mirror/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-02-01 11:31:51 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions
mariadb/storage/innobase/rem/rem0cmp.cc
Marko Mäkelä a4948dafcd MDEV-11369 Instant ADD COLUMN for InnoDB 
For InnoDB tables, adding, dropping and reordering columns has
required a rebuild of the table and all its indexes. Since MySQL 5.6
(and MariaDB 10.0) this has been supported online (LOCK=NONE), allowing
concurrent modification of the tables.

This work revises the InnoDB ROW_FORMAT=REDUNDANT, ROW_FORMAT=COMPACT
and ROW_FORMAT=DYNAMIC so that columns can be appended instantaneously,
with only minor changes performed to the table structure. The counter
innodb_instant_alter_column in INFORMATION_SCHEMA.GLOBAL_STATUS
is incremented whenever a table rebuild operation is converted into
an instant ADD COLUMN operation.

ROW_FORMAT=COMPRESSED tables will not support instant ADD COLUMN.

Some usability limitations will be addressed in subsequent work:

MDEV-13134 Introduce ALTER TABLE attributes ALGORITHM=NOCOPY
and ALGORITHM=INSTANT
MDEV-14016 Allow instant ADD COLUMN, ADD INDEX, LOCK=NONE

The format of the clustered index (PRIMARY KEY) is changed as follows:

(1) The FIL_PAGE_TYPE of the root page will be FIL_PAGE_TYPE_INSTANT,
and a new field PAGE_INSTANT will contain the original number of fields
in the clustered index ('core' fields).
If instant ADD COLUMN has not been used or the table becomes empty,
or the very first instant ADD COLUMN operation is rolled back,
the fields PAGE_INSTANT and FIL_PAGE_TYPE will be reset
to 0 and FIL_PAGE_INDEX.

(2) A special 'default row' record is inserted into the leftmost leaf,
between the page infimum and the first user record. This record is
distinguished by the REC_INFO_MIN_REC_FLAG, and it is otherwise in the
same format as records that contain values for the instantly added
columns. This 'default row' always has the same number of fields as
the clustered index according to the table definition. The values of
'core' fields are to be ignored. For other fields, the 'default row'
will contain the default values as they were during the ALTER TABLE
statement. (If the column default values are changed later, those
values will only be stored in the .frm file. The 'default row' will
contain the original evaluated values, which must be the same for
every row.) The 'default row' must be completely hidden from
higher-level access routines. Assertions have been added to ensure
that no 'default row' is ever present in the adaptive hash index
or in locked records. The 'default row' is never delete-marked.

(3) In clustered index leaf page records, the number of fields must
reside between the number of 'core' fields (dict_index_t::n_core_fields
introduced in this work) and dict_index_t::n_fields. If the number
of fields is less than dict_index_t::n_fields, the missing fields
are replaced with the column value of the 'default row'.
Note: The number of fields in the record may shrink if some of the
last instantly added columns are updated to the value that is
in the 'default row'. The function btr_cur_trim() implements this
'compression' on update and rollback; dtuple::trim() implements it
on insert.

(4) In ROW_FORMAT=COMPACT and ROW_FORMAT=DYNAMIC records, the new
status value REC_STATUS_COLUMNS_ADDED will indicate the presence of
a new record header that will encode n_fields-n_core_fields-1 in
1 or 2 bytes. (In ROW_FORMAT=REDUNDANT records, the record header
always explicitly encodes the number of fields.)

We introduce the undo log record type TRX_UNDO_INSERT_DEFAULT for
covering the insert of the 'default row' record when instant ADD COLUMN
is used for the first time. Subsequent instant ADD COLUMN can use
TRX_UNDO_UPD_EXIST_REC.

This is joint work with Vin Chen (陈福荣) from Tencent. The design
that was discussed in April 2017 would not have allowed import or
export of data files, because instead of the 'default row' it would
have introduced a data dictionary table. The test
rpl.rpl_alter_instant is exactly as contributed in pull request #408.
The test innodb.instant_alter is based on a contributed test.

The redo log record format changes for ROW_FORMAT=DYNAMIC and
ROW_FORMAT=COMPACT are as contributed. (With this change present,
crash recovery from MariaDB 10.3.1 will fail in spectacular ways!)
Also the semantics of higher-level redo log records that modify the
PAGE_INSTANT field is changed. The redo log format version identifier
was already changed to LOG_HEADER_FORMAT_CURRENT=103 in MariaDB 10.3.1.

Everything else has been rewritten by me. Thanks to Elena Stepanova,
the code has been tested extensively.

When rolling back an instant ADD COLUMN operation, we must empty the
PAGE_FREE list after deleting or shortening the 'default row' record,
by calling either btr_page_empty() or btr_page_reorganize(). We must
know the size of each entry in the PAGE_FREE list. If rollback left a
freed copy of the 'default row' in the PAGE_FREE list, we would be
unable to determine its size (if it is in ROW_FORMAT=COMPACT or
ROW_FORMAT=DYNAMIC) because it would contain more fields than the
rolled-back definition of the clustered index.

UNIV_SQL_DEFAULT: A new special constant that designates an instantly
added column that is not present in the clustered index record.

len_is_stored(): Check if a length is an actual length. There are
two magic length values: UNIV_SQL_DEFAULT, UNIV_SQL_NULL.

dict_col_t::def_val: The 'default row' value of the column.  If the
column is not added instantly, def_val.len will be UNIV_SQL_DEFAULT.

dict_col_t: Add the accessors is_virtual(), is_nullable(), is_instant(),
instant_value().

dict_col_t::remove_instant(): Remove the 'instant ADD' status of
a column.

dict_col_t::name(const dict_table_t& table): Replaces
dict_table_get_col_name().

dict_index_t::n_core_fields: The original number of fields.
For secondary indexes and if instant ADD COLUMN has not been used,
this will be equal to dict_index_t::n_fields.

dict_index_t::n_core_null_bytes: Number of bytes needed to
represent the null flags; usually equal to UT_BITS_IN_BYTES(n_nullable).

dict_index_t::NO_CORE_NULL_BYTES: Magic value signalling that
n_core_null_bytes was not initialized yet from the clustered index
root page.

dict_index_t: Add the accessors is_instant(), is_clust(),
get_n_nullable(), instant_field_value().

dict_index_t::instant_add_field(): Adjust clustered index metadata
for instant ADD COLUMN.

dict_index_t::remove_instant(): Remove the 'instant ADD' status
of a clustered index when the table becomes empty, or the very first
instant ADD COLUMN operation is rolled back.

