mariadb/sql/rpl_record.cc

349 lines
11 KiB
C++
Raw Normal View History

/* Copyright 2007 MySQL AB. 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., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
#include "mysql_priv.h"
#include "rpl_rli.h"
#include "rpl_record.h"
#include "slave.h" // Need to pull in slave_print_msg
#include "rpl_utility.h"
#include "rpl_rli.h"
/**
Pack a record of data for a table into a format suitable for
transfer via the binary log.
The format for a row in transfer with N fields is the following:
ceil(N/8) null bytes:
One null bit for every column *regardless of whether it can be
null or not*. This simplifies the decoding. Observe that the
number of null bits is equal to the number of set bits in the
@c cols bitmap. The number of null bytes is the smallest number
of bytes necessary to store the null bits.
Padding bits are 1.
N packets:
Each field is stored in packed format.
@param table Table describing the format of the record
@param cols Bitmap with a set bit for each column that should
be stored in the row
@param row_data Pointer to memory where row will be written
@param record Pointer to record that should be packed. It is
assumed that the pointer refers to either @c
record[0] or @c record[1], but no such check is
made since the code does not rely on that.
@return The number of bytes written at @c row_data.
*/
#if !defined(MYSQL_CLIENT)
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
size_t
pack_row(TABLE *table, MY_BITMAP const* cols,
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar *row_data, const uchar *record)
{
Field **p_field= table->field, *field;
int const null_byte_count= (bitmap_bits_set(cols) + 7) / 8;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar *pack_ptr = row_data + null_byte_count;
uchar *null_ptr = row_data;
my_ptrdiff_t const rec_offset= record - table->record[0];
my_ptrdiff_t const def_offset= table->s->default_values - table->record[0];
BUG#29549 (Endians: test failures on Solaris): Refactoring code to add parameter to pack() and unpack() functions with purpose of indicating if data should be packed in little-endian or native order. Using new functions to always pack data for binary log in little-endian order. The purpose of this refactoring is to allow proper implementation of endian-agnostic pack() and unpack() functions. Eliminating several versions of virtual pack() and unpack() functions in favor for one single virtual function which is overridden in subclasses. Implementing pack() and unpack() functions for some field types that packed data in native format regardless of the value of the st_table_share::db_low_byte_first flag. The field types that were packed in native format regardless are: Field_real, Field_decimal, Field_tiny, Field_short, Field_medium, Field_long, Field_longlong, and Field_blob. Before the patch, row-based logging wrote the rows incorrectly on big-endian machines where the storage engine defined its own low_byte_first() to be FALSE on big-endian machines (the default is TRUE), while little-endian machines wrote the fields in correct order. The only known storage engine that does this is NDB. In effect, this means that row-based replication from or to a big-endian machine where the table was using NDB as storage engine failed if the other engine was either non-NDB or on a little-endian machine. With this patch, row-based logging is now always done in little-endian order, while ORDER BY uses the native order if the storage engine defines low_byte_first() to return FALSE for big-endian machines. In addition, the max_data_length() function available in Field_blob was generalized to the entire Field hierarchy to give the maximum number of bytes that Field::pack() will write.
2007-10-11 18:18:05 +02:00
DBUG_ENTER("pack_row");
/*
We write the null bits and the packed records using one pass
through all the fields. The null bytes are written little-endian,
i.e., the first fields are in the first byte.
*/
unsigned int null_bits= (1U << 8) - 1;
// Mask to mask out the correct but among the null bits
unsigned int null_mask= 1U;
for ( ; (field= *p_field) ; p_field++)
{
DBUG_PRINT("debug", ("null_mask=%d; null_ptr=%p; row_data=%p; null_byte_count=%d",
null_mask, null_ptr, row_data, null_byte_count));
if (bitmap_is_set(cols, p_field - table->field))
{
my_ptrdiff_t offset;
if (field->is_null(rec_offset))
{
offset= def_offset;
null_bits |= null_mask;
}
else
{
offset= rec_offset;
null_bits &= ~null_mask;
/*
We only store the data of the field if it is non-null
For big-endian machines, we have to make sure that the
length is stored in little-endian format, since this is the
format used for the binlog.
