2001-02-17 13:19:19 +01:00
|
|
|
/************************************************************************
|
|
|
|
|
The index tree adaptive search
|
|
|
|
|
|
|
|
|
|
(c) 1996 Innobase Oy
|
|
|
|
|
|
|
|
|
|
Created 2/17/1996 Heikki Tuuri
|
|
|
|
|
*************************************************************************/
|
|
|
|
|
|
|
|
|
|
#ifndef btr0sea_h
|
|
|
|
|
#define btr0sea_h
|
|
|
|
|
|
|
|
|
|
#include "univ.i"
|
|
|
|
|
|
|
|
|
|
#include "rem0rec.h"
|
|
|
|
|
#include "dict0dict.h"
|
|
|
|
|
#include "btr0types.h"
|
|
|
|
|
#include "mtr0mtr.h"
|
|
|
|
|
#include "ha0ha.h"
|
|
|
|
|
|
|
|
|
|
/*********************************************************************
|
|
|
|
|
Creates and initializes the adaptive search system at a database start. */
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
btr_search_sys_create(
|
|
|
|
|
/*==================*/
|
|
|
|
|
ulint hash_size); /* in: hash index hash table size */
|
|
|
|
|
/************************************************************************
|
|
|
|
|
Returns search info for an index. */
|
|
|
|
|
UNIV_INLINE
|
|
|
|
|
btr_search_t*
|
|
|
|
|
btr_search_get_info(
|
|
|
|
|
/*================*/
|
|
|
|
|
/* out: search info; search mutex reserved */
|
|
|
|
|
dict_index_t* index); /* in: index */
|
|
|
|
|
/*********************************************************************
|
|
|
|
|
Creates and initializes a search info struct. */
|
|
|
|
|
|
|
|
|
|
btr_search_t*
|
|
|
|
|
btr_search_info_create(
|
|
|
|
|
/*===================*/
|
|
|
|
|
/* out, own: search info struct */
|
|
|
|
|
mem_heap_t* heap); /* in: heap where created */
|
|
|
|
|
/*************************************************************************
|
|
|
|
|
Updates the search info. */
|
|
|
|
|
UNIV_INLINE
|
|
|
|
|
void
|
|
|
|
|
btr_search_info_update(
|
|
|
|
|
/*===================*/
|
|
|
|
|
dict_index_t* index, /* in: index of the cursor */
|
|
|
|
|
btr_cur_t* cursor);/* in: cursor which was just positioned */
|
|
|
|
|
/**********************************************************************
|
|
|
|
|
Tries to guess the right search position based on the hash search info
|
|
|
|
|
of the index. Note that if mode is PAGE_CUR_LE, which is used in inserts,
|
|
|
|
|
and the function returns TRUE, then cursor->up_match and cursor->low_match
|
|
|
|
|
both have sensible values. */
|
|
|
|
|
|
|
|
|
|
ibool
|
|
|
|
|
btr_search_guess_on_hash(
|
|
|
|
|
/*=====================*/
|
Applied innodb-5.1-ss269 snapshot.
Fixed BUGS:
#3300: "UPDATE statement with no index column in where condition locks
all rows"
Implement semi-consistent read to reduce lock conflicts at the cost
of breaking serializability.
ha_innobase::unlock_row(): reset the "did semi consistent read" flag
ha_innobase::was_semi_consistent_read(),
ha_innobase::try_semi_consistent_read(): new methods
row_prebuilt_t, row_create_prebuilt(): add field row_read_type for
keeping track of semi-consistent reads
row_vers_build_for_semi_consistent_read(),
row_sel_build_committed_vers_for_mysql(): new functions
row_search_for_mysql(): implement semi-consistent reads
#9802: "Foreign key checks disallow alter table".
Added test cases.
#12456: "Cursor shows incorrect data - DML does not affect,
probably caching"
This patch implements a high-granularity read view to be used with
cursors. In this high-granularity consistent read view modifications
done by the creating transaction after the cursor is created or
future transactions are not visible. But those modifications that
transaction did before the cursor was created are visible.
#12701: "Support >4GB buffer pool and log files on 64-bit Windows"
Do not call os_file_create_tmpfile() at runtime. Instead, create all
tempfiles at startup and guard access to them with mutexes.
#13778: "If FOREIGN_KEY_CHECKS=0, one can create inconsistent FOREIGN KEYs".
When FOREIGN_KEY_CHECKS=0 we still need to check that datatypes between
foreign key references are compatible.
#14189: "VARBINARY and BINARY variables: trailing space ignored with InnoDB"
innobase_init(): Assert that
DATA_MYSQL_BINARY_CHARSET_COLL == my_charset_bin.number.
dtype_get_pad_char(): Do not pad VARBINARY or BINARY columns.
row_ins_cascade_calc_update_vec(): Refuse ON UPDATE CASCADE when trying
to change the length of a VARBINARY column that refers to or is referenced
by a BINARY column. BINARY columns are no longer padded on comparison,
and thus they cannot be padded on storage either.
#14747: "Race condition can cause btr_search_drop_page_hash_index() to crash"
Note that buf_block_t::index should be protected by btr_search_latch
or an s-latch or x-latch on the index page.
btr_search_drop_page_hash_index(): Read block->index while holding
btr_search_latch and use the cached value in the loop. Remove some
redundant assertions.
#15108: "mysqld crashes when innodb_log_file_size is set > 4G"
#15308: "Problem of Order with Enum Column in Primary Key"
#15550: "mysqld crashes in printing a FOREIGN KEY error in InnoDB"
row_ins_foreign_report_add_err(): When printing the parent record,
use the index in the parent table rather than the index in the child table.
#15653: "Slow inserts to InnoDB if many thousands of .ibd files"
Keep track on unflushed modifications to file spaces. When there are tens
of thousands of file spaces, flushing all files in fil_flush_file_spaces()
would be very slow.
fil_flush_file_spaces(): Only flush unflushed file spaces.
fil_space_t, fil_system_t: Add a list of unflushed spaces.
#15991: "innodb-file-per-table + symlink database + rename = cr"
os_file_handle_error(): Map the error codes EXDEV, ENOTDIR, and EISDIR
to the new code OS_FILE_PATH_ERROR. Treat this code as OS_FILE_PATH_ERROR.
