----------------------------------------------------------
revno: 2630.13.6
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-3288
timestamp: Fri 2008-07-11 20:22:44 +0400
message:
WL#3288, step 1: ensure that the SQL layer always closes an open
cursor (rnd or index read) before closing a handler.
sql/handler.h:
Assert that the read is closed in handler destructor.
sql/sql_select.cc:
Remove JOIN::table which was a piece of redundancy. The problem was
that JOIN::cleanup() works only if JOIN::table is not null,
but JOIN::cleanup also assigns JOIN::table to NULL. This assignment
is apparently there for safety, from the times when we had no support for correlated
subqueries. Indeed, in case of a evaluation of a correlated subquery more than once it led
to JOIN::cleanup doing nothing, and leaving the rnd or index read open.
In do_select(), make sure we call JOIN::join_free() even in case of an
error.
sql/sql_select.h:
Remove JOIN::table, JOIN::all_tables has the same functionality.
----------------------------------------------------------
revno: 2630.13.6
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-3288
timestamp: Fri 2008-07-11 20:22:44 +0400
message:
WL#3288, step 1: ensure that the SQL layer always closes an open
cursor (rnd or index read) before closing a handler.
----------------------------------------------------------
revno: 2630.22.8
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-runtime
timestamp: Sun 2008-08-10 18:49:52 +0400
message:
Get rid of typedef struct for the most commonly used types:
TABLE, TABLE_SHARE, LEX. This simplifies use of tags
and forward declarations.
----------------------------------------------------------
revno: 2630.22.8
committer: Konstantin Osipov <konstantin@mysql.com>
branch nick: mysql-6.0-runtime
timestamp: Sun 2008-08-10 18:49:52 +0400
message:
Get rid of typedef struct for the most commonly used types:
TABLE, TABLE_SHARE, LEX. This simplifies use of tags
and forward declarations.
removed if server_id changes
When MySQL crashes (or a snapshot is taken which simulates
a crash), then it is possible that internal XA
transactions (used to sync the binary log and InnoDB)
can be left in a PREPARED state, whereas they should be
rolled back. This is done when the server_id changes
before the restart occurs.
This patch releases he restriction that the server_id
should be consistent if the XID is to be considerred
valid. The rollback phase should then be able to
clean up all pending XA transactions.
removed if server_id changes
When MySQL crashes (or a snapshot is taken which simulates
a crash), then it is possible that internal XA
transactions (used to sync the binary log and InnoDB)
can be left in a PREPARED state, whereas they should be
rolled back. This is done when the server_id changes
before the restart occurs.
This patch releases he restriction that the server_id
should be consistent if the XID is to be considerred
valid. The rollback phase should then be able to
clean up all pending XA transactions.
When a storage engine failed to initialize before allocated slot number,
the slot number would be 0, and when later finalizing this plugin, it would
accidentally unplug the storage engine currently uses slot 0.
This patch fixed this problem by add a new macro value HA_SLOT_UNDEF to
distinguish undefined slot number from slot 0.
When a storage engine failed to initialize before allocated slot number,
the slot number would be 0, and when later finalizing this plugin, it would
accidentally unplug the storage engine currently uses slot 0.
This patch fixed this problem by add a new macro value HA_SLOT_UNDEF to
distinguish undefined slot number from slot 0.
Despite copying the value of the old table's row type
we don't always have to mark row type as being specified.
Innodb uses this to check if it can do fast ALTER TABLE
or not.
Fixed by correctly flagging the presence of row_type
only when it's actually changed.
Added a test case for 39200.
Despite copying the value of the old table's row type
we don't always have to mark row type as being specified.
Innodb uses this to check if it can do fast ALTER TABLE
or not.
Fixed by correctly flagging the presence of row_type
only when it's actually changed.
Added a test case for 39200.