dict_table_t: Add the accessors is_instant(), is_temporary(),
supports_instant().

dict_table_t::instant_add_column(): Adjust metadata for
instant ADD COLUMN.

dict_table_t::rollback_instant(): Adjust metadata on the rollback
of instant ADD COLUMN.

prepare_inplace_alter_table_dict(): First create the ctx->new_table,
and only then decide if the table really needs to be rebuilt.
We must split the creation of table or index metadata from the
creation of the dictionary table records and the creation of
the data. In this way, we can transform a table-rebuilding operation
into an instant ADD COLUMN operation. Dictionary objects will only
be added to cache when table rebuilding or index creation is needed.
The ctx->instant_table will never be added to cache.

dict_table_t::add_to_cache(): Modified and renamed from
dict_table_add_to_cache(). Do not modify the table metadata.
Let the callers invoke dict_table_add_system_columns() and if needed,
set can_be_evicted.

dict_create_sys_tables_tuple(), dict_create_table_step(): Omit the
system columns (which will now exist in the dict_table_t object
already at this point).

dict_create_table_step(): Expect the callers to invoke
dict_table_add_system_columns().

pars_create_table(): Before creating the table creation execution
graph, invoke dict_table_add_system_columns().

row_create_table_for_mysql(): Expect all callers to invoke
dict_table_add_system_columns().

create_index_dict(): Replaces row_merge_create_index_graph().

innodb_update_n_cols(): Renamed from innobase_update_n_virtual().
Call my_error() if an error occurs.

btr_cur_instant_init(), btr_cur_instant_init_low(),
btr_cur_instant_root_init():
Load additional metadata from the clustered index and set
dict_index_t::n_core_null_bytes. This is invoked
when table metadata is first loaded into the data dictionary.

dict_boot(): Initialize n_core_null_bytes for the four hard-coded
dictionary tables.

dict_create_index_step(): Initialize n_core_null_bytes. This is
executed as part of CREATE TABLE.

dict_index_build_internal_clust(): Initialize n_core_null_bytes to
NO_CORE_NULL_BYTES if table->supports_instant().

row_create_index_for_mysql(): Initialize n_core_null_bytes for
CREATE TEMPORARY TABLE.

commit_cache_norebuild(): Call the code to rename or enlarge columns
in the cache only if instant ADD COLUMN is not being used.
(Instant ADD COLUMN would copy all column metadata from
instant_table to old_table, including the names and lengths.)

PAGE_INSTANT: A new 13-bit field for storing dict_index_t::n_core_fields.
This is repurposing the 16-bit field PAGE_DIRECTION, of which only the
least significant 3 bits were used. The original byte containing
PAGE_DIRECTION will be accessible via the new constant PAGE_DIRECTION_B.

page_get_instant(), page_set_instant(): Accessors for the PAGE_INSTANT.

page_ptr_get_direction(), page_get_direction(),
page_ptr_set_direction(): Accessors for PAGE_DIRECTION.

page_direction_reset(): Reset PAGE_DIRECTION, PAGE_N_DIRECTION.

page_direction_increment(): Increment PAGE_N_DIRECTION
and set PAGE_DIRECTION.

rec_get_offsets(): Use the 'leaf' parameter for non-debug purposes,
and assume that heap_no is always set.
Initialize all dict_index_t::n_fields for ROW_FORMAT=REDUNDANT records,
even if the record contains fewer fields.

rec_offs_make_valid(): Add the parameter 'leaf'.

rec_copy_prefix_to_dtuple(): Assert that the tuple is only built
on the core fields. Instant ADD COLUMN only applies to the
clustered index, and we should never build a search key that has
more than the PRIMARY KEY and possibly DB_TRX_ID,DB_ROLL_PTR.
All these columns are always present.

dict_index_build_data_tuple(): Remove assertions that would be
duplicated in rec_copy_prefix_to_dtuple().

rec_init_offsets(): Support ROW_FORMAT=REDUNDANT records whose
number of fields is between n_core_fields and n_fields.

cmp_rec_rec_with_match(): Implement the comparison between two
MIN_REC_FLAG records.

trx_t::in_rollback: Make the field available in non-debug builds.

trx_start_for_ddl_low(): Remove dangerous error-tolerance.
A dictionary transaction must be flagged as such before it has generated
any undo log records. This is because trx_undo_assign_undo() will mark
the transaction as a dictionary transaction in the undo log header
right before the very first undo log record is being written.

btr_index_rec_validate(): Account for instant ADD COLUMN

row_undo_ins_remove_clust_rec(): On the rollback of an insert into
SYS_COLUMNS, revert instant ADD COLUMN in the cache by removing the
last column from the table and the clustered index.

row_search_on_row_ref(), row_undo_mod_parse_undo_rec(), row_undo_mod(),
trx_undo_update_rec_get_update(): Handle the 'default row'
as a special case.

dtuple_t::trim(index): Omit a redundant suffix of an index tuple right
before insert or update. After instant ADD COLUMN, if the last fields
of a clustered index tuple match the 'default row', there is no
need to store them. While trimming the entry, we must hold a page latch,
so that the table cannot be emptied and the 'default row' be deleted.

btr_cur_optimistic_update(), btr_cur_pessimistic_update(),
row_upd_clust_rec_by_insert(), row_ins_clust_index_entry_low():
Invoke dtuple_t::trim() if needed.

row_ins_clust_index_entry(): Restore dtuple_t::n_fields after calling
row_ins_clust_index_entry_low().

rec_get_converted_size(), rec_get_converted_size_comp(): Allow the number
of fields to be between n_core_fields and n_fields. Do not support
infimum,supremum. They are never supposed to be stored in dtuple_t,
because page creation nowadays uses a lower-level method for initializing
them.

rec_convert_dtuple_to_rec_comp(): Assign the status bits based on the
number of fields.

btr_cur_trim(): In an update, trim the index entry as needed. For the
'default row', handle rollback specially. For user records, omit
fields that match the 'default row'.

btr_cur_optimistic_delete_func(), btr_cur_pessimistic_delete():
Skip locking and adaptive hash index for the 'default row'.

row_log_table_apply_convert_mrec(): Replace 'default row' values if needed.
In the temporary file that is applied by row_log_table_apply(),
we must identify whether the records contain the extra header for
instantly added columns. For now, we will allocate an additional byte
for this for ROW_T_INSERT and ROW_T_UPDATE records when the source table
has been subject to instant ADD COLUMN. The ROW_T_DELETE records are
fine, as they will be converted and will only contain 'core' columns
(PRIMARY KEY and some system columns) that are converted from dtuple_t.

rec_get_converted_size_temp(), rec_init_offsets_temp(),
rec_convert_dtuple_to_temp(): Add the parameter 'status'.

