2006-04-11 15:45:10 +02:00
|
|
|
/* Copyright (C) 2006 MySQL AB & MySQL Finland AB & TCX DataKonsult AB
|
|
|
|
|
|
|
|
|
|
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
|
2007-03-02 11:20:23 +01:00
|
|
|
the Free Software Foundation; version 2 of the License.
|
2006-04-11 15:45:10 +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.
|
|
|
|
|
|
|
|
|
|
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 */
|
|
|
|
|
|
|
|
|
|
/* Testing of the basic functions of a MARIA table */
|
|
|
|
|
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
#include "maria_def.h"
|
2006-04-11 15:45:10 +02:00
|
|
|
#include <my_getopt.h>
|
|
|
|
|
#include <m_string.h>
|
2007-06-04 13:07:18 +02:00
|
|
|
#include "ma_control_file.h"
|
|
|
|
|
#include "ma_loghandler.h"
|
WL#3071 Maria checkpoint
Ability for flush_pagecache_blocks() to flush only certain pages of
a file, as instructed by an option "filter" pointer-to-function argument;
Checkpoint and background dirty page flushing use that to flush only
pages which have been dirty for long enough and bitmap pages.
Fix for a bug in flush_cached_blocks() (no idea if it could produce
a bug in real life, but theoretically it is).
Testing checkpoint in ma_test_recovery via ma_test1 and ma_test2.
Background checkpoint & dirty pages flush thread is still disabled
by default in ha_maria.
mysql-test/r/maria.result:
result update
storage/maria/ha_maria.cc:
blank after function comment
storage/maria/ma_checkpoint.c:
Using an enum instead of 0/1/2 (applying Sanja's review comments).
The comment about "this is an horizon" can be removed as Sanja
created translog_next_LSN() which parse_checkpoint_record() uses.
Variables in ma_checkpoint_background() cannot be declared in the
for() as their value must not be reset at each iteration!
storage/maria/ma_pagecache.c:
adding to flush_pagecache_blocks() optional arguments 'filter'
(pointer to function) and 'filter_arg'; if filter!=NULL this function
will be called for each block of the file and will reply if this
block and following ones should be flushed or not (3 possible
replies).
Fixing a bug when flush_cached_blocks() skips a pinned page: it has
to unset PCBLOCK_IN_FLUSH set by flush_pagecache_blocks_int().
storage/maria/ma_pagecache.h:
flush_pagecache_blocks() is changed to take "filter" and "filter_arg"
arguments. "filter", if it is not NULL, may return one value
among enum pagecache_flush_filter_result.
storage/maria/ma_recovery.c:
open_count=0 when closing tables at the end of recovery.
storage/maria/ma_test1.c:
Optional checkpoints (-H#) at various stages (stages similar
to --testflag), for testing of checkpoints.
storage/maria/ma_test2.c:
Optional checkpoints (-H#) at various stages (stages similar
to -t), for testing of checkpoints.
storage/maria/ma_test_recovery.expected:
Result update: the results of the additional test run with -H#
(checkpoints) are added here. They are exactly identical to without
checkpoints except that the index's Root (printed by maria_chk)
is more correct when using checkpoints. This is because checkpoint
flushed the state, so it happens to be correct, while no-checkpoint
does not flush the state, and recovery does not recover indexes
so Root is never fixed. When we recover indices, this will go away.
storage/maria/ma_test_recovery:
We duplicate the loop of tests to add an additional run with
checkpoints at various stages, to see if maria_read_log
uses them fine.
2007-10-17 16:55:26 +02:00
|
|
|
#include "ma_checkpoint.h"
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
#include "trnman.h"
|
2007-06-04 13:07:18 +02:00
|
|
|
|
|
|
|
|
extern PAGECACHE *maria_log_pagecache;
|
|
|
|
|
extern const char *maria_data_root;
|
2006-04-11 15:45:10 +02:00
|
|
|
|
|
|
|
|
#define MAX_REC_LENGTH 1024
|
|
|
|
|
|
|
|
|
|
static void usage();
|
|
|
|
|
|
WL#3071 Maria checkpoint
Ability for flush_pagecache_blocks() to flush only certain pages of
a file, as instructed by an option "filter" pointer-to-function argument;
Checkpoint and background dirty page flushing use that to flush only
pages which have been dirty for long enough and bitmap pages.
Fix for a bug in flush_cached_blocks() (no idea if it could produce
a bug in real life, but theoretically it is).
Testing checkpoint in ma_test_recovery via ma_test1 and ma_test2.
Background checkpoint & dirty pages flush thread is still disabled
by default in ha_maria.
mysql-test/r/maria.result:
result update
storage/maria/ha_maria.cc:
blank after function comment
storage/maria/ma_checkpoint.c:
Using an enum instead of 0/1/2 (applying Sanja's review comments).
The comment about "this is an horizon" can be removed as Sanja
created translog_next_LSN() which parse_checkpoint_record() uses.
Variables in ma_checkpoint_background() cannot be declared in the
for() as their value must not be reset at each iteration!
storage/maria/ma_pagecache.c:
adding to flush_pagecache_blocks() optional arguments 'filter'
(pointer to function) and 'filter_arg'; if filter!=NULL this function
will be called for each block of the file and will reply if this
block and following ones should be flushed or not (3 possible
replies).
Fixing a bug when flush_cached_blocks() skips a pinned page: it has
to unset PCBLOCK_IN_FLUSH set by flush_pagecache_blocks_int().
storage/maria/ma_pagecache.h:
flush_pagecache_blocks() is changed to take "filter" and "filter_arg"
arguments. "filter", if it is not NULL, may return one value
among enum pagecache_flush_filter_result.
storage/maria/ma_recovery.c:
open_count=0 when closing tables at the end of recovery.
storage/maria/ma_test1.c:
Optional checkpoints (-H#) at various stages (stages similar
to --testflag), for testing of checkpoints.
storage/maria/ma_test2.c:
Optional checkpoints (-H#) at various stages (stages similar
to -t), for testing of checkpoints.
storage/maria/ma_test_recovery.expected:
Result update: the results of the additional test run with -H#
(checkpoints) are added here. They are exactly identical to without
checkpoints except that the index's Root (printed by maria_chk)
is more correct when using checkpoints. This is because checkpoint
flushed the state, so it happens to be correct, while no-checkpoint
does not flush the state, and recovery does not recover indexes
so Root is never fixed. When we recover indices, this will go away.
storage/maria/ma_test_recovery:
We duplicate the loop of tests to add an additional run with
checkpoints at various stages, to see if maria_read_log
uses them fine.
2007-10-17 16:55:26 +02:00
|
|
|
static int rec_pointer_size=0, flags[50], testflag, checkpoint;
|
2006-04-11 15:45:10 +02:00
|
|
|
static int key_field=FIELD_SKIP_PRESPACE,extra_field=FIELD_SKIP_ENDSPACE;
|
|
|
|
|
static int key_type=HA_KEYTYPE_NUM;
|
|
|
|
|
static int create_flag=0;
|
2007-10-19 23:24:22 +02:00
|
|
|
static ulong blob_length;
|
2007-01-18 20:38:14 +01:00
|
|
|
static enum data_file_type record_type= DYNAMIC_RECORD;
|
2006-04-11 15:45:10 +02:00
|
|
|
|
|
|
|
|
static uint insert_count, update_count, remove_count;
|
|
|
|
|
static uint pack_keys=0, pack_seg=0, key_length;
|
|
|
|
|
static uint unique_key=HA_NOSAME;
|
2007-09-05 01:57:53 +02:00
|
|
|
static uint die_in_middle_of_transaction;
|
|
|
|
|
static my_bool pagecacheing, null_fields, silent, skip_update, opt_unique;
|
|
|
|
|
static my_bool verbose, skip_delete, transactional;
|
2006-04-11 15:45:10 +02:00
|
|
|
static MARIA_COLUMNDEF recinfo[4];
|
|
|
|
|
static MARIA_KEYDEF keyinfo[10];
|
|
|
|
|
static HA_KEYSEG keyseg[10];
|
|
|
|
|
static HA_KEYSEG uniqueseg[10];
|
|
|
|
|
|
|
|
|
|
static int run_test(const char *filename);
|
|
|
|
|
static void get_options(int argc, char *argv[]);
|
2007-10-04 19:33:42 +02:00
|
|
|
static void create_key(uchar *key,uint rownr);
|
|
|
|
|
static void create_record(uchar *record,uint rownr);
|
|
|
|
|
static void update_record(uchar *record);
|
2006-04-11 15:45:10 +02:00
|
|
|
|
2007-09-04 09:53:52 +02:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
These are here only for testing of recovery with undo. We are not
|
|
|
|
|
including maria_def.h here as this test is also to be an example of
|
|
|
|
|
how to use maria outside of the maria directory
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
extern int _ma_flush_table_files(MARIA_HA *info, uint flush_data_or_index,
|
|
|
|
|
enum flush_type flush_type_for_data,
|
|
|
|
|
enum flush_type flush_type_for_index);
|
|
|
|
|
#define MARIA_FLUSH_DATA 1
|
|
|
|
|
|
|
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
int main(int argc,char *argv[])
|
|
|
|
|
{
|
|
|
|
|
MY_INIT(argv[0]);
|
|
|
|
|
my_init();
|
|
|
|
|
get_options(argc,argv);
|
2007-06-04 13:07:18 +02:00
|
|
|
maria_data_root= ".";
|
2007-04-19 17:48:36 +02:00
|
|
|
/* Maria requires that we always have a page cache */
|
2007-06-04 13:07:18 +02:00
|
|
|
if (maria_init() ||
|
First part of redo/undo for key pages
Added key_nr to st_maria_keydef for faster keyinfo->keynr conversion
For transactional tables, shift record number in keys up with 1 bit to have place to indicate if transid follows
Checksum for MyISAM now ignores NULL and not used part of VARCHAR
Renamed some variables that caused shadow compiler warnings
Moved extra() call when waiting for tables to not be used to after tables are removed from cache.
Fixed crashing bugs when using Maria TEMPORARY tables with TRUNCATE. Removed 'hack' code in sql directory to go around this bug.
pagecache_unlock_by_ulink() now has extra argument to say if page was changed.
Give error message if we fail to open control file
Mark page cache variables as not flushable
include/maria.h:
Made min page cache larger (needed for pinning key page)
Added key_nr to st_maria_keydef for faster keyinfo->keynr conversion
Added write_comp_flag to move some runtime code to maria_open()
include/my_base.h:
Added new error message to be used when handler initialization failed
include/my_global.h:
Renamed dummy to swap_dummy to avoid conflicts with local 'dummy' variables
include/my_handler.h:
Added const to some parameters
mysys/array.c:
More DBUG
mysys/my_error.c:
Fixed indentation
mysys/my_handler.c:
Added const to some parameters
Added missing error messages
sql/field.h:
Renamed variables to avoid variable shadowing
sql/handler.h:
Renamed parameter to avoid variable name conflict
sql/item.h:
Renamed variables to avoid variable shadowing
sql/log_event_old.h:
Renamed variables to avoid variable shadowing
sql/set_var.h:
Renamed variables to avoid variable shadowing
sql/sql_delete.cc:
Removed maria hack for temporary tables
Fixed indentation
sql/sql_table.cc:
Moved extra() call when waiting for tables to not be used to after tables are removed from cache.
This was needed to ensure we don't do a PREPARE_FOR_DROP or similar call while the table is still in use.
sql/table.cc:
Copy page_checksum from share
Removed Maria hack
storage/maria/Makefile.am:
Added new files
storage/maria/ha_maria.cc:
Renamed records -> record_count and info -> create_info to avoid variable name conflicts
Mark page cache variables as not flushable
storage/maria/ma_blockrec.c:
Moved _ma_unpin_all_pages() to ma_key_recover.c
Moved init of info->pinned_pages to ma_open.c
Moved _ma_finalize_row() to maria_key_recover.h
Renamed some variables to avoid variable name conflicts
Mark page_link.changed for blocks we change directly
Simplify handling of undo link when writing LOGREC_UNDO_ROW_INSERT (old code crashed when having redo for index)
storage/maria/ma_blockrec.h:
Removed extra empty line
storage/maria/ma_checkpoint.c:
Remove not needed trnman.h
storage/maria/ma_close.c:
Free pinned pages (which are now always allocated)
storage/maria/ma_control_file.c:
Give error message if we fail to open control file
storage/maria/ma_delete.c:
Changes for redo logging (first part, logging of underflow not yet done)
- Log undo-key-delete
- Log delete of key
- Updated arguments to _ma_fetch_keypage(), _ma_dispose(), _ma_write_keypage(), _ma_insert()
- Added new arguments to some functions to be able to write redo information
- Mark key pages as changed when we write with PAGECACHE_LOCK_LEFT_WRITELOCKED
Remove one not needed _ma_write_keypage() in d_search() when upper level will do the write anyway
Changed 2 bmove_upp() to bmove() as this made code easer to understand
More function comments
Indentation fixes
storage/maria/ma_ft_update.c:
New arguments to _ma_write_keypage()
storage/maria/ma_loghandler.c:
Fixed some DBUG_PRINT messages
Simplify code
Added new log entrys for key page redo
Renamed some variables to avoid variable name shadowing
storage/maria/ma_loghandler.h:
Moved some defines here
Added define for storing key number on key pages
Added new translog record types
Added enum for type of operations in LOGREC_REDO_INDEX
storage/maria/ma_open.c:
Always allocate info.pinned_pages (we need now also for normal key page usage)
Update keyinfo->key_nr
Added virtual functions to convert record position o number to be stored on key pages
Update keyinfo->write_comp_flag to value of search flag to be used when writing key
storage/maria/ma_page.c:
Added redo for key pages
- Extended _ma_fetch_keypage() with type of lock to put on page and address to used MARIA_PINNED_PAGE
- _ma_fetch_keypage() now pin's pages if needed
- Extended _ma_write_keypage() with type of locks to be used
- ma_dispose() now locks info->s->state.key_del from other threads
- ma_dispose() writes redo log record
- ma_new() locks info->s->state.key_del from other threads if it was used
- ma_new() now pins read page
Other things:
- Removed some not needed arguments from _ma_new() and _ma_dispose)
- Added some new variables to simplify code
- If EXTRA_DEBUG is used, do crc on full page to catch not unitialized bytes
storage/maria/ma_pagecache.h:
Applied patch from Sanja to add extra argument to pagecache_unlock_by_ulink() to mark if page was changed
Added some defines for pagecache priority levels that one can use
storage/maria/ma_range.c:
Added new arguments for call to _ma_fetch_keypage()
storage/maria/ma_recovery.c:
- Added hooks for new translog types:
REDO_INDEX, REDO_INDEX_NEW_PAGE, REDO_INDEX_FREE_PAGE, UNDO_KEY_INSERT, UNDO_KEY_DELETE and
UNDO_KEY_DELETE_WITH_ROOT.
- Moved variable declarations to start of function (portability fixes)
- Removed some not needed initializations
- Set only relevant state changes for each redo/undo entry
storage/maria/lockman.c:
Removed end space
storage/maria/ma_check.c:
Removed end space
storage/maria/ma_create.c:
Removed end space
storage/maria/ma_locking.c:
Removed end space
storage/maria/ma_packrec.c:
Removed end space
storage/maria/ma_pagecache.c:
Removed end space
storage/maria/ma_panic.c:
Removed end space
storage/maria/ma_rt_index.c:
Added new arguments for call to _ma_fetch_keypage(), _ma_write_keypage(), _ma_dispose() and _ma_new()
Fixed indentation
storage/maria/ma_rt_key.c:
Added new arguments for call to _ma_fetch_keypage()
storage/maria/ma_rt_split.c:
Added new arguments for call to _ma_new()
Use new keypage header
Added new arguments for call to _ma_write_keypage()
storage/maria/ma_search.c:
Updated comments & indentation
Added new arguments for call to _ma_fetch_keypage()
Made some variables and arguments const
Added virtual functions for converting row position to number to be stored in key
use MARIA_RECORD_POS of record position instead of my_off_t
Record in MARIA_KEY_PARAM how page was changed one key insert (needed for REDO)
storage/maria/ma_sort.c:
Removed end space
storage/maria/ma_statrec.c:
Updated arguments for call to _ma_rec_pos()
storage/maria/ma_test1.c:
Fixed too small buffer to init_pagecache()
Fixed bug when using insert_count and test_flag
storage/maria/ma_test2.c:
Use more resonable pagecache size
Remove not used code
Reset blob_length to fix wrong output message
storage/maria/ma_test_all.sh:
Fixed wrong test
storage/maria/ma_write.c:
Lots of new code to handle REDO of key pages
No logic changes because of REDO code, mostly adding new arguments and adding new code for logging
Added new arguments for calls to _ma_fetch_keypage(), _ma_write_keypage() and similar functions
Move setting of comp_flag in ma_ck_wrte_btree() from runtime to maria_open()
Zerofill new used pages for:
- To remove possible sensitive data left in buffer
- To get idenitical data on pages after running redo
- Better compression of pages if archived
storage/maria/maria_chk.c:
Added information if table is crash safe
storage/maria/maria_def.h:
New virtual function to convert between record position on key and normal record position
Aded mutex and extra variables to handle locking of share->state.key_del
Moved some structure variables to get things more aligned
Added extra arguments to MARIA_KEY_PARAM to be able to remember what was changed on key page on key insert
Added argument to MARIA_PINNED_PAGE to indicate if page was changed
Updated prototypes for functions
Added some structures for signaling changes in REDO handling
storage/maria/unittest/ma_pagecache_single.c:
Updated arguments for changed function calls
storage/myisam/mi_check.c:
Made calc_check_checksum virtual
storage/myisam/mi_checksum.c:
Update checksums to ignore null columns
storage/myisam/mi_create.c:
Mark if table has null column (to know when we have to use mi_checksum())
storage/myisam/mi_open.c:
Added virtual function for calculating checksum to be able to easily ignore NULL fields
storage/myisam/mi_test2.c:
Fixed bug
storage/myisam/myisamdef.h:
Added virtual function for calculating checksum during check table
Removed ha_key_cmp() as this is in handler.h
storage/maria/ma_key_recover.c:
New BitKeeper file ``storage/maria/ma_key_recover.c''
storage/maria/ma_key_recover.h:
New BitKeeper file ``storage/maria/ma_key_recover.h''
storage/maria/ma_key_redo.c:
New BitKeeper file ``storage/maria/ma_key_redo.c''
2007-11-14 18:08:06 +01:00
|
|
|
(init_pagecache(maria_pagecache, maria_block_size * 16, 0, 0,
|
2007-12-04 22:23:42 +01:00
|
|
|
maria_block_size, MY_WME) == 0) ||
|
- WL#3072 Maria Recovery:
Recovery of state.records (the count of records which is stored into
the header of the index file). For that, state.is_of_lsn is introduced;
logic is explained in ma_recovery.c (look for "Recovery of the state").