* Finished Monty and Jani's merge
* Some InnoDB tests still fail (because it's old xtradb code run against
newer testsuite). They are expected to go after mergning with the latest
xtradb.
those keywords do nothing in 5.1 (they are meant for future versions, for example featuring the Maria engine)
so they are here removed from the syntax. Adding those keywords to future versions when needed is:
- WL#5034 "Add TRANSACTIONA=0|1 and PAGE_CHECKSUM=0|1 clauses to CREATE TABLE"
- WL#5037 "New ROW_FORMAT value for CREATE TABLE: PAGE"
mysql-test/r/create.result:
test that syntax is not accepted
mysql-test/t/create.test:
test that syntax is not accepted
sql/handler.cc:
remove ROW_FORMAT=PAGE
sql/handler.h:
Mark unused objects, but I don't remove them by fear of breaking any plugin which includes this file
(see also table.h)
sql/lex.h:
removing syntax
sql/sql_show.cc:
removing output of noise keywords in SHOW CREATE TABLE and INFORMATION_SCHEMA.TABLES
sql/sql_table.cc:
removing TRANSACTIONAL
sql/sql_yacc.yy:
removing syntax
sql/table.cc:
removing TRANSACTIONAL, PAGE_CHECKSUM. Their place in the frm file is not reclaimed,
for compatibility with older 5.1.
sql/table.h:
Mark unused objects, but I don't remove them by fear of breaking any plugin which includes this file
(and there are several engines which use the content TABLE_SHARE and thus rely on a certain binary
layout of this structure).
"CREATE TABLE TRANSACTIONAL PAGE_CHECKSUM ROW_FORMAT=PAGE accepted,
does nothing".
Put back stubs for members of structures that are shared between
sql/ and pluggable storage engines. to not break ABI unnecessarily.
To be NULL-merged into 5.4, where we do break the ABI already.
PAGE_CHECKSUM ROW_FORMAT=PAGE accepted, does nothing"
Remove unused code that would lead to warnings when compiling
sql_yacc.yy.
sql/handler.h:
Remove unused defines.
sql/sql_yacc.yy:
Remove unused grammar.
sql/table.h:
Remove unused TABLE members.
"CREATE TABLE TRANSACTIONAL PAGE_CHECKSUM ROW_FORMAT=PAGE accepted,
does nothing".
Put back stubs for members of structures that are shared between
sql/ and pluggable storage engines. to not break ABI unnecessarily.
To be NULL-merged into 5.4, where we do break the ABI already.
those keywords do nothing in 5.1 (they are meant for future versions, for example featuring the Maria engine)
so they are here removed from the syntax. Adding those keywords to future versions when needed is:
- WL#5034 "Add TRANSACTIONA=0|1 and PAGE_CHECKSUM=0|1 clauses to CREATE TABLE"
- WL#5037 "New ROW_FORMAT value for CREATE TABLE: PAGE"
- Added a handler call (prepare_index_scan()) to inform storage engines that an index scan is about to take place.
- Extended the maximun key parts for an index from 16 to 32
- Extended MyISAM and Maria engines to support up to 32 parts
Added checks for return value from ha_index_init()
include/my_handler.h:
Extended number of key parts for MyISAM and Maria from 16 to 32
include/my_pthread.h:
Ensure we always have 256M of stack.
(Required to be able to handle the current number of keys and key parts in MyISAM)
mysql-test/r/create.result:
Extended to test for 32 key parts
mysql-test/r/myisam.result:
Test that we can create 32 but not 33 key parts
mysql-test/r/ps_1general.result:
Length of ref is now 2048 as we can have more key parts
mysql-test/r/ps_2myisam.result:
Length of ref is now 2048 as we can have more key parts
mysql-test/r/ps_3innodb.result:
Length of ref is now 2048 as we can have more key parts
mysql-test/r/ps_4heap.result:
Length of ref is now 2048 as we can have more key parts
mysql-test/r/ps_5merge.result:
Length of ref is now 2048 as we can have more key parts
mysql-test/suite/maria/r/maria.result:
Max key length is now 1208 bytes
mysql-test/suite/maria/r/maria3.result:
Max key length is now 1208 bytes
mysql-test/suite/maria/r/ps_maria.result:
Max key length is now 1208 byte
mysql-test/t/create.test:
Extended to test for 32 key parts
mysql-test/t/myisam.test:
Test that we can create 32 but not 33 key parts
sql/handler.cc:
Check return value from ha_index_init()
sql/handler.h:
Added a handler call (prepare_index_scan()) to inform storage engines that an index scan is about to take place.
sql/sql_select.cc:
Checks all return values from ha_index_init()
Call prepare_index_scan()) to inform storage engines that an index scan is about to take place.