REC_INFO_DEFAULT_ROW = REC_INFO_MIN_REC_FLAG | REC_STATUS_COLUMNS_ADDED:
An info_bits constant for distinguishing the 'default row' record.

rec_comp_status_t: An enum of the status bit values.

rec_leaf_format: An enum that replaces the bool parameter of
rec_init_offsets_comp_ordinary().
2017-10-06 09:50:10 +03:00

1263 lines
33 KiB
C++
Raw Blame History

/*****************************************************************************
Copyright (c) 1994, 2016, Oracle and/or its affiliates. All Rights Reserved.
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 the Free Software
Foundation; version 2 of the License.
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.,
51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
/*******************************************************************//**
@file rem/rem0cmp.cc
Comparison services for records
Created 7/1/1994 Heikki Tuuri
************************************************************************/
#include "ha_prototypes.h"
#include "rem0cmp.h"
#include "handler0alter.h"
#include "srv0srv.h"
#include <gstream.h>
#include <spatial.h>
#include <gis0geo.h>
#include <page0cur.h>
#include <algorithm>
/* ALPHABETICAL ORDER
==================
The records are put into alphabetical order in the following
way: let F be the first field where two records disagree.
If there is a character in some position n where the
records disagree, the order is determined by comparison of
the characters at position n, possibly after
collating transformation. If there is no such character,
but the corresponding fields have different lengths, then
if the data type of the fields is paddable,
shorter field is padded with a padding character. If the
data type is not paddable, longer field is considered greater.
Finally, the SQL null is bigger than any other value.
At the present, the comparison functions return 0 in the case,
where two records disagree only in the way that one
has more fields than the other. */
/** Compare two data fields.
@param[in] prtype precise type
@param[in] a data field
@param[in] a_length length of a, in bytes (not UNIV_SQL_NULL)
@param[in] b data field
@param[in] b_length length of b, in bytes (not UNIV_SQL_NULL)
@return positive, 0, negative, if a is greater, equal, less than b,
respectively */
UNIV_INLINE
int
innobase_mysql_cmp(
ulint prtype,
const byte* a,
unsigned int a_length,
const byte* b,
unsigned int b_length)
{
#ifdef UNIV_DEBUG
switch (prtype & DATA_MYSQL_TYPE_MASK) {
case MYSQL_TYPE_BIT:
case MYSQL_TYPE_STRING:
case MYSQL_TYPE_VAR_STRING:
case MYSQL_TYPE_TINY_BLOB:
case MYSQL_TYPE_MEDIUM_BLOB:
case MYSQL_TYPE_BLOB:
case MYSQL_TYPE_LONG_BLOB:
case MYSQL_TYPE_VARCHAR:
break;
default:
ut_error;
}
#endif /* UNIV_DEBUG */
uint cs_num = (uint) dtype_get_charset_coll(prtype);
if (CHARSET_INFO* cs = get_charset(cs_num, MYF(MY_WME))) {
return(cs->coll->strnncollsp(
cs, a, a_length, b, b_length));
}
ib::fatal() << "Unable to find charset-collation " << cs_num;
return(0);
}
/*************************************************************//**
Returns TRUE if two columns are equal for comparison purposes.
@return TRUE if the columns are considered equal in comparisons */
ibool
cmp_cols_are_equal(
/*===============*/
const dict_col_t* col1, /*!< in: column 1 */
const dict_col_t* col2, /*!< in: column 2 */
ibool check_charsets)
/*!< in: whether to check charsets */
{
if (dtype_is_non_binary_string_type(col1->mtype, col1->prtype)
&& dtype_is_non_binary_string_type(col2->mtype, col2->prtype)) {
/* Both are non-binary string types: they can be compared if
and only if the charset-collation is the same */
if (check_charsets) {
return(dtype_get_charset_coll(col1->prtype)
== dtype_get_charset_coll(col2->prtype));
} else {
return(TRUE);
}
}
if (dtype_is_binary_string_type(col1->mtype, col1->prtype)
&& dtype_is_binary_string_type(col2->mtype, col2->prtype)) {
/* Both are binary string types: they can be compared */
return(TRUE);
}
if (col1->mtype != col2->mtype) {
return(FALSE);
}
if (col1->mtype == DATA_INT
&& (col1->prtype & DATA_UNSIGNED)
!= (col2->prtype & DATA_UNSIGNED)) {
/* The storage format of an unsigned integer is different
from a signed integer: in a signed integer we OR
0x8000... to the value of positive integers. */
return(FALSE);
}
return(col1->mtype != DATA_INT || col1->len == col2->len);
}
/** Compare two DATA_DECIMAL (MYSQL_TYPE_DECIMAL) fields.
TODO: Remove this function. Everything should use MYSQL_TYPE_NEWDECIMAL.
@param[in] a data field
@param[in] a_length length of a, in bytes (not UNIV_SQL_NULL)
@param[in] b data field
@param[in] b_length length of b, in bytes (not UNIV_SQL_NULL)
@return positive, 0, negative, if a is greater, equal, less than b,
respectively */
static ATTRIBUTE_COLD
int
cmp_decimal(
const byte* a,
unsigned int a_length,
const byte* b,
unsigned int b_length)
{
int swap_flag;
/* Remove preceding spaces */
for (; a_length && *a == ' '; a++, a_length--) { }
for (; b_length && *b == ' '; b++, b_length--) { }
if (*a == '-') {
swap_flag = -1;
if (*b != '-') {
return(swap_flag);
}
a++; b++;
a_length--;
b_length--;
} else {
swap_flag = 1;
if (*b == '-') {
return(swap_flag);
}
}
while (a_length > 0 && (*a == '+' || *a == '0')) {
a++; a_length--;
}
while (b_length > 0 && (*b == '+' || *b == '0')) {