This fixes the crash on RENAME TABLE when the .ibd file is a symbolic link
to a different file system.
#16157: "InnoDB crashes when main location settings are empty"
This patch is from Heikki.
#16298: "InnoDB segfaults in INSERTs in upgrade of 4.0 -> 5.0 tables
with VARCHAR BINARY"
dict_load_columns(): Set the charset-collation code
DATA_MYSQL_BINARY_CHARSET_COLL for those binary string columns
that lack a charset-collation code, i.e., the tables were created
with an older version of MySQL/InnoDB than 4.1.2.
#16229: "MySQL/InnoDB uses full explicit table locks in trigger processing"
Take a InnoDB table lock only if user has explicitly requested a table
lock. Added some additional comments to store_lock() and external_lock().
#16387: "InnoDB crash when dropping a foreign key <table>_ibfk_0"
Do not mistake TABLENAME_ibfk_0 for auto-generated id.
dict_table_get_highest_foreign_id(): Ignore foreign constraint
identifiers starting with the pattern TABLENAME_ibfk_0.
#16582: "InnoDB: Error in an adaptive hash index pointer to page"
Account for a race condition when dropping the adaptive hash index
for a B-tree page.
btr_search_drop_page_hash_index(): Retry the operation if a hash index
with different parameters was built meanwhile. Add diagnostics for the
case that hash node pointers to the page remain.
btr_search_info_update_hash(), btr_search_info_update_slow():
Document the parameter "info" as in/out.
#16814: "SHOW INNODB STATUS format error in LATEST FOREIGN KEY ERROR
section"
Add a missing newline to the LAST FOREIGN KEY ERROR section in SHOW
INNODB STATUS output.
dict_foreign_error_report(): Always print a newline after invoking
dict_print_info_on_foreign_key_in_create_format().
#16827: "Better InnoDB error message if ibdata files omitted from my.cnf"
#17126: "CHECK TABLE on InnoDB causes a short hang during check of adaptive
hash"
CHECK TABLE blocking other queries, by releasing the btr_search_latch
periodically during the adaptive hash table validation.
#17405: "Valgrind: conditional jump or move depends on unititialised values"
buf_block_init(): Reset magic_n, buf_fix_count and io_fix to avoid
testing uninitialized variables.
2006-03-10 17:22:21 +01:00
|
|
|
/* out: TRUE if succeeded */
|
2001-02-17 13:19:19 +01:00
|
|
|
dict_index_t* index, /* in: index */
|
|
|
|
|
btr_search_t* info, /* in: index search info */
|
|
|
|
|
dtuple_t* tuple, /* in: logical record */
|
|
|
|
|
ulint mode, /* in: PAGE_CUR_L, ... */
|
Applied innodb-5.1-ss269 snapshot.
Fixed BUGS:
#3300: "UPDATE statement with no index column in where condition locks
all rows"
Implement semi-consistent read to reduce lock conflicts at the cost
of breaking serializability.
ha_innobase::unlock_row(): reset the "did semi consistent read" flag
ha_innobase::was_semi_consistent_read(),
ha_innobase::try_semi_consistent_read(): new methods
row_prebuilt_t, row_create_prebuilt(): add field row_read_type for
keeping track of semi-consistent reads
row_vers_build_for_semi_consistent_read(),
row_sel_build_committed_vers_for_mysql(): new functions
row_search_for_mysql(): implement semi-consistent reads
#9802: "Foreign key checks disallow alter table".
Added test cases.
#12456: "Cursor shows incorrect data - DML does not affect,
probably caching"
This patch implements a high-granularity read view to be used with
cursors. In this high-granularity consistent read view modifications
done by the creating transaction after the cursor is created or
future transactions are not visible. But those modifications that
transaction did before the cursor was created are visible.
#12701: "Support >4GB buffer pool and log files on 64-bit Windows"
Do not call os_file_create_tmpfile() at runtime. Instead, create all
tempfiles at startup and guard access to them with mutexes.
#13778: "If FOREIGN_KEY_CHECKS=0, one can create inconsistent FOREIGN KEYs".
When FOREIGN_KEY_CHECKS=0 we still need to check that datatypes between
foreign key references are compatible.
#14189: "VARBINARY and BINARY variables: trailing space ignored with InnoDB"
innobase_init(): Assert that
DATA_MYSQL_BINARY_CHARSET_COLL == my_charset_bin.number.
dtype_get_pad_char(): Do not pad VARBINARY or BINARY columns.
row_ins_cascade_calc_update_vec(): Refuse ON UPDATE CASCADE when trying
to change the length of a VARBINARY column that refers to or is referenced
by a BINARY column. BINARY columns are no longer padded on comparison,
and thus they cannot be padded on storage either.
#14747: "Race condition can cause btr_search_drop_page_hash_index() to crash"
Note that buf_block_t::index should be protected by btr_search_latch
or an s-latch or x-latch on the index page.
btr_search_drop_page_hash_index(): Read block->index while holding
btr_search_latch and use the cached value in the loop. Remove some
redundant assertions.
#15108: "mysqld crashes when innodb_log_file_size is set > 4G"
#15308: "Problem of Order with Enum Column in Primary Key"
#15550: "mysqld crashes in printing a FOREIGN KEY error in InnoDB"
row_ins_foreign_report_add_err(): When printing the parent record,
use the index in the parent table rather than the index in the child table.
#15653: "Slow inserts to InnoDB if many thousands of .ibd files"
Keep track on unflushed modifications to file spaces. When there are tens
of thousands of file spaces, flushing all files in fil_flush_file_spaces()
would be very slow.
fil_flush_file_spaces(): Only flush unflushed file spaces.
fil_space_t, fil_system_t: Add a list of unflushed spaces.
#15991: "innodb-file-per-table + symlink database + rename = cr"
os_file_handle_error(): Map the error codes EXDEV, ENOTDIR, and EISDIR
to the new code OS_FILE_PATH_ERROR. Treat this code as OS_FILE_PATH_ERROR.
This fixes the crash on RENAME TABLE when the .ibd file is a symbolic link
to a different file system.
#16157: "InnoDB crashes when main location settings are empty"
This patch is from Heikki.
