mariadb/mysql-test/main/create-big.test

553 lines
17 KiB
Text
Raw Normal View History

5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# Tests for various concurrency-related aspects of CREATE TABLE ... SELECT
# and CREATE TABLE like implementation.
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
#
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# Note that we don't test general CREATE TABLE ... SELECT/LIKE functionality
# here as it is already covered by create.test. We are more interested in
# extreme cases.
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
#
# This test takes rather long time so let us run it only in --big-test mode
--source include/big_test.inc
# We need the Debug Sync Facility.
--source include/have_debug_sync.inc
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# Some of tests below also use binlog to check that statements are
# executed and logged in correct order
--source include/have_binlog_format_mixed_or_statement.inc
# Save the initial number of concurrent sessions.
--source include/count_sessions.inc
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Create auxilliary connections
connect (addconroot1, localhost, root,,);
connect (addconroot2, localhost, root,,);
connect (addconroot3, localhost, root,,);
connection default;
--disable_warnings
drop table if exists t1,t2,t3,t4,t5;
--enable_warnings
set debug_sync='RESET';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
#
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# Tests for concurrency problems in CREATE TABLE ... SELECT
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
#
# We introduce delays between various stages of table creation
# and check that other statements dealing with this table cannot
# interfere during those delays.
#
# What happens in situation when other statement messes with
# table to be created before it is created ?
# Concurrent CREATE TABLE
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send create table t1 (j char(5));
connection addconroot2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "create table t1 (j char(5))";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--error ER_TABLE_EXISTS_ERROR
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
show create table t1;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent CREATE TABLE ... SELECT
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send create table t1 select 'Test' as j;
connection addconroot2;
# Wait until the above CREATE TABLE t1 is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "create table t1 select 'Test' as j";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--error ER_TABLE_EXISTS_ERROR
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
show create table t1;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent CREATE TABLE LIKE
create table t3 (j char(5));
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send create table t1 like t3;
connection addconroot2;
# Wait until the above CREATE TABLE t1 is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "create table t1 like t3";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--error ER_TABLE_EXISTS_ERROR
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
show create table t1;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent RENAME TABLE
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send rename table t3 to t1;
connection addconroot2;
# Wait until the above RENAME TABLE is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "rename table t3 to t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--error ER_TABLE_EXISTS_ERROR
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
show create table t1;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent ALTER TABLE RENAME
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
Bug#13982017: ALTER TABLE RENAME ENDS UP WITH ERROR 1050 (42S01) Fixed by backport of: ------------------------------------------------------------ revno: 3402.50.156 committer: Jon Olav Hauglid <jon.hauglid@oracle.com> branch nick: mysql-trunk-test timestamp: Wed 2012-02-08 14:10:23 +0100 message: Bug#13417754 ASSERT IN ROW_DROP_DATABASE_FOR_MYSQL DURING DROP SCHEMA This assert could be triggered if an InnoDB table was being moved to a different database using ALTER TABLE ... RENAME, while this database concurrently was being dropped by DROP DATABASE. The reason for the problem was that no metadata lock was taken on the target database by ALTER TABLE ... RENAME. DROP DATABASE was therefore not blocked and could remove the database while ALTER TABLE ... RENAME was executing. This could cause the assert in InnoDB to be triggered. This patch fixes the problem by taking a IX metadata lock on the target database before ALTER TABLE ... RENAME starts moving a table to a different database. Note that this problem did not occur with RENAME TABLE which already takes the correct metadata locks. Also note that this patch slightly changes the behavior of ALTER TABLE ... RENAME. Before, the statement would abort and return an error if a lock on the target table name could not be taken immediately. With this patch, ALTER TABLE ... RENAME will instead block and wait until the lock can be taken (or until we get a lock timeout). This also means that it is possible to get ER_LOCK_DEADLOCK errors in this situation since we allow ALTER TABLE ... RENAME to wait and not just abort immediately.
2012-06-01 09:31:24 +02:00
--send alter table t3 rename to t1
connection addconroot2;
# Wait until the above ALTER TABLE RENAME is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "alter table t3 rename to t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
Bug#13982017: ALTER TABLE RENAME ENDS UP WITH ERROR 1050 (42S01) Fixed by backport of: ------------------------------------------------------------ revno: 3402.50.156 committer: Jon Olav Hauglid <jon.hauglid@oracle.com> branch nick: mysql-trunk-test timestamp: Wed 2012-02-08 14:10:23 +0100 message: Bug#13417754 ASSERT IN ROW_DROP_DATABASE_FOR_MYSQL DURING DROP SCHEMA This assert could be triggered if an InnoDB table was being moved to a different database using ALTER TABLE ... RENAME, while this database concurrently was being dropped by DROP DATABASE. The reason for the problem was that no metadata lock was taken on the target database by ALTER TABLE ... RENAME. DROP DATABASE was therefore not blocked and could remove the database while ALTER TABLE ... RENAME was executing. This could cause the assert in InnoDB to be triggered. This patch fixes the problem by taking a IX metadata lock on the target database before ALTER TABLE ... RENAME starts moving a table to a different database. Note that this problem did not occur with RENAME TABLE which already takes the correct metadata locks. Also note that this patch slightly changes the behavior of ALTER TABLE ... RENAME. Before, the statement would abort and return an error if a lock on the target table name could not be taken immediately. With this patch, ALTER TABLE ... RENAME will instead block and wait until the lock can be taken (or until we get a lock timeout). This also means that it is possible to get ER_LOCK_DEADLOCK errors in this situation since we allow ALTER TABLE ... RENAME to wait and not just abort immediately.
