Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
mysqld.cc, ha_innodb.cc:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
sql/ha_innodb.cc:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
sql/mysqld.cc:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
innobase/include/srv0start.h:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
innobase/include/ut0dbg.h:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
innobase/ut/ut0dbg.c:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
innobase/ut/ut0mem.c:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
innobase/srv/srv0start.c:
Changes for NetWare to exit the InnoDB gracefully instead of crashing the server (patch by PRam@novell.com, polished a little by Heikki Tuuri)
Adding a compilation define so that on recent Darwin mysqld does not react to SIGHUP and SIGQUIT:
this fixes a rpl000001 problem on our Powermac G5 machine (popping after an upgrade from 10.3.4 to
10.3.5) and is expected to fix
BUG#2030 "relay bin log on slave resets multiple times a second"
(i.e. under some Mac OS X Panther versions, mysqld receives many SIGHUP and SIGQUIT).
So this fix is more a problem-hider than a real understanding of why mysqld receives so many signals.
Note that we saw other problems on this OS where mysqld reacts to Ctrl-Z but apparently only once,
where using SSL seems to make the problem more frequent...
configure.in:
Escaping [] so that darwin[7-8] gets into 'configure' and not darwin7-8
Adding a compilation define so that on recent Darwin mysqld does not react to SIGHUP and SIGQUIT.
sql/mysqld.cc:
optionally ignore SIGHUP and SIGQUIT, for some Mac OS X Panther versions which send too many of those.
from the select list and TEXT field
make setup_copy_fields to insert Item_copy_string for blobs in
the beginning of the copy_funcs (push_back instead of push_front)
the thing is that Item_copy_string::copy for function can call
Item_copy_string::val_int for blob via Item_ref.
But if Item_copy_string::copy for blob isn't called before,
it's value will be wrong.
So all the Item_copy_string::copy for blobs should be called before
Item_copy_string::copy for functions.
mysql-test/r/having.result:
added test case for
Bug #4358 Problem with HAVING clause that uses
alias from the select list and TEXT field
mysql-test/t/having.test:
added test case for
Bug #4358 Problem with HAVING clause that uses
alias from the select list and TEXT field
sql/sql_select.cc:
make setup_copy_fields to insert Item_copy_string for blobs in
the beginning of the copy_funcs (push_back instead of push_front)
the thing is that Item_copy_string::copy for function can call
Item_copy_string::val_int for blob via Item_ref.
But if Item_copy_string::copy for blob isn't called before,
it's value will be wrong.
So all the Item_copy_string::copy for blobs should be called before
Item_copy_string::copy for functions.
fixed
Bug #4358 Problem with HAVING clause that
uses alias from the select list and TEXT field
Fix for server hang bug
mysql-test/r/heap.result:
Bug#4411
Test for server hang bug
mysql-test/t/heap.test:
Bug#4411
Test for server hang bug
sql/opt_sum.cc:
Bug#4411
Allow code to exit the loop.
BitKeeper/etc/logging_ok:
Logging to logging@openlogging.org accepted
sql/ha_heap.cc:
added calling fn_format(name,..) for name before heap_delete_table
as it's done before heap_create
fixed Bug #4973 Memory is not released when HEAP table is dropped
innobase/include/os0file.h:
Improve the comment of os_file_create_tmpfile()
innobase/os/os0file.c:
os_file_create_tmpfile(): Use create_temp_file()
via innobase_mysql_tmpfile() unless UNIV_HOTBACKUP is defined
sql/ha_innodb.cc:
Added innobase_mysql_tmpfile(), a wrapper around create_temp_file()
Solved performance problems by enabling write buffer.
sql/sql_select.cc:
bug#4497 - Serious regression if disk based TMP table is used.
Solved performance problems by enabling write buffer.
Supplied no test case, as it required too much data to see the performance regression.