*/
#ifndef DBUG_OFF
BUG#29549 (Endians: test failures on Solaris): Refactoring code to add parameter to pack() and unpack() functions with purpose of indicating if data should be packed in little-endian or native order. Using new functions to always pack data for binary log in little-endian order. The purpose of this refactoring is to allow proper implementation of endian-agnostic pack() and unpack() functions. Eliminating several versions of virtual pack() and unpack() functions in favor for one single virtual function which is overridden in subclasses. Implementing pack() and unpack() functions for some field types that packed data in native format regardless of the value of the st_table_share::db_low_byte_first flag. The field types that were packed in native format regardless are: Field_real, Field_decimal, Field_tiny, Field_short, Field_medium, Field_long, Field_longlong, and Field_blob. Before the patch, row-based logging wrote the rows incorrectly on big-endian machines where the storage engine defined its own low_byte_first() to be FALSE on big-endian machines (the default is TRUE), while little-endian machines wrote the fields in correct order. The only known storage engine that does this is NDB. In effect, this means that row-based replication from or to a big-endian machine where the table was using NDB as storage engine failed if the other engine was either non-NDB or on a little-endian machine. With this patch, row-based logging is now always done in little-endian order, while ORDER BY uses the native order if the storage engine defines low_byte_first() to return FALSE for big-endian machines. In addition, the max_data_length() function available in Field_blob was generalized to the entire Field hierarchy to give the maximum number of bytes that Field::pack() will write.
2007-10-11 18:18:05 +02:00
const uchar *old_pack_ptr= pack_ptr;
#endif
BUG#29549 (Endians: test failures on Solaris): Refactoring code to add parameter to pack() and unpack() functions with purpose of indicating if data should be packed in little-endian or native order. Using new functions to always pack data for binary log in little-endian order. The purpose of this refactoring is to allow proper implementation of endian-agnostic pack() and unpack() functions. Eliminating several versions of virtual pack() and unpack() functions in favor for one single virtual function which is overridden in subclasses. Implementing pack() and unpack() functions for some field types that packed data in native format regardless of the value of the st_table_share::db_low_byte_first flag. The field types that were packed in native format regardless are: Field_real, Field_decimal, Field_tiny, Field_short, Field_medium, Field_long, Field_longlong, and Field_blob. Before the patch, row-based logging wrote the rows incorrectly on big-endian machines where the storage engine defined its own low_byte_first() to be FALSE on big-endian machines (the default is TRUE), while little-endian machines wrote the fields in correct order. The only known storage engine that does this is NDB. In effect, this means that row-based replication from or to a big-endian machine where the table was using NDB as storage engine failed if the other engine was either non-NDB or on a little-endian machine. With this patch, row-based logging is now always done in little-endian order, while ORDER BY uses the native order if the storage engine defines low_byte_first() to return FALSE for big-endian machines. In addition, the max_data_length() function available in Field_blob was generalized to the entire Field hierarchy to give the maximum number of bytes that Field::pack() will write.
2007-10-11 18:18:05 +02:00
pack_ptr= field->pack(pack_ptr, field->ptr + offset,
field->max_data_length(), TRUE);
DBUG_PRINT("debug", ("field: %s; pack_ptr: 0x%lx;"
" pack_ptr':0x%lx; bytes: %d",
field->field_name, (ulong) old_pack_ptr,
(ulong) pack_ptr,
(int) (pack_ptr - old_pack_ptr)));
}
null_mask <<= 1;
if ((null_mask & 0xFF) == 0)
{
DBUG_ASSERT(null_ptr < row_data + null_byte_count);
null_mask = 1U;
*null_ptr++ = null_bits;
null_bits= (1U << 8) - 1;
}
}
}
/*
Write the last (partial) byte, if there is one
*/
if ((null_mask & 0xFF) > 1)
{
DBUG_ASSERT(null_ptr < row_data + null_byte_count);
*null_ptr++ = null_bits;
}
/*
The null pointer should now point to the first byte of the
packed data. If it doesn't, something is very wrong.