2012-06-01 09:31:24 +02:00
connection addconroot1;
--error ER_TABLE_EXISTS_ERROR
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
show create table t1;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent ALTER TABLE RENAME which also adds column
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
Bug#13982017: ALTER TABLE RENAME ENDS UP WITH ERROR 1050 (42S01) Fixed by backport of: ------------------------------------------------------------ revno: 3402.50.156 committer: Jon Olav Hauglid <jon.hauglid@oracle.com> branch nick: mysql-trunk-test timestamp: Wed 2012-02-08 14:10:23 +0100 message: Bug#13417754 ASSERT IN ROW_DROP_DATABASE_FOR_MYSQL DURING DROP SCHEMA This assert could be triggered if an InnoDB table was being moved to a different database using ALTER TABLE ... RENAME, while this database concurrently was being dropped by DROP DATABASE. The reason for the problem was that no metadata lock was taken on the target database by ALTER TABLE ... RENAME. DROP DATABASE was therefore not blocked and could remove the database while ALTER TABLE ... RENAME was executing. This could cause the assert in InnoDB to be triggered. This patch fixes the problem by taking a IX metadata lock on the target database before ALTER TABLE ... RENAME starts moving a table to a different database. Note that this problem did not occur with RENAME TABLE which already takes the correct metadata locks. Also note that this patch slightly changes the behavior of ALTER TABLE ... RENAME. Before, the statement would abort and return an error if a lock on the target table name could not be taken immediately. With this patch, ALTER TABLE ... RENAME will instead block and wait until the lock can be taken (or until we get a lock timeout). This also means that it is possible to get ER_LOCK_DEADLOCK errors in this situation since we allow ALTER TABLE ... RENAME to wait and not just abort immediately.
2012-06-01 09:31:24 +02:00
--send alter table t3 rename to t1, add k int
connection addconroot2;
# Wait until the above ALTER TABLE RENAME is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "alter table t3 rename to t1, add k int";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
Bug#13982017: ALTER TABLE RENAME ENDS UP WITH ERROR 1050 (42S01) Fixed by backport of: ------------------------------------------------------------ revno: 3402.50.156 committer: Jon Olav Hauglid <jon.hauglid@oracle.com> branch nick: mysql-trunk-test timestamp: Wed 2012-02-08 14:10:23 +0100 message: Bug#13417754 ASSERT IN ROW_DROP_DATABASE_FOR_MYSQL DURING DROP SCHEMA This assert could be triggered if an InnoDB table was being moved to a different database using ALTER TABLE ... RENAME, while this database concurrently was being dropped by DROP DATABASE. The reason for the problem was that no metadata lock was taken on the target database by ALTER TABLE ... RENAME. DROP DATABASE was therefore not blocked and could remove the database while ALTER TABLE ... RENAME was executing. This could cause the assert in InnoDB to be triggered. This patch fixes the problem by taking a IX metadata lock on the target database before ALTER TABLE ... RENAME starts moving a table to a different database. Note that this problem did not occur with RENAME TABLE which already takes the correct metadata locks. Also note that this patch slightly changes the behavior of ALTER TABLE ... RENAME. Before, the statement would abort and return an error if a lock on the target table name could not be taken immediately. With this patch, ALTER TABLE ... RENAME will instead block and wait until the lock can be taken (or until we get a lock timeout). This also means that it is possible to get ER_LOCK_DEADLOCK errors in this situation since we allow ALTER TABLE ... RENAME to wait and not just abort immediately.