#16298: "InnoDB segfaults in INSERTs in upgrade of 4.0 -> 5.0 tables
with VARCHAR BINARY"
dict_load_columns(): Set the charset-collation code
DATA_MYSQL_BINARY_CHARSET_COLL for those binary string columns
that lack a charset-collation code, i.e., the tables were created
with an older version of MySQL/InnoDB than 4.1.2.
#16229: "MySQL/InnoDB uses full explicit table locks in trigger processing"
Take a InnoDB table lock only if user has explicitly requested a table
lock. Added some additional comments to store_lock() and external_lock().
#16387: "InnoDB crash when dropping a foreign key <table>_ibfk_0"
Do not mistake TABLENAME_ibfk_0 for auto-generated id.
dict_table_get_highest_foreign_id(): Ignore foreign constraint
identifiers starting with the pattern TABLENAME_ibfk_0.
#16582: "InnoDB: Error in an adaptive hash index pointer to page"
Account for a race condition when dropping the adaptive hash index
for a B-tree page.
btr_search_drop_page_hash_index(): Retry the operation if a hash index
with different parameters was built meanwhile. Add diagnostics for the
case that hash node pointers to the page remain.
btr_search_info_update_hash(), btr_search_info_update_slow():
Document the parameter "info" as in/out.
#16814: "SHOW INNODB STATUS format error in LATEST FOREIGN KEY ERROR
section"
Add a missing newline to the LAST FOREIGN KEY ERROR section in SHOW
INNODB STATUS output.
dict_foreign_error_report(): Always print a newline after invoking
dict_print_info_on_foreign_key_in_create_format().
#16827: "Better InnoDB error message if ibdata files omitted from my.cnf"
#17126: "CHECK TABLE on InnoDB causes a short hang during check of adaptive
hash"
CHECK TABLE blocking other queries, by releasing the btr_search_latch
periodically during the adaptive hash table validation.
#17405: "Valgrind: conditional jump or move depends on unititialised values"
buf_block_init(): Reset magic_n, buf_fix_count and io_fix to avoid
testing uninitialized variables.
2006-03-10 17:22:21 +01:00
|
|
|
ulint latch_mode, /* in: BTR_SEARCH_LEAF, ... */
|
|
|
|
|
btr_cur_t* cursor, /* out: tree cursor */
|
2001-02-17 13:19:19 +01:00
|
|
|
ulint has_search_latch,/* in: latch mode the caller
|
|
|
|
|
currently has on btr_search_latch:
|
|
|
|
|
RW_S_LATCH, RW_X_LATCH, or 0 */
|
|
|
|
|
mtr_t* mtr); /* in: mtr */
|
|
|
|
|
/************************************************************************
|
|
|
|
|
Moves or deletes hash entries for moved records. If new_page is already hashed,
|
|
|
|
|
then the hash index for page, if any, is dropped. If new_page is not hashed,
|
|
|
|
|
and page is hashed, then a new hash index is built to new_page with the same
|
|
|
|
|
parameters as page (this often happens when a page is split). */
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
btr_search_move_or_delete_hash_entries(
|
|
|
|
|
/*===================================*/
|
2004-12-02 18:45:07 +01:00
|
|
|
page_t* new_page, /* in: records are copied
|
|
|
|
|
to this page */
|
|
|
|
|
page_t* page, /* in: index page */
|
|
|
|
|
dict_index_t* index); /* in: record descriptor */
|
2001-02-17 13:19:19 +01:00
|
|
|
/************************************************************************
|
|
|
|
|
Drops a page hash index. */
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
btr_search_drop_page_hash_index(
|
|
|
|
|
/*============================*/
|
|
|
|
|
page_t* page); /* in: index page, s- or x-latched */
|
|
|
|
|
/************************************************************************
|
|
|
|
|
Drops a page hash index when a page is freed from a fseg to the file system.
|
|
|
|
|
Drops possible hash index if the page happens to be in the buffer pool. */
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
btr_search_drop_page_hash_when_freed(
|
|
|
|
|
/*=================================*/
|
|
|
|
|
ulint space, /* in: space id */
|
|
|
|
|
ulint page_no); /* in: page number */
|
|
|
|
|
/************************************************************************
|
|
|
|
|
Updates the page hash index when a single record is inserted on a page. */
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
btr_search_update_hash_node_on_insert(
|
|
|
|
|
/*==================================*/
|
|
|
|
|
btr_cur_t* cursor);/* in: cursor which was positioned to the
|
|
|
|
|
place to insert using btr_cur_search_...,
|
|
|
|
|
and the new record has been inserted next
|
|
|
|
|
to the cursor */
|
|
|
|
|
/************************************************************************
|
|
|
|
|
Updates the page hash index when a single record is inserted on a page. */
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
btr_search_update_hash_on_insert(
|
|
|
|
|
/*=============================*/
|
|
|
|
|
btr_cur_t* cursor);/* in: cursor which was positioned to the
|
|
|
|
|
place to insert using btr_cur_search_...,
|
|
|
|
|
and the new record has been inserted next
|
|
|
|
|
to the cursor */
|
|
|
|
|
/************************************************************************
|
|
|
|
|
Updates the page hash index when a single record is deleted from a page. */
|
|
|
|
|
|
|
|
|
|
void
|
|
|
|
|
btr_search_update_hash_on_delete(
|
|
|
|
|
/*=============================*/
|
|
|
|
|
btr_cur_t* cursor);/* in: cursor which was positioned on the
|
|
|
|
|
record to delete using btr_cur_search_...,
|
|
|
|
|
the record is not yet deleted */
|
|
|
|
|
/************************************************************************
|
|
|
|
|
Validates the search system. */
|
|
|
|
|
|
|
|
|
|
ibool
|
2004-12-08 13:33:09 +01:00
|
|
|
btr_search_validate(void);
|
|
|
|
|
/*======================*/
|
2004-12-02 18:45:07 +01:00
|
|
|
/* out: TRUE if ok */
|
2001-02-17 13:19:19 +01:00
|
|
|
|
|
|
|
|
/* The search info struct in an index */
|
|
|
|
|
|
|
|
|
|
struct btr_search_struct{
|
2006-09-21 09:39:09 +02:00
|
|
|
/* The following fields are not protected by any latch.
