2006-12-31 01:02:27 +01:00
|
|
|
/* Copyright (C) 2000-2006 MySQL AB
|
2000-08-14 15:27:19 +04:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
|
|
it under the terms of the GNU General Public License as published by
|
2006-12-23 20:17:15 +01:00
|
|
|
the Free Software Foundation; version 2 of the License.
|
2000-08-14 15:27:19 +04:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
This program is distributed in the hope that it will be useful,
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
GNU General Public License for more details.
|
2000-08-14 15:27:19 +04:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
You should have received a copy of the GNU General Public License
|
|
|
|
along with this program; if not, write to the Free Software
|
|
|
|
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
|
|
|
|
|
|
|
|
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
/*
|
|
|
|
MyISAM MERGE tables
|
|
|
|
|
|
|
|
A MyISAM MERGE table is kind of a union of zero or more MyISAM tables.
|
|
|
|
|
|
|
|
Besides the normal form file (.frm) a MERGE table has a meta file
|
|
|
|
(.MRG) with a list of tables. These are paths to the MyISAM table
|
|
|
|
files. The last two components of the path contain the database name
|
|
|
|
and the table name respectively.
|
|
|
|
|
|
|
|
When a MERGE table is open, there exists an TABLE object for the MERGE
|
|
|
|
table itself and a TABLE object for each of the MyISAM tables. For
|
|
|
|
abbreviated writing, I call the MERGE table object "parent" and the
|
|
|
|
MyISAM table objects "children".
|
|
|
|
|
|
|
|
A MERGE table is almost always opened through open_and_lock_tables()
|
|
|
|
and hence through open_tables(). When the parent appears in the list
|
|
|
|
of tables to open, the initial open of the handler does nothing but
|
|
|
|
read the meta file and collect a list of TABLE_LIST objects for the
|
|
|
|
children. This list is attached to the parent TABLE object as
|
|
|
|
TABLE::child_l. The end of the children list is saved in
|
|
|
|
TABLE::child_last_l.
|
|
|
|
|
|
|
|
Back in open_tables(), add_merge_table_list() is called. It updates
|
|
|
|
each list member with the lock type and a back pointer to the parent
|
|
|
|
TABLE_LIST object TABLE_LIST::parent_l. The list is then inserted in
|
|
|
|
the list of tables to open, right behind the parent. Consequently,
|
|
|
|
open_tables() opens the children, one after the other. The TABLE
|
|
|
|
references of the TABLE_LIST objects are implicitly set to the open
|
|
|
|
tables. The children are opened as independent MyISAM tables, right as
|
|
|
|
if they are used by the SQL statement.
|
|
|
|
|
|
|
|
TABLE_LIST::parent_l is required to find the parent 1. when the last
|
|
|
|
child has been opened and children are to be attached, and 2. when an
|
|
|
|
error happens during child open and the child list must be removed
|
|
|
|
from the queuery list. In these cases the current child does not have
|
|
|
|
TABLE::parent set or does not have a TABLE at all respectively.
|
|
|
|
|
|
|
|
When the last child is open, attach_merge_children() is called. It
|
|
|
|
removes the list of children from the open list. Then the children are
|
|
|
|
"attached" to the parent. All required references between parent and
|
|
|
|
children are set up.
|
|
|
|
|
|
|
|
The MERGE storage engine sets up an array with references to the
|
|
|
|
low-level MyISAM table objects (MI_INFO). It remembers the state of
|
|
|
|
the table in MYRG_INFO::children_attached.
|
|
|
|
|
|
|
|
Every child TABLE::parent references the parent TABLE object. That way
|
|
|
|
TABLE objects belonging to a MERGE table can be identified.
|
|
|
|
TABLE::parent is required because the parent and child TABLE objects
|
|
|
|
can live longer than the parent TABLE_LIST object. So the path
|
|
|
|
child->pos_in_table_list->parent_l->table can be broken.
|
|
|
|
|
|
|
|
If necessary, the compatibility of parent and children is checked.
|
|
|
|
This check is necessary when any of the objects are reopend. This is
|
|
|
|
detected by comparing the current table def version against the
|
|
|
|
remembered child def version. On parent open, the list members are
|
|
|
|
initialized to an "impossible"/"undefined" version value. So the check
|
|
|
|
is always executed on the first attach.
|
|
|
|
|
|
|
|
The version check is done in myisammrg_attach_children_callback(),
|
|
|
|
which is called for every child. ha_myisammrg::attach_children()
|
|
|
|
initializes 'need_compat_check' to FALSE and
|
|
|
|
myisammrg_attach_children_callback() sets it ot TRUE if a table
|
|
|
|
def version mismatches the remembered child def version.
|
|
|
|
|
|
|
|
Finally the parent TABLE::children_attached is set.
|
|
|
|
|
|
|
|
---
|
|
|
|
|
|
|
|
On parent open the storage engine structures are allocated and initialized.
|
|
|
|
They stay with the open table until its final close.
|
|
|
|
|
|
|
|
|
|
|
|
*/
|
|
|
|
|
2005-05-26 12:09:14 +02:00
|
|
|
#ifdef USE_PRAGMA_IMPLEMENTATION
|
2000-07-31 21:29:14 +02:00
|
|
|
#pragma implementation // gcc: Class implementation
|
|
|
|
#endif
|
|
|
|
|
2006-08-19 10:23:22 -07:00
|
|
|
#define MYSQL_SERVER 1
|
2000-07-31 21:29:14 +02:00
|
|
|
#include "mysql_priv.h"
|
2006-08-19 10:23:22 -07:00
|
|
|
#include <mysql/plugin.h>
|
2000-07-31 21:29:14 +02:00
|
|
|
#include <m_ctype.h>
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
#include "../myisam/ha_myisam.h"
|
2000-07-31 21:29:14 +02:00
|
|
|
#include "ha_myisammrg.h"
|
2006-08-19 10:23:22 -07:00
|
|
|
#include "myrg_def.h"
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2006-04-13 13:49:29 -07:00
|
|
|
|
2006-09-29 17:19:02 -07:00
|
|
|
static handler *myisammrg_create_handler(handlerton *hton,
|
|
|
|
TABLE_SHARE *table,
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
MEM_ROOT *mem_root)
|
2005-11-07 16:25:06 +01:00
|
|
|
{
|
2006-09-29 17:19:02 -07:00
|
|
|
return new (mem_root) ha_myisammrg(hton, table);
|
2005-11-07 16:25:06 +01:00
|
|
|
}
|
|
|
|
|
2005-07-19 22:21:12 +04:00
|
|
|
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
/**
|
|
|
|
@brief Constructor
|
|
|
|
*/
|
|
|
|
|
2006-09-29 17:19:02 -07:00
|
|
|
ha_myisammrg::ha_myisammrg(handlerton *hton, TABLE_SHARE *table_arg)
|
|
|
|
:handler(hton, table_arg), file(0)
|
2005-07-19 22:21:12 +04:00
|
|
|
{}
|
|
|
|
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
|
|
|
|
/**
|
|
|
|
@brief Destructor
|
|
|
|
*/
|
|
|
|
|
|
|
|
ha_myisammrg::~ha_myisammrg(void)
|
|
|
|
{}
|
|
|
|
|
|
|
|
|
2005-04-27 11:25:08 +02:00
|
|
|
static const char *ha_myisammrg_exts[] = {
|
|
|
|
".MRG",
|
|
|
|
NullS
|
|
|
|
};
|
2007-01-31 16:15:20 +04:00
|
|
|
extern int table2myisam(TABLE *table_arg, MI_KEYDEF **keydef_out,
|
|
|
|
MI_COLUMNDEF **recinfo_out, uint *records_out);
|
|
|
|
extern int check_definition(MI_KEYDEF *t1_keyinfo, MI_COLUMNDEF *t1_recinfo,
|
|
|
|
uint t1_keys, uint t1_recs,
|
|
|
|
MI_KEYDEF *t2_keyinfo, MI_COLUMNDEF *t2_recinfo,
|
|
|
|
uint t2_keys, uint t2_recs, bool strict);
|
2007-06-14 16:18:01 +05:00
|
|
|
static void split_file_name(const char *file_name,
|
|
|
|
LEX_STRING *db, LEX_STRING *name);
|
|
|
|
|
|
|
|
|
2007-06-06 04:42:41 +05:00
|
|
|
extern "C" void myrg_print_wrong_table(const char *table_name)
|
|
|
|
{
|
2007-06-14 16:18:01 +05:00
|
|
|
LEX_STRING db, name;
|
|
|
|
char buf[FN_REFLEN];
|
|
|
|
split_file_name(table_name, &db, &name);
|
|
|
|
memcpy(buf, db.str, db.length);
|
|
|
|
buf[db.length]= '.';
|
|
|
|
memcpy(buf + db.length + 1, name.str, name.length);
|
|
|
|
buf[db.length + name.length + 1]= 0;
|
2007-06-06 04:42:41 +05:00
|
|
|
push_warning_printf(current_thd, MYSQL_ERROR::WARN_LEVEL_ERROR,
|
|
|
|
ER_ADMIN_WRONG_MRG_TABLE, ER(ER_ADMIN_WRONG_MRG_TABLE),
|
2007-06-14 16:18:01 +05:00
|
|
|
buf);
|
2007-06-06 04:42:41 +05:00
|
|
|
}
|
|
|
|
|
2005-04-27 11:25:08 +02:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
const char **ha_myisammrg::bas_ext() const
|
2005-04-27 11:25:08 +02:00
|
|
|
{
|
|
|
|
return ha_myisammrg_exts;
|
|
|
|
}
|
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2004-11-23 22:57:04 +01:00
|
|
|
const char *ha_myisammrg::index_type(uint key_number)
|
|
|
|
{
|
|
|
|
return ((table->key_info[key_number].flags & HA_FULLTEXT) ?
|
|
|
|
"FULLTEXT" :
|
|
|
|
(table->key_info[key_number].flags & HA_SPATIAL) ?
|
|
|
|
"SPATIAL" :
|
|
|
|
(table->key_info[key_number].algorithm == HA_KEY_ALG_RTREE) ?
|
|
|
|
"RTREE" :
|
|
|
|
"BTREE");
|
|
|
|
}
|
|
|
|
|
2003-11-18 13:47:27 +02:00
|
|
|
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
/**
|
|
|
|
@brief Callback function for open of a MERGE parent table.
|
|
|
|
|
|
|
|
@detail This function adds a TABLE_LIST object for a MERGE child table
|
|
|
|
to a list of tables of the parent TABLE object. It is called for
|
|
|
|
each child table.
|
|
|
|
|
|
|
|
The list of child TABLE_LIST objects is kept in the TABLE object of
|
|
|
|
the parent for the whole life time of the MERGE table. It is
|
|
|
|
inserted in the statement list behind the MERGE parent TABLE_LIST
|
|
|
|
object when the MERGE table is opened. It is removed from the
|
|
|
|
statement list after the last child is opened.