2012-06-01 09:31:24 +02:00
connection addconroot1;
--error ER_TABLE_EXISTS_ERROR
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
show create table t1;
drop table t1,t3;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# What happens if other statement sneaks in after the table
# creation but before its opening ?
set debug_sync='create_table_select_before_open SIGNAL parked WAIT_FOR go';
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent DROP TABLE
set debug_sync='create_table_select_before_open SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send drop table t1;
connection addconroot2;
# Wait until the above DROP TABLE is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "drop table t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent RENAME TABLE
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send rename table t1 to t2;
connection addconroot2;
# Wait until the above RENAME TABLE is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "rename table t1 to t2";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
drop table t2;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent SELECT
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send select * from t1;
connection addconroot2;
# Wait until the above SELECT is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "select * from t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent INSERT
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send insert into t1 values (2);
connection addconroot2;
# Wait until the above INSERT is blocked due to CREATE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "insert into t1 values (2)";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
select * from t1;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent CREATE TRIGGER
set @a:=0;
set debug_sync='create_table_select_before_create SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send create trigger t1_bi before insert on t1 for each row set @a:=1;
connection addconroot2;
# Wait until the above CREATE TRIGGER is blocked due to CREATE TABLE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "create trigger t1_bi before insert on t1 for each row set @a:=1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
select @a;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Okay, now the same tests for the potential gap between open and lock
set debug_sync='create_table_select_before_lock SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent DROP TABLE
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send drop table t1;
connection addconroot2;
# Wait until the above DROP TABLE is blocked due to CREATE TABLE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "drop table t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent RENAME TABLE
set debug_sync='create_table_select_before_lock SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send rename table t1 to t2;
connection addconroot2;
# Wait until the above RENAME TABLE is blocked due to CREATE TABLE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "rename table t1 to t2";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
drop table t2;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent SELECT
set debug_sync='create_table_select_before_lock SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send select * from t1;
connection addconroot2;
# Wait until the above SELECT is blocked due to CREATE TABLE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "select * from t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent INSERT
set debug_sync='create_table_select_before_lock SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send insert into t1 values (2);
connection addconroot2;
# Wait until the above INSERT INTO t1 is blocked due to CREATE TABLE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "insert into t1 values (2)";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
select * from t1;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent CREATE TRIGGER
set @a:=0;
set debug_sync='create_table_select_before_lock SIGNAL parked WAIT_FOR go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
--send create table t1 select 1 as i;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send create trigger t1_bi before insert on t1 for each row set @a:=1;
connection addconroot2;
# Wait until the above CREATE TRIGGER is blocked due to CREATE TABLE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "create trigger t1_bi before insert on t1 for each row set @a:=1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
select @a;
drop table t1;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
# Concurrent DROP TABLE
set debug_sync='create_table_before_check_if_exists SIGNAL parked WAIT_FOR go';
--send create table if not exists t1 select 1 as i
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send drop table t1
connection addconroot2;
# Wait until the above DROP TABLE is blocked due to CREATE TABLE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "drop table t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
# Concurrent CREATE TRIGGER
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
create table t1 (i int);
set @a:=0;
set debug_sync='create_table_before_check_if_exists SIGNAL parked WAIT_FOR go';
--send create table if not exists t1 select 1 as i
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send create trigger t1_bi before insert on t1 for each row set @a:=1
connection addconroot2;
# Wait until the above DROP TABLE is blocked due to CREATE TABLE
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info like "create trigger%";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
Fix for: Bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables" Bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" Bug #24738 "CREATE TABLE ... SELECT is not isolated properly" Bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" Deadlock occured when one tried to execute CREATE TABLE IF NOT EXISTS ... SELECT statement under LOCK TABLES which held read lock on target table. Attempt to execute the same statement for already existing target table with triggers caused server crashes. Also concurrent execution of CREATE TABLE ... SELECT statement and other statements involving target table suffered from various races (some of which might've led to deadlocks). Finally, attempt to execute CREATE TABLE ... SELECT in case when a temporary table with same name was already present led to the insertion of data into this temporary table and creation of empty non-temporary table. All above problems stemmed from the old implementation of CREATE TABLE ... SELECT in which we created, opened and locked target table without any special protection in a separate step and not with the rest of tables used by this statement. This underminded deadlock-avoidance approach used in server and created window for races. It also excluded target table from prelocking causing problems with trigger execution. The patch solves these problems by implementing new approach to handling of CREATE TABLE ... SELECT for base tables. We try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. We still use old approach for creation of temporary tables. Note that we have separate fix for 5.0 since there we use slightly different less intrusive approach. mysql-test/r/create.result: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/r/trigger.result: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" mysql-test/t/create.test: Extended test coverage for CREATE TABLE ... SELECT. In particular added tests for bug #24508 "Inconsistent results of