Fixed indentation
sql/table.cc:
Fixed wrong types for key_length (rest of code assumed this was 32 bit)
sql/unireg.h:
Extended the maximun key parts for an index from 16 to 32
storage/maria/ha_maria.cc:
Don't allocate HA_CHECK on the stack in functions where we call repair() as HA_CHECK is HUGE and will overflow stack
storage/myisam/ha_myisam.cc:
Don't allocate HA_CHECK on the stack in functions where we call repair() as HA_CHECK is HUGE and will overflow stack
storage/myisam/mi_check.c:
Fixed wrong check if value overflow
tests/mysql_client_test.c:
Added fflush() to fix output in case of error
Fixed wrong check of 'ref' length in EXPLAIN
Remove custom DLL loader code from innodb plugin code, use
symbols exported from mysqld.
storage/innodb_plugin/handler/ha_innodb.cc:
Remove a Win32 workaround for current_thd.
The original problem that innodb plugin used
value of TLS variable across DLL boundaries is
solved in MySQL server (current_thd is a function
not TLS variable now)
storage/innodb_plugin/handler/handler0alter.cc:
Remove custom delay loader
storage/innodb_plugin/handler/handler0vars.h:
Remove custom delay loader
storage/innodb_plugin/handler/i_s.cc:
Remove custom delay loader
storage/innodb_plugin/handler/win_delay_loader.cc:
Remove custom delay loader
storage/innodb_plugin/plug.in:
Remove commented out MYSQL_PLUGIN_STATIC,
CMake would not parse that correctly
A backport of fix for "BUG40092 - Storage engine API uses
time_t datatype".
Starting from MSVC C++ 2005 (v8), the default size of time_t
is changed from 32-bit to 64-bit. As the result, the binaries
built with pre-v8 MSVC C++ do not work with the binaries
(storage engine plugins) built with v8 or after (server
crashes).
Fixed storage engine API to use datatype with known size
(ulong) instead of time_t.
sql/handler.h:
Bug#39802 On Windows, 32-bit time_t should be enforced
Change create_time, check_time, update_time
in the ha_statistics and PARTITION_INFO
structures to ulong.
A backport of fix for "BUG40092 - Storage engine API uses
time_t datatype".
Starting from MSVC C++ 2005 (v8), the default size of time_t
is changed from 32-bit to 64-bit. As the result, the binaries
built with pre-v8 MSVC C++ do not work with the binaries
(storage engine plugins) built with v8 or after (server
crashes).
Fixed storage engine API to use datatype with known size
(ulong) instead of time_t.
Problem: in 5.0 'check table for upgrade' doesn't detect
incompatible collation changes made in 5.0.48.
Fix: backport #39585 fix to 5.0
sql/handler.cc:
Fix for bug#40984: backport fix from 39585 into 5.0
- backport of #39585 fix
sql/handler.h:
Fix for bug#40984: backport fix from 39585 into 5.0
- backport of #39585 fix
Problem was that partitioning cached the table flags.
These flags could change due to TRANSACTION LEVEL changes.
Solution was to remove the cache and always return the table flags
from the first partition (if the handler was initialized).
mysql-test/r/partition_innodb_stmt.result:
Bug#39084: Getting intermittent errors with statement-based binary logging
New test result file.
mysql-test/t/partition_innodb_stmt.test:
Bug#39084: Getting intermittent errors with statement-based binary logging
New test file.
sql/ha_partition.cc:
Bug#39084: Getting intermittent errors with statement-based binary logging
Removed m_table_flags, and added m_handler_status.