|
|
|
|
|
Unfortunately, this means that they must be aligned to
|
|
|
|
|
the machine word, i.e., they cannot be turned into bit-fields. */
|
2001-02-17 13:19:19 +01:00
|
|
|
page_t* root_guess; /* the root page frame when it was last time
|
|
|
|
|
fetched, or NULL */
|
2006-09-21 09:39:09 +02:00
|
|
|
ulint hash_analysis; /* when this exceeds BTR_SEARCH_HASH_ANALYSIS,
|
|
|
|
|
the hash analysis starts; this is reset if no
|
2001-02-17 13:19:19 +01:00
|
|
|
success noticed */
|
|
|
|
|
ibool last_hash_succ; /* TRUE if the last search would have
|
|
|
|
|
succeeded, or did succeed, using the hash
|
|
|
|
|
index; NOTE that the value here is not exact:
|
|
|
|
|
it is not calculated for every search, and the
|
|
|
|
|
calculation itself is not always accurate! */
|
2006-09-21 09:39:09 +02:00
|
|
|
ulint n_hash_potential;
|
|
|
|
|
/* number of consecutive searches
|
|
|
|
|
which would have succeeded, or did succeed,
|
|
|
|
|
using the hash index;
|
|
|
|
|
the range is 0 .. BTR_SEARCH_BUILD_LIMIT + 5 */
|
Applied innodb-5.1-ss269 snapshot.
Fixed BUGS:
#3300: "UPDATE statement with no index column in where condition locks
all rows"
Implement semi-consistent read to reduce lock conflicts at the cost
of breaking serializability.
ha_innobase::unlock_row(): reset the "did semi consistent read" flag
ha_innobase::was_semi_consistent_read(),
ha_innobase::try_semi_consistent_read(): new methods
row_prebuilt_t, row_create_prebuilt(): add field row_read_type for
keeping track of semi-consistent reads
row_vers_build_for_semi_consistent_read(),
row_sel_build_committed_vers_for_mysql(): new functions
row_search_for_mysql(): implement semi-consistent reads
#9802: "Foreign key checks disallow alter table".
Added test cases.
#12456: "Cursor shows incorrect data - DML does not affect,
probably caching"
This patch implements a high-granularity read view to be used with
cursors. In this high-granularity consistent read view modifications
done by the creating transaction after the cursor is created or
future transactions are not visible. But those modifications that
transaction did before the cursor was created are visible.
#12701: "Support >4GB buffer pool and log files on 64-bit Windows"
Do not call os_file_create_tmpfile() at runtime. Instead, create all
tempfiles at startup and guard access to them with mutexes.
#13778: "If FOREIGN_KEY_CHECKS=0, one can create inconsistent FOREIGN KEYs".
When FOREIGN_KEY_CHECKS=0 we still need to check that datatypes between
foreign key references are compatible.
#14189: "VARBINARY and BINARY variables: trailing space ignored with InnoDB"
innobase_init(): Assert that
DATA_MYSQL_BINARY_CHARSET_COLL == my_charset_bin.number.
dtype_get_pad_char(): Do not pad VARBINARY or BINARY columns.
row_ins_cascade_calc_update_vec(): Refuse ON UPDATE CASCADE when trying
to change the length of a VARBINARY column that refers to or is referenced
by a BINARY column. BINARY columns are no longer padded on comparison,
and thus they cannot be padded on storage either.
#14747: "Race condition can cause btr_search_drop_page_hash_index() to crash"
Note that buf_block_t::index should be protected by btr_search_latch
or an s-latch or x-latch on the index page.
btr_search_drop_page_hash_index(): Read block->index while holding
btr_search_latch and use the cached value in the loop. Remove some
redundant assertions.
#15108: "mysqld crashes when innodb_log_file_size is set > 4G"
#15308: "Problem of Order with Enum Column in Primary Key"
#15550: "mysqld crashes in printing a FOREIGN KEY error in InnoDB"
row_ins_foreign_report_add_err(): When printing the parent record,
use the index in the parent table rather than the index in the child table.
#15653: "Slow inserts to InnoDB if many thousands of .ibd files"
Keep track on unflushed modifications to file spaces. When there are tens
of thousands of file spaces, flushing all files in fil_flush_file_spaces()
would be very slow.
fil_flush_file_spaces(): Only flush unflushed file spaces.
fil_space_t, fil_system_t: Add a list of unflushed spaces.
#15991: "innodb-file-per-table + symlink database + rename = cr"
os_file_handle_error(): Map the error codes EXDEV, ENOTDIR, and EISDIR
to the new code OS_FILE_PATH_ERROR. Treat this code as OS_FILE_PATH_ERROR.
This fixes the crash on RENAME TABLE when the .ibd file is a symbolic link
to a different file system.
#16157: "InnoDB crashes when main location settings are empty"
This patch is from Heikki.
#16298: "InnoDB segfaults in INSERTs in upgrade of 4.0 -> 5.0 tables
with VARCHAR BINARY"
dict_load_columns(): Set the charset-collation code
DATA_MYSQL_BINARY_CHARSET_COLL for those binary string columns
that lack a charset-collation code, i.e., the tables were created
with an older version of MySQL/InnoDB than 4.1.2.
#16229: "MySQL/InnoDB uses full explicit table locks in trigger processing"
Take a InnoDB table lock only if user has explicitly requested a table
lock. Added some additional comments to store_lock() and external_lock().
#16387: "InnoDB crash when dropping a foreign key <table>_ibfk_0"
Do not mistake TABLENAME_ibfk_0 for auto-generated id.
dict_table_get_highest_foreign_id(): Ignore foreign constraint
identifiers starting with the pattern TABLENAME_ibfk_0.
#16582: "InnoDB: Error in an adaptive hash index pointer to page"
Account for a race condition when dropping the adaptive hash index
for a B-tree page.
btr_search_drop_page_hash_index(): Retry the operation if a hash index
with different parameters was built meanwhile. Add diagnostics for the
case that hash node pointers to the page remain.
btr_search_info_update_hash(), btr_search_info_update_slow():
Document the parameter "info" as in/out.
#16814: "SHOW INNODB STATUS format error in LATEST FOREIGN KEY ERROR
section"
Add a missing newline to the LAST FOREIGN KEY ERROR section in SHOW
INNODB STATUS output.
dict_foreign_error_report(): Always print a newline after invoking
dict_print_info_on_foreign_key_in_create_format().