CREATE TABLE ... SELECT when temporary table exists" and bug #20662 "Infinite loop in CREATE TABLE IF NOT EXISTS ... SELECT with locked tables". mysql-test/t/trigger.test: Added test case for bug #20903 "Crash when using CREATE TABLE .. SELECT and triggers" sql/lock.cc: Now for creation of name-lock placeholder lock_table_name() uses auxiliary function table_cache_insert_placeholder(). sql/mysql_priv.h: Removed declaration of non-existing build_table_path() routine. The former mysql_create_table_internal() was renamed to mysql_create_table_no_lock() and now exposed to other modules to give them opportunity of creation of tables in cases when name-lock is already obtained. reopen_name_locked_table() now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added declaration of auxiliary function table_cache_insert_placeholder() which is used for creation of table placeholders for name-locking. Added declaration of lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. Changed signature of unlink_open_table() function to simplify its use and make it useful for table placeholders and tables that are only open. Added auxiliary drop_open_table() routine. Moved declaration of refresh_version to table.h header to make it accessible from inline methods of TABLE class. MYSQL_OPEN_IGNORE_LOCKED_TABLES flag is no longer used. Instead MYSQL_OPEN_TEMPORARY_ONLY option was added. sql/sql_base.cc: Added support for the new approach to the handling of CREATE TABLE ... SELECT for base tables. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. Note significant distinctions of this placeholder from the placeholder used for normal name-lock: 1) It is treated like open table by other name-locks so it does not allow name-lock taking operations like DROP TABLE or RENAME TABLE to proceed. 2) it is linked into THD::open_tables list and automatically removed during close_thread_tables() call open_tables(): Implemented logic described above. To do this added auxiliary check_if_table_exists() function. Removed support for MYSQL_OPEN_IGNORE_LOCKED_TABLES option which is no longer used. Added MYSQL_OPEN_TEMPORARY_ONLY which is used to restrict search for temporary tables only. close_cached_tables()/close_thread_table()/reopen_tables()/ close_old_data_files()/table_is_used()/remove_table_from_cache(): Added support for open placeholders (note that we also use them when we need to re-open tables during flush). unlink_open_table(): Changed function signature to simplify its use and to make useful for open placeholders and tables which are only open and not locked. Added auxiliary drop_open_table() routine. reopen_name_locked_table(): Now has 3rd argument which controls linking in of table being opened into THD::open_tables (this is useful in cases when placeholder used for name-locking is already linked into this list). Added auxiliary table_cache_insert_placeholder() routine which simplifies creation of placeholders used for name-locking. Added lock_table_name_if_not_cached() which can be used to take an exclusive name-lock on table if there are no records for it in table cache. sql/sql_handler.cc: Adjusted mysql_ha_mark_tables_for_reopen() routine to properly handle placeholders which now can be linked into open tables list. sql/sql_insert.cc: Introduced new approach to handling of base tables in CREATE TABLE ... SELECT statement. Now we try to open and lock table to be created at the same time as the rest of tables used by this statement. If such table does not exist at this moment we create and place in the table cache special placeholder for it which prevents its creation or any other usage by other threads. By doing this we avoid races which existed with previous approach in which we created, opened and locked target in separate step without any special protection. This also allows properly calculate prelocking set in cases when target table already exists and has some on insert triggers. Note that we don't employ the same approach for temporary tables (this is okay as such tables are unaffected by other threads). Changed create_table_from_items() and methods of select_create class to implement this approach. sql/sql_parse.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_prepare.cc: The new approach to handling of CREATE TABLE ... SELECT for base tables assumes that all tables (including table to be created) are opened and (or) locked at the same time. So in cases when we create base table we have to pass to open_and_lock_tables() table list which includes target table. sql/sql_table.cc: Changed mysql_create_table(), mysql_create_like_table() and mysql_alter_table() (in rename case) to obtain exclusive name-lock on the non-temporary table which is going to be created (to which we going to rename). This ensures that not only destination table doesn't exist on disk but also that there are no placeholder in table cache for it (i.e. there is no CREATE TABLE ... SELECT operation in progress for it). Note that to avoid deadlocks while taking these name-locks this code assumes that existence of any record for table in table cache (even name-lock) means that table exists. Altough such check can lead to false positives these should occur only in case of highly concurrent DDL operations on the table and should not break binary logging. Renamed mysql_create_table_internal() to mysql_create_table_no_lock() and made it accessible from other files to give them ability to create table in situation when name-lock is already obtained or not relevant. Adjusted calls to reopen_name_locked_table(), which now takes extra argument, which controls linking of open table into THD::open_tables list. Removed redundant setting of table's 'version' field before calls to close_cached_table(). This function will set it to 0 itself anyway. sql/sql_trigger.cc: reopen_name_locked_tables() now has one more argument which controls linking of opened table into the THD::open_tables list. sql/sql_yacc.yy: The new approach to handling of CREATE TABLE ... SELECT statement for base tables assumes that all tables including table to be created are open and (or) locked at the same time. Therefore we need to set correct lock for target table. sql/table.h: Moved declaration of refresh_version variable from mysql_priv.h to make it accessible from inline methods of TABLE class. Renamed TABLE::locked_by_flush member to open_placeholder since now it is also used for taking exclusive name-lock and not only by flush. Introduced TABLE::is_name_opened() helper method which can be used to distinguish TABLE instances corresponding to open tables or placeholders for them from closed instances (e.g. due to their old version). Also introduced TABLE::needs_reopen_or_name_lock() helper which allows to check if TABLE instance corresponds to outdated version of table or to name-lock placeholder. Introduced TABLE_LIST::create member which marks elements of table list corresponds to the table to be created. Adjusted TABLE_LIST::placeholder() method to take into account name-lock placeholders for tables to be created (this, for example, allows to properly handle such placeholders in lock_tables()). Finally, moved currently unused TABLE::open_next/open_prev members under ifdef NOT_YET. mysql-test/r/create_select-big.result: New BitKeeper file ``mysql-test/r/create_select-big.result'' mysql-test/t/create_select-big.test: New BitKeeper file ``mysql-test/t/create_select-big.test''
2007-05-11 19:51:03 +02:00
select @a;
select * from t1;
drop table t1;
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# Tests for possible concurrency issues with CREATE TABLE ... LIKE