Added checks that all partitions have the same
table flags.
Moved some variable initializations.
Updated some comments.
Fixed typo initialise -> initialize
Changed HA_EXTTA_NO_READCHECK to do nothing, since it
is only used in ha_open, which is called for every
partition in ::open anyway.
sql/ha_partition.h:
Bug#39084: Getting intermittent errors with statement-based binary logging
Removed m_table_flags, and added m_handler_status.
Always return the first partitions table flags, instead of using
cached table flags.
Added define of enabled/disabled partitioning table flags
Fixed type initialise -> initialize
Updated some comments.
sql/handler.cc:
Bug#39084: Getting intermittent errors with statement-based binary logging
Fixed type initialise -> initialize.
sql/handler.h:
Bug#39084: Getting intermittent errors with statement-based binary logging
Added comment to understand where the cached value is set.
Problem was that partitioning cached the table flags.
These flags could change due to TRANSACTION LEVEL changes.
Solution was to remove the cache and always return the table flags
from the first partition (if the handler was initialized).
upgrade from <=5.0.46 to >=5.0.48
Problem: 'check table .. for upgrade' doesn't detect
incompatible collation changes made in 5.0.48.
Fix: check for incompatible collation changes.
sql/handler.cc:
Fix for bug#39585: innodb and myisam corruption after binary
upgrade from <=5.0.46 to >=5.0.48
- check for incompatible collation changes made in 5.0.48:
bug #29461
latin7_general_ci
latin7_general_cs
latin7_estonian_cs
latin2_hungarian_ci
koi8u_general_ci
cp1251_ukrainian_ci
cp1250_general_ci
bug #29499, bug #27562
ascii_general_ci
upgrade from <=5.0.46 to >=5.0.48
Problem: 'check table .. for upgrade' doesn't detect
incompatible collation changes made in 5.0.48.
Fix: check for incompatible collation changes.
InnoDB Plugin locks table
This is a pre fix update that does the change to the handler api.
This is done since there are already changes in this version,
so the real fix does not need to change the api.
sql/handler.h:
Bug#37453: Dropping/creating index on partitioned table with
InnoDB Plugin locks table
The check for which alter table flags a handler have is done through the
handlerton, which will not work correctly for any partitioned table.
It must be through the handler interface.
To be able to fix this I have to add a virtual function to the handler class.
InnoDB Plugin locks table
This is a pre fix update that does the change to the handler api.
This is done since there are already changes in this version,
so the real fix does not need to change the api.
"Trigger fired multiple times leads to gaps in auto_increment sequence".
The bug was that if a trigger fired multiple times inside a top
statement (for example top-statement is a multi-row INSERT,
and trigger is ON INSERT), and that trigger inserted into an auto_increment
column, then gaps could be observed in the auto_increment sequence,
even if there were no other users of the database (no concurrency).
It was wrong usage of THD::auto_inc_intervals_in_cur_stmt_for_binlog.
Note that the fix changes "class handler", I'll tell the Storage Engine API team.
mysql-test/r/trigger-trans.result:
result; before the bugfix, the sequence was 1,2,4,6,8,10,12...
mysql-test/t/trigger-trans.test:
test for BUG#31612
sql/handler.cc:
See revision comment of handler.h.
As THD::auto_inc_intervals_in_cur_stmt_for_binlog is cumulative
over all trigger invokations by the top statement, the
second invokation of the trigger arrived in handler::update_auto_increment()
with already one interval in
THD::auto_inc_intervals_in_cur_stmt_for_binlog. The method thus
believed it had already reserved one interval for that invokation,
thus reserved a twice larger interval (heuristic when we don't know
how large the interval should be: we grow by powers of two). InnoDB
thus increased its internal per-table auto_increment counter by 2
while only one row was to be inserted. Hence a gap in the sequence.
The fix is to use the new handler::auto_inc_intervals_count.