#16827: "Better InnoDB error message if ibdata files omitted from my.cnf"
#17126: "CHECK TABLE on InnoDB causes a short hang during check of adaptive
hash"
CHECK TABLE blocking other queries, by releasing the btr_search_latch
periodically during the adaptive hash table validation.
#17405: "Valgrind: conditional jump or move depends on unititialised values"
buf_block_init(): Reset magic_n, buf_fix_count and io_fix to avoid
testing uninitialized variables.
2006-03-10 17:22:21 +01:00
|
|
|
/*----------------------*/
|
2001-02-17 13:19:19 +01:00
|
|
|
ulint n_fields; /* recommended prefix length for hash search:
|
|
|
|
|
number of full fields */
|
|
|
|
|
ulint n_bytes; /* recommended prefix: number of bytes in
|
2006-09-21 09:39:09 +02:00
|
|
|
an incomplete field;
|
2007-01-05 03:51:34 +01:00
|
|
|
see also BTR_PAGE_MAX_REC_SIZE */
|
2006-09-21 09:39:09 +02:00
|
|
|
ibool left_side; /* TRUE or FALSE, depending on whether
|
2001-02-17 13:19:19 +01:00
|
|
|
the leftmost record of several records with
|
|
|
|
|
the same prefix should be indexed in the
|
|
|
|
|
hash index */
|
|
|
|
|
/*----------------------*/
|
Applied innodb-5.1-ss475 snapshot.
* Fix BUG#15650: "DELETE with LEFT JOIN crashes server with innodb_locks_unsafe_for binlog"
* Fix BUG#17134: "Partitions: uncommitted changes are visible"
* Fix BUG#17992: "Partitions: InnoDB, somehow rotten table after UPDATE"
row0ins.c: MySQL's partitioned table code does not set preduilt->sql_stat_start right
if it does an insert in the same statement after doing a search first in the same
partition table. We now write trx id always to the buffer, not just when flag
sql_stat_start is on. This will waste CPU time very sightly.
* Fix BUG#18077: "InnoDB uses full explicit table locks in stored FUNCTION"
* Fix BUG#18238: "When locks exhaust the buffer pool, InnoDB does not roll back the trx"
* Fix BUG#18252" "Disk space leak in updates of InnoDB BLOB rows in 5.0 and 5.1"
* Fix BUG#18283: "When InnoDB returns error 'lock table full', MySQL can write to binlog too much"
* Fix BUG#18350: "Use consistent read in CREATE ... SELECT ... if innodb_locks_unsafe_for_binlog"
* Fix BUG#18384: "InnoDB memory leak on duplicate key errors in 5.0 if row has many columns"
* Fix BUG#18934: "InnoDB crashes when table uses column names like DB_ROW_ID"
Refuse tables that use reserved column names.
* InnoDB's SQL parser:
- Add support for UNSIGNED types, EXIT keyword, quoted identifiers, user-function callbacks
for processing results of FETCH statements, bound literals, DATA_VARCHAR for bound literals.
- Allow bound literals of type non-INTEGER to be of length 0.
- Add make_flex.sh and update lexer/parser generation documentation.
- Add comment clarifying the difference between 'alias' and 'indirection' fields in sym_node_t.
- Remove never reached duplicate code in pars_set_dfield_type().
- Rewrite pars_info datatypes and APIs, add a few helper functions.
- Since the functions definitions in pars_info_t are accessed after pars_sql() returns
in the query graph execution stage, we can't free pars_info_t in pars_sql(). Instead,
make pars_sql() transfer ownership of pars_info_t to the created query graph, and
make que_graph_free() free it if needed.
- Allow access to system columns like DB_ROW_ID.
* Use bound literals in row_truncate_table_for_mysql, row_drop_table_for_mysql,
row_discard_tablespace_for_mysql, and row_rename_table_for_mysql.
* Setting an isolation level of the transaction to read committed weakens the locks for
this session similarly like the option innodb_locks_unsafe_for binlog. This patch removes
alnost all gap locking (used in next-key locking) and makes MySQL to release the row locks
on the rows which does not belong to result set. Additionally, nonlocking selects on
INSERT INTO SELECT, UPDATE ... (SELECT ...), and CREATE ... SELECT ... use a nonlocking
consistent read. If a binlog is used, then binlog format should be set to row based
binloging to make the execution of the complex SQL statements.
* Disable the statistic variables btr_search_n_hash_fail and n_hash_succ, n_hash_fail,
n_patt_succ, and n_searches of btr_search_t in builds without #ifdef UNIV_SEARCH_PERF_STAT.
* Make innodb.test faster. Group all consistent read test cases to a one test case and
wait their lock timeout after all have been send to the server. Decrease amount of rows
inserted in a certain test - this has no effect on the effectiveness of the test and
reduces the running time by ~10 sec. Remove temporary work-arounds from innodb.result
now that ALTER TABLE DROP FOREIGN KEY works once again.
* Make innodb_unsafe_binlog.test faster. Grout all consistent read test cases to a one
test case amd wait their lock timeout after all have been sent to the server. Remove
unnecessary option --loose_innodb_lock_wait_timeout.
* Print dictionary memory size in SHOW INNODB STATUS.
* Fix memory leaks in row_create_table_for_mysql() in rare corner cases.
* Remove code related to clustered tables. They were never implemented, and the
implementation would be challenging with ROW_FORMAT=COMPACT. Remove the table types
DICT_TABLE_CLUSTER_MEMBER and DICT_TABLE_CLUSTER and all related tests and functions.
dict_table_t: Remove mix_id, mix_len, mix_id_len, mix_id_buf, and cluster_name.
plan_t: Remove mixed_index.
dict_create_sys_tables_tuple(): Set MIX_ID=0, MIX_LEN=0, CLUSTER_NAME=NULL when
inserting into SYS_TABLES.
dict_tree_check_search_tuple(): Enclose in #ifdef UNIV_DEBUG.
* Move calling of thr_local_free() from trx_free_for_mysql() to
innobase_close_connection().
2006-04-23 10:48:31 +02:00
|
|
|
#ifdef UNIV_SEARCH_PERF_STAT
|
2001-02-17 13:19:19 +01:00
|
|
|
ulint n_hash_succ; /* number of successful hash searches thus
|
|
|
|
|
far */
|
|
|
|
|
ulint n_hash_fail; /* number of failed hash searches */
|
|
|
|
|
ulint n_patt_succ; /* number of successful pattern searches thus
|
|
|
|
|
far */
|
|
|
|
|
ulint n_searches; /* number of searches */
|
Applied innodb-5.1-ss475 snapshot.