#
# Bug #18950 "create table like does not obtain LOCK_open"
# Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other
# connections"
#
# Again the idea of this test is that we introduce artificial delays on
# various stages of table creation and check that concurrent statements
# for tables from CREATE TABLE ... LIKE are not interfering.
--disable_warnings
drop table if exists t1,t2;
--enable_warnings
set debug_sync='RESET';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# What happens if some statements sneak in right after we have
# acquired locks and opened source table ?
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
create table t1 (i int);
set debug_sync='create_table_like_after_open SIGNAL parked WAIT_FOR go';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# Reset binlog to have clear start
reset master;
--send create table t2 like t1;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# DML on source table should be allowed to run concurrently
insert into t1 values (1);
# And DDL should wait
--send drop table t1;
connection addconroot2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "drop table t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
show create table t2;
drop table t2;
# Let us check that statements were executed/binlogged in correct order
Bug #49741 test files contain explicit references to bin/relay-log positions Some of the test cases reference to binlog position and these position numbers are written into result explicitly. It is difficult to maintain if log event format changes. There are a couple of cases explicit position number appears, we handle them in different ways A. 'CHANGE MASTER ...' with MASTER_LOG_POS or/and RELAY_LOG_POS options Use --replace_result to mask them. B. 'SHOW BINLOG EVENT ...' Replaced by show_binlog_events.inc or wait_for_binlog_event.inc. show_binlog_events.inc file's function is enhanced by given $binlog_file and $binlog_limit. C. 'SHOW SLAVE STATUS', 'show_slave_status.inc' and 'show_slave_status2.inc' For the test cases just care a few items in the result of 'SHOW SLAVE STATUS', only the items related to each test case are showed. 'show_slave_status.inc' is rebuild, only the given items in $status_items will be showed. 'check_slave_is_running.inc' and 'check_slave_no_error.inc' and 'check_slave_param.inc' are auxiliary files helping to show running status and error information easily. mysql-test/extra/binlog_tests/binlog.test: It only cares whether current binlog file index is changed, so it is ok with 'show_master_status.inc' instead of 'show mater status'. mysql-test/extra/binlog_tests/blackhole.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/extra/rpl_tests/rpl_deadlock.test: Use 'check_slave_is_running.inc' instead of 'show_slave_status2.inc'. mysql-test/extra/rpl_tests/rpl_extraMaster_Col.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/extra/rpl_tests/rpl_extraSlave_Col.test: Use 'wait_for_slave_sql_error.inc' and 'ait_for_slave_sql_error_and_skip.inc' instead of 'show slave status'. mysql-test/extra/rpl_tests/rpl_flsh_tbls.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/extra/rpl_tests/rpl_get_master_version_and_clock.test: It is need now to give a error number, so use 'wait_for_slave_io_to_stop.inc' instead of 'wait_for_slave_io_error.inc'. mysql-test/extra/rpl_tests/rpl_insert_delayed.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/extra/rpl_tests/rpl_log.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. se 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/extra/rpl_tests/rpl_max_relay_size.test: se 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/extra/rpl_tests/rpl_ndb_apply_status.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/extra/rpl_tests/rpl_reset_slave.test: Use 'show_slave_status.inc' instead of 'show_slave_status2.inc' statement. Use 'check_slave_no_error.inc' to simplify the check that there is no error. mysql-test/extra/rpl_tests/rpl_row_basic.test: Use 'check_slave_is_running.inc' to verify that Slave threads are running well. Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. mysql-test/extra/rpl_tests/rpl_row_tabledefs.test: Use 'show_slave_error_status_and_skip.inc' instead of 'show slave status'. mysql-test/include/check_slave_is_running.inc: To make sure both sql and io thread are running well. If not, the test will be aborted. mysql-test/include/check_slave_no_error.inc: To make sure both sql and io thread have no error. If not, the test will be aborted. mysql-test/include/get_relay_log_pos.inc: According to the position of a log event in master binlog file, find the peer position of a log event in relay log file. mysql-test/include/rpl_stmt_seq.inc: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/include/show_binlog_events.inc: Add two options $binlog_file and $binlog_limit for showing binlog events from different binlog files or/and given different limits on position or row number. mysql-test/include/show_rpl_debug_info.inc: Add 