InnoDB: Implement tmpfile() differently on Windows (Bug #3998)
innobase/dict/dict0dict.c:
Check the return value of os_file_create_tmpfile(),
as it can now return NULL
innobase/include/os0file.h:
Note that os_file_create_tmpfile() can now return NULL
innobase/include/srv0srv.h:
Add a new server flag (srv_innodb_status) to disable
the creation of innodb_status.<pid> files
innobase/lock/lock0lock.c:
Check the return value of os_file_create_tmpfile(),
as it can now return NULL
innobase/os/os0file.c:
os_file_create_tmpfile(): separate implementation for Win32;
errors will be reported but will not cause assertion failure
innobase/srv/srv0srv.c:
Add a new server flag (srv_innodb_status) to disable
the creation of innodb_status.<pid> files
innobase/srv/srv0start.c:
innobase_start_or_create_for_mysql(): create srv_monitor_file
with tmpfile() or with a visible name "innodb_status.<pid>",
depending on the setting of the flag srv_innodb_status.
sql/ha_innodb.cc:
innobase_init(): initialize srv_innodb_status
update_table_comment(), get_foreign_key_create_info(): replace
tmpfile() with os_file_create_tmpfile()
sql/ha_innodb.h:
Add new Boolean flag, innobase_create_status_file.
sql/mysqld.cc:
Add new Boolean flag, innodb_status_file
We should allow 19 as length of newly created TIMESTAMP fields.
mysql-test/r/type_timestamp.result:
Added test of TIMESTAMP(19) support.
mysql-test/t/type_timestamp.test:
Added test of TIMESTAMP(19) support.
sql/sql_parse.cc:
add_field_to_list(): TIMESTAMP columns should also support 19 as length since it is
length of 4.1 compatible representation.
BitKeeper/etc/logging_ok:
Logging to logging@openlogging.org accepted
sql/share/czech/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/danish/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/dutch/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/english/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/estonian/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/french/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/german/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/greek/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/hungarian/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/japanese/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/korean/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/norwegian-ny/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/norwegian/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/polish/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/portuguese/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/romanian/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/russian/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/slovak/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/spanish/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/swedish/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
sql/share/ukrainian/errmsg.txt:
Remove trailing \n from ER_READY (it is now printed in mysqld.cc)
Bug #4810 "deadlock with KILL when the victim was in a wait state"
(I included mutex unlock into exit_cond() for future safety)
and BUG#4827 "KILL while START SLAVE may lead to replication slave crash"
sql/lock.cc:
we did exit_cond() before unlock(LOCK_open), which led to deadlocks with THD::awake(). Fixing this.
sql/log.cc:
mutex unlock is now included in exit_cond()
sql/repl_failsafe.cc:
we did exit_cond() before unlock(LOCK_rpl_status), which led to deadlocks with THD::awake(). Fixing this.
sql/slave.cc:
we did exit_cond() before unlock(cond_lock), which led to deadlocks with THD::awake(). Fixing this.
Fixing also that if killed while waiting for slave thread to start, we don't release the mutex
(that caused a double release of the mutex => crash).
sql/sql_class.h:
comments about exit_cond()/enter_cond().
Mutex unlock is now included in exit_cond() so that it's always done in the good order.
sql/sql_table.cc:
unlock is now included in exit_cond().
sql/sql_class.h:
removing the assertion as I don't want to include assert.h (inclusion of assert.h has already been
reworked in 4.1, so I'll move the assertion to 4.1). Assertion makes compilation fail if non-debug.
happen on SMP machines, when a thread is going to wait on a condition and it is KILLed at the
same time.
Cleaning code a bit by adding a test in enter_cond() that we have the mutex (was already the case in all places
where it's called except one which is fixed here).
sql/log.cc:
safe_mutex_assert_owner() is now in THD::enter_cond()
sql/slave.cc:
lock mutex before waiting on condition.
sql/sql_class.cc:
THD::awake(): before locking the mutex, let's test it's not zero;
in theory indeed, the killer thread may see current_cond non-zero and current_mutex zero
(order of assignments is not guaranteed by POSIX).
A comment noting that there is still a small chance a KILL does not work and needs
being re-issued.
sql/sql_class.h:
Assert in enter_cond() that we have the mutex.
It is already the case in all places where we call enter_cond(), so better ensure it there.
The problem was that (for any storage engine), the created temporary table was not removed if CREATE SELECT failed (because
of a constraint violation for example). This was not consistent with the manual and with CREATE SELECT (no TEMPORARY).
sql/sql_insert.cc:
Fix for BUG#4551 "Temporary InnoDB tables not replicated properly with CREATE TABLE .. SELECT"
The problem was that (for any storage engine), the created temporary table was not removed if CREATE SELECT failed (because
of a constraint violation for example). This was not consistent with the manual and with CREATE SELECT (no TEMPORARY).
And it led to the above bug, because the binlogging of CREATE SELECT is done by select_insert::send_eof() (same function
as INSERT SELECT) and so, if the table is transactional and there is a failure, the statement is considered as rolled
back and so nothing is written in the binlog. So temp table MUST be deleted.