* Fix BUG#15650: "DELETE with LEFT JOIN crashes server with innodb_locks_unsafe_for binlog"
* Fix BUG#17134: "Partitions: uncommitted changes are visible"
* Fix BUG#17992: "Partitions: InnoDB, somehow rotten table after UPDATE"
row0ins.c: MySQL's partitioned table code does not set preduilt->sql_stat_start right
if it does an insert in the same statement after doing a search first in the same
partition table. We now write trx id always to the buffer, not just when flag
sql_stat_start is on. This will waste CPU time very sightly.
* Fix BUG#18077: "InnoDB uses full explicit table locks in stored FUNCTION"
* Fix BUG#18238: "When locks exhaust the buffer pool, InnoDB does not roll back the trx"
* Fix BUG#18252" "Disk space leak in updates of InnoDB BLOB rows in 5.0 and 5.1"
* Fix BUG#18283: "When InnoDB returns error 'lock table full', MySQL can write to binlog too much"
* Fix BUG#18350: "Use consistent read in CREATE ... SELECT ... if innodb_locks_unsafe_for_binlog"
* Fix BUG#18384: "InnoDB memory leak on duplicate key errors in 5.0 if row has many columns"
* Fix BUG#18934: "InnoDB crashes when table uses column names like DB_ROW_ID"
Refuse tables that use reserved column names.
* InnoDB's SQL parser:
- Add support for UNSIGNED types, EXIT keyword, quoted identifiers, user-function callbacks
for processing results of FETCH statements, bound literals, DATA_VARCHAR for bound literals.
- Allow bound literals of type non-INTEGER to be of length 0.
- Add make_flex.sh and update lexer/parser generation documentation.
- Add comment clarifying the difference between 'alias' and 'indirection' fields in sym_node_t.
- Remove never reached duplicate code in pars_set_dfield_type().
- Rewrite pars_info datatypes and APIs, add a few helper functions.
- Since the functions definitions in pars_info_t are accessed after pars_sql() returns
in the query graph execution stage, we can't free pars_info_t in pars_sql(). Instead,
make pars_sql() transfer ownership of pars_info_t to the created query graph, and
make que_graph_free() free it if needed.
- Allow access to system columns like DB_ROW_ID.
* Use bound literals in row_truncate_table_for_mysql, row_drop_table_for_mysql,
row_discard_tablespace_for_mysql, and row_rename_table_for_mysql.
* Setting an isolation level of the transaction to read committed weakens the locks for
this session similarly like the option innodb_locks_unsafe_for binlog. This patch removes
alnost all gap locking (used in next-key locking) and makes MySQL to release the row locks
on the rows which does not belong to result set. Additionally, nonlocking selects on
INSERT INTO SELECT, UPDATE ... (SELECT ...), and CREATE ... SELECT ... use a nonlocking
consistent read. If a binlog is used, then binlog format should be set to row based
binloging to make the execution of the complex SQL statements.
* Disable the statistic variables btr_search_n_hash_fail and n_hash_succ, n_hash_fail,
n_patt_succ, and n_searches of btr_search_t in builds without #ifdef UNIV_SEARCH_PERF_STAT.
* Make innodb.test faster. Group all consistent read test cases to a one test case and
wait their lock timeout after all have been send to the server. Decrease amount of rows
inserted in a certain test - this has no effect on the effectiveness of the test and
reduces the running time by ~10 sec. Remove temporary work-arounds from innodb.result
now that ALTER TABLE DROP FOREIGN KEY works once again.
* Make innodb_unsafe_binlog.test faster. Grout all consistent read test cases to a one
test case amd wait their lock timeout after all have been sent to the server. Remove
unnecessary option --loose_innodb_lock_wait_timeout.
* Print dictionary memory size in SHOW INNODB STATUS.
* Fix memory leaks in row_create_table_for_mysql() in rare corner cases.
* Remove code related to clustered tables. They were never implemented, and the
implementation would be challenging with ROW_FORMAT=COMPACT. Remove the table types
DICT_TABLE_CLUSTER_MEMBER and DICT_TABLE_CLUSTER and all related tests and functions.
dict_table_t: Remove mix_id, mix_len, mix_id_len, mix_id_buf, and cluster_name.
plan_t: Remove mixed_index.
dict_create_sys_tables_tuple(): Set MIX_ID=0, MIX_LEN=0, CLUSTER_NAME=NULL when
inserting into SYS_TABLES.
dict_tree_check_search_tuple(): Enclose in #ifdef UNIV_DEBUG.
* Move calling of thr_local_free() from trx_free_for_mysql() to
innobase_close_connection().
2006-04-23 10:48:31 +02:00
|
|
|
#endif /* UNIV_SEARCH_PERF_STAT */
|
2006-09-21 09:39:09 +02:00
|
|
|
#ifdef UNIV_DEBUG
|
|
|
|
|
ulint magic_n; /* magic number */
|
|
|
|
|
# define BTR_SEARCH_MAGIC_N 1112765
|
|
|
|
|
#endif /* UNIV_DEBUG */
|
2001-02-17 13:19:19 +01:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/* The hash index system */
|
|
|
|
|
|
|
|
|
|
typedef struct btr_search_sys_struct btr_search_sys_t;
|
|
|
|
|
|
|
|
|
|
struct btr_search_sys_struct{
|
|
|
|
|
hash_table_t* hash_index;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
extern btr_search_sys_t* btr_search_sys;
|
|
|
|
|
|
|
|
|
|
/* The latch protecting the adaptive search system: this latch protects the
|
2003-01-18 14:01:56 +01:00
|
|
|
(1) hash index;
|
|
|
|
|
(2) columns of a record to which we have a pointer in the hash index;
|
|
|
|
|
|
|
|
|
|
but does NOT protect:
|
|
|
|
|
|
|
|
|
|
(3) next record offset field in a record;
|
|
|
|
|
(4) next or previous records on the same page.
|
|
|
|
|
|
|
|
|
|
Bear in mind (3) and (4) when using the hash index.
|
|
|
|
|
*/
|
2001-02-17 13:19:19 +01:00
|
|
|
|
|
|
|
|
extern rw_lock_t* btr_search_latch_temp;
|
|
|
|
|
|
|
|
|
|
#define btr_search_latch (*btr_search_latch_temp)
|
|
|
|
|
|
2004-03-16 17:55:44 +01:00
|
|
|
#ifdef UNIV_SEARCH_PERF_STAT
|
2001-02-17 13:19:19 +01:00
|
|
|
extern ulint btr_search_n_succ;
|
|
|
|
|
extern ulint btr_search_n_hash_fail;
|
Applied innodb-5.1-ss475 snapshot.