'SELECT NOW()' in the debug information. mysql-test/include/show_slave_status.inc: It's more clean and tidy Only the given columns of slave status are printed. mysql-test/include/test_fieldsize.inc: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave stutus'. mysql-test/include/wait_for_binlog_event.inc: Use show_rpl_debug_info.inc instead of 'SHOW BINLOG EVENTS'. mysql-test/include/wait_for_slave_io_error.inc: Add $slave_io_errno and $show_slave_io_error, it waits only a given error. mysql-test/include/wait_for_slave_param.inc: Use die instead of exit. mysql-test/include/wait_for_slave_sql_error.inc: Add $slave_sql_errno and $show_slave_sql_error, it waits only a given error. mysql-test/include/wait_for_status_var.inc: Use die instead of exit. mysql-test/r/flush_block_commit_notembedded.result: It checks whether somethings are binlogged, so we using 'show_binlog_event.inc' instead of 'show master status'. mysql-test/r/multi_update.result: It checks whether somethings are binlogged, so we using 'show_binlog_event.inc' instead of 'show master status'. mysql-test/suite/binlog/r/binlog_innodb.result: It checks whether somethings are binlogged, so we using 'show_binlog_event.inc' instead of 'show master status'. mysql-test/suite/binlog/r/binlog_row_binlog.result: Position in the result of 'show master status' is replaced by '#'. mysql-test/suite/binlog/r/binlog_stm_binlog.result: Position in the result of 'show master status' is replaced by '#'. mysql-test/suite/binlog/t/binlog_innodb.test: It checks whether somethings are binlogged, so we use 'show_binlog_event.inc' instead of 'show master status'. mysql-test/suite/binlog/t/binlog_stm_binlog.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/bugs/r/rpl_bug36391.result: Position in the result of 'show master status' is replaced by '#'. mysql-test/suite/bugs/t/rpl_bug12691.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/bugs/t/rpl_bug36391.test: 'show master status' is replaced by 'show_master_status.inc'. Position in the result of 'show master status' is replaced by '#'. mysql-test/suite/engines/funcs/r/rpl_000015.result: It checks whether somethings are binlogged, so we using 'show_binlog_event.inc' instead of 'show master status'. mysql-test/suite/engines/funcs/t/rpl_000015.test: Use 'check_slave_is_running.inc' to verify that Slave threads are running well. mysql-test/suite/engines/funcs/t/rpl_REDIRECT.test: Use 'query_vertical SHOW SLAVE STATUS' instead of 'show slave status'. There is no status columns in the result file, for no slave exists on master's server. mysql-test/suite/engines/funcs/t/rpl_change_master.test: This test just care whether Read_Master_Log_Pos is equal to Exec_Master_Log_Pos after 'CHANGE MASTER ..'. So 'show slave status' is removed and just check the value of Read_Master_Log_Pos and Exec_Master_Log_Pos. mysql-test/suite/engines/funcs/t/rpl_empty_master_crash.test: We doesn't really need the statement. mysql-test/suite/engines/funcs/t/rpl_flushlog_loop.test: Just show Relay_Log_File, running status and error informations. Use 'check_slave_is_running.inc' to verify that Slave threads are running well. mysql-test/suite/engines/funcs/t/rpl_loaddata_s.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/engines/funcs/t/rpl_log_pos.test: Mask the explicit positions in the result file. Use 'check_slave_no_error.inc' to simplify the check that there is no error. Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave stutus'. mysql-test/suite/engines/funcs/t/rpl_rbr_to_sbr.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/engines/funcs/t/rpl_row_drop.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/engines/funcs/t/rpl_row_inexist_tbl.test: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave stutus'. mysql-test/suite/engines/funcs/t/rpl_row_until.test: Use 'check_slave_param.inc' to check whether SQL Thread stop at a right position, and use binlog position variables instead of explicit number in the 'CHANGE MASTER' statements. Mask the explicit binary log positions in the result file. mysql-test/suite/engines/funcs/t/rpl_server_id1.test: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave stutus'. Use 'check_slave_no_error.inc' to simplify the check that there is no error. mysql-test/suite/engines/funcs/t/rpl_server_id2.test: It doesn't really need in this test. mysql-test/suite/engines/funcs/t/rpl_slave_status.test: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave stutus'. mysql-test/suite/engines/funcs/t/rpl_switch_stm_row_mixed.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/manual/t/rpl_replication_delay.test: Use 'show_slave_status.inc' instead of 'show slave status'. mysql-test/suite/parts/t/rpl_partition.