b++; b_length--;
}
if (a_length != b_length) {
if (a_length < b_length) {
return(-swap_flag);
}
return(swap_flag);
}
while (a_length > 0 && *a == *b) {
a++; b++; a_length--;
}
if (a_length == 0) {
return(0);
}
if (*a <= *b) {
swap_flag = -swap_flag;
}
return(swap_flag);
}
/*************************************************************//**
Innobase uses this function to compare two geometry data fields
@return 1, 0, -1, if a is greater, equal, less than b, respectively */
static
int
cmp_geometry_field(
/*===============*/
ulint mtype, /*!< in: main type */
ulint prtype, /*!< in: precise type */
const byte* a, /*!< in: data field */
unsigned int a_length, /*!< in: data field length,
not UNIV_SQL_NULL */
const byte* b, /*!< in: data field */
unsigned int b_length) /*!< in: data field length,
not UNIV_SQL_NULL */
{
double x1, x2;
double y1, y2;
ut_ad(prtype & DATA_GIS_MBR);
if (a_length < sizeof(double) || b_length < sizeof(double)) {
return(0);
}
/* Try to compare mbr left lower corner (xmin, ymin) */
x1 = mach_double_read(a);
x2 = mach_double_read(b);
y1 = mach_double_read(a + sizeof(double) * SPDIMS);
y2 = mach_double_read(b + sizeof(double) * SPDIMS);
if (x1 > x2) {
return(1);
} else if (x2 > x1) {
return(-1);
}
if (y1 > y2) {
return(1);
} else if (y2 > y1) {
return(-1);
}
/* left lower corner (xmin, ymin) overlaps, now right upper corner */
x1 = mach_double_read(a + sizeof(double));
x2 = mach_double_read(b + sizeof(double));
y1 = mach_double_read(a + sizeof(double) * SPDIMS + sizeof(double));
y2 = mach_double_read(b + sizeof(double) * SPDIMS + sizeof(double));
if (x1 > x2) {
return(1);
} else if (x2 > x1) {
return(-1);
}
if (y1 > y2) {
return(1);
} else if (y2 > y1) {
return(-1);
}
return(0);
}
/*************************************************************//**
Innobase uses this function to compare two gis data fields
@return 1, 0, -1, if mode == PAGE_CUR_MBR_EQUAL. And return
1, 0 for rest compare modes, depends on a and b qualifies the
relationship (CONTAINT, WITHIN etc.) */
static
int
cmp_gis_field(
/*============*/
page_cur_mode_t mode, /*!< in: compare mode */
const byte* a, /*!< in: data field */
unsigned int a_length, /*!< in: data field length,
not UNIV_SQL_NULL */
const byte* b, /*!< in: data field */
unsigned int b_length) /*!< in: data field length,
not UNIV_SQL_NULL */
{
if (mode == PAGE_CUR_MBR_EQUAL) {
/* TODO: Since the DATA_GEOMETRY is not used in compare
function, we could pass it instead of a specific type now */
return(cmp_geometry_field(DATA_GEOMETRY, DATA_GIS_MBR,
a, a_length, b, b_length));
} else {
return(rtree_key_cmp(mode, a, a_length, b, b_length));
}
}
/** Compare two data fields.
@param[in] mtype main type
@param[in] prtype precise type
@param[in] a data field
@param[in] a_length length of a, in bytes (not UNIV_SQL_NULL)
@param[in] b data field
@param[in] b_length length of b, in bytes (not UNIV_SQL_NULL)
@return positive, 0, negative, if a is greater, equal, less than b,
respectively */
static
int
cmp_whole_field(
ulint mtype,
ulint prtype,
const byte* a,
unsigned int a_length,
const byte* b,
unsigned int b_length)
{
float f_1;
float f_2;
double d_1;
double d_2;
switch (mtype) {
case DATA_DECIMAL:
return(cmp_decimal(a, a_length, b, b_length));
case DATA_DOUBLE:
d_1 = mach_double_read(a);
d_2 = mach_double_read(b);
if (d_1 > d_2) {
return(1);
} else if (d_2 > d_1) {
return(-1);
}
return(0);
case DATA_FLOAT:
f_1 = mach_float_read(a);
f_2 = mach_float_read(b);
if (f_1 > f_2) {
return(1);
} else if (f_2 > f_1) {
return(-1);
}
return(0);
case DATA_VARCHAR:
case DATA_CHAR:
return(my_charset_latin1.coll->strnncollsp(
&my_charset_latin1,
a, a_length, b, b_length));
case DATA_BLOB:
if (prtype & DATA_BINARY_TYPE) {
ib::error() << "Comparing a binary BLOB"
" using a character set collation!";
ut_ad(0);
}
/* fall through */
case DATA_VARMYSQL:
case DATA_MYSQL:
return(innobase_mysql_cmp(prtype,
a, a_length, b, b_length));
case DATA_GEOMETRY:
return(cmp_geometry_field(mtype, prtype, a, a_length, b,
b_length));
default:
ib::fatal() << "Unknown data type number " << mtype;
}
return(0);
}
/** Compare two data fields.
@param[in] mtype main type
@param[in] prtype precise type
@param[in] data1 data field
@param[in] len1 length of data1 in bytes, or UNIV_SQL_NULL
@param[in] data2 data field
@param[in] len2 length of data2 in bytes, or UNIV_SQL_NULL
@return the comparison result of data1 and data2
@retval 0 if data1 is equal to data2
@retval negative if data1 is less than data2
@retval positive if data1 is greater than data2 */
inline
int
cmp_data(
ulint mtype,
ulint prtype,
const byte* data1,
ulint len1,
const byte* data2,
ulint len2)
{
ut_ad(len1 != UNIV_SQL_DEFAULT);
ut_ad(len2 != UNIV_SQL_DEFAULT);
if (len1 == UNIV_SQL_NULL || len2 == UNIV_SQL_NULL) {
if (len1 == len2) {
return(0);
}
/* We define the SQL null to be the smallest possible
value of a field. */
return(len1 == UNIV_SQL_NULL ? -1 : 1);
}
ulint pad;
switch (mtype) {
case DATA_FIXBINARY:
case DATA_BINARY:
if (dtype_get_charset_coll(prtype)
!= DATA_MYSQL_BINARY_CHARSET_COLL) {
pad = 0x20;
break;
}
/* fall through */
case DATA_INT:
case DATA_SYS_CHILD:
case DATA_SYS:
pad = ULINT_UNDEFINED;
break;
case DATA_GEOMETRY:
ut_ad(prtype & DATA_BINARY_TYPE);
pad = ULINT_UNDEFINED;
if (prtype & DATA_GIS_MBR) {