The net gain is that in case of crash, we now recover state.records,
and it is idempotent (ma_test_recovery tests it).
state.checksum is not recovered yet, mail sent for discussion.
- WL#3071 Maria Checkpoint: preparation for it, by protecting
all modifications of the state in memory or on disk with intern_lock
(with the exception of the really-often-modified state.records,
which is now protected with the log's lock, see ma_recovery.c
(look for "Recovery of the state"). Also, if maria_close() sees that
Checkpoint is looking at this table it will not my_free() the share.
- don't compute row's checksum twice in case of UPDATE (correction
to a bugfix I made yesterday).
storage/maria/ha_maria.cc:
protect state write with intern_lock (against Checkpoint)
storage/maria/ma_blockrec.c:
* don't reset trn->rec_lsn in _ma_unpin_all_pages(), because it
should wait until we have corrected the allocation in the bitmap
(as the REDO can serve to correct the allocation during Recovery);
introducing _ma_finalize_row() for that.
* In a changeset yesterday I moved computation of the checksum
into write_block_record(), to fix a bug in UPDATE. Now I notice
that maria_update() already computes the checksum, it's just that
it puts it into info->cur_row while _ma_update_block_record()
uses info->new_row; so, removing the checksum computation from
write_block_record(), putting it back into allocate_and_write_block_record()
(which is called only by INSERT and UNDO_DELETE), and copying
cur_row->checksum into new_row->checksum in _ma_update_block_record().
storage/maria/ma_check.c:
new prototypes, they will take intern_lock when writing the state;
also take intern_lock when changing share->kfile. In both cases
this is to protect against Checkpoint reading/writing the state or reading
kfile at the same time.
Not updating create_rename_lsn directly at end of write_log_record_for_repair()
as it wouldn't have intern_lock.
storage/maria/ma_close.c:
Checkpoint builds a list of shares (under THR_LOCK_maria), then it
handles each such share (under intern_lock) (doing flushing etc);
if maria_close() freed this share between the two, Checkpoint
would see a bad pointer. To avoid this, when building the list Checkpoint
marks each share, so that maria_close() knows it should not free it
and Checkpoint will free it itself.
Extending the zone covered by intern_lock to protect against
Checkpoint reading kfile, writing state.
storage/maria/ma_create.c:
When we update create_rename_lsn, we also update is_of_lsn to
the same value: it is logical, and allows us to test in maria_open()
that the former is not bigger than the latter (the contrary is a sign
of index header corruption, or severe logging bug which hinders
Recovery, table needs a repair).
_ma_update_create_rename_lsn_on_disk() also writes is_of_lsn;
it now operates under intern_lock (protect against Checkpoint),
a shortcut function is available for cases where acquiring
intern_lock is not needed (table's creation or first open).
storage/maria/ma_delete.c:
if table is transactional, "records" is already decremented
when logging UNDO_ROW_DELETE.
storage/maria/ma_delete_all.c:
comments
storage/maria/ma_extra.c:
Protect modifications of the state, in memory and/or on disk,
with intern_lock, against a concurrent Checkpoint.
When state goes to disk, update it's is_of_lsn (by calling
the new _ma_state_info_write()).
In HA_EXTRA_FORCE_REOPEN, don't set share->changed to 0 (undoing
a change I made a few days ago) and ASK_MONTY
storage/maria/ma_locking.c:
no real code change here.
storage/maria/ma_loghandler.c:
Log-write-hooks for updating "state.records" under log's mutex
when writing/updating/deleting a row or deleting all rows.
storage/maria/ma_loghandler_lsn.h:
merge (make LSN_ERROR and LSN_REPAIRED_BY_MARIA_CHK different)
storage/maria/ma_open.c:
When opening a table verify that is_of_lsn >= create_rename_lsn; if
false the header must be corrupted.
_ma_state_info_write() is split in two: _ma_state_info_write_sub()
which is the old _ma_state_info_write(), and _ma_state_info_write()
which additionally takes intern_lock if requested (to protect
against Checkpoint) and updates is_of_lsn.
_ma_open_keyfile() should change kfile.file under intern_lock
to protect Checkpoint from reading a wrong kfile.file.
storage/maria/ma_recovery.c:
Recovery of state.records: when the REDO phase sees UNDO_ROW_INSERT
which has a LSN > state.is_of_lsn it increments state.records.
Same for UNDO_ROW_DELETE and UNDO_ROW_PURGE.
When closing a table during Recovery, we know its state is at least
as new as the current log record we are looking at, so increase
is_of_lsn to the LSN of the current log record.
storage/maria/ma_rename.c:
update for new behaviour of _ma_update_create_rename_lsn_on_disk().
storage/maria/ma_test1.c:
update to new prototype
storage/maria/ma_test2.c:
update to new prototype (actually prototype was changed days ago,
but compiler does not complain about the extra argument??)
storage/maria/ma_test_recovery.expected:
new result file of ma_test_recovery. Improvements: record
count read from index's header is now always correct.
storage/maria/ma_test_recovery:
"rm" fails if file does not exist. Redirect stderr of script.
storage/maria/ma_write.c:
if table is transactional, "records" is already incremented when
logging UNDO_ROW_INSERT. Comments.
storage/maria/maria_chk.c:
update is_of_lsn too
storage/maria/maria_def.h:
- MARIA_STATE_INFO::is_of_lsn which is used by Recovery. It is stored
into the index file's header.
- Checkpoint can now mark a table as "don't free this", and maria_close()
can reply "ok then you will free it".
- new functions
storage/maria/maria_pack.c:
update for new name
2007-09-07 15:02:30 +02:00
|
|
|
ma_control_file_create_or_open() ||
|
2007-06-04 13:07:18 +02:00
|
|
|
(init_pagecache(maria_log_pagecache,
|
|
|
|
|
TRANSLOG_PAGECACHE_SIZE, 0, 0,
|
2007-12-04 22:23:42 +01:00
|
|
|
TRANSLOG_PAGE_SIZE, MY_WME) == 0) ||
|
2007-06-04 13:07:18 +02:00
|
|
|
translog_init(maria_data_root, TRANSLOG_FILE_SIZE,
|
|
|
|
|
0, 0, maria_log_pagecache,
|
2007-12-15 14:44:29 +01:00
|
|
|
TRANSLOG_DEFAULT_FLAGS, 0) ||
|
WL#3071 - Maria checkpoint
- serializing calls to flush_pagecache_blocks_int() on the same file
to avoid known concurrency bugs
- having that, we can now enable the background thread, as the
flushes it does are now supposedly safe in concurrent situations.
- new type of flush FLUSH_KEEP_LAZY: when the background checkpoint
thread is flushing a packet of dirty pages between two checkpoints,
it uses this flush type, indeed if a file is already being flushed
by another thread it's smarter to move on to the next file than wait.
- maria_checkpoint_frequency renamed to maria_checkpoint_interval.
include/my_sys.h:
new type of flushing for the page cache: FLUSH_KEEP_LAZY
mysql-test/r/maria.result:
result update
mysys/mf_keycache.c:
indentation. No FLUSH_KEEP_LAZY support in key cache.
storage/maria/ha_maria.cc:
maria_checkpoint_frequency was somehow a hidden part of the
Checkpoint API and that was not good. Now we have checkpoint_interval,
local to ha_maria.cc, which serves as container for the user-visible
maria_checkpoint_interval global variable; setting it calls
update_checkpoint_interval which passes the new value to
ma_checkpoint_init(). There is no hiding anymore.
By default, enable background thread which does checkpoints
every 30 seconds, and dirty page flush in between. That thread takes
a checkpoint when it ends, so no need for maria_hton_panic to take one.
The | is | and not ||, because maria_panic() must always be called.
frequency->interval.
storage/maria/ma_checkpoint.c:
Use FLUSH_KEEP_LAZY for background thread when it flushes packets of
dirty pages between two checkpoints: it is smarter to move on to
the next file than wait for it to have been completely flushed, which
may take long.
Comments about flush concurrency bugs moved from ma_pagecache.c.
Removing out-of-date comment.
frequency->interval.
create_background_thread -> (interval>0).
In ma_checkpoint_background(), some variables need to be preserved
between iterations.
storage/maria/ma_checkpoint.h:
new prototype
storage/maria/ma_pagecache.c:
- concurrent calls of flush_pagecache_blocks_int() on the same file
cause bugs (see @note in that function); we fix them by serializing
in this situation. For that we use a global hash of (file, wqueue).
When flush_pagecache_blocks_int() starts it looks into the hash,
using the file as key. If not found, it inserts (file,wqueue) into the
hash, flushes the file, and finally removes itself from the hash and
wakes up any waiter in the queue. If found, it adds itself to the
wqueue and waits.
- As a by-product, we can remove changed_blocks_is_incomplete
and replace it by scanning the hash, replace the sleep() by a queue wait.
- new type of flush FLUSH_KEEP_LAZY: when flushing a file, if it's
already being flushed by another thread (even partially), return
immediately.
storage/maria/ma_pagecache.h:
In pagecache, a hash of files currently being flushed (i.e. there
is a call to flush_pagecache_blocks_int() for them).
storage/maria/ma_recovery.c:
new prototype
storage/maria/ma_test1.c:
new prototype
storage/maria/ma_test2.c:
new prototype
2007-10-19 14:15:13 +02:00
|
|
|
(transactional && (trnman_init(0) || ma_checkpoint_init(0))))
|
2007-06-04 13:07:18 +02:00
|
|
|
{
|
First part of redo/undo for key pages
Added key_nr to st_maria_keydef for faster keyinfo->keynr conversion
For transactional tables, shift record number in keys up with 1 bit to have place to indicate if transid follows
Checksum for MyISAM now ignores NULL and not used part of VARCHAR
Renamed some variables that caused shadow compiler warnings
Moved extra() call when waiting for tables to not be used to after tables are removed from cache.
Fixed crashing bugs when using Maria TEMPORARY tables with TRUNCATE. Removed 'hack' code in sql directory to go around this bug.
pagecache_unlock_by_ulink() now has extra argument to say if page was changed.
Give error message if we fail to open control file
Mark page cache variables as not flushable
include/maria.h:
Made min page cache larger (needed for pinning key page)
Added key_nr to st_maria_keydef for faster keyinfo->keynr conversion
Added write_comp_flag to move some runtime code to maria_open()
include/my_base.h:
Added new error message to be used when handler initialization failed
include/my_global.h:
Renamed dummy to swap_dummy to avoid conflicts with local 'dummy' variables
include/my_handler.h:
Added const to some parameters
mysys/array.c:
More DBUG
mysys/my_error.c:
Fixed indentation
mysys/my_handler.c:
Added const to some parameters
Added missing error messages
sql/field.h:
Renamed variables to avoid variable shadowing
sql/handler.h:
Renamed parameter to avoid variable name conflict
sql/item.h:
Renamed variables to avoid variable shadowing
sql/log_event_old.h:
Renamed variables to avoid variable shadowing
sql/set_var.h:
Renamed variables to avoid variable shadowing
sql/sql_delete.cc:
Removed maria hack for temporary tables
Fixed indentation
sql/sql_table.cc:
Moved extra() call when waiting for tables to not be used to after tables are removed from cache.
This was needed to ensure we don't do a PREPARE_FOR_DROP or similar call while the table is still in use.
sql/table.cc:
Copy page_checksum from share
Removed Maria hack
storage/maria/Makefile.am:
Added new files
storage/maria/ha_maria.cc:
Renamed records -> record_count and info -> create_info to avoid variable name conflicts
Mark page cache variables as not flushable
storage/maria/ma_blockrec.c:
Moved _ma_unpin_all_pages() to ma_key_recover.c
Moved init of info->pinned_pages to ma_open.c
Moved _ma_finalize_row() to maria_key_recover.h
Renamed some variables to avoid variable name conflicts
Mark page_link.changed for blocks we change directly
Simplify handling of undo link when writing LOGREC_UNDO_ROW_INSERT (old code crashed when having redo for index)
storage/maria/ma_blockrec.h:
Removed extra empty line
storage/maria/ma_checkpoint.c:
Remove not needed trnman.h
storage/maria/ma_close.c:
Free pinned pages (which are now always allocated)
storage/maria/ma_control_file.c:
Give error message if we fail to open control file
storage/maria/ma_delete.c:
Changes for redo logging (first part, logging of underflow not yet done)
- Log undo-key-delete
- Log delete of key
- Updated arguments to _ma_fetch_keypage(), _ma_dispose(), _ma_write_keypage(), _ma_insert()
- Added new arguments to some functions to be able to write redo information
- Mark key pages as changed when we write with PAGECACHE_LOCK_LEFT_WRITELOCKED
Remove one not needed _ma_write_keypage() in d_search() when upper level will do the write anyway
Changed 2 bmove_upp() to bmove() as this made code easer to understand
More function comments
Indentation fixes
storage/maria/ma_ft_update.c:
New arguments to _ma_write_keypage()
storage/maria/ma_loghandler.c:
Fixed some DBUG_PRINT messages
Simplify code
Added new log entrys for key page redo
Renamed some variables to avoid variable name shadowing
storage/maria/ma_loghandler.h:
Moved some defines here
Added define for storing key number on key pages
Added new translog record types
Added enum for type of operations in LOGREC_REDO_INDEX
storage/maria/ma_open.c:
Always allocate info.pinned_pages (we need now also for normal key page usage)
Update keyinfo->key_nr
Added virtual functions to convert record position o number to be stored on key pages
Update keyinfo->write_comp_flag to value of search flag to be used when writing key
storage/maria/ma_page.c:
Added redo for key pages
- Extended _ma_fetch_keypage() with type of lock to put on page and address to used MARIA_PINNED_PAGE
- _ma_fetch_keypage() now pin's pages if needed
- Extended _ma_write_keypage() with type of locks to be used
- ma_dispose() now locks info->s->state.key_del from other threads
- ma_dispose() writes redo log record
- ma_new() locks info->s->state.key_del from other threads if it was used
- ma_new() now pins read page
Other things:
- Removed some not needed arguments from _ma_new() and _ma_dispose)
- Added some new variables to simplify code
- If EXTRA_DEBUG is used, do crc on full page to catch not unitialized bytes
storage/maria/ma_pagecache.h:
Applied patch from Sanja to add extra argument to pagecache_unlock_by_ulink() to mark if page was changed
Added some defines for pagecache priority levels that one can use
storage/maria/ma_range.c:
Added new arguments for call to _ma_fetch_keypage()
storage/maria/ma_recovery.c:
- Added hooks for new translog types:
REDO_INDEX, REDO_INDEX_NEW_PAGE, REDO_INDEX_FREE_PAGE, UNDO_KEY_INSERT, UNDO_KEY_DELETE and
UNDO_KEY_DELETE_WITH_ROOT.