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/suite/rpl/include/rpl_mixed_ddl.inc: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/rpl/include/rpl_mixed_dml.inc: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/rpl/t/rpl_000015.test: Use 'show_slave_status.inc' instead of 'show_slave_status2.inc'. mysql-test/suite/rpl/t/rpl_binlog_grant.test: Use 'wait_for_binlog_event.inc' instead of 'show binlog events' statement. mysql-test/suite/rpl/t/rpl_bug33931.test: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave stutus'. mysql-test/suite/rpl/t/rpl_change_master.test: This test just care whether Read_Master_Log_Pos is equal to Exec_Master_Log_Pos after 'CHANGE MASTER ..'. So 'show slave status' is removed and just check the value of Read_Master_Log_Pos and Exec_Master_Log_Pos. mysql-test/suite/rpl/t/rpl_critical_errors.test: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave stutus'. mysql-test/suite/rpl/t/rpl_dual_pos_advance.test: Mask the explicit position numbers in result file. It is restricted running on SBR, for it want to binlog 'set @a=1' statement. mysql-test/suite/rpl/t/rpl_empty_master_crash.test: It doesn't need in this test. mysql-test/suite/rpl/t/rpl_flushlog_loop.test: UUse 'check_slave_is_running.inc' and 'show_slave_status.inc' instead of 'show slave status' statement. mysql-test/suite/rpl/t/rpl_get_master_version_and_clock.test: Use 'wait_for_slave_io_error.inc' to wait the given io thread error happening. mysql-test/suite/rpl/t/rpl_grant.test: It doesn't need in this test. mysql-test/suite/rpl/t/rpl_incident.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/suite/rpl/t/rpl_known_bugs_detection.test: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave stutus'. mysql-test/suite/rpl/t/rpl_loaddata_fatal.test: Use 'wait_for_slave_sql_error_and_skip.inc' to wait the given sql thread error happening and then skip the event. There is no need to print the result of 'show slave stutus'. mysql-test/suite/rpl/t/rpl_log_pos.test: Use 'wait_for_slave_io_error.inc' to wait the given io thread error happening. There is no need to print the result of 'show slave status'. mysql-test/suite/rpl/t/rpl_rbr_to_sbr.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/rpl/t/rpl_replicate_do.test: Use 'show_slave_status.inc' instead of 'show slave status'. mysql-test/suite/rpl/t/rpl_rotate_logs.test: Use 'show_slave_status.inc' instead of 'show_slave_status2.inc'. mysql-test/suite/rpl/t/rpl_row_basic_11bugs.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/rpl/t/rpl_row_create_table.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/rpl/t/rpl_row_drop.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/rpl/t/rpl_row_until.test: Use 'check_slave_param.inc' to check whether SQL Thread stop at a right position, and use binlog position variables instead of explicit number in the 'CHANGE MASTER' statements. mysql-test/suite/rpl/t/rpl_skip_error.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/suite/rpl/t/rpl_slave_load_remove_tmpfile.test: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave status'. mysql-test/suite/rpl/t/rpl_slave_skip.test: Use 'check_slave_param.inc' to check whether SQL Thread stop at a right position, and mask the explicit position number in the 'CHANGE MASTER' statements. mysql-test/suite/rpl/t/rpl_sp.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/suite/rpl/t/rpl_ssl.test: Use 'show_slave_status.inc' instead of 'show slave status'. mysql-test/suite/rpl/t/rpl_ssl1.test: Use 'show_slave_status.inc' instead of 'show slave status'. mysql-test/suite/rpl/t/rpl_stm_until.test: Use 'check_slave_param.inc' to check whether SQL Thread stop at a right position, and use binlog position variables instead of explicit number in the 'CHANGE MASTER' statements. mysql-test/suite/rpl/t/rpl_temporary_errors.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/suite/rpl_ndb/t/rpl_ndb_basic.test: Use 'wait_for_slave_sql_error.inc' to wait the given sql thread error happening. There is no need to print the result of 'show slave status'. mysql-test/suite/rpl_ndb/t/rpl_ndb_circular.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/suite/rpl_ndb/t/rpl_ndb_circular_simplex.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/suite/rpl_ndb/t/rpl_ndb_idempotent.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/suite/rpl_ndb/t/rpl_ndb_multi.test: Mask master_log_pos and master_log_file mysql-test/suite/rpl_ndb/t/rpl_ndb_sync.test: Use 'check_slave_is_running.inc' instead of 'show slave status' statement. mysql-test/suite/rpl_ndb/t/rpl_truncate_7ndb.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/t/alter_table-big.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/t/create-big.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/t/ctype_cp932_binlog_stm.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement. mysql-test/t/flush_block_commit_notembedded.test: It checks whether somethings are binlogged, so we using 'show_binlog_event.inc' instead of 'show master status'. mysql-test/t/multi_update.test: It checks whether somethings are binlogged, so we using 'wait_binlog_event.inc' instead of 'show master status'. mysql-test/t/sp_trans_log.test: Use 'show_binlog_events.inc' instead of 'show binlog events' statement.