|
|
|
|
|
|
|
|
All memeory used for the child TABLE_LIST objects and the strings
|
|
|
|
referred by it are taken from the parent TABLE::mem_root. Thus they
|
|
|
|
are all freed implicitly at the final close of the table.
|
|
|
|
|
|
|
|
TABLE::child_l -> TABLE_LIST::next_global -> TABLE_LIST::next_global
|
|
|
|
# # ^ # ^
|
|
|
|
# # | # |
|
|
|
|
# # +--------- TABLE_LIST::prev_global
|
|
|
|
# # |
|
|
|
|
# |<--- TABLE_LIST::prev_global |
|
|
|
|
# |
|
|
|
|
TABLE::child_last_l -----------------------------------------+
|
|
|
|
|
|
|
|
@param[in] callback_param data pointer as given to myrg_parent_open()
|
|
|
|
@param[in] filename file name of MyISAM table
|
|
|
|
without extension.
|
|
|
|
|
|
|
|
@return status
|
|
|
|
@retval 0 OK
|
|
|
|
@retval != 0 Error
|
|
|
|
*/
|
|
|
|
|
|
|
|
static int myisammrg_parent_open_callback(void *callback_param,
|
|
|
|
const char *filename)
|
|
|
|
{
|
|
|
|
ha_myisammrg *ha_myrg;
|
|
|
|
TABLE *parent;
|
|
|
|
TABLE_LIST *child_l;
|
|
|
|
const char *db;
|
|
|
|
const char *table_name;
|
|
|
|
uint dirlen;
|
|
|
|
char dir_path[FN_REFLEN];
|
|
|
|
DBUG_ENTER("myisammrg_parent_open_callback");
|
|
|
|
|
|
|
|
/* Extract child table name and database name from filename. */
|
|
|
|
dirlen= dirname_length(filename);
|
|
|
|
if (dirlen >= FN_REFLEN)
|
|
|
|
{
|
|
|
|
/* purecov: begin inspected */
|
|
|
|
DBUG_PRINT("error", ("name too long: '%.64s'", filename));
|
|
|
|
my_errno= ENAMETOOLONG;
|
|
|
|
DBUG_RETURN(1);
|
|
|
|
/* purecov: end */
|
|
|
|
}
|
|
|
|
table_name= filename + dirlen;
|
|
|
|
dirlen--; /* Strip off trailing '/'. */
|
|
|
|
memcpy(dir_path, filename, dirlen);
|
|
|
|
dir_path[dirlen]= '\0';
|
|
|
|
db= base_name(dir_path);
|
|
|
|
dirlen-= db - dir_path; /* This is now the length of 'db'. */
|
|
|
|
DBUG_PRINT("myrg", ("open: '%s'.'%s'", db, table_name));
|
|
|
|
|
|
|
|
ha_myrg= (ha_myisammrg*) callback_param;
|
|
|
|
parent= ha_myrg->table_ptr();
|
|
|
|
|
|
|
|
/* Get a TABLE_LIST object. */
|
|
|
|
if (!(child_l= (TABLE_LIST*) alloc_root(&parent->mem_root,
|
|
|
|
sizeof(TABLE_LIST))))
|
|
|
|
{
|
|
|
|
/* purecov: begin inspected */
|
|
|
|
DBUG_PRINT("error", ("my_malloc error: %d", my_errno));
|
|
|
|
DBUG_RETURN(1);
|
|
|
|
/* purecov: end */
|
|
|
|
}
|
|
|
|
bzero((char*) child_l, sizeof(TABLE_LIST));
|
|
|
|
|
|
|
|
/* Set database (schema) name. */
|
|
|
|
child_l->db_length= dirlen;
|
|
|
|
child_l->db= strmake_root(&parent->mem_root, db, dirlen);
|
|
|
|
/* Set table name. */
|
|
|
|
child_l->table_name_length= strlen(table_name);
|
|
|
|
child_l->table_name= strmake_root(&parent->mem_root, table_name,
|
|
|
|
child_l->table_name_length);
|
|
|
|
/* Convert to lowercase if required. */
|
|
|
|
if (lower_case_table_names && child_l->table_name_length)
|
|
|
|
child_l->table_name_length= my_casedn_str(files_charset_info,
|
|
|
|
child_l->table_name);
|
|
|
|
/* Set alias. */
|
|
|
|
child_l->alias= child_l->table_name;
|
|
|
|
|
|
|
|
/* Initialize table map to 'undefined'. */
|
|
|
|
child_l->init_child_def_version();
|
|
|
|
|
|
|
|
/* Link TABLE_LIST object into the parent list. */
|
|
|
|
if (!parent->child_last_l)
|
|
|
|
{
|
|
|
|
/* Initialize parent->child_last_l when handling first child. */
|
|
|
|
parent->child_last_l= &parent->child_l;
|
|
|
|
}
|
|
|
|
*parent->child_last_l= child_l;
|
|
|
|
child_l->prev_global= parent->child_last_l;
|
|
|
|
parent->child_last_l= &child_l->next_global;
|
|
|
|
|
|
|
|
DBUG_RETURN(0);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
@brief Callback function for attaching a MERGE child table.
|
|
|
|
|
|
|
|
@detail This function retrieves the MyISAM table handle from the
|
|
|
|
next child table. It is called for each child table.
|
|
|
|
|
|
|
|
@param[in] callback_param data pointer as given to
|
|
|
|
myrg_attach_children()
|
|
|
|
|
|
|
|
@return pointer to open MyISAM table structure
|
|
|
|
@retval !=NULL OK, returning pointer
|
|
|
|
@retval NULL, my_errno == 0 Ok, no more child tables
|
|
|
|
@retval NULL, my_errno != 0 error
|
|
|
|
*/
|
|
|
|
|
|
|
|
static MI_INFO *myisammrg_attach_children_callback(void *callback_param)
|
|
|
|
{
|
|
|
|
ha_myisammrg *ha_myrg;
|
|
|
|
TABLE *parent;
|
|
|
|
TABLE *child;
|
|
|
|
TABLE_LIST *child_l;
|
|
|
|
MI_INFO *myisam;
|
|
|
|
DBUG_ENTER("myisammrg_attach_children_callback");
|
|
|
|
|
|
|
|
my_errno= 0;
|
|
|
|
ha_myrg= (ha_myisammrg*) callback_param;
|
|
|
|
parent= ha_myrg->table_ptr();
|
|
|
|
|
|
|
|
/* Get child list item. */
|
|
|
|
child_l= ha_myrg->next_child_attach;
|
|
|
|
if (!child_l)
|
|
|
|
{
|
|
|
|
DBUG_PRINT("myrg", ("No more children to attach"));
|
|
|
|
DBUG_RETURN(NULL);
|
|
|
|
}
|
|
|
|
child= child_l->table;
|
|
|
|
DBUG_PRINT("myrg", ("child table: '%s'.'%s' 0x%lx", child->s->db.str,
|
|
|
|
child->s->table_name.str, (long) child));
|
|
|
|
/*
|
|
|
|
Prepare for next child. Used as child_l in next call to this function.
|
|
|
|
We cannot rely on a NULL-terminated chain.
|
|
|
|
*/
|
|
|
|
if (&child_l->next_global == parent->child_last_l)
|
|
|
|
{
|
|
|
|
DBUG_PRINT("myrg", ("attaching last child"));
|
|
|
|
ha_myrg->next_child_attach= NULL;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
ha_myrg->next_child_attach= child_l->next_global;
|
|
|
|
|
|
|
|
/* Set parent reference. */
|
|
|
|
child->parent= parent;
|
|
|
|
|
|
|
|
/*
|
|
|
|
Do a quick compatibility check. The table def version is set when
|
|
|
|
the table share is created. The child def version is copied
|
|
|
|
from the table def version after a sucessful compatibility check.
|
|
|
|
We need to repeat the compatibility check only if a child is opened
|
|
|
|
from a different share than last time it was used with this MERGE
|
|
|
|
table.
|
|
|
|
*/
|
|
|
|
DBUG_PRINT("myrg", ("table_def_version last: %lu current: %lu",
|
|
|
|
(ulong) child_l->get_child_def_version(),
|
|
|
|
(ulong) child->s->get_table_def_version()));
|
|
|
|
if (child_l->get_child_def_version() != child->s->get_table_def_version())
|
|
|
|
ha_myrg->need_compat_check= TRUE;
|
|
|
|
|
|
|
|
/*
|
|
|
|
If parent is temporary, children must be temporary too and vice
|
|
|
|
versa. This check must be done for every child on every open because
|
|
|
|
the table def version can overlap between temporary and
|
|
|
|
non-temporary tables. We need to detect the case where a
|
|
|
|
non-temporary table has been replaced with a temporary table of the
|
|
|
|
same version. Or vice versa. A very unlikely case, but it could
|
|
|
|
happen.
|
|
|
|
*/
|
|
|
|
if (child->s->tmp_table != parent->s->tmp_table)
|
|
|
|
{
|
|
|
|
DBUG_PRINT("error", ("temporary table mismatch parent: %d child: %d",
|
|
|
|
parent->s->tmp_table, child->s->tmp_table));
|
|
|
|
my_errno= HA_ERR_WRONG_MRG_TABLE_DEF;
|
|
|
|
goto err;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Extract the MyISAM table structure pointer from the handler object. */
|
|
|
|
if ((child->file->ht->db_type != DB_TYPE_MYISAM) ||
|
|
|
|
!(myisam= ((ha_myisam*) child->file)->file_ptr()))
|
|
|
|
{
|
|
|
|
DBUG_PRINT("error", ("no MyISAM handle for child table: '%s'.'%s' 0x%lx",
|
|
|
|
child->s->db.str, child->s->table_name.str,
|
|
|
|
(long) child));
|
|
|
|
my_errno= HA_ERR_WRONG_MRG_TABLE_DEF;
|
|
|
|
}
|
|
|
|
DBUG_PRINT("myrg", ("MyISAM handle: 0x%lx my_errno: %d",
|
|
|
|
(long) myisam, my_errno));
|
|
|
|
|
|
|
|
err:
|
|
|
|
DBUG_RETURN(my_errno ? NULL : myisam);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
@brief Open a MERGE parent table, not its children.
|
|
|
|
|
|
|
|
@detail This function initializes the MERGE storage engine structures
|
|
|
|
and adds a child list of TABLE_LIST to the parent TABLE.