return(cmp_whole_field(mtype, prtype,
data1, (unsigned) len1,
data2, (unsigned) len2));
}
break;
case DATA_BLOB:
if (prtype & DATA_BINARY_TYPE) {
pad = ULINT_UNDEFINED;
break;
}
/* fall through */
default:
return(cmp_whole_field(mtype, prtype,
data1, (unsigned) len1,
data2, (unsigned) len2));
}
ulint len;
int cmp;
if (len1 < len2) {
len = len1;
len2 -= len;
len1 = 0;
} else {
len = len2;
len1 -= len;
len2 = 0;
}
if (len) {
#if defined __i386__ || defined __x86_64__ || defined _M_IX86 || defined _M_X64
/* Compare the first bytes with a loop to avoid the call
overhead of memcmp(). On x86 and x86-64, the GCC built-in
(repz cmpsb) seems to be very slow, so we will be calling the
libc version. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43052
tracks the slowness of the GCC built-in memcmp().
We compare up to the first 4..7 bytes with the loop.
The (len & 3) is used for "normalizing" or
"quantizing" the len parameter for the memcmp() call,
in case the whole prefix is equal. On x86 and x86-64,
the GNU libc memcmp() of equal strings is faster with
len=4 than with len=3.
On other architectures than the IA32 or AMD64, there could
be a built-in memcmp() that is faster than the loop.
We only use the loop where we know that it can improve
the performance. */
for (ulint i = 4 + (len & 3); i > 0; i--) {
cmp = int(*data1++) - int(*data2++);
if (cmp) {
return(cmp);
}
if (!--len) {
break;
}
}
if (len) {
#endif /* IA32 or AMD64 */
cmp = memcmp(data1, data2, len);
if (cmp) {
return(cmp);
}
data1 += len;
data2 += len;
#if defined __i386__ || defined __x86_64__ || defined _M_IX86 || defined _M_X64
}
#endif /* IA32 or AMD64 */
}
cmp = (int) (len1 - len2);
if (!cmp || pad == ULINT_UNDEFINED) {
return(cmp);
}
len = 0;
if (len1) {
do {
cmp = static_cast<int>(
mach_read_from_1(&data1[len++]) - pad);
} while (cmp == 0 && len < len1);
} else {
ut_ad(len2 > 0);
do {
cmp = static_cast<int>(
pad - mach_read_from_1(&data2[len++]));
} while (cmp == 0 && len < len2);
}
return(cmp);
}
/** Compare a GIS data tuple to a physical record.
@param[in] dtuple data tuple
@param[in] rec B-tree record
@param[in] offsets rec_get_offsets(rec)
@param[in] mode compare mode
@retval negative if dtuple is less than rec */
int
cmp_dtuple_rec_with_gis(
/*====================*/
const dtuple_t* dtuple, /*!< in: data tuple */
const rec_t* rec, /*!< in: physical record which differs from
dtuple in some of the common fields, or which
has an equal number or more fields than
dtuple */
const ulint* offsets,/*!< in: array returned by rec_get_offsets() */
page_cur_mode_t mode) /*!< in: compare mode */
{
const dfield_t* dtuple_field; /* current field in logical record */
ulint dtuple_f_len; /* the length of the current field
in the logical record */
ulint rec_f_len; /* length of current field in rec */
const byte* rec_b_ptr; /* pointer to the current byte in
rec field */
int ret = 0; /* return value */
dtuple_field = dtuple_get_nth_field(dtuple, 0);
dtuple_f_len = dfield_get_len(dtuple_field);
rec_b_ptr = rec_get_nth_field(rec, offsets, 0, &rec_f_len);
ret = cmp_gis_field(
mode, static_cast<const byte*>(dfield_get_data(dtuple_field)),
(unsigned) dtuple_f_len, rec_b_ptr, (unsigned) rec_f_len);
return(ret);
}
/** Compare a GIS data tuple to a physical record in rtree non-leaf node.
We need to check the page number field, since we don't store pk field in
rtree non-leaf node.
@param[in] dtuple data tuple
@param[in] rec R-tree record
@param[in] offsets rec_get_offsets(rec)
@retval negative if dtuple is less than rec */
int
cmp_dtuple_rec_with_gis_internal(
const dtuple_t* dtuple,
const rec_t* rec,
const ulint* offsets)
{
const dfield_t* dtuple_field; /* current field in logical record */
ulint dtuple_f_len; /* the length of the current field
in the logical record */
ulint rec_f_len; /* length of current field in rec */
const byte* rec_b_ptr; /* pointer to the current byte in
rec field */
int ret = 0; /* return value */
dtuple_field = dtuple_get_nth_field(dtuple, 0);
dtuple_f_len = dfield_get_len(dtuple_field);
rec_b_ptr = rec_get_nth_field(rec, offsets, 0, &rec_f_len);
ret = cmp_gis_field(
PAGE_CUR_WITHIN,
static_cast<const byte*>(dfield_get_data(dtuple_field)),
(unsigned) dtuple_f_len, rec_b_ptr, (unsigned) rec_f_len);
if (ret != 0) {
return(ret);
}
dtuple_field = dtuple_get_nth_field(dtuple, 1);
dtuple_f_len = dfield_get_len(dtuple_field);
rec_b_ptr = rec_get_nth_field(rec, offsets, 1, &rec_f_len);
return(cmp_data(dtuple_field->type.mtype,
dtuple_field->type.prtype,
static_cast<const byte*>(dtuple_field->data),
dtuple_f_len,
rec_b_ptr,
rec_f_len));
}
/** Compare two data fields.
@param[in] mtype main type
@param[in] prtype precise type
@param[in] data1 data field
@param[in] len1 length of data1 in bytes, or UNIV_SQL_NULL
@param[in] data2 data field
@param[in] len2 length of data2 in bytes, or UNIV_SQL_NULL
@return the comparison result of data1 and data2
@retval 0 if data1 is equal to data2
@retval negative if data1 is less than data2
@retval positive if data1 is greater than data2 */
int
cmp_data_data(
ulint mtype,
ulint prtype,
const byte* data1,
ulint len1,
const byte* data2,
ulint len2)
{
return(cmp_data(mtype, prtype, data1, len1, data2, len2));
}
/** Compare a data tuple to a physical record.
@param[in] dtuple data tuple