2010-05-24 15:54:08 +02:00
source include/show_binlog_events.inc;
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
# Now check the gap between table creation and binlogging
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
create table t1 (i int);
set debug_sync='create_table_like_before_binlog SIGNAL parked WAIT_FOR go';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
reset master;
--send create table t2 like t1;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send insert into t2 values (1);
connection addconroot2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "insert into t2 values (1)";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
drop table t2;
set debug_sync='create_table_like_before_binlog SIGNAL parked WAIT_FOR go';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
--send create table t2 like t1;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send drop table t2;
connection addconroot2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "drop table t2";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
set debug_sync='create_table_like_before_binlog SIGNAL parked WAIT_FOR go';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
--send create table t2 like t1;
connection addconroot1;
set debug_sync='now WAIT_FOR parked';
--send drop table t1;
connection addconroot2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock" and
info = "drop table t1";
--source include/wait_condition.inc
set debug_sync='now SIGNAL go';
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
connection default;
--reap
connection addconroot1;
--reap
connection default;
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
drop table t2;
disconnect addconroot1;
disconnect addconroot2;
disconnect addconroot3;
5.1 version of fix for: Bug #23667 "CREATE TABLE LIKE is not isolated from alteration by other connections" Bug #18950 "CREATE TABLE LIKE does not obtain LOCK_open" As well as: Bug #25578 "CREATE TABLE LIKE does not require any privileges on source table". The first and the second bugs resulted in various errors and wrong binary log order when one tried to execute concurrently CREATE TABLE LIKE statement and DDL statements on source table or DML/DDL statements on its target table. The problem was caused by incomplete protection/table-locking against concurrent statements implemented in mysql_create_like_table() routine. We solve it by simply implementing such protection in proper way. Most of actual work for 5.1 was already done by fix for bug 20662 and preliminary patch changing locking in ALTER TABLE. The third bug allowed user who didn't have any privileges on table create its copy and therefore circumvent privilege check for SHOW CREATE TABLE. This patch solves this problem by adding privilege check, which was missing. Finally it also removes some duplicated code from mysql_create_like_table() and thus fixes bug #26869 "TABLE_LIST::table_name_length inconsistent with TABLE_LIST::table_name". mysql-test/r/create-big.result: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/r/create.result: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/r/grant2.result: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". mysql-test/t/create-big.test: Added test coverage for concurrency-related issues with CREATE TABLE LIKE. mysql-test/t/create.test: Adjusted error-code in the test case after refactoring code that implements CREATE TABLE ... LIKE. mysql-test/t/disabled.def: Recent code changes ensured that CREATE TABLE LIKE statement is properly isolated against other statements, so synchronization.test should no longer fail (see fix for bug 20662 and preliminary patch for bug 23667 changing ALTER TABLE locking). mysql-test/t/grant2.test: Added test for bug#25578 "CREATE TABLE LIKE does not require any privileges on source table". sql/handler.h: Introduced new flag for HA_CREATE_INFO::options in order to be able to distinguish CREATE TABLE ... LIKE from other types of CREATE TABLE. sql/mysql_priv.h: mysql_create_like_table() now takes source table name not as a Table_ident object but as regular table list element. sql/sql_lex.h: Removed LEX::like_name member. Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_parse.cc: CREATE TABLE ... LIKE implementation now uses statement's table list for storing information about the source table. We also use flag in LEX::create_info.options for distinguishing it from other types of CREATE TABLE. Finally CREATE TABLE ... LIKE now requires the same privileges on the source tables as SHOW CREATE TABLE. Moved this privilege check to check_show_create_table_access() function. sql/sql_partition.cc: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list. sql/sql_table.cc: mysql_create_like_table(): - Commented and cleaned-up a bit code which is responsible for achieving isolation from concurrent statements. Most of actual work was done by fix for bug 20662 and preliminary patch changing locking locking in ALTER TABLE, so here we do minor things like relaxing locking on source table (we don't need lock on it, to have it open is enough) and adjusting code to make it more friendly against code implementing I_S. - Get rid of duplicated code related to source database/table name handling. All these operations are already done in st_select_lex::add_table_to_list(), so we achieve the same effect by including source table into the statement's table list. sql/sql_yacc.yy: Now we use special flag in LEX::create_info::options for distinguishing CREATE TABLE ... LIKE from other types of CREATE TABLE and store name of source table as regular element in statement's table list.
2007-05-23 13:26:16 +02:00
set debug_sync='RESET';
source include/show_binlog_events.inc;
# Check that all connections opened by test cases in this file are really
# gone so execution of other tests won't be affected by their presence.
--source include/wait_until_count_sessions.inc