|
|
|
|
|
|
|
|
@param[in] name MERGE table path name
|
|
|
|
@param[in] mode read/write mode, unused
|
|
|
|
@param[in] test_if_locked open flags
|
|
|
|
|
|
|
|
@return status
|
|
|
|
@retval 0 OK
|
|
|
|
@retval -1 Error, my_errno gives reason
|
|
|
|
*/
|
|
|
|
|
|
|
|
int ha_myisammrg::open(const char *name, int mode __attribute__((unused)),
|
|
|
|
uint test_if_locked)
|
|
|
|
{
|
|
|
|
DBUG_ENTER("ha_myisammrg::open");
|
|
|
|
DBUG_PRINT("myrg", ("name: '%s' table: 0x%lx", name, (long) table));
|
|
|
|
DBUG_PRINT("myrg", ("test_if_locked: %u", test_if_locked));
|
|
|
|
|
|
|
|
/* Save for later use. */
|
|
|
|
this->test_if_locked= test_if_locked;
|
|
|
|
|
|
|
|
/* retrieve children table list. */
|
|
|
|
my_errno= 0;
|
|
|
|
if (!(file= myrg_parent_open(name, myisammrg_parent_open_callback, this)))
|
|
|
|
{
|
|
|
|
DBUG_PRINT("error", ("my_errno %d", my_errno));
|
|
|
|
DBUG_RETURN(my_errno ? my_errno : -1);
|
|
|
|
}
|
|
|
|
DBUG_PRINT("myrg", ("MYRG_INFO: 0x%lx", (long) file));
|
|
|
|
DBUG_RETURN(0);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
@brief Attach children to a MERGE table.
|
|
|
|
|
|
|
|
@detail Let the storage engine attach its children through a callback
|
|
|
|
function. Check table definitions for consistency.
|
|
|
|
|
|
|
|
@note Special thd->open_options may be in effect. We can make use of
|
|
|
|
them in attach. I.e. we use HA_OPEN_FOR_REPAIR to report the names
|
|
|
|
of mismatching child tables. We cannot transport these options in
|
|
|
|
ha_myisammrg::test_if_locked because they may change after the
|
|
|
|
parent is opened. The parent is kept open in the table cache over
|
|
|
|
multiple statements and can be used by other threads. Open options
|
|
|
|
can change over time.
|
|
|
|
|
|
|
|
@return status
|
|
|
|
@retval 0 OK
|
|
|
|
@retval != 0 Error, my_errno gives reason
|
|
|
|
*/
|
|
|
|
|
|
|
|
int ha_myisammrg::attach_children(void)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
MYRG_TABLE *u_table;
|
|
|
|
MI_COLUMNDEF *recinfo;
|
|
|
|
MI_KEYDEF *keyinfo;
|
|
|
|
uint recs;
|
|
|
|
uint keys= table->s->keys;
|
|
|
|
int error;
|
|
|
|
DBUG_ENTER("ha_myisammrg::attach_children");
|
|
|
|
DBUG_PRINT("myrg", ("table: '%s'.'%s' 0x%lx", table->s->db.str,
|
|
|
|
table->s->table_name.str, (long) table));
|
|
|
|
DBUG_PRINT("myrg", ("test_if_locked: %u", this->test_if_locked));
|
|
|
|
DBUG_ASSERT(!this->file->children_attached);
|
|
|
|
|
|
|
|
/*
|
|
|
|
Initialize variables that are used, modified, and/or set by
|
|
|
|
myisammrg_attach_children_callback().
|
|
|
|
'next_child_attach' traverses the chain of TABLE_LIST objects
|
|
|
|
that has been compiled during myrg_parent_open(). Every call
|
|
|
|
to myisammrg_attach_children_callback() moves the pointer to
|
|
|
|
the next object.
|
|
|
|
'need_compat_check' is set by myisammrg_attach_children_callback()
|
|
|
|
if a child fails the table def version check.
|
|
|
|
'my_errno' is set by myisammrg_attach_children_callback() in
|
|
|
|
case of an error.
|
|
|
|
*/
|
|
|
|
next_child_attach= table->child_l;
|
|
|
|
need_compat_check= FALSE;
|
|
|
|
my_errno= 0;
|
2003-11-18 13:47:27 +02:00
|
|
|
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
if (myrg_attach_children(this->file, this->test_if_locked |
|
|
|
|
current_thd->open_options,
|
|
|
|
myisammrg_attach_children_callback, this))
|
2001-12-02 14:34:01 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_PRINT("error", ("my_errno %d", my_errno));
|
|
|
|
DBUG_RETURN(my_errno ? my_errno : -1);
|
2001-12-02 14:34:01 +02:00
|
|
|
}
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_PRINT("myrg", ("calling myrg_extrafunc"));
|
2002-03-22 22:55:08 +02:00
|
|
|
myrg_extrafunc(file, query_cache_invalidate_by_MyISAM_filename_ref);
|
2000-07-31 21:29:14 +02:00
|
|
|
if (!(test_if_locked == HA_OPEN_WAIT_IF_LOCKED ||
|
|
|
|
test_if_locked == HA_OPEN_ABORT_IF_LOCKED))
|
2002-07-23 18:31:22 +03:00
|
|
|
myrg_extra(file,HA_EXTRA_NO_WAIT_LOCK,0);
|
2000-07-31 21:29:14 +02:00
|
|
|
info(HA_STATUS_NO_LOCK | HA_STATUS_VARIABLE | HA_STATUS_CONST);
|
|
|
|
if (!(test_if_locked & HA_OPEN_WAIT_IF_LOCKED))
|
2002-07-23 18:31:22 +03:00
|
|
|
myrg_extra(file,HA_EXTRA_WAIT_LOCK,0);
|
2002-09-30 14:54:16 +03:00
|
|
|
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
/*
|
|
|
|
The compatibility check is required only if one or more children do
|
|
|
|
not match their table def version from the last check. This will
|
|
|
|
always happen at the first attach because the reference child def
|
|
|
|
version is initialized to 'undefined' at open.
|
|
|
|
*/
|
|
|
|
DBUG_PRINT("myrg", ("need_compat_check: %d", need_compat_check));
|
|
|
|
if (need_compat_check)
|
2007-01-31 16:15:20 +04:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
TABLE_LIST *child_l;
|
|
|
|
|
|
|
|
if (table->s->reclength != stats.mean_rec_length && stats.mean_rec_length)
|
2007-01-31 16:15:20 +04:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_PRINT("error",("reclength: %lu mean_rec_length: %lu",
|
|
|
|
table->s->reclength, stats.mean_rec_length));
|
2007-06-06 04:42:41 +05:00
|
|
|
if (test_if_locked & HA_OPEN_FOR_REPAIR)
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
myrg_print_wrong_table(file->open_tables->table->filename);
|
|
|
|
error= HA_ERR_WRONG_MRG_TABLE_DEF;
|
|
|
|
goto err;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
Both recinfo and keyinfo are allocated by my_multi_malloc(), thus
|
|
|
|
only recinfo must be freed.
|
|
|
|
*/
|
|
|
|
if ((error= table2myisam(table, &keyinfo, &recinfo, &recs)))
|
|
|
|
{
|
|
|
|
/* purecov: begin inspected */
|
|
|
|
DBUG_PRINT("error", ("failed to convert TABLE object to MyISAM "
|
|
|
|
"key and column definition"));
|
|
|
|
goto err;
|
|
|
|
/* purecov: end */
|
|
|
|
}
|
|
|
|
for (u_table= file->open_tables; u_table < file->end_table; u_table++)
|
|
|
|
{
|
|
|
|
if (check_definition(keyinfo, recinfo, keys, recs,
|
|
|
|
u_table->table->s->keyinfo, u_table->table->s->rec,
|
|
|
|
u_table->table->s->base.keys,
|
|
|
|
u_table->table->s->base.fields, false))
|
2007-06-06 04:42:41 +05:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_PRINT("error", ("table definition mismatch: '%s'",
|
|
|
|
u_table->table->filename));
|
|
|
|
error= HA_ERR_WRONG_MRG_TABLE_DEF;
|
|
|
|
if (!(this->test_if_locked & HA_OPEN_FOR_REPAIR))
|
|
|
|
{
|
|
|
|
my_free((uchar*) recinfo, MYF(0));
|
|
|
|
goto err;
|
|
|
|
}
|
|
|
|
myrg_print_wrong_table(u_table->table->filename);
|
2007-06-06 04:42:41 +05:00
|
|
|
}
|
2007-01-31 16:15:20 +04:00
|
|
|
}
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
my_free((uchar*) recinfo, MYF(0));
|
|
|
|
if (error == HA_ERR_WRONG_MRG_TABLE_DEF)
|
|
|
|
goto err;
|
|
|
|
|
|
|
|
/* All checks passed so far. Now update child def version. */
|
|
|
|
for (child_l= table->child_l; ; child_l= child_l->next_global)
|
|
|
|
{
|
|
|
|
child_l->set_child_def_version(
|
|
|
|
child_l->table->s->get_table_def_version());
|
|
|
|
|
|
|
|
if (&child_l->next_global == table->child_last_l)
|
|
|
|
break;
|
|
|
|
}
|
2007-01-31 16:15:20 +04:00
|
|
|
}
|
2002-09-30 14:54:16 +03:00
|
|
|
#if !defined(BIG_TABLES) || SIZEOF_OFF_T == 4
|
|
|
|
/* Merge table has more than 2G rows */
|
2005-01-06 13:00:13 +02:00
|
|
|
if (table->s->crashed)
|
2007-01-31 16:15:20 +04:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_PRINT("error", ("MERGE table marked crashed"));
|
2007-01-31 16:15:20 +04:00
|
|
|
error= HA_ERR_WRONG_MRG_TABLE_DEF;
|
2002-09-30 14:54:16 +03:00
|
|
|
goto err;
|
2007-01-31 16:15:20 +04:00
|
|
|
}
|
2002-09-30 14:54:16 +03:00
|
|
|
#endif
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_RETURN(0);
|
|
|
|
|
2002-09-30 14:54:16 +03:00
|
|
|
err:
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
myrg_detach_children(file);
|
|
|
|
DBUG_RETURN(my_errno= error);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
|
|
|
|
/**
|
|
|
|
@brief Detach all children from a MERGE table.
|
|
|
|
|
|
|
|
@note Detach must not touch the children in any way.
|
|
|
|
They may have been closed at ths point already.
|
|
|
|
All references to the children should be removed.
|
|
|
|
|
|
|
|
@return status
|
|
|
|
@retval 0 OK
|
|
|
|
@retval != 0 Error, my_errno gives reason
|
|
|
|
*/
|
|
|
|
|
|
|
|
int ha_myisammrg::detach_children(void)
|
|
|
|
{
|
|
|
|
DBUG_ENTER("ha_myisammrg::detach_children");
|
|
|
|
DBUG_ASSERT(this->file && this->file->children_attached);
|
|
|
|
|
|
|
|
if (myrg_detach_children(this->file))
|
|
|
|
{
|
|
|
|
/* purecov: begin inspected */
|
|
|
|
DBUG_PRINT("error", ("my_errno %d", my_errno));
|
|
|
|
DBUG_RETURN(my_errno ? my_errno : -1);
|
|
|
|
/* purecov: end */
|
|
|
|
}
|
|
|
|
DBUG_RETURN(0);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
@brief Close a MERGE parent table, not its children.
|
|
|
|
|
|
|
|
@note The children are expected to be closed separately by the caller.
|
|
|
|
|
|
|
|
@return status
|
|
|
|
@retval 0 OK
|
|
|
|
@retval != 0 Error, my_errno gives reason
|
|
|
|
*/
|
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
int ha_myisammrg::close(void)
|
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
int rc;
|
|
|
|
DBUG_ENTER("ha_myisammrg::close");
|
|
|
|
/*
|
|
|
|
Children must not be attached here. Unless the MERGE table has no
|
|
|
|
children. In this case children_attached is always true.