Note that the trigger's statement knows how many rows it is going
to insert, but provides estimation_rows_to_insert == 0 (see comments
in sql_insert.cc why triggers don't call handler::ha_start_bulk_insert()).
* removing white space at end of line
* we don't need to maintain THD::auto_inc_intervals_in_cur_stmt_for_binlog
if no binlogging or if row-based binlogging. Using auto_inc_intervals_count in
the heuristic makes the heuristic independent of binary logging, which is good.
sql/handler.h:
THD::auto_inc_intervals_in_cur_stmt_for_binlog served
- for binlogging
- as a heuristic when we have no estimation of how many records the
statement will insert.
But the first goal needs to be cumulative over all statements which
form a binlog event, while the second one needs to be attached to each
statement. THD::auto_inc_intervals_in_cur_stmt_for_binlog is cumulative,
leading to BUG#31612. So we introduce handler::auto_inc_intervals_count
for the second goal. See the revision comment of handler.cc.
A smaller issue was that, even when the binlog event was only one
statement (no triggers, no stored functions),
THD::auto_inc_intervals_in_cur_stmt.nb_elements() could be lower than
the number of reserved intervals (fooling the heuristic), because its
append() method collapses two contiguous intervals in one.
Note that as auto_inc_intervals_count is in class 'handler' and not
in class 'THD', it does not need to be handled in
THD::reset|restore_sub_statement_state().
sql/log.cc:
Comment is wrong: if auto_increment is second, in handler::update_auto_increment()
'append' is false and so auto_inc_intervals_in_cur_stmt_for_binlog
is empty, we do not come here.
sql/sql_class.h:
comment
"Trigger fired multiple times leads to gaps in auto_increment sequence".
The bug was that if a trigger fired multiple times inside a top
statement (for example top-statement is a multi-row INSERT,
and trigger is ON INSERT), and that trigger inserted into an auto_increment
column, then gaps could be observed in the auto_increment sequence,
even if there were no other users of the database (no concurrency).
It was wrong usage of THD::auto_inc_intervals_in_cur_stmt_for_binlog.
Note that the fix changes "class handler", I'll tell the Storage Engine API team.
InnoDB Plugin locks table
The fast/on-line add/drop index handler calls was not implemented
whithin the partitioning.
This implements it in the partitioning handler.
Since this is only used by the not included InnoDB plugin, there
is no test case. (Have tested it manually with the plugin, and
it does not allow unique indexes not including partitioning
function, or removal of pk, which in innodb generates a new pk,
which is not in the partitioning function.)
NOTE: This introduces a new handler method, and because of that
changes the storage engine api. (One cannot use a handlerton to
see the capabilities of a table's handler if it is partitioned.
So I added a wrapper function in the handler that defaults to
the handlerton function, which the partitioning handler overrides.
sql/ha_partition.cc:
Bug#37453: Dropping/creating index on partitioned table with
InnoDB Plugin locks table
Added support for fast/on-line add/drop index.
Implemented alter_table_flags as bit-or of the partitioned
hton and the first partitions alter_table_flags.
It is only to forward the calls for the other functions:
check_if_incompatible_data
add_index
prepare_drop_index
final_drop_index
to all parts handler
sql/ha_partition.h:
Bug#37453: Dropping/creating index on partitioned table with
InnoDB Plugin locks table
Added support for fast/on-line add/drop index.
sql/handler.h:
Bug#37453: Dropping/creating index on partitioned table with
InnoDB Plugin locks table
Added the function on handler level, defaulting to use
the handlerton function, but a handler can override it.
Needed for partitioned tables.
NOTE: Change of storage engine api.
sql/sql_partition.cc:
Bug#37453: Dropping/creating index on partitioned table with
InnoDB Plugin locks table
Using the new handler function, instead of the handlerton
function. This works better with the partitioning handler.
sql/sql_table.cc:
Bug#37453: Dropping/creating index on partitioned table with
InnoDB Plugin locks table
Using the new handler function, instead of the handlerton
function. This works better with the partitioning handler.
Also using new process info for 'manage keys' (kind of fix
for bug-37550).