BUG#4506 "mysqlbinlog --position --read-from-remote-server has wrong "# at" lines",
BUG#4553 "Multi-table DROP TABLE replicates improperly for nonexistent table" with a test file.
It was not possible to add a test for BUG#4506 as in the test suite we must use --short-form
which does not display the "# at" lines.
client/mysqlbinlog.cc:
Fix for BUG#4506 "mysqlbinlog --position --read-from-remote-server has wrong "# at" lines"
when reading a remote binlog, the start position is not always BIN_LOG_HEADER_SIZE (4).
sql/sql_table.cc:
Fix for BUG#4553 "Multi-table DROP TABLE replicates improperly for nonexistent table"
we must my_error() _before_ we write to the binlog, so that a meaningful error code is available
in thd->net.last_errno for storage of the DROP TABLE statement into the binlog.
sql/sql_load.cc:
A fix for a long standing bug in LOAD DATA .. LOCAL ..'
When the error occurs, a link is broken instead of simply returning
the error message and maintaining the same connection.
innobase/include/lock0lock.h:
Improve comments regarding LOCK_TABLE_EXP
innobase/include/row0mysql.h:
Rename row_unlock_table_for_mysql() to row_unlock_tables_for_mysql()
and improve its comment
innobase/include/trx0trx.h:
Rename n_tables_locked to n_lock_table_exp
innobase/lock/lock0lock.c:
Rename n_tables_locked to n_lock_table_exp
Increment n_lock_table_exp already in lock_table_create()
Replace some ut_ad() assertions with ut_a()
innobase/row/row0mysql.c:
Rename n_tables_locked to n_lock_table_exp
Rename row_unlock_table_for_mysql() to row_unlock_tables_for_mysql()
and improve its comment
innobase/trx/trx0trx.c:
Rename n_tables_locked to n_lock_table_exp
sql/ha_innodb.cc:
Rename n_tables_locked to n_lock_table_exp
Rename row_unlock_table_for_mysql() to row_unlock_tables_for_mysql()
processlist on slave":
we now report in SHOW PROCESSLIST that we are writing to the temp
files or loading the table. When we are writing to the tmp file:
| 3 | system user | | | Connect | 6 | Making temp file /tmp/SQL_LOAD-2-1-2.data |
and when we are actually loading the .data temp file into the table:
| 3 | system user | | test | Connect | 2 | | LOAD DATA INFILE '/tmp/SQL_LOAD-2-1-2.data' INTO TABLE `t` <...> |
sql/log_event.cc:
Replication of LOAD DATA INFILE:
we now report in SHOW PROCESSLIST that we are creating
the temp files or loading the table.
Plus removing a line which had a comment "should not be needed"
and a guarding assertion which we have never heard fail (and logic
says it should not fail).
Redesigned the handler close functions so that they are usable
at different places where waiting for closing tables is done.
mysql-test/r/flush_table.result:
bug#3565 - HANDLER and FLUSH TABLE/TABLES deadlock.
Added the test results.
mysql-test/t/flush_table.test:
bug#3565 - HANDLER and FLUSH TABLE/TABLES deadlock.
Activated old test case.
Added new test cases.
sql/mysql_priv.h:
bug#3565 - HANDLER and FLUSH TABLE/TABLES deadlock.
Made mysql_ha_close() more flexible.
Removed mysql_ha_closeall(), which closed only one table despite its name.
Added mysql_ha_close_list(), which closes the complete list or all tables and does not lock.
Removed a duplicate declaration.
sql/sql_base.cc:
bug#3565 - HANDLER and FLUSH TABLE/TABLES deadlock.
Added proper close calls before some wait points to avoid deadlocks
or infinite loops.
sql/sql_handler.cc:
bug#3565 - HANDLER and FLUSH TABLE/TABLES deadlock.
Redesigned the internal function find_table_ptr_by_name().
It can now suppress locking and tells if the requested table
has been flushed by itself.
Extended mysql_ha_close() so that it can now suppres locking
and error reporting. That way it can be used at more places and the
old function mysql_ha_closeall() is now obsolete.
Added a new function mysql_ha_close_list() which closes a whole list
of HANDLER tables or all HANDLER tables, if the list is NULL.
Furthermore is can close all 'old' (marked for flush) HANDLER tables.
sql/sql_table.cc:
bug#3565 - HANDLER and FLUSH TABLE/TABLES deadlock.
Replaced the obsolte mysql_ha_closeall() by the new mysql_ha_close().