@param[in] rec B-tree record
@param[in] offsets rec_get_offsets(rec)
@param[in] n_cmp number of fields to compare
@param[in,out] matched_fields number of completely matched fields
@return the comparison result of dtuple and rec
@retval 0 if dtuple is equal to rec
@retval negative if dtuple is less than rec
@retval positive if dtuple is greater than rec */
int
cmp_dtuple_rec_with_match_low(
const dtuple_t* dtuple,
const rec_t* rec,
const ulint* offsets,
ulint n_cmp,
ulint* matched_fields)
{
ulint cur_field; /* current field number */
int ret; /* return value */
ut_ad(dtuple_check_typed(dtuple));
ut_ad(rec_offs_validate(rec, NULL, offsets));
cur_field = *matched_fields;
ut_ad(n_cmp > 0);
ut_ad(n_cmp <= dtuple_get_n_fields(dtuple));
ut_ad(cur_field <= n_cmp);
ut_ad(cur_field <= rec_offs_n_fields(offsets));
if (cur_field == 0) {
ulint rec_info = rec_get_info_bits(rec,
rec_offs_comp(offsets));
ulint tup_info = dtuple_get_info_bits(dtuple);
if (UNIV_UNLIKELY(rec_info & REC_INFO_MIN_REC_FLAG)) {
ret = !(tup_info & REC_INFO_MIN_REC_FLAG);
goto order_resolved;
} else if (UNIV_UNLIKELY(tup_info & REC_INFO_MIN_REC_FLAG)) {
ret = -1;
goto order_resolved;
}
}
/* Match fields in a loop */
for (; cur_field < n_cmp; cur_field++) {
const byte* rec_b_ptr;
const dfield_t* dtuple_field
= dtuple_get_nth_field(dtuple, cur_field);
const byte* dtuple_b_ptr
= static_cast<const byte*>(
dfield_get_data(dtuple_field));
const dtype_t* type
= dfield_get_type(dtuple_field);
ulint dtuple_f_len
= dfield_get_len(dtuple_field);
ulint rec_f_len;
/* We should never compare against an externally
stored field. Only clustered index records can
contain externally stored fields, and the first fields
(primary key fields) should already differ. */
ut_ad(!rec_offs_nth_extern(offsets, cur_field));
/* We should never compare against instantly added columns.
Columns can only be instantly added to clustered index
leaf page records, and the first fields (primary key fields)
should already differ. */
ut_ad(!rec_offs_nth_default(offsets, cur_field));
rec_b_ptr = rec_get_nth_field(rec, offsets, cur_field,
&rec_f_len);
ut_ad(!dfield_is_ext(dtuple_field));
ret = cmp_data(type->mtype, type->prtype,
dtuple_b_ptr, dtuple_f_len,
rec_b_ptr, rec_f_len);
if (ret) {
goto order_resolved;
}
}
ret = 0; /* If we ran out of fields, dtuple was equal to rec
up to the common fields */
order_resolved:
*matched_fields = cur_field;
return(ret);
}
/** Get the pad character code point for a type.
@param[in] type
@return pad character code point
@retval ULINT_UNDEFINED if no padding is specified */
UNIV_INLINE
ulint
cmp_get_pad_char(
const dtype_t* type)
{
switch (type->mtype) {
case DATA_FIXBINARY:
case DATA_BINARY:
if (dtype_get_charset_coll(type->prtype)
== DATA_MYSQL_BINARY_CHARSET_COLL) {
/* Starting from 5.0.18, do not pad
VARBINARY or BINARY columns. */
return(ULINT_UNDEFINED);
}
/* Fall through */
case DATA_CHAR:
case DATA_VARCHAR:
case DATA_MYSQL:
case DATA_VARMYSQL:
/* Space is the padding character for all char and binary
strings, and starting from 5.0.3, also for TEXT strings. */
return(0x20);
case DATA_GEOMETRY:
/* DATA_GEOMETRY is binary data, not ASCII-based. */
return(ULINT_UNDEFINED);
case DATA_BLOB:
if (!(type->prtype & DATA_BINARY_TYPE)) {
return(0x20);
}
/* Fall through */
default:
/* No padding specified */
return(ULINT_UNDEFINED);
}
}
/** Compare a data tuple to a physical record.
@param[in] dtuple data tuple
@param[in] rec B-tree or R-tree index record
@param[in] index index tree
@param[in] offsets rec_get_offsets(rec)
@param[in,out] matched_fields number of completely matched fields
@param[in,out] matched_bytes number of matched bytes in the first
field that is not matched
@return the comparison result of dtuple and rec
@retval 0 if dtuple is equal to rec
@retval negative if dtuple is less than rec
@retval positive if dtuple is greater than rec */
int
cmp_dtuple_rec_with_match_bytes(
const dtuple_t* dtuple,
const rec_t* rec,
const dict_index_t* index,
const ulint* offsets,
ulint* matched_fields,
ulint* matched_bytes)
{
ulint n_cmp = dtuple_get_n_fields_cmp(dtuple);
ulint cur_field; /* current field number */
ulint cur_bytes;
int ret; /* return value */
ut_ad(dtuple_check_typed(dtuple));
ut_ad(rec_offs_validate(rec, index, offsets));
ut_ad(!(REC_INFO_MIN_REC_FLAG
& dtuple_get_info_bits(dtuple)));
ut_ad(!(REC_INFO_MIN_REC_FLAG
& rec_get_info_bits(rec, rec_offs_comp(offsets))));
cur_field = *matched_fields;
cur_bytes = *matched_bytes;
ut_ad(n_cmp <= dtuple_get_n_fields(dtuple));
ut_ad(cur_field <= n_cmp);
ut_ad(cur_field + (cur_bytes > 0) <= rec_offs_n_fields(offsets));
/* Match fields in a loop; stop if we run out of fields in dtuple
or find an externally stored field */
while (cur_field < n_cmp) {
const dfield_t* dfield = dtuple_get_nth_field(
dtuple, cur_field);
const dtype_t* type = dfield_get_type(dfield);
ulint dtuple_f_len = dfield_get_len(dfield);
const byte* dtuple_b_ptr;
const byte* rec_b_ptr;
ulint rec_f_len;
dtuple_b_ptr = static_cast<const byte*>(
dfield_get_data(dfield));
ut_ad(!rec_offs_nth_default(offsets, cur_field));
rec_b_ptr = rec_get_nth_field(rec, offsets,
cur_field, &rec_f_len);
ut_ad(!rec_offs_nth_extern(offsets, cur_field));
/* If we have matched yet 0 bytes, it may be that one or
both the fields are SQL null, or the record or dtuple may be