|
|
|
|
*/
|
|
|
|
DBUG_ASSERT(!this->file->children_attached || !this->file->tables);
|
|
|
|
rc= myrg_close(file);
|
|
|
|
file= 0;
|
|
|
|
DBUG_RETURN(rc);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::write_row(uchar * buf)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ENTER("ha_myisammrg::write_row");
|
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_write_count);
|
2006-07-11 17:28:50 -07:00
|
|
|
|
|
|
|
if (file->merge_insert_method == MERGE_INSERT_DISABLED || !file->tables)
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_RETURN(HA_ERR_TABLE_READONLY);
|
2006-07-11 17:28:50 -07:00
|
|
|
|
2004-10-01 18:54:06 +04:00
|
|
|
if (table->timestamp_field_type & TIMESTAMP_AUTO_SET_ON_INSERT)
|
|
|
|
table->timestamp_field->set_time();
|
2001-09-22 17:40:57 +03:00
|
|
|
if (table->next_number_field && buf == table->record[0])
|
2006-08-30 13:20:39 -07:00
|
|
|
{
|
|
|
|
int error;
|
|
|
|
if ((error= update_auto_increment()))
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_RETURN(error); /* purecov: inspected */
|
2006-08-30 13:20:39 -07:00
|
|
|
}
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_RETURN(myrg_write(file,buf));
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::update_row(const uchar * old_data, uchar * new_data)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_update_count);
|
2004-10-01 18:54:06 +04:00
|
|
|
if (table->timestamp_field_type & TIMESTAMP_AUTO_SET_ON_UPDATE)
|
|
|
|
table->timestamp_field->set_time();
|
2002-11-11 22:43:07 +01:00
|
|
|
return myrg_update(file,old_data,new_data);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::delete_row(const uchar * buf)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_delete_count);
|
2002-11-11 22:43:07 +01:00
|
|
|
return myrg_delete(file,buf);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
2007-08-13 16:11:25 +03:00
|
|
|
int ha_myisammrg::index_read_map(uchar * buf, const uchar * key,
|
|
|
|
key_part_map keypart_map,
|
|
|
|
enum ha_rkey_function find_flag)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_key_count);
|
2007-01-29 10:40:26 +01:00
|
|
|
int error=myrg_rkey(file,buf,active_index, key, keypart_map, find_flag);
|
2000-08-14 15:27:19 +04:00
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
2007-08-13 16:11:25 +03:00
|
|
|
int ha_myisammrg::index_read_idx_map(uchar * buf, uint index, const uchar * key,
|
|
|
|
key_part_map keypart_map,
|
|
|
|
enum ha_rkey_function find_flag)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_key_count);
|
2007-01-29 10:40:26 +01:00
|
|
|
int error=myrg_rkey(file,buf,index, key, keypart_map, find_flag);
|
2000-08-14 15:27:19 +04:00
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
2007-08-13 16:11:25 +03:00
|
|
|
int ha_myisammrg::index_read_last_map(uchar *buf, const uchar *key,
|
|
|
|
key_part_map keypart_map)
|
2002-01-12 15:42:54 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_key_count);
|
2007-01-29 10:40:26 +01:00
|
|
|
int error=myrg_rkey(file,buf,active_index, key, keypart_map,
|
2002-01-12 15:42:54 +02:00
|
|
|
HA_READ_PREFIX_LAST);
|
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2002-01-12 15:42:54 +02:00
|
|
|
}
|
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::index_next(uchar * buf)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_next_count);
|
2000-08-14 15:27:19 +04:00
|
|
|
int error=myrg_rnext(file,buf,active_index);
|
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::index_prev(uchar * buf)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_prev_count);
|
2000-08-14 15:27:19 +04:00
|
|
|
int error=myrg_rprev(file,buf, active_index);
|
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
2000-08-14 15:27:19 +04:00
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::index_first(uchar * buf)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_first_count);
|
2000-08-14 15:27:19 +04:00
|
|
|
int error=myrg_rfirst(file, buf, active_index);
|
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::index_last(uchar * buf)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_last_count);
|
2000-08-14 15:27:19 +04:00
|
|
|
int error=myrg_rlast(file, buf, active_index);
|
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::index_next_same(uchar * buf,
|
|
|
|
const uchar *key __attribute__((unused)),
|
2002-11-25 22:18:44 +01:00
|
|
|
uint length __attribute__((unused)))
|
|
|
|
{
|
2007-11-26 18:58:54 +03:00
|
|
|
int error;
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_next_count);
|
2007-11-26 18:58:54 +03:00
|
|
|
do
|
|
|
|
{
|
|
|
|
error= myrg_rnext_same(file,buf);
|
|
|
|
} while (error == HA_ERR_RECORD_DELETED);
|
2002-11-25 22:18:44 +01:00
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
int ha_myisammrg::rnd_init(bool scan)
|
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
return myrg_reset(file);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::rnd_next(uchar *buf)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_rnd_next_count);
|
2000-07-31 21:29:14 +02:00
|
|
|
int error=myrg_rrnd(file, buf, HA_OFFSET_ERROR);
|
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
int ha_myisammrg::rnd_pos(uchar * buf, uchar *pos)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2007-03-02 08:43:45 -08:00
|
|
|
ha_statistic_increment(&SSV::ha_read_rnd_count);
|
2004-11-07 22:39:27 +01:00
|
|
|
int error=myrg_rrnd(file, buf, my_get_ptr(pos,ref_length));
|
2000-07-31 21:29:14 +02:00
|
|
|
table->status=error ? STATUS_NOT_FOUND: 0;
|
2002-11-11 22:43:07 +01:00
|
|
|
return error;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
void ha_myisammrg::position(const uchar *record)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2006-12-15 00:51:37 +02:00
|
|
|
ulonglong row_position= myrg_position(file);
|
|
|
|
my_store_ptr(ref, ref_length, (my_off_t) row_position);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
2004-05-16 14:48:32 +03:00
|
|
|
|
|
|
|
ha_rows ha_myisammrg::records_in_range(uint inx, key_range *min_key,
|
|
|
|
key_range *max_key)
|
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2004-05-16 14:48:32 +03:00
|
|
|
return (ha_rows) myrg_records_in_range(file, (int) inx, min_key, max_key);
|
2002-11-17 18:54:23 +01:00
|
|
|
}
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2004-05-16 14:48:32 +03:00
|
|
|
|
2006-08-10 22:55:20 +08:00
|
|
|
int ha_myisammrg::info(uint flag)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2006-12-15 00:51:37 +02:00
|
|
|
MYMERGE_INFO mrg_info;
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2006-12-15 00:51:37 +02:00
|
|
|
(void) myrg_status(file,&mrg_info,flag);
|
2002-09-30 14:54:16 +03:00
|
|
|
/*
|
|
|
|
The following fails if one has not compiled MySQL with -DBIG_TABLES
|
|
|
|
and one has more than 2^32 rows in the merge tables.
|
|
|
|
*/
|
2007-01-27 03:46:45 +02:00
|
|
|
stats.records = (ha_rows) mrg_info.records;
|
|
|
|
stats.deleted = (ha_rows) mrg_info.deleted;
|
2002-09-30 14:54:16 +03:00
|
|
|
#if !defined(BIG_TABLES) || SIZEOF_OFF_T == 4
|
2006-12-15 00:51:37 +02:00
|
|
|
if ((mrg_info.records >= (ulonglong) 1 << 32) ||
|
|
|
|
(mrg_info.deleted >= (ulonglong) 1 << 32))
|
2005-01-06 13:00:13 +02:00
|
|
|
table->s->crashed= 1;
|
2002-09-30 14:54:16 +03:00
|
|
|
#endif
|
2007-01-27 03:46:45 +02:00
|
|
|
stats.data_file_length= mrg_info.data_file_length;
|
|
|
|
errkey= mrg_info.errkey;
|
2005-01-06 13:00:13 +02:00
|
|
|
table->s->keys_in_use.set_prefix(table->s->keys);
|
2007-01-27 03:46:45 +02:00
|
|
|
stats.mean_rec_length= mrg_info.reclength;
|
2006-02-11 21:51:43 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
The handler::block_size is used all over the code in index scan cost
|
|
|
|
calculations. It is used to get number of disk seeks required to
|
|
|
|
retrieve a number of index tuples.
|
|
|
|
If the merge table has N underlying tables, then (assuming underlying
|
|
|
|
tables have equal size, the only "simple" approach we can use)
|
|
|
|
retrieving X index records from a merge table will require N times more
|
|
|
|
disk seeks compared to doing the same on a MyISAM table with equal
|
|
|
|
number of records.
|
|
|
|
In the edge case (file_tables > myisam_block_size) we'll get
|
|
|
|
block_size==0, and index calculation code will act as if we need one
|
|
|
|
disk seek to retrieve one index tuple.
|
|
|
|
|
|
|
|
TODO: In 5.2 index scan cost calculation will be factored out into a
|
|
|
|
virtual function in class handler and we'll be able to remove this hack.
|
|
|
|
*/
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
stats.block_size= 0;
|
2006-02-11 21:51:43 +03:00
|
|
|
if (file->tables)
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
stats.block_size= myisam_block_size / file->tables;
|
2006-02-11 21:51:43 +03:00
|
|
|
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
stats.update_time= 0;
|
2001-11-03 15:18:09 +02:00
|
|
|
#if SIZEOF_OFF_T > 4
|
2000-07-31 21:29:14 +02:00
|
|
|
ref_length=6; // Should be big enough
|
2001-11-03 15:18:09 +02:00
|
|
|
#else
|
|
|
|
ref_length=4; // Can't be > than my_off_t
|
|
|
|
#endif
|
2002-12-10 00:06:05 +01:00
|
|
|
if (flag & HA_STATUS_CONST)
|
|
|
|
{
|
2006-12-15 00:51:37 +02:00
|
|
|
if (table->s->key_parts && mrg_info.rec_per_key)
|
2006-10-06 15:17:42 +05:00
|
|
|
{
|
|
|
|
#ifdef HAVE_purify
|
|
|
|
/*
|
|
|
|
valgrind may be unhappy about it, because optimizer may access values
|
|
|
|
between file->keys and table->key_parts, that will be uninitialized.
|
|
|
|
It's safe though, because even if opimizer will decide to use a key
|
|
|
|
with such a number, it'll be an error later anyway.