- Moved variable declarations to start of function (portability fixes)
- Removed some not needed initializations
- Set only relevant state changes for each redo/undo entry
storage/maria/lockman.c:
Removed end space
storage/maria/ma_check.c:
Removed end space
storage/maria/ma_create.c:
Removed end space
storage/maria/ma_locking.c:
Removed end space
storage/maria/ma_packrec.c:
Removed end space
storage/maria/ma_pagecache.c:
Removed end space
storage/maria/ma_panic.c:
Removed end space
storage/maria/ma_rt_index.c:
Added new arguments for call to _ma_fetch_keypage(), _ma_write_keypage(), _ma_dispose() and _ma_new()
Fixed indentation
storage/maria/ma_rt_key.c:
Added new arguments for call to _ma_fetch_keypage()
storage/maria/ma_rt_split.c:
Added new arguments for call to _ma_new()
Use new keypage header
Added new arguments for call to _ma_write_keypage()
storage/maria/ma_search.c:
Updated comments & indentation
Added new arguments for call to _ma_fetch_keypage()
Made some variables and arguments const
Added virtual functions for converting row position to number to be stored in key
use MARIA_RECORD_POS of record position instead of my_off_t
Record in MARIA_KEY_PARAM how page was changed one key insert (needed for REDO)
storage/maria/ma_sort.c:
Removed end space
storage/maria/ma_statrec.c:
Updated arguments for call to _ma_rec_pos()
storage/maria/ma_test1.c:
Fixed too small buffer to init_pagecache()
Fixed bug when using insert_count and test_flag
storage/maria/ma_test2.c:
Use more resonable pagecache size
Remove not used code
Reset blob_length to fix wrong output message
storage/maria/ma_test_all.sh:
Fixed wrong test
storage/maria/ma_write.c:
Lots of new code to handle REDO of key pages
No logic changes because of REDO code, mostly adding new arguments and adding new code for logging
Added new arguments for calls to _ma_fetch_keypage(), _ma_write_keypage() and similar functions
Move setting of comp_flag in ma_ck_wrte_btree() from runtime to maria_open()
Zerofill new used pages for:
- To remove possible sensitive data left in buffer
- To get idenitical data on pages after running redo
- Better compression of pages if archived
storage/maria/maria_chk.c:
Added information if table is crash safe
storage/maria/maria_def.h:
New virtual function to convert between record position on key and normal record position
Aded mutex and extra variables to handle locking of share->state.key_del
Moved some structure variables to get things more aligned
Added extra arguments to MARIA_KEY_PARAM to be able to remember what was changed on key page on key insert
Added argument to MARIA_PINNED_PAGE to indicate if page was changed
Updated prototypes for functions
Added some structures for signaling changes in REDO handling
storage/maria/unittest/ma_pagecache_single.c:
Updated arguments for changed function calls
storage/myisam/mi_check.c:
Made calc_check_checksum virtual
storage/myisam/mi_checksum.c:
Update checksums to ignore null columns
storage/myisam/mi_create.c:
Mark if table has null column (to know when we have to use mi_checksum())
storage/myisam/mi_open.c:
Added virtual function for calculating checksum to be able to easily ignore NULL fields
storage/myisam/mi_test2.c:
Fixed bug
storage/myisam/myisamdef.h:
Added virtual function for calculating checksum during check table
Removed ha_key_cmp() as this is in handler.h
storage/maria/ma_key_recover.c:
New BitKeeper file ``storage/maria/ma_key_recover.c''
storage/maria/ma_key_recover.h:
New BitKeeper file ``storage/maria/ma_key_recover.h''
storage/maria/ma_key_redo.c:
New BitKeeper file ``storage/maria/ma_key_redo.c''
2007-11-14 18:08:06 +01:00
|
|
|
fprintf(stderr, "Error in initialization\n");
|
2007-06-04 13:07:18 +02:00
|
|
|
exit(1);
|
|
|
|
|
}
|
2006-04-11 15:45:10 +02:00
|
|
|
|
|
|
|
|
exit(run_test("test1"));
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
static int run_test(const char *filename)
|
|
|
|
|
{
|
|
|
|
|
MARIA_HA *file;
|
2007-11-20 16:42:16 +01:00
|
|
|
int i,j= 0,error,deleted,rec_length,uniques=0;
|
2007-09-05 01:57:53 +02:00
|
|
|
uint offset_to_key;
|
2006-04-11 15:45:10 +02:00
|
|
|
ha_rows found,row_count;
|
2007-10-04 19:33:42 +02:00
|
|
|
uchar record[MAX_REC_LENGTH],key[MAX_REC_LENGTH],read_record[MAX_REC_LENGTH];
|
2006-04-11 15:45:10 +02:00
|
|
|
MARIA_UNIQUEDEF uniquedef;
|
|
|
|
|
MARIA_CREATE_INFO create_info;
|
|
|
|
|
|
2007-09-04 09:53:52 +02:00
|
|
|
if (die_in_middle_of_transaction)
|
|
|
|
|
null_fields= 1;
|
|
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
bzero((char*) recinfo,sizeof(recinfo));
|
2007-01-18 20:38:14 +01:00
|
|
|
bzero((char*) &create_info,sizeof(create_info));
|
2006-04-11 15:45:10 +02:00
|
|
|
|
|
|
|
|
/* First define 2 columns */
|
2007-01-18 20:38:14 +01:00
|
|
|
create_info.null_bytes= 1;
|
|
|
|
|
recinfo[0].type= key_field;
|
2007-04-19 12:18:56 +02:00
|
|
|
recinfo[0].length= (key_field == FIELD_BLOB ? 4+portable_sizeof_char_ptr :
|
2006-04-11 15:45:10 +02:00
|
|
|
key_length);
|
|
|
|
|
if (key_field == FIELD_VARCHAR)
|
2007-01-18 20:38:14 +01:00
|
|
|
recinfo[0].length+= HA_VARCHAR_PACKLENGTH(key_length);
|
|
|
|
|
recinfo[1].type=extra_field;
|
2007-04-19 12:18:56 +02:00
|
|
|
recinfo[1].length= (extra_field == FIELD_BLOB ? 4 + portable_sizeof_char_ptr : 24);
|
2006-04-11 15:45:10 +02:00
|
|
|
if (extra_field == FIELD_VARCHAR)
|
2007-01-18 20:38:14 +01:00
|
|
|
recinfo[1].length+= HA_VARCHAR_PACKLENGTH(recinfo[1].length);
|
2007-09-11 00:58:15 +02:00
|
|
|
recinfo[1].null_bit= null_fields ? 2 : 0;
|
|
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
if (opt_unique)
|
|
|
|
|
{
|
2007-01-18 20:38:14 +01:00
|
|
|
recinfo[2].type=FIELD_CHECK;
|
|
|
|
|
recinfo[2].length=MARIA_UNIQUE_HASH_LENGTH;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
2007-01-18 20:38:14 +01:00
|
|
|
rec_length= recinfo[0].length+recinfo[1].length+recinfo[2].length;
|
2006-04-11 15:45:10 +02:00
|
|
|
|
|
|
|
|
if (key_type == HA_KEYTYPE_VARTEXT1 &&
|
|
|
|
|
key_length > 255)
|
|
|
|
|
key_type= HA_KEYTYPE_VARTEXT2;
|
|
|
|
|
|
|
|
|
|
/* Define a key over the first column */
|
|
|
|
|
keyinfo[0].seg=keyseg;
|
|
|
|
|
keyinfo[0].keysegs=1;
|
Completion of merge of mysql-5.1 into mysql-maria.
Manually imported changes done to MyISAM (include/myisam.h,
storage/myisam/*, sql/ha_myisam.*, mysql-test/t/myisam.test,
mysql-test/t/ps_2myisam.test) the last
months into Maria (tedious, should do it more frequently in the
future), including those not done at the previous 5.1->Maria merge
(please in the future don't forget to apply MyISAM changes to Maria
when you merge 5.1 into Maria).
Note: I didn't try to import anything which could be MyISAM-related
in other tests of mysql-test (I didn't want to dig in all csets),
but as QA is working to make most tests re-usable for other engines
(Falcon), it is likely that we'll benefit from this and just have
to set engine=Maria somewhere to run those tests on Maria.
func_group and partition tests fail but they already do in main 5.1
on my machine. No Valgrind error in t/*maria*.test.
Monty: please see the commit comment of maria.result and check.
BitKeeper/deleted/.del-ha_maria.m4:
Delete: config/ac-macros/ha_maria.m4
configure.in:
fix for the new way of enabling engines
include/maria.h:
importing changes done to MyISAM the last months into Maria
include/my_handler.h:
importing changes done to MyISAM the last months into Maria
include/myisam.h:
importing changes done to MyISAM the last months into Maria
mysql-test/r/maria.result:
identical to myisam.result, except the engine name in some places
AND in the line testing key_block_size=1000000000000000000:
Maria gives a key block size of 8192 while MyISAM gives 4096;
is it explainable by the difference between MARIA_KEY_BLOCK_LENGTH
and the same constant in MyISAM? Monty?
mysql-test/r/ps_maria.result:
identical to ps_2myisam.result (except the engine name in some places)
mysql-test/t/maria.test:
instead of engine=maria everywhere, I use @@storage_engine (reduces
the diff with myisam.test).
importing changes done to MyISAM the last months into Maria
mysys/my_handler.c:
importing changes done to MyISAM the last months into Maria
sql/ha_maria.cc:
importing changes done to MyISAM the last months into Maria
sql/ha_maria.h:
importing changes done to MyISAM the last months into Maria
sql/mysqld.cc:
unneeded
storage/maria/Makefile.am:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_check.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_create.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_delete_table.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_dynrec.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_extra.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_ft_boolean_search.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_ft_eval.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_ft_nlq_search.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_ft_parser.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_ft_test1.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_ft_update.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_ftdefs.h:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_key.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_open.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_page.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_rkey.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_rsamepos.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_rt_index.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_rt_mbr.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_search.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_sort.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_test1.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_test2.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_test3.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_update.c:
importing changes done to MyISAM the last months into Maria
storage/maria/ma_write.c:
importing changes done to MyISAM the last months into Maria
storage/maria/maria_chk.c:
importing changes done to MyISAM the last months into Maria
storage/maria/maria_def.h:
importing changes done to MyISAM the last months into Maria
storage/maria/maria_ftdump.c:
importing changes done to MyISAM the last months into Maria
storage/maria/maria_pack.c:
importing changes done to MyISAM the last months into Maria
2006-08-10 16:36:54 +02:00
|
|
|
keyinfo[0].block_length= 0; /* Default block length */
|
2006-04-11 15:45:10 +02:00
|
|
|
keyinfo[0].key_alg=HA_KEY_ALG_BTREE;
|
|
|
|
|
keyinfo[0].seg[0].type= key_type;
|
|
|
|
|
keyinfo[0].seg[0].flag= pack_seg;
|
|
|
|
|
keyinfo[0].seg[0].start=1;
|
|
|
|
|
keyinfo[0].seg[0].length=key_length;
|
|
|
|
|
keyinfo[0].seg[0].null_bit= null_fields ? 2 : 0;
|
|
|
|
|
keyinfo[0].seg[0].null_pos=0;
|
|
|
|
|
keyinfo[0].seg[0].language= default_charset_info->number;
|
|
|
|
|
if (pack_seg & HA_BLOB_PART)
|
|
|
|
|
{
|
|
|
|
|
keyinfo[0].seg[0].bit_start=4; /* Length of blob length */
|
|
|
|
|
}
|
|
|
|
|
keyinfo[0].flag = (uint8) (pack_keys | unique_key);
|
|
|
|
|
|
2007-07-02 19:45:15 +02:00
|
|
|
bzero((uchar*) flags,sizeof(flags));
|
2006-04-11 15:45:10 +02:00
|
|
|
if (opt_unique)
|
|
|
|
|
{
|
|
|
|
|
uint start;
|
|
|
|
|
uniques=1;
|
|
|
|
|
bzero((char*) &uniquedef,sizeof(uniquedef));
|
|
|
|
|
bzero((char*) uniqueseg,sizeof(uniqueseg));
|
|
|
|
|
uniquedef.seg=uniqueseg;
|
|
|
|
|
uniquedef.keysegs=2;
|
|
|
|
|
|
|
|
|
|
/* Make a unique over all columns (except first NULL fields) */
|
|
|
|
|
for (i=0, start=1 ; i < 2 ; i++)
|
|
|
|
|
{
|
|
|
|
|
uniqueseg[i].start=start;
|
2007-01-18 20:38:14 +01:00
|
|
|
start+=recinfo[i].length;
|
|
|
|
|
uniqueseg[i].length=recinfo[i].length;
|
2006-04-11 15:45:10 +02:00
|
|
|
uniqueseg[i].language= default_charset_info->number;
|
|
|
|
|
}
|
|
|
|
|
uniqueseg[0].type= key_type;
|
|
|
|
|
uniqueseg[0].null_bit= null_fields ? 2 : 0;
|
|
|
|
|
uniqueseg[1].type= HA_KEYTYPE_TEXT;
|
|
|
|
|
if (extra_field == FIELD_BLOB)
|
|
|
|
|
{
|
|
|
|
|
uniqueseg[1].length=0; /* The whole blob */
|
|
|
|
|
uniqueseg[1].bit_start=4; /* long blob */
|
|
|
|
|
uniqueseg[1].flag|= HA_BLOB_PART;
|
|
|
|
|
}
|
|
|
|
|
else if (extra_field == FIELD_VARCHAR)
|
2007-01-18 20:38:14 +01:00
|
|
|
{
|
2006-04-11 15:45:10 +02:00
|
|
|
uniqueseg[1].flag|= HA_VAR_LENGTH_PART;
|
2007-01-18 20:38:14 +01:00
|
|
|
uniqueseg[1].type= (HA_VARCHAR_PACKLENGTH(recinfo[1].length-1) == 1 ?
|
|
|
|
|
HA_KEYTYPE_VARTEXT1 : HA_KEYTYPE_VARTEXT2);
|
|
|
|
|
}
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
uniques=0;
|
|
|
|
|
|
2007-09-05 01:57:53 +02:00
|
|
|
offset_to_key= test(null_fields);
|
2007-09-11 00:58:15 +02:00
|
|
|
if (key_field == FIELD_BLOB || key_field == FIELD_VARCHAR)
|
2007-09-05 01:57:53 +02:00
|
|
|
offset_to_key+= 2;
|
|
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
if (!silent)
|
2007-01-18 20:38:14 +01:00
|
|
|
printf("- Creating maria file\n");
|
2006-04-11 15:45:10 +02:00
|
|
|
create_info.max_rows=(ulong) (rec_pointer_size ?
|
|
|
|
|
(1L << (rec_pointer_size*8))/40 :
|
|
|
|
|
0);
|
This patch is a collection of patches from from Sanja, Sergei and Monty.
Added logging and pinning of pages to block format.
Integration of transaction manager, log handler.
Better page cache intergration
Split trnman.h into two files, so that we don't have to include my_atomic.h into C++ programs.
Renaming of structures, more comments, more debugging etc.
Fixed problem with small head block + long varchar.
Added extra argument to delete_record() and update_record() (needed for UNDO logging)
Small changes to interface of pagecache and log handler.
Change initialization of log_record_type_descriptors to not be depending on enum order.
Use array of LEX_STRING's to send data to log handler
Added 'dummy' transaction option to MARIA_INFO so that we can always assume 'trn' exists.
include/lf.h:
Interface fixes
Rename of structures
(Patch from Sergei via Sanja)
include/my_atomic.h:
More comments
include/my_global.h:
Added MY_ERRPTR
include/pagecache.h:
Added undo LSN when unlocking pages
mysql-test/r/maria.result:
Updated results
mysql-test/t/maria.test:
Added autocommit around lock tables
(Patch from Sanja)
mysys/lf_alloc-pin.c:
Post-review fixes, simple optimizations
More comments
Struct slot renames
Check amount of memory on stack
(Patch from Sergei)
mysys/lf_dynarray.c:
More comments
mysys/lf_hash.c:
More comments
After review fixes
(Patch from Sergei)
storage/maria/ha_maria.cc:
Split trnman.h into two files, so that we don't have to include my_atomic.h into the .cc program.
(Temporary fix to avoid bug in gcc)
Move out all deferencing of the transaction structure.
Transaction manager integrated (Patch from Sergei)
storage/maria/ha_maria.h:
Added prototype for start_stmt()
storage/maria/lockman.c:
Function call rename
storage/maria/ma_bitmap.c:
Mark deleted pages free from page cache
storage/maria/ma_blockrec.c:
Offset -> rownr
More debugging
Fixed problem with small head block + long varchar
Added logging of changed pages
Added logging of undo (Including only loggging of changed fields in case of update)
Added pinning/unpinning of all changed pages
More comments
Added free_full_pages() as the same code was used in several places.
fill_rows_parts() renamed as fill_insert_undo_parts()
offset -> rownr
Added some optimization of not transactional tables
_ma_update_block_record() has new parameter, as we need original row to do efficent undo for update
storage/maria/ma_blockrec.h:
Added ROW_EXTENTS_ON_STACK
Changed prototype for update and delete of row
storage/maria/ma_check.c:
Added original row to delete_record() call
storage/maria/ma_control_file.h:
Added ifdefs for C++
storage/maria/ma_delete.c:
Added original row to delete_record() call
(Needed for efficent undo logging)
storage/maria/ma_dynrec.c:
Added extra argument to delete_record() and update_record()
Removed not used variable
storage/maria/ma_init.c:
Initialize log handler
storage/maria/ma_loghandler.c:
Removed not used variable
Change initialization of log_record_type_descriptors to not be depending on enum order
Use array of LEX_STRING's to send data to log handler
storage/maria/ma_loghandler.h:
New defines
Use array of LEX_STRING's to send data to log handler
storage/maria/ma_open.c:
Added 'dummy' transaction option to MARIA_INFO so that we can always assume 'trn' exists.