*/
DBUG_ASSERT(null_ptr == row_data + null_byte_count);
BUG#29549 (Endians: test failures on Solaris): Refactoring code to add parameter to pack() and unpack() functions with purpose of indicating if data should be packed in little-endian or native order. Using new functions to always pack data for binary log in little-endian order. The purpose of this refactoring is to allow proper implementation of endian-agnostic pack() and unpack() functions. Eliminating several versions of virtual pack() and unpack() functions in favor for one single virtual function which is overridden in subclasses. Implementing pack() and unpack() functions for some field types that packed data in native format regardless of the value of the st_table_share::db_low_byte_first flag. The field types that were packed in native format regardless are: Field_real, Field_decimal, Field_tiny, Field_short, Field_medium, Field_long, Field_longlong, and Field_blob. Before the patch, row-based logging wrote the rows incorrectly on big-endian machines where the storage engine defined its own low_byte_first() to be FALSE on big-endian machines (the default is TRUE), while little-endian machines wrote the fields in correct order. The only known storage engine that does this is NDB. In effect, this means that row-based replication from or to a big-endian machine where the table was using NDB as storage engine failed if the other engine was either non-NDB or on a little-endian machine. With this patch, row-based logging is now always done in little-endian order, while ORDER BY uses the native order if the storage engine defines low_byte_first() to return FALSE for big-endian machines. In addition, the max_data_length() function available in Field_blob was generalized to the entire Field hierarchy to give the maximum number of bytes that Field::pack() will write.
2007-10-11 18:18:05 +02:00
DBUG_DUMP("row_data", row_data, pack_ptr - row_data);
DBUG_RETURN(static_cast<size_t>(pack_ptr - row_data));
}
#endif
/**
Unpack a row into @c table->record[0].
The function will always unpack into the @c table->record[0]
record. This is because there are too many dependencies on where
the various member functions of Field and subclasses expect to
write.
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
The row is assumed to only consist of the fields for which the corresponding
bit in bitset @c cols is set; the other parts of the record are left alone.
At most @c colcnt columns are read: if the table is larger than
that, the remaining fields are not filled in.
@param rli Relay log info
@param table Table to unpack into
@param colcnt Number of columns to read from record
@param row_data
Packed row data
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
@param cols Pointer to bitset describing columns to fill in
@param row_end Pointer to variable that will hold the value of the
one-after-end position for the row
@param master_reclength
Pointer to variable that will be set to the length of the
record on the master side
@retval 0 No error
@retval ER_NO_DEFAULT_FOR_FIELD
Returned if one of the fields existing on the slave but not on the
master does not have a default value (and isn't nullable)
*/
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
int
unpack_row(Relay_log_info const *rli,
TABLE *table, uint const colcnt,
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar const *const row_data, MY_BITMAP const *cols,
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
uchar const **const row_end, ulong *const master_reclength)
{
DBUG_ENTER("unpack_row");
DBUG_ASSERT(row_data);
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
size_t const master_null_byte_count= (bitmap_bits_set(cols) + 7) / 8;
int error= 0;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar const *null_ptr= row_data;
uchar const *pack_ptr= row_data + master_null_byte_count;
Field **const begin_ptr = table->field;
Field **field_ptr;
Field **const end_ptr= begin_ptr + colcnt;
DBUG_ASSERT(null_ptr < row_data + master_null_byte_count);
// Mask to mask out the correct bit among the null bits
unsigned int null_mask= 1U;
// The "current" null bits
unsigned int null_bits= *null_ptr++;
uint i= 0;
table_def *tabledef= ((Relay_log_info*)rli)->get_tabledef(table);
for (field_ptr= begin_ptr ; field_ptr < end_ptr && *field_ptr ; ++field_ptr)
{
Field *const f= *field_ptr;
/*
No need to bother about columns that does not exist: they have
gotten default values when being emptied above.