the predefined minimum record. */
if (cur_bytes == 0) {
if (dtuple_f_len == UNIV_SQL_NULL) {
if (rec_f_len == UNIV_SQL_NULL) {
goto next_field;
}
ret = -1;
goto order_resolved;
} else if (rec_f_len == UNIV_SQL_NULL) {
/* We define the SQL null to be the
smallest possible value of a field
in the alphabetical order */
ret = 1;
goto order_resolved;
}
}
switch (type->mtype) {
case DATA_FIXBINARY:
case DATA_BINARY:
case DATA_INT:
case DATA_SYS_CHILD:
case DATA_SYS:
break;
case DATA_BLOB:
if (type->prtype & DATA_BINARY_TYPE) {
break;
}
/* fall through */
default:
ret = cmp_data(type->mtype, type->prtype,
dtuple_b_ptr, dtuple_f_len,
rec_b_ptr, rec_f_len);
if (!ret) {
goto next_field;
}
cur_bytes = 0;
goto order_resolved;
}
/* Set the pointers at the current byte */
rec_b_ptr += cur_bytes;
dtuple_b_ptr += cur_bytes;
/* Compare then the fields */
for (const ulint pad = cmp_get_pad_char(type);;
cur_bytes++) {
ulint rec_byte = pad;
ulint dtuple_byte = pad;
if (rec_f_len <= cur_bytes) {
if (dtuple_f_len <= cur_bytes) {
goto next_field;
}
if (rec_byte == ULINT_UNDEFINED) {
ret = 1;
goto order_resolved;
}
} else {
rec_byte = *rec_b_ptr++;
}
if (dtuple_f_len <= cur_bytes) {
if (dtuple_byte == ULINT_UNDEFINED) {
ret = -1;
goto order_resolved;
}
} else {
dtuple_byte = *dtuple_b_ptr++;
}
if (dtuple_byte < rec_byte) {
ret = -1;
goto order_resolved;
} else if (dtuple_byte > rec_byte) {
ret = 1;
goto order_resolved;
}
}
next_field:
cur_field++;
cur_bytes = 0;
}
ut_ad(cur_bytes == 0);
ret = 0; /* If we ran out of fields, dtuple was equal to rec
up to the common fields */
order_resolved:
*matched_fields = cur_field;
*matched_bytes = cur_bytes;
return(ret);
}
/** Compare a data tuple to a physical record.
@see cmp_dtuple_rec_with_match
@param[in] dtuple data tuple
@param[in] rec B-tree record
@param[in] offsets rec_get_offsets(rec); may be NULL
for ROW_FORMAT=REDUNDANT
@return the comparison result of dtuple and rec
@retval 0 if dtuple is equal to rec
@retval negative if dtuple is less than rec
@retval positive if dtuple is greater than rec */
int
cmp_dtuple_rec(
const dtuple_t* dtuple,
const rec_t* rec,
const ulint* offsets)
{
ulint matched_fields = 0;
ut_ad(rec_offs_validate(rec, NULL, offsets));
return(cmp_dtuple_rec_with_match(dtuple, rec, offsets,
&matched_fields));
}
/**************************************************************//**
Checks if a dtuple is a prefix of a record. The last field in dtuple
is allowed to be a prefix of the corresponding field in the record.
@return TRUE if prefix */
ibool
cmp_dtuple_is_prefix_of_rec(
/*========================*/
const dtuple_t* dtuple, /*!< in: data tuple */
const rec_t* rec, /*!< in: physical record */
const ulint* offsets)/*!< in: array returned by rec_get_offsets() */
{
ulint n_fields;
ulint matched_fields = 0;
ut_ad(rec_offs_validate(rec, NULL, offsets));
n_fields = dtuple_get_n_fields(dtuple);
if (n_fields > rec_offs_n_fields(offsets)) {
ut_ad(0);
return(FALSE);
}
cmp_dtuple_rec_with_match(dtuple, rec, offsets, &matched_fields);
return(matched_fields == n_fields);
}
/*************************************************************//**
Compare two physical record fields.
@retval positive if rec1 field is greater than rec2
@retval negative if rec1 field is less than rec2
@retval 0 if rec1 field equals to rec2 */
static MY_ATTRIBUTE((nonnull, warn_unused_result))
int
cmp_rec_rec_simple_field(
/*=====================*/
const rec_t* rec1, /*!< in: physical record */
const rec_t* rec2, /*!< in: physical record */
const ulint* offsets1,/*!< in: rec_get_offsets(rec1, ...) */
const ulint* offsets2,/*!< in: rec_get_offsets(rec2, ...) */
const dict_index_t* index, /*!< in: data dictionary index */
ulint n) /*!< in: field to compare */
{
const byte* rec1_b_ptr;
const byte* rec2_b_ptr;
ulint rec1_f_len;
ulint rec2_f_len;
const dict_col_t* col = dict_index_get_nth_col(index, n);
ut_ad(!rec_offs_nth_extern(offsets1, n));
ut_ad(!rec_offs_nth_extern(offsets2, n));
rec1_b_ptr = rec_get_nth_field(rec1, offsets1, n, &rec1_f_len);
rec2_b_ptr = rec_get_nth_field(rec2, offsets2, n, &rec2_f_len);
return(cmp_data(col->mtype, col->prtype,
rec1_b_ptr, rec1_f_len, rec2_b_ptr, rec2_f_len));
}
/** Compare two physical records that contain the same number of columns,
none of which are stored externally.
@retval positive if rec1 (including non-ordering columns) is greater than rec2
@retval negative if rec1 (including non-ordering columns) is less than rec2
@retval 0 if rec1 is a duplicate of rec2 */
int
cmp_rec_rec_simple(
/*===============*/
const rec_t* rec1, /*!< in: physical record */
const rec_t* rec2, /*!< in: physical record */
const ulint* offsets1,/*!< in: rec_get_offsets(rec1, ...) */
const ulint* offsets2,/*!< in: rec_get_offsets(rec2, ...) */
const dict_index_t* index, /*!< in: data dictionary index */
struct TABLE* table) /*!< in: MySQL table, for reporting
duplicate key value if applicable,
or NULL */
{
ulint n;
ulint n_uniq = dict_index_get_n_unique(index);
bool null_eq = false;
ut_ad(rec_offs_n_fields(offsets1) >= n_uniq);
ut_ad(rec_offs_n_fields(offsets2) == rec_offs_n_fields(offsets2));
ut_ad(rec_offs_comp(offsets1) == rec_offs_comp(offsets2));
for (n = 0; n < n_uniq; n++) {
int cmp = cmp_rec_rec_simple_field(