Store in MARIA_SHARE->page_type if pages will have up to date LSN's
storage/maria/ma_pagecache.c:
Don't decrease number of readers when using pagecache_write()/pagecache_read()
In pagecache_write() decrement request count if page was left pinned
Added pagecache_delete_pages()
Removed some casts
Make trace output consistent with rest of code
Simplify calling of DBUG_ASSERT(0)
Only update LSN if the LSN is bigger than what's already on the page
Added LSN parameter pagecache_unpin_page(), pagecache_unpin(), and pagecache_unlock()
(Part of patch from Sanja)
storage/maria/ma_static.c:
Added 'dummy' transaction option to MARIA_INFO so that we can always assume 'trn' exists.
Added default page cache
storage/maria/ma_statrec.c:
Added extra argument to delete_record() and update_record()
storage/maria/ma_test1.c:
Added option -T for transactions
storage/maria/ma_test2.c:
Added option -T for transactions
storage/maria/ma_test_all.sh:
Test with transactions
storage/maria/ma_update.c:
Changed prototype for update of row
storage/maria/maria_def.h:
Changed prototype for update & delete of row as block records need to access the old row
Store in MARIA_SHARE->page_type if pages will have up to date LSN's
Added MARIA_MAX_TREE_LEVELS to allow us to calculate the number of possible pinned pages we may need.
Removed not used 'empty_bits_buffer'
Added pointer to transaction object
Added array for pinned pages
Added log_row_parts array for logging of field data.
Added MARIA_PINNED_PAGE to store pinned pages
storage/maria/trnman.c:
Added accessor functions to transaction object
Added missing DBUG_RETURN()
More debugging
More comments
Changed // comment of code to #ifdef NOT_USED
Transaction manager integrated.
Post review fixes
Part of patch originally from Sergei
storage/maria/trnman.h:
Split trnman.h into two files, so that we don't have to include my_atomic.h into the .cc program.
(Temporary fix to avoid bug in gcc)
storage/maria/unittest/ma_pagecache_single.c:
Added missing argument
Added SKIP_BIG_TESTS
(Patch from Sanja)
storage/maria/unittest/ma_test_loghandler-t.c:
Test logging with new LEX_STRING parameter
(Patch from Sanja)
storage/maria/unittest/ma_test_loghandler_multigroup-t.c:
Test logging with new LEX_STRING parameter
(Patch from Sanja)
storage/maria/unittest/ma_test_loghandler_multithread-t.c:
Test logging with new LEX_STRING parameter
(Patch from Sanja)
storage/maria/unittest/ma_test_loghandler_pagecache-t.c:
Test logging with new LEX_STRING parameter
(Patch from Sanja)
storage/maria/unittest/trnman-t.c:
Stack overflow detection
(Patch from Sergei)
unittest/unit.pl:
Command-line options --big and --verbose
(Patch from Sergei)
unittest/mytap/tap.c:
Detect --big
(Patch from Sergei)
unittest/mytap/tap.h:
Skip_big_tests and SKIP_BIG_TESTS
(Patch from Sergei)
storage/maria/trnman_public.h:
New BitKeeper file ``storage/maria/trnman_public.h''
2007-05-29 19:13:56 +02:00
|
|
|
create_info.transactional= transactional;
|
2007-01-18 20:38:14 +01:00
|
|
|
if (maria_create(filename, record_type, 1, keyinfo,2+opt_unique,recinfo,
|
2006-04-11 15:45:10 +02:00
|
|
|
uniques, &uniquedef, &create_info,
|
|
|
|
|
create_flag))
|
|
|
|
|
goto err;
|
|
|
|
|
if (!(file=maria_open(filename,2,HA_OPEN_ABORT_IF_LOCKED)))
|
|
|
|
|
goto err;
|
|
|
|
|
if (!silent)
|
|
|
|
|
printf("- Writing key:s\n");
|
|
|
|
|
|
2007-08-29 08:03:10 +02:00
|
|
|
if (maria_begin(file))
|
|
|
|
|
goto err;
|
2006-04-11 15:45:10 +02:00
|
|
|
my_errno=0;
|
|
|
|
|
row_count=deleted=0;
|
|
|
|
|
for (i=49 ; i>=1 ; i-=2 )
|
|
|
|
|
{
|
First part of redo/undo for key pages
Added key_nr to st_maria_keydef for faster keyinfo->keynr conversion
For transactional tables, shift record number in keys up with 1 bit to have place to indicate if transid follows
Checksum for MyISAM now ignores NULL and not used part of VARCHAR
Renamed some variables that caused shadow compiler warnings
Moved extra() call when waiting for tables to not be used to after tables are removed from cache.
Fixed crashing bugs when using Maria TEMPORARY tables with TRUNCATE. Removed 'hack' code in sql directory to go around this bug.
pagecache_unlock_by_ulink() now has extra argument to say if page was changed.
Give error message if we fail to open control file
Mark page cache variables as not flushable
include/maria.h:
Made min page cache larger (needed for pinning key page)
Added key_nr to st_maria_keydef for faster keyinfo->keynr conversion
Added write_comp_flag to move some runtime code to maria_open()
include/my_base.h:
Added new error message to be used when handler initialization failed
include/my_global.h:
Renamed dummy to swap_dummy to avoid conflicts with local 'dummy' variables
include/my_handler.h:
Added const to some parameters
mysys/array.c:
More DBUG
mysys/my_error.c:
Fixed indentation
mysys/my_handler.c:
Added const to some parameters
Added missing error messages
sql/field.h:
Renamed variables to avoid variable shadowing
sql/handler.h:
Renamed parameter to avoid variable name conflict
sql/item.h:
Renamed variables to avoid variable shadowing
sql/log_event_old.h:
Renamed variables to avoid variable shadowing
sql/set_var.h:
Renamed variables to avoid variable shadowing
sql/sql_delete.cc:
Removed maria hack for temporary tables
Fixed indentation
sql/sql_table.cc:
Moved extra() call when waiting for tables to not be used to after tables are removed from cache.
This was needed to ensure we don't do a PREPARE_FOR_DROP or similar call while the table is still in use.
sql/table.cc:
Copy page_checksum from share
Removed Maria hack
storage/maria/Makefile.am:
Added new files
storage/maria/ha_maria.cc:
Renamed records -> record_count and info -> create_info to avoid variable name conflicts
Mark page cache variables as not flushable
storage/maria/ma_blockrec.c:
Moved _ma_unpin_all_pages() to ma_key_recover.c
Moved init of info->pinned_pages to ma_open.c
Moved _ma_finalize_row() to maria_key_recover.h
Renamed some variables to avoid variable name conflicts
Mark page_link.changed for blocks we change directly
Simplify handling of undo link when writing LOGREC_UNDO_ROW_INSERT (old code crashed when having redo for index)
storage/maria/ma_blockrec.h:
Removed extra empty line
storage/maria/ma_checkpoint.c:
Remove not needed trnman.h
storage/maria/ma_close.c:
Free pinned pages (which are now always allocated)
storage/maria/ma_control_file.c:
Give error message if we fail to open control file
storage/maria/ma_delete.c:
Changes for redo logging (first part, logging of underflow not yet done)
- Log undo-key-delete
- Log delete of key
- Updated arguments to _ma_fetch_keypage(), _ma_dispose(), _ma_write_keypage(), _ma_insert()
- Added new arguments to some functions to be able to write redo information
- Mark key pages as changed when we write with PAGECACHE_LOCK_LEFT_WRITELOCKED
Remove one not needed _ma_write_keypage() in d_search() when upper level will do the write anyway
Changed 2 bmove_upp() to bmove() as this made code easer to understand
More function comments
Indentation fixes
storage/maria/ma_ft_update.c:
New arguments to _ma_write_keypage()
storage/maria/ma_loghandler.c:
Fixed some DBUG_PRINT messages
Simplify code
Added new log entrys for key page redo
Renamed some variables to avoid variable name shadowing
storage/maria/ma_loghandler.h:
Moved some defines here
Added define for storing key number on key pages
Added new translog record types
Added enum for type of operations in LOGREC_REDO_INDEX
storage/maria/ma_open.c:
Always allocate info.pinned_pages (we need now also for normal key page usage)
Update keyinfo->key_nr
Added virtual functions to convert record position o number to be stored on key pages
Update keyinfo->write_comp_flag to value of search flag to be used when writing key
storage/maria/ma_page.c:
Added redo for key pages
- Extended _ma_fetch_keypage() with type of lock to put on page and address to used MARIA_PINNED_PAGE
- _ma_fetch_keypage() now pin's pages if needed
- Extended _ma_write_keypage() with type of locks to be used
- ma_dispose() now locks info->s->state.key_del from other threads
- ma_dispose() writes redo log record
- ma_new() locks info->s->state.key_del from other threads if it was used
- ma_new() now pins read page
Other things:
- Removed some not needed arguments from _ma_new() and _ma_dispose)
- Added some new variables to simplify code
- If EXTRA_DEBUG is used, do crc on full page to catch not unitialized bytes
storage/maria/ma_pagecache.h:
Applied patch from Sanja to add extra argument to pagecache_unlock_by_ulink() to mark if page was changed
Added some defines for pagecache priority levels that one can use
storage/maria/ma_range.c:
Added new arguments for call to _ma_fetch_keypage()
storage/maria/ma_recovery.c:
- Added hooks for new translog types:
REDO_INDEX, REDO_INDEX_NEW_PAGE, REDO_INDEX_FREE_PAGE, UNDO_KEY_INSERT, UNDO_KEY_DELETE and
UNDO_KEY_DELETE_WITH_ROOT.
- Moved variable declarations to start of function (portability fixes)
- Removed some not needed initializations
- Set only relevant state changes for each redo/undo entry
storage/maria/lockman.c:
Removed end space
storage/maria/ma_check.c:
Removed end space
storage/maria/ma_create.c:
Removed end space
storage/maria/ma_locking.c:
Removed end space
storage/maria/ma_packrec.c:
Removed end space
storage/maria/ma_pagecache.c:
Removed end space
storage/maria/ma_panic.c:
Removed end space
storage/maria/ma_rt_index.c:
Added new arguments for call to _ma_fetch_keypage(), _ma_write_keypage(), _ma_dispose() and _ma_new()
Fixed indentation
storage/maria/ma_rt_key.c:
Added new arguments for call to _ma_fetch_keypage()
storage/maria/ma_rt_split.c:
Added new arguments for call to _ma_new()
Use new keypage header
Added new arguments for call to _ma_write_keypage()
storage/maria/ma_search.c:
Updated comments & indentation
Added new arguments for call to _ma_fetch_keypage()
Made some variables and arguments const
Added virtual functions for converting row position to number to be stored in key
use MARIA_RECORD_POS of record position instead of my_off_t
Record in MARIA_KEY_PARAM how page was changed one key insert (needed for REDO)
storage/maria/ma_sort.c:
Removed end space
storage/maria/ma_statrec.c:
Updated arguments for call to _ma_rec_pos()
storage/maria/ma_test1.c:
Fixed too small buffer to init_pagecache()
Fixed bug when using insert_count and test_flag
storage/maria/ma_test2.c:
Use more resonable pagecache size
Remove not used code
Reset blob_length to fix wrong output message
storage/maria/ma_test_all.sh:
Fixed wrong test
storage/maria/ma_write.c:
Lots of new code to handle REDO of key pages
No logic changes because of REDO code, mostly adding new arguments and adding new code for logging
Added new arguments for calls to _ma_fetch_keypage(), _ma_write_keypage() and similar functions
Move setting of comp_flag in ma_ck_wrte_btree() from runtime to maria_open()
Zerofill new used pages for:
- To remove possible sensitive data left in buffer
- To get idenitical data on pages after running redo
- Better compression of pages if archived
storage/maria/maria_chk.c:
Added information if table is crash safe
storage/maria/maria_def.h:
New virtual function to convert between record position on key and normal record position
Aded mutex and extra variables to handle locking of share->state.key_del
Moved some structure variables to get things more aligned
Added extra arguments to MARIA_KEY_PARAM to be able to remember what was changed on key page on key insert
Added argument to MARIA_PINNED_PAGE to indicate if page was changed
Updated prototypes for functions
Added some structures for signaling changes in REDO handling
storage/maria/unittest/ma_pagecache_single.c:
Updated arguments for changed function calls
storage/myisam/mi_check.c:
Made calc_check_checksum virtual
storage/myisam/mi_checksum.c:
Update checksums to ignore null columns
storage/myisam/mi_create.c:
Mark if table has null column (to know when we have to use mi_checksum())
storage/myisam/mi_open.c:
Added virtual function for calculating checksum to be able to easily ignore NULL fields
storage/myisam/mi_test2.c:
Fixed bug
storage/myisam/myisamdef.h:
Added virtual function for calculating checksum during check table
Removed ha_key_cmp() as this is in handler.h
storage/maria/ma_key_recover.c:
New BitKeeper file ``storage/maria/ma_key_recover.c''
storage/maria/ma_key_recover.h:
New BitKeeper file ``storage/maria/ma_key_recover.h''
storage/maria/ma_key_redo.c:
New BitKeeper file ``storage/maria/ma_key_redo.c''
2007-11-14 18:08:06 +01:00
|
|
|
if (insert_count-- == 0)
|
|
|
|
|
{
|
|
|
|
|
if (testflag)
|
|
|
|
|
break;
|
|
|
|
|
VOID(maria_close(file));
|
|
|
|
|
exit(0);
|
|
|
|
|
}
|
2006-04-11 15:45:10 +02:00
|
|
|
j=i%25 +1;
|
|
|
|
|
create_record(record,j);
|
|
|
|
|
error=maria_write(file,record);
|
|
|
|
|
if (!error)
|
|
|
|
|
row_count++;
|
|
|
|
|
flags[j]=1;
|
|
|
|
|
if (verbose || error)
|
|
|
|
|
printf("J= %2d maria_write: %d errno: %d\n", j,error,my_errno);
|
|
|
|
|
}
|
|
|
|
|
|
2007-09-04 09:53:52 +02:00
|
|
|
if (maria_commit(file) || maria_begin(file))
|
|
|
|
|
goto err;
|
|
|
|
|
|
WL#3071 Maria checkpoint
Ability for flush_pagecache_blocks() to flush only certain pages of
a file, as instructed by an option "filter" pointer-to-function argument;
Checkpoint and background dirty page flushing use that to flush only
pages which have been dirty for long enough and bitmap pages.
Fix for a bug in flush_cached_blocks() (no idea if it could produce
a bug in real life, but theoretically it is).
Testing checkpoint in ma_test_recovery via ma_test1 and ma_test2.
Background checkpoint & dirty pages flush thread is still disabled
by default in ha_maria.
mysql-test/r/maria.result:
result update
storage/maria/ha_maria.cc:
blank after function comment
storage/maria/ma_checkpoint.c:
Using an enum instead of 0/1/2 (applying Sanja's review comments).
The comment about "this is an horizon" can be removed as Sanja
created translog_next_LSN() which parse_checkpoint_record() uses.
Variables in ma_checkpoint_background() cannot be declared in the
for() as their value must not be reset at each iteration!
storage/maria/ma_pagecache.c:
adding to flush_pagecache_blocks() optional arguments 'filter'
(pointer to function) and 'filter_arg'; if filter!=NULL this function
will be called for each block of the file and will reply if this
block and following ones should be flushed or not (3 possible
replies).
Fixing a bug when flush_cached_blocks() skips a pinned page: it has
to unset PCBLOCK_IN_FLUSH set by flush_pagecache_blocks_int().
storage/maria/ma_pagecache.h:
flush_pagecache_blocks() is changed to take "filter" and "filter_arg"
arguments. "filter", if it is not NULL, may return one value
among enum pagecache_flush_filter_result.
storage/maria/ma_recovery.c:
open_count=0 when closing tables at the end of recovery.
storage/maria/ma_test1.c:
Optional checkpoints (-H#) at various stages (stages similar
to --testflag), for testing of checkpoints.
storage/maria/ma_test2.c:
Optional checkpoints (-H#) at various stages (stages similar
to -t), for testing of checkpoints.
storage/maria/ma_test_recovery.expected:
Result update: the results of the additional test run with -H#
(checkpoints) are added here. They are exactly identical to without
checkpoints except that the index's Root (printed by maria_chk)
is more correct when using checkpoints. This is because checkpoint
flushed the state, so it happens to be correct, while no-checkpoint
does not flush the state, and recovery does not recover indexes
so Root is never fixed. When we recover indices, this will go away.
storage/maria/ma_test_recovery:
We duplicate the loop of tests to add an additional run with
checkpoints at various stages, to see if maria_read_log
uses them fine.
