b6089c4814 100644
|
--- a/storage/innobase/handler/ha_innodb.cc
|
+++ b/storage/innobase/handler/ha_innodb.cc
|
@@ -1421,8 +1421,9 @@ static void innodb_drop_database(handlerton*, char *path)
|
"WHILE 1 = 1 LOOP\n"
|
" FETCH fkf INTO fk;\n"
|
" IF (SQL % NOTFOUND) THEN EXIT; END IF;\n"
|
- " IF SUBSTR(fk, 0, LENGTH(:db)) <> :db THEN EXIT; END IF;\n"
|
- " DELETE FROM SYS_FOREIGN_COLS WHERE ID=fk;\n"
|
+ " IF TO_BINARY(SUBSTR(fk, 0, LENGTH(:db)))<>TO_BINARY(:db)"
|
+ " THEN EXIT; END IF;\n"
|
+ " DELETE FROM SYS_FOREIGN_COLS WHERE TO_BINARY(ID)=TO_BINARY(fk);\n"
|
" DELETE FROM SYS_FOREIGN WHERE CURRENT OF fkf;\n"
|
"END LOOP;\n"
|
"CLOSE fkf;\n" |
In earlier versions (before MDEV-25691 cleaned up the code in 10.6.1), there is an even more severe bug: not only unrelated constraints but also unrelated tables are being dropped:
|
10.5 9bbedcdd590f2c25c79bd9f367a7f54c1fcefdcb |
CURRENT_TEST: innodb.alter_foreign_crash
|
mysqltest: At line 42: query 'select * from child' failed: 1146: Table 'bug.child' doesn't exist
|
That problem probably affects all previous InnoDB versions, so I would not call the milder bug in 10.6 a regression:
|
10.2 629449172a5b0a6975663ca1ac420789e00b941d |
CURRENT_TEST: innodb.alter_foreign_crash
|
mysqltest: At line 42: query 'select * from child' failed: 1146: Table 'bug.child' doesn't exist
|
Attachments
Issue Links
- is caused by
-
MDEV-25691 Simplify handlerton::drop_database for InnoDB
-
- Closed
-
- relates to
-
-
- Closed
-
-
-
- Closed
-