rec1, rec2, offsets1, offsets2, index, n);
if (cmp) {
return(cmp);
}
/* If the fields are internally equal, they must both
be NULL or non-NULL. */
ut_ad(rec_offs_nth_sql_null(offsets1, n)
== rec_offs_nth_sql_null(offsets2, n));
if (rec_offs_nth_sql_null(offsets1, n)) {
ut_ad(!(dict_index_get_nth_col(index, n)->prtype
& DATA_NOT_NULL));
null_eq = true;
}
}
/* If we ran out of fields, the ordering columns of rec1 were
equal to rec2. Issue a duplicate key error if needed. */
if (!null_eq && table && dict_index_is_unique(index)) {
/* Report erroneous row using new version of table. */
innobase_rec_to_mysql(table, rec1, index, offsets1);
return(0);
}
/* Else, keep comparing so that we have the full internal
order. */
for (; n < dict_index_get_n_fields(index); n++) {
int cmp = cmp_rec_rec_simple_field(
rec1, rec2, offsets1, offsets2, index, n);
if (cmp) {
return(cmp);
}
/* If the fields are internally equal, they must both
be NULL or non-NULL. */
ut_ad(rec_offs_nth_sql_null(offsets1, n)
== rec_offs_nth_sql_null(offsets2, n));
}
/* This should never be reached. Internally, an index must
never contain duplicate entries. */
ut_ad(0);
return(0);
}
/** Compare two B-tree records.
@param[in] rec1 B-tree record
@param[in] rec2 B-tree record
@param[in] offsets1 rec_get_offsets(rec1, index)
@param[in] offsets2 rec_get_offsets(rec2, index)
@param[in] index B-tree index
@param[in] nulls_unequal true if this is for index cardinality
statistics estimation, and innodb_stats_method=nulls_unequal
or innodb_stats_method=nulls_ignored
@param[out] matched_fields number of completely matched fields
within the first field not completely matched
@return the comparison result
@retval 0 if rec1 is equal to rec2
@retval negative if rec1 is less than rec2
@retval positive if rec2 is greater than rec2 */
int
cmp_rec_rec_with_match(
const rec_t* rec1,
const rec_t* rec2,
const ulint* offsets1,
const ulint* offsets2,
const dict_index_t* index,
bool nulls_unequal,
ulint* matched_fields)
{
ulint rec1_n_fields; /* the number of fields in rec */
ulint rec1_f_len; /* length of current field in rec */
const byte* rec1_b_ptr; /* pointer to the current byte
in rec field */
ulint rec2_n_fields; /* the number of fields in rec */
ulint rec2_f_len; /* length of current field in rec */
const byte* rec2_b_ptr; /* pointer to the current byte
in rec field */
ulint cur_field = 0; /* current field number */
int ret = 0; /* return value */
ulint comp;
ut_ad(rec1 != NULL);
ut_ad(rec2 != NULL);
ut_ad(index != NULL);
ut_ad(rec_offs_validate(rec1, index, offsets1));
ut_ad(rec_offs_validate(rec2, index, offsets2));
ut_ad(rec_offs_comp(offsets1) == rec_offs_comp(offsets2));
comp = rec_offs_comp(offsets1);
rec1_n_fields = rec_offs_n_fields(offsets1);
rec2_n_fields = rec_offs_n_fields(offsets2);
/* Test if rec is the predefined minimum record */
if (UNIV_UNLIKELY(rec_get_info_bits(rec1, comp)
& REC_INFO_MIN_REC_FLAG)) {
ret = UNIV_UNLIKELY(rec_get_info_bits(rec2, comp)
& REC_INFO_MIN_REC_FLAG)
? 0 : -1;
goto order_resolved;
} else if (UNIV_UNLIKELY
(rec_get_info_bits(rec2, comp)
& REC_INFO_MIN_REC_FLAG)) {
ret = 1;
goto order_resolved;
}
/* Match fields in a loop */
for (; cur_field < rec1_n_fields && cur_field < rec2_n_fields;
cur_field++) {
ulint mtype;
ulint prtype;
/* If this is node-ptr records then avoid comparing node-ptr
field. Only key field needs to be compared. */
if (cur_field == dict_index_get_n_unique_in_tree(index)) {
break;
}
if (dict_index_is_ibuf(index)) {
/* This is for the insert buffer B-tree. */
mtype = DATA_BINARY;
prtype = 0;
} else {
const dict_col_t* col;
col = dict_index_get_nth_col(index, cur_field);
mtype = col->mtype;
prtype = col->prtype;
/* If the index is spatial index, we mark the
prtype of the first field as MBR field. */
if (cur_field == 0 && dict_index_is_spatial(index)) {
ut_ad(DATA_GEOMETRY_MTYPE(mtype));
prtype |= DATA_GIS_MBR;
}
}
/* We should never encounter an externally stored field.
Externally stored fields only exist in clustered index
leaf page records. These fields should already differ
in the primary key columns already, before DB_TRX_ID,
DB_ROLL_PTR, and any externally stored columns. */
ut_ad(!rec_offs_nth_extern(offsets1, cur_field));
ut_ad(!rec_offs_nth_extern(offsets2, cur_field));
ut_ad(!rec_offs_nth_default(offsets1, cur_field));
ut_ad(!rec_offs_nth_default(offsets2, cur_field));
rec1_b_ptr = rec_get_nth_field(rec1, offsets1,
cur_field, &rec1_f_len);
rec2_b_ptr = rec_get_nth_field(rec2, offsets2,
cur_field, &rec2_f_len);
if (nulls_unequal
&& rec1_f_len == UNIV_SQL_NULL
&& rec2_f_len == UNIV_SQL_NULL) {
ret = -1;
goto order_resolved;
}
ret = cmp_data(mtype, prtype,
rec1_b_ptr, rec1_f_len,
rec2_b_ptr, rec2_f_len);
if (ret) {
goto order_resolved;
}
}
/* If we ran out of fields, rec1 was equal to rec2 up
to the common fields */
ut_ad(ret == 0);
order_resolved:
*matched_fields = cur_field;
return(ret);
}
#ifdef UNIV_COMPILE_TEST_FUNCS
#ifdef HAVE_UT_CHRONO_T
void
test_cmp_data_data(ulint len)
{
int i;
static byte zeros[64];
if (len > sizeof zeros) {
len = sizeof zeros;
}
ut_chrono_t ch(__func__);
for (i = 1000000; i > 0; i--) {
i += cmp_data(DATA_INT, 0, zeros, len, zeros, len);
}
}
#endif /* HAVE_UT_CHRONO_T */
#endif /* UNIV_COMPILE_TEST_FUNCS */
Reference in a new issue View git blame Copy permalink