2007-10-17 16:55:26 +02:00
|
|
|
if (checkpoint == 1 && ma_checkpoint_execute(CHECKPOINT_MEDIUM, FALSE))
|
|
|
|
|
goto err;
|
|
|
|
|
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
if (testflag == 1)
|
|
|
|
|
goto end;
|
|
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
/* Insert 2 rows with null values */
|
|
|
|
|
if (null_fields)
|
|
|
|
|
{
|
|
|
|
|
create_record(record,0);
|
|
|
|
|
error=maria_write(file,record);
|
|
|
|
|
if (!error)
|
|
|
|
|
row_count++;
|
|
|
|
|
if (verbose || error)
|
|
|
|
|
printf("J= NULL maria_write: %d errno: %d\n", error,my_errno);
|
|
|
|
|
error=maria_write(file,record);
|
|
|
|
|
if (!error)
|
|
|
|
|
row_count++;
|
|
|
|
|
if (verbose || error)
|
|
|
|
|
printf("J= NULL maria_write: %d errno: %d\n", error,my_errno);
|
|
|
|
|
flags[0]=2;
|
|
|
|
|
}
|
|
|
|
|
|
WL#3071 Maria checkpoint
Ability for flush_pagecache_blocks() to flush only certain pages of
a file, as instructed by an option "filter" pointer-to-function argument;
Checkpoint and background dirty page flushing use that to flush only
pages which have been dirty for long enough and bitmap pages.
Fix for a bug in flush_cached_blocks() (no idea if it could produce
a bug in real life, but theoretically it is).
Testing checkpoint in ma_test_recovery via ma_test1 and ma_test2.
Background checkpoint & dirty pages flush thread is still disabled
by default in ha_maria.
mysql-test/r/maria.result:
result update
storage/maria/ha_maria.cc:
blank after function comment
storage/maria/ma_checkpoint.c:
Using an enum instead of 0/1/2 (applying Sanja's review comments).
The comment about "this is an horizon" can be removed as Sanja
created translog_next_LSN() which parse_checkpoint_record() uses.
Variables in ma_checkpoint_background() cannot be declared in the
for() as their value must not be reset at each iteration!
storage/maria/ma_pagecache.c:
adding to flush_pagecache_blocks() optional arguments 'filter'
(pointer to function) and 'filter_arg'; if filter!=NULL this function
will be called for each block of the file and will reply if this
block and following ones should be flushed or not (3 possible
replies).
Fixing a bug when flush_cached_blocks() skips a pinned page: it has
to unset PCBLOCK_IN_FLUSH set by flush_pagecache_blocks_int().
storage/maria/ma_pagecache.h:
flush_pagecache_blocks() is changed to take "filter" and "filter_arg"
arguments. "filter", if it is not NULL, may return one value
among enum pagecache_flush_filter_result.
storage/maria/ma_recovery.c:
open_count=0 when closing tables at the end of recovery.
storage/maria/ma_test1.c:
Optional checkpoints (-H#) at various stages (stages similar
to --testflag), for testing of checkpoints.
storage/maria/ma_test2.c:
Optional checkpoints (-H#) at various stages (stages similar
to -t), for testing of checkpoints.
storage/maria/ma_test_recovery.expected:
Result update: the results of the additional test run with -H#
(checkpoints) are added here. They are exactly identical to without
checkpoints except that the index's Root (printed by maria_chk)
is more correct when using checkpoints. This is because checkpoint
flushed the state, so it happens to be correct, while no-checkpoint
does not flush the state, and recovery does not recover indexes
so Root is never fixed. When we recover indices, this will go away.
storage/maria/ma_test_recovery:
We duplicate the loop of tests to add an additional run with
checkpoints at various stages, to see if maria_read_log
uses them fine.
2007-10-17 16:55:26 +02:00
|
|
|
if (checkpoint == 2 && ma_checkpoint_execute(CHECKPOINT_MEDIUM, FALSE))
|
|
|
|
|
goto err;
|
|
|
|
|
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
if (testflag == 2)
|
2007-09-04 09:53:52 +02:00
|
|
|
{
|
2007-09-11 11:11:22 +02:00
|
|
|
printf("Terminating after inserts\n");
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
goto end;
|
2007-09-04 09:53:52 +02:00
|
|
|
}
|
|
|
|
|
|
2007-09-09 18:15:10 +02:00
|
|
|
if (maria_commit(file) || maria_begin(file))
|
|
|
|
|
goto err;
|
|
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
if (!skip_update)
|
|
|
|
|
{
|
|
|
|
|
if (opt_unique)
|
|
|
|
|
{
|
|
|
|
|
if (!silent)
|
|
|
|
|
printf("- Checking unique constraint\n");
|
2007-11-20 16:42:16 +01:00
|
|
|
create_record(record,j); /* Check last created row */
|
2006-04-11 15:45:10 +02:00
|
|
|
if (!maria_write(file,record) || my_errno != HA_ERR_FOUND_DUPP_UNIQUE)
|
|
|
|
|
{
|
|
|
|
|
printf("unique check failed\n");
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (!silent)
|
|
|
|
|
printf("- Updating rows\n");
|
|
|
|
|
|
|
|
|
|
/* Update first last row to force extend of file */
|
|
|
|
|
if (maria_rsame(file,read_record,-1))
|
|
|
|
|
{
|
|
|
|
|
printf("Can't find last row with maria_rsame\n");
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
memcpy(record,read_record,rec_length);
|
|
|
|
|
update_record(record);
|
|
|
|
|
if (maria_update(file,read_record,record))
|
|
|
|
|
{
|
|
|
|
|
printf("Can't update last row: %.*s\n",
|
|
|
|
|
keyinfo[0].seg[0].length,read_record+1);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Read through all rows and update them */
|
2007-01-18 20:38:14 +01:00
|
|
|
assert(maria_scan_init(file) == 0);
|
|
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
found=0;
|
2007-01-18 20:38:14 +01:00
|
|
|
while ((error= maria_scan(file,read_record)) == 0)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
2007-09-09 18:15:10 +02:00
|
|
|
if (--update_count == 0) { VOID(maria_close(file)) ; exit(0) ; }
|
2006-04-11 15:45:10 +02:00
|
|
|
memcpy(record,read_record,rec_length);
|
|
|
|
|
update_record(record);
|
|
|
|
|
if (maria_update(file,read_record,record))
|
|
|
|
|
{
|
|
|
|
|
printf("Can't update row: %.*s, error: %d\n",
|
|
|
|
|
keyinfo[0].seg[0].length,record+1,my_errno);
|
|
|
|
|
}
|
|
|
|
|
found++;
|
|
|
|
|
}
|
|
|
|
|
if (found != row_count)
|
|
|
|
|
printf("Found %ld of %ld rows\n", (ulong) found, (ulong) row_count);
|
2007-01-18 20:38:14 +01:00
|
|
|
maria_scan_end(file);
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
|
WL#3071 Maria checkpoint
Ability for flush_pagecache_blocks() to flush only certain pages of
a file, as instructed by an option "filter" pointer-to-function argument;
Checkpoint and background dirty page flushing use that to flush only
pages which have been dirty for long enough and bitmap pages.
Fix for a bug in flush_cached_blocks() (no idea if it could produce
a bug in real life, but theoretically it is).
Testing checkpoint in ma_test_recovery via ma_test1 and ma_test2.
Background checkpoint & dirty pages flush thread is still disabled
by default in ha_maria.
mysql-test/r/maria.result:
result update
storage/maria/ha_maria.cc:
blank after function comment
storage/maria/ma_checkpoint.c:
Using an enum instead of 0/1/2 (applying Sanja's review comments).
The comment about "this is an horizon" can be removed as Sanja
created translog_next_LSN() which parse_checkpoint_record() uses.
Variables in ma_checkpoint_background() cannot be declared in the
for() as their value must not be reset at each iteration!
storage/maria/ma_pagecache.c:
adding to flush_pagecache_blocks() optional arguments 'filter'
(pointer to function) and 'filter_arg'; if filter!=NULL this function
will be called for each block of the file and will reply if this
block and following ones should be flushed or not (3 possible
replies).
Fixing a bug when flush_cached_blocks() skips a pinned page: it has
to unset PCBLOCK_IN_FLUSH set by flush_pagecache_blocks_int().
storage/maria/ma_pagecache.h:
flush_pagecache_blocks() is changed to take "filter" and "filter_arg"
arguments. "filter", if it is not NULL, may return one value
among enum pagecache_flush_filter_result.
storage/maria/ma_recovery.c:
open_count=0 when closing tables at the end of recovery.
storage/maria/ma_test1.c:
Optional checkpoints (-H#) at various stages (stages similar
to --testflag), for testing of checkpoints.
storage/maria/ma_test2.c:
Optional checkpoints (-H#) at various stages (stages similar
to -t), for testing of checkpoints.
storage/maria/ma_test_recovery.expected:
Result update: the results of the additional test run with -H#
(checkpoints) are added here. They are exactly identical to without
checkpoints except that the index's Root (printed by maria_chk)
is more correct when using checkpoints. This is because checkpoint
flushed the state, so it happens to be correct, while no-checkpoint
does not flush the state, and recovery does not recover indexes
so Root is never fixed. When we recover indices, this will go away.
storage/maria/ma_test_recovery:
We duplicate the loop of tests to add an additional run with
checkpoints at various stages, to see if maria_read_log
uses them fine.
2007-10-17 16:55:26 +02:00
|
|
|
if (checkpoint == 3 && ma_checkpoint_execute(CHECKPOINT_MEDIUM, FALSE))
|
|
|
|
|
goto err;
|
|
|
|
|
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
if (testflag == 3)
|
2007-09-09 18:15:10 +02:00
|
|
|
{
|
2007-09-11 11:11:22 +02:00
|
|
|
printf("Terminating after updates\n");
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
goto end;
|
2007-09-09 18:15:10 +02:00
|
|
|
}
|
2006-04-11 15:45:10 +02:00
|
|
|
if (!silent)
|
|
|
|
|
printf("- Reopening file\n");
|
2007-08-29 08:03:10 +02:00
|
|
|
if (maria_commit(file))
|
|
|
|
|
goto err;
|
|
|
|
|
if (maria_close(file))
|
|
|
|
|
goto err;
|
|
|
|
|
if (!(file=maria_open(filename,2,HA_OPEN_ABORT_IF_LOCKED)))
|
|
|
|
|
goto err;
|
|
|
|
|
if (maria_begin(file))
|
|
|
|
|
goto err;
|
2007-01-18 20:38:14 +01:00
|
|
|
if (!skip_delete)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
if (!silent)
|
|
|
|
|
printf("- Removing keys\n");
|
|
|
|
|
|
|
|
|
|
for (i=0 ; i <= 10 ; i++)
|
|
|
|
|
{
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
/*
|
|
|
|
|
If you want to debug the problem in ma_test_recovery with BLOBs
|
|
|
|
|
(see @todo there), you can break out of the loop after just one
|
|
|
|
|
delete, it is enough, like this:
|
|
|
|
|
if (i==1) break;
|
|
|
|
|
*/
|
2006-04-11 15:45:10 +02:00
|
|
|
/* testing */
|
2007-01-18 20:38:14 +01:00
|
|
|
if (remove_count-- == 0)
|
|
|
|
|
{
|
|
|
|
|
fprintf(stderr,
|
|
|
|
|
"delete-rows number of rows deleted; Going down hard!\n");
|
2007-09-11 00:58:15 +02:00
|
|
|
goto end;
|
2007-01-18 20:38:14 +01:00
|
|
|
}
|
2006-04-11 15:45:10 +02:00
|
|
|
j=i*2;
|
|
|
|
|
if (!flags[j])
|
|
|
|
|
continue;
|
|
|
|
|
create_key(key,j);
|
|
|
|
|
my_errno=0;
|
2007-07-27 16:11:40 +02:00
|
|
|
if ((error = maria_rkey(file, read_record, 0, key,
|
|
|
|
|
HA_WHOLE_KEY, HA_READ_KEY_EXACT)))
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
if (verbose || (flags[j] >= 1 ||
|
|
|
|
|
(error && my_errno != HA_ERR_KEY_NOT_FOUND)))
|
|
|
|
|
printf("key: '%.*s' maria_rkey: %3d errno: %3d\n",
|
2007-09-05 01:57:53 +02:00
|
|
|
(int) key_length,key+offset_to_key,error,my_errno);
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
error=maria_delete(file,read_record);
|
|
|
|
|
if (verbose || error)
|
|
|
|
|
printf("key: '%.*s' maria_delete: %3d errno: %3d\n",
|
2007-09-05 01:57:53 +02:00
|
|
|
(int) key_length, key+offset_to_key, error, my_errno);
|
2006-04-11 15:45:10 +02:00
|
|
|
if (! error)
|
|
|
|
|
{
|
|
|
|
|
deleted++;
|
|
|
|
|
flags[j]--;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
}
|
2007-09-05 01:57:53 +02:00
|
|
|
|
WL#3071 Maria checkpoint
Ability for flush_pagecache_blocks() to flush only certain pages of
a file, as instructed by an option "filter" pointer-to-function argument;
Checkpoint and background dirty page flushing use that to flush only
pages which have been dirty for long enough and bitmap pages.
Fix for a bug in flush_cached_blocks() (no idea if it could produce
a bug in real life, but theoretically it is).
Testing checkpoint in ma_test_recovery via ma_test1 and ma_test2.
Background checkpoint & dirty pages flush thread is still disabled
by default in ha_maria.
mysql-test/r/maria.result:
result update
storage/maria/ha_maria.cc:
blank after function comment
storage/maria/ma_checkpoint.c:
Using an enum instead of 0/1/2 (applying Sanja's review comments).
The comment about "this is an horizon" can be removed as Sanja
created translog_next_LSN() which parse_checkpoint_record() uses.
Variables in ma_checkpoint_background() cannot be declared in the
for() as their value must not be reset at each iteration!
storage/maria/ma_pagecache.c:
adding to flush_pagecache_blocks() optional arguments 'filter'
(pointer to function) and 'filter_arg'; if filter!=NULL this function
will be called for each block of the file and will reply if this
block and following ones should be flushed or not (3 possible
replies).
Fixing a bug when flush_cached_blocks() skips a pinned page: it has
to unset PCBLOCK_IN_FLUSH set by flush_pagecache_blocks_int().
storage/maria/ma_pagecache.h:
flush_pagecache_blocks() is changed to take "filter" and "filter_arg"
arguments. "filter", if it is not NULL, may return one value
among enum pagecache_flush_filter_result.
storage/maria/ma_recovery.c:
open_count=0 when closing tables at the end of recovery.
storage/maria/ma_test1.c:
Optional checkpoints (-H#) at various stages (stages similar
to --testflag), for testing of checkpoints.
storage/maria/ma_test2.c:
Optional checkpoints (-H#) at various stages (stages similar
to -t), for testing of checkpoints.
storage/maria/ma_test_recovery.expected:
Result update: the results of the additional test run with -H#
(checkpoints) are added here. They are exactly identical to without
checkpoints except that the index's Root (printed by maria_chk)
is more correct when using checkpoints. This is because checkpoint
flushed the state, so it happens to be correct, while no-checkpoint
does not flush the state, and recovery does not recover indexes
so Root is never fixed. When we recover indices, this will go away.
storage/maria/ma_test_recovery:
We duplicate the loop of tests to add an additional run with
checkpoints at various stages, to see if maria_read_log
uses them fine.