|
|
|
|
*/
|
|
|
|
bzero((char*) table->key_info[0].rec_per_key,
|
2006-10-08 22:34:32 +05:00
|
|
|
sizeof(table->key_info[0].rec_per_key) * table->s->key_parts);
|
2006-10-06 15:17:42 +05:00
|
|
|
#endif
|
2002-12-10 00:06:05 +01:00
|
|
|
memcpy((char*) table->key_info[0].rec_per_key,
|
2006-12-15 00:51:37 +02:00
|
|
|
(char*) mrg_info.rec_per_key,
|
2006-10-06 15:17:42 +05:00
|
|
|
sizeof(table->key_info[0].rec_per_key) *
|
2006-10-06 15:29:30 +05:00
|
|
|
min(file->keys, table->s->key_parts));
|
2006-10-06 15:17:42 +05:00
|
|
|
}
|
2002-12-10 00:06:05 +01:00
|
|
|
}
|
2006-08-10 22:55:20 +08:00
|
|
|
return 0;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int ha_myisammrg::extra(enum ha_extra_function operation)
|
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
if (operation == HA_EXTRA_ATTACH_CHILDREN)
|
|
|
|
{
|
|
|
|
int rc= attach_children();
|
|
|
|
if (!rc)
|
|
|
|
(void) extra(HA_EXTRA_NO_READCHECK); // Not needed in SQL
|
|
|
|
return(rc);
|
|
|
|
}
|
|
|
|
else if (operation == HA_EXTRA_DETACH_CHILDREN)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
Note that detach must not touch the children in any way.
|
|
|
|
They may have been closed at ths point already.
|
|
|
|
*/
|
|
|
|
int rc= detach_children();
|
|
|
|
return(rc);
|
|
|
|
}
|
|
|
|
|
2001-09-27 21:45:48 +03:00
|
|
|
/* As this is just a mapping, we don't have to force the underlying
|
|
|
|
tables to be closed */
|
|
|
|
if (operation == HA_EXTRA_FORCE_REOPEN ||
|
|
|
|
operation == HA_EXTRA_PREPARE_FOR_DELETE)
|
2002-11-11 22:43:07 +01:00
|
|
|
return 0;
|
|
|
|
return myrg_extra(file,operation,0);
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
This changeset is largely a handler cleanup changeset (WL#3281), but includes fixes and cleanups that was found necessary while testing the handler changes
Changes that requires code changes in other code of other storage engines.
(Note that all changes are very straightforward and one should find all issues
by compiling a --debug build and fixing all compiler errors and all
asserts in field.cc while running the test suite),
- New optional handler function introduced: reset()
This is called after every DML statement to make it easy for a handler to
statement specific cleanups.
(The only case it's not called is if force the file to be closed)
- handler::extra(HA_EXTRA_RESET) is removed. Code that was there before
should be moved to handler::reset()
- table->read_set contains a bitmap over all columns that are needed
in the query. read_row() and similar functions only needs to read these
columns
- table->write_set contains a bitmap over all columns that will be updated
in the query. write_row() and update_row() only needs to update these
columns.
The above bitmaps should now be up to date in all context
(including ALTER TABLE, filesort()).
The handler is informed of any changes to the bitmap after
fix_fields() by calling the virtual function
handler::column_bitmaps_signal(). If the handler does caching of
these bitmaps (instead of using table->read_set, table->write_set),
it should redo the caching in this code. as the signal() may be sent
several times, it's probably best to set a variable in the signal
and redo the caching on read_row() / write_row() if the variable was
set.
- Removed the read_set and write_set bitmap objects from the handler class
- Removed all column bit handling functions from the handler class.
(Now one instead uses the normal bitmap functions in my_bitmap.c instead
of handler dedicated bitmap functions)
- field->query_id is removed. One should instead instead check
table->read_set and table->write_set if a field is used in the query.
- handler::extra(HA_EXTRA_RETRIVE_ALL_COLS) and
handler::extra(HA_EXTRA_RETRIEVE_PRIMARY_KEY) are removed. One should now
instead use table->read_set to check for which columns to retrieve.
- If a handler needs to call Field->val() or Field->store() on columns
that are not used in the query, one should install a temporary
all-columns-used map while doing so. For this, we provide the following
functions:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->read_set);
field->val();
dbug_tmp_restore_column_map(table->read_set, old_map);
and similar for the write map:
my_bitmap_map *old_map= dbug_tmp_use_all_columns(table, table->write_set);
field->val();
dbug_tmp_restore_column_map(table->write_set, old_map);
If this is not done, you will sooner or later hit a DBUG_ASSERT
in the field store() / val() functions.
(For not DBUG binaries, the dbug_tmp_restore_column_map() and
dbug_tmp_restore_column_map() are inline dummy functions and should
be optimized away be the compiler).
- If one needs to temporary set the column map for all binaries (and not
just to avoid the DBUG_ASSERT() in the Field::store() / Field::val()
methods) one should use the functions tmp_use_all_columns() and
tmp_restore_column_map() instead of the above dbug_ variants.
- All 'status' fields in the handler base class (like records,
data_file_length etc) are now stored in a 'stats' struct. This makes
it easier to know what status variables are provided by the base
handler. This requires some trivial variable names in the extra()
function.
- New virtual function handler::records(). This is called to optimize
COUNT(*) if (handler::table_flags() & HA_HAS_RECORDS()) is true.
(stats.records is not supposed to be an exact value. It's only has to
be 'reasonable enough' for the optimizer to be able to choose a good
optimization path).
- Non virtual handler::init() function added for caching of virtual
constants from engine.
- Removed has_transactions() virtual method. Now one should instead return
HA_NO_TRANSACTIONS in table_flags() if the table handler DOES NOT support
transactions.
- The 'xxxx_create_handler()' function now has a MEM_ROOT_root argument
that is to be used with 'new handler_name()' to allocate the handler
in the right area. The xxxx_create_handler() function is also
responsible for any initialization of the object before returning.
For example, one should change:
static handler *myisam_create_handler(TABLE_SHARE *table)
{
return new ha_myisam(table);
}
->
static handler *myisam_create_handler(TABLE_SHARE *table, MEM_ROOT *mem_root)
{
return new (mem_root) ha_myisam(table);
}
- New optional virtual function: use_hidden_primary_key().
This is called in case of an update/delete when
(table_flags() and HA_PRIMARY_KEY_REQUIRED_FOR_DELETE) is defined
but we don't have a primary key. This allows the handler to take precisions
in remembering any hidden primary key to able to update/delete any
found row. The default handler marks all columns to be read.
- handler::table_flags() now returns a ulonglong (to allow for more flags).
- New/changed table_flags()
- HA_HAS_RECORDS Set if ::records() is supported
- HA_NO_TRANSACTIONS Set if engine doesn't support transactions
- HA_PRIMARY_KEY_REQUIRED_FOR_DELETE
Set if we should mark all primary key columns for
read when reading rows as part of a DELETE
statement. If there is no primary key,
all columns are marked for read.
- HA_PARTIAL_COLUMN_READ Set if engine will not read all columns in some
cases (based on table->read_set)
- HA_PRIMARY_KEY_ALLOW_RANDOM_ACCESS
Renamed to HA_PRIMARY_KEY_REQUIRED_FOR_POSITION.
- HA_DUPP_POS Renamed to HA_DUPLICATE_POS
- HA_REQUIRES_KEY_COLUMNS_FOR_DELETE
Set this if we should mark ALL key columns for
read when when reading rows as part of a DELETE
statement. In case of an update we will mark
all keys for read for which key part changed
value.
- HA_STATS_RECORDS_IS_EXACT
Set this if stats.records is exact.
(This saves us some extra records() calls
when optimizing COUNT(*))
- Removed table_flags()
- HA_NOT_EXACT_COUNT Now one should instead use HA_HAS_RECORDS if
handler::records() gives an exact count() and
HA_STATS_RECORDS_IS_EXACT if stats.records is exact.
- HA_READ_RND_SAME Removed (no one supported this one)
- Removed not needed functions ha_retrieve_all_cols() and ha_retrieve_all_pk()
- Renamed handler::dupp_pos to handler::dup_pos
- Removed not used variable handler::sortkey
Upper level handler changes:
- ha_reset() now does some overall checks and calls ::reset()
- ha_table_flags() added. This is a cached version of table_flags(). The
cache is updated on engine creation time and updated on open.
MySQL level changes (not obvious from the above):
- DBUG_ASSERT() added to check that column usage matches what is set
in the column usage bit maps. (This found a LOT of bugs in current
column marking code).
- In 5.1 before, all used columns was marked in read_set and only updated
columns was marked in write_set. Now we only mark columns for which we
need a value in read_set.
- Column bitmaps are created in open_binary_frm() and open_table_from_share().
(Before this was in table.cc)
- handler::table_flags() calls are replaced with handler::ha_table_flags()
- For calling field->val() you must have the corresponding bit set in
table->read_set. For calling field->store() you must have the
corresponding bit set in table->write_set. (There are asserts in
all store()/val() functions to catch wrong usage)
- thd->set_query_id is renamed to thd->mark_used_columns and instead
of setting this to an integer value, this has now the values:
MARK_COLUMNS_NONE, MARK_COLUMNS_READ, MARK_COLUMNS_WRITE
Changed also all variables named 'set_query_id' to mark_used_columns.
- In filesort() we now inform the handler of exactly which columns are needed
doing the sort and choosing the rows.
- The TABLE_SHARE object has a 'all_set' column bitmap one can use
when one needs a column bitmap with all columns set.
(This is used for table->use_all_columns() and other places)
- The TABLE object has 3 column bitmaps:
- def_read_set Default bitmap for columns to be read
- def_write_set Default bitmap for columns to be written
- tmp_set Can be used as a temporary bitmap when needed.
The table object has also two pointer to bitmaps read_set and write_set
that the handler should use to find out which columns are used in which way.
- count() optimization now calls handler::records() instead of using
handler->stats.records (if (table_flags() & HA_HAS_RECORDS) is true).
- Added extra argument to Item::walk() to indicate if we should also
traverse sub queries.
- Added TABLE parameter to cp_buffer_from_ref()
- Don't close tables created with CREATE ... SELECT but keep them in
the table cache. (Faster usage of newly created tables).
New interfaces:
- table->clear_column_bitmaps() to initialize the bitmaps for tables
at start of new statements.
- table->column_bitmaps_set() to set up new column bitmaps and signal
the handler about this.
- table->column_bitmaps_set_no_signal() for some few cases where we need
to setup new column bitmaps but don't signal the handler (as the handler
has already been signaled about these before). Used for the momement
only in opt_range.cc when doing ROR scans.
- table->use_all_columns() to install a bitmap where all columns are marked
as use in the read and the write set.
- table->default_column_bitmaps() to install the normal read and write
column bitmaps, but not signaling the handler about this.
This is mainly used when creating TABLE instances.
- table->mark_columns_needed_for_delete(),
table->mark_columns_needed_for_delete() and
table->mark_columns_needed_for_insert() to allow us to put additional
columns in column usage maps if handler so requires.
(The handler indicates what it neads in handler->table_flags())
- table->prepare_for_position() to allow us to tell handler that it
needs to read primary key parts to be able to store them in
future table->position() calls.
(This replaces the table->file->ha_retrieve_all_pk function)
- table->mark_auto_increment_column() to tell handler are going to update
columns part of any auto_increment key.