* Fix BUG#15650: "DELETE with LEFT JOIN crashes server with innodb_locks_unsafe_for binlog"
* Fix BUG#17134: "Partitions: uncommitted changes are visible"
* Fix BUG#17992: "Partitions: InnoDB, somehow rotten table after UPDATE"
row0ins.c: MySQL's partitioned table code does not set preduilt->sql_stat_start right
if it does an insert in the same statement after doing a search first in the same
partition table. We now write trx id always to the buffer, not just when flag
sql_stat_start is on. This will waste CPU time very sightly.
* Fix BUG#18077: "InnoDB uses full explicit table locks in stored FUNCTION"
* Fix BUG#18238: "When locks exhaust the buffer pool, InnoDB does not roll back the trx"
* Fix BUG#18252" "Disk space leak in updates of InnoDB BLOB rows in 5.0 and 5.1"
* Fix BUG#18283: "When InnoDB returns error 'lock table full', MySQL can write to binlog too much"
* Fix BUG#18350: "Use consistent read in CREATE ... SELECT ... if innodb_locks_unsafe_for_binlog"
* Fix BUG#18384: "InnoDB memory leak on duplicate key errors in 5.0 if row has many columns"
* Fix BUG#18934: "InnoDB crashes when table uses column names like DB_ROW_ID"
Refuse tables that use reserved column names.
* InnoDB's SQL parser:
- Add support for UNSIGNED types, EXIT keyword, quoted identifiers, user-function callbacks
for processing results of FETCH statements, bound literals, DATA_VARCHAR for bound literals.
- Allow bound literals of type non-INTEGER to be of length 0.
- Add make_flex.sh and update lexer/parser generation documentation.
- Add comment clarifying the difference between 'alias' and 'indirection' fields in sym_node_t.
- Remove never reached duplicate code in pars_set_dfield_type().
- Rewrite pars_info datatypes and APIs, add a few helper functions.
- Since the functions definitions in pars_info_t are accessed after pars_sql() returns
in the query graph execution stage, we can't free pars_info_t in pars_sql(). Instead,
make pars_sql() transfer ownership of pars_info_t to the created query graph, and
make que_graph_free() free it if needed.
- Allow access to system columns like DB_ROW_ID.
* Use bound literals in row_truncate_table_for_mysql, row_drop_table_for_mysql,
row_discard_tablespace_for_mysql, and row_rename_table_for_mysql.
* Setting an isolation level of the transaction to read committed weakens the locks for
this session similarly like the option innodb_locks_unsafe_for binlog. This patch removes
alnost all gap locking (used in next-key locking) and makes MySQL to release the row locks
on the rows which does not belong to result set. Additionally, nonlocking selects on
INSERT INTO SELECT, UPDATE ... (SELECT ...), and CREATE ... SELECT ... use a nonlocking
consistent read. If a binlog is used, then binlog format should be set to row based
binloging to make the execution of the complex SQL statements.
* Disable the statistic variables btr_search_n_hash_fail and n_hash_succ, n_hash_fail,
n_patt_succ, and n_searches of btr_search_t in builds without #ifdef UNIV_SEARCH_PERF_STAT.
* Make innodb.test faster. Group all consistent read test cases to a one test case and
wait their lock timeout after all have been send to the server. Decrease amount of rows
inserted in a certain test - this has no effect on the effectiveness of the test and
reduces the running time by ~10 sec. Remove temporary work-arounds from innodb.result
now that ALTER TABLE DROP FOREIGN KEY works once again.
* Make innodb_unsafe_binlog.test faster. Grout all consistent read test cases to a one
test case amd wait their lock timeout after all have been sent to the server. Remove
unnecessary option --loose_innodb_lock_wait_timeout.
* Print dictionary memory size in SHOW INNODB STATUS.
* Fix memory leaks in row_create_table_for_mysql() in rare corner cases.
* Remove code related to clustered tables. They were never implemented, and the
implementation would be challenging with ROW_FORMAT=COMPACT. Remove the table types
DICT_TABLE_CLUSTER_MEMBER and DICT_TABLE_CLUSTER and all related tests and functions.
dict_table_t: Remove mix_id, mix_len, mix_id_len, mix_id_buf, and cluster_name.
plan_t: Remove mixed_index.
dict_create_sys_tables_tuple(): Set MIX_ID=0, MIX_LEN=0, CLUSTER_NAME=NULL when
inserting into SYS_TABLES.
dict_tree_check_search_tuple(): Enclose in #ifdef UNIV_DEBUG.
* Move calling of thr_local_free() from trx_free_for_mysql() to
innobase_close_connection().
2006-04-23 10:48:31 +02:00
|
|
|
#endif /* UNIV_SEARCH_PERF_STAT */
|
2001-02-17 13:19:19 +01:00
|
|
|
|
|
|
|
|
/* After change in n_fields or n_bytes in info, this many rounds are waited
|
|
|
|
|
before starting the hash analysis again: this is to save CPU time when there
|
|
|
|
|
is no hope in building a hash index. */
|
|
|
|
|
|
|
|
|
|
#define BTR_SEARCH_HASH_ANALYSIS 17
|
|
|
|
|
|
|
|
|
|
/* Limit of consecutive searches for trying a search shortcut on the search
|
|
|
|
|
pattern */
|
|
|
|
|
|
|
|
|
|
#define BTR_SEARCH_ON_PATTERN_LIMIT 3
|
|
|
|
|
|
|
|
|
|
/* Limit of consecutive searches for trying a search shortcut using the hash
|
|
|
|
|
index */
|
|
|
|
|
|
|
|
|
|
#define BTR_SEARCH_ON_HASH_LIMIT 3
|
|
|
|
|
|
2001-10-10 21:47:08 +02:00
|
|
|
/* We do this many searches before trying to keep the search latch over calls
|
|
|
|
|
from MySQL. If we notice someone waiting for the latch, we again set this
|
|
|
|
|
much timeout. This is to reduce contention. */
|
|
|
|
|
|
|
|
|
|
#define BTR_SEA_TIMEOUT 10000
|
|
|
|
|
|
2001-02-17 13:19:19 +01:00
|
|
|
#ifndef UNIV_NONINL
|
|
|
|
|
#include "btr0sea.ic"
|
|
|
|
|
#endif
|
|
|
|
|
|
Applied innodb-5.1-ss269 snapshot.