*/
if (bitmap_is_set(cols, field_ptr - begin_ptr))
{
if ((null_mask & 0xFF) == 0)
{
DBUG_ASSERT(null_ptr < row_data + master_null_byte_count);
null_mask= 1U;
null_bits= *null_ptr++;
}
DBUG_ASSERT(null_mask & 0xFF); // One of the 8 LSB should be set
/* Field...::unpack() cannot return 0 */
DBUG_ASSERT(pack_ptr != NULL);
if ((null_bits & null_mask) && f->maybe_null())
f->set_null();
else
{
f->set_notnull();
/*
We only unpack the field if it was non-null.
Use the master's size information if available else call
normal unpack operation.
2007-08-03 17:12:00 +02:00
*/
uint16 const metadata= tabledef->field_metadata(i);
#ifndef DBUG_OFF
BUG#29549 (Endians: test failures on Solaris): Refactoring code to add parameter to pack() and unpack() functions with purpose of indicating if data should be packed in little-endian or native order. Using new functions to always pack data for binary log in little-endian order. The purpose of this refactoring is to allow proper implementation of endian-agnostic pack() and unpack() functions. Eliminating several versions of virtual pack() and unpack() functions in favor for one single virtual function which is overridden in subclasses. Implementing pack() and unpack() functions for some field types that packed data in native format regardless of the value of the st_table_share::db_low_byte_first flag. The field types that were packed in native format regardless are: Field_real, Field_decimal, Field_tiny, Field_short, Field_medium, Field_long, Field_longlong, and Field_blob. Before the patch, row-based logging wrote the rows incorrectly on big-endian machines where the storage engine defined its own low_byte_first() to be FALSE on big-endian machines (the default is TRUE), while little-endian machines wrote the fields in correct order. The only known storage engine that does this is NDB. In effect, this means that row-based replication from or to a big-endian machine where the table was using NDB as storage engine failed if the other engine was either non-NDB or on a little-endian machine. With this patch, row-based logging is now always done in little-endian order, while ORDER BY uses the native order if the storage engine defines low_byte_first() to return FALSE for big-endian machines. In addition, the max_data_length() function available in Field_blob was generalized to the entire Field hierarchy to give the maximum number of bytes that Field::pack() will write.
2007-10-11 18:18:05 +02:00
uchar const *const old_pack_ptr= pack_ptr;
#endif
BUG#29549 (Endians: test failures on Solaris): Refactoring code to add parameter to pack() and unpack() functions with purpose of indicating if data should be packed in little-endian or native order. Using new functions to always pack data for binary log in little-endian order. The purpose of this refactoring is to allow proper implementation of endian-agnostic pack() and unpack() functions. Eliminating several versions of virtual pack() and unpack() functions in favor for one single virtual function which is overridden in subclasses. Implementing pack() and unpack() functions for some field types that packed data in native format regardless of the value of the st_table_share::db_low_byte_first flag. The field types that were packed in native format regardless are: Field_real, Field_decimal, Field_tiny, Field_short, Field_medium, Field_long, Field_longlong, and Field_blob. Before the patch, row-based logging wrote the rows incorrectly on big-endian machines where the storage engine defined its own low_byte_first() to be FALSE on big-endian machines (the default is TRUE), while little-endian machines wrote the fields in correct order. The only known storage engine that does this is NDB. In effect, this means that row-based replication from or to a big-endian machine where the table was using NDB as storage engine failed if the other engine was either non-NDB or on a little-endian machine. With this patch, row-based logging is now always done in little-endian order, while ORDER BY uses the native order if the storage engine defines low_byte_first() to return FALSE for big-endian machines. In addition, the max_data_length() function available in Field_blob was generalized to the entire Field hierarchy to give the maximum number of bytes that Field::pack() will write.