2007-10-17 16:55:26 +02:00
|
|
|
if (checkpoint == 4 && ma_checkpoint_execute(CHECKPOINT_MEDIUM, FALSE))
|
|
|
|
|
goto err;
|
|
|
|
|
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
if (testflag == 4)
|
|
|
|
|
{
|
2007-09-11 11:11:22 +02:00
|
|
|
printf("Terminating after deletes\n");
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
goto end;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
if (!silent)
|
|
|
|
|
printf("- Reading rows with key\n");
|
2007-01-18 20:38:14 +01:00
|
|
|
record[1]= 0; /* For nicer printf */
|
2006-04-11 15:45:10 +02:00
|
|
|
for (i=0 ; i <= 25 ; i++)
|
|
|
|
|
{
|
|
|
|
|
create_key(key,i);
|
|
|
|
|
my_errno=0;
|
2007-07-27 16:11:40 +02:00
|
|
|
error=maria_rkey(file,read_record,0,key,HA_WHOLE_KEY,HA_READ_KEY_EXACT);
|
2006-04-11 15:45:10 +02:00
|
|
|
if (verbose ||
|
|
|
|
|
(error == 0 && flags[i] == 0 && unique_key) ||
|
|
|
|
|
(error && (flags[i] != 0 || my_errno != HA_ERR_KEY_NOT_FOUND)))
|
|
|
|
|
{
|
|
|
|
|
printf("key: '%.*s' maria_rkey: %3d errno: %3d record: %s\n",
|
2007-09-05 01:57:53 +02:00
|
|
|
(int) key_length,key+offset_to_key,error,my_errno,record+1);
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (!silent)
|
|
|
|
|
printf("- Reading rows with position\n");
|
2007-01-18 20:38:14 +01:00
|
|
|
if (maria_scan_init(file))
|
|
|
|
|
{
|
|
|
|
|
fprintf(stderr, "maria_scan_init failed\n");
|
|
|
|
|
goto err;
|
|
|
|
|
}
|
|
|
|
|
|
2006-04-11 15:45:10 +02:00
|
|
|
for (i=1,found=0 ; i <= 30 ; i++)
|
|
|
|
|
{
|
|
|
|
|
my_errno=0;
|
2007-01-18 20:38:14 +01:00
|
|
|
if ((error= maria_scan(file, read_record)) == HA_ERR_END_OF_FILE)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
if (found != row_count-deleted)
|
|
|
|
|
printf("Found only %ld of %ld rows\n", (ulong) found,
|
|
|
|
|
(ulong) (row_count - deleted));
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
if (!error)
|
|
|
|
|
found++;
|
|
|
|
|
if (verbose || (error != 0 && error != HA_ERR_RECORD_DELETED &&
|
|
|
|
|
error != HA_ERR_END_OF_FILE))
|
|
|
|
|
{
|
|
|
|
|
printf("pos: %2d maria_rrnd: %3d errno: %3d record: %s\n",
|
|
|
|
|
i-1,error,my_errno,read_record+1);
|
|
|
|
|
}
|
|
|
|
|
}
|
2007-10-04 19:33:42 +02:00
|
|
|
maria_scan_end(file);
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
|
|
|
|
|
end:
|
|
|
|
|
if (die_in_middle_of_transaction)
|
|
|
|
|
{
|
|
|
|
|
/* As commit record is not done, UNDO entries needs to be rolled back */
|
|
|
|
|
switch (die_in_middle_of_transaction) {
|
|
|
|
|
case 1:
|
|
|
|
|
/*
|
|
|
|
|
Flush changed pages go to disk. That will also flush log. Recovery
|
|
|
|
|
will skip REDOs and apply UNDOs.
|
|
|
|
|
*/
|
2007-12-10 01:32:00 +01:00
|
|
|
_ma_flush_table_files(file, MARIA_FLUSH_DATA | MARIA_FLUSH_INDEX,
|
|
|
|
|
FLUSH_RELEASE, FLUSH_RELEASE);
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
break;
|
|
|
|
|
case 2:
|
|
|
|
|
/*
|
|
|
|
|
Just flush log. Pages are likely to not be on disk. Recovery will
|
|
|
|
|
then execute REDOs and UNDOs.
|
|
|
|
|
*/
|
|
|
|
|
if (translog_flush(file->trn->undo_lsn))
|
|
|
|
|
goto err;
|
|
|
|
|
break;
|
|
|
|
|
case 3:
|
|
|
|
|
/*
|
|
|
|
|
Flush nothing. Pages and log are likely to not be on disk. Recovery
|
|
|
|
|
will then do nothing.
|
|
|
|
|
*/
|
|
|
|
|
break;
|
2007-12-10 01:32:00 +01:00
|
|
|
case 4:
|
|
|
|
|
/*
|
|
|
|
|
Flush changed data pages go to disk. Changed index pages are not
|
|
|
|
|
flushed. Recovery will skip some REDOs and apply UNDOs.
|
|
|
|
|
*/
|
|
|
|
|
_ma_flush_table_files(file, MARIA_FLUSH_DATA, FLUSH_RELEASE,
|
|
|
|
|
FLUSH_RELEASE);
|
|
|
|
|
/*
|
|
|
|
|
We have to flush log separately as the redo for the last key page
|
|
|
|
|
may not be flushed
|
|
|
|
|
*/
|
|
|
|
|
if (translog_flush(file->trn->undo_lsn))
|
|
|
|
|
goto err;
|
|
|
|
|
break;
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
}
|
|
|
|
|
printf("Dying on request without maria_commit()/maria_close()\n");
|
|
|
|
|
exit(0);
|
|
|
|
|
}
|
|
|
|
|
|
2007-08-29 08:03:10 +02:00
|
|
|
if (maria_commit(file))
|
|
|
|
|
goto err;
|
2007-01-18 20:38:14 +01:00
|
|
|
if (maria_close(file))
|
|
|
|
|
goto err;
|
2006-04-11 15:45:10 +02:00
|
|
|
maria_end();
|
|
|
|
|
my_end(MY_CHECK_ERROR);
|
|
|
|
|
|
|
|
|
|
return (0);
|
|
|
|
|
err:
|
|
|
|
|
printf("got error: %3d when using maria-database\n",my_errno);
|
|
|
|
|
return 1; /* skip warning */
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2007-10-04 19:33:42 +02:00
|
|
|
static void create_key_part(uchar *key,uint rownr)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
if (!unique_key)
|
|
|
|
|
rownr&=7; /* Some identical keys */
|
|
|
|
|
if (keyinfo[0].seg[0].type == HA_KEYTYPE_NUM)
|
|
|
|
|
{
|
2007-10-04 19:33:42 +02:00
|
|
|
sprintf((char*) key,"%*d",keyinfo[0].seg[0].length,rownr);
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
else if (keyinfo[0].seg[0].type == HA_KEYTYPE_VARTEXT1 ||
|
|
|
|
|
keyinfo[0].seg[0].type == HA_KEYTYPE_VARTEXT2)
|
|
|
|
|
{ /* Alpha record */
|
|
|
|
|
/* Create a key that may be easily packed */
|
|
|
|
|
bfill(key,keyinfo[0].seg[0].length,rownr < 10 ? 'A' : 'B');
|
2007-10-04 19:33:42 +02:00
|
|
|
sprintf((char*) key+keyinfo[0].seg[0].length-2,"%-2d",rownr);
|
2006-04-11 15:45:10 +02:00
|
|
|
if ((rownr & 7) == 0)
|
|
|
|
|
{
|
|
|
|
|
/* Change the key to force a unpack of the next key */
|
2007-01-18 20:38:14 +01:00
|
|
|
bfill(key+3,keyinfo[0].seg[0].length-5,rownr < 10 ? 'a' : 'b');
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{ /* Alpha record */
|
|
|
|
|
if (keyinfo[0].seg[0].flag & HA_SPACE_PACK)
|
2007-10-04 19:33:42 +02:00
|
|
|
sprintf((char*) key,"%-*d",keyinfo[0].seg[0].length,rownr);
|
2006-04-11 15:45:10 +02:00
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
/* Create a key that may be easily packed */
|
|
|
|
|
bfill(key,keyinfo[0].seg[0].length,rownr < 10 ? 'A' : 'B');
|
2007-10-04 19:33:42 +02:00
|
|
|
sprintf((char*) key+keyinfo[0].seg[0].length-2,"%-2d",rownr);
|
2006-04-11 15:45:10 +02:00
|
|
|
if ((rownr & 7) == 0)
|
|
|
|
|
{
|
|
|
|
|
/* Change the key to force a unpack of the next key */
|
|
|
|
|
key[1]= (rownr < 10 ? 'a' : 'b');
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2007-10-04 19:33:42 +02:00
|
|
|
static void create_key(uchar *key,uint rownr)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
if (keyinfo[0].seg[0].null_bit)
|
|
|
|
|
{
|
|
|
|
|
if (rownr == 0)
|
|
|
|
|
{
|
|
|
|
|
key[0]=1; /* null key */
|
2007-01-18 20:38:14 +01:00
|
|
|
key[1]=0; /* For easy print of key */
|
2006-04-11 15:45:10 +02:00
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
*key++=0;
|
|
|
|
|
}
|
|
|
|
|
if (keyinfo[0].seg[0].flag & (HA_BLOB_PART | HA_VAR_LENGTH_PART))
|
|
|
|
|
{
|
|
|
|
|
uint tmp;
|
|
|
|
|
create_key_part(key+2,rownr);
|
2007-10-04 19:33:42 +02:00
|
|
|
tmp=strlen((char*) key+2);
|
2006-04-11 15:45:10 +02:00
|
|
|
int2store(key,tmp);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
create_key_part(key,rownr);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2007-10-04 19:33:42 +02:00
|
|
|
static uchar blob_key[MAX_REC_LENGTH];
|
|
|
|
|
static uchar blob_record[MAX_REC_LENGTH+20*20];
|
2006-04-11 15:45:10 +02:00
|
|
|
|
|
|
|
|
|
2007-10-04 19:33:42 +02:00
|
|
|
static void create_record(uchar *record,uint rownr)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
2007-10-04 19:33:42 +02:00
|
|
|
uchar *pos;
|
2006-04-11 15:45:10 +02:00
|
|
|
bzero((char*) record,MAX_REC_LENGTH);
|
|
|
|
|
record[0]=1; /* delete marker */
|
|
|
|
|
if (rownr == 0 && keyinfo[0].seg[0].null_bit)
|
|
|
|
|
record[0]|=keyinfo[0].seg[0].null_bit; /* Null key */
|
|
|
|
|
|
|
|
|
|
pos=record+1;
|
2007-01-18 20:38:14 +01:00
|
|
|
if (recinfo[0].type == FIELD_BLOB)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
uint tmp;
|
2007-10-04 19:33:42 +02:00
|
|
|
uchar *ptr;
|
2006-04-11 15:45:10 +02:00
|
|
|
create_key_part(blob_key,rownr);
|
2007-10-04 19:33:42 +02:00
|
|
|
tmp=strlen((char*) blob_key);
|
2006-04-11 15:45:10 +02:00
|
|
|
int4store(pos,tmp);
|
|
|
|
|
ptr=blob_key;
|
|
|
|
|
memcpy_fixed(pos+4,&ptr,sizeof(char*));
|
2007-01-18 20:38:14 +01:00
|
|
|
pos+=recinfo[0].length;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
2007-01-18 20:38:14 +01:00
|
|
|
else if (recinfo[0].type == FIELD_VARCHAR)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
2007-01-18 20:38:14 +01:00
|
|
|
uint tmp, pack_length= HA_VARCHAR_PACKLENGTH(recinfo[0].length-1);
|
2006-04-11 15:45:10 +02:00
|
|
|
create_key_part(pos+pack_length,rownr);
|
2007-10-04 19:33:42 +02:00
|
|
|
tmp= strlen((char*) pos+pack_length);
|
2006-04-11 15:45:10 +02:00
|
|
|
if (pack_length == 1)
|
|
|
|
|
*(uchar*) pos= (uchar) tmp;
|
|
|
|
|
else
|
|
|
|
|
int2store(pos,tmp);
|
2007-01-18 20:38:14 +01:00
|
|
|
pos+= recinfo[0].length;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
create_key_part(pos,rownr);
|
2007-01-18 20:38:14 +01:00
|
|
|
pos+=recinfo[0].length;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
2007-01-18 20:38:14 +01:00
|
|
|
if (recinfo[1].type == FIELD_BLOB)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
uint tmp;
|
2007-10-04 19:33:42 +02:00
|
|
|
uchar *ptr;;
|
|
|
|
|
sprintf((char*) blob_record,"... row: %d", rownr);
|
|
|
|
|
strappend((char*) blob_record,max(MAX_REC_LENGTH-rownr,10),' ');
|
|
|
|
|
tmp=strlen((char*) blob_record);
|
2006-04-11 15:45:10 +02:00
|
|
|
int4store(pos,tmp);
|
|
|
|
|
ptr=blob_record;
|
|
|
|
|
memcpy_fixed(pos+4,&ptr,sizeof(char*));
|
|
|
|
|
}
|
2007-01-18 20:38:14 +01:00
|
|
|
else if (recinfo[1].type == FIELD_VARCHAR)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
uint tmp, pack_length= HA_VARCHAR_PACKLENGTH(recinfo[1].length-1);
|
2007-10-04 19:33:42 +02:00
|
|
|
sprintf((char*) pos+pack_length, "... row: %d", rownr);
|
|
|
|
|
tmp= strlen((char*) pos+pack_length);
|
2006-04-11 15:45:10 +02:00
|
|
|
if (pack_length == 1)
|
2007-10-06 11:39:26 +02:00
|
|
|
*pos= (uchar) tmp;
|
2006-04-11 15:45:10 +02:00
|
|
|
else
|
|
|
|
|
int2store(pos,tmp);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
2007-10-04 19:33:42 +02:00
|
|
|
sprintf((char*) pos,"... row: %d", rownr);
|
|
|
|
|
strappend((char*) pos,recinfo[1].length,' ');
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* change row to test re-packing of rows and reallocation of keys */
|
|
|
|
|
|
2007-10-04 19:33:42 +02:00
|
|
|
static void update_record(uchar *record)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
2007-10-04 19:33:42 +02:00
|
|
|
uchar *pos=record+1;
|
2007-01-18 20:38:14 +01:00
|
|
|
if (recinfo[0].type == FIELD_BLOB)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
2007-10-04 19:33:42 +02:00
|
|
|
uchar *column,*ptr;
|
2006-04-11 15:45:10 +02:00
|
|
|
int length;
|
|
|
|
|
length=uint4korr(pos); /* Long blob */
|
|
|
|
|
memcpy_fixed(&column,pos+4,sizeof(char*));
|
|
|
|
|
memcpy(blob_key,column,length); /* Move old key */
|
|
|
|
|
ptr=blob_key;
|
|
|
|
|
memcpy_fixed(pos+4,&ptr,sizeof(char*)); /* Store pointer to new key */
|
|
|
|
|
if (keyinfo[0].seg[0].type != HA_KEYTYPE_NUM)
|
|
|
|
|
default_charset_info->cset->casedn(default_charset_info,
|
2007-10-04 19:33:42 +02:00
|
|
|
(char*) blob_key, length,
|
|
|
|
|
(char*) blob_key, length);
|
2007-01-18 20:38:14 +01:00
|
|
|
pos+=recinfo[0].length;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
2007-01-18 20:38:14 +01:00
|
|
|
else if (recinfo[0].type == FIELD_VARCHAR)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
2007-01-18 20:38:14 +01:00
|
|
|
uint pack_length= HA_VARCHAR_PACKLENGTH(recinfo[0].length-1);
|
2006-04-11 15:45:10 +02:00
|
|
|
uint length= pack_length == 1 ? (uint) *(uchar*) pos : uint2korr(pos);
|
|
|
|
|
default_charset_info->cset->casedn(default_charset_info,
|
2007-10-04 19:33:42 +02:00
|
|
|
(char*) pos + pack_length, length,
|
|
|
|
|
(char*) pos + pack_length, length);
|
2007-01-18 20:38:14 +01:00
|
|
|
pos+=recinfo[0].length;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
if (keyinfo[0].seg[0].type != HA_KEYTYPE_NUM)
|
|
|
|
|
default_charset_info->cset->casedn(default_charset_info,
|
2007-10-04 19:33:42 +02:00
|
|
|
(char*) pos, keyinfo[0].seg[0].length,
|
|
|
|
|
(char*) pos, keyinfo[0].seg[0].length);
|
2007-01-18 20:38:14 +01:00
|
|
|
pos+=recinfo[0].length;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
|
2007-01-18 20:38:14 +01:00
|
|
|
if (recinfo[1].type == FIELD_BLOB)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
2007-10-04 19:33:42 +02:00
|
|
|
uchar *column;
|
2006-04-11 15:45:10 +02:00
|
|
|
int length;
|
|
|
|
|
length=uint4korr(pos);
|
|
|
|
|
memcpy_fixed(&column,pos+4,sizeof(char*));
|
|
|
|
|
memcpy(blob_record,column,length);
|
|
|
|
|
bfill(blob_record+length,20,'.'); /* Make it larger */
|
|
|
|
|
length+=20;
|
|
|
|
|
int4store(pos,length);
|
|
|
|
|
column=blob_record;
|
|
|
|
|
memcpy_fixed(pos+4,&column,sizeof(char*));
|
|
|
|
|
}
|
2007-01-18 20:38:14 +01:00
|
|
|
else if (recinfo[1].type == FIELD_VARCHAR)
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
/* Second field is longer than 10 characters */
|
|
|
|
|
uint pack_length= HA_VARCHAR_PACKLENGTH(recinfo[1].length-1);
|
|
|
|
|
uint length= pack_length == 1 ? (uint) *(uchar*) pos : uint2korr(pos);
|
2007-01-18 20:38:14 +01:00
|
|
|
pos= record+ recinfo[1].offset;
|
|
|
|
|
bfill(pos+pack_length+length,recinfo[1].length-length-pack_length,'.');
|
|
|
|
|
length=recinfo[1].length-pack_length;
|
2006-04-11 15:45:10 +02:00
|
|
|
if (pack_length == 1)
|
|
|
|
|
*(uchar*) pos= (uchar) length;
|
|
|
|
|
else
|
|
|
|
|
int2store(pos,length);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
2007-01-18 20:38:14 +01:00
|
|
|
bfill(pos+recinfo[1].length-10,10,'.');
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
static struct my_option my_long_options[] =
|
|
|
|
|
{
|
WL#3071 Maria checkpoint
Ability for flush_pagecache_blocks() to flush only certain pages of
a file, as instructed by an option "filter" pointer-to-function argument;
Checkpoint and background dirty page flushing use that to flush only
pages which have been dirty for long enough and bitmap pages.
Fix for a bug in flush_cached_blocks() (no idea if it could produce
a bug in real life, but theoretically it is).
Testing checkpoint in ma_test_recovery via ma_test1 and ma_test2.
Background checkpoint & dirty pages flush thread is still disabled
by default in ha_maria.
mysql-test/r/maria.result:
result update
storage/maria/ha_maria.cc:
blank after function comment
storage/maria/ma_checkpoint.c:
Using an enum instead of 0/1/2 (applying Sanja's review comments).