Fixed BUGS:
#3300: "UPDATE statement with no index column in where condition locks
all rows"
Implement semi-consistent read to reduce lock conflicts at the cost
of breaking serializability.
ha_innobase::unlock_row(): reset the "did semi consistent read" flag
ha_innobase::was_semi_consistent_read(),
ha_innobase::try_semi_consistent_read(): new methods
row_prebuilt_t, row_create_prebuilt(): add field row_read_type for
keeping track of semi-consistent reads
row_vers_build_for_semi_consistent_read(),
row_sel_build_committed_vers_for_mysql(): new functions
row_search_for_mysql(): implement semi-consistent reads
#9802: "Foreign key checks disallow alter table".
Added test cases.
#12456: "Cursor shows incorrect data - DML does not affect,
probably caching"
This patch implements a high-granularity read view to be used with
cursors. In this high-granularity consistent read view modifications
done by the creating transaction after the cursor is created or
future transactions are not visible. But those modifications that
transaction did before the cursor was created are visible.
#12701: "Support >4GB buffer pool and log files on 64-bit Windows"
Do not call os_file_create_tmpfile() at runtime. Instead, create all
tempfiles at startup and guard access to them with mutexes.
#13778: "If FOREIGN_KEY_CHECKS=0, one can create inconsistent FOREIGN KEYs".
When FOREIGN_KEY_CHECKS=0 we still need to check that datatypes between
foreign key references are compatible.
#14189: "VARBINARY and BINARY variables: trailing space ignored with InnoDB"
innobase_init(): Assert that
DATA_MYSQL_BINARY_CHARSET_COLL == my_charset_bin.number.
dtype_get_pad_char(): Do not pad VARBINARY or BINARY columns.
row_ins_cascade_calc_update_vec(): Refuse ON UPDATE CASCADE when trying
to change the length of a VARBINARY column that refers to or is referenced
by a BINARY column. BINARY columns are no longer padded on comparison,
and thus they cannot be padded on storage either.
#14747: "Race condition can cause btr_search_drop_page_hash_index() to crash"
Note that buf_block_t::index should be protected by btr_search_latch
or an s-latch or x-latch on the index page.
btr_search_drop_page_hash_index(): Read block->index while holding
btr_search_latch and use the cached value in the loop. Remove some
redundant assertions.
#15108: "mysqld crashes when innodb_log_file_size is set > 4G"
#15308: "Problem of Order with Enum Column in Primary Key"
#15550: "mysqld crashes in printing a FOREIGN KEY error in InnoDB"
row_ins_foreign_report_add_err(): When printing the parent record,
use the index in the parent table rather than the index in the child table.
#15653: "Slow inserts to InnoDB if many thousands of .ibd files"
Keep track on unflushed modifications to file spaces. When there are tens
of thousands of file spaces, flushing all files in fil_flush_file_spaces()
would be very slow.
fil_flush_file_spaces(): Only flush unflushed file spaces.
fil_space_t, fil_system_t: Add a list of unflushed spaces.
#15991: "innodb-file-per-table + symlink database + rename = cr"
os_file_handle_error(): Map the error codes EXDEV, ENOTDIR, and EISDIR
to the new code OS_FILE_PATH_ERROR. Treat this code as OS_FILE_PATH_ERROR.
This fixes the crash on RENAME TABLE when the .ibd file is a symbolic link
to a different file system.
#16157: "InnoDB crashes when main location settings are empty"
This patch is from Heikki.
#16298: "InnoDB segfaults in INSERTs in upgrade of 4.0 -> 5.0 tables
with VARCHAR BINARY"
dict_load_columns(): Set the charset-collation code
DATA_MYSQL_BINARY_CHARSET_COLL for those binary string columns
that lack a charset-collation code, i.e., the tables were created
with an older version of MySQL/InnoDB than 4.1.2.
#16229: "MySQL/InnoDB uses full explicit table locks in trigger processing"
Take a InnoDB table lock only if user has explicitly requested a table
lock. Added some additional comments to store_lock() and external_lock().
#16387: "InnoDB crash when dropping a foreign key <table>_ibfk_0"
Do not mistake TABLENAME_ibfk_0 for auto-generated id.
dict_table_get_highest_foreign_id(): Ignore foreign constraint
identifiers starting with the pattern TABLENAME_ibfk_0.
#16582: "InnoDB: Error in an adaptive hash index pointer to page"
Account for a race condition when dropping the adaptive hash index
for a B-tree page.
btr_search_drop_page_hash_index(): Retry the operation if a hash index
with different parameters was built meanwhile. Add diagnostics for the
case that hash node pointers to the page remain.
btr_search_info_update_hash(), btr_search_info_update_slow():
Document the parameter "info" as in/out.
#16814: "SHOW INNODB STATUS format error in LATEST FOREIGN KEY ERROR
section"
Add a missing newline to the LAST FOREIGN KEY ERROR section in SHOW
INNODB STATUS output.
dict_foreign_error_report(): Always print a newline after invoking
dict_print_info_on_foreign_key_in_create_format().
#16827: "Better InnoDB error message if ibdata files omitted from my.cnf"
#17126: "CHECK TABLE on InnoDB causes a short hang during check of adaptive
hash"
CHECK TABLE blocking other queries, by releasing the btr_search_latch
periodically during the adaptive hash table validation.
#17405: "Valgrind: conditional jump or move depends on unititialised values"
buf_block_init(): Reset magic_n, buf_fix_count and io_fix to avoid
testing uninitialized variables.
2006-03-10 17:22:21 +01:00
|
|
|
#endif
|