2007-10-11 18:18:05 +02:00
pack_ptr= f->unpack(f->ptr, pack_ptr, metadata, TRUE);
DBUG_PRINT("debug", ("field: %s; metadata: 0x%x;"
" pack_ptr: 0x%lx; pack_ptr': 0x%lx; bytes: %d",
f->field_name, metadata,
(ulong) old_pack_ptr, (ulong) pack_ptr,
(int) (pack_ptr - old_pack_ptr)));
}
null_mask <<= 1;
}
i++;
}
/*
throw away master's extra fields
*/
uint max_cols= min(tabledef->size(), cols->n_bits);
for (; i < max_cols; i++)
{
if (bitmap_is_set(cols, i))
{
if ((null_mask & 0xFF) == 0)
{
DBUG_ASSERT(null_ptr < row_data + master_null_byte_count);
null_mask= 1U;
null_bits= *null_ptr++;
}
DBUG_ASSERT(null_mask & 0xFF); // One of the 8 LSB should be set
if (!((null_bits & null_mask) && tabledef->maybe_null(i)))
pack_ptr+= tabledef->calc_field_size(i, (uchar *) pack_ptr);
null_mask <<= 1;
}
}
/*
We should now have read all the null bytes, otherwise something is
really wrong.
*/
DBUG_ASSERT(null_ptr == row_data + master_null_byte_count);
BUG#29549 (Endians: test failures on Solaris): Refactoring code to add parameter to pack() and unpack() functions with purpose of indicating if data should be packed in little-endian or native order. Using new functions to always pack data for binary log in little-endian order. The purpose of this refactoring is to allow proper implementation of endian-agnostic pack() and unpack() functions. Eliminating several versions of virtual pack() and unpack() functions in favor for one single virtual function which is overridden in subclasses. Implementing pack() and unpack() functions for some field types that packed data in native format regardless of the value of the st_table_share::db_low_byte_first flag. The field types that were packed in native format regardless are: Field_real, Field_decimal, Field_tiny, Field_short, Field_medium, Field_long, Field_longlong, and Field_blob. Before the patch, row-based logging wrote the rows incorrectly on big-endian machines where the storage engine defined its own low_byte_first() to be FALSE on big-endian machines (the default is TRUE), while little-endian machines wrote the fields in correct order. The only known storage engine that does this is NDB. In effect, this means that row-based replication from or to a big-endian machine where the table was using NDB as storage engine failed if the other engine was either non-NDB or on a little-endian machine. With this patch, row-based logging is now always done in little-endian order, while ORDER BY uses the native order if the storage engine defines low_byte_first() to return FALSE for big-endian machines. In addition, the max_data_length() function available in Field_blob was generalized to the entire Field hierarchy to give the maximum number of bytes that Field::pack() will write.
2007-10-11 18:18:05 +02:00
DBUG_DUMP("row_data", row_data, pack_ptr - row_data);
*row_end = pack_ptr;
if (master_reclength)
{
if (*field_ptr)
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
*master_reclength = (*field_ptr)->ptr - table->record[0];
else
*master_reclength = table->s->reclength;
}
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
DBUG_RETURN(error);
}
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
/**
Fills @c table->record[0] with default values.
First @c empty_record() is called and then, additionally, fields are
initialized explicitly with a call to @c set_default().
For optimization reasons, the explicit initialization can be skipped for
first @c skip fields. This is useful if later we are going to fill these
fields from other source (e.g. from a Rows replication event).
If @c check is true, fields are explicitly initialized only if they have
default value or can be NULL. Otherwise error is reported.
@param table Table whose record[0] buffer is prepared.
@param skip Number of columns for which default value initialization
should be skipped.
@param check Indicates if errors should be checked when setting default
values.
@returns 0 on success or a handler level error code
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
*/
int prepare_record(TABLE *const table,
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
const uint skip, const bool check)
{
DBUG_ENTER("prepare_record");
int error= 0;
empty_record(table);
if (skip >= table->s->fields) // nothing to do
DBUG_RETURN(0);
/* Explicit initialization of fields */
for (Field **field_ptr= table->field+skip ; *field_ptr ; ++field_ptr)
{
uint32 const mask= NOT_NULL_FLAG | NO_DEFAULT_VALUE_FLAG;
Field *const f= *field_ptr;
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
if (check && ((f->flags & mask) == mask))
{
my_error(ER_NO_DEFAULT_FOR_FIELD, MYF(0), f->field_name);
error = HA_ERR_ROWS_EVENT_APPLY;
}
else
f->set_default();
}
DBUG_RETURN(error);
}
#endif // HAVE_REPLICATION