- table->mark_columns_used_by_index() to mark all columns that is part of
an index. It will also send extra(HA_EXTRA_KEYREAD) to handler to allow
it to quickly know that it only needs to read colums that are part
of the key. (The handler can also use the column map for detecting this,
but simpler/faster handler can just monitor the extra() call).
- table->mark_columns_used_by_index_no_reset() to in addition to other columns,
also mark all columns that is used by the given key.
- table->restore_column_maps_after_mark_index() to restore to default
column maps after a call to table->mark_columns_used_by_index().
- New item function register_field_in_read_map(), for marking used columns
in table->read_map. Used by filesort() to mark all used columns
- Maintain in TABLE->merge_keys set of all keys that are used in query.
(Simplices some optimization loops)
- Maintain Field->part_of_key_not_clustered which is like Field->part_of_key
but the field in the clustered key is not assumed to be part of all index.
(used in opt_range.cc for faster loops)
- dbug_tmp_use_all_columns(), dbug_tmp_restore_column_map()
tmp_use_all_columns() and tmp_restore_column_map() functions to temporally
mark all columns as usable. The 'dbug_' version is primarily intended
inside a handler when it wants to just call Field:store() & Field::val()
functions, but don't need the column maps set for any other usage.
(ie:: bitmap_is_set() is never called)
- We can't use compare_records() to skip updates for handlers that returns
a partial column set and the read_set doesn't cover all columns in the
write set. The reason for this is that if we have a column marked only for
write we can't in the MySQL level know if the value changed or not.
The reason this worked before was that MySQL marked all to be written
columns as also to be read. The new 'optimal' bitmaps exposed this 'hidden
bug'.
- open_table_from_share() does not anymore setup temporary MEM_ROOT
object as a thread specific variable for the handler. Instead we
send the to-be-used MEMROOT to get_new_handler().
(Simpler, faster code)
Bugs fixed:
- Column marking was not done correctly in a lot of cases.
(ALTER TABLE, when using triggers, auto_increment fields etc)
(Could potentially result in wrong values inserted in table handlers
relying on that the old column maps or field->set_query_id was correct)
Especially when it comes to triggers, there may be cases where the
old code would cause lost/wrong values for NDB and/or InnoDB tables.
- Split thd->options flag OPTION_STATUS_NO_TRANS_UPDATE to two flags:
OPTION_STATUS_NO_TRANS_UPDATE and OPTION_KEEP_LOG.
This allowed me to remove some wrong warnings about:
"Some non-transactional changed tables couldn't be rolled back"
- Fixed handling of INSERT .. SELECT and CREATE ... SELECT that wrongly reset
(thd->options & OPTION_STATUS_NO_TRANS_UPDATE) which caused us to loose
some warnings about
"Some non-transactional changed tables couldn't be rolled back")
- Fixed use of uninitialized memory in ha_ndbcluster.cc::delete_table()
which could cause delete_table to report random failures.
- Fixed core dumps for some tests when running with --debug
- Added missing FN_LIBCHAR in mysql_rm_tmp_tables()
(This has probably caused us to not properly remove temporary files after
crash)
- slow_logs was not properly initialized, which could maybe cause
extra/lost entries in slow log.
- If we get an duplicate row on insert, change column map to read and
write all columns while retrying the operation. This is required by
the definition of REPLACE and also ensures that fields that are only
part of UPDATE are properly handled. This fixed a bug in NDB and
REPLACE where REPLACE wrongly copied some column values from the replaced
row.
- For table handler that doesn't support NULL in keys, we would give an error
when creating a primary key with NULL fields, even after the fields has been
automaticly converted to NOT NULL.
- Creating a primary key on a SPATIAL key, would fail if field was not
declared as NOT NULL.
Cleanups:
- Removed not used condition argument to setup_tables
- Removed not needed item function reset_query_id_processor().
- Field->add_index is removed. Now this is instead maintained in
(field->flags & FIELD_IN_ADD_INDEX)
- Field->fieldnr is removed (use field->field_index instead)
- New argument to filesort() to indicate that it should return a set of
row pointers (not used columns). This allowed me to remove some references
to sql_command in filesort and should also enable us to return column
results in some cases where we couldn't before.
- Changed column bitmap handling in opt_range.cc to be aligned with TABLE
bitmap, which allowed me to use bitmap functions instead of looping over
all fields to create some needed bitmaps. (Faster and smaller code)
- Broke up found too long lines
- Moved some variable declaration at start of function for better code
readability.
- Removed some not used arguments from functions.
(setup_fields(), mysql_prepare_insert_check_table())
- setup_fields() now takes an enum instead of an int for marking columns
usage.
- For internal temporary tables, use handler::write_row(),
handler::delete_row() and handler::update_row() instead of
handler::ha_xxxx() for faster execution.
- Changed some constants to enum's and define's.
- Using separate column read and write sets allows for easier checking
of timestamp field was set by statement.
- Remove calls to free_io_cache() as this is now done automaticly in ha_reset()
- Don't build table->normalized_path as this is now identical to table->path
(after bar's fixes to convert filenames)
- Fixed some missed DBUG_PRINT(.."%lx") to use "0x%lx" to make it easier to
do comparision with the 'convert-dbug-for-diff' tool.
Things left to do in 5.1:
- We wrongly log failed CREATE TABLE ... SELECT in some cases when using
row based logging (as shown by testcase binlog_row_mix_innodb_myisam.result)
Mats has promised to look into this.
- Test that my fix for CREATE TABLE ... SELECT is indeed correct.
(I added several test cases for this, but in this case it's better that
someone else also tests this throughly).
Lars has promosed to do this.
2006-06-04 18:52:22 +03:00
|
|
|
int ha_myisammrg::reset(void)
|
|
|
|
{
|
|
|
|
return myrg_reset(file);
|
|
|
|
}
|
2002-07-23 18:31:22 +03:00
|
|
|
|
|
|
|
/* To be used with WRITE_CACHE, EXTRA_CACHE and BULK_INSERT_BEGIN */
|
|
|
|
|
|
|
|
int ha_myisammrg::extra_opt(enum ha_extra_function operation, ulong cache_size)
|
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2002-12-12 21:01:32 +02:00
|
|
|
if ((specialflag & SPECIAL_SAFE_MODE) && operation == HA_EXTRA_WRITE_CACHE)
|
2002-11-11 22:43:07 +01:00
|
|
|
return 0;
|
|
|
|
return myrg_extra(file, operation, (void*) &cache_size);
|
2002-07-23 18:31:22 +03:00
|
|
|
}
|
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
int ha_myisammrg::external_lock(THD *thd, int lock_type)
|
|
|
|
{
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
DBUG_ASSERT(this->file->children_attached);
|
2002-11-11 22:43:07 +01:00
|
|
|
return myrg_lock_database(file,lock_type);
|
2000-08-14 15:27:19 +04:00
|
|
|
}
|
2000-07-31 21:29:14 +02:00
|
|
|
|
|
|
|
uint ha_myisammrg::lock_count(void) const
|
|
|
|
{
|
2007-12-11 15:32:10 +01:00
|
|
|
/*
|
|
|
|
Return the real lock count even if the children are not attached.
|
|
|
|
This method is used for allocating memory. If we would return 0
|
|
|
|
to another thread (e.g. doing FLUSH TABLE), and attach the children
|
|
|
|
before the other thread calls store_lock(), then we would return
|
|
|
|
more locks in store_lock() than we claimed by lock_count(). The
|
|
|
|
other tread would overrun its memory.
|
|
|
|
*/
|
2002-11-11 22:43:07 +01:00
|
|
|
return file->tables;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
THR_LOCK_DATA **ha_myisammrg::store_lock(THD *thd,
|
|
|
|
THR_LOCK_DATA **to,
|
|
|
|
enum thr_lock_type lock_type)
|
|
|
|
{
|
2002-11-07 03:54:00 +02:00
|
|
|
MYRG_TABLE *open_table;
|
2007-12-11 15:32:10 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
This method can be called while another thread is attaching the
|
|
|
|
children. If the processor reorders instructions or write to memory,
|
|
|
|
'children_attached' could be set before 'open_tables' has all the
|
|
|
|
pointers to the children. Use of a mutex here and in
|
|
|
|
myrg_attach_children() forces consistent data.
|
|
|
|
*/
|
|
|
|
pthread_mutex_lock(&this->file->mutex);
|
|
|
|
|
|
|
|
/*
|
|
|
|
When MERGE table is open, but not yet attached, other threads
|
|
|
|
could flush it, which means call mysql_lock_abort_for_thread()
|
|
|
|
on this threads TABLE. 'children_attached' is FALSE in this
|
|
|
|
situaton. Since the table is not locked, return no lock data.