The comment about "this is an horizon" can be removed as Sanja
created translog_next_LSN() which parse_checkpoint_record() uses.
Variables in ma_checkpoint_background() cannot be declared in the
for() as their value must not be reset at each iteration!
storage/maria/ma_pagecache.c:
adding to flush_pagecache_blocks() optional arguments 'filter'
(pointer to function) and 'filter_arg'; if filter!=NULL this function
will be called for each block of the file and will reply if this
block and following ones should be flushed or not (3 possible
replies).
Fixing a bug when flush_cached_blocks() skips a pinned page: it has
to unset PCBLOCK_IN_FLUSH set by flush_pagecache_blocks_int().
storage/maria/ma_pagecache.h:
flush_pagecache_blocks() is changed to take "filter" and "filter_arg"
arguments. "filter", if it is not NULL, may return one value
among enum pagecache_flush_filter_result.
storage/maria/ma_recovery.c:
open_count=0 when closing tables at the end of recovery.
storage/maria/ma_test1.c:
Optional checkpoints (-H#) at various stages (stages similar
to --testflag), for testing of checkpoints.
storage/maria/ma_test2.c:
Optional checkpoints (-H#) at various stages (stages similar
to -t), for testing of checkpoints.
storage/maria/ma_test_recovery.expected:
Result update: the results of the additional test run with -H#
(checkpoints) are added here. They are exactly identical to without
checkpoints except that the index's Root (printed by maria_chk)
is more correct when using checkpoints. This is because checkpoint
flushed the state, so it happens to be correct, while no-checkpoint
does not flush the state, and recovery does not recover indexes
so Root is never fixed. When we recover indices, this will go away.
storage/maria/ma_test_recovery:
We duplicate the loop of tests to add an additional run with
checkpoints at various stages, to see if maria_read_log
uses them fine.
2007-10-17 16:55:26 +02:00
|
|
|
{"checkpoint", 'H', "Checkpoint at specified stage", (uchar**) &checkpoint,
|
|
|
|
|
(uchar**) &checkpoint, 0, GET_INT, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
|
2006-04-11 15:45:10 +02:00
|
|
|
{"checksum", 'c', "Undocumented",
|
|
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
|
|
|
|
#ifndef DBUG_OFF
|
|
|
|
|
{"debug", '#', "Undocumented",
|
|
|
|
|
0, 0, 0, GET_STR, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
|
|
|
|
|
#endif
|
2007-01-18 20:38:14 +01:00
|
|
|
{"delete-rows", 'd', "Abort after this many rows has been deleted",
|
2007-07-02 19:45:15 +02:00
|
|
|
(uchar**) &remove_count, (uchar**) &remove_count, 0, GET_UINT, REQUIRED_ARG,
|
2007-01-18 20:38:14 +01:00
|
|
|
1000, 0, 0, 0, 0, 0},
|
2006-04-11 15:45:10 +02:00
|
|
|
{"help", '?', "Display help and exit",
|
|
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-07-02 19:45:15 +02:00
|
|
|
{"insert-rows", 'i', "Undocumented", (uchar**) &insert_count,
|
|
|
|
|
(uchar**) &insert_count, 0, GET_UINT, REQUIRED_ARG, 1000, 0, 0, 0, 0, 0},
|
2007-01-18 20:38:14 +01:00
|
|
|
{"key-alpha", 'a', "Use a key of type HA_KEYTYPE_TEXT",
|
2006-04-11 15:45:10 +02:00
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-01-18 20:38:14 +01:00
|
|
|
{"key-binary-pack", 'B', "Undocumented",
|
2006-04-11 15:45:10 +02:00
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-01-18 20:38:14 +01:00
|
|
|
{"key-blob", 'b', "Undocumented",
|
2007-10-19 23:24:22 +02:00
|
|
|
(uchar**) &blob_length, (uchar**) &blob_length,
|
|
|
|
|
0, GET_ULONG, OPT_ARG, 0, 0, 0, 0, 0, 0},
|
2007-07-02 19:45:15 +02:00
|
|
|
{"key-cache", 'K', "Undocumented", (uchar**) &pagecacheing,
|
|
|
|
|
(uchar**) &pagecacheing, 0, GET_BOOL, NO_ARG, 0, 0, 0, 0, 0, 0},
|
|
|
|
|
{"key-length", 'k', "Undocumented", (uchar**) &key_length,
|
|
|
|
|
(uchar**) &key_length, 0, GET_UINT, REQUIRED_ARG, 6, 0, 0, 0, 0, 0},
|
First part of redo/undo for key pages
Added key_nr to st_maria_keydef for faster keyinfo->keynr conversion
For transactional tables, shift record number in keys up with 1 bit to have place to indicate if transid follows
Checksum for MyISAM now ignores NULL and not used part of VARCHAR
Renamed some variables that caused shadow compiler warnings
Moved extra() call when waiting for tables to not be used to after tables are removed from cache.
Fixed crashing bugs when using Maria TEMPORARY tables with TRUNCATE. Removed 'hack' code in sql directory to go around this bug.
pagecache_unlock_by_ulink() now has extra argument to say if page was changed.
Give error message if we fail to open control file
Mark page cache variables as not flushable
include/maria.h:
Made min page cache larger (needed for pinning key page)
Added key_nr to st_maria_keydef for faster keyinfo->keynr conversion
Added write_comp_flag to move some runtime code to maria_open()
include/my_base.h:
Added new error message to be used when handler initialization failed
include/my_global.h:
Renamed dummy to swap_dummy to avoid conflicts with local 'dummy' variables
include/my_handler.h:
Added const to some parameters
mysys/array.c:
More DBUG
mysys/my_error.c:
Fixed indentation
mysys/my_handler.c:
Added const to some parameters
Added missing error messages
sql/field.h:
Renamed variables to avoid variable shadowing
sql/handler.h:
Renamed parameter to avoid variable name conflict
sql/item.h:
Renamed variables to avoid variable shadowing
sql/log_event_old.h:
Renamed variables to avoid variable shadowing
sql/set_var.h:
Renamed variables to avoid variable shadowing
sql/sql_delete.cc:
Removed maria hack for temporary tables
Fixed indentation
sql/sql_table.cc:
Moved extra() call when waiting for tables to not be used to after tables are removed from cache.
This was needed to ensure we don't do a PREPARE_FOR_DROP or similar call while the table is still in use.
sql/table.cc:
Copy page_checksum from share
Removed Maria hack
storage/maria/Makefile.am:
Added new files
storage/maria/ha_maria.cc:
Renamed records -> record_count and info -> create_info to avoid variable name conflicts
Mark page cache variables as not flushable
storage/maria/ma_blockrec.c:
Moved _ma_unpin_all_pages() to ma_key_recover.c
Moved init of info->pinned_pages to ma_open.c
Moved _ma_finalize_row() to maria_key_recover.h
Renamed some variables to avoid variable name conflicts
Mark page_link.changed for blocks we change directly
Simplify handling of undo link when writing LOGREC_UNDO_ROW_INSERT (old code crashed when having redo for index)
storage/maria/ma_blockrec.h:
Removed extra empty line
storage/maria/ma_checkpoint.c:
Remove not needed trnman.h
storage/maria/ma_close.c:
Free pinned pages (which are now always allocated)
storage/maria/ma_control_file.c:
Give error message if we fail to open control file
storage/maria/ma_delete.c:
Changes for redo logging (first part, logging of underflow not yet done)
- Log undo-key-delete
- Log delete of key
- Updated arguments to _ma_fetch_keypage(), _ma_dispose(), _ma_write_keypage(), _ma_insert()
- Added new arguments to some functions to be able to write redo information
- Mark key pages as changed when we write with PAGECACHE_LOCK_LEFT_WRITELOCKED
Remove one not needed _ma_write_keypage() in d_search() when upper level will do the write anyway
Changed 2 bmove_upp() to bmove() as this made code easer to understand
More function comments
Indentation fixes
storage/maria/ma_ft_update.c:
New arguments to _ma_write_keypage()
storage/maria/ma_loghandler.c:
Fixed some DBUG_PRINT messages
Simplify code
Added new log entrys for key page redo
Renamed some variables to avoid variable name shadowing
storage/maria/ma_loghandler.h:
Moved some defines here
Added define for storing key number on key pages
Added new translog record types
Added enum for type of operations in LOGREC_REDO_INDEX
storage/maria/ma_open.c:
Always allocate info.pinned_pages (we need now also for normal key page usage)
Update keyinfo->key_nr
Added virtual functions to convert record position o number to be stored on key pages
Update keyinfo->write_comp_flag to value of search flag to be used when writing key
storage/maria/ma_page.c:
Added redo for key pages
- Extended _ma_fetch_keypage() with type of lock to put on page and address to used MARIA_PINNED_PAGE
- _ma_fetch_keypage() now pin's pages if needed
- Extended _ma_write_keypage() with type of locks to be used
- ma_dispose() now locks info->s->state.key_del from other threads
- ma_dispose() writes redo log record
- ma_new() locks info->s->state.key_del from other threads if it was used
- ma_new() now pins read page
Other things:
- Removed some not needed arguments from _ma_new() and _ma_dispose)
- Added some new variables to simplify code
- If EXTRA_DEBUG is used, do crc on full page to catch not unitialized bytes
storage/maria/ma_pagecache.h:
Applied patch from Sanja to add extra argument to pagecache_unlock_by_ulink() to mark if page was changed
Added some defines for pagecache priority levels that one can use
storage/maria/ma_range.c:
Added new arguments for call to _ma_fetch_keypage()
storage/maria/ma_recovery.c:
- Added hooks for new translog types:
REDO_INDEX, REDO_INDEX_NEW_PAGE, REDO_INDEX_FREE_PAGE, UNDO_KEY_INSERT, UNDO_KEY_DELETE and
UNDO_KEY_DELETE_WITH_ROOT.
- Moved variable declarations to start of function (portability fixes)
- Removed some not needed initializations
- Set only relevant state changes for each redo/undo entry
storage/maria/lockman.c:
Removed end space
storage/maria/ma_check.c:
Removed end space
storage/maria/ma_create.c:
Removed end space
storage/maria/ma_locking.c:
Removed end space
storage/maria/ma_packrec.c:
Removed end space
storage/maria/ma_pagecache.c:
Removed end space
storage/maria/ma_panic.c:
Removed end space
storage/maria/ma_rt_index.c:
Added new arguments for call to _ma_fetch_keypage(), _ma_write_keypage(), _ma_dispose() and _ma_new()
Fixed indentation
storage/maria/ma_rt_key.c:
Added new arguments for call to _ma_fetch_keypage()
storage/maria/ma_rt_split.c:
Added new arguments for call to _ma_new()
Use new keypage header
Added new arguments for call to _ma_write_keypage()
storage/maria/ma_search.c:
Updated comments & indentation
Added new arguments for call to _ma_fetch_keypage()
Made some variables and arguments const
Added virtual functions for converting row position to number to be stored in key
use MARIA_RECORD_POS of record position instead of my_off_t
Record in MARIA_KEY_PARAM how page was changed one key insert (needed for REDO)
storage/maria/ma_sort.c:
Removed end space
storage/maria/ma_statrec.c:
Updated arguments for call to _ma_rec_pos()
storage/maria/ma_test1.c:
Fixed too small buffer to init_pagecache()
Fixed bug when using insert_count and test_flag
storage/maria/ma_test2.c:
Use more resonable pagecache size
Remove not used code
Reset blob_length to fix wrong output message
storage/maria/ma_test_all.sh:
Fixed wrong test
storage/maria/ma_write.c:
Lots of new code to handle REDO of key pages
No logic changes because of REDO code, mostly adding new arguments and adding new code for logging
Added new arguments for calls to _ma_fetch_keypage(), _ma_write_keypage() and similar functions
Move setting of comp_flag in ma_ck_wrte_btree() from runtime to maria_open()
Zerofill new used pages for:
- To remove possible sensitive data left in buffer
- To get idenitical data on pages after running redo
- Better compression of pages if archived
storage/maria/maria_chk.c:
Added information if table is crash safe
storage/maria/maria_def.h:
New virtual function to convert between record position on key and normal record position
Aded mutex and extra variables to handle locking of share->state.key_del
Moved some structure variables to get things more aligned
Added extra arguments to MARIA_KEY_PARAM to be able to remember what was changed on key page on key insert
Added argument to MARIA_PINNED_PAGE to indicate if page was changed
Updated prototypes for functions
Added some structures for signaling changes in REDO handling
storage/maria/unittest/ma_pagecache_single.c:
Updated arguments for changed function calls
storage/myisam/mi_check.c:
Made calc_check_checksum virtual
storage/myisam/mi_checksum.c:
Update checksums to ignore null columns
storage/myisam/mi_create.c:
Mark if table has null column (to know when we have to use mi_checksum())
storage/myisam/mi_open.c:
Added virtual function for calculating checksum to be able to easily ignore NULL fields
storage/myisam/mi_test2.c:
Fixed bug
storage/myisam/myisamdef.h:
Added virtual function for calculating checksum during check table
Removed ha_key_cmp() as this is in handler.h
storage/maria/ma_key_recover.c:
New BitKeeper file ``storage/maria/ma_key_recover.c''
storage/maria/ma_key_recover.h:
New BitKeeper file ``storage/maria/ma_key_recover.h''
storage/maria/ma_key_redo.c:
New BitKeeper file ``storage/maria/ma_key_redo.c''
2007-11-14 18:08:06 +01:00
|
|
|
{"key-multiple", 'm', "Don't use unique keys",
|
2006-04-11 15:45:10 +02:00
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-01-18 20:38:14 +01:00
|
|
|
{"key-prefix_pack", 'P', "Undocumented",
|
2006-04-11 15:45:10 +02:00
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-01-18 20:38:14 +01:00
|
|
|
{"key-space_pack", 'p', "Undocumented",
|
2006-04-11 15:45:10 +02:00
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-01-18 20:38:14 +01:00
|
|
|
{"key-varchar", 'w', "Test VARCHAR keys",
|
2006-04-11 15:45:10 +02:00
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-01-18 20:38:14 +01:00
|
|
|
{"null-fields", 'N', "Define fields with NULL",
|
2007-07-02 19:45:15 +02:00
|
|
|
(uchar**) &null_fields, (uchar**) &null_fields, 0, GET_BOOL, NO_ARG,
|
2006-04-11 15:45:10 +02:00
|
|
|
0, 0, 0, 0, 0, 0},
|
2007-01-18 20:38:14 +01:00
|
|
|
{"row-fixed-size", 'S', "Fixed size records",
|
|
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
|
|
|
|
{"rows-in-block", 'M', "Store rows in block format",
|
2006-04-11 15:45:10 +02:00
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-07-02 19:45:15 +02:00
|
|
|
{"row-pointer-size", 'R', "Undocumented", (uchar**) &rec_pointer_size,
|
|
|
|
|
(uchar**) &rec_pointer_size, 0, GET_INT, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
|
2006-04-11 15:45:10 +02:00
|
|
|
{"silent", 's', "Undocumented",
|
2007-07-02 19:45:15 +02:00
|
|
|
(uchar**) &silent, (uchar**) &silent, 0, GET_BOOL, NO_ARG, 0, 0, 0, 0,
|
|
|
|
|
0, 0},
|
|
|
|
|
{"skip-delete", 'U', "Don't test deletes", (uchar**) &skip_delete,
|
|
|
|
|
(uchar**) &skip_delete, 0, GET_BOOL, NO_ARG, 0, 0, 0, 0, 0, 0},
|
|
|
|
|
{"skip-update", 'D', "Don't test updates", (uchar**) &skip_update,
|
|
|
|
|
(uchar**) &skip_update, 0, GET_BOOL, NO_ARG, 0, 0, 0, 0, 0, 0},
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
{"testflag", 't', "Stop test at specified stage", (uchar**) &testflag,
|
|
|
|
|
(uchar**) &testflag, 0, GET_INT, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
|
2007-09-04 09:53:52 +02:00
|
|
|
{"test-undo", 'A',
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
"Abort hard. Used for testing recovery with undo",
|
2007-09-04 09:53:52 +02:00
|
|
|
(uchar**) &die_in_middle_of_transaction,
|
|
|
|
|
(uchar**) &die_in_middle_of_transaction,
|
WL#3072 Maria Recovery
misc fixes of execution of UNDOs in the UNDO phase:
- into the CLR_END, store the LSN of the _previous_ UNDO (we debated
what was best, so far we're going with "previous"; later we can change
to "current" if needed), and store the type of record which is being
undone (needed to know how to update state.records when we see the
CLR_END during the REDO phase).
- declaring all UNDOs and CLR_END as "compressed"
- when executing an UNDO in the UNDO phase, state.records is updated
as a hook when writing CLR_END (needed for "recovery of the state"),
and so is trn->undo_lsn (needed for when we have checkpoints).
- bugfix (execution of UNDO_ROW_DELETE didn't store the correct checksum
into the re-inserted row, maria_chk -r thus threw the row away).
- modifications of ma_test1: where to stop is now driven by --testflag;
--test-undo just tells how to stop (flush data, flush log, nothing).