|
|
|
|
*/
|
|
|
|
if (!this->file->children_attached)
|
|
|
|
goto end; /* purecov: tested */
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2002-11-07 03:54:00 +02:00
|
|
|
for (open_table=file->open_tables ;
|
|
|
|
open_table != file->end_table ;
|
|
|
|
open_table++)
|
2000-07-31 21:29:14 +02:00
|
|
|
{
|
2002-11-07 03:54:00 +02:00
|
|
|
*(to++)= &open_table->table->lock;
|
|
|
|
if (lock_type != TL_IGNORE && open_table->table->lock.type == TL_UNLOCK)
|
|
|
|
open_table->table->lock.type=lock_type;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
2007-12-11 15:32:10 +01:00
|
|
|
|
|
|
|
end:
|
|
|
|
pthread_mutex_unlock(&this->file->mutex);
|
2002-11-11 22:43:07 +01:00
|
|
|
return to;
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
|
|
|
|
2003-06-27 16:29:10 +03:00
|
|
|
|
|
|
|
/* Find out database name and table name from a filename */
|
|
|
|
|
|
|
|
static void split_file_name(const char *file_name,
|
|
|
|
LEX_STRING *db, LEX_STRING *name)
|
|
|
|
{
|
2003-11-20 22:06:25 +02:00
|
|
|
uint dir_length, prefix_length;
|
2003-06-27 16:29:10 +03:00
|
|
|
char buff[FN_REFLEN];
|
|
|
|
|
|
|
|
db->length= 0;
|
2003-11-20 22:06:25 +02:00
|
|
|
strmake(buff, file_name, sizeof(buff)-1);
|
2003-06-27 16:29:10 +03:00
|
|
|
dir_length= dirname_length(buff);
|
|
|
|
if (dir_length > 1)
|
|
|
|
{
|
|
|
|
/* Get database */
|
|
|
|
buff[dir_length-1]= 0; // Remove end '/'
|
|
|
|
prefix_length= dirname_length(buff);
|
|
|
|
db->str= (char*) file_name+ prefix_length;
|
|
|
|
db->length= dir_length - prefix_length -1;
|
|
|
|
}
|
|
|
|
name->str= (char*) file_name+ dir_length;
|
|
|
|
name->length= (uint) (fn_ext(name->str) - name->str);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2001-03-06 15:24:08 +02:00
|
|
|
void ha_myisammrg::update_create_info(HA_CREATE_INFO *create_info)
|
|
|
|
{
|
2002-11-11 22:43:07 +01:00
|
|
|
DBUG_ENTER("ha_myisammrg::update_create_info");
|
2003-06-27 16:29:10 +03:00
|
|
|
|
2001-03-06 15:24:08 +02:00
|
|
|
if (!(create_info->used_fields & HA_CREATE_USED_UNION))
|
|
|
|
{
|
2002-11-07 03:54:00 +02:00
|
|
|
MYRG_TABLE *open_table;
|
2001-03-06 15:24:08 +02:00
|
|
|
THD *thd=current_thd;
|
2003-06-27 16:29:10 +03:00
|
|
|
|
2001-03-06 15:24:08 +02:00
|
|
|
create_info->merge_list.next= &create_info->merge_list.first;
|
2001-08-10 17:05:54 +03:00
|
|
|
create_info->merge_list.elements=0;
|
2001-03-06 15:24:08 +02:00
|
|
|
|
2002-11-07 03:54:00 +02:00
|
|
|
for (open_table=file->open_tables ;
|
|
|
|
open_table != file->end_table ;
|
|
|
|
open_table++)
|
2001-03-06 15:24:08 +02:00
|
|
|
{
|
|
|
|
TABLE_LIST *ptr;
|
2003-06-27 16:29:10 +03:00
|
|
|
LEX_STRING db, name;
|
2007-02-27 19:31:49 +02:00
|
|
|
LINT_INIT(db.str);
|
2003-06-27 16:29:10 +03:00
|
|
|
|
2001-03-06 15:24:08 +02:00
|
|
|
if (!(ptr = (TABLE_LIST *) thd->calloc(sizeof(TABLE_LIST))))
|
|
|
|
goto err;
|
2003-06-27 16:29:10 +03:00
|
|
|
split_file_name(open_table->table->filename, &db, &name);
|
2005-01-06 13:00:13 +02:00
|
|
|
if (!(ptr->table_name= thd->strmake(name.str, name.length)))
|
2003-06-27 16:29:10 +03:00
|
|
|
goto err;
|
|
|
|
if (db.length && !(ptr->db= thd->strmake(db.str, db.length)))
|
2001-03-06 15:24:08 +02:00
|
|
|
goto err;
|
2003-06-27 16:29:10 +03:00
|
|
|
|
2001-08-10 17:05:54 +03:00
|
|
|
create_info->merge_list.elements++;
|
WL#3817: Simplify string / memory area types and make things more consistent (first part)
The following type conversions was done:
- Changed byte to uchar
- Changed gptr to uchar*
- Change my_string to char *
- Change my_size_t to size_t
- Change size_s to size_t
Removed declaration of byte, gptr, my_string, my_size_t and size_s.
Following function parameter changes was done:
- All string functions in mysys/strings was changed to use size_t
instead of uint for string lengths.
- All read()/write() functions changed to use size_t (including vio).
- All protocoll functions changed to use size_t instead of uint
- Functions that used a pointer to a string length was changed to use size_t*
- Changed malloc(), free() and related functions from using gptr to use void *
as this requires fewer casts in the code and is more in line with how the
standard functions work.
- Added extra length argument to dirname_part() to return the length of the
created string.
- Changed (at least) following functions to take uchar* as argument:
- db_dump()
- my_net_write()
- net_write_command()
- net_store_data()
- DBUG_DUMP()
- decimal2bin() & bin2decimal()
- Changed my_compress() and my_uncompress() to use size_t. Changed one
argument to my_uncompress() from a pointer to a value as we only return
one value (makes function easier to use).
- Changed type of 'pack_data' argument to packfrm() to avoid casts.
- Changed in readfrm() and writefrom(), ha_discover and handler::discover()
the type for argument 'frmdata' to uchar** to avoid casts.
- Changed most Field functions to use uchar* instead of char* (reduced a lot of
casts).
- Changed field->val_xxx(xxx, new_ptr) to take const pointers.
Other changes:
- Removed a lot of not needed casts
- Added a few new cast required by other changes
- Added some cast to my_multi_malloc() arguments for safety (as string lengths
needs to be uint, not size_t).
- Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done
explicitely as this conflict was often hided by casting the function to
hash_get_key).
- Changed some buffers to memory regions to uchar* to avoid casts.
- Changed some string lengths from uint to size_t.
- Changed field->ptr to be uchar* instead of char*. This allowed us to
get rid of a lot of casts.
- Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar
- Include zlib.h in some files as we needed declaration of crc32()
- Changed MY_FILE_ERROR to be (size_t) -1.
- Changed many variables to hold the result of my_read() / my_write() to be
size_t. This was needed to properly detect errors (which are
returned as (size_t) -1).
- Removed some very old VMS code
- Changed packfrm()/unpackfrm() to not be depending on uint size
(portability fix)
- Removed windows specific code to restore cursor position as this
causes slowdown on windows and we should not mix read() and pread()
calls anyway as this is not thread safe. Updated function comment to
reflect this. Changed function that depended on original behavior of
my_pwrite() to itself restore the cursor position (one such case).
- Added some missing checking of return value of malloc().
- Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow.
- Changed type of table_def::m_size from my_size_t to ulong to reflect that
m_size is the number of elements in the array, not a string/memory
length.
- Moved THD::max_row_length() to table.cc (as it's not depending on THD).
Inlined max_row_length_blob() into this function.
- More function comments
- Fixed some compiler warnings when compiled without partitions.
- Removed setting of LEX_STRING() arguments in declaration (portability fix).
- Some trivial indentation/variable name changes.
- Some trivial code simplifications:
- Replaced some calls to alloc_root + memcpy to use
strmake_root()/strdup_root().
- Changed some calls from memdup() to strmake() (Safety fix)
- Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
|
|
|
(*create_info->merge_list.next) = (uchar*) ptr;
|
|
|
|
create_info->merge_list.next= (uchar**) &ptr->next_local;
|
2001-03-06 15:24:08 +02:00
|
|
|
}
|
|
|
|
*create_info->merge_list.next=0;
|
|
|
|
}
|
2001-09-22 17:40:57 +03:00
|
|
|
if (!(create_info->used_fields & HA_CREATE_USED_INSERT_METHOD))
|
|
|
|
{
|
|
|
|
create_info->merge_insert_method = file->merge_insert_method;
|
|
|
|
}
|
2001-03-06 15:24:08 +02:00
|
|
|
DBUG_VOID_RETURN;
|
|
|
|
|
|
|
|
err:
|
|
|
|
create_info->merge_list.elements=0;
|
|
|
|
create_info->merge_list.first=0;
|
|
|
|
DBUG_VOID_RETURN;
|
|
|
|
}
|
2000-07-31 21:29:14 +02:00
|
|
|
|
2003-06-27 16:29:10 +03:00
|
|
|
|
2000-07-31 21:29:14 +02:00
|
|
|
int ha_myisammrg::create(const char *name, register TABLE *form,
|
|
|
|
HA_CREATE_INFO *create_info)
|
|
|
|
{
|
2005-01-06 13:00:13 +02:00
|
|
|
char buff[FN_REFLEN];
|
|
|
|
const char **table_names, **pos;
|
2000-09-14 02:39:07 +03:00
|
|
|
TABLE_LIST *tables= (TABLE_LIST*) create_info->merge_list.first;
|
2003-06-27 16:29:10 +03:00
|
|
|
THD *thd= current_thd;
|
2005-04-28 18:28:50 +02:00
|
|
|
uint dirlgt= dirname_length(name);
|
2000-09-14 02:39:07 +03:00
|
|
|
DBUG_ENTER("ha_myisammrg::create");
|
|
|
|
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
/* Allocate a table_names array in thread mem_root. */
|
2005-01-06 13:00:13 +02:00
|
|
|
if (!(table_names= (const char**)
|
|
|
|
thd->alloc((create_info->merge_list.elements+1) * sizeof(char*))))
|
2003-12-30 13:14:21 +02:00
|
|
|
DBUG_RETURN(HA_ERR_OUT_OF_MEM);
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
|
|
|
|
/* Create child path names. */
|
2004-07-16 01:15:55 +03:00
|
|
|
for (pos= table_names; tables; tables= tables->next_local)
|
2002-11-05 00:00:51 +01:00
|
|
|
{
|
2005-01-06 13:00:13 +02:00
|
|
|
const char *table_name;
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
Construct the path to the MyISAM table. Try to meet two conditions:
|
|
|
|
1.) Allow to include MyISAM tables from different databases, and
|
|
|
|
2.) allow for moving DATADIR around in the file system.
|
|
|
|
The first means that we need paths in the .MRG file. The second
|
|
|
|
means that we should not have absolute paths in the .MRG file.
|
|
|
|
The best, we can do, is to use 'mysql_data_home', which is '.'
|
|
|
|
in mysqld and may be an absolute path in an embedded server.
|
|
|
|
This means that it might not be possible to move the DATADIR of
|
|
|
|
an embedded server without changing the paths in the .MRG file.
|
|
|
|
|
|
|
|
Do the same even for temporary tables. MERGE children are now
|
|
|
|
opened through the table cache. They are opened by db.table_name,
|
|
|
|
not by their path name.
|
|
|
|
*/
|
|
|
|
uint length= build_table_filename(buff, sizeof(buff),
|
|
|
|
tables->db, tables->table_name, "", 0);
|
|
|
|
/*
|
|
|
|
If a MyISAM table is in the same directory as the MERGE table,
|
|
|
|
we use the table name without a path. This means that the
|
|
|
|
DATADIR can easily be moved even for an embedded server as long
|
|
|
|
as the MyISAM tables are from the same database as the MERGE table.