- ma_test_recovery: testing of the UNDO phase, more testing of the
REDO phase, identification of a bug.
storage/maria/ma_blockrec.c:
- bugfix: execution of UNDO_ROW_DELETE didn't store the correct
checksum into the row (leading to "maria_chk -r" eliminating the
re-inserted row, net effect was that rollback appeared to have
rolled back no deletion). Reason was that write_block_record() used
info->cur_row.checksum, while "row" can be != &info->cur_row
(case of UNDO_ROW_DELETE). After fixing this, problems with
_ma_update_block_record() appeared; indeed checksum was computed
by allocate_and_write_block_record() while _ma_update_block_record()
directly calls write_block_record(). Solution is to compute checksum
in write_block_record() instead.
- when executing an UNDO, we now pass the LSN of the _previous_ UNDO
to block_format functions. This LSN can be 0 (if the being-executed UNDO
was the transaction's first UNDO), so "undo_lsn==0" cannot work
anymore to indicate "this is not UNDO work". Using undo_lsn==LSN_ERROR
instead (this is an impossible LSN).
- store into CLR_END the type of log record which was undone
(INSERT/UPDATE/DELETE); needed for Recovery to know if/how it has
to update state.records if it sees this CLR_END in the REDO phase.
- when writing the CLR_END in _ma_apply_undo_row_insert(),
the place to store file's id is log_data+LSN_STORE_SIZE.
- in _ma_apply_undo_row_insert(), the records-- is moved
to a hook when writing the CLR_END (this way it is under log's mutex
which is needed for "recovery of the state")
storage/maria/ma_loghandler.c:
- all UNDOs, and CLR_END, start with the LSN of another UNDO; so
we can declare them "compressed".
- write_hook_for_clr_end() to set trn->undo_lsn (to the previous
UNDO's LSN) under log's lock (like UNDOs set trn->undo_lsn under log's
lock), and also update, if appropriate, state.records.
- reset share->id to 0 when deassigning; not useful for now but
sounds logical.
storage/maria/ma_recovery.c:
- if no table is found for a REDO, it's not an error; for an UNDO, it is
- in the REDO phase, when we see a CLR_END we must update trn->undo_lsn
and sometimes state.records.
- in the UNDO phase, when we execute an UNDO_ROW_INSERT:
* update trn->undo_lsn only after executing the record
* store the _previous_ undo_lsn into the CLR_END
- at the end of the REDO phase, when we recreate TRN objects, they
have already their long id in the log (either via a
LOGREC_LONG_TRANSACTION_ID, or in a checkpoint record), don't write
a new, useless LOGREC_LONG_TRANSACTION_ID for them.
storage/maria/ma_test1.c:
* where to stop execution is now driven by --testflag and not --test-undo
(ma_test2 already has --testflag for the same purpose). This allows
us to do a clean stop (with commit) at any point.
* --test-undo=# tells how to abort (flush all pages (which implies
flushing log) or only log or nothing); all such "ways of crashing"
are tested in ma_test_recovery
storage/maria/ma_test_recovery:
* Testing execution of UNDOs, with and without BLOBs.
* Testing idempotency of REDOs.
* See @todo for a probable bug with BLOBs.
* maria_chk -rq instead of -r, as with -q it nicely stops on any
problem in the data file (like the checksum bug see comment of
ma_blockrec.c).
* Testing if log was written by UNDO phase (often expected),
not written by REDO phase (always expected).
* Less output on the screen, compares with expected output in the end.
* some shell thingies like "set --" and $# are courtesy of
Danny and Pekka.
storage/maria/maria_read_log.c:
when only displaying the records, don't do an UNDO phase
storage/maria/ma_test_recovery.expected:
This is the expected output of a great part of ma_test_recovery.
ma_test_recovery compares its output to the expected output
and tells if different.
If we look at this file it mentions differences in checksum
(normal, it's not recovered yet) and in records count
(getting a correct records' count when recovery starts on an
already existing table, like when testing rollback,
is coded but not yet pushed).
2007-09-06 16:04:36 +02:00
|
|
|
0, GET_INT, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
|
2007-07-02 19:45:15 +02:00
|
|
|
{"transactional", 'T',
|
|
|
|
|
"Test in transactional mode. (Only works with block format)",
|
|
|
|
|
(uchar**) &transactional, (uchar**) &transactional, 0, GET_BOOL, NO_ARG,
|
This patch is a collection of patches from from Sanja, Sergei and Monty.
Added logging and pinning of pages to block format.
Integration of transaction manager, log handler.
Better page cache intergration
Split trnman.h into two files, so that we don't have to include my_atomic.h into C++ programs.
Renaming of structures, more comments, more debugging etc.
Fixed problem with small head block + long varchar.
Added extra argument to delete_record() and update_record() (needed for UNDO logging)
Small changes to interface of pagecache and log handler.
Change initialization of log_record_type_descriptors to not be depending on enum order.
Use array of LEX_STRING's to send data to log handler
Added 'dummy' transaction option to MARIA_INFO so that we can always assume 'trn' exists.
include/lf.h:
Interface fixes
Rename of structures
(Patch from Sergei via Sanja)
include/my_atomic.h:
More comments
include/my_global.h:
Added MY_ERRPTR
include/pagecache.h:
Added undo LSN when unlocking pages
mysql-test/r/maria.result:
Updated results
mysql-test/t/maria.test:
Added autocommit around lock tables
(Patch from Sanja)
mysys/lf_alloc-pin.c:
Post-review fixes, simple optimizations
More comments
Struct slot renames
Check amount of memory on stack
(Patch from Sergei)
mysys/lf_dynarray.c:
More comments
mysys/lf_hash.c:
More comments
After review fixes
(Patch from Sergei)
storage/maria/ha_maria.cc:
Split trnman.h into two files, so that we don't have to include my_atomic.h into the .cc program.
(Temporary fix to avoid bug in gcc)
Move out all deferencing of the transaction structure.
Transaction manager integrated (Patch from Sergei)
storage/maria/ha_maria.h:
Added prototype for start_stmt()
storage/maria/lockman.c:
Function call rename
storage/maria/ma_bitmap.c:
Mark deleted pages free from page cache
storage/maria/ma_blockrec.c:
Offset -> rownr
More debugging
Fixed problem with small head block + long varchar
Added logging of changed pages
Added logging of undo (Including only loggging of changed fields in case of update)
Added pinning/unpinning of all changed pages
More comments
Added free_full_pages() as the same code was used in several places.
fill_rows_parts() renamed as fill_insert_undo_parts()
offset -> rownr
Added some optimization of not transactional tables
_ma_update_block_record() has new parameter, as we need original row to do efficent undo for update
storage/maria/ma_blockrec.h:
Added ROW_EXTENTS_ON_STACK
Changed prototype for update and delete of row
storage/maria/ma_check.c:
Added original row to delete_record() call
storage/maria/ma_control_file.h:
Added ifdefs for C++
storage/maria/ma_delete.c:
Added original row to delete_record() call
(Needed for efficent undo logging)
storage/maria/ma_dynrec.c:
Added extra argument to delete_record() and update_record()
Removed not used variable
storage/maria/ma_init.c:
Initialize log handler
storage/maria/ma_loghandler.c:
Removed not used variable
Change initialization of log_record_type_descriptors to not be depending on enum order
Use array of LEX_STRING's to send data to log handler
storage/maria/ma_loghandler.h:
New defines
Use array of LEX_STRING's to send data to log handler
storage/maria/ma_open.c:
Added 'dummy' transaction option to MARIA_INFO so that we can always assume 'trn' exists.
Store in MARIA_SHARE->page_type if pages will have up to date LSN's
storage/maria/ma_pagecache.c:
Don't decrease number of readers when using pagecache_write()/pagecache_read()
In pagecache_write() decrement request count if page was left pinned
Added pagecache_delete_pages()
Removed some casts
Make trace output consistent with rest of code
Simplify calling of DBUG_ASSERT(0)
Only update LSN if the LSN is bigger than what's already on the page
Added LSN parameter pagecache_unpin_page(), pagecache_unpin(), and pagecache_unlock()
(Part of patch from Sanja)
storage/maria/ma_static.c:
Added 'dummy' transaction option to MARIA_INFO so that we can always assume 'trn' exists.
Added default page cache
storage/maria/ma_statrec.c:
Added extra argument to delete_record() and update_record()
storage/maria/ma_test1.c:
Added option -T for transactions
storage/maria/ma_test2.c:
Added option -T for transactions
storage/maria/ma_test_all.sh:
Test with transactions
storage/maria/ma_update.c:
Changed prototype for update of row
storage/maria/maria_def.h:
Changed prototype for update & delete of row as block records need to access the old row
Store in MARIA_SHARE->page_type if pages will have up to date LSN's
Added MARIA_MAX_TREE_LEVELS to allow us to calculate the number of possible pinned pages we may need.
Removed not used 'empty_bits_buffer'
Added pointer to transaction object
Added array for pinned pages
Added log_row_parts array for logging of field data.
Added MARIA_PINNED_PAGE to store pinned pages
storage/maria/trnman.c:
Added accessor functions to transaction object
Added missing DBUG_RETURN()
More debugging
More comments
Changed // comment of code to #ifdef NOT_USED
Transaction manager integrated.
Post review fixes
Part of patch originally from Sergei
storage/maria/trnman.h:
Split trnman.h into two files, so that we don't have to include my_atomic.h into the .cc program.
(Temporary fix to avoid bug in gcc)
storage/maria/unittest/ma_pagecache_single.c:
Added missing argument
Added SKIP_BIG_TESTS
(Patch from Sanja)
storage/maria/unittest/ma_test_loghandler-t.c:
Test logging with new LEX_STRING parameter
(Patch from Sanja)
storage/maria/unittest/ma_test_loghandler_multigroup-t.c:
Test logging with new LEX_STRING parameter
(Patch from Sanja)
storage/maria/unittest/ma_test_loghandler_multithread-t.c:
Test logging with new LEX_STRING parameter
(Patch from Sanja)
storage/maria/unittest/ma_test_loghandler_pagecache-t.c:
Test logging with new LEX_STRING parameter
(Patch from Sanja)
storage/maria/unittest/trnman-t.c:
Stack overflow detection
(Patch from Sergei)
unittest/unit.pl:
Command-line options --big and --verbose
(Patch from Sergei)
unittest/mytap/tap.c:
Detect --big
(Patch from Sergei)
unittest/mytap/tap.h:
Skip_big_tests and SKIP_BIG_TESTS
(Patch from Sergei)
storage/maria/trnman_public.h:
New BitKeeper file ``storage/maria/trnman_public.h''
2007-05-29 19:13:56 +02:00
|
|
|
0, 0, 0, 0, 0, 0},
|
2007-07-02 19:45:15 +02:00
|
|
|
{"unique", 'C', "Undocumented", (uchar**) &opt_unique,
|
|
|
|
|
(uchar**) &opt_unique, 0, GET_BOOL, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2007-08-31 09:19:54 +02:00
|
|
|
{"update-rows", 'u', "Max number of rows to update", (uchar**) &update_count,
|
2007-07-02 19:45:15 +02:00
|
|
|
(uchar**) &update_count, 0, GET_UINT, REQUIRED_ARG, 1000, 0, 0, 0, 0, 0},
|
|
|
|
|
{"verbose", 'v', "Be more verbose", (uchar**) &verbose,
|
|
|
|
|
(uchar**) &verbose, 0, GET_BOOL, NO_ARG, 0, 0, 0, 0, 0, 0},
|
2006-04-11 15:45:10 +02:00
|
|
|
{"version", 'V', "Print version number and exit",
|
|
|
|
|
0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0},
|
|
|
|
|
{ 0, 0, 0, 0, 0, 0, GET_NO_ARG, NO_ARG, 0, 0, 0, 0, 0, 0}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
static my_bool
|
|
|
|
|
get_one_option(int optid, const struct my_option *opt __attribute__((unused)),
|
2007-06-09 13:52:17 +02:00
|
|
|
char *argument __attribute__((unused)))
|
2006-04-11 15:45:10 +02:00
|
|
|
{
|
|
|
|
|
switch(optid) {
|
|
|
|
|
case 'a':
|
|
|
|
|
key_type= HA_KEYTYPE_TEXT;
|
|
|
|
|
break;
|
|
|
|
|
case 'c':
|
|
|
|
|
create_flag|= HA_CREATE_CHECKSUM;
|
|
|
|
|
break;
|
|
|
|
|
case 'R': /* Length of record pointer */
|
|
|
|
|
if (rec_pointer_size > 3)
|
|
|
|
|
rec_pointer_size=0;
|
|
|
|
|
break;
|
|
|
|
|
case 'P':
|
|
|
|
|
pack_keys= HA_PACK_KEY; /* Use prefix compression */
|
|
|
|
|
break;
|
|
|
|
|
case 'B':
|
|
|
|
|
pack_keys= HA_BINARY_PACK_KEY; /* Use binary compression */
|
|
|
|
|
break;
|
2007-01-18 20:38:14 +01:00
|
|
|
case 'M':
|
|
|
|
|
record_type= BLOCK_RECORD;
|
|
|
|
|
break;
|
2006-04-11 15:45:10 +02:00
|
|
|
case 'S':
|
|
|
|
|
if (key_field == FIELD_VARCHAR)
|
|
|
|
|
{
|
|
|
|
|
create_flag=0; /* Static sized varchar */
|
2007-01-18 20:38:14 +01:00
|
|
|
record_type= STATIC_RECORD;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
else if (key_field != FIELD_BLOB)
|
|
|
|
|
{
|
|
|
|
|
key_field=FIELD_NORMAL; /* static-size record */
|
|
|
|
|
extra_field=FIELD_NORMAL;
|
2007-01-18 20:38:14 +01:00
|
|
|
record_type= STATIC_RECORD;
|
2006-04-11 15:45:10 +02:00
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
case 'p':
|
|
|
|
|
pack_keys=HA_PACK_KEY; /* Use prefix + space packing */
|
|
|
|
|
pack_seg=HA_SPACE_PACK;
|
|
|
|
|
key_type=HA_KEYTYPE_TEXT;
|
|
|
|
|
break;
|
|
|
|
|
case 'm':
|
|
|
|
|
unique_key=0;
|
|
|
|
|
break;
|
|
|
|
|
case 'b':
|
|
|
|
|
key_field=FIELD_BLOB; /* blob key */
|
|
|
|
|
extra_field= FIELD_BLOB;
|
|
|
|
|
pack_seg|= HA_BLOB_PART;
|
|
|
|
|
key_type= HA_KEYTYPE_VARTEXT1;
|
2007-01-18 20:38:14 +01:00
|
|
|
if (record_type == STATIC_RECORD)
|
|
|
|
|
record_type= DYNAMIC_RECORD;
|
2006-04-11 15:45:10 +02:00
|
|
|
break;
|
|
|
|
|
case 'k':
|
|
|
|
|
if (key_length < 4 || key_length > HA_MAX_KEY_LENGTH)
|
|
|
|
|
{
|
|
|
|
|
fprintf(stderr,"Wrong key length\n");
|
|
|
|
|
exit(1);
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
case 'w':
|
|
|
|
|
key_field=FIELD_VARCHAR; /* varchar keys */
|
|
|
|
|
extra_field= FIELD_VARCHAR;
|
|
|
|
|
key_type= HA_KEYTYPE_VARTEXT1;
|
|
|
|
|
pack_seg|= HA_VAR_LENGTH_PART;
|
2007-01-18 20:38:14 +01:00
|
|
|
if (record_type == STATIC_RECORD)
|
|
|
|
|
record_type= DYNAMIC_RECORD;
|
2006-04-11 15:45:10 +02:00
|
|
|
break;
|
|
|
|
|
case 'K': /* Use key cacheing */
|
2007-04-04 22:37:09 +02:00
|
|
|
pagecacheing=1;
|
2006-04-11 15:45:10 +02:00
|
|
|
break;
|
|
|
|
|
case 'V':
|
|
|
|
|
printf("test1 Ver 1.2 \n");
|
|
|
|
|
exit(0);
|
|
|
|
|
case '#':
|
2007-12-04 22:23:42 +01:00
|
|
|
DBUG_PUSH(argument);
|
2006-04-11 15:45:10 +02:00
|
|
|
break;
|
|
|
|
|
case '?':
|
|
|
|
|
usage();
|
|
|
|
|
exit(1);
|
|
|
|
|
}
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/* Read options */
|
|
|
|
|
|
|
|
|
|
static void get_options(int argc, char *argv[])
|
|
|
|
|
{
|
|
|
|
|
int ho_error;
|
|
|
|
|
|
|
|
|
|
if ((ho_error=handle_options(&argc, &argv, my_long_options, get_one_option)))
|
|
|
|
|
exit(ho_error);
|
|
|
|
|
|
|
|
|
|
return;
|
|
|
|
|
} /* get options */
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
static void usage()
|
|
|
|
|
{
|
|
|
|
|
printf("Usage: %s [options]\n\n", my_progname);
|
|
|
|
|
my_print_help(my_long_options);
|
|
|
|
|
my_print_variables(my_long_options);
|
|
|
|
|
}
|