|
|
|
|
*/
|
|
|
|
if ((dirname_length(buff) == dirlgt) && ! memcmp(buff, name, dirlgt))
|
|
|
|
table_name= tables->table_name;
|
2002-11-05 00:00:51 +01:00
|
|
|
else
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
if (! (table_name= thd->strmake(buff, length)))
|
|
|
|
DBUG_RETURN(HA_ERR_OUT_OF_MEM); /* purecov: inspected */
|
|
|
|
|
2002-11-05 00:00:51 +01:00
|
|
|
*pos++= table_name;
|
|
|
|
}
|
2000-09-14 02:39:07 +03:00
|
|
|
*pos=0;
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
|
|
|
|
/* Create a MERGE meta file from the table_names array. */
|
2005-12-31 09:01:26 +04:00
|
|
|
DBUG_RETURN(myrg_create(fn_format(buff,name,"","",
|
|
|
|
MY_RESOLVE_SYMLINKS|
|
|
|
|
MY_UNPACK_FILENAME|MY_APPEND_EXT),
|
2005-01-06 13:00:13 +02:00
|
|
|
table_names,
|
2001-09-22 17:40:57 +03:00
|
|
|
create_info->merge_insert_method,
|
|
|
|
(my_bool) 0));
|
2000-07-31 21:29:14 +02:00
|
|
|
}
|
2001-02-17 19:04:33 +02:00
|
|
|
|
2003-06-27 16:29:10 +03:00
|
|
|
|
2001-02-17 19:04:33 +02:00
|
|
|
void ha_myisammrg::append_create_info(String *packet)
|
|
|
|
{
|
2003-06-27 16:29:10 +03:00
|
|
|
const char *current_db;
|
|
|
|
uint db_length;
|
|
|
|
THD *thd= current_thd;
|
2005-11-23 22:45:02 +02:00
|
|
|
MYRG_TABLE *open_table, *first;
|
2003-06-27 16:29:10 +03:00
|
|
|
|
2001-09-22 17:40:57 +03:00
|
|
|
if (file->merge_insert_method != MERGE_INSERT_DISABLED)
|
|
|
|
{
|
2005-11-20 20:47:07 +02:00
|
|
|
packet->append(STRING_WITH_LEN(" INSERT_METHOD="));
|
2001-09-27 21:45:48 +03:00
|
|
|
packet->append(get_type(&merge_insert_method,file->merge_insert_method-1));
|
2001-09-22 17:40:57 +03:00
|
|
|
}
|
2005-11-20 20:47:07 +02:00
|
|
|
packet->append(STRING_WITH_LEN(" UNION=("));
|
2001-02-17 19:04:33 +02:00
|
|
|
|
2005-11-23 22:45:02 +02:00
|
|
|
current_db= table->s->db.str;
|
|
|
|
db_length= table->s->db.length;
|
2003-06-27 16:29:10 +03:00
|
|
|
|
2002-11-07 03:54:00 +02:00
|
|
|
for (first=open_table=file->open_tables ;
|
|
|
|
open_table != file->end_table ;
|
|
|
|
open_table++)
|
2001-02-17 19:04:33 +02:00
|
|
|
{
|
2003-06-27 16:29:10 +03:00
|
|
|
LEX_STRING db, name;
|
2007-02-27 19:31:49 +02:00
|
|
|
LINT_INIT(db.str);
|
|
|
|
|
2003-06-27 16:29:10 +03:00
|
|
|
split_file_name(open_table->table->filename, &db, &name);
|
2002-11-07 03:54:00 +02:00
|
|
|
if (open_table != first)
|
2001-02-17 19:04:33 +02:00
|
|
|
packet->append(',');
|
2003-06-27 16:29:10 +03:00
|
|
|
/* Report database for mapped table if it isn't in current database */
|
|
|
|
if (db.length &&
|
|
|
|
(db_length != db.length ||
|
|
|
|
strncmp(current_db, db.str, db.length)))
|
|
|
|
{
|
|
|
|
append_identifier(thd, packet, db.str, db.length);
|
|
|
|
packet->append('.');
|
|
|
|
}
|
|
|
|
append_identifier(thd, packet, name.str, name.length);
|
2001-02-17 19:04:33 +02:00
|
|
|
}
|
|
|
|
packet->append(')');
|
|
|
|
}
|
2005-07-22 23:43:59 +03:00
|
|
|
|
|
|
|
|
|
|
|
bool ha_myisammrg::check_if_incompatible_data(HA_CREATE_INFO *info,
|
|
|
|
uint table_changes)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
For myisammrg, we should always re-generate the mapping file as this
|
|
|
|
is trivial to do
|
|
|
|
*/
|
|
|
|
return COMPATIBLE_DATA_NO;
|
|
|
|
}
|
2006-04-13 13:49:29 -07:00
|
|
|
|
2007-06-07 13:53:23 +05:00
|
|
|
|
2007-06-06 04:42:41 +05:00
|
|
|
int ha_myisammrg::check(THD* thd, HA_CHECK_OPT* check_opt)
|
|
|
|
{
|
|
|
|
return HA_ADMIN_OK;
|
|
|
|
}
|
2007-06-07 13:53:23 +05:00
|
|
|
|
|
|
|
|
2006-09-30 12:49:46 -07:00
|
|
|
extern int myrg_panic(enum ha_panic_function flag);
|
|
|
|
int myisammrg_panic(handlerton *hton, ha_panic_function flag)
|
|
|
|
{
|
|
|
|
return myrg_panic(flag);
|
|
|
|
}
|
|
|
|
|
2006-09-15 10:28:00 -07:00
|
|
|
static int myisammrg_init(void *p)
|
2006-05-28 14:51:01 +02:00
|
|
|
{
|
2006-09-30 12:49:46 -07:00
|
|
|
handlerton *myisammrg_hton;
|
|
|
|
|
2006-09-15 10:28:00 -07:00
|
|
|
myisammrg_hton= (handlerton *)p;
|
|
|
|
|
2006-09-30 12:49:46 -07:00
|
|
|
myisammrg_hton->db_type= DB_TYPE_MRG_MYISAM;
|
|
|
|
myisammrg_hton->create= myisammrg_create_handler;
|
|
|
|
myisammrg_hton->panic= myisammrg_panic;
|
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Bug 25038 - Waiting TRUNCATE
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Bug 19627 - temporary merge table locking
Bug 27660 - Falcon: merge table possible
Bug 30273 - merge tables: Can't lock file (errno: 155)
The problems were:
Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE
corrupts a MERGE table
1. A thread trying to lock a MERGE table performs busy waiting while
REPAIR TABLE or a similar table administration task is ongoing on
one or more of its MyISAM tables.
2. A thread trying to lock a MERGE table performs busy waiting until all
threads that did REPAIR TABLE or similar table administration tasks
on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK
TABLES. The difference against problem #1 is that the busy waiting
takes place *after* the administration task. It is terminated by
UNLOCK TABLES only.
3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the
lock. This does *not* require a MERGE table. The first FLUSH TABLES
can be replaced by any statement that requires other threads to
reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke
the problem.
Bug 26867 - LOCK TABLES + REPAIR + merge table result in
memory/cpu hogging
Trying DML on a MERGE table, which has a child locked and
repaired by another thread, made an infinite loop in the server.
Bug 26377 - Deadlock with MERGE and FLUSH TABLE
Locking a MERGE table and its children in parent-child order
and flushing the child deadlocked the server.
Bug 25038 - Waiting TRUNCATE
Truncating a MERGE child, while the MERGE table was in use,
let the truncate fail instead of waiting for the table to
become free.
Bug 25700 - merge base tables get corrupted by
optimize/analyze/repair table
Repairing a child of an open MERGE table corrupted the child.
It was necessary to FLUSH the child first.
Bug 30275 - Merge tables: flush tables or unlock tables
causes server to crash
Flushing and optimizing locked MERGE children crashed the server.
Bug 19627 - temporary merge table locking
Use of a temporary MERGE table with non-temporary children
could corrupt the children.
Temporary tables are never locked. So we do now prohibit
non-temporary chidlren of a temporary MERGE table.
Bug 27660 - Falcon: merge table possible
It was possible to create a MERGE table with non-MyISAM children.
Bug 30273 - merge tables: Can't lock file (errno: 155)
This was a Windows-only bug. Table administration statements
sometimes failed with "Can't lock file (errno: 155)".
These bugs are fixed by a new implementation of MERGE table open.
When opening a MERGE table in open_tables() we do now add the
child tables to the list of tables to be opened by open_tables()
(the "query_list"). The children are not opened in the handler at
this stage.
After opening the parent, open_tables() opens each child from the
now extended query_list. When the last child is opened, we remove
the children from the query_list again and attach the children to
the parent. This behaves similar to the old open. However it does
not open the MyISAM tables directly, but grabs them from the already
open children.
When closing a MERGE table in close_thread_table() we detach the
children only. Closing of the children is done implicitly because
they are in thd->open_tables.
For more detail see the comment at the top of ha_myisammrg.cc.
Changed from open_ltable() to open_and_lock_tables() in all places
that can be relevant for MERGE tables. The latter can handle tables
added to the list on the fly. When open_ltable() was used in a loop
over a list of tables, the list must be temporarily terminated
after every table for open_and_lock_tables().
table_list->required_type is set to FRMTYPE_TABLE to avoid open of
special tables. Handling of derived tables is suppressed.
These details are handled by the new function
open_n_lock_single_table(), which has nearly the same signature as
open_ltable() and can replace it in most cases.
In reopen_tables() some of the tables open by a thread can be
closed and reopened. When a MERGE child is affected, the parent
must be closed and reopened too. Closing of the parent is forced
before the first child is closed. Reopen happens in the order of
thd->open_tables. MERGE parents do not attach their children
automatically at open. This is done after all tables are reopened.
So all children are open when attaching them.
Special lock handling like mysql_lock_abort() or mysql_lock_remove()
needs to be suppressed for MERGE children or forwarded to the parent.
This depends on the situation. In loops over all open tables one
suppresses child lock handling. When a single table is touched,
forwarding is done.
Behavioral changes:
===================
This patch changes the behavior of temporary MERGE tables.
Temporary MERGE must have temporary children.
The old behavior was wrong. A temporary table is not locked. Hence
even non-temporary children were not locked. See
Bug 19627 - temporary merge table locking.
You cannot change the union list of a non-temporary MERGE table
when LOCK TABLES is in effect. The following does *not* work:
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...;
LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE;
ALTER TABLE m1 ... UNION=(t1,t2) ...;
However, you can do this with a temporary MERGE table.
You cannot create a MERGE table with CREATE ... SELECT, neither
as a temporary MERGE table, nor as a non-temporary MERGE table.
CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...;
Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
|
|
|
myisammrg_hton->flags= HTON_NO_PARTITION;
|
2006-09-15 10:28:00 -07:00
|
|
|
|
2006-05-28 14:51:01 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
struct st_mysql_storage_engine myisammrg_storage_engine=
|
2006-09-26 21:26:04 -07:00
|
|
|
{ MYSQL_HANDLERTON_INTERFACE_VERSION };
|
2006-05-28 14:51:01 +02:00
|
|
|
|
2006-04-13 13:49:29 -07:00
|
|
|
mysql_declare_plugin(myisammrg)
|
|
|
|
{
|
|
|
|
MYSQL_STORAGE_ENGINE_PLUGIN,
|
2006-05-28 14:51:01 +02:00
|
|
|
&myisammrg_storage_engine,
|
|
|
|
"MRG_MYISAM",
|
2006-04-13 13:49:29 -07:00
|
|
|
"MySQL AB",
|
2006-05-28 14:51:01 +02:00
|
|
|
"Collection of identical MyISAM tables",
|
2006-10-05 00:41:29 -07:00
|
|
|
PLUGIN_LICENSE_GPL,
|
2006-05-28 14:51:01 +02:00
|
|
|
myisammrg_init, /* Plugin Init */
|
2006-04-13 13:49:29 -07:00
|
|
|
NULL, /* Plugin Deinit */
|
2006-05-04 19:39:47 +03:00
|
|
|
0x0100, /* 1.0 */
|
2006-08-30 14:27:29 -07:00
|
|
|
NULL, /* status variables */
|
|
|
|
NULL, /* system variables */
|
|
|
|
NULL /* config options */
|
2006-04-13 13:49:29 -07:00
|
|
|
}
|
|
|
|
mysql_declare